united States
Environmental Protection
Agency
Water
OillCH OT
Drinking Water (WH-550)
Washington, DC 20460
cr« 3/u/5j-oh-vwh\u;
June 1934
National Statistical
Assessment of Rural Water
Conditions
Volume li
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)'
1. REPORT NO. |2.
EPA 570/9-84-004-B"
3. RECIPIENT'S ACCESSION NO.
pRR & 222348
4. TITLE AND SUBTITLE
National Statistical Assessment of Rural
Water Conditions, Volume : jj
6. REPORT DATE
June 1984
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Department of Rural Sociology
Cornell University
B. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Cornell University
Department of Rural Socialocy
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
Office of Drinking Water (WH-550)
Washington, DC 20460
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
Four Volume Set
16. ABSTRACT
This study considered five dominant dimensions of the status of
domestic^water: quality, quantity, availability, cost and
affordability. Rural residents were asked about health effects
but the results were modest in that very few rural residents
reported adverse health conditions which they associated with the
water supply.
With enactment of the Safe Drinking Water Act of 1974, Congress set
in motion two major efforts to develop systematic, current data
on rural water supplies across the nation. First, in response to
growing concern with the quality of drinking water and its effects
on human health, the Safe Drinking Water Act provided for a uniform,
national set of water quality standards and extended the monitoring'
and regulatory responsibility of the US Government over smaller water
supplies. Second, the Act mandated a one-time national statistical
assessment of the current status of rural domestic water characteris-
tics. This document fulfills that mandate.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Rural water
drinking water
Rural water conditions
water supply systems
water quality
water quantity
18. DISTRIBUTION STATEMENT
19. SECURITY CLASS (This Report)
non-sensitive
21. NO. OF PAGES
•m
20. SECURITY CLASS (This page)
22. PRICE
EPA Form 2220-1 (R«*. 4-77) previous edition i* omolets
I
l
-------
-------
V
Status of US Rural Water Supply
Congress recognized in the Safe Drinking Water Act that an assessment of
rural water conditions required investigation of a number of interrelated compon-
ents of water supply. All of those components—not just water quality—were
studied in the NSA. This chapter, the central descriptive chapter of the NSA
report, presents a comprehensive account of the status of household water
conditions in rural America. Findings are arranged according to their relevance to
the various dimensions of rural water supply. Later chapters of the report explore
the relationships between these findings and other NSA data.
Congress sensed that rural water conditions were best described by water
quality, quantity, and availability, and it specified that these three factors were to
be studied in the rural water survey. At the same time, the Safe Drinking Water
Act mandated national, legally enforceable actions affecting a number of aspects
of public water supply. The broad scope of the legislation required an equally
broad study of the technical, economic, and institutional aspects of water
conditions. In view of that orientation, it became clear to EPA officials that the
three designated factors—quality, quantity, and availability—would have to be
v-1
-------
V - 2
defined comprehensively, and that other factors would have to be included in the
study.
In light of these considerations, the status of rural water supplies was
described in the NSA in terms of five primary factors: quality, quantity,
availability, cost, and affordability. In addition, the survey questioned rural water
users about the effects of water quality, quantity, and availability on their
households. Each of these broad subjects is explored in detail in this chapter.
The status of rural water supplies described here was determined in the
rural household. This emphasis was in keeping with the Congressional directive to
obtain information on the number of rural residents who had inadequate service,
limited access to supply, exposure to waterborne health risks, or outright water-
borne illnesses. The new federal drinking water regulations reiterated this concern
by requiring that most quality standards be met in the consumer's household, at his
tap, rather than just at the supply facility or at the source. This approach
recognized that conditions at other points along the distribution system were
important, but that household water conditions had to be judged in the consumer's
home.
In this chapter, as is the practice in the NSA, variables describing status
include both laboratory-measured values (for water quality, for example) and
perceived values (such as the user's evaluation of the water's taste and appear-
ance). This approach allowed analysis of conditions which required laboratory
measurement and of conditions which needed to be assessed by personal appraisal.
Laboratory-measured values are discussed first, and perceived values second.
QUALITY
Of the five primary factors used in the NSA to delineate rural water
conditions, quality is taken up first for both historical and pragmatic reasons.*
Historically, from porous vessel filtration of water in ancient Egypt to deactivation
-------
V -3
of the infamous cholera-contaminated Broad Street well in nineteenth-century
England, a major concern has been with the purity of drinking water. Even the
NSA stemmed originally from worry about US waterborne disease outbreaks in the
1960s. This worry was intensified by the discovery of substandard drinking water in
many community water supply facilities and in consumers' homes in 1969 during a
nationwide study (the US Public Health Service Community Water Supply Survey,
cited under Reference 1).
Pragmatically, household water quality was the factor which could be
studied most thoroughly in the NSA. In this context, 'quality of water' referred to
the suitability of water for human use. The major consideration was that the water
not present a health threat to human beings. However, the NSA concept of water
quality also included aesthetic and economic considerations. Overall, the NSA
concept of quality was consistent with the definition of a "functionally ideal"
public water supply as adopted by the Board of Directors of the American Water
Works Association in 1968. The definition stated, in part: "Ideally, water delivered
to the consumer should be clear, colorless, tasteless, and odorless. It should
contain no pathogenic organisms and be free from biological forms which may be
harmful to human health or aesthetically objectionable. It should not contain
concentrations of chemicals which may be physiologically harmful, esthetically
objectionable, or economically damaging. The water should not be corrosive or
incrusting to, or leave deposits on, water-conveying structures through which it
passes, or in which it may be retained, including pipes, tanks, water heaters, and
2
plumbing fixtures."
Major emphasis in the NSA was given to bacteriological, physical, and
chemical water constituents which traditionally have characterized water quality
(see Table V-l). Many of the NSA measurements were relevant to new federal
drinking water regulations. The major focus was on measurements of health-
related constituents of water, particularly in the subsample of 10 percent of the
-------
V -4
Tabde V-l
Constituents Measured in NSA Survey
Category
Constituent
Has Primary (P),
Secondary (S),
or
No (N) MCL
Measured in All
NSA Household
Samples or only
in Group II
Subsample
Microbial
Physical
and
Chemical
Inorganic
Organic
Total coliform P
Fecal coliform N
Fecal streptococcus N
Standard plate count N
Fecal coliform/fecal
streptococcus ratio N
Turbidity P
Color S
Temperature N
Specific conductance N
Total dissolved solids
(as determined from
conductance) S
Hardness
(as determined from
calcium and magnesium) N
Calcium N
Magnesium N
Nitrate-N P
Sulfates S
Iron S
Manganese S
Sodium N
Lead P
Arsenic P
Selenium P
Fluoride P
Cadmium P
Mercury P
Chromium P
Barium P
Silver P
Endrin P
Lindane P
Methoxychlor P
Toxaphene P
2,4-D P
2,4,5-TP P
All
All
All
All
All
All
All
All
All
All
All
All
All
All
All
All
AH
All
All
Subsample
Subsample
Subsample
Subsample
Subsample
Subsample
Subsample
Subsample
Subsample
Subsample
Subsample
Subsample
Subsample
Subsample
-------
V -5
Table V-l (continued)
Category
Constituent
Has Primary (P),
Secondary (S),
or
No (N) MCL
Measured in All
NSA Household
Samples or only
in Group II
Subs am pie
Radioactive
Gross alpha
Gross beta
~Radium 226
~Radium 228
~Uranium
~Stontium-89
*Strontium-90
~Cesium-134
~Tritium
~Iodine-131
P
P
P
P
P
P
P
P
P
P
Subsample
Subsample
~Measured only if the laboratory analyst considered gross alpha or gross beta
readings sufficient to warrant further investigation.
-------
V -6
NSA water specimens (see "Constituents Studied in NSA," below), but some
determinations were related more to aesthetic and economic considerations.
The regulations provide one, but only one, body of standards for interpret-
ing the implications of NSA findings. In this chapter, federal standards are
compared with other existing criteria and standards in order to present a broader
context for interpreting the NSA findings.
As to the specific terminology used in the federal regulations, there are
two levels of standards which have been established. One level is the interim
primary Maximum Contaminant Level (MCL). Primary MCLs are numbers which
refer to specific concentrations of individual constituents. The specific concentra-
tions cannot be exceeded in public drinking water supplies which have fifteen or
more connections or which regularly serve 25 or more people. The requirement is
mandatory since the constituents in question are regarded as possible health
threats if they are present in excessive concentrations.
The other regulatory level is the secondary Maximum Contaminant Level.
Secondary MCLs also are numbers which specify concentrations of constituents,
but the specifications are recommended, not legally enforceable by the US
government. The constituents involved are considered to have aesthetic or
economic consequences, but only minor or uncertain health effects.
NSA WATER QUALITY REFERENCE VALUES
In order to assess household water quality, it was desirable to develop a set
of reference values for all of the constituents- studied in the NSA. The federal
primary and secondary drinking water regulations (which were developed to assess
the quality of community water systems) composed one set of standards which
provided appropriate bases, but other standards also were consulted in developing
the NSA reference values (see Table V-2).
-------
V -7
Tahle V-2
NSA Reference Values for Constituents Measured
in NSA Survey
Constituent
NSA Reference Value
(milligrams per
liter of water, unless
ot her wise not ed)
~Basis for
Reference
Value
Purpose or Effect
of Constituent
Total coliform
bacteria
Fecal coliform
bacteria
Fecal
streptococci
Fecal coliform/
fecal strepto-
coccus ratio
Not more than one bacterium
per 100 milliliters of water
Complete absence of bacteria
in a 100-milliliter sample
None
None
MCL(P) Indicator of
infectious disease
potential
EPA Indicator of
infectious disease
potential
Indicator of
possible infectious
disease potential
Indicator of human
versus animal
contamination
Standard
plate count
500 colony-forming units
per one milliliter of water
NRC General indicator
of bacteria level
Turbidity
None
Aesthetic, health
Color
15 color units
MCL(S)
Aesthetic
Tem perature
None
Aesthetic
Specific
conductance
(normalized at
25° C)
None
Used for estimating
total dissolved solids
Total dissolved 500 MCL(S) Economic, aesthetic
solids (as derived
from specific
conductance)
-------
V -8
Table V-2 (continued)
Constituent
NSA Reference Value
(milligrams per
liter of water, unless
otherwise noted)
~Basis for
Reference Purpose or Effect
Value of Constituent
Hardness
Calcium
Magnesium
Nitrate-N
Sulfates
Iron
Manganese
Sodium
Lead
Arsenic
Selenium
Fluoride
Cadmium
Mercury
None
None
125
10
250
0.3
0.05
More stringent: 20
Less stringent: 100
0.05
0.05
0.01
1.4
0.01
0.002
Various
NRC
MCL(P)
MCL(P)
MCL(P)
MCL(P)
MCL(P)
MCL(P)
Economic
Aesthetic,
economic
Aesthetic,
economic,
health
MCL(P) Health
MCL(5) Aesthetic, health
MCL(S) Aesthetic
MCL(S) Economic,
aesthetic
Health
Health
Health
Health
Health
Health
Health
-------
V -9
Table V-2 (continued)
Constituent
NSA Reference Value
(milligrams per
liter of water, unless
otherwise noted)
*Basis for
Reference
Value
Purpose or Effect
of Constituent
Chromium
0.05
MCL(P)
Health
Barium
1
MCL(P)
Health
Silver
0.05
MCL(P)
Health
Endrin
0.0002
MCL(P)
Health
Lindane
0.004
MCL(P)
Health
Methoxychlor
0.1
MCL(P)
Health
Toxaphene
0.005
MCL(P)
Health
2, 4-D
0.1
MCL(P)
Health
2, 4, 5-TP
0.01
MCL(P)
Health
Gross alpha
radioactivity
See Figure V-28
MCL(P)
Health
Gross beta
radioactivity
50 pCi
MCL(P)
Health
Radium 226
Radium 228
Other radio-
nuclides
(uranium,
strontium-89,
strontium-90,
cesium-134,
tritium,
iodine-131)
These constituents
were not measured
frequently enough
to provide independent
national estimates
(See text for details
about NSA reference values.)
Health
*See text for details: MCL(P) indicates interim primary Maximum Contaminant
Level, MCL(S) indicates secondary Maximum Contaminant Level; EPA stands for
US Environmental Protection Agency, NRC for the National Research Council, and
"Various" for several sources which are described in the text.
-------
V - 10
As to the federal drinking water regulations, there was an important
distinction between the requirements for arriving at measurements under the EPA
regulations and the procedure followed in the NSA. Under the regulations
3
governing administration of the interim primary MCLs, sampling was to be done
over designated periods of time, and compliance with the MCL requirement was
judged not on the basis of just one finding, but on the basis of the average of
several findings. The secondary drinking water regulations were not federally
enforceable, but the same sort of sampling and averaging process was envisioned
for their application
The sampling and averaging process is used to monitor public drinking
water supplies systematically and to identify situations in which excessive concen-
trations of certain materials appear to pose persistent problems. The averaging
provision reduces the chance that a single, temporary elevation of one substance
would bring the supply into noncompliance with regulations. In the NSA, on the
other hand, snly one set of specimens was collected. All of the NSA findings thus
were based on single-specimen values, not on average values for multiple collec-
tions. The MCL itself, however, had validity as a measure of health or aesthetic
consequence. That value, then, frequently was used as the basis for the NSA
reference value. The MCLs and other standards are discussed and compared where
appropriate in this report, and the basis for the NSA reference value is stated.
As a note on terminology used in subsequent sections on laboratory
findings, measured values which are larger than the NSA reference values are
reported as exceeding (surpassing or being above) the reference values. Measured
values which are equal to or less than the reference values may be reported as
within or below the reference values. The terms "household" and "supply" are used
interchangeably. In this regard, it is important to note that the set of specimens
for quality studies was drawn from the one major water supply in each household.
-------
V - 11
In 97.0 percent of households, that supply provided drinking water as well as
satisfying other domestic water requirements (see Chapter IV).
HEALTH-RELATED CHARACTERISTICS
The emphasis on traditional health-related characteristics was felt to be in
keeping with recent public health trends. Microorganisms have been the cause of
the largest reported outbreaks of waterborne disease in the US during the past
twenty years, even though chemical and radioactive contaminants have posed new
problems. The number of outbreaks (defined as at least two cases of certain
specified illnesses), after declining from 1938 through 1950, began to rise after
1950 (see Figure V-l).
Whatever the reason for the increase, the outbreaks were attributed
primarily to microscopic organisms rather than to chemical contaminants. This
tendency is apparent in the types of illness outbreaks from 1971 to 1974 (see Table
V-3). During that period, 90.9 percent of the outbreaks were caused by micro-
organisms—only 9.1 percent of the outbreaks were caused by chemical contamin-
ants. Nevertheless, the number of individual cases of chemical poisoning (474) was
larger than the number for several infectious diseases.
A complication in considering chemical contamination is that effects of
industrial pollutants, ranging from asbestos to exotic organic chemicals, may be
serious but so subtle or delayed as to avoid detection by present methods. To add
to the difficulty, only a small portion of organic contaminants in water have been
identified at all. According to the NRC: "Although approximately 90 percent of
the volatile organics in drinking water have been identified and quantified, these
represent no more than ten percent of the total organic material. Only five to ten
percent of the nonvolatile organic compounds, which comprise the remaining 90
percent of the total organic material in water, have been identified."*'
-------
V - 12
Figure V-1
Average Annual Number of Waterborne Disease Outbreaks,* 1938 - 1975
1938" 1 1941- 1 1946- 1 1951" ' 1956" 1 1961" 1 1966" 1 1971-
1940 1945 1950 • 1955 I960 1965 1970 1975
Years
* An outbreak consisted of at feast two coses of certain reportable illnesses.
Source: National Academy of Sciences. Drinking Water and Health. Washington,DC. National Academy of Sciences, 1977, p.64.
-------
V - 13
Table V-3
Summary of Water borne Illness Outbreaks
and Cases, 1971-74
Illness
Outbreaks
Cases
Acute gastroenteritis
(cause unknown)
46
7,992
Bacterial
Shigellosis
Typhoid fever
Salmonellosis
13
4
2
2,747
222
37
Viral
Infectious hepatitis
13
351
Protozoan
Giardiasis
(includes 4,800 cases in one
outbreak at Rome, New York)
12
5,127
Chemical
Chemical poisoning
9
474
Total
99
16,950
Source: Adapted from National Academy of Sciences.
Drinking Water and Health. Washington, DC: National
Academy of Sciences, 1977, p. 65.
-------
V - 14
As to radioactive contaminants, the NRC concluded that "the radiation
associated with most water supplies is such a small portion of the normal
background to which all human being are exposed that it is difficult, if not
impossible, to measure any adverse health effects with certainty."
The pattern of federal regulations for drinking water is in keeping with the
state of scientific knowledge about water contamination. Thus, heavy emphasis is
on acceptable bacterial content of water, coupled with surveillance of public water
supplies to ensure that the limits are not exceeded. Considerable emphasis also is
given to traditional tests for water turbidity, an optical measure of suspended
substances in water which may harbor or protect microorganisms. Less emphasis is
given to surveillance for inorganic materials, although mandatory limits are
established for those with apparent health effects, except in noncommunity
systems where long-term exposure is assumed to be limited. The emphasis on
organic materials (limited to insecticides and herbicides at the time of the NSA) is
on substances known to have serious toxic properties. Emphasis also is given to
surveillance for certain levels of radioactivity, with little expectation of finding
significant levels in public drinking water except in isolated cases.
In a broader context, the relation of water quality to human health is
complex, and it is necessary to identify those aspects of the relationship which
were considered in the NSA study. Public health officials generally relate water
quality to the occurrence of specific, potentially dangerous substances in drinking
water supplies. Health nutritionists, on the other hand, often relate water quality
as well to the presence of certain substances required in the human diet. The
public health official's main concern is prevention of waterborne illness; the
nutritionist's main concern is dietary adequacy.
The emphasis in the NSA investigation was on the traditional concern of
public health officials with levels of materials that might have adverse health
effects. Thus, even though a number of constituents measured in the NSA had
-------
V - 15
potentially beneficial effects in small amounts, the emphasis was on the hazard
they migfit create in larger amounts.
Despite the NSA focus on potential health hazards in water, it was not
possible within the confines of the study to trace the implications of water
conditions which indicated potential health problems. To do this would have
required expensive, time-consuming epidemiological research to link particular
health problems with conditions in household water supplies. In some cases, long
periods of time would have been needed to identify cumulative, delayed, or
interrelated health effects. Such research on a national scale would have been far
beyond the scope of the project. In addition, it was possible to take only one set of
specimens of water from each household at one particular point in time. This was
adequate to provide an indication of overall national and regional situations. One
sample could not represent the range of conditions which occurred during an entire
year, however.
Faced, with these problems, NSA investigators foresaw serious difficulties
in one section of the Congressional mandate for the rural water survey. That
section directed that the survey include "consideration of the number of residents
in each rural area . . . who have experienced incidents of chronic or acute illness,
which may be attributed to the absence or inadequacy of a drinking water supply
system."
Possible methods for meeting this directive were reviewed by EPA and by
NSA investigators. Advice was sought from the US Health Interview Survey, the
Health Examination and Nutrition Survey, the Office of Vital Statistics, and the
Center for Disease Control. These government organizations have had extensive
experience with clinical and epidemiological techniques applicable to public health
studies. Officials from these organizations pointed out the many obstacles to
obtaining useful information in an extensive, one-time, cross-sectional survey such
as the NSA. For example, clinical examination of respondents would be far too
-------
V - 16
limited to provide meaningful public health conclusions. Available statistical data
from health departments would be incomplete and outdated. Epidemiological
studies would be of limited value without solid evidence of localized outbreaks.
Perhaps most important, integration of data to provide a reliable national picture
of public health problems related to drinking water would be risky and very likely
misleading.
In view of these considerations, NSA investigators concluded that the NSA
could not provide a direct, comprehensive assessment of the adverse health effects
of rural water supplies. Nevertheless, questions about perceived water-related
illness were included in the NSA interview schedule. Responses, however, may
have been more significant indicators of the residents' awareness of possible links
between health problems and water supplies than they were of actual health
effects of the supplies. The results of these questions and others related to
possible health effects of water supplies are discussed in the section of this chapter
entitled "Effects of Quality, Quantity, and Availability."
In summary, the only measure of health effects in the NSA is strictly a
measure of potential health hazards posed by constituents found in the water
specimens collected at rural households. Measurement of the constituents provides
an indication of the immediate risk to persons living in the households. In addition,
the analysis of "at-risk" US rural households could contribute to public health
statistical investigations and to assessment of the need for countermeasures.
WATER QUALITY DATA—PERSPECTIVES
Every scientific endeavor strives to collect, transport, process, synthesize,
and report with minimal distortion. The practical hope is to reduce errors to small
random perturbations.
From the time that the NSA was planned, a variety of procedures were
included to minimize the possibilities for error. Interviewers were intensively
-------
V - 17
trained on every aspect of the survey during a two-week training course. Their
work in the field was checked by multiple independent examinations of their
completed forms, through telephone follow-ups, and by actual on-site, follow-up
visits by supervisors at randomly selected households. Water specimens were
preserved at the time of collection—ice for microbiological samples, mercuric
chloride for the nitrate samples and nitric acid for the metals and radiation
samples. Every effort was made to insure that the collection was careful and
consistent. Only EPA laboratories or EPA-approved laboratories were used to
analyze the water.
Nevertheless, errors invariably arise in any large-scale study. Hence, even
after data collection was completed, further checking was employed to uncover
and correct errors. The effects of errors can be characterized in two ways. There
are errors which leave the estimate of the mean unaffected but cause the data to
be dispersed. This imprecision is reflected in large standard deviations. Errors of
this nature are difficult to correct since they are not consistent in their effect.
The second type of error is usually called bias. It causes the estimates of means to
be displaced, but may not affect the standard deviation. The two types of error
may occur together.
Two approaches were used throughout the NSA for assessing the presence
of errors in the collected data. First, data sets with results that had some overlap
with the content of the NSA were examined to see how well the overlapping results
aligned. Second, extensive internal examinations were undertaken.
Regarding the first approach, comparisons of NSA results with indepen-
dently collected studies of a similar nature were made. This was the approach used
in part of Chapter III and Appendix B. Similar findings indicate either that (1) both
efforts reasonably reflect the true situation or (2) both err but happen to arrive at
the same result. Since the latter case was not likely, a pattern of consistent
results was taken to mean that the total effect of error was not meaningful.
-------
V - 18
Since the N5A was the first national investigation of rural water condi-
tions, there were no completely comparable data sets. Though the US Census had
some overlap with NSA's socioeconomic data, comparable data sets of national
rural water quality data did not exist. Other data sets differed from the NSA in
important ways. Some were not national in scope. Most were not rural oriented.
Some were not systematically taken. In some, water samples had not been
appropriately preserved, or the water samples had been analyzed with incompatible
techniques, were incompatibly reported, and so forth. (The EPA's federal reporting
data system, FRDS, does collect water quality data for community water systems,
including rural systems. It holds promise of becoming a comparable sourCe for
those systems affected by federal monitoring guidelines.) Though they were not
directly comparable, these other data sets generally had lower values than the
NSA, particularly among the metals readings. The lack of comparable data
necessitated the use of the second, internal approach to detection of errors.
The second, more tedious approach involved an internal examination of
NSA research procedures to try and identify sources of consistent error. An
internal data examination is generally less desirable because it requires more
effort and is necessarily inconclusive since all possible error sources cannot be
examined. The best attainable conclusion from such an approach would be that of
the possible problems examined, none showed a systematic error pattern.
The internal examination of the water quality data was crucial to some
NSA water quality findings, especially the metals results, which generally had
higher readings and a higher proportion of samples with high readings than was
expected by many professionals. Lead, cadmium, and mercury were the most
notable standouts. Work by other researchers (to be discussed later) provided a
strong indication that the preservative ampules of nitric acid could have affected
the lead and cadmium results. Internal detection of errors in the water quality
data was conducted in two steps. Step one involved an examination of interviewer
-------
V - 19
behavior, collection, and transporting of the water specimens. Step two involved
an examination of laboratory procedures. The results of that examination,
presented below, include qualifications and cautions which are appropriate in order
to establish a realistic basis for interpreting the laboratory measures of water
quality. More detailed documentation of the NSA's internal assessment of the
laboratories' performance is contained in Appendix C. Appendix C is not bound
with the NSA report but can be obtained by contacting the Director, The Office of
Drinking Water, US Environmental Protection Agency, 401 M Street S.W.,
Washington, DC, 20460.
Quality assurance results
Investigation of interviewer, handling, and transporting procedures did not
identify any systematic error source. There were no systematic patterns of
inordinately high or low findings among interviewers that were inconsistent with
readings for other interviewers in the same general geographic area.
Seven laboratories participated in the NSA water quality assessment. Each
of the laboratories performed standard checks to assure consistency, accuracy, and
validity. But, the results of these checks are not standardly requested by the data
user. While NSA researchers did request these quality assurance data, they were
not always available or interpretable for a variety of reasons. For those
laboratories which could be checked, original laboratory notebooks were compared
with NSA reporting forms, data key punched, and finally with the computer
analysis tapes. An error rate ranging from 1 to 3 percent was apparent in
comparing the laboratory notebooks to the computer tape (many of those errors
were corrected in the checking process). Every number reported from the
laboratories was independently checked at least twice, but usually four times.
Some important transcription errors were discovered and corrected for mercury.
Some metals specimens, preserved with nitric acid, sent to the EPA laboratory in
-------
V - 20
Las Vegas were inadvertently misplaced and were not analyzed for as long as nine
months. These specimens were mistakenly transferred to an unrefrigerated
warehouse. An examination of the readings for those misplaced specimens did not
reveal an identifiable effect from the improper handling. However, a thorough
resolution of that question was not possible without experimentation.
The course of investigating the validity and reliability of the data did not
identify any procedure or laboratory effect which would suggest aberrant results.
But again, some important avenues of inquiry were unavailable. For instance,
randomly interspersed blank specimens of laboratory pure water were not sent to
the laboratories. They could have indicated bias in drawing, handling, or analyzing
the specimens. Specimens from the same household, drawn at the same time, were
not sent to more than one laboratory. So no direct test of laboratory bias was
available. A variety of other experiments testing storage time, container
characteristics, preservation technique, and so on would have been useful but were
not performed.
The following is a summary of specific findings of the quality assurance
which was derived from laboratory records.
Microbiology
Three laboratories participated in the bacteriological investigation of NSA
water specimens. The ERCO (Energy Resources Company, Inc.) of Boston,
Massachusetts, did the bulk of the work. The Madison County Environmental
Center of Edwardsville, Illinois, and the Colorado State Health Department of
Denver, Colorado, were the other laboratories.
A polypropylene, autoclaved, sealed, one quart container was used to
collect the water. During the collection, the bottle lid was held suspended face
down. Neither it nor the interior of the bottle were allowed to come in contact
with anything but the flowing tap water. The filled container was then sealed in a
-------
V - 21
plastic bag and placed in a styrofoam shipping container. Ice, in a separate sealed
plastic bag, was packed with the specimen. The styrofoam box was closed, sealed,
and sent to the nearest of the three laboratories, usually by airplane. The
laboratories picked up the incoming specimens and began the analysis within thirty
hours from the time of collection. If this deadline was not met, the water was
discarded and the interviewer instructed to collect a new water specimen.
A detailed system of checks was used to test the validity of the organism
identification and the accuracy of the count. The EPA's Office of Research and
Development, Environmental Monitoring and Support Laboratory (EMSL) followed
the quality assurance data (QA) from the microbiological analyses and drew the
following conclusion upon transmitting the detailed results of their monitoring:
"The importance of a vigorous QA program was evidenced in the
detection and resolution, early in the study, of technical problems
and differences that occurred in laboratory operations and data
reporting by participants. We believe the . . . QA protocol (for
microbiology) was conscientiously followed. The quality control
report forms were monitored regularly by EMSL. We conclude that
the QA program (for microbiology) was appropriate for the study
and did confirm the validity of the test data."
Radiation
All radiological investigations were conducted at the EPA Environmental
Systems Monitoring Laboratory at Las Vegas. A reorganization of the laboratory
following the NSA survey resulted in the dispersion of personnel and the loss of
records. There were, therefore, no data available to describe the results of quality
control efforts for the radiation results.
Chlorinated hydrocarbons
Two laboratories participated in the examination of chlorinated hydro-
carbons: The South Carolina Epidemiologic Study Laboratory and the Mississippi
State Chemistry Laboratory. As with the radiological examination, no quality
-------
V - 22
control data were recovered regarding laboratory procedures with the NSA data.
However, results on the EMSL check samples were available for the first month of
the NSA data collection. While no examination of NSA specimens occurred in
these laboratories until some time later, these were the only quality control data
available. At the time the check samples were analyzed, the laboratory did
perform within EPA's acceptance limits.
Physical, chemical, and inorganic analyses
Two laboratories participated in this segment of the NSA investigation.
Both are EPA laboratories: one, the Environmental iWonitoring Systems Laboratory,
is located in Las Vegas, Nevada; the other, the Environmental Monitoring and
Support Laboratory, is in Cincinnati, Ohio. Both laboratories performed analyses
on the same constituents. The Cincinnati laboratory received about two-thirds of
the specimens while the Las Vegas laboratory did one-third. The specimens were
divided between the laboratories largely on the basis of which laboratory was
closest to the sampled household.
Tables V-A, V-B, and V-C, at the end of this section, display results of
quality assurance procedures for some of the data generated by the Cincinnati and
Las Vegas laboratories.
There are three types of inquiry reflected in the tables. Table V-A displays
the differences between the value reported for a household and a duplicate reading
on a separate aliquot from the same sample. Ideally, the differences should tend
toward a mean and standard deviation of zero. Assuming homogeneity throughout
a particular water sample, the duplicate reading provided an indication of variabil-
ity in the measuring process. The data shown were not for the entire NSA sample,
but for the subset on which quality assurance data were aggregated. Thus, the
columns showing ranges of data may not correspond to ranges presented elsewhere
in the report when the full NSA data is discussed.
-------
V - 23
Table V-B reports percentages of spike recovery. Spiking involves purpose-
fully adding a known amount of the substance being investigated to a tested
aliquot, then retesting to see what percent of the spike is recovered in the new
reading. The process can reflect both validity—whether the chemistry is measur-
ing what is intended—and the precision of the range. The mean should tend toward
100 percent and the standard deviation to zero.
Table V-C displays the results of measuring known standard solutions
interspersed with the test samples. The differences between the expected value
and the measured result should tend toward a mean and a standard deviation of
zero. These results were only for those analysis runs for which the measured
standards were within the laboratory acceptance limits. All results from runs in
which the standard readings were unacceptable were automatically discarded and
the water retested.
The data in the tables came from a selected sample (usually 5 or 10
percent) of the cases reported in the NSA. Some of the apparent instability was
directly a result of having few data points. This caution is particularly relevant for
those constituents studied only in the NSA 10 percent subsample.
The lack of data in some parts of the tables does not necessarily mean the
work was not done by the laboratory, rather that laboratory quality assurance
procedures are not normally reported to the data users. Incomplete communica-
tion, unavailable records, lack of time, and the like made some of these data
unavailable. Examination of some of the laboratory records suggested that less
care was taken in transcribing quality assurance results than was taken for the
primary findings. That may account for some but certainly not all of the
variability reflected in the tables. Some of the quality assurance results suggest
that more careful laboratory control could have been exercised.
The results for two parameters need some special qualification. Turbidity
and color were measured at the laboratory—not in the field. Ideally, they should
-------
V - 24
be assessed at the time the water is drawn. There can be precipitation or other
physical, chemical, or biological actions from the time of drawing to laboratory
analysis which could change the readings for both turbidity and color. The EPA
considered that consistently accurate field readings by interviewers would be more
of a problem than the possible inaccuracies induced by waiting to take the readings
under controlled laboratory conditions. The effect of that decision upon the data is
unknown.
Some of the constituents were measured on a 10 percent subsample of
interviewed households. That reduction in data reduces the statistical confidence
which can be associated with point estimates (such as means, medians, and
standard deviations). Cadmium and mercury were among the subsample constitu-
ents an£, as previously mentioned, they were found in greater concentrations and in
greater proportions of household water supplies than expected by many profes-
sionals, suggesting a possible bias or that the smaller case base resulted in an
unrepresentative sample, or both.
Edward Calabrese et al. reported in the Bulletin of Environmental
Contamination and Toxicology (September, 1979, pp. 107-111) that the preservative
ampules purchased from the same corporation as those used in the NSA could bias
lead and cadmium results in drinking water analyses. Apparently the thin line of
blue paint marking the appropriate breakpoint on the ampule neck contained lead
and cadmium. Shaking the acid from the opened ampule could contaminate the
water sample with sufficient paint to alter subsequent measurement of the two
metals. If the mean contamination elevation in NSA water samples was the same
as discovered by Calabrese, then the NSA's estimates are artifically elevated by 36
parts per billion lead and 0.92 parts per billion cadmium. While there is no way to
ascertain the true effect on each water sample, it is likely that the average effect
was probably similar to that found by Calabrese. (The manufacturer no longer
employs the painted marking line.)
-------
V - 25
Quality assurance summary and conclusions
Bias in the data will shift the estimated concentrations either higher or
lower than what was actually present in the water. Two groups of data, radiation
and chlorinated hydrocarbons, had uniformly low readings. There were no data
available for inquiring whether these findings were biased. If they were biased
high, then the implications are not serious. But, if they were biased low, then the
NSA suggests inappropriate complacency. For the other water quality data,
extensive quality assurance information was investigated but no bias (other than
for lead and cadmium) was identified. Nevertheless, if the data are biased high,
then the NSA suggests problems which are not real. If the data are biased low,
then the NSA underestimates the severity of problems with rural domestic water
supplies.
The extensive inquiry into sources of error in the data has not identified
any problem which repudiates the findings, though the Calabrese report does
qualify the NSA's lead and cadmium results. Still, not all the important possible
error sources have been eliminated. Some of the results suggest water problems
that do not square with some long-standing professional expectations. These data
should therefore be viewed with realistic caution and the appropriate scientific
scepticism. They should not be the basis for permanently definitive conclusions,
neither should they be ignored. They represent the first nationally systematic
investigation of rural domestic water supplies. They identify and clarify possible
concerns for rural domestic supplies and can be a valuable guide for subsequent
inquiries. . •
-------
V - 26
Table V-A
Water Chemistry Laboratory Quality Assurance Results.
Original and Duplicate Readings
Constituent
and
Laboratory
Range of the
Original Readings
Lowest Highest
Value Value
Distribution Characteristics
of the Original Readings
Minus the Duplicate Readings
Highest
Absolute Standard
Value Mean Deviation
Turbidity (NTU)
Cincinnati 0.1 22.0 0.4 -0.03 0.08
Las Vegas UA* UA UA UA UA
Color (std. color units)
Cincinnati 2.0 50.0 UA UA UA
Las Vegas UA UA UA UA UA
Specific
Conductance (micromhos)
Cincinnati 24.0 2131.0 4.0 -0.04 0.42
Las Vegas UA UA UA UA UA
Calcium (mg/1)
Cincinnati 0.5 58.0 21.4 -0.14 2.69
Las Vegas 0.1 582.5 94.9 -1.44 10.52
Magnesium (mg/1)
Cincinnati 0.1 59.0 2.8 -0.01 0.52
Las Vegas 0.1 138.4 2.9 0.04 0.62
Nitrate-N (mg/1)
Cincinnati 0.3 19.2 0.4 0.02 0.09
Las Vegas 0.0 18.6 0.7 0.02 0.17
Sulfates (mg/1)
Cincinnati 15.0 320.0 10.0 0.17 1.13
Las Vegas 0.6 36.4 1.2 0.08 0.40
Iron (mg/1)
Cincinnati 0.10 5.15 1.00 0.008 0.161
Las Vegas 0.00 7.35 1.00 0.109 0.128
-------
V - 27
Table V-A continued
Constituent
and
Laboratory
Range of the
Original Readings
Lowest Highest
Value Value
Distribution Characteristics
of the Original Readings
Minus the Duplicate Readings
Highest
Absolute Standard
Value Mean Deviation
Manganese (mg/1)
Cincinnati
Las Vegas
Sodium (mg/1)
Cincinnati
Las Vegas
Lead (mg/1)
Cincinnati
Las Vegas
Arsenic (mg/1)
Cincinnati
Las Vegas
Selenium (mg/1)
Cincinnati
Las Vegas
Fluoride (mg/1)
Cincinnati
Las Vegas
Cadmium (mg/1)
Cincinnati
Las Vegas
Mercury (mg/1)
Cincinnati
Las Vegas
Chromium (mg/1)
Cincinnati
Las Vegas
UA
0.00
1.0
0.9
0.005
0.002
0.005
0.001
0.005
0.005
UA
0.10
0.002
0.002
0.001
0.001
0.005
0.003
UA
1.73
254.0
1025.0
0.200
0.131
0.005
0.021
0.005
0.014
UA
0.91
0.020
0.012
0.006
0.010
0.005
0.0L6
UA
0.02
4.0
29.0
0.012
0.030
0.000
0.006
0.000
0.005
UA
1.62
0.000
0.000
0.004
0.000
0.000
0.013
UA
-0.003
0.00
0.95
-0.000
-0.001
0.000
0.000
0.000
0.000
UA
-0.18
0.000
-0.001
0.000
0.000
0.000
0.004
UA
0.005
0.63
4.91
0.002
0.005
0.000
0.006
0.000
0.004
UA
0.54
0.000
0.001
0.000
0.000
0.000
0.008
-------
V - 28
Table V-A continued
Constituent
and
Laboratory
Range of the
Original Readings
Lowest
Value
Highest
Value
Distribution Characteristics
of the Original Readings
Minus the Duplicate Readings
Highest
Absolute Standard
Value Mean Deviation
Barium (mg/1)
Cincinnati
Las Vegas
Silver (mg/1)
Cincinnati
Las V egas
0.2
0.0
0.030
0.010
0.3
0.4
0.080
0.020
0.1
0.0
0.030
0.000
-0.312
-0.045
0.006
-0.010
0.366
0.100
0.016
0.020
*UA - Unavailable.
-------
V - 29
Table V-B
Water Chemistry Laboratory Quality Assurance Results.
Recovery of Spikes
Constituent
and
Laboratory
Range of
Spikes Used
Lowest
Value
Highest
Value
Lowest
Percent
Distribution Characteristics of the
Percent of Recovered Spike
Highest
Percent
Mean
Standard
Deviation
Turbidity (NTU)
Cincinnati
Las Vegas
NA*
NA
Color (std. color units)
Cincinnati NA
Las Vegas » NA
Specific
Conductance (micromhos)
Cincinnati UA**
Las Vegas UA
Calcium (mg/1)
Cincinnati
Las Vegas
Magnesium (mg/1)
Cincinnati
Las Vegas
Nitrate-N (mg/1)
Cincinnati
Las Vegas
Sulfates (mg/1)
Cincinnati
Las Vegas
Iron (mg/1)
Cincinnati
Las Vegas
5.0
2.0
1.0
1.0
0.5
0.3
10.0
9.6
0.50
0.50
NA
NA
NA
NA
UA
UA
5.0
20.0
1.0
20.0
1.0
7.5
40.0
9.6
1.00
2.00
NA
NA
NA
NA
UA
UA
52.5
57.7
76.8
10.7
25.0
3.26
44.4
89.0
50.0
2.5
NA
NA
NA
NA
UA
UA
131.7
124.8
131.3
297.1
160.0
254.14
105.3
112.2
122.7
210.8
NA
NA
NA
NA
UA
UA
93.1
93.0
101.8
101.0
84.3
64.2
90.8
99.6
92.9
89.7
NA
NA
NA
NA
UA
UA
11.8
14.1
7.8
28.3
18.1
48.9
13.0
4.3
12.6
31.7
-------
V -30
Table V-B continued
Constituent
and
Laboratory
Range of the
Spikes Used
Lowest
Value
Highest
Value
Distribution Characteristics of the
Percent of Recovered Spike
Lowest Highest Standard
Percent Percent Mean Deviation
Manganese (mg/1)
Cincinnati
Las Vegas
Sodium (mg/1)
Cincinnati
Las Vegas
Lead (mg/1)
Cincinnati
Las Vegas
Arsenic (mg/1)
Cincinnati
Las Vegas
Selenium (mg/1)
Cincinnati
Las Vegas
Fluoride (mg/1)
Cincinnati
Las Vegas
Cadmium (mg/1)
Cincinnati
Las Vegas
Mercury (mg/1)
Cincinnati
Las Vegas
Chromium (mg/1)
Cincinnati
Las Vegas
0.50
0.50
5.0
5.0
0.010
0.020
UI***
0.200
UA
0.200
0.50
1.10
0.010
0.050
0.001
0.001
UI
0.200
1.00
1.00
10.0
10.0
0.100
0.200
UI
0.400
UA
0.400
1.00
2.00
0.010
0.050
0.001
0.005
UI
0.200
20.2
0.6
27.2
0.0
15.0
14.3
UI
84.6
UA
11.1
83.3
34.4
50.0
83.1
60.0
75.7
UI
97.0
179.2
205.3
186.1
292.4
125.0
177.1
UI
117.0
UA
73.6
116.7
83.3
214.3
101.9
112.5
103.7
UI
218.9
97.5
97.7
102.1
105.2
84.9
81.3
UI
99.2
UA
49.8
96.5
60.2
105.3
92.5
81.1
95.3
UI
128.6
13.4
45.2
18.3
42.1
21.7
33.0
UI
16.5
UA
27.0
11.6
14.1
46.6
13.3
19.9
7.1
UI
60.2
-------
V -31
Table V-B continued
Constituent
and
Laboratory
Range of the
Spikes Used
Lowest
Value
Highest
Value
Lowest
Percent
Distribution Characteristics of the
Percent of Recovered Spike
Highest
Percent
Mean
Standard
•Deviation
Barium (mg/1)
Cincinnati
Las Vegas
Silver (mg/1)
Cincinnati
Las Vegas
0.5
1.0
0.500
0.200
1.0
1.0
0.500
0.400
68.3
78.0
49.1
77.3
121 .4
156.1
128.3
97.0
90.7
109.9
83.9
88.4
17.8
41.0
24.6
9.4
*NA
**UA
***UI
- not applicable.
- Unavailable.
- Uninterpretable.
-------
V -32
Table V-C
Water Chemistry Laboratory Quality Assurance Results.
Performance Results on Laboratory Standard Solution Tests
Constituent
and
Laboratory
Range of Standard
Solutions Used
Lowest
Value
Highest
Value
Distribution Characteristics of Known Standard
Solutions Minus the Measure of the Standard
Lowest Highest Standard
Value Value Mean Deviation
Turbidity (NTU)
Cincinnati
Las Vegas
UA**
10.0
Color (std. color units)
Cincinnati
Las Vegas
NA*
NA
Specific
Conductance (micromhos)
Cincinnati UA
Las Vegas 586.0
Calcium (mg/1)
Cincinnati
Las Vegas
Magnesium (mg/1)
Cincinnati
Las Vegas
Nitrate-N (mg/1)
Cincinnati
Las Vegas
Sulfates (mg/1)
Cincinnati
Las Vegas
Iron (mg/1)
Cincinnati
Las Vegas
UA
5.0
UA
5.0
UA
1.0
UA
UA
UA
0.50
UA
10.0
NA
NA
UA
720.0
UA
100.0
UA
100.0
UA
1.0
UA
UA
UA
10.00
UA
0.0
NA
NA
UA
-12.0
UA
-7.9
UA
-4.9
UA
0.0
UA"
UA
UA
-4.80
UA
0.5
NA
NA
UA
11.0
UA
5.6
UA
6.1
UA
0.1
UA
UA
UA
1.00
UA
0.08
NA
NA
UA
-0.04
UA
-0.61
UA
-0.27
UA
0.01
UA
UA
UA
-0.033
UA
0.10
NA
NA
UA
4.30
UA
2.02
UA
1.02
UA
0.01
UA
UA
UA
0.521
-------
V -33
Table V-C continued
Constituent
and
Laboratory
Range of Standard
Solutions Used
Distribution Characteristics of Known Standard
Solutions Minus the Measure of the Standard
Lowest
Value
Highest
Value
Lowest
Value
Highest
Value
Mean
Standard
Deviation
Manganese (mg/1)
Cincinnati
Las Vegas
Sodium (mg/1)
Cincinnati
Las Vegas
Lead (mg/1)
Cincinnati
Las Vegas
Arsenic (mg/1)
Cincinnati
Las Vegas
Selenium (mg/1)
Cincinnati
Las Vegas
Fluoride (mg/1)
Cincinnati
Las Vegas
Cadmium (mg/1)
Cincinnati
Las Vegas
Mercury (mg/1)
Cincinnati
Las Vegas
Chromium (mg/1)
Cincinnati
Las Vegas
UA
0.20
UA
10.0
UA
UA
UA
0.100
UA
UA
UA
UA
UA
UA
UA
UA
UA
0.100
UA
5.00
UA
100.0
UA
UA
UA
0.500
UA
UA
UA
UA
UA
UA
UA
UA
UA
0.500
UA
-0.08
UA
-6.0
UA
UA
UA
-0.023
UA
UA
UA
UA
UA
UA
UA
UA
UA
-0.038
UA
2.02
UA
3.8
UA
UA
UA
0.035
UA
UA
UA
UA
UA
UA
UA
UA
UA
0.000
UA
0.036
UA
-0.51
UA
UA
UA
0.000
UA
UA
UA
UA
UA
UA
UA'
UA
UA
-0.012
UA
0.250
UA
1.36
UA
UA
UA
0.015
UA
UA
UA
UA
UA
UA
UA
UA
UA
0.011
-------
V -34
Table V-C continued
Constituent
and
Laboratory
Range of Standard
Solutions Used
Distribution Characteristics of Known Standard
Solutions Minus the Measure of the Standard
Lowest
Value
Highest
Value
Lowest
Value
Highest
Value
Mean
Standard
Deviation
Bariun (mg/1)
Cincinnati UA UA
Las Vegas 1.0 5.0
Silver (mg/1)
Cincinnati UA UA
Las Vegas 0.100 0.500
UA
-0.1
UA
-0.010
UA
1.0
UA
0.010
UA UA
0.10 0.29
UA UA
0.000 0.010
*NA - Not applicable.
**UA - Unavailable.
-------
V -35
METHOD OF PRESENTATION
The NSA results which follow are grouped in four broad categories—(1)
bacterial content, (2) physical or chemical properties (turbidity, color, tempera-
ture, specific conductance, or hardness), (3) inorganic and organic constituents,
and (4) radioactivity. Background information for the results comes from a
variety of sources (see References), but primarily from the national interim
primary drinking water regulations and the 1977 report by the National Research
Council (NRC) entitled Drinking Water and Health.^ The NRC report reviews the
subject of water quality in extensive detail and critically assesses recent
regulatory approaches.
The focus in this report is on the findings for the various NSA constitu-
ents. Specific laboratory techniques used to obtain the findings are described only
in general terms, if at all. Details about the laboratory procedures, beyond those
cited in the preceding section, are found in Appendix C. In reviewing the NSA
findings, it is helpful to keep in mind that some of the constituents were studied
for each of the 2,654 NSA sample households, and the results were then projected
to rural America. Some of the constituents, however, were studied only for a
special 267-household, 10-percent subsample of the surveyed households. The
groupings are set out in Table V-l. In reporting the laboratory results, there is a
coincidental grouping in which constituents studied at the full sample of house-
holds are reported on in sequence, from the total coliform findings through the
manganese findings. Then, in reporting the laboratory results for the constituents
at the subsample households, there is a subsequent grouping which includes the
9
arsenic findings through the radionuclide findings. One major consequence of the
different sample sizes is the statistical confidence with which the results can be
projected to all of rural America (see Appendix B).
-------
V - 36
Within the full-sample and subsample groupings, the presentation of the
NSA findings follows the same order of reporting generally used in previous
chapters. That is, findings are presented first for households in the nation as a
whole; second, for households according to their location in different geographic
regions; third, for households according to their location inside or outside of
SMSAs (regardless of geographic region); fourth, for households according to their
location in large rural communities, small rural communities, or other rural areas;
and, fifth, for households according to the type of water system they use
(community, intermediate, or individual). After the presentation of findings in
these different groupings, the health-related implications of the findings are
discussed, when warranted.
To supplement the narrative discussion of each constituent, graphs are
presented which plot the distribution of most constituents in household supplies
across the nation and within each of the four geographic regions. In addition,
graphs on semilog scales are used to present the national distribution of values for
certain constituents which were present in household supplies in a very wide range
of concentrations. Graphs are not presented for the other subnational compari-
sons, but differences in the various groupings are set forth fully in the text.
As to the statistical analyses, there are several summary statistics which
are employed routinely, most notably medians and percentages. Means generally
are not used for two reasons. The first reason is that, for many of the
constituents, the mean would be biased toward a slightly too-large reading. This
is a result of the standard laboratory practice of reporting a lower limit of
detection. That is, when concentrations become very small in relation to the
measurement capacity of the laboratory instruments, the measurements begin to
be unreliable. Rather than induce an artificial degree of precision into the data, a
lower detection limit is established, and the limit is assigned to households when
the measured quantity is lower than that detection limit. Hence, in calculating
-------
V - 37
the mean, each household with very small concentrations gets a value which,
though small (the lower detection limit), is usually larger than the true concentra-
tion, thereby slightly inflating the mean.
The second reason is that the mean, being the average of values for all
households, would be strongly influenced by a few very large values, which
sometimes occurred in the NSA. The median, being the midpoint between the
highest and lowest values in a distribution, is not affected by extremely high or
low values; nor is it affected by the lower limit of detection. The median is often
more useful than the mean for summarizing data trends, and it is used almost
exclusively in discussion of the NSA laboratory results. Since a formula is used
for calculating the median, it is sometimes expressed as a fraction even though
the counts may be in whole numbers.
Percentages are used to indicate relative proportions of households in a
certain range of constituent values. They are especially helpful in establishing
perspective for the overall situation. They do not directly indicate the numbers
of households involved, however, and it should be kept in mind that relatively
small percentages may actually refer to a large number of households—for
example, 5.0 percent of households may be equivalent to as many as 1.1 million
households.
Finally, a special note is required to aid interpretation of findings in one
grouping—size of place. As explained in Chapter II, one grouping of households
for analysis in the NSA was according to the size of place in which they were
located: places with 1,000 to 2,500 people (sometimes referred to in this report
as large rural places or large rural communities), places with fewer than 1,000
people (sometimes referred to as small rural places or small rural communities),
and other rural areas—roughly, those areas which range from open country to
informally recognized communities.
-------
V - 38
On the basis of the NSA findings, about 82 percent of rural households are
located in other rural areas, compared to about 11 percent in large rural
communities and 7 percent in small rural communities. Households within each of
these three classifications are not distributed evenly throughout the US, however.
In Chapter III, it was pointed out that about 70 percent of all rural households are
located in two regions—the South (42.3 percent) and the North Central (28.3
percent). Therefore, rural households within the three size-of-place classifica-
tions are located predominantly in the South and North Central.
For a number of NSA constituents—calcium, magnesium, nitrates, arse-
nic, and others—size-of-place findings are influenced by the regional distribution
of households. The presence of these constituents tends to be related at least in
part to geological or environmental factors which vary from region to region.
Thus, concentrations of the constituents are likely to be greatest in households
located in size-of-place classifications which happen to be located predominantly
in regions where the constituents are naturally present in greater-than-usual
quantities. This situation is particularly apparent in findings for households in
small rural communities. About one-half of all the households located in small
rural communities are in the North Central. Coincidentally, the largest concen-
trations of a number of constituents also are in the North Central. Thus, as
discussed in the reports which follow, households in small rural communities
sometimes show larger-than-expected levels of constituents, not primarily be-
cause of some unique aspect related to small communities, but because of the
geographic location of the households. On the other hand, some findings for
households in small rural communities appear to indicate that the category has
certain aspects which require special consideration.
-------
V - 39
BACTERIAL CONTENT
Four standard tests of bacteriological conditions were performed on ail
NSA household water specimens. The procedures measured content of total
coliform bacteria, fecal coliform bacteria, fecal streptococcal bacteria, and total
bacteria as determined by the standard plate count method. These tests had one
feature in common: they did not measure content of disease-causing organisms
directly; instead, they produced results which could determine, or at least
indicate, whether there was disease potential.
This indirect approach to monitoring pathogens in water is the best
available, despite its inability to measure disease threat directly. In the proceed-
ings from the water quality workshop sponsored by the National Science Founda-
tion, it is pointed out that: "For some pathogens, even the qualitative techniques
for demonstration of their presence in water are quite unreliable and, even when
the pathogens are found, there is no way to tell if they are viable and virulent
enough to establish an infection."^
In view of this, an indirect method for specimen analysis is necessary:
"The way around the problems of enumerating pathogens and establishing their
virulence," the editor continues, "is to measure the degree of contamination of
the water with fecal material. The amount of fecal material in the water is
measured by enumerating nonpathogenic fecal bacteria for which reasonably
accurate microbiological techniques are available. The basis for this is the
assumption that if the water is contaminated with the feces of a large number of
people, the person-to-person variation in excretion of pathogens and indicators
will be averaged out, and there will be a more-or-less stable ratio of indicators to
pathogens. The (numerical) value of the ratio of indicators to pathogens is a
function of the number of excreters of the pathogen in the population which in
turn should be related to the incidence of the disease."
-------
V - 40
Total conform bacteria
Assay of coliform bacteria has been a standard test of drinking water
quality for 70 years. The standard admittedly is imperfect, but it has been a
reliable tool in drinking water sanitation. The coliform bacteria generally do not
themselves cause disease. Rather, they are "indicator organisms" which are
present in human and animal feces as well as in other organic materials and
therefore are often indicative of fecal contamination. The fecal wastes, in turn,
may contain disease organisms which can cause typhoid fever, salmonellosis,
gastroenteritis, and other intestinal diseases. Technically, the coliform group
includes all of the aerobic and facultative anaerobic, Gram-negative, non-spore-
forming, rod-shaped bacteria which ferment lactose with gas formation within US
hours at 35° C. (For the membrane filter technique described below, this
definition is altered to the bacteria, as described here, which produce a dark
g
colony with a metallic sheen within 2k hours of incubation.)
To a much lesser extent, the coliform bacteria signal possible hazard from
pathogenic protozoa or intestinal worms (helminths) excreted in human feces.
Also, to a certain extent, the indicator bacteria may warn of possible viral
disease, but in this regard, the test is especially weak. This is particularly so
because many viruses can substantially outlast the coliform bacteria indicators.
As long as the indicator bacteria survive with the viruses, the indicator warning
tag is intact. If the viruses survive even though the bacteria die, however, the
warning is lost.
Special care was .used in handling the water specimens intended for
bacteriological analysis. As soon as the water was drawn, the container was
packed in ice and shipped by air to the laboratory. Any specimen which could not
be analyzed within 30 hours was discarded, and a new specimen was taken from
the same household. Despite these procedures, the count of viable organisms in
each specimen was expected to be lower than when the water was first drawn.
-------
V - 41
The bacteriological results, therefore, were a conservative estimate of the
number of indicator organisms which were actually present in the tap water.
Conceptually, the total coliform test is best regarded not as an indicator of
water quality at all, but rather as an indicator of possible fecal contamination or of
the effectiveness of disinfection, since potable water should be free of coliform
organisms. According to the NRC: "It has been reported repeatedly in the
literature that the presence of any type of coliform organism in drinking water is
undesirable. The regulations essentially demand that coliform-free water be
distributed to consumers. Wolf has ably summarized: 'The drinking water standard
presently in use (approximately one coliform per 100 ml.) is, in a sense, a standard
of expedience. It does not entirely exclude the possibility of acquiring an intestinal
infection. It is attainable by the economic development of available water
supplies, their disinfection, and, if need be, treatment in purification works by
9
economically feasible methods. It is not a standard of perfection'."
In assessing the NSA findings in terms of overall implications for health in
rural America, it is important to keep in mind that the total coliform count is
primarily useful as an indication of the status of sanitation in water supplies. That
is, the presence of coliform organisms indicates pollution in the supply which
should be corrected by measures such as treatment, protection of the supply
source, or even change in the supply source. In addition to providing this warning,
the total coliform count provides an indication of possible disease potential in the
water supply.
Many attempts have been made to establish some direct 'relationship
between a specific number of coliform organisms and the presence of disease
organisms, but a consistent relationship has not been proven. It is impossible to
predict how many, or what kind, of disease organisms can be expected at a
particular coliform concentration. For this and a number of other reasons, there is
little scientific evidence pointing to a reliable "threshold" at which the level of
-------
V - k2
coliform bacteria is associated statistically with increased incidence of disease. It
is possible, however—on theoretical grounds only—to state that water supplies
with more than one coliform bacteria per 100 milliliters "can be responsible for
water borne disease, both gastroenteritis and typhoid fever." ^
In addition to this theoretical conclusion, one 1953 study by Albert H.
Stevenson indicates increased incidence of disease in water used for swimming
when total coliform organisms in the water exceed at least 2,300 per 100
milliliters.^ Although the Stevenson study is being reevaluated by the National
Environmental Research Center, the study findings "added much weight to the
rationale of establishing a coliform standard for drinking-water sources," according
12
to the NRC. A level of more than 2,300 coliform organisms per 100 milliliters
thus is taken in the NSA as a possible indication of increased hazards to human
health.
Most attempts to associate total coliform counts with some range of
pathogen counts have been made with reference to water sources contaminated
with feces from many people. Rural water supplies, which are often individual
wells located on users' household premises, draw water from sources which
generally are not subject to fecal contamination by large numbers of people.
Coliform bacteria in such supplies are likely to originate from a few individuals or
animals, or from decaying organic material. In this sense, the total coliform test is
even less interpretable for rural households than for community water systems.
Nevertheless, whenever viable coliforms occur in a water supply, they indicate that
the supply is not completely protected. Their presence, regardless of origin,
signals a possible health hazard to anyone who consumes the water. However,
while such supplies constitute a continuing health risk, it is very possible for users
to show no adverse health effects, even after many years of exposure.
The levels of total coliform bacteria in the NSA were determined by the
membrane-filter technique. According to the technique, coliform organisms were
-------
V -M
those which produced a dark colony (generally purplish-green) with a metallic sheen
within 24 hours of incubation on an appropriate culture medium. The colonies were
counted under magnification, and the number counted was reported as the total
number of coliform organisms (on the assumption that most of the colonies had
each been produced by just one of the organisms originally present). In a very
small number of samples, the membrane-filter technique was not usable for
technical reasons, and the somewhat slower most-probable-number technique was
used instead. This alternative technique was employed so infrequently that the
following NSA results can be assumed to be based on the membrane-filter
procedure.
In monitoring coliform levels, suppliers are given some leeway. The
number of specimens to be taken depends on the size of the population served;
when there are more than four coliform bacteria in a single 100-milliliter
specimen, intensified monitoring is required until the average concentration in
specimens is less than one coliform bacterium per 100 milliliters of water. Here,
the interim primary MCL is assumed to be one coliform bacterium per 100
milliliters, and that value is taken as the NSA reference value.
— Total coliform bacteria levels in rural supplies
In rural America, 28.9 percent or 6A million households had major supplies
with two or more coliform organisms per 100 milliliters of water (Figures V-2,
V-2a). Thus, more than one out of every four rural households exceeded the NSA
reference value and were served by water supplies which needed attention such as
further water quality studies, disinfection, or some other protective measure.
When larger concentrations were considered, 3.7 percent of all rural
households—a total of about 813,000—had total coliform levels exceeding 1,000
organisms per 100 milliliters. The levels exceeded 2,300 organisms at 2A percent
of all rural households.
-------
V - w
Figure V-2
Total Coliform in US Rural Household Supplies
REFERENCE VALUE
71.1% 28.9%
Lowest value:
Highest value:
Median:
Interval width:
Confluent growth
0.3
2.0
Number of households: 21,974,000
c «~-
I.
T
40
[I n, p n p D ~ ~ 0 0_
60 80 100 120 140
I >
160
20
Bocteria Colonies/100 milliliters
Figure V-2a. Cumulative Distribution of Total Coliform
REFERENCE VALUE
71.1%
100
95
90
» 85
80
70
65
60
o OO •—L
O OOO O
o OOO o
(flCDO O
O OO
o oof
o 00<7>
O OO®
O OO®
confluent growth
Bocterio Colonies/100 milliliters
NOTE: Cumulative distribution is plotted on semilog paper. Base: 21,97^,000 households.
-------
V - 45
Although total coliform bacterial colonies were estimated in whole num-
bers (as were colonies of fecal streptococcal and fecal coliform bacteria, to be
discussed shortly), the median and mean values were derived by formula, and they
were not whole numbers. The median was the most meaningful summary number
since the mean (10,500 total coliform bacterial colonies) was strongly influenced by
very large values which occurred in a relatively small percentage of rural
households. The median, then, was 0.3 (i.e., less than one) total coliform bacterial
colonies in major household water supplies in rural America.
In terms of regional variations (Figures V-2b through V-2e), the NSA
reference value was exceeded most commonly in the South and West. There, about
32 percent and 31 percent, respectively, of the rural households were above the
reference value. The North Central had the lowest proportion of households above
the reference value—24.4 percent. The results indicated, however, that the
contamination was not localized in any one particular region, but rather was
pervasive throughout rural America.
In comparing the results for SMSA and nonSMSA households, it was found
that 18.3 percent of the SMSA households exceeded the NSA reference value,
compared to 33.9 percent of the nonSMSA households. Similarly, supplies in other
rural areas were contaminated more commonly than those in places of larger
population. (About 31 percent of other-rural-area household supplies were above
the reference value, compared to 20 percent of supplies in small communities and
18 percent of supplies in large communities.)
A striking contrast was that households served by intermediate or indi-
vidual systems exceeded the reference value more than 40 percent of the
time—nearly three times as often as households using community systems. Of the
6.4 million rural households above the NSA reference value, 4.7 million were served
by these small water systems. At the same time, 1.7 million households using
community systems also exceeded the reference value.
-------
V - 46
Regional Variation in Total Coliform in US Rural Household Supplies
Figure V-2b. Northeast
REFERENCE VALUE
71.7% 28.3%
Lowest value:
Highest value:
Median:
Interval width:
Confluent growth
0.2
2.0
Number of households: 3,693,000
1—
60
I >
160
20
40
r
80
100
120
140
Bocteria Colonies/I00 milliliters
Figure V-2c. North Central
REFERENCE VALUE
75.6% 24 4%
Lowest value: 0
Highest value: Confluent growth
Median: 0.2
Interval width: 2.0
Number of households: 6,213,000
1
100
r>-
160
20
40
I
60
80
120
140
Bacteria Colonies/100 milliliters
-------
V - k7
Regional Variation in Total Coliform (continued)
Figure V-2d. South
REFERENCE VALUE
68--
67-,
5-
4-
2-
I-
68.3% 31.7%
Lowest value: 0
Highest value: Confluent growth
Median: 0.3
Interval width: 2.0
Number of households: 9,291,000
1 I 1 'I ' 1 T
I 20 40
To do 1 ib o
Bacteria Colonies/100 milliliters
iio"
140
160
Figure V-2e. West
REFERENCE VALUE
70 •
694-
68 -
69.4% 30 6%
Lowest value:
Highest value:
Median:
Interval width:
0
Confluent growth
OA
2.0
Number of households:
2,777,000
20 40 60 80 100 120
Bacteria Colonies/100 milliliters
40
160
-------
V - 48
In terms of possible health implications for rural America, the values
discussed in the introductory material above provide some perspective in assessing
the NSA data. On theoretical grounds, then, there was the chance for waterborne,
bacterially induced disease in some of the 6.4 million rural households with more
than one coliform organism per 100 milliliters of water. The major implication for
many of those households, however, was not that an immediate health threat
existed (a hazard which could not be directly proven on the basis of the total
coliform tests, as explained above), but rather that the supplies required further
assessment to determine the need for remedial action. Considerable evidence
points to increasing health risk as the concentration of total coliform bacteria
rises, however, and in this light a number of rural American households had
potentially serious conditions in their water supplies. For example, the 527,000
household supplies in the nation that had levels exceeding 2,300 coliform organisms
were at levels even greater than those which have been associated with increased
incidence of disease in some public bathing waters.
The potential problems from bacterial contamination were pervasive in
rural America. They were not limited to any particular region of the country. On
the other hand, the problems were most prominent in households in areas classified
as other rural (mostly in open country) which were served by individual or
intermediate systems. This pattern was consistent with other studies pointing to
contamination being associated most often with individual wells in areas outside of
• • 13
major communities.
Fecal coliform bacteria
Because of the generalized nature of the total coliform bacteria test, other
tests have been proposed. In particular, a test for fecal coliforms has been favored
to assess the recentness of fecal contamination in water supplies. The test has
gained acceptance primarily for evaluation of recreational and shellfish waters
-------
V -w
Ik
rather than for drinking water, however. The usual test for fecal coliforms is,
in essence, an extension of the total coliform test, but with a different medium and
with the incubation temperature raised from 35° C to kk.5° C. The advantage is
that at the higher temperature, nonfecal coliform bacteria (such as those originat-
ing from soil and plants) generally do not grow well. These bacteria are included in
the total coliform count (above), but they have less significance for human health.
Although the test for fecal coliforms provides an indication of recent contamina-
tion from human or animal feces, fecal coliforms are less numerous than total
coliforms, and thus provide a less sensitive indication of pollution than does the
total coliform content.
Furthermore, as with other indicator tests, the one for fecal coliforms does
not pinpoint specific pathogens. Rather, presence of fecal coliforms indicates the
possibility of pathogens and accompanying health risks. However, fecal coliforms
can die at a faster rate than some pathogens, a situation which diminishes their
usefulness as indicator organisms.
As with total coliform organisms, various attempts have been made to
establish a direct relationship between a specific number of fecal coliforms and the
presence of disease organisms. For example, Salmonella bacteria are potential
pathogens which can be detected fairly easily in water, and for convenience they
have been studied in association with fecal coliforms. This has been done to give
investigators a feeling for the possibility of a direct relationship between concen-
trations of fecal coliforms and the occurrence of disease organisms. The studies
have not shown a precise, reliable relationship. However, Edwin E. Geldreich has
t
reported that "field data from numerous fresh water and estuarine pollution studies
indicate a sharp increase in the frequency of Salmonella detection when fecal
coliform densities are above 200 organisms per 100 milliliters."^
This generalized finding does not mean that the incidence of waterborne
disease will necessarily increase exactly at this threshold concentration of fecal
-------
V -50
coliforms. There is marked variation in the occurrence of Salmonella bacteria and
other disease organisms regardless of the concentration of fecal coliforms, and
there is great variation in susceptibility of persons exposed to the organisms.
Nevertheless, Geldreich concludes that the evidence supports a limit of no more
than 200 fecal coliforms per 100 milliliters for water to be used for recreation such
as swimming.
There is no federal standard for concentration of fecal coliform bacteria in
drinking water, but the presence of even one such organism is taken as indication
of fecal contamination which requires attention. The EPA makes this statement:
"Although the total coliform group is the prime measurement of potable water
quality, the use of a fecal coliform measurement in untreated potable supplies will
yield valuable supplemental information. Any untreated potable supply that
contains one or more fecal coliforms per 100 milliliters should receive immediate
disinfection."^ In the NSA, the reference value was zero, or the complete
absence of fecal coliform bacteria in a 100-milliliter sample.
— Fecal coliform bacteria levels in rural supplies
About 12.2 percent of all rural households (approximately 2.7 million) had
supplies with one or more fecal coliform bacteria (Figures V-3, V-3a) and thus
exceeded the NSA reference value. The level exceeded 200 organisms per 100
milliliters at about 350,000 households (1.6 percent of all households), and it ranged
up to concentrations which were too dense to count (confluent growth) at 109,000
households (0.5 percent of all households). The median for the rural US was 0.1
fecal coliform colonies.
In terms of regional variations, the percentage of households showing fecal
coliforms was the same (14.0 percent) in three regions: the Northeast, South, and
West (Figures V-3b through V-3e). In the North Centrsil, the proportion was six
percentage points less. The evidence thus was that, as with total coliform
-------
recai i^oiiiorm in uo Kurai nousenoia supplies
REFERENCE VALUE
12.2%
Lowest value:
Highest value:
Median:
Interval width:
Confluent growth
0.1
2.0
Number of households: 21,97^,000
Bacteria Colonies/100 milliliters
Figure V-3a. Cumulative Distribution of Fecal Coliform
REFERENCE VALUE
78.8% 12.2 •/,
5 90
o
o
CM
o oc>
O OOO
IfiCOO
oo„
°o2
oo2
cDCOO
o
o
o
o
o
OJ
o oo
O OOO
O OOO
O OOO
<0®o
o
Bacteria Colonies/100 milliliters
NOTE: Cumulative distribution is plotted on semilog paper. Base: 21,97^,000 households.
confluent
o oo
o ooo
o OOff>
(OQOO)
growth
Reproduced from
best available copy.
-------
V - 52
Regional Variation in Fecal Coliform in US Rural Household Supplies
Figure V-3b. Northeast
REFERENCE VALUE
86.0% 14.0%
6-
5-
o 4 —
£ 3-
2-
Lowest value:
Highest value:
Median:
Interval width:
Confluent growth
0.1
2.0
Number of households:
3,693,000
20
30 40
Bacteria Colonies/100 milliliters
—
50
60
—i—
70
—r*
80
Figure V-3c. North Central
REFERENCE VALUE
92.0% 8.0%
5-
3-
2-
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
0
1200
0.0
2.0
6,213,000
—J—
50
ir
20
I
30
I
40
I
60
70
Bacteria Colonies/100 milliliters
-------
V - 53
Regional Variation in Fecal Coiiform (continued)
Figure V-3d. South
REFERENCE VALUE
86.0% 14.0%
6-
- 4-
o
c
o 3-
£
2-
I •
0-
"T-
10
—r
20
Lowest value:
Highest value:
Median:
Interval width:
Confluent growth
0.1
2.0
Number of households: 9,291,000
—r~
50
—T"
60
30
I
40
70
80
Bacteria Colonies/100 milliliters
Figure V-3e. West
REFERENCE VALUE
86.0% 14 0%
Lowest value:
Highest value:
Median:
Interval width:
Confluent growth
0.1
2.0
Number of households: 2,777,000
10
I
20
30
40
-1-
50
I
60
-r~
70
—T*-
80
Bacteria Colonies/10 0 milliliters
-------
V - 54
contamination, contamination from fecal coliform bacteria occurred across the US
and was not localized in one region. Even in the North Central, although the
percentage of households exceeding the reference value was smaller than else-
where (8.0 percent), the number of households involved was about 400,000—more
than in the West and Northeast.
There were one or more fecal coliforms per 100 milliliters in twice as large
a proportion of nonSMSA households as SMS A households—in 15.0 percent of the
former and 7.0 percent of the latter. Also, this contamination occurred about
three times as frequently in households in other rural areas as in households in
either of the other two size-of-place classifications (14.0 percent in other rural
areas, compared to less than 5 percent in large and small rural communities).
Fecal coliform bacteria were found five times more frequently in rural
households served by individual or intermediate systems as in households served by
community systems. Specifically, 4.0 percent of households served by community
systems had fecal coliforms, as opposed to 20.0 percent of households served by
individual or intermediate systems.
As to the potential health effects of the fecal coliform levels in rural
households, the presence of even one fecal coliform bacterium in a 100-milliliter
specimen is viewed with concern by public health authorities. The supplies in 2.7
million households with at least one organism thus presented potentially significant
problems. All of the supplies were candidates for prompt study and possible
treatment. Furthermore, in about 350,000 of these households, the fecal coliform
level was greater than 200 organisms per 100 milliliters. That concentration was in
excess of the limit viewed by some authorities as excessive even for public
swimming water (as noted before), and it represented a potentially serious sanitary
problem.
Generally, variations in the degree of contamination by region and other
groupings were similar to those for total coliform bacteria. Overall, the most
-------
V - 55
serious contamination problems involving either total or fecal coliform bacteria
existed in rural households which were outside of SMSAs, located in other rural
areas, or served by intermediate or individual systems.
Fecal streptococci
Fecal streptococcal bacteria include a variety of strains which have
different origins and survival rates. The organisms have been studied as possible
indicator organisms, but they have not proven suitable for drinking water analysis
because of low recovery rates, poor agreement between various assay procedures,
and uncertainty about their health significance.^ Thus, there is no NSA reference
value for levels of fecal streptococcal bacteria. Values for concentration of fecal
coliforms and fecal streptococci together, however, have been used in sanitary
evaluations (see below).
— Fecal streptococci levels in rural supplies
At least one organism was found in water supplies of 19.0 percent of all
rural households (about 4.2 million). There were ten or more fecal streptococci per
100 milliliters at about 10 percent of all rural households; 100 or more fecal
streptococci at 3.5 percent of all rural households. The median value for the level
of fecal streptococci in major household water supplies was 0.12 (Figures V-4,
V-4a).
In terms of variations in the different NSA groupings, the medians were
close to the national median of 0.12 in most classifications. The slight variations
which did exist tended to repeat the pattern which was observed for other
bacteriological indicators. That is, the values tended to be somewhat larger in
rural households which were located in regions other than the North Central
(Figures V-4b through V-4e), outside of SMSAs, in other rural areas, or which were
served by intermediate or individual systems.
-------
V -56
Figure V-4
FecaJ Streptococcus in US Rurai Household Supplies
82 _
81 -
80.
Lowest value: 0
Highest value: Confluent growth
Median: 0.1
Interval width: 2.0
Number of households: 21,97^,000
I _
0.
—I—
30
40
-I—
10
—I—
20
50
60
70
80
Bacteria Coionies/100 milliliters
Figure V-4a. Cumulative Distribution of Fecal Streptococcus
100-1
90-
<£00O
OOO
to CO <75
Bacteria Colonies/100 milliliters
confluent growth
NOTE: Cumulative distribution is plotted on semilog paper. Base: 21,97
-------
V - 57
Regional Variation in Fecal Streptococcus in US Rural Household Supplies
Figure V-^b. Northeast
82-
81-
80-
8-
7
6-
5-
4-
3
2-
I-
—r~
20
Lowest vaiue:
Highest vaiue:
Median:
Interval width:
Confluent growth
0.1
2.0
Number of households: 3,693,000
—I—
60
1 >
80
i
30
40 50
Bacterio Colonies/100 milliliters
I
70
Figure V-4c. North Central
86-
85-h
84-
6-
5-
4-
3-
2-
1 1-
10
I
30
Lowest value:
Highest value:
Median:
Interval width:
Confluent growth
0.1
2.0
Number of households: 6,213,000
—r~
50
1 >
80
20
40
I
60
70
Bacteria Colonies/100 milliliters
-------
V -58
Regional Variation in Fecal Streptococcus (continued)
Figure V-W. South
80-1
79--i
Lowest value:
Highest value:
Median:
Interval width:
Confluent growth
0.1
2.0
Number of households:
i 1 1 r-
10 20
—r~
60
9,291,000
—r~
70
-r*-
80
30 40 50
Bacteria Colonies/100 milliliters
Figure V-Ue. West
81-
80--
79-
8H
7-
6-
5-
4-
3-
2-
I-
0-
Lowest value: 0
Highest value: Confluent growth
Median:* 0.1
Interval width: 2.0
Number of households: 2,777,000
l
10
20
30 40 50
Bacteria Colonies/100 milliliters
1
60
I
70
80
-------
V - 59
Fecal coliform/fecal streptococci ratio
The ratio of fecal coliforms to fecal streptococci (FC/FS ratio) in water
specimens supplements the value obtained for total coliforms. As pointed out
above, total coliform content may warn of contamination, but the contamination
may be from a variety of fecal and plant materials. Furthermore, fecal
contamination may be from either human beings or animals. Investigators have
discovered, however, that the FC/FS ratio of the specimen helps indicate whether
human or animal wastes are implicated. In some instances, even the type of animal
can be suggested by the ratio.
As explained by EPA microbiologist Edwin E. Geldreich: "The ratio of fecal
coliform to fecal streptococcus is four-to-one or higher in human fecal material.
However, this ratio is reversed in the feces of other warm-blooded animals, so that
the ratio of fecal coliform to fecal streptococcus would be 0.6 or less. When data
are carefully developed through the use of sensitive media, this unique relationship
can be a useful bacteriological tool to characterize fecal pollution sources of
human origin through domestic sewage discharges, of farm animal origin through
18
feedlot drainage, or of wildlife or pet animal origin in storm water runoff."
To determine the FC/FS ratio in rural households, NSA investigators first
tabulated independently the concentrations of fecal coliform and fecal strepto-
coccal bacteria in water specimens studied for all surveyed households—as
reported in the preceding sections. The investigators then compared the values for
both constituents to obtain the FC/FS ratios, and the results were projected to
rural American households. The ratios themselves did not indicate the magnitude
of bacterial contamination in households since small numbers of the organisms
could produce the same ratios as very large numbers. The ratios did provide an
indication of the origin of the pollution, however.
-------
V - 60
— FC/FS ratios in rural supplies
Either fecal coliform or fecal streptococci—or both—appeared in five
million, or 22.9 percent, of rural household water supplies. Concentrations of fecal
coliforms and fecal streptococci were found together in 1.8 million rural house-
holds (8.3 percent of all rural households). The reported FC/FS ratios thus provided
insight into the source of pollution only at those 1.8 million households. Put
another way, in those households for which fecal coliform or fecal streptococci,
but not both, were identified, no interpretation regarding the bacterial source was
attempted. In those households, a meaningful ratio could not be formulated since
only one of the two elements in the ratio was present.
Among the 8.3 percent of all rural households in which the ratio could be
formulated, most (4.9 percent) had FC/FS ratios which were 0.6 or lower (Figure
V-5), suggesting contamination from animal feces. At the other end of the scale,
the FC/FS ratios were four-to-one or greater in 1.5 percent of all rural households,
suggesting contamination from human feces. In addition, the ratios at 2.0 percent
of all rural households suggested contamination of mixed animal and human origin
or plant origin (the FC/FS ratios were between 0.6 and 4.0).
As was true for the nation as a whole, the evidence indicated a preponder-
ance of contamination from animal or mixed origin as opposed to human origin in
all regions of the US (Figures V-5a through V-5d).
In both SMSA and nonSMSA households with fecally contaminated supplies,
the evidence pointed to contamination of animal or mixed origin. Specifically, the
FC/FS-ratios suggested1 contamination with animal feces in 2.6 percent of SMSA
households as opposed to 5.9 percent of nonSMSA households. The ratios suggested
human contamination in 1.0 percent of SMSA households as opposed to 1.6 percent
of nonSMSA households.
Predominantly animal contamination also was implicated in households
grouped according to size of place. The FC/FS ratios suggested animal
-------
V -61
Figure V-5
Fecal Coliform/Fecal Streptococcus Ratios in US Rural Household Supplies
77.1%
4.5%
4.9 %
0< <0.6
ANIMAL
2.0%
fc
1.5 %
:o.6<7r <4.0!———rr
!fCETERMtNAT£; 4'0iiT?
' HUMAN
No fecal coliform; Either fecal
no fecal coliform or fecal
streptococcus streptococcus
Both fecal coliform and fecal streptococcus
-------
V - 62
Regional Variation in Fecal Coliform/Fecal Streptococcus Ratios
in US Rural Household Supplies
Figure V-5a. Northeast
77.0%
13.2%
5.6%
0<# £0.6
ANIMAL
2.0%
2.2%
r
0.6<#<4.0j
INDETERMINATE!
»
4.0 <£
HUMAN
No fecal coliform; Either fecal Both fecal coliform and fecal streptococcus
no fecal coliform or fecal
streptococcus streptococcus
Figure V-5b. North Centred
82.4%
3.3 %
OC-ff-SO.6
No fecal coliform;
no fecal
streptococcus
Either fecal
coliform or fecal
streptococcus
1.0 %
0.9 %
;0.6 < 4 0j 4 0<4E-
!lND€TERMINATE! HUMAN *
Both fecal coliform and recal streptococcus
-------
V -63
Regional Variation in Fecal Coliform/Fecal Streptococcus Ratios (continued)
Figure V-5c. South
7 4.4%
15.8%
5.7 %
0
-------
V - 6 k
contamination at 1.0 percent of large-community households (about 24,000) and
human contamination at 0.8 percent of those households (about 20,000); animal
contamination at 2.5 percent of small-community households (about 38,000) and
human contamination at 0.0 percent of those households; and animal contamination
at 5.6 percent of households in other rural areas (about one million) with human
contamination at 1.7 percent of those households (about 300,000).
The fecal bacteria also were much more prevalent in households served by
individual or intermediate systems than in households served by community
systems. The FC/FS ratios suggested that the contamination was primarily of
animal or mixed origin in major household supplies from individual or intermediate
systems. For major household supplies from community water systems, only 0.8
percent appeared to have predominantly animal fecal contamination, and just 0.6
percent appeared to have human fecal contamination. On the other hand, 9.2 and
7.9 percent of household supplies from individual and intermediate systems,
respectively, were in the range indicating animal contamination. At the same
time, 2.2 and 2.7 percent of the respective supplies were in the range indicating
human contamination. The more favorable findings for community water systems
were not surprising since larger systems generally have disinfection programs and
usually exercise closer control over system maintenance.
The NSA findings generally were consistent with what one might expect in
rural America. That is, fecal contamination in rural household supplies could be
primarily from animal wastes contaminating wells. To a lesser extent, the
contamination could be from local human waste disposal systems. Across the
nation, about 23 percent of all households showed indications of fecal contamina-
tion (from fecal coliform, fecal streptococci, or both). According to the FC/FS
ratio where it could be applied, the indication was that about 5 percent of the
households were contaminated by animal fecal material, and 1.5 percent by human
sewage.
-------
V - 65
As to the health implications of the NSA findings, the primary usefulness
of the NSA data is to indicate potential contamination sources. Fecal contamin-
ation of human origin, of course, indicates a serious sanitary problem. On the
other hand, fecal contamination from warm-blooded animals also can introduce
19
organisms which cause illness in man. Thus, the NSA findings cannot be taken as
an indication of the relative seriousness of different types of fecal contamination
in rural households. The findings may, however, help focus on the aspects of water
supply systems which need attention.
Standard plate count
Another bacteriological test which is a useful supplement to the enumer-
ation of coliform bacteria is the standard plate count (SPC). Technically, the SPC
begins with inoculating a known volume of a water specimen in a culture dish
containing a nutrient agar medium. The culture is incubated for 48 hours, and the
organisms which grow as colonies on the agar plates represent a fraction of the
total population of bacteria in the water. Allowable SPC bacterial numbers vary in
different health department jurisdictions, but a limit of 500 colony-forming units
20
per milliliter of water is recommended by the National Research Council. This
limit also is suggested, although not required, in the EPA's National Interim
21
Primary Drinking Water Regulations. The limit is used as the NSA reference
value.
— Standard plate count values in rural supplies
The SPC values exceeded 500 colonies per milliliter of water in the
supplies of 19.3 percent of all rural households; about 10 percent of all households
had counts of 2,000 or more (Figures V-6, V-6a). The median SPC value in rural
households was about 42 colonies per milliliter.
-------
V -66
Figure V-6
Standard Plate Count in US Rural Household Supplies
3 9.
3
REFERENCE VALUE
I 0_
80 7%
9_
•£
O
8-
7_
Lowest value:
Highest value:
Median:
Interval width:
Confluent growth
42.3
20.0
Number of households: 21,974,000
o
6_
c
o
f
4_
3_
2_
" 800
Bacteria Colonies/milliliter
-r>-
1600
200
400 500 600
1200
1400
1000
Figure V-6a. Cumulative Distribution of Standard Plate Count
REFERENCE VALUE
00-T
80 7%
90
80'
70
60
50'
S 40
30
20
O O OO t
o o OOO
N V tOCOO
O OO L_
O OOO
o ooo
*3- 10COO
O OOUJ
O OO Oi
O Oo &
O 00
O OO
confluent growth
Bocteria Colonies/milliliter
NOTE: Cumulative distribution is plotted on semilog paper. Base: 21,974,000 households.
-------
V - 67
As to regional variations (Figures V-6b through V-6e), households exceeded
the reference value most frequently in the West (in 24.8 percent of households
there) and in the South (in 22.8 percent of households there). Households exceeded
the reference value least often in the Northeast (in 10.2 percent of households
there). In the North Central, 17.1 percent of rural households were above the
reference value.
As was generally true for the results of other bacterial measurements, SPC
values in excess of the NSA reference value occurred more often among nonSMSA
households (22.0 percent, compared to 13.8 percent of SMSA households) and more
often among households located in other rural areas (20.5 percent, compared to
15.0 percent in large rural communities and 11.7 percent in small communities).
High values also occurred more often among households served by individual
systems (26.6 percent) than among households served either by intermediate
systems or community systems (17.8 percent and 13.9 percent, respectively).
Summary of bacteriological findings
Despite its shortcomings (discussed in more detail earlier in this chapter),
the total coliform count is regarded by many professionals as the best available
general indicator of bacteriological water quality. Accordingly, 28.9 percent of all
rural households at the time of the NSA survey had supplies with total coliform
levels (two or more bacteria per 100 milliliters of water) which exceeded the NSA
reference value and which therefore probably required further assessment and
possible remedial action. The levels in some of the supplies were high enough to
raise the possibility of imminent health consequences. Rural supplies were above
the reference value more often in the South and West than in other regions, more
often outside of SMSAs than inside SMSAs, and more often in other rural areas than
in large or small rural communities. High values also were much more common
-------
V - 68
Regional Variation in Standard Plate Count in US Rural Household Supplies
Figure V-6b. Northeast
55-i
54
IO-A-,
REFERENCE VALUE
89.8% 10.2%
9-
8-
7-
6-
5-
4 -
3-
2
Lowest value: 0
Highest value: 360,000
Median: 14.5
Interval width: 20.0
Number of households: 3,693,000
1000
200
1400
—r*-
1600
200
400 500 600
800
Bacteria Colonies/milliliter
Figure V-6c. North Central
39-
38-
37-
15-
U»
"O
0 14-
s.
9
1 'J"
o
c 6"
o
©
0. 5-
4-
3-
2-
REFERENCE VALUE
82.9% 17 1%
Lowest value: 0
Highest value: Confluent growth
Mediaq: 33.0
Interval width: 2.0
Number of households: 6,213,000
-i
1600
200 400 500 600 800 1000
Bacteria Colonies/milliliter
1200
1400
-------
V -69
Regional Variation in Standard Plate Count (continued)
Figure V-6d. South
35-1
34-n
33-
10-
9 -
8-
7-
6-
5-
4
3-
2-
I-
(L
REFERENCE VALUE
77.2%
2 2.8%
Lowest value:
Highest value:
Median:
Interval width:
Confluent growth
61.6
20.0
Number of households: 9,291,000
i >
1600
200
400 500 600
800
1000
1200
1400
8acteria Colonies/milliliter
Figure V-6e. West
33-|
32
10
9-
8-
7-
6
5-
4-
3
2-
I -
REFERENCE VALUE
75.2% 248%
Lowest value:
Highest value:
Median:
Interval width:
Confluent growth
76.1
20.0
Number of households: 2,777,000
1 T
400 500 600
—r>~
1600
200
800 1000
Bacteria Colonies /milliliter
1200
1400
-------
V - 70
among supplies served by individual or intermediate systems than among supplies
provided by community systems.
On the basis of NSA findings, 12.2 percent of all rural households had
supplies which may have required attention because of the presence of at least one
fecal coliform bacterium. In 1.6 percent of rural households (350,000), the fecal
coliform level was higher than that which has been associated with increased
occurrence of at least one organism with disease potential (200 colonies), indicat-
ing the possibility of heightened threat of disease. High values were proportion-
ately most common outside of SMSAs, in other rural areas, and among households
served by individual or intermediate systems.
Fecal streptococci were found in 1.8 million rural water supplies which also
had fecal coliform bacteria. The ratio of these two organisms in the supplies
suggested that contamination from animal wastes alone (a potential in 4.9 percent
of all rural households) outweighed contamination from human wastes alone (a
potential in 1.5 percent of all rural supplies). This trend was apparent in all regions
of the US and also in households grouped by size of place and size of system.
In regard to general levels of bacteria as determined by the SPC, values
exceeding 500 per milliliter of water were found most often in the West and South.
The high values were more frequent among nonSMSA households, among households
located in other rural areas, and among households served by individual systems.
PHYSICAL AND CHEMICAL CHARACTERISTICS
Three physical characteristics—turbidity, color, and temperature—were
determined in all NSA water specimens, as were two general chemical character-
istics—hardness and conductance. Measurements of turbidity, color, and specific
conductance were made in EPA laboratories. The concentrations of calcium and
magnesium in all NSA water specimens were used to measure hardness.
-------
V - 71
Temperature of the water was recorded by NSA interviewers before the specimens
were collected from the households. All five characteristics represented important
overall aspects of water quality.
Turbidity
Turbidity refers to the optical effect in water caused by suspended matter
such as clay, silt, and organic particles, as well as by plankton and other
microscopic organisms. These materials, in turn, may have direct health effects,
or indirect effects resulting from reactions with other constituents (like chlorine,
which can combine with some organic materials in the formation of trihalo-
22
methanes). One of the important effects of the turbidity-chlorine interaction is
the dissipation of the disinfecting power of chlorine through reactions with the
suspended matter other than bacteria and viruses. The turbidity particles also can
shield bacteria and viruses from the chlorine. An effective means of removing the
risk associated with chlorine-resistant cysts of pathogenic protozoa and helminth
23
eggs is sedimentation or filtration, both of which remove turbidity to a great
degree; therefore, turbidity suggests a potential for disease risk which treatment
might alleviate. (Of course, clear water does not necessarily mean there is no
health risk.)
Turbidity measurements in the laboratory are based on light scattering and
absorbing properties of the suspended substances in the water. The amount of light
scattered and detected depends on the number, size, shape, and refractive index of
this particles, the wavelength spectrum of the incident light, and the geometry and
detection characteristics of the analytic equipment. Turbidity results do not
identify the substances in the water: more selective techniques must be used to
test for specific substances. However, high levels of turbidity do provide an
indication that adverse health effects are possible. Furthermore, the levels may
make the water unattractive enough to prompt people to use another supply.
-------
V - 72
Unacceptably high turbidity can be an economic liability in its effects as well as in
the cost of its removal.
The NRC does not recommend a specific standard for unacceptable
turbidity. Instead, the council cautions that health department standards and
equipment vary, and that the test requires standardization. Furthermore, the NRC
advises, the nature of the test is such that low turbidity measurements do not
guarantee that water is potable. Federal regulations, on the other hand, do set
interim Maximum Contaminant Levels (MCLs) for turbidity for both community
and noncommunity water supply systems using surface water. The MCL ranges
from one to five turbidity units, depending on a number of factors.
The federal interim regulations, however, are difficult to apply as a direct
gauge of NSA findings. For example, the regulations require testing in the
distribution system between the water treatment plant and the main distribution
pipes, but the NSA specimens were taken at the user's tap. The regulations
strictly apply only to suppliers with surface water sources, but NSA water supplies
were from a number of sources, some of which were not identified (as was noted in
Chapter IV). Furthermore, turbidity measurements are best made on the day on
2*i
which a specimen is obtained. This procedure was not followed in the NSA, since
all specimens were shipped to central laboratories for later analysis.
Water supplies that are to be treated with chlorine ideally should have very
low turbidity levels—less than one nephelometric turbidity unit—in order to
minimize interference with the disinfection. Turbidity becomes perceptible at
about the level of five turbidity units. Turbidity constitutes a general indication of
potential problems. In the NSA, however, more selective bacteriological, physical,
and chemical measures were available to specify the substances present. No NSA
reference value was chosen for turbidity, since NSA findings on particular
constituents specify the problems which turbidity indicates in a general way.
-------
V - 73
N5A turbidity data will be divided and presented in two ways. First,
conditions will be described according to the prevalence of rural water supplies
with one turbidity unit. The object is to indicate the proportion of supplies that
might require attention if chlorination were adopted. Second, the level of five
turbidity units will be used to indicate the proportion of supplies with turbidity that
is aesthetically perceptible.
Turbidity was measured in the NSA by use of an instrument called a
nephelometer, and the measured findings were expressed in nephelometric turbidity
units (NTUs).
— Turbidity levels in rural water supplies
Turbidity levels were one NTU or less in 83.5 percent of all rural
households (Figures V-7, V-7a). For 16.5 percent—a total of 3.6 million house-
holds—turbidity exceeded one NTU. About one million of these households had
turbidity of more than five NTUs. The latter households—*f.8 percent of all rural
households—had turbidity levels which were aesthetically perceptible. In all, the
measured levels ranged from less than 0.05 NTU to 132 NTUs; the rural US median
was 0.3 NTU.
Levels exceeded one and five NTUs most often in rural households in the
North Central, where 23.8 percent were above one NTU and 6.8 percent were over
five NTUs. The West had the lowest percentages: 8.5 percent exceeded one NTU
and 1.7 percent exceeded five NTUs. Values larger than one NTU were discovered
in 10.1 percent of households in the Northeast and 16.6 percent of households in the
South (Figures V-7b through V-7e). There were 2.9 and 5.2 percent of households in
the Northeast and South, respectively, above five NTUs.
As to results of other NSA comparisons, turbidity values were over one
NTU in 18.2 percent of nonSMSA households and over five NTUs in 5.2 percent,
compared to 12.9 and 3.9 percent of SMS A households. Supplies with high turbidity
-------
V - 7k
Figure V-7
Turbidity in US Rural Household Supplies
5_"
Lowest value:
Highest value:
Median:
Interval width:
< 0.05
132.0
0.3
0.02
Number of households: 21,97^,000
"oTe
NTU
"Ti"
0.2
0.4
0.6
1.0
1.4
1.6
Figure V-7a. Cumulative Distribution of Turbidity
NTU
NOTE: Cumulative distribution is plotted on semilog paper. Base: 21,97^,000 households.
-------
V - 75
Regional Variation in Turbidity in US Rural Household Supplies
Figure V-7b. Northeast
Lowest value: < 0.05
Highest value: 67.0
Median: 0.2
Interval width: 0.02
Number of households: 3,693,000
—I r
0.2
—I—
0.4
—I—
0.6
!
0.8
1—1 '1 I'
1.0
—I—
1.2
T
1.4
T
1.6
NTU
Figure V-7c. North Central
20-1
15-
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
0.06
132.0
O.t
0.02
6,213,000
10-
0.2
oU
0.6
—I—
0.8
NTU
—
1.0
I—
1.2
—1—
1.4
1.6
-------
V - 76
Regional Variation in Turbidity (continued)
Figure V-7d. South
5-
Lowest value: 0.06
Highest value: 127.0
Median: 0.3
Interval width: 0.02
Number of households: 9,291,000
"I 1 1 1 1 1 T
0.2 0.4 0.6
n 1 r
i.o
_1—
1.2
-i 1 r
1.4
0.8
NTU
1.6
Figure V-7e. West
0.2
0.4
0.6
—I—
0.8
NTU
Lowest value:
Highest value:
Median:
Interval width:
0.08
32.5
OA
0.02
Number of households: 2,777,000
i
1.0
i i>
1.6
-------
V - 77
levels were found proportionately more often among households located in small
rural communities (21.8 percent above one NTU and 5.0 percent above five NTUs).
Comparable proportions were 10.8 and 1.7 percent among households in large rural
communities, and 16.8 and 5.2 percent among households in other rural areas.
The difference in the size-of-place comparison probably was influenced by
the distribution of rural households (discussed earlier, in the section entitled
"Constituents Studied in NSA"). That is, rural communities with fewer than 1,000
people were concentrated in the North Central and the South, the two regions
having the largest proportions of high-value households. If adjustment were made
for this situation, it is likely that the proportion of households over one NTU in
small rural communities would be similar to the proportion in other rural areas.
The proportions over one NTU in these two classifications would be greater than
the proportion in large rural communities, however.
In the size-of-system comparison, households served by individual and
intermediate systems were more than twice as likely to have high values as
households served by community systems. Specifically, there were 24.0 percent of
households with individual systems, 24.7 percent of households on intermediate
systems, and 8.9 percent of households on community systems above one NTU.
Above five NTUs, the proportions were 8.7 percent for individual, 6.7 percent for
intermediate, and 1.3 percent for community systems. The proportion of house-
holds served by individual systems with readings above five NTUs (8.7 percent) was
almost as large as the percentage over one NTU (8.9 percent) among households
served by community systems.
The relative differences in the NSA comparisons generally were antici-
pated. Correction of excessive turbidity may require treatment processes such as
sedimentation, coagulation, and filtration which are used most extensively by
community systems. Water provided by these systems would be expected to have
fewer extreme levels of turbidity, and it is these larger systems which more often
-------
V - 78
serve households in SMS As and in larger communities. Nonetheless, because of the
likelihood of chlorination among community systems and because of the negative
interaction between chlorine and turbidity, the relatively low levels of turbidity
found in households served by community systems are potentially significant. For
the one million community-system households with turbidity levels over one NTU,
chlorination would give rise to a potential hazard of trihalomethanes. In addition,
the substances causing the turbid conditions may themselves constitute health
hazards, as they would in supplies from systems of any size. All these risks would
be even greater among those 1.1 million rural households with turbidity levels over
five NTUs.
Despite the relative differences among proportions of households with high
turbidity, the median level of turbidity in rural American households was similar
throughout all of the groupings except the regional one. Specifically, the national
median of 0.3 NTU held fairly steady in rural households whether they were in
SMSAs or not, whether they were in rural communities or in other rural areas, or
whether they were served by individual, intermediate, or community systems. The
median was lower in the Northeast, however (0.2 NTU), than in other regions; in
the South, it was at about the national level, while it was somewhat higher in the
West and North Central (OA NTU in both regions).
Color
Water is colored primarily by natural organic matter, but also by certain
industrial wastes and some metallic complexes. The standard laboratory procedure
for measuring color is based on the use of solutions that contain known concentra-
tions of a color-producing chemical. Solutions containing a range of concentrations
of the chemical are assigned corresponding, arbitrary color values. Through the
use of a color comparison device, the specimen to be studied is visually matched
-------
V - 79
according to color intensity with the closest standard solution, and the scale value
of the standard solution becomes the measured value for the specimen.
At less than five color units, color in water is not discernible, according to
25
the EPA. At more than fifteen color units, the color is displeasing to most
people. In view of this, the EPA has promulgated a secondary MCL of fifteen color
units. That standard was selected as the NSA reference value.
— Color unit values in rural supplies
Most rural water supplies were within the NSA reference value for color
(Figure V-8). The reference value was exceeded in only 2.3 percent of all rural
households (a total of about 513,000). The median value for the rural US was only
3.6 color units. The maximum value was 80 units, recorded for the supplies at
about 25,000 rural households.
Among the relatively small number of households with color unit values
surpassing the reference value, disproportionately more were located in the North
Central and South (3A percent and 2.6 percent, respectively, compared to 1.6
percent in the West and 0.5 percent in the Northeast—see Figures V-8a through V-
8d), and outside of SMSAs (2.8 percent, compared to 1.4 percent inside SMSAs).
Although the proportion of households above the reference value was greater
among households in small rural communities (5A percent) than in large communi-
ties (1.5 percent) or other rural areas (2.2 percent), the difference probably
occurred because a disproportionately large number of small communities were in
the North Central and South, which had the greatest proportions of over-reference-
value households in the regional comparison. A slightly larger proportion of
households served by individual as opposed to intermediate or community systems
had color values exceeding fifteen standard color units (3.0 percent of individual-
system households, compared to 1.9 percent of both intermediate-system and
community-system households).
-------
V - 80
Figure V-8
Color in US Rural Household Supplies
35_
30.
20_
I 5_
I0_
5.
REFERENCE VALUE
97.7%
Lvilxi
2.3 %
Lowest value:
Highest value:
Median:
Interval width:
I
SO
3.6
1.0
Number of households: 21,97^,000
10 15 20
-p—
30
40
50
—T—
60
-1—
70
—r
80
Standard Color Units
-------
V - 81
Regional Variation in Color in US Rural Household Supplies
Figure V-8a. Northeast
REFERENCE VALUE
99.5%
15-
0.5%
—I—
20
Lowest value: 1
Highest value: 50
Median: 2.9
Interval width: 1.0
Number of households: 3,693,000
-f-
50
6*0
70
1T"
30
40
Standard Color Units
Figure V-8b. North Central
REFERENCE VALUE
45
96.6%
40-
30-
Lowest value: 2
Highest value: 80
Median: it. 6
Interval width: 1.0
Number of households: 6,213,000
25-
5-
40
70
80
20
30
50
60
Standard Color Units
-------
V - 82
Regional Variation in Color (continued)
Figure V-8c. South
40-i REFERENCE VALUE
97.4%
35
25-
20-
10-
5-
kv
2.6%
-f P-
Lowest value:
Highest value:
Median:
Interval width:
2
80
3.2
1.0
Number of households: 9,291,000
—I T
30
1 1 1 r>-
10 15 20
40
I T
50
60
70
Standard Color Units
80
Figure V-8d. West
REFERENCE VALUE
98.4% 1.6%
tk
Lowest value: 2
Highest value: 60
Median: 5.0
Interval width: 1.0
Number of households: 2,777,000
10 15
~I-
20
30 40
Standard Color Units
50
60
-------
V - 83
Temperature
Temperature influences the palatability of water, but it also may affect its
healthfulness. The Drexel University Workshop Microbiology Panel observed that:
"Water temperature especially influences survival of enteric organisms, with higher
temperatures promoting inactivation and low temperatures promoting survival. In
cold weather viruses introduced into aquatic environments may persist for several
weeks to months, and become widely dispersed from the source of contamination.
Coupling this longer viral persistence with the fact that disinfection action is
slower in cold water temperatures, there may be some undefined increase in risk of
viral breakthrough into potable water which was processed from poor quality raw
+ .,26
water."
NSA water samples were drawn from the cold water tap, but the
temperature range did not seem great enough to warrant conclusions about possible
survival of viruses, and no attempt was made to measure viral contamination.
Also, though data collection spanned three seasons (summer and fall of 1978, and
winter of 1978-79), only one set of specimens was drawn at any particular sampled
household, and most specimens were obtained from June through October 1978.
Therefore, conclusions about water supply temperature have to be qualified.
Consequently, temperature values in the NSA were used mainly in reference to the
aesthetic acceptability of the water supplies.
In regard to aesthetic acceptability, the EPA concludes that "most
individuals find that water having a temperature between 50° and 60° F (10° to
16° C) is most palatable." Authors of the State of California Water Quality
Criteria state that "for drinking purposes, water with a temperature of 10° C is
usually satisfactory. Temperatures of 15° C or higher are usually objectionable."^
One report quoted in the Water Quality Criteria indicates that public water supply
temperatures in excess of 19° C invariably result in complaints from consumers.
-------
V - 84
In the NSA, temperature was measured by opening the tap and inserting a
thermometer into the flowing water until the temperature reading stabilized. This
usually took about 30 seconds. The NSA temperature readings therefore are
probably higher than would have been recorded if sufficient water had been run to
flush the household plumbing. However, NSA temperatures do reflect the
temperatures of household water as it generally was used during those periods of
the year when households were visited for the NSA. Of course, there could be a
large range of temperatures at any particular household, depending upon how long
water was allowed to run. This makes it difficult to describe the household
condition based on the NSA temperature readings. Consequently, 20° C will be
used as a descriptor for dividing the data and aiding in the presentation of
results—but no formal reference value will be chosen for temperature.
The possible health consequences of domestic water temperature cannot be
assessed on the basis of the NSA data. Further, no criterion for a minimum
temperature was considered in the NSA since the focus of professional attention
has been on the agreeability of warm water rather than cold water. In fact, the
lowest temperatures recorded in rural households were within 5° C of the lower
end of the range viewed as favorable by the EPA.
— Temperature levels in rural supplies
The temperature of major water supplies in rural households, as measured
at the cold water tap, ranged from 5° C through 39° C. Both the mean and the
median temperatures were nearly 20° C. Supplies at 43.8 percent of all rural
households (a total of 9.6 million) were above 20° C (Figure V-9).
Mean and median temperatures were relatively warm in rural household
water supplies in all regions (Figures V-9a through V-9d), but especially in the
South and West. The median water temperature was 22.5° C among Southern
households, 19.9° C among Western households. At the same time, the percentage
-------
V - 85
Figure V-9
Temperature in US Rural Household Supplies
J~L
-1—1—i—r~
5
Ln
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
5.0
39.0
19.7
1.0
21,974,000
| i i i lit
10 15
I1" I ""J"1 1 " T 1 ""p-l
20 25
"^9
degrees Celsius
-------
V - 86
Regional Variation in Temperature in US Rural Household Supplies
Figure V-9a. Northeast
~
ru
Ln
Lowest value: 7.0
Highest value: 30.0
Median: 17.7
Interval width: 1.0
Number of households: 3,693,000
' I
5
-H-
~1—n—i" i' |—r
10 15
lio"
-i—i—i—i—i—r
25
30
degrees Celsius
Figure V-9b. North Centred
12-1
II
10
9-
8
7
l/l
2
-c 6
0)
if)
3
o
* 5
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
6.0
30.0
17.1
1.0
6,213,000
Ln
T 1 1
10
-1—l—l—|—r-
15
-i—I—1—I—i—i—r
20
degrees Celsius
-------
V - 87
Regional Variation in Temperature (continued)
Figure V-9c. South
Lowest value:
Highest value:
Median:
Interval width:
Number of households: 9,291,000
39.0
22.5
9-
8-
7-
5-
4-
3-
2-
25
30
20
degrees Celsius
35
39
Figure V-9d. West
IO-i
Lowest value:
Highest value:
Median:
interval width:
Number of households: 2,777,000
5.0
35.0
20.0
9-
8-
7-
6-
5-
4-
3-
2-
25
20
degrees Celsius
30
35
-------
V - 88
of households having water warmer than 20° C was much greater in the South than
in any other region. The percentage above 20° C in the South was 69.0 percent—a
total of 6A million households. In the West, about 45 percent of households were
over 20° C, compared to about 20 percent in both the Northeast and North Central.
Water temperatures were similar among SMSA and nonSMSA households
(the medians were 19.9° C and 19.7° C, respectively), but showed variation in the
size-of-place and size-of-system comparisons. The median temperature was
slightly warmer among households located in large rural communities (20.8° C)
than among those in small rural communities (19.2° C) or other rural areas
(19.7° C). Also, the percentage of households above 20° C in large rural com-
munities was 53.0 percent, more than ten percentage points higher than for
households in small communities or other rural areas. This finding was in part
attributable to the large concentration (about 45 percent) of all rural large-
community households in the South, where the water tended to be warmer.
Both the median temperature and the percentage of households over 20° C
showed a progressive pattern in the size-of-system comparison: lowest among
households served by individual systems, highest among those served by community
systems. Specifically, median temperatures were about 18°, 19°, and 21° C,
respectively, in households served by individual, intermediate, and community
systems. The respective over-20° C rates were about 27 percent, 38 percent, and
59 percent.
In summary, the NSA data showed that domestic water in rural America
tended to be unsuitably warm—at least by established criteria. This was
particularly the case for households located in large rural communities and served
by community systems. The situation was most prominent in the South and West.
Large community systems may deliver water from surface sources, and
may transmit it for considerable distances, which may account for the higher water
temperatures that were found in community supplies. Depth of pipes in the ground,
-------
V - 89
the weather, latitude, and the source of the water are known to affect water
temperature. Indeed, the NSA regional differences were consistent with the
warmer climate of the South and with the predominance of community systems in
the West (see Chapter IV).
Specific conductance
Specific conductance is an electrical measurement which provides an
indication of the concentration of dissolved mineral salts. The measurement is
particularly helpful in determining suitability of water for irrigation, but it also is
generally useful in determining domestic water quality. For example, increasing
conductance may indicate rising content of dissolved mineral salts from natural or
industrial origin, but the source of the increase and the chemical composition of
the substances must be determined by other tests. Specific conductance is the
reciprocal of the electrical resistance measured between two electrodes one
centimeter apart and one square centimeter in cross-section; conductance is
measured at the existing water temperature and corrected to 25° C.
Conductance values can be related to drinking water quality only insofar as
the values are indicative of the concentration of total dissolved solids in water.
The dissolved solids are primarily mineral salts which, in large amounts, can
increase water hardness, corrosivity, and an unpleasant, "salty" taste. The 1962 US
Public Health Service standards suggested a limit of 500 milligrams of total
dissolved solids per liter of drinking water. This same limit is proposed for a
federal secondary MCL, and was used as the NSA reference value. Conductance in
micromhos per centimeter can be converted to roughly equivalent values of total
dissolved solids (in milligrams per liter) by multiplying the micromho values by
0.65.29' 30
The implications for water quality in terms of total-dissolved-solid content
is taken up here after the discussion of conductance findings. The conductance
-------
V - 90
findings themselves, presented immediately below, provide the background for the
discussion of water quality. However, conductance is a general indicator rather
than a precise measurement of water quality, and no reference value for
conductance was set in the NSA.
— Conductance values in rural supplies
Specific conductance showed a wide range of values (Figures V-10, V-lOa).
At the lower end of the scale, the supplies in some 8,000 rural households had a
conductance of seven micromhos per centimeter. At the other extreme, the
supplies at about 7,000. households had a conductance of 9,152 micromhos. The
mean in rural US households was 473.2; the median was 381.7.
Conductance values varied considerably from region to region (Figures V-
10b through V-lOe), as was to be expected on the basis of different geological and
environmental features. The lowest median values in household supplies were in
the Northeast and South (240.1 and 251.7, respectively); the highest values were in
the North Central and West (600.4 and 444.0, respectively). Median values in the
other NSA groupings (SMSA/nonSMSA, size of place, and size of system) showed
some—but not necessarily meaningful—variation. The most prominent variation
occurred in the size-of-place comparison. The median value among households in
small rural communities (484.3) was higher than in either large rural communities
(388.9) or other rural areas (368.1). The variation was partially attributable to a
disproportionate number of small-community households being located in the North
Central, where conductance values were highest.
— Levels of estimated total dissolved solids in rural supplies
As indicated above, values for specific conductance can be converted to
approximately equivalent values of total dissolved solids by multiplying by 0.65. It
must be emphasized that the conversion provides only an approximation of the
-------
V -91
Figure V-10
Specific Conductance in US Rural Household Supplies
Lowest value: 7.0
Highest value: 9152.0
Median: 381.7
Interval width: 20.0
Number of households: 21,974,000
3-
2-
400
600
1000
1200
1400
1600
micromhos
Figure V-lOa. Cumulative Distribution of Specific Conductance
PT
"T
|
I-
»
I
—
:
i
•
-
Ill
»i—
•
-
1
-1
. li
~t~T
-
l
M
. |.
T
1
• : t i. • i ,nrrrrrr i-tv * >• i 1 ••
• t r
•i i
• i
V>J ^ w w ^ - - -
11
1
•1
— <\i (0 0DO O O O O
i :: .ii- .
II 1
•. i1
!f
micromhos
oo
O o-
<£ ®
NOTE: Cumulative distribution is plotted on semilog paper. Base: 21,974,000 households.
-------
V - 92
Regional Variation in Specific Conductance in US Rural Household Supplies
Figure V-lOb. Northeast
J1
U
Lowest value: 14.0
Highest value: 8240.0
Median: 240.1
Interval width: 20.0
Number of households: 3,693,000
il
lh
j
400
r-—r r
000 1200
H4L4-
n
200
600
800
micromhos
1400
1600
Figure V-lOc. North Central
A
X
Lowest value: 74.0
Highest value: 9152.0
Median: 600.4
Interval width: 20.0
Number of households: 6,213,000
H
[r
LtH
^in
—r——r
1000 1200
200
400
600
800
micromhos
1400
1600
-------
V - 93
Regional Variation in Specific Conductance (continued)
7 Figure V-lOd. South
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
12.0
2600.0
251.7
20.0
9,291,000
n nfl
I 1
IOOO 1200
i P
1600
micromhos
Figure V-lOe. West
6-i
5-
4-
3-
2-
n
n
11
PJ
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
7.0
2935.0
m.o
20.0
2,777,000
h
ji
—, r
800 1000
Ji- il n p l^i
1200 1400
200
400
600
1600
micromhos
-------
V - 94
concentration of total dissolved solids in water supplies. The conductance values
vary according to the specific mineral content of the water, and the more exact
way to measure total dissolved solids is to allow a specimen of water to evaporate,
then to weigh the solid residue. Despite this major qualification, the converted
values for total dissolved solids provide a measure of water quality which is easier
to assess than the conductance values alone. Specifically, the converted values can
be compared to an NSA reference value based on the national secondary MCL for
total dissolved solids. Again, the reference value provides only a rough comparison
point since, among other things, the secondary MCL on which the reference value
is based assumes testing by the evaporative method. The NSA reference value is
500 milligrams of total dissolved solids per liter of water.
Rural US water supplies showed a median value of total dissolved solids
that was about one-half of the reference value. Fully 85.3 percent of supplies were
below the reference value, with values of 500 milligrams per liter or less (Figure
V-ll).
The proportion of supplies above the reference value varied notably from
region to region (Figures V-lla through V-lld). The lowest percentage was in the
Northeast, where only 5.0 percent of households had supplies with more than 500
milligrams of total dissolved solids per liter of water. In contrast, 23.9 percent of
the households in the North Central and 22.2 percent of the households in the West
had supplies which exceeded 500 milligrams per liter. In the South, 10.2 percent
exceeded that level.
»
Although there -were prominent regional differences, no substantial varia-
tion appeared in the other NSA comparisons (SMSA/nonSMSA, size of place, and
size of system).
-------
V - 95
Figure V-li
Estimated Total Dissolved Solids in US Rural Household Supplies
10-
9-
8-
7-
6-
5-
4-
3-
2-
REFERENCE VALUE
85.3% 14.7%
Lowest value:
Highest vaiue;
Medians
Interval width:
Number of households:
59HS.S
248.0
~0.0
21,97 if, 000
-1.
—i r
2800
I2D0
1600
milligrams/liter
2000
2400
3200
-------
V - 96
Regional Variation in Estimated Total Dissolved Solids
in US Rural Household Supplies
Figure V-l la. Northeast
REFERENCE VALUE
95.0% 5.0%
20-
15-
5-
_1
JL
Lowest value: 9.1
Highest value: 5356.0
Median: 156.1
Interval width: <*0.0
Number of households: 3,693,000
n 1—i i 1 1 1 1 1 1 1 j 1 1 1
400 800 1200 1600 2000 1400 2800 3200
500
milligrams/liter
Figure V-lib. North Central
REFERENCE VALUE
15-
10-
76.1% 23.9%
fl
El
1 J r
400
500
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
48.1
5948.8
390.3
k 0.0
6,213,000
1_
qnni n
—i—
1600
1
2000
1
2400
—i r
2800
800
T
1200
3200
milligrams/liter
-------
V -97
Regional Variation in Estimated Total Dissolved Solids (continued)
Figure V-llc. South
REFERENCE VALUE
89.8% 102%
I5-|
:ju
10-
5-:
\pl
1 I
400
500
Lowest value: 7.8
Highest value: 1690.0
Median: 163.6
Interval width: W.O
Number of households: 9,291,000
nn
800
—I
1200
T 1 1
1600
milligrams/liter
Figure V-l Id. West
REFERENCE VALUE
77.8% 2 2 2%
15-
X
LnJ
-i 1 r
400
500
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
<*.6
1907.8
283.6
40.0
2,777,000
1—
800 1200
milligrams/liter
1600
-r~
-------
V - 98
Hardness
Water's "hardness" refers to its capacity to neutralize soap; substances
which form insoluble curds with soap cause hardness. The characteristic of
hardness presents a dilemma for water users. On the one hand, hardness (caused
most often by calcium and magnesium salts) retards the cleaning action of soaps
and detergents and causes a buildup of scale deposits in hot water pipes and
cooking pots. On the other hand, artificial "softening" of water by the ion-
exchange or lime-soda ash processes may increase the sodium content of water and
31
thus make it unsuitable for people restricted to low sodium diets. Furthermore,
32
naturally soft water tends to be more corrosive than hard water. As a result,
soft water can dissolve metals such as cadmium and lead from water pipes or other
containers. These and other metals may have specific adverse health or aesthetic
effects, some of which are described later.
In addition, the NRC reports that the preponderance of evidence indicates
33
that the softer the water, the higher the incidence of cardiovascular disease.
Furthermore, investigators have attributed certain disease-protective effects to
the very substances in hard water which are removed in water softeners—calcium
and magnesium. All of the theories regarding soft-water problems, however, need
extensive testing before final conclusions can be reached, the NRC observes.
According to the NRC, water with less than 75 milligrams of calcium
carbonate equivalent per liter generally is considered soft, and water with more is
considered hard. A more complete categorization has been done by researchers
C. N. Durfor and Edith Becker
-------
V - 99
Hardness Range
(in milligrams of calcium
carbonate equivalents
per liter of water)
0-60
61 - 120
121 - 180
More than 180
Description
Soft
Moderately hard
Hard
Very hard
Although hardness is caused most often by calcium and magnesium
salts, small levels of hardness can be caused by metals, including iron and
35
manganese. Although these two metals were assayed in NSA specimens, their
concentrations were not large enough to alter hardness to a prominent degree. In
the NSA, hardness thus was calculated only on the basis of concentrations of
calcium and magnesium in water. The combined concentrations of the two
•
substances were converted into equivalent quantities of calcium carbonate, which
were used to express hardness. To do this, the concentration of calcium in
milligrams per liter was multiplied by 0.0499 to convert to milliequivalents per
liter. Similarly, the concentration of magnesium in milligrams per liter was
multiplied by 0.08226 to convert to milliequivalents per liter. The sum of the
milliequivalent values was multiplied by 50 to obtain calcium carbonate equivalent
values; those values were used to express hardness.
An NSA reference value was not set for hardness because of the
unresolved questions about its potentially contradictory aesthetic and health
effects. In fact, the EPA considered establishing a secondary MCL for hardness,
but concluded that "available information is not sufficient at this time to balance
the aesthetic desirability of setting a limit for hardness against the potential
37
health risk of water softening."
-------
V - 100
Because of this uncertainty, no attempt is made to interpret the potential
health effects of the NSA findings for hardness. However, the findings are
compared in a general way with values in the hardness range estimates quoted
above. This comparison provides some insight into potential aesthetic and
economic effects of hardness, but not into health consequences.
— Hardness In rural supplies
Hardness, expressed as calcium carbonate equivalent units, ranged from a
low of 0.1 to a high of just over 1,800 in rural US water supplies (Figures V-12,
V-12a). The median was 111.6, which was in the "moderately hard" range
according to the Durfor and Becker scale. Hardness at 36.6 percent of households
was in the "soft" range (0 through 60). At the other end of the scale, a similar
proportion of household supplies (35.3 percent) were in the "very hard" range (more
than 180).
Hardness varied prominently according to region (Figures V-12b through
V-12e). This was anticipated since the condition is influenced by geological
characteristics. Medians were 55.9 and 68.5 in the South and Northeast, respec-
tively. The supplies in those regions thus tended to be soft to moderately hard. In
contrast, the medians in the West and North Central were 156.5 and 255.6,
respectively. Supplies in those regions thus tended to be very hard—particularly in
the North Central, where the median was more than twice that for the nation.
In contrast to the sharp differences that appeared in the regional compar-
ison, few mdjor differences were seen in the comparison of SMSA and nonSMSA
households or in the size-of-system comparison. Medians were nearly identical
(about 111) for both SMSA and nonSMSA supplies. Medians also were similar for
households with individual, intermediate, and tommunity systems—although those
with intermediate systems tended to have slightly less hard water (a median of
-------
V - 101
Figure V-12
Hardness as Calcium Carbonate (CaCC>3) in US Rural Household Supplies
13—1
12-
0-
9-
<0
8-
4)
V>
3
o
X
7-
Lowest value:
Highest value:
Median:
Interval width:
Number of households: 21,974,000
o
1805.4
6-
c
©
u
10.0
£
5-
4-.
2-
In-1
100
200
300
400
milligrams/liter
500
600
700
800
Figure V-12a. Cumulative Distribution of Hardness as Calcium Carbonate
iooq
90^
80^
70-
Hi —
2
o "
O 60-
to
3
O
Z 50:
o Z
£ 40-
o
®
Q.
30
20-
10-
0
—f—
O
(0 CO
o o 1
=r=*r
-H I H
o o o
o
o
o
o"
J±
o
o °-
00 o
o
o
o
o
milligrams/liter
O O
O ^
o o
(0 00 O
o
o
o
o
CO
NOTE: Cumulative distribution is plotted on semilog paper. Base: 21,974,000 households.
-------
V - 102
Regional Variation in Hardness as Calcium Carbonate (CaCO,)
in US Rural Household Supplies
Figure V-12b. Northeast
10-1
9-
8-
7
6-
5-
4-
3-
2-
I -
0
if
1
U
—i—
100
Lowest value:
Highest value:
Median:
Interval width:
2.5
636.8
6S.5
10.0
Number of households: 3,693,000
200
300
400
milligrams/liter
500
—I 1
600
Figure V-12c. North Central
o
9-
8-
7-
6-
5-
4-
3-
2-
I-
fin
%i
lU
P, nW
urn
nJ
ii
Lowest value:
Highest value:
Median:
Interval width:
0.1
1631 .Li
255.6
10.0
Number of households: 6,213,000
"I I
200
I I I i i
300 400
-A
r*~
800
100
500
600 700
milligrams/liter
-------
V - 103
Regional Variation in Hardness as Calcium Carbonate (continued)
18-
Figure V-12d. South
17-
I
10-
o 8-
T 6H
4-
3-
2-
n
u
i r
200
Lowest value:
Highest value:
Median:
Interval width:
0.7
1805.4
55.9
10.0
Number of households: 9,291,000
—r»-
800
100
300
I r
400
500
600
700
milligrams/liter
Figure V-12e. West
8-7
7-
T= 6-
5-
4_ J
3-
2-
-rJUl
1J
\r
U1
Lowest value:
0.3
Highest value:
1607.2
Median:
156.5
Interval width:
10.0
Number of households:
2,777,000
260
, W
300 400
f=4-
n , r
—f*-
800
100
400
milligrams/liter
500
dAo
700
-------
V - 104
106.7, compared to about 112 for households with individual and community
systems).
The size-of-place comparison showed more variation. The median for
households in small rural communities was 195A, more than twice the median value
for households in large rural communities (91.8) and considerably larger than the
value for households in other rural areas (109.2). Once again, this difference was
at least in part attributable to a disproportionate number of small-rural-community
households being located in the North Central, where water hardness was much
greater than in other regions.
Summary of physical and chemical characteristics
Of the characteristics in this section of the NSA study, turbidity probably
is the most comprehensive, but least specific, indicator of water quality. Gener-
ally, domestic water with low turbidity offers less interference to industrial and
commercial applications, is easier to disinfect, offers less opportunity for prolifer-
38
ation of bacteria, and may be less susceptible to taste and odor problems. Nearly
84 percent of rural household supplies had readings lower than one NTU. High
values, on the other hand, were particularly prominent in the North Central and
South, and in- households served by individual or intermediate systems as opposed to
community systems.
Households which had more than 500 milligrams per liter of total dissolved
solids (as derived from specific conductance) also were most prominent in the
North Central, where about one of every four households had excessive concentra-
tions. The proportion above this reference value was nearly as large in the West,
but much smaller in the South and Northeast.
As to water hardness, the North Central once again had more than its share
of supplies with very hard water. The median in the North Central (255.6 calcium
-------
V - 105
carbonate equivalent units) was more than twice the median value for the nation,
and considerably larger than the median for any other region.
Color was not a problem in rural supplies. Water temperature, on the other
hand, was a potential problem. Measured temperature was above 20° C at 43.8
percent of all rural households. The proportion of households with warm water was
most prominent in the South, where about seven of every ten households had water
temperatures over 20° C. The proportion over 20° C in the West was less than that
in the South, but large in comparison to the Northeast and North Central. An
important consideration, however, is that the 20° C mark was derived from
conventional abstract measures of desirable water temperature. Household resi-
dents, on the other hand, may consider water that is warmer than 20° C
acceptable.
INORGANIC CONSTITUENTS
A number of inorganic substances, ranging from chemical compounds to
heavy metals, have recognized health effects. Standards for acceptable levels of a
number of the substances have been established by public health authorities over
the years. Constituents with potential health, economic, or aesthetic effects were
studied in NSA water specimens. Some of the substances—calcium, magnesium,
nitrates, sulfates, iron, manganese, sodium, and lead—were measured in all of the
NSA specimens. Others—arsenic, barium, cadmium, chromium, mercury, selenium,
silver, and fluoride—were measured only in a special subsample of the specimens.
Specimens from this subsample, which were designated as "Group II," comprised 10
percent of the total number of specimens obtained in the study. The former
(Group I) substances tended to be those which were traditionally acknowledged by
public health experts as being important in public water supplies. Some (nitrates
and lead) had federal interim primary Maximum Contaminant Levels (MCLs); some
(sulfates, iron, and manganese) were included in the national secondary regulations;
-------
V - 106
and one (sodium) was under consideration by the EPA. The latter substances
—those assayed in the Group II subsample—all had interim primary MCLs, but
there was less expectation of finding them in problematic quantities. The federal
MCLs for inorganic substances were used primarily as guides in determining NSA
reference values, as explained earlier in this chapter.
Calcium
Calcium compounds are common in water. The element is essential to
human nutrition, and the diet should include about seven-tenths to two grams of
calcium per day, an amount considerably greater than that found in water—even
hard water. Some evidence implicates excessive calcium and magnesium in
drinking water as predisposing people to kidney or bladder stones, but other
evidence points to deficiency of calcium in water as being a more serious
. problem.^
The situation is summarized in California's exhaustive reference source,
kO
Water Quality Criteria: "So far as can be determined at the present time,
calcium limits are desirable for domestic supplies not because of a hazard to
health, but because calcium may be disadvantageous for other household uses, such
as washing, bathing, and laundering, and because it tends to cause incrustations on
cooking utensils and water heaters. Hibbard has recommended the following
limiting concentrations of calcium in waters for domestic use:
Drinking and Cooking 30 milligrams per liter
Washing 10 milligrams per liter
Laundry 0 milligrams per liter."
Because of the uncertainty about the effects of specific levels of calcium
in domestic water, the NSA findings were not compared directly to a reference
-------
V - 107
value. The findings thus are presented without an attempt to analyze their
significance for water quality.
— Calcium levels in rural supplies
Calcium was detected in US rural supplies in amounts ranging from less
than 0.05 to 582 milligrams per liter of water (Figures V-13, V-13a). The median
for the rural US was 30.0 milligrams of calcium per liter; the mean was 41.0.
Calcium was one of the two constituents used to determine hardness, and
regional differences in calcium concentrations paralleled those reported for
hardness (Figures V-13b through V-13e). Thus, median values were smallest among
household supplies in the South (17.0 milligrams per liter) and in the Northeast
(19.3 milligrams per liter), and largest among household supplies in the West (40.0
milligrams per liter) and in the North Central (68.0 milligrams per liter).
Results of other NSA groupings also paralleled those for hardness. Thus,
the median concentration among both SMSA and nonSMSA households was about 30
milligrams of calcium per liter, as was the median concentration among households
served by each size of system, whether individual, intermediate, or community.
The size-of-place differences also paralleled those for hardness, with
median values largest among households located in places of less than 1,000 people
(small rural communities). These differences were partially attributable to the
regional distribution of households, as described in the discussion of hardness,
above.
Magnesium
As one of the most abundant elements in the earth's crust, magnesium is
widely distributed in ores and minerals. Magnesium salts generally are very
soluble, and large concentrations are found in water.
-------
V - 108
Figure V-13
Calcium in US Rural Household Supplies
lO—i
Lowest value: < 0.05
Highest value: 582.0
Median: 30.0
2.0
Number of households: 21,974,000
V)
o
•C
«
V)
2
x
Interval width:
o
c
e
u
a>
Q.
120
0
20
40
60
80
100
140
160
milligrams/liter
Figure V-13a. Cumulative Distribution of Calcium
100
90
80
70
60
50
« 40
30
20
10CDO O
o oo°. °
o oo
o o
o oou
NOTE: Cumulative distribution is plotted on semilog paper. Base: 21,974,000 households.
-------
V - 109
Regional Variation in Calcium in US Rural Household Supplies
Figure V-13b. Northeast
6-
5-
K
Lowest value:
Highest value:
Median:
Interval width:
1.0
199.0
19.3
2.0
Number of households: 3,693,000
U
U
¦P^n-n
I 1 1 1 I 1 I 'T
20 40 60 80
milligrams/liter
Figure V-13c. North Central
7-
6-
5-
4-
3—
Lowest value: < 0.05
Highest value: * 522.0
Median: 68.0
Interval width: 2.0
Number of households: 6,213,000
rOl
IT
J1
\
LrwLnjJ]
20
40
60
80
100
120
140
milligrams/liter
160
-------
V - 110
Regional Variation in Calcium (continued)
Figure V-13d. South
ii-i
10-J
9-
8-
7-
Lowest value:
Highest value:
Median:
Interval width:
Number of households: 9,291,000
582.0
17.0
6-
5-
4-
3-
2-
40
100
20
60
20
60
80
milligrams/liter
Figure V-13e. West
Lowest value: 0.1
Highest value: 451.0
Median: 40.0
Interval width: 2.0
Number of households: 2,777,000
3
2
I-
,-j0
J1
nJ
iJ
-flr^
—!—
40
—]—
60
[L
—i—
140
—r>~
160
20
80
100
120
milligrams/liter
-------
V - 111
Magnesium is an essential element in the human diet; excessive amounts
seldom are hazardous since they usually are excreted before harm is done. Large
concentrations of magnesium sulfate in drinking water may cause diarrhea at first,
but the body apparently counteracts the effect with time, as it does the laxative
effect of sulfate (see below). In the past, the US Public Health Service has
recommended a maximum concentration of 100 milligrams per liter of water (in
1925), and 125 milligrams per liter (in 1942 and 1946). The recommendation was
dropped in 1962, however, and the new federal regulations do not set limits for the
substance. The World Health Organization's international standards specify 150
41
milligrams of magnesium per liter of drinking water as excessive.
To provide a descriptive guide, NSA investigators selected a reference
value of 125 milligrams of magnesium per liter of water. This level was the same
as the former Public Health Service standard, but lower than the World Health
Organization recommendation. Selection of the more stringent reference value
was in accord with the general NSA procedure of using the more conservative
measure, but the World Health Organization recommendation was also used as a
reference in evaluating potential health consequences.
— Magnesium levels in rural supplies
As expected, magnesium was detected in nearly all rural water supplies.
The concentrations varied greatly—from a low of less than 0.002 milligrams per
liter to a- high of 142.6 milligrams per liter (Figure V-14). However, only 0.1
percent of households (21,000) had supplies with values exceeding the NSA
reference value of 125 milligrams per liter. The US median value was 7.1
milligrams of magnesium per liter.
Magnesium was one of the two constituents used to calculate hardness, and
concentrations of the metal varied in a regional pattern similar to that for
hardness. The smallest median values were observed among households in the
-------
V - 112
Figure V-14
Magnesium in US Rural Household Supplies
50-
40-
30-
20-
10-
Lowest values
Highest values
Median:
Interval width:
0.002
142.6
7.1
5.0
Number of households: 21,97^,000
-r~
10
20
—1—
30
1 1 1—
40
milligrams/liter
REFERENCE VALUE
99.9% 0.1%
50
125
135
"I—I
-------
V - 113
South (2.9 milligrams per liter) and in the Northeast (4.4 milligrams per liter); the
largest values were found among households in the West (14.9 milligrams per liter)
and in the North Central (19.1 milligrams per liter). In conjunction with these
findings, household supplies exceeding the NSA reference value were found only in
the West and North Central, where median values were highest (Figures V-14a
through V-14d).
The medians and the proportions of households exceeding the NSA refer-
ence value were not strongly influenced by location of households inside or outside
SMSAs. There were variations according to the size of system serving the
household, but the variations were too small to be meaningful. Again, the most
prominent variation occurred in the size-of-place grouping. There, the median
value among households located in small rural communities was 11.3 milligrams of
magnesium per liter, more than twice the value among households in large rural
communities, and about one and one-half the value for households in other rural
areas. This difference was in part attributable to the regional distribution of
households, as explained in the discussion of hardness, above.
There did not appear to be serious health consequences associated with the
NSA findings for magnesium. On the basis of present research knowledge, even the
highest concentrations of magnesium appeared to present no imminent health
hazard. That is, the highest concentration (142.6 milligrams per liter), which
occurred in some 8,000 households, exceeded the NSA reference value but not the
World Health Organization recommendation. High concentrations were found most
often in the West and North Central.
Nitrates
Nitrogenous materials tend to be converted to nitrates in lakes, streams,
and groundwater. Most naturally occurring nitrogenous compounds enter the water
in organic matter. A small amount is in precipitation. Concentrated amounts
-------
v - m
Regional Variation in Magnesium in US Rural Household Supplies
Figure V-14a. Northeast
6O-1
50-
40-
¦? 30-
Q- 20-
REFERENCE VALUE
100 0% 0.0%
Lowest value: 0.01
Highest value: 56.1
Median:
Interval width: 5.0
Number of households: 3,693,000
10-
20
30
1 1 I
40
milligrams/liter
I
50
125
Figure V-14b. North Central
20-1
10-
REFERENCE VALUE
99 9% 0.1%
Lowest value: < 0.002
Highest value: * 128 .5
Median: 19.1
Interval width: 5.0
Number o! households: 6,213,000
[
10
—r~
20
—r~
30
1 1 f—
40
milligrams/liter
I
50
125
-------
V - 115
Regional Variation in Magnesium (continued)
Figure V-14c. South
REFERENCE VALUE
100.0%
70-i
° 60-1
20-
10-
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
< 0.002
102.6
2.9
5.0
9,291,000
T~~
10
—r_
20
30
40
milligrams/liter
j—r
100
I
110
0.0%
120 125
Figure V-14d. West
30-i
20-
10-
REFERENCE VALUE
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
< 0.002
1^2.6
U.9
5,0
2,777,000
r~
10
—j—
20
—p..
30
40
milligrams/liter
—I—
50
99.5% 0.5%
125
—!
135
-------
V - 116
enter from municipal and industrial wastewater pipes, refuse dumps, animal feed
lots, and septic tanks. Another, more diffuse source of nitrates is inorganic
chemical fertilizer applied to the land.
Nitrates themselves generally are not direct health hazards. Nitrates in
the human intestine, however, can be converted by bacterial action into nitrites.
This is a particular problem in infants because the pH in the stomachs of infants
tends toward alkalinity (pH of 5-7) because the acid secretion pattern found in
older humans is not yet developed. As a consequence, nitrate-converting bacteria
are able to inhabit areas of the small intestine closer to the stomach. The
converted nitrites then have longer residence time, and therefore absorption time,
in the intestine than if acidity from the stomach forced the bacteria farther down
h 2
the tract. Absorbed nitrites oxidize hemoglobin in the blood to methemoglobin,
which blocks the crucial function of oxygen transport to the body's tissues. The
result can be severe oxygen depletion (the "blue baby" syndrome).
Generally, standards for the maximum acceptable concentration of nitrates
in drinking water are linked to the anoxic threat in infants. According to the NRC,
a value of about ten milligrams of nitrate-N (nitrate content expressed as
equivalent nitrogen) per liter of water is the maximum level at which no adverse
health effects have been observed. This value is the same as the federal interim
primary MCL. The NRC cautions, however, that "there is little margin of safety in
43
this value."
The NSA reference value was the same as the interim primary MCL—ten
#
milligrams of nitrate-N per liter of water.
— Nitrate-N levels in rural America
Large concentrations of nitrate-N previously had been found in shallow
wells, particularly in the Midwest, and it was assumed that a fairly large number
of rural supplies might exceed the NSA reference value. Fortunately, this was not
-------
V - 117
the case. In the rural US, 97.3 percent of households were below the reference
value; only 2.7 percent were above it (Figures V-15, V-15a). Despite some larger
values, the mean level in rural America was only 1.7 milligrams of nitrate-N per
liter, and the median was only 0.3.
As expected, those households above the reference value were predomin-
antly in the North Central and West, regions which have been associated with
excessive nitrate-N values in the past.^ The percentage of North Central
households above the reference value was 5.8, twice the national average. The
percentage of Western households exceeding the reference value was 4.0. Only 1.3
percent of households in the South and 0.3 percent in the Northeast surpassed the
reference value (Figures V-15b through V-15e). Also as anticipated, a larger
proportion of nonSMSA households had values greater than the reference value
—3.2 percent, compared to 1.7 percent for SMS A households.
As to size-of-place variation, the proportion of households above the
reference value was close to the national average in other rural areas, but half
again as large as the national average in both large and small rural communities
(4.2 percent and 4.7 percent, respectively). This finding was surprising since in
previous studies many of the supplies with excessive amounts of nitrates were on
farms, which most often are located in other rural areas. Again, the difference
may have been attributable in part to differential distributions in the NSA sample.
That is, a disproportionately large number of small-rural-community households
(about 48 percent) were located in the North Central, where the proportion of
households beyond the nitrogen reference value was largest. At the same time, a
disproportionately small number of small-rural-community households (about 8
percent) were located in the Northeast, where the proportion of households above
the nitrogen reference value was smallest. The overall efiect was to increase the
over-reference-value rate in the small-rural-community category. A similar but
-------
V - 118
Figure V-15
Nitrate-N in US Rural Household Supplies
REFERENCE VALUE
55-i
97 3%
50-
Lowest value:
Highest value;
Median:
Interval width:
99.5
0.3
0.2
Number of households: 21,971,000
45_
o
o
c
o
I
milligrams/liter
Figure V-15a. Cumulative Distribution of Nitrate-N
100
95
90
85
80
75
§ 70
65
60
55
50
milligrams/liter
NOTE: Cumulative distribution is plotted on semilog paper. Base: 21,97^,000 households.
-------
V - 119
Regional Variation in Nitrate-N in US Rural Household Supplies
Figure V-15b. Northeast
REFERENCE VALUE
45n
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
13.0
40-
0.2
3,693,000
10-
5-
milligroms/liter
Figure V-15c. North Central
REFERENCE VALUE
—
<
>
94.2%
Lowest value: <0.3
Highest value: 99.5
Median: 0.3
Interval width: 0.2
'I Number of households: 6,213,000
5. 8 %
1 1 ! 1 1 1 1 1 I
— i i ~ i r r i >
milligrams/liter
-------
V - 120
Regional Variation in Nitrate-N (continued)
Figure V-15d. South
REFERENCE VALUE
60->
98.7%
55-
Lowest value:
Highest value:
Median:
Interval width:
Number of households: 9,291,000
67.0
15-
10-
5-
milligrams/liter
Figure V-15e. West
Lowest value:
Highest value:
Median:
Interval width:
< 0.3
19.8
0.589
0.20
Number of households: 2,777,000
l/l
ILtLtuTiJ
EL
REFERENCE VALUE
96.0% 4.0%
I 1 1
6 8 10
milligrams/liter
I
16
12
i
14
-------
V - 121
much weaker effect influenced the over-reference-value rate in the large-rural-
community category.
Regarding the size-of-system findings, there was an inverse relationship
between the proportion of households exceeding the reference value and the size of
system serving the households. There were 4.1 percent of households served by
individual systems above the reference value, compared to 3.0 percent of house-
holds served by intermediate systems and 1.6 percent served by community
systems. Individual systems were most prevalent among households in other rural
areas, less prevalent among households in small rural communities, and least
prevalent among households in large rural communities. Community systems, on
the other hand, were least common among households in other rural areas and most
common among households in large rural communities. The overall pattern, then,
indicated that the size of system was more important than the size of place in
determining the over-reference-value rate. This was consistent with other findings
pointing to a much higher incidence of large nitrate concentrations in individual
wells in open country than in community systems in larger rural places.^
Although relatively few rural supplies surpassed the N5A reference value,
it is important to keep in mind that those that did (about 603,000) could pose an
important health threat in the form of increased risk to infants aged about four
months or less. On the basis of NSA data, this risk of exposure to high nitrates was
greatest among Western or North Central households served by individual or
intermediate systems.
Sulfates
Sulfates are natural constituents of water. They also are generated by
human activities. They enter the water from sediments, precipitation, domestic
wastes, and industrial wastes. Sulfates tend to remain dissolved in water unless
they are removed artificially.
-------
V - 122
In view of the prevalence and persistence of sulfates in drinking water, it is
fortunate that the substances have relatively minor health effects. At concentra-
tions exceeding 500 milligrams per liter of water, sulfates can cause diarrhea. The
level at which this laxative effect occurs is assumed to be greater than 600
milligrams per liter of water in the national secondary drinking water regula-
tions.*^ Sulfate compounds can cause detectable tastes at concentrations of 300
to 400 milligrams per liter, according to the regulations.
Regular users of sulfate-containing water apparently develop resistance to
the laxative effect. The main hazard is to travelers or visitors who drink the water
infrequently. Similarly, although excessive levels of sulfates may taint the water,
the taste may be acceptable to those who use the water regularly.
The secondary MCL for sulfates, as well as the NSA reference value, is 250
milligrams of sulfate per liter of water.
— Sulfate levels in rural supplies
Concentrations of sulfates in rural supplies were generally well within the
NSA reference value (Figures V-16, V-16a). Fully 96.0 percent of all rural
households were below the reference value; only 4.0 percent were above it. The
median sulfate level in rural US supplies was 17.0 milligrams per liter of water.
Median concentrations of sulfates were more than 30 milligrams per liter
in both the North Central and West, but only half that great in the Northeast and
South. The occurrence of over-reference-value supplies reflected this pattern: the
rates were greater than 7 percent among major household supplies in the North
Central and West, but smaller than 1 percent in the Northeast and South (Figures
V^16b through V-16e). The rate was somewhat higher among nonSMSA households
(4.8 percent) than among those located inside SMSAs (2.2 percent), and higher
among households in small rural communities (7.5 percent) than among those in
large rural communities (2.6 percent) or other rural areas (3.8 percent). However,
-------
V - 123
Figure V-16
Sulfates in US Rural Household Supplies
54-
53-
REFERENCE VALUE
H
96.0% 4.0%
Lowest values
Highest value:
Median:
Interval widths
Number of households!
< 9.S
> 999.5
17.0
5.0
21,97*#,000
milligrams/liter
Figure V- 16a. Cumulative Distribution of Sulfates
REFERENCE VALUE
100
95'
90
96.0%
4.0 %
CP CO
<£ CO
milligrams/liter
NOTE: Cumulative distribution is plotted on semilog paper. Base: 21,974,000 households.
-------
v - m
Regional Variation in Sulfates in US Rural Household Supplies
Figure V-16b. Northeast
60-1
59-
10-
9-
8-
7-
6-
5-
4-
3-
2
I -
Vn
1—
50
\d]SL
REFERENCE VALUE
99.5%
Lowest value:
<15.0
Highest value:
655.0
Median:
16.5
Interval width:
5.0
Number of households:
3,693,000
I
150
n-.
0.5%
100
I r
200 250
milligrams/liter
I 1 T
300 350
Figure V-16c. North Central
34-t
33
6
5
4
3
2-
I-
fL
REFERENCE VALUE
92.6% 7.4%
Lowest value: '15.0
Highest value: > 999.5
Median: f 35. S
Interval width: 5.0
Number of households: 6,213,000
JU Ln-
50
150
200 250
milligrams/liter
300
350
400
-------
V - 125
Regional Variation in Sulfates (continued)
Figure V-16d. South
REFERENCE VALUE
—
*
99.3%
Lowest value: <15.0
Highest value: > 999.5
Median: 15.0
Interval width: 5.0
Number of households: 9,291,000
if]
Wji n
0.7 %
1 1 1 1 1 1)11
i i i 1 1 1
0 50 100 150 200 250 300 350 400
milligrams/liter
Figure V-16e. West
46-,
45-
44-
e
6-
5-
REFERENCE VALUE
88.3% 11.7%
Lowest value:
Highest value:
Median:
Interval width<
Number of households:
< 9.8
> 999.5
32.if
5.0
2,777,000
r"1, ^
400
50
200 250
milligrams/liter
300
350
-------
V - 126
the size-of-place findings were influenced by a disproportionate number of small-
community households being located in the North Central, where the median
sulfate level was relatively high.
In regard to the size-of-system grouping, the incidence of households over
the reference value was exactly the same for households using community systems
(4.2 percent) as for those using individual systems; that rate was two and one-half
times larger than for households served by intermediate systems (1.7 percent).
Median concentrations, however, were the same for households with individual or
intermediate systems (15.0 milligrams per liter of water), and greater for house-
holds with community systems (24.0 milligrams per liter).
Regardless of the size-of-system differences, the levels in excess of the
reference value did not represent a serious health problem. Among those 4.0
percent of households (about 870,000) above the reference value, few had concen-
trations equal to those which were assumed by the EPA to be a potential cause of
diarrhea. Specifically, in 0.9 percent of all rural households—roughly one out of
four of the over-reference-value group—the concentrations were higher than 600
milligrams of sulfate per liter. In those households, visitors unaccustomed to the
drinking water might have difficulties. The remaining 3.1 percent of the rural
supplies with high levels had concentrations between 250 milligrams and 600
milligrams of sulfate per liter. In those households, supplies may have had
objectionable tastes, but they would be expected to have posed few gastrointestinal
problems.
Iron
Iron is a common natural constituent of water. The amount ingested from
48
water is small in comparison to that consumed in food, however. The concentra-
tions of iron normally in water thus pose no known threat to human health. On
the other hand, excessive amounts of iron in water promote a reddish-brown
-------
V - 127
discoloration in laundry, stain water fixtures, and cause an astringent or bitter
taste in drinking water. In view of these objectionable characteristics, EPA has
established a secondary MCL of 0.3 milligrams per liter of water. That MCL value
was used as the NSA reference value.
— Iron concentrations in rural supplies
Approximately eight out of ten rural households were below the NSA
reference value for iron (Figures V-17, V-17a). In fact, the median for supplies in
the rural US was only 0.10 milligrams of iron per liter—one-third of the reference
value. Despite the favorable overall situation, supplies in 18.7 percent of rural
households were above the reference value. In 2.6 percent of rural households
(about 570,000), the concentration was more than ten times the reference value.
The reference value was exceeded most often in households in the North
Central, where 28.2 percent of households were high. Over-reference values in the
South and Northeast were 17.0 and 16.0 percent, respectively. In the West, by
contrast, only 7.0 percent of households were above 0.3 milligrams per liter (see
Figures V-17b through V-17e). Consistent with these findings, the highest median
concentrations of iron occurred in household supplies in the Northeast, South, and
North Central (0.1 milligrams per liter in each of the three regions) rather than in
the West (0.05 milligrams per liter).
The proportion of nonSMSA households above the NSA reference value for
iron was greater than that for SMS A households (21.0 percent versus 13.8 percent).
In the size-of-place comparison, supplies with high values were located at least
twice as often in small rural communities and other rural areas (23.3 percent and
19.5 percent, respectively) as in large rural communities (9A percent). In addition,
households served by individual and intermediate systems were over the reference
value about four times as often as households served by community systems. The
exact proportions were 29.9 percent of households served by individual systems and
-------
V - 128
Figure V-17
Iron in US Rural Household Supplies
REFERENCE VALUE
40-i
18.7 %
35-
Lowest value:
Highest value:
Median:
Interval width: 0.01
Number of households: 21,97<>,000
2 30-
25-
20-
5-
o.io
0 20
0 30
0.40
0.50
0.60
0.70
0.80
milligroms/liter
Figure V-17a. Cumulative Distribution of Iron
REFERENCE VALUE
milligrams/liter
NOTE: Cumulative distribution is plotted on semilog paper. Base: 21,974,000 households.
-------
V - 129
Regional Variation in Iron in US Rural Household Supplies
Figure V-17b. Northeast
60-]
5 5-
5-
REFERENCE VALUE
84.0% 16 0%
SJlil
Lowest value: 0.10
Highest value: 29.60
Median: 0.10
Interval width: 0.01
Number of households: 3,693,000
0.60
0.70
—r*~
0.80
0.10
0 20
0.30
0.40
milligrams/liter
0 50
Figure V-17c. North Central
REFERENCE VALUE
71. 8%
¦: 0.05
11.20
0.10
Lowest value:
Highest value:
Median:
Interval width:
Number of households: 6,213,000
25-
5-
0.80
0.50
0.60
0.70
0.10
0.20
0.30
0.40
milligrams /liter
-------
V - 130
Regional Variation in Iron (continued)
Figure V-17d. South
45-
REFERENCE VALUE
83 0% 17.0%
—I
0.20
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
< 0.03
34.70
0.10
0.01
9,291,000
—
0.80
0.10
0.30
n l
0.40
milligrams/liter
0.50
0.60
0.70
Figure V-17e. West
6 5-t
60-
5-
REFERENCE VALUE
93.0% 7.0%
1
n
0.20
,—n-,,
Lowest value: < 0.02
Highest value: 5.70
Median: 0.05
Interval width: 0.01
Number of households: 2,777,000
1
0.70
1 r*~
0.80
0.10
0.30
0.40
milligrams/liter
0.50
0.60
-------
V - 131
28.7 percent among those served by intermediate systems, compared to 7.7 percent
of households served by community systems.
The NSA reference value for iron provided a dividing point in terms of non-
health-related, objectionable characteristics of iron compounds in domestic water.
In that regard, a fairly large number of rural households (4.1 million) had
concentrations which exceeded the reference value, and residents therefore faced
potential problems from discolored laundry or from bitter or astringent tastes. The
overall situation in rural America, however, was not serious since so many rural
supplies were within the reference value.
Manganese
Like iron, manganese is a natural constituent in water. This metal also
poses little danger to health in concentrations which can be expected in water. In
fact, manganese is an essential trace element which is important to proper enzyme
function in human beings, as is iron.
Manganese is less abundant in waterborne compounds than is iron, but its
objectionable characteristics are much the same: it stains laundry and water
fixtures, and has an unpleasant taste. In view of these effects, the EPA has
adopted a secondary MCL of 0.05 milligrams of manganese per liter of water. That
also was the NSA reference value.
— Manganese levels in rural supplies
Manganese resembles iron in chemical behavior, and it frequently is found
kS
with iron in groundwater. It is less abundant in rocks, however, and thus its
concentration in water usually is less than that of iron.^ This overall picture is
consistent with that found in the NSA. The largest concentration of manganese in
major household supplies of the rural US was 10.2 milligrams per liter—about one-
third the highest concentration of iron. The median concentration of manganese in
-------
V - 132
rural supplies was 0.029 milligrams per liter—compared to 0.1 milligrams for iron.
Overall, 85.8 percent of rural households were below the manganese reference
value; 14.2 percent of households were above it (Figures V-18, V-18a). By
comparison, 18.7 percent of rural households were over the iron reference value.
The proportion of households with high manganese values varied in differ-
ent NSA comparative groupings in a pattern similar to that for iron. As seen in
Figures V-18b through V-18e, high readings occurred most often among households
located in the North Central (19.9 percent) and Northeast (16.9 percent), less often
in the South (12.3 percent), and least often in the West (4.7 percent). High readings
were more frequent among nonSMSA households (16.3 percent) than among SMS A
households (9.9 percent), and more frequent among households located in small
rural communities (21.7 percent) or other rural areas (14.0 percent) than among
households in large rural communities (11.4 percent). As with iron concentrations,
manganese concentrations exceeding the reference value were proportionately
nrtore frequent (about three times so) among households served by individual or
intermediate systems than among households served by community systems, which
had an over-reference-value rate of 7.2 percent. Treatment methods to reduce
manganese concentrations were not generally used by intermediate and individual
systems, and rural households served by community systems were less likely to
have excessive amounts of the metal, indicating that those systems were better
able to deal with the problem.
As with iron compounds, the reference value provided a dividing point for
non-health-related, objectionable characteristics of manganese. From that per-
spective, there were potential problems with taste and staining properties in the
supplies at 3.1 million rural households (those above the reference value). The
general pattern of the extent of the problems was very similar to that for iron, and
since iron and manganese frequently occur together, the two substances represent
-------
V - 133
Figure V-18
Manganese in US Rural Household Supplies
55n
50-
REFERENCE VALUE
85.8% 14.2%
45-
- <
T3
o 25-
3
x
r 20-
10-
Lowest value:
Highest value:
Median:
Interval width:
< 0.01
10.20
0.029
0.01 •
Number of households: 21,97^,000
I i r—rur~'"r~' r
0 0.05 0.10 0.20
030 0.40 0.50 0.60 0.70 0.80
milligrams/liter
Figure V-18a. Cumulative Distribution of Manganese
REFERENCE VALUE
85.8% 14.2%
CO 00 O
milligrams/liter
NOTE: Cumulative distribution is plotted on semilog paper. Base: 21,97^,000 households.
-------
V - 135
Regional Variation in Manganese (continued)
Figure V-18d. South
60-
REFERENCE VALUE
87.7% 12.3%
0.20
Lowest value: < 0.01
Highest value: 10.20
Median: 0.030
Interval width: 0.01
Number of households: 9,291,000
T"
T
T
—i r
0 60
—i—
0.70
—
0 80
0.50 0 10
0.30 0.40
milligrams/liter
0.50
Figure V-18e. West
I
REFERENCE VALUE
95.3%
4.7%
r-n
Lowest value:
Highest value:
Median:
Interval width:
< 0.01
L.l<*
0.012
0.01
Number of households: 2,777,000
H 1 1-
0.40
milligrams/liter
!
0.50
n
0.80
0.05 0.10
0.20
0.30
0.60
0.70
-------
V - 136
an aesthetic difficulty for a number of households in the rural US—particularly in
the North Central.
Sodium
The sodium ion is ubiquitous in water. It comes naturally from surface and
underground deposits of salts such as sodium chloride, from the sodium aluminum
silicates and similar minerals, from rainfall which contains evaporated salt water
particles, and from seawater which enters fresh water aquifers. It also comes from
sodium chloride used to de-ice roads and from a number of substances in municipal
and industrial wastewater. Other sources in domestic water supplies are the ion-
exchange or the lime-soda ash water softening processes.
Sodium plays an essential role in the regulation of fluid balances in the
body. Yet adults in the US apparently routinely take in more than ten times the
amount of sodium which they require.^ * According to the NRC: "An impressive
amount of evidence has accumulated over the last several decades that sodium
taken in excess of physiologic need is important in inducing an age-related increase
in blood pressure that culminates in hypertension in genetically susceptible people."
In view of this hazard, the NRC cautions: "Concentrations (of sodium)
should be maintained at the lowest practicable levels, and trends toward increasing
concentrations of sodium in water supplies as a result of deicing and water-
softening procedures should be discouraged. Optimal concentrations of sodium
52
should be regarded as the lowest feasible."
A standard procedure in toxicology is to determine the highest concen-
tration of a substance at which no adverse effect is observed. This "threshold"
value is determined on the basis of animal experiments and public health
experience. A safety factor then is applied, which produces a much lower value to
be used as the one considered safe for humans. This value then, can be used as a
basis for maximum permissible concentrations of the substance in food, water, and
-------
V - 137
air—based on a number of related considerations, such as total exposure to the
substance and its rate of dietary intake. This approach was used in setting official
standards which became the bases for NSA reference values regarding a number of
inorganic substances, but the situation was more complicated for sodium. The
NRC states: "Specification of a 'no-observed-adverse-health-effect' level in water
for a substance like sodium for which the effect is associated with total dietary
intake and for which food intake is already greater than a desirable level is
impossible."
One arbitrary approach is to aim for an allowable maximum level which
offers some protection to persons in the US who are required to use sodium-
restricted diets because they have hypertension. Persons who must severely limit
their sodium intake should not use drinking or cooking water which contains more
than twenty milligrams of sodium per liter of water. The EPA recognizes the
desirability of having supplies with a sodium content of less than twenty milligrams
per liter, but the position of the agency is that "regulation of sodium by a
maximum contaminant level is a relatively inflexible, very expensive means of
dealing with a problem which varies greatly from person to person."^ A massive
control program would be necessary since, by EPA's estimate, about 40 percent of
US public water supplies have a natural or added sodium content greater than
twenty milligrams per liter. In view of this situation, the EPA advises that sodium
monitoring programs are the most practical countermeasures: if excessive levels of
sodium are found in domestic water, users can be warned and, if necessary, they
can then use alternative water sources.
There clearly are problems with complete reliance on a monitoring
program. For example, monitoring may be of uneven quality in different areas;
users may not always receive sufficient warning about excessive amounts of sodium
and, even if they are warned, there may be no practical alternative water sources
for them to use. In view of these considerations, NSA investigators decided that
-------
V - 138
the most useful approach in analysis of NSA data would be to use two reference
values. One reference value selected was twenty milligrams of sodium per liter;
the other was 100 milligrams. The former, more stringent, value was appropriate
for persons who must limit their sodium intake in the extreme—to a total of about
54
500 milligrams of sodium each day from all sources. Indeed, the "EPA suggests
that sodium levels of 20 mg/1 or less in drinking water be considered as optimal."^
The latter, less stringent, value would offer some advantage to persons who require
a less restrictive low-sodium diet.^
— Sodium levels in rural America
The median sodium concentration in rural households was about twelve
milligrams per liter of water. In terms of the more stringent NSA reference value,
the sodium concentration exceeded twenty milligrams per liter of water in 39.1
percent of rural households (Figures V-19, V-19a). This was consistent with the
EPA's estimate that about 40 percent of all US public water supplies had
concentrations of this amount. As to the less stringent NSA reference value, the
sodium concentration was higher than 100 milligrams per liter in 14.3 percent of all
rural households. This was consistent with data used by the EPA which indicated
that 14.6 percent of all public water supplies had concentrations of this amount.
The agreement between the NSA findings and earlier EPA estimates reinforced
existing impressions of the scope of the potential problem in the US.
Household sodium concentrations varied prominently in different regions of
the US (Figures V-19b through V-19e). Median concentrations were highest in the
West (24.72 milligrams per liter) and in the North Central (20.05 milligrams per
liter). Those median concentrations were more than twice as large as the ones in
the South (9.29 milligrams per liter) and in the Northeast (8.2 milligrams per liter).
The percentage of households exceeding either of the NSA reference values varied
in a pattern similar to that for the median concentrations. That is, the proportion
-------
V - 139
Figure V-19
Sodium in US Rurai Household Supplies
13
12-
II-
10-
9-
999.5
12.0
2.0
Number of households: 21,97^ ,000
100
T
120
140
160
Figure V- 19a. Cumulative Distribution of Sodium
REFERENCE VALUES
100-1
90
80-
70~
60-jj
50-
40
30-
20~
10-
0
60.9%
'III Mil
tc co
o o'o'1
39 1% 85.7%
14.3%
z±
o
C\J
O
O O
<£> R
O
..1...
o
04
milligrams/liter
o o o
o o o"
^ (0 coo
o
o
o
O O O <0
o o o
o o o
00
NOTE: Cumulative distribution is plotted on semilog paper. Base: 21,97^,000 households.
-------
V - 140
Regional Variation in Sodium in US Rural Household Supplies
15-
14-
13-
12-
II-
10
9-
8-
5-
4
3-
2-
Figure V-19b. Northeast
REFERENCE VALUES
78.7%
21.3%
94.0%
"Irnr
6.0%
Lowest value: 1.0
Highest value: 404.0
Median: 8.2
Interval width: . 2.0
Number of households: 3,693,000
(Ma
120
20
40
60 80 100
milligrams/liter
I
140
160
Figure V-19c. North Central
REFERENCE VALUES
II-
10-
9-
8-
"O
0 7-
.e
©
1 6"
"o
_ 5.
c
03
-------
V - 141
Regional Variation in Sodium (continued)
Figure V-19d. South
REFERENCE VALUES
16-
10-
9-
8-
Lowest value:
Highest value:
Median:
Interval width:
Number of households: 9,291,000
a>
"3
O
I
695.3
9.3
o
5-
c
©
u
a.
3-
2-
140
00
120
20
40
80
milligrams/liter
60
Figure V-19e. West
10-
9-
8-
7-
6-
¦5 5-
4-
3-
2
ul
REFERENCE VALUES
45.1%
54.9%
85.0%
KJlJlJ
fl
20
I
40
In
LTtJ
-1—
60
i I
80
milligrams/liter
X.
15.0%
Lowest value:
Highest value:
Median:
Interval width:
0.2
WO.O
24.7
2.0
Number of households: 2,777,000
Jh.
100
120
I
140
160
-------
V - 142
of households above either reference value was generally larger in the West and in
the North Central than in the South and Northeast.
Variations in median sodium concentrations and in proportions of house-
holds exceeding either reference value were much less evident in other NSA
groupings. One possible exception was a tendency for larger median concentrations
and more values in excess of the reference values to be found among households
served by community systems than among those served by individual or inter-
mediate systems. This tendency was particularly apparent in regard to the
reference value of twenty milligrams per liter: about 44 percent of households
served by community systems (4.8 million) exceeded that reference value. The
contrasting percentages were about 35 percent for households served by individual
systems and 31 percent for households served by intermediate systems.
As noted above, the NSA reference values for sodium were health-related,
but they did not represent a professional consensus on the level at which specific
health risks could be expected in the general population. In addition, it would have
been necessary to know something about the health status and dietary habits of
individuals in NSA households to determine the possible consequences of large
sodium concentrations in water supplies. That is, only a certain percentage of
rural residents had the specific diseases (such as hypertension, renal disease, and
cirrhosis of the liver) which require restricted sodium intake. Identification of
these individuals and a study of their dietary experiences were far beyond the
scope of the NSA.
Lead
Since Roman times, lead in water has been linked to health problems. In
the centuries since then, investigators have found no beneficial effects for human
health. They have found persistent quantities of the substance, however, in the air
we breathe, food we eat, and water we drink.
-------
V - 143
Most of this exposure stems from sources related to human activities.
Lead in the air in cities, for example, often comes from the exhaust of vehicles
burning leaded gasoline. Ingested lead frequently is in the form of leaded paint
peelings in old dwellings—children find the peelings attractive and eat them. Lead
in domestic water most often comes from water pipes—naturally soft, acidic water
dissolves lead from service connections, lead-lined household piping, and soldered
joints.
Human exposure to lead must be limited since the metal can accumulate in
the body and cause serious damage to the kidney, liver, brain, reproductive
systems, and central nervous system. At subtoxic levels, lead interferes with the
functioning of red blood cells; at toxic levels, it may destroy the red blood cells.
Mental retardation in children is a common result of lead poisoning.
Permissible levels of this metal in air, food, and water are established in
light of anticipated total intake of the metal as people go about their daily
activities. This is the customary way to set limits for a number of toxic substances
which have cumulative effects. Concentrations in each exposure source—air, food,
and water—must be low enough to guard against dangerous overall accumulation of
the substance in the human body.
In the case of lead, intake among city dwellers frequently is so large that
the cumulative total—50 to 60 micrograms of lead per day—is at the theoretical
level at which adverse effects can be expected. In view of this, the NRC cautions
that existing US regulations for waterborne lead may not provide adequate
protection: "Results of studies in the Boston area indicate that increased blood
levels of lead will occur in children when the water supply contains 0.05 - 0.1
milligrams of lead per liter of water. Thus, the interim limit of 0.05 milligrams of
lead per liter may not provide the margin of safety to safeguard the high-risk
population of urban areas ... It is concluded that the no-observed-adverse-health-
-------
v - m
effect level cannot be set with assurance at any value greater than 0.025
milligrams per liter of water
The NSA reference value for lead was the same as the interim primary
MCL—0.05 milligrams of lead per liter of water.
— Lead levels in rural America
Large concentrations of several heavy metals studied in the NSA were
much more pervasive in rural US water supplies than was anticipated. Lead was
one of those metals—cadmium and mercury, discussed below, were the others.
The reference value for lead was exceeded in 16.6 percent of rural
households—a total of 3.6 million (Figures V-20, V-20a). The range of US rural
lead concentrations was from less than 0.005 milligrams of lead per liter of water
to 0.97 milligrams; the median was 0.008 milligrams; the mean was 0.03 milli-
grams.
In the South, 23.1 percent of- households were over the reference value
—the largest proportion in any region. The lowest rate was in the Northeast
(Figures V-20b through V-20e); nonetheless, 9.6 percent of household supplies in the
Northeast had concentrations which exceeded the reference value. Moreover,
although 9.6 percent was lower than the proportion in any other region, it was
considerably higher than was anticipated on the basis of the EPA's experience.
That experience had indicated that less than 4 percent of public water supplies had
58
lead concentrations exceeding the reference value. It might have been antici-
pated that large lead concentrations would be more common in households served
by individual systems with lead pipes or lead-alloy-soldered pipes, and this might
have led the EPA to expect larger concentrations among NSA samples (the EPA
estimate was based on studies of community systems as opposed to rural household
supplies). This expectation also would not have been borne out, however; in
the NSA size-of-system comparison, a larger proportion—and a larger total
-------
V - 145
Figure V-20
Lead in US Rurai Household Supplies
42-i
4 I-
40-
REFERENCE VALUE
83 4% 16.6%
U>
? 8-
.C
4)
«/>
3
£ 7-
O
c 6"
u
Q. 5-
4-
3-
2-
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
<0.003
0.970
0.008
0.002
21,974,000
, , j
0.04 0.05 0.06
milligrams/liter
Figure V-20a. Cumulative Distribution of Lead
REFERENCE VALUE
100
90
80
70
60
50
§..40-
30-
Z0-
10-
milligrams/liter
NOTE: Cumulative distribution is plotted on semilog paper. Base: 21,97^,000 households.
-------
V - 146
Regional Variation in Lead in US Rural Household Supplies
Figure V-20b. Northeast
REFERENCE VALUE
9.6%
62-
Lowest value:
Highest value:
Median:
< 0.005
0.970
6-
< 0.005
0.002
3,693,000
Interval width:
5-
Number of households:
4-
3-
2-
r
0.08
milligrams/liter
0.02
0.04 0.05 0.06
Figure V-20c. North Central
REFERENCE VALUE
89 2%
9-
8-
fl
1
Ln
Ln
i 1 r
0 0 02
In.
10.8%
Lowest value:
Highest value:
Med ian:
Interval width:
Number of households:
0.003
0.600
0.007
0.002
6,213,000
r|1 Hl\\ -r^Lf^ r-^n
0.04 0.05 0.06
0.08 0.10
milligrams/liter
0.12
i r
0.14
0.16
-------
V - 147
Regional Variation in Lead (continued)
Figure V-20d. South
REFERENCE VALUE
371
36-
6-
5-
4-
3
2
I -
0
Ln
1
ii
76.9%
U1M
n 1 r
0.02
23.1%
Lowest value:
Highest value:
Median:
Interval width:
< 0.005
0.330
0.011
0.002
Number of households: 9,291,000
1 1
r^Lflrfjln
8 O.IO
0 04 0.05 0 06 0.08
milligrams/liter
0 12
0.14
0.16
10-
9-
8-
7
6-
5-
4
3
2-
I-
Figure V-20e. West
28-i REFERENCE VALUE
2t\ []
tin
83 1%
h
u
ii
16.9
1 1 i i r r
0 0.02 0.04 0.05 0.06
Lowest value:
Highest value:
Median:
Interval width:
< 0.005
0.302
0.010
0.002
Number of households: 2,777,000
—I 1 1—
0.08 0.10
milligrams/liter
—n»"
0.16
0.12
0.14
-------
V - 148
number—of households served by community systems were over the reference
value, compared to households served by individual systems. Specifically, 17.7
percent of households served by community systems (1.9 million) were high with
regard to the lead reference value, compared to 14.1 percent of households using
individual systems (1.2 million). The largest proportion (20.5 percent)—but the
smallest number (458,000)—of high-value households were served by intermediate
systems.
As to variation in the other NSA groupings, proportions over the lead
reference value were higher among nonSMSA households (18.3 percent, compared
to 12.9 percent of SMSA households), and somewhat higher in small rural
communities (18.1 percent, compared to 15.1 percent in large rural communities
and 16.6 percent in other rural areas).
The health implications of the NSA findings are potentially serious. The
risk from lead intake is dependent on several factors, but it is particularly
troublesome for children who are exposed to large amounts of the metal. Such
exposure is most common in urban areas, as noted above. In rural areas, where
airborne lead levels usually would not be expected to be as large, total exposure
would be lower. The NSA data provide no indication of total exposure to lead, and
the findings about concentrations in the water thus must be interpreted in isolation
from other relevant considerations. Nevertheless, the NSA findings indicate that
lead intake could reach hazardous levels in some rural households.
The mean total dietary intake of lead by North American adults is
59
estimated to be from 0.2 to 0.3 milligrams per day. Roughly one-tenth of this
intake (about 0.026 milligrams per day) is attributed to lead in water used for
drinking or cooking. This estimate of the contribution from domestic water is
based on the assumption that the average lead concentration in tap water is 0.013
milligrams per liter, and that daily water consumption by adults is two liters.^
The NSA findings, however, showed a mean lead concentration in rural US
-------
V - 149
household supplies of 0.03 milligrams of lead per liter—over two times the
assumed average level. In fact, although the mean is influenced by relatively large
concentrations in a small number of households, about 38 percent of rural
households exceeded the anticipated level of 0.013 milligrams per liter. The
concentrations in those households are not necessarily cause for concern until they
approach the reference value of 0.05 milligrams per liter, but they demonstrate the
potential for unexpectedly heavy exposure to lead in the diet of rural Americans.
That exposure becomes particularly noteworthy in rural households in
which the lead concentration in the major water supply is large enough to push
total dietary intake to relatively high levels. It has been estimated that total
intake of appreciably more than 0.6 milligrams of lead per day on a regular basis
and over a long period of time may lead to dangerous accumulation of the metal in
the human body.*'* Assuming that a rural adult American eats about 0.2 milligrams
of lead in food each day (the per capita estimate for North Americans, discussed
above), the additional intake from water in a small proportion of rural households
would be enough to raise total intake well above 0.6 milligrams per day. For
example, supplies at about 110,000 households had concentrations of at least 0.3
milligrams of lead per liter. An adult in one of these households, then, might in the
course of one day consume two liters of water with a total of 0.6 milligrams of
lead, plus food containing 0.2 milligrams. His total intake that day would be 0.8
milligrams of lead, which could be hazardous if continued regularly even for as
62
short a period as four years because of accumulation in the body.
The situation probably is most serious in households .with children aged one
to three years. Children of this age drink only about half as much water as do
63
adults, but they are more susceptible to adverse effects of lead. As a result,
their situation should be carefully evaluated in households with supplies over the
reference value—a total of 3.6 million households in rural America.
-------
V - 150
Arsenic
This solid, brittle chemical element has a fearsome reputation as a poison.
Yet its toxic effects differ strikingly according to the chemical compound, the
route by which it is taken in, and the duration of the exposure. For example, the
lethal dose of the more toxic arsenic compounds may be one to 25 milligrams per
kilogram of body weight in animals; the lethal dose for less toxic compounds may
be ten to 400 times this amount.
Small concentrations of arsenic are common in US waters. The material
probably comes mostly from natural geological sources, although some is from
industrial sources such as smelters.
Specific health effects of waterborne arsenic in the US are unclear, but
there does appear to be a potential, yet-to-be-defined risk. According to the NRC:
"The evidence for an association between arsenic and disease in some human
populations has been further strengthened by recent epidemiological studies such as
those conducted in the waters of Puget Sound, in local water supplies such as those
in Lassen County, California; Perham, Minnesota; Lane County, Oregon; Antofa-
gasta, Chile; and on the southwest coast of Taiwan. Skin lesions, including cancer,
and a circulatory disorder referred to as 'blackfoot' are major clinical problems
where chronic exposure to arsenic exists. Human disease associated with arsenic is
not exactly duplicated in animals, although misuse of arsenicals results in disease
in dogs and in cattle. . . The different forms of arsenic that exist in the
environment may account for differences in clinical manifestations between
different localities."^
In view of the indirect, epidemiological evidence of a link between skin
cancer and large concentrations of arsenic in drinking water, the current US
interim primary MCL of 0.05 milligrams per liter may not provide an adequate
margin of safety, according to the NRC. The NRC does not specify a recom-
mended level, however.
-------
V - 151
In the NSA survey, arsenic was one of eight inorganic substances studied
exclusively in 10 percent of the household specimens—the Group II subsample. As
noted previously, all the Group II constituents are reported on in sequence in the
following pages, beginning with arsenic and ending with the radionuclides.
The reference value used for comparative purposes in the NSA was the
same as the interim primary MCL of 0.05 milligrams of arsenic per liter of water.
— Arsenic levels in rural supplies
Arsenic was above the minimum detection limit in about 17 percent of
rural supplies (Figures V-21, V-21a), but the concentrations rarely exceeded the
reference value (0.05 milligrams per liter). In fact, detected concentrations in
most supplies were half or less of that allowed by the reference value. Levels
exceeded the reference value in only 0.8 percent of rural households. The highest
value was 0.179 milligrams of arsenic per liter (in 0.3 percent of households).
The median value was consistently less than 0.005 milligrams of arsenic
(less than one-tenth the reference value) in all rural households, regardless of NSA
grouping. The only exception occurred in the West, where household supplies had a
slightly higher median value, 0.008 milligrams of arsenic per liter of water (Figures
V-21b through V-21e).
Consistent with the regional distribution of median values, the proportion
of households exceeding the reference value was greatest in the West—2.1 percent
of households there. The proportion was 1.8 percent in the North Central. Within
the statistical limits of NSA findings, the results showed that the reference value
was not exceeded in households in the Northeast or South.
The proportion of households over the reference value for arsenic was
greater among nonSMSA households (1.2 percent, compared to 0.0 percent of SMS A
households). Small rural communities had a much greater proportion of house-
holds above the reference value—6.6 percent, compared to 0.0 percent in large
-------
V - 152
Figure V-21
Arsenic in US Rural Household Supplies
70-,
2 65-
3
X
*
10-
REFERENCE VALUE
992% 0.8%
'-pnn-.or/l p
Lowest value:
Highest value:
Median:
Interval width:
< 0.002
0.179
< 0.005
0.001
Number of households: 21,97* ,000
1 T
0.060
0.010
0.020
0.030
0.040
milligrams/liter
0 050
i i r*—
0 070 0 080
Figure V-21a. Cumulative Distribution of Arsenic
REFERENCE VALUE
100;
90;
80;
70 ¦
«0
s
o _
» 60 3
3
O
Z 5°:
o
§ 40-
u
4)
Q.
30-
20-
10-
99.2%
irit
-—:rir
|.... _:.L..
3i±
r4
0 8%
±0
¦i—
life
m
o
o
6
CJ
o
o
o
o
6
(0
o
o
6
®
o o
O ^
o
CJ
o
ttH—
o o o
milligrams/liter
o
o
o
o
o
0)
r-
o
NOTE: Cumulative distribution is plotted on semilog paper. Base: 21,974,000 households.
-------
V - 153
Regional Variation in Arsenic in US Rural Household Supplies
Figure V-21b. Northeast
lOO—i
95-
<
REFERENCE VALUE
100.0% 00%
Lowest value:
Highest value:
Median:
Interval width:
< 0.005
0.006
< 0.005
0.001
Number of households: 3,693,000
0.010 0.020 0.030 0.040 0 050
milligrams/liter
Figure V-21c. North Centred
50-,
45-
15-
5-
LJ
L
hi
r—'i i r
o.oio o 020
REFERENCE VALUE
98.2% 1.8%
Lowest value: < 0.002
Highest value: 0.082
Median: < 0.005
Interval width: 0.001
Number of households: 6,213,000
I T
0.060
0.b70 ' oljgo
0.030
0.040
milligrams/liter
0.050
-------
V - 15k
Regional Variation in Arsenic (continued)
Figure V-21d. South
85-
10
REFERENCE VALUE
100.0% 0.0%
IT
0 0.010 0.020
Lowest value: < 0.002
Highest value: 0.019
Median: < 0.005
Interval width: 0.001
Number of households: 9,291,000
0.030 0.040
milligrams/liter
0 050
Figure V-21e. West
REFERENCE VALUE
9 7.9% 2.1%
0 I i r
0 0.010 0.020
Lowest value: < 0.002
Highest value: 0.179
Median: 0.008
Interval width: 0.001
Number of households: 2,777,000
0.030
0 040
milligrams/liter
0.050
0.060
0.070
0.080
-------
V - 155
communities and OA percent in other rural areas—but all of the high-value small-
community households were located in the North Central, where the reference
value was exceeded most often. The pattern of high values showed little variation
in the size-of-system grouping.
On the basis of the NSA findings, arsenic was not pervasive in US rural
household water supplies. When found, it appeared in relatively low concentra-
tions. Even the highest concentrations posed no identifiable immediate health
threat.
Selenium
This nonmetallic element resembles sulfur chemically. The substance is
one of those which has a dual influence on human health: too little results in
nutritional insufficiency; too much (in certain forms) produces a variety of adverse
symptoms.
Selenium toxicity may be modified by the presence of other elements such
as arsenic, mercury, cadmium, and thallium. In particular, selenium toxicity in
laboratory animals is alleviated or prevented by administration of sodium
+ 65
arsenate.
Water-soluble selenium compounds come from both natural and artificial
sources. According to the NRC, "there is a wide variation in concentration of
selenium, depending on geologic location. Thus, groundwaters and surface waters
may contain significant amounts of selenium, particularly in areas where there is
an excess of selenium in rocks and soils; in other areas, there may be little (if any)
66
detectable selenium in the water." Despite the likelihood of selenium in many
water sources, the NRC adds "there is little in the literature to indicate that
surface waters contain toxic amounts of selenium; in fact, it is likely that there is
an insufficient amount of selenium in the water alone to provide the nutrient
requirement of most animals, but concentrations may vary in different places."
-------
V - 156
In view of this, the NRC advises: "Rather than concern for toxicity, the
literature indicates that there is a greater potential for a deficiency. Consider-
ation should be given to raising the current permitted levels in water of the United
States (the U.S. interim primary MCL is 0.01 milligrams of selenium per liter of
water)."^
In the NSA, the interim primary MCL was used as the reference value.
— Selenium levels In rural supplies
Supplies at 86.3 percent of all rural households were below the selenium
reference value; those at 13.7 percent were above (Figures V-22, V-22a). Selenium
levels were as high as 0.114 milligrams per liter of water (in 0.3 percent of rural
i
households). The median level in rural US households was less than 0.005
milligrams per liter.
As was frequently true for constituents studied in the NSA, variation
among the total percentage of households exceeding the MCL was most pronounced
according to region rather than other groupings (Figures V-22b through V-22e).
Thus, 41.3 percent of the households in the West had supplies with selenium
concentrations over the reference value. At the other extreme, no households in
the Northeast and only 2.1 percent of households in the South had supplies with
concentrations above the reference value. In the North Central, 25.7 percent of
households had supplies exceeding the reference value.
In general, variations in the proportions of supplies exceeding the reference
value were not prominent in other comparisons. One possible exception was that in
households located in small rural communities, supplies exceeded the reference
value less often (6.6 percent) than in households located in large rural places (16.5
percent) or in other rural areas (14.0 percent). Another possible exception was that
supplies also exceeded the reference value less often in households served by
-------
V - 157
Figure V-22
Selenium in US Rural Household Supplies
REFERENCE VALUE
86.3% 13.7%
77-
76-
2 3-
2-
Lfl
rL
JUH
Lowest value:
Highest value:
Median:
Interval width:
< 0.005
0.114
< 0.005
0.001
Number of households: 21,974,000
,
0.05
i
O.OI
0.02
0.03
0.04
0.06
0.07
milligrams/litgr
0.08
Figure V-22a. Cumulative Distribution of Selenium
REFERENCE VALUE
86.3 /o 13.7 /o
milligrams/liter
NOTE: Cumulative distribution is plotted on semilog paper. Base: 21,974,000 households.
-------
V - 158
Regional Variation in Selenium in US Rural Household Supplies
Figure V-22b. Northeast
REFERENCE VALUE
too-
99-
98-
100 0%
e-
0.0%
0 0.005 0.010
milligrams/liter
Lowest values < 0,005
Highest value: < 0.005
Median: < 0.005
Interval width: 0.001
Number of households: 3,693,000
74.3% 25.7%
Figure V-22c. North Central
56-, REFERENCE VALUE
55-
54-
U"
Lowest value;
Highest value;
Median:
Interval width:
Number of households:
< 0.005
0.11#
< 0.005
0.001*
6,213,000
0 050
o.'oeo '
—i ( 1 >¦
0.070 0.080
0 010
0.020
0.030
0.040
milligrams/liter
-------
V - 159
Regional Variation in Selenium (continued)
Figure V-22d. South
REFERENCE VALUE
97.9%
94-t
9 3-
3 92-
- 4-
3-
2-
1
2.1%
iu
r
0 010 0.020
milligrams/liter
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
< 0.005
0.025
< 0.005
0.001
9,291,000
Figure V-22e. West
38-
37-
10-
8-
6-
REFERENCE VALUE
^ 4-
3-
I -
58.7%
41.3%
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
< 0.005
0.058
0.009
0.001
2,777,000
0.010 0.020 0.030
milligrams/liter
0.040 0.050
-------
V - 160
individual or community systems than in households served by intermediate
systems.
As to possible health implications, the NSA findings indicated that selen-
ium was not present in rural household supplies in sufficient quantities to pose a
health threat. At the same time, the findings suggested the need for caution in
raising the present interim primary MCL as is suggested by the NRC (see above).
The highest concentration in the NSA was 0.114 milligrams of selenium per liter, a
concentration which by itself did not represent a health threat. However, dietary
68
intake of selenium from other sources totals about 0.2 milligrams a day.
69
Assuming that a person drinks two liters of water a day, total daily selenium
intake could be about OA3 milligrams a day in households with 0.114 milligrams of
selenium per liter of drinking water. According to data relied on by the EPA, signs
of selenium toxicity have occurred at an estimated intake as low as 0.7 milligrams
of selenium per day. There is still a margin of safety between this lower threshold
danger point and the exposure level which could be anticipated in the households
with the largest selenium concentrations, but the margin is not great, particularly
if exposed persons drink more than two liters of water a day and eat food with
more than average amounts of selenium.
Fluoride
Fluorine exists naturally as fluoride. The concentration of fluoride in
water depends principally on the solubility of local fluoride-containing rocks. In
addition, fluoride compounds such as sodium fluoride have been added to drinking
water supplies for more than 30 years in the US as a countermeasure against tooth
decay (caries).
The use of fluorides in drinking water has prompted a continuing public
debate. Fluorides in appropriate concentrations reduce the incidence of tooth
decay, but in excessive amounts fluorides can cause mottling of the teeth.
-------
V - 161
These conditions are not likely to develop, however, according to the NRC.
The council offers this summation:
"There is no generally accepted evidence that anyone has been
harmed by drinking water with fluoride concentrations considered
optimal for the annual mean temperatures in the temperate zones. It
seems likely, however, that objectionable dental fluorosis mottling
occurred in two children with diabetes insipidus. Bone changes,
possibly undesirable, have been noted in patients being dialyzed
against large volumes of fluoridated water. Similar changes can be
expected in the rare renal patient with a long history of renal
insufficiency and a high fluid intake that includes large amounts of
tea. With this particular combination of circumstances, the lowest
drinking-water concentration of fluoride associated with symptoma-
tic skeletal fluorosis that has been reported to date is three ppm
(equivalent to three milligrams per liter of water), outside of
countries such as India. It should be possible for the medical
profession to avoid the possible adverse effects of fluoride under the
conditions described above, thereby making it unnecessary to limit
the concentrations of fluoride in order to protect these rare patients.
On the basis of studies done more than fifteen years ago, occasional
objectionable mottling would be expected to occur in communities in
the hotter regions of the United States with water that contains
fluoride at one ppm or higher and in any community with water that
contains fluoride at two ppm or higher. However, this may not be the
case today; more liberal provisional limits seem appropriate while
studies are conducted to clarify the subject.
"The possibility of fluoride causing other adverse effects (allergic
responses, mongolism, and cancer) or beneficial effects other than
decreased dental caries has not been adequately documented to carry
weight in the practical decision about the desirable levels of fluoride.
The questions of mongolism and cancer have been raised on the basis
of epidemiological data for which there is contrary evidence and the
risk factors involved in any case are too low to establish a causal
association. The allergic responses claimed by some reports are
based on clinical observations and in some case double blind tests.
The reservation in accepting these at face value is the lack of similar
reports in much larger numbers of people who have been exposed to
considerably more fluoride than was involved in the original observa-
tions. From a scientific point of view none of these effects can be
ruled out, but the available data are rather limited or easily improved
so further study is indicated.
Interim primary MCLs for fluoride have been established by the EPA
according to a schedule which takes into account local air temperature. (The
* hotter the climate, the more water consumed and thus the greater the amount of
fluoride taken in—a situation of particular significance to children, whose teeth
-------
V - 162
are most susceptible to fluoride mottling.) The maximum contaminant levels for
fluoride are set according to the annual average of the maximum daily air
temperatures for the location in which the community water system is situated:
Temperature
(degrees,
Fahrenheit)
Temperature
(degrees,
Celsius)
MCL
(milligrams
per liter)
53.7 and below
12.0 and below
2.4
53.8 to 58.3
12.1 to 14.6
2.2
58.4 to 63.8
14.7 to 17.6
2.0
63.9 to 70.6
17.7 to 21.4
1.8
70.7 to 79.2
21.5 to 26.2
1.6
79.3 to 90.5
26.3 to 32.5
1.4
In describing the background for the national regulations, the EPA states:
"Excessive fluoride in drinking water supplies produces objection-
able dental fluorosis which increases with increasing fluoride concen-
tration above the recommended upper control limits. In the United
States, this is the only harmful effect observed to result from
fluoride found in drinking water. Other expected effects from
excessively high intake levels are: (a) bone changes when water
containing 8-20 mg fluoride per liter (8-20 mg/1) is consumed over a
long period of time; (b) crippling fluorosis when 20 or more mg of
fluoride from all sources is consumed per day for 20 or more years;
(c) death when 2,250-4,500 mg of fluoride (5,000-10,000 mg sodium
fluoride) is consumed in a single dose."71
Fluoride concentration was determined in water specimens from the Group
II NSA subsample. The reference value used was the lowest allowable MCL value:
1.4 milligrams of fluoride per liter of water. Data on local air temperatures were
not readily available, thus prohibiting the use of a set of reference values to
parallel the MCL. The selection of 1.4 milligrams of* fluoride per liter of water,
the lowest value in the range of MCLs, was in line with the NSA policy of selecting
the most conservative value when no other selection criterion was available. In the
case of fluoride, the MCLs are very close to the concentrations at which known
-------
V - 163
adverse health effects are noticed. The use of 1.4 milligrams per liter as a
reference value provides the greatest margin of safety.
— Fluoride levels in rural supplies
Concentrations of fluoride generally were well below the MCL-related
reference value of 1.4 milligrams per liter of water (Figure V-23). Only 2.5
percent of rural households had supplies with concentrations which exceeded that
level. In fact, fewer than 1 percent of all rural households had values (3.02
milligrams per liter) which exceeded the maximum, temperature-based, interim
primary MCL of 2.4 milligrams of fluoride per liter of water. The median value for
rural US water supplies was 0.20 milligrams of fluoride per liter, or about one-third
the reference value.
The reference value for fluoride was exceeded proportionately more often
in the West than in other regions (Figures V-23a through V-23d). The reference
. value was exceeded somewhat more often among nonSMSA households than among
SMSA households, and more than twice as often in households in small rural places
than in households in large rural places or other rural areas.
Pronounced variations occurred according to the size of system serving
rural households. The fluoride reference value was exceeded most often in
households served by intermediate systems (in about 7 percent of those households).
The reference value was exceeded least often in households served by individual
systems (in slightly less than 1 percent of those households). The reference value
was exceeded in only about 3 percent of households using community systems. On
the other hand, median fluoride values were lower among households served by
individual systems and intermediate systems (0.10 milligrams per liter in both
cases) than among households served by community systems (0.38 milligrams per
liter). This finding may reflect the effect of fluoridation in some community
systems.
-------
V - 164
Figure V-23
Fluoride in US Rural Household Supplies
REFERENCE VALUE
97 5% 2.5%
50-i _
U Lrij
'III'
0.50
i—i—r
1 r**
.00
Lowest value:
Highest value:
Median:
interval width:
0.10
3.02
0.20
0.1
Number of households: 21,974,000
-p—i—r~i I rw-
1.40 2.00
milligrams/liter
i—i—i—i—i—r
2.50 3.00
¦P
-------
V - 165
Regional Variation in Fluoride in US Rural Household Supplies
Figure V-23a. Northeast
2 85-
80-
5-
s\
REFERENCE VALUE
100.0% 0.0o/c
n£
Lowest value: <0.10
Highest value: 1.10
Median; 0.11
Interval width: 0.1
Number of households: 3,693,000
0.5 1.00
milligrams/liter
1.40
Figure V-23b. North Central
n
REFERENCE VALUE
98.2% 1.8%
-t r
0.50
i ~
.00
Lowest value: < 0.10
Highest value: 2. IS
Median: 0.33
Interval width: 0.1
Number of households: 6,213,000
~
1.40
2.00
milligrams/liter
-------
V - 166
Regional Variation in Fluoride (continued)
Figure V-23c. South
REFERENCE VALUE
97.3% 2.7%
L^J
^ ~A
i i i i i ' —i i |
0.50 1.00
a
Lowest v^lye:
Highest value:
Median:
Interval width:
Number of households:
' I '
2.00
-R-
< 0.10
3.02
0.16
0.1
9,291,000
T—1—I—I—I—r
2.50 3.00
1.40
milligrams/liter
Figure V-23d. West
REFERENCE VALUE
93.8% 6.2%
20-
ru
' 11
0.50
-fl.
' r
1.00
1.40
milligrams/liter
Lowest value: <0.10
Highest value: 1.90
Median: 0.38
Interval width: 0.1
Number of households: 2,777,000
T
1.90
-------
V - 167
NSA findings suggested no serious health consequences from the levels of
fluoride in rural water supplies. The highest concentration of the substance (3.02
milligrams per liter) was below the range at which long-term bone changes might
be a hazard. The most prominent effect of excessive fluoride intake by human
beings is mottling of the teeth, and this might be a problem in those households
72
with concentrations higher than two milligrams of fluoride per liter. That
concentration is equivalent to the two ppm (parts per million) threshold which some
research has linked to increased incidence of mottling (see above). The NSA
findings revealed that only about 1 percent of rural households had concentrations
of at least two milligrams of fluoride per liter. These levels were found only in the
South and North Central—with the highest value (3.02 milligrams per liter)
occurring in the South. This finding indicated that mottling of teeth was a
potential but isolated problem primarily in the South and North Central, even
though the largest proportion of households over the reference value was in the
West. That is, although a larger proportion of households exceeded the reference
value in the West than in the other regions, the largest values actually were found
in the South and North Central.
Cadmium
Cadmium is a soft, silver-white, blue-tinged element that is chemically
related to mercury. The main source of cadmium in source water has been
assumed to be industrial discharges which release the metal into the water
directly, or indirectly through atmospheric emissions which contaminate precipita-
73
tion. The most serious health risk is to persons who breathe the metal in
industrial emissions or in cigarette smoke. Waterborne cadmium poisoning,
however, has produced more than 200 cases of severe degenerative bone disease in
Japan. In the US, the concern is centered primarily on possible long-term
7k
development of hypertension caused by continued exposure to cadmium.
-------
V - 168
According to the EPA: "The average concentration of cadmium in drinking
water from community supplies is 1.3 micrograms per liter in the United States.
Slight amounts are common, with 63 percent of samples taken at household taps
showing one microgram per liter or more."^
Despite the prevalence of the substance, the EPA estimates that "only 0.3
percent of tap samples would be expected to exceed the limits of ten micrograms
per liter (equivalent to 0.01 milligrams per liter, the official interim primary
MCL)." The NSA reference value was the same as the interim primary MCL—0.01
milligrams per liter.
— Cadmium levels in rural supplies
A far larger proportion of rural households exceeded the cadmium refer-
ence value than was anticipated on the basis of existing estimates. Fewer than 1
percent of US rural households were expected to have readings beyond the
reference value (see above), but instead 16.8 percent did (Figures V-24, V-24a).
The highest recorded concentration was 0.046 milligrams of cadmium per liter of
water; the median for the rural US was less than 0.002 milligrams per liter.
Although the median values for cadmium concentrations were at or near
the limit of detection in each region, the proportion of households over the
reference value changed considerably from region to region (Figures V-24b through
V-24e). By far the largest proportion of households with high levels was in the
West, where supplies in 27.1 percent of rural households had levels in excess of the
reference value. In sharp contrast, supplies in only 1.6 perce'nt of Northeast rural
households had concentrations that high. The reference value was exceeded in 20.7
percent of North Central households and in 17.3 percent of Southern households.
Supplies in excess of the reference value occurred in 21.4 percent of SMS A
households, compared to 14.3 percent of nonSMSA households. The reference value
was surpassed more than twice as often in large rural places (19.8 percent) and
-------
V - 169
Figure V-24
Cadmium in US Rural Household Supplies
4-
REFERENCE VALUE
83.2% 16.8%
ls\
S\
K.1J
1 i i i i
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
<0.002
0.046
<0.002
0.001
21,974,000
4=^3-
' i **
0.040
0.005
O.OIO
I 1 i
0.015
-r—r
0.020
0.025
0.030
0.035
milligrams/liter
Figure V-24a. Cumulative Distribution of Cadmium
REFERENCE VALUE
100
95
90
80
75
70
65
60
milligrams/liter
NOTE: Cumulative distribution is plotted on semilog paper. Base: 21,974,000 households.
-------
V - 170
Regional Variation in Cadmium in US Rural Household Supplies
Figure V-2
0.040
0.010
0.015
0.030
milligrams/liter
-------
V - 171
Regional Variation in Cadmium (continued)
Figure V-24d. South
REFERENCE VALUE
65-|
64-
5
4 -
3-
2-
82.7%
Ln
'i 1 1
0 0.005
17.3%
' I
0.015
£
Lowest value:
Highest values
Median:
Interval width:
< 0.002
0.043
< 0.002
0.001
Number of households: 9,291,000
XL
1—i—i—i—i—i—i—i—f—i—|—r-
0.020 0.025 0.030
i I i f » » I i
0.035 0.040
-P
0.010
milligrams/liter
43-i
42-
e
10-
9
8
7
6-
5-
4
3
2-
Figure V-2^e. West
REFERENCE VALUE
72.9%
L
27.1%
0.005 O.O'IO
Lowest value:
Highest value:
Median:
Interval width:
< 0.002
0.040
0.003
0.001
Number of households: 2,777,000
0.035 ' ' 0.0 40
0.015 0.020
milligrams/liter
0.025
0.030
-------
V - 172
other rural areas (17.3 percent) than it was in small rural places (7.3 percent).
Supplies exceeding the reference value were much more common among households
served by community or intermediate systems than among those using individual
systems. Specifically, the reference value was exceeded in only 7.9 percent of
households served by individual systems, as opposed to 26.9 percent of households
served by intermediate systems and 21.2 percent of households served by commun-
ity systems.
Cadmium compounds enter the water from a number of sources.^ Despite
the many possible sources, one in particular is consistent with the NSA data—
namely, the technological features of water transmission and distribution facilities.
Cadmium frequently is an impurity in zinc, lead, and complex copper ores.^ The
metal also can be present as an impurity in zinc-galvanized pipes, and it is used
78
itself as an anti-corrosion coating for some metallic parts. Cadmium also is used
in the formulation of silver-brazing alloys which are used to join iron, copper,
79
nickel, and silver-base alloys. The metal sometimes is used as an alloy with
copper. Furthermore, cadmium compounds are used as stabilizers in some plastic
products, and the metal has been shown experimentally to leach out into water in
gn
black polyethylene pipes.
Water transmitted through the pipes of community systems, particularly in
the West where long-range piping often is necessary, is in contact with transmis-
sion equipment for longer periods of time than is water in individual systems.
Thus, there would be more opportunity for exposure to cadmium compounds in the
equipment. Consistent with this consideration, supplies exceeded the reference
value considerably more often among households served by intermediate and
community systems than among those served by individual systems. Further,
values in excess of the reference value also were most frequent in the West, where
the proportion of households with community systems was substantially larger than
in other regions (see Chapter IV).
-------
V - 173
As to overall health implications, the NSA results indicated the need for
further assessment of cadmium contamination since a sizable proportion of rural
households were over the NSA reference value in every region except the
Northeast. As to immediate risk, even the highest recorded values were far below
those which have caused direct toxic effects in human beings. Furthermore, even
the largest concentrations were considerably lower than those which have been
81
associated with chronic effects. On the other hand, the potential long-term
cumulative influence of cadmium ingestion on the scale occurring in the rural US is
82
uncertain.
Mercury
Mercury is one of the least abundant elements, and its presence in surface
source water is associated mainly with industrial discharges. Historically, the
toxicity of mercury became apparent when workers in the felt hat industry became
mentally unstable after being exposed to the metal in their trade.
Mercury is present in soil and rock. The concentration usually averages
only about 0.05 parts per million, but it can range from one to 30 parts per million
in some geological areas with sediment and volcanic rock containing large amounts
83
of cinnabar (HgS). Weathering of rocks and deposits may contribute to the
amount of mercurial compounds in the sediment of streams and lakes. Ground
water, depending upon its aggressiveness and the geology of its surroundings, may
pick up the compounds.
In addition to the natural sources of mercury, inorganic mercurial salts
have been discharged by industry. The main concern has been that the natural or
industrial inorganic mercurial compounds can be converted by naturally occurring
microorganisms into organic methylmercury compounds, which are of greater
84
hazard to human beings. The greatest potential hazard is consumption of fresh-
water fish which contain the methylmercury compounds. Contaminated fish flesh
-------
V - 174
may contain concentrations of methylmercury which far exceed allowable federal
limits.
On the other hand, there has been no historical indication that mercurial
compounds in US drinking water supplies are present in sufficient quantities or in
specific chemical forms which pose a threat to hurrtan health. (This is the
conclusion of the NRC in reviewing information about the subject.)
In formulating the interim primary MCL for mercury in drinking water, the
EPA took a more cautious approach. The background statement for the regulation
begins: "Environmental exposure of the population to mercury and its compounds
poses an unwarranted threat to man's health. Since conditions indicate an
increasing possibility that mercurials may be present in drinking water, there is a
85
need for a guideline that will protect the health of the water consumer."
The interim primary MCL for mercury is 0.002 milligrams per liter of
water. That is the reference value used in the NSA.
— Mercury levels in rural supplies
As with cadmium, the reference value for mercury was exceeded in a far
larger proportion of rural households than had been anticipated. Fully 24.1 percent
of households had more than the reference value level of 0.002 milligrams of
mercury per liter (Figures V-25, V-25a). In fact, the median value for rural
supplies was 0.001, a level relatively close to the reference value. Furthermore,
the mean value for rural supplies was 0.003, or about one and one-half times the
reference value. The largest recorded concentration was 0.25 milligrams per liter.
The median value for mercury concentrations was 0.001 milligrams per
liter across ail regions and across all other NSA groupings, indicating the
pervasiveness of the metal in water supplies. The proportions of households above
the reference value varied considerably from region to region, however (Figures
V-25b through V-25e). Specifically, proportions exceeding the reference value for
-------
V - 175
Figure V-25
Mercury In US Rural Household Supplies
REFERENCE VALUE
75.9% 24.1%
55-i
50 -
| 15-1
o- 10-
5-
J1
±1
L
Lowest value:
Highest value:
Median;
Interval width:
< 0.0002
0.250
0.001
0.0002
Number of households: 21,974,000
-P—r rcf
#¦
0 0.0020 0.0040 0.0060 0.0080 0.0100
milligrams/liter
I r
0.0120
0.0140
0.0160
10 0-1
90
80^
70-
> "
| ~
\ 60-
!J
>
i 40
20:
10-
Figure V-25a. Cumulative Distribution of Mercury
REFERENCE VALUE
-•I-
75.9%
>—r—i-
T
O
o
o
o
r ! !
24.1%
I •
T
o*o
o o
9 °
o o
.. i:
l:. :t:
T+i-
TTT
i:;:
,:i.;
'Mir
-H
• o
o
o
o
o
o
o
o
o
o
o
CM
o
o
o
o
o
o
o
CO
o
o
o
3 o
o o
9 o
o r\
O
O
CM
O
o
O
o
o
§§
® o
o o
milligrams/liter
o o
o o
O *5
CO CM
6 o
NOTE: Cumulative distribution is plotted on semilog paper. Base: 21,974,000 households.
-------
V - 176
Regional Variation in Mercury in US Rural Household Supplies
Figure V-25b. Northeast
REFERENCE VALUE
78 0%
75-
70-
10-
5-
22.0%
Lowest value:
0.0005
Highest value:
o.
-------
V - 177
Regional Variation in Mercury (continued)
Figure V-25d. South
REFERENCE VALUE
75.0%
65-1
60-
10-
5-
Ln
25.0%
n p n
Lowest value:
< 0.0002
Highest value:
0.016
Median:
0.001
Interval width:
0.0002
Number of households:
9,291,000
A
P p p
t 1 1 1 r—
0 0 0020 0.0040 0.0060 0.0080 0.0100 0.0120 0.0140 0.0160
milligrams/liter
Figure V-25e. West
4 0-i
35-
25-
20-
REFERENCE VALUE
89.6%
10.4%
Lowest value: < 0,0002
Highest value: 0.012
Median: 0.000
Interval width: 0.0002
Number of households: 2,777,000
t 1 1—'i " i"—r—i 1 1 rf
0.0020 0.0040 0.0060 0.0080 0.0100 0.0120
milligrams/liter
-------
V - 178
mercury were 22.0 percent in the Northeast, 31.8 percent in the North Central,
<
25.0 percent in the South, and 10.4 percent in the West.
As to patterns among the other NSA groupings, about one-fourth of
households surpassed the NSA reference value both inside and outside SMSAs, in
each size-of-place category, and regardless of the size of the supply system.
Exceptions were a lower rate in large-community households (16.2 percent) and a
higher rate in intermediate-system households (36.0 percent).
The health implications of the NSA findings are cause for concern. As
noted above, the main threat from mercury is from the methylmercuric (organic)
form. Mercury content was determined in NSA samples by flameless (cold vapor)
atomic absorption spectrophotometry. The method is based on the specific light-
absorbing characteristics of the metal being studied (in this case, mercury), and it
does not determine the original chemical form of the metal. Thus, the NSA results
tell us only that mercury was present in certain amounts in many US supplies. The
results do not tell us whether the metal was in the highly toxic methylmercuric
form, or whether it was in less dangerous inorganic forms. In this regard, EPA
authorities have assumed that less than 0.1 percent of the mercury in water is in
86
the toxic organic form. On the other hand, researchers have pointed out that
natural mechanisms exist in the environment which convert inorganic mercurial
compounds into organic ones, so that the presence of mercury in whatever form
87
must be taken seriously.
According to Swedish authorities cited by the NRC, clinical manifestations
of mercurial poisoning may occur in some persons who consume 0.3 milligrams of
88
methylmercury per day. In the NSA, 0.5 percent of all rural households had
levels of 0.25 milligrams of mercury per liter of water. Since people are assumed
89
by the EPA to drink an average of two liters of water per day, persons in those
93,000 households faced potential direct health consequences if methylmercury
-------
V - 179
were present, especially if they also ate food, such as fish, containing methyl-
mercury.
As to the sources of the contamination, the NSA findings did not provide
clear indications for further investigation. Mercury contamination was pervasive
regardless of household location with respect to SMSAs or size of place, and
regardless of the size of the water system. Mercury is a particularly difficult
element to maintain in water specimens from the time of collection until assay in
the laboratory. The difficulty usually results in inaccurately low readings. It is not
known whether such a bias exists in the NSA mercury data, but the possibility
indicates even more strongly the need for reinvestigation.
Chromium
Only trace amounts of this metal are found in US waters because chromium
compounds are not particularly soluble. The main source of the substance is
industrial wastewater. Small amounts of chromium are essential to glucose
metabolism in human beings. Excessive levels of chromium are poisonous, but the
toxicity depends on the chemical form of the compound. Sufficient amounts of
hexavalent chromium produce gastrointestinal bleeding, and inhaled industrial
chromate may cause cancer of the respiratory tract. Trivalent chromium, on the
other hand, is relatively nontoxic and is the form essential in the human diet.
These complexities make it difficult for public health authorities to
establish meaningful limits for chromium in drinking water. The European
standards of the World Health Organization, as well as the Japanese standards, set
the acceptable limit at 0.05 milligrams of chromium per -liter of water. The
standard is specifically for hexavalent chromium rather than for total chromium,
however. The EPA, on the other hand, has set a national interim primary standard
of 0.05 milligrams of total chromium per liter. The NSA assays, in water
specimens in the Group II subsample, were for total chromium content as
-------
V - 180
determined by atomic absorption spectrophotometry, with atomization of the
specimen in a flame.
The NRC observes: "The present interim drinking water standard of 0.05
milligrams per liter is less than the no-observed-adverse-health-effect level.
Consideration should be given to setting the chromium limit in terms of the
hexavalent form. Extensive work is urgently needed to establish the role of dietary
chromium with regard to atherosclerosis and glucose metabolism as well as its
90
possible carcinogenic effects at low levels in lifetime feeding studies."
It was decided to utilize the MCL value of 0.05 milligrams per liter as the
NSA reference value.
— Chromium levels in rural supplies
Only trace amounts of chromium were present in rural supplies. The
highest value was 0.012 milligrams per liter of water—only one-fourth of the MCL-
based reference value of 0.05 milligrams per liter. The mean and median in rural
US supplies both were recorded at the minimum level of detection, less than 0.005
milligrams of chromium per liter of water.
The means and medians did not vary from region to region or in any other
NSA grouping. In view of this and the very low concentrations of chromium
encountered in rural supplies, graphic plots of the distributions have been omitted
from this report. On the basis of the NSA findings, chromium did not represent a
health problem in rural water supplies.
Barium
An alkaline earth metal, barium occurs in trace amounts in most surface
waters. Barium usually is in the form of dissolved barium sulfate in natural waters,
however, and because of the low solubility of that compound, concentrations of
barium ions are typically low. In sufficient amounts (0.8 to 0.9 grams), barium
-------
V - 181
chloride can be a deadly poison because it overstimulates the muscles, especially
the heart muscles. However, the EPA reports that: "No study appears to have
been made of the amounts of barium that may be tolerated in drinking water or of
effects from prolonged feeding of barium salts from which an acceptable water
91
guideline may be set."
The national interim primary MCL of one milligram of barium per liter of
water is based on extrapolation from effects of industrial exposure to dusts of
soluble barium salts. The NSA reference value also was one milligram of barium
per liter of water.
— Barium levels in rural supplies
Trace amounts of barium compounds are frequent in water, and small
amounts of the metal were found in most rural supplies. Despite the prevalence of
the substance, concentrations exceeded the reference value in only 0.3 percent of
all households (Figure V-26). The level in those households was 1.35 milligrams of
barium per liter—35 percent greater than the reference value.
The few household supplies which exceeded the reference value occurred in
the South (Figures V-26a through V-26d). Because the number of households
exceeding the NSA reference value was small, and since the households all were in
the South, analysis by other than regional groupings was not reliable.
Although values in excess of the reference value require attention, those
92
for barium in the NSA did not appear to be large enough to pose a health risk.
Silver
Large amounts of colloidal silver can be fatal, but only very small amounts
of any silver compounds are found in US waters. Trace amounts in drinking water
may come from natural or industrial sources. The NRC advises that: "Since silver
ion has not been detected in water supplies in concentrations greater than half the
-------
V - 182
Figure V-26
Barium in US Rural Household Supplies
70-1
69-
9-
8-
1 7-
c.
©
I 6-
o
S 5-
u
<5
Q- 4 -
3-
2-
I-
REFERENCE VALUE
997% 0.3%
LnflJ
0.20
T
0.40
JEL
0.60 0.80
milligrams/liter
Lowest values
Highest value:
Median:
Interval width:
Number of households:
< 0.03
IA
0.2
0.02
21,974,000
I 00
120
1 T
-------
V - 183
Regional Variation in Barium in US Rural Household Supplies
Figure V-26a. Northeast
6-
REFERENCE VALUE
100.0% 0.0%
run
"1 P
0.20 0.40 0.60
milligrams/liter
0.80
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
0.2
0.5
0.2
0.02
3,693,000
1.00
Figure V-26b. North Central
13-
REFERENCE VALUE
100.0% 0.0%
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
0.1
0.7
0.2
0.02
6,213,000
0.20 0.40 0.60
milligrams/liter
0.80
1.00
-------
V - 184
Regional Variation in Barium (continued)
Figure V-26c. South
3
x
81-
80-
tfl \
| 10-
"o
s \
a>
a. 4—
100.0% 0 0%
Lowest value: 0.1
Highest value: 0.5
Median: 0.2
Interval width: 0.02
Number of households: 2,777,000
3-
2-
0.20 0.40 0.6 0
milligrams/liter
0 80
.00
-------
V - 185
no-observed-adverse-health-effect level, regulation of its concentration as a
93
primary standard would appear to be unnecessary."
Silver compounds can be used as disinfectants in water, however. The EPA
thus has set a national interim primary MCL of 0.05 milligrams of silver per liter
of water. That MCL was the reference value used in the NSA.
— Silver levels in rural supplies
Silver was more prevalent than anticipated on the basis of findings in other
Q(i
surveys. In addition, the proportion of larger values was greater than antici-
pated. That is, 4.1 percent of US rural households surpassed the reference value,
which was a considerably larger proportion than was found in several studies of
95
public water supplies. The largest NSA concentration was 0.1 milligrams per
liter—twice the NSA reference value (Figure V-27).
Median concentrations of the metal in ail regions of the US were close to
the national median of 0.028 milligrams per liter, indicating that silver was found
in similar quantities in household supplies regardless of region (Figures V-27a
through V-27d). One possible exception was in the West, where the median
concentration was 0.021, slightly lower than in other regions. Similarly, the
proportion of households above the reference value was slightly smaller in the West
than in the other three regions.
Median silver concentrations and the proportions of households exceeding
the reference value did not vary notably in either the SMSA/nonSMSA or size-of-
place comparison. In the size-of-system grouping, however, a larger proportion of
households with individual systems exceeded the reference value (7.1 percent) than
did households served by either intermediate (3.4 percent) or community systems
(2.1 percent).
The potential health consequences of the NSA findings are difficult to
assess. Even the largest concentrations in rural households are far below those
-------
V - 186
Figure V-27
Silver in US Rural Household Supplies
60-
55-
35-
30-
REFERENCE VALUE
95.9% 4.1%
Lowest value:
Highest values
Medianr
Interval width:
Number of households:
< 0.020
0.100
0.02S
0.01
21,974,000
•0.02 0.04 0 05 0 06
milligrams/liter
0 08
0 10
9
-------
V - 187
Regional Variation in Silver in US Rural Household Supplies
Figure V-27a. Northeast
REFERENCE VALUE
85-,
80-
5-
952% 4.8%
—I 1 1—
0 0.01 0 03 0.05 0.07
milligrams/liter
Lowest value: < 0.030
Highest value: 0.070
Median: 0.031
Interval width: 0.01
Number of households: 3,693,000
Figure V-27b. North Central
REFERENCE VALUE
50-
45-
5-
96.3% 3.7%
0 0 01
1
0.03
0.05 0 07
milligrams/liter
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
< 0.020
0.080
0.026
0.01
6,213,000
-------
V - 188
Regional Variation in Silver (continued)
Figure V-27c. South
REFERENCE VALUE
7 5-
70-
2 15-
10-
5-
95.2%
0.01
003
4.8%
Lowest value:
Highest value:
Median:
Interval width:
Number of households:
< 0.020
0.100
0.030
0.01
9,291,000
0 05 0 07
milligrams/liter
0.09
Figure V-27d. West
REFERENCE VALUE
90-,
85-
15-
10-
5-
97 9%
0.01
0 03
2.1%
Lowest value: < 0.020
Highest value: 0.080
Median: 0.021
Interval width: O.Ol
Number of households: 2,777,000
0.05 0.07
milligrams/liter
-------
V - 189
which are considered possibly toxic to human beings. Chronic ingestion of large
amounts of silver can produce argyria, a condition which causes unsightly blue-gray
discoloration of the skin. Estimates from studies of industrial exposure to silver
show that the condition develops after gradual accumulation of one through five
grams of silver in the human body. The NRC estimates that, based on a 50 percent
retention rate for silver taken into the body, it would take 55 years for a person to
consume enough silver in a supply with 0.05 milligrams per liter (the NSA reference
96
value) to acquire argyria. Thus, even at the rural households above the reference
value, the consequences appeared to be limited to possible but uncertain long-term
effects.
Summary of inorganic constituent findings
Inorganic substances studied in NSA specimens ranged from those largely
with aesthetic effects, such as manganese, to those predominately with health
effects, such as lead. Certain of the constituents have received particular
attention from investigators in the past, and it was anticipated that they might be
associated with significant findings in the NSA. Among those constituents were
calcium and magnesium, key elements in water hardness; nitrates, substances
which pose a special health risk for infants; sulfates, substances which may make
water distasteful and even cause diarrhea; and iron and manganese, elements which
cause aesthetic and economic problems.
Calcium and magnesium were present in sufficient amounts to produce
moderately hard water in rural US supplies, but the substances themselves were not
implicated in any direct health effects by the NSA findings. Nitrates were not
discovered in large concentrations in rural supplies—fully 97.3 percent of rural
households were below the NSA reference value. Sulfates were potential problems
in 4.0 percent of rural households, but more often for aesthetic rather than for
serious health reasons. Iron and manganese concentrations posed aesthetic
-------
V - 190
problems in a number of rural households, but about eight out of ten rural
households met the N5A reference value for each metal.
Although households which exceeded the NSA reference value for any of
these common constituents faced potential problems which required assessment
and attention, the difficulties were not beyond the scope anticipated.
There were potentially important problems posed by several other inor-
ganic substances. Those substances, all heavy metals, were lead, cadmium, and
mercury. Lead was studied in all NSA sample households, and on the basis of the
results, 16.6 percent of all rural American households (numbering 3.6 million) were
above the NSA reference value for lead. Cadmium and mercury were studied in
NSA subsample households, and on the basis of the results, 16.8 percent and 24.1
percent of US rural households, respectively, exceeded the NSA reference value.
The possible health implications of these findings are important. The three
metals have different physiological effects, but the effects are known to be
potentially serious, as is detailed in the separate reports on each of the heavy
metals, above.
Study of a number of other inorganic substances in NSA subsample
households showed other persistent, but less consequential, problems. Supplies at
13.7 percent of rural households surpassed the NSA reference value for selenium,
and although the concentrations were not large enough to pose health threats, some
were large enough to suggest the need for caution in adjusting federal selenium
standards, as has been discussed by the NRC. About 4 percent of rural households
had values beyond the reference value for silver, a situation which posed no
immediate health threat but which involved a substantial proportion of rural
households.
As to some of the other NSA constituents, amounts of arsenic, chromium,
and barium in rural households generally were very small and posed few health
-------
V - 191
concerns. Levels of fluoride also generally were well within the NSA reference
value.
In retrospect, the NSA findings point up the need to evaluate water quality
in terms of all of the components of the supply—from the source water to the
tap—rather than focusing too sharply on only one component, such as the source
water. For example, the high levels of cadmium in the NSA are difficult to explain
without reference to possible contamination from elements within the transmission
and distribution system, such as the pipes. The same is true for the high levels of
lead, and it may be true for the high levels of mercury. It has been customary, of
course, to assess the possible contribution of lead pipes and lead-base solder to lead
contamination in drinking water. This concern with possible contamination from
transmission technology should be extended to assessment of other substances as
well, including cadmium, mercury, sodium, and silver. Furthermore, this assess-
ment must include both old and new technology. For example, older metal water
pipes may be suspected as possible causes of heavy-metal contamination of
drinking water, but plastic pipes must be assessed as well. The possibility of
cadmium leaching out into water from plastic pipes has been discussed above, but
the same possibility also exists for lead and mercury, a concern put forward by the
97
World Health Organization.
ORGANIC CONSTITUENTS
Whereas public health concern about inorganic substances dates back many
years in the US, fears about organic materials in drinking water are more recent.
0
In its official statement entitled Interim Primary Drinking Water Regulations—
Control of Organic Chemical Contaminants in Drinking Water, February 9, 1978,
the EPA presented its view of the situation. Several passages are relevant as
background information:
-------
V - 192
"More than 700 specific organic chemicals have been identified in
various drinking water supplies in the United States. These com-
pounds result from such diverse sources as industrial and municipal
discharges, urban and rural runoff, and natural decomposition of
vegetative and animal matter, as well as from water and sewage
chlorination practices. Compositions and concentrations vary from
virtually nil in protected ground water to substantial levels in many
surface waters and contaminated ground waters.
"Organic chemical contaminants in drinking water can be divided
into two major classes: those of natural origin and those of synthetic
origin. The natural substances represent by far the greatest portion
and consist primarily of undefined humus and fulvous materials and
others produced by normal organic decomposition or biotic transfor-
mation and are not known to be harmful in themselves.
"The synthetic chemicals in water can be subdivided into two
groups. The first group consists of those chemicals that result from
water treatment practices (e.g. trihalomethanes). Recent EPA
studies indicate that, except for certain cases, the trihalomethanes
constitute the largest portion of the identifiable synthetic chemicals
in drinking water. Unlike other synthetic chemicals, chloroform and
other trihalomethanes are formed during the treatment process.
They are thus found in virtually every drinking water supply that is
disinfected with chlorine, and not uncommonly at concentrations of
several hundred parts per billion (ppb or micrograms per liter).
"The second group of synthetic chemicals consists of those
chemicals introduced as a result of point and nonpoint sources of
pollution. Nationally, both surface water and to a lesser degree
ground waters are contaminated with a variety of these pollution-
related synthetic organic chemicals ranging generally from the lower
molecular weight halogenated hydrocarbons and monocyclic aromatic
compounds to higher molecular weight pesticides, polycyclic aroma-
tic compounds, and pesticide-like compounds.
"These classes of compounds have been found in drinking water
using gas chromatography or gas chromatography/mass spectroscopy.
However, the large bulk of organic matter (primarily natural products
but also higher molecular weight synthetics) in water is not amenable
to detection by these commonly used methods. Those organic
contaminants which have been identified in drinking water constitute
only a small percentage of the total amount of organic matter
present." 98
Generally, scientific knowledge has developed slowly in defining the
sources and consequences of organic materials in drinking water and drinking water
sources. As a result, the EPA's strategy has been to establish regulations at a
graduated pace while trying to complete the required surveys and laboratory
research needed to support the regulations.
-------
V - 193
At the time of the NSA survey, US interim primary MCLs had been
established for a limited number of organic chemicals. In addition, federal
guidelines suggested that permissible limits of five well-known chlorinated hydro-
carbon insecticides—DDT, aldrin, dieldrin, chlordane, and heptachlor—should be
established on the basis of certain specified research reports which assessed the
99
health hazards of the materials.
The organic chemicals for which MCLs had been established at the time of
the NSA were four other chlorinated hydrocarbon insecticides and two chloro-
phenoxy herbicides. The MCLs and NSA reference values are presented below.
The chlorinated hydrocarbon insecticides were endrin, lindane, methoxychlor, and
toxaphene. The chlorophenoxy herbicides were 2,4-D and 2,4,5-TP. Levels of
these six substances were assessed in water specimens from the Group n NSA
subsample. In comparing the findings for the six organic constituents with the
findings for inorganic constituents, it is necessary to note that the former are
reported in micrograms per liter (parts per billion), the latter generally in
milligrams per liter (parts per million).
— Chlorinated hydrocarbon insecticide levels in rural supplies
Although NSA subsample specimens were examined for four chlorinated
hydrocarbon insecticides (endrin, lindane, methoxychlor, and toxaphene), only two
of the chemicals—lindane and methoxychlor—were discovered in any of the
specimens. Since the four insecticides were found so rarely, and since the findings
for lindane and methoxychlor were so similar, the results for all four chemicals are
discussed here in one section rather than being considered separately.
The NSA reference values corresponded to the interim primary MCLs. The
reference values were equivalent to 0.2 micrograms of endrin per liter of water,
four micrograms of lindane per liter of water, 100 micrograms of methoxychlor per
liter of water, and five micrograms of toxaphene per liter of water.
-------
v - m
Endrin and toxaphene. In the NSA laboratory work, analytic reporting
procedures were designed to indicate endrin only in concentrations exceeding 0.00S
micrograms per liter, and to report toxaphene only in concentrations exceeding
0.17 micrograms per liter. Neither chemical was reported above those values in
any of the NSA households. It is possible, of course, that some rural supplies had
minute concentrations of endrin or toxaphene which were lower than the detection
levels. Even so, those levels would have been considerably less than the reference
values since the detection levels were so low. On the basis of the NSA findings,
then, neither endrin nor toxaphene posed a health threat in rural water supplies.
Lindane. Lindane was detected in 1.6 percent of all rural household water
supplies—about 347,000—but the largest concentration reported was only 0.08
micrograms per liter, one-fiftieth of the reference value.
The largest number of lindane-contaminated supplies (192,000 households)
were located in the West, at a concentration of 0.006 micrograms per liter. The
other contaminated supplies were in the South, where an estimated 64,000
households had the highest concentration found in the NSA (0.08 micrograms per
liter). The chemical was not discovered in North Central or Northeastern
households.
All of the lindane-contaminated supplies were in nonSMSA households
located in other rural areas. The largest number of households (208,000) were
served by community systems. A total of 68,000 households were served by
intermediate systems, and the same number by individual systems. The largest
concentration of lindane (0.08 micrograms per liter) was found among households
using individual systems.
Methoxychlor. Only 1.0 percent of all rural household supplies had
detectable amounts of methoxychlor. Those 224,000 supplies had 0.09 micrograms
of methoxychlor per liter, far less than the reference value.
-------
V - 195
All of the supplies with methoxychlor contamination were in the West. The
households all were outside of SMSAs, located in other rural areas, and served by
community systems.
Health considerations for lindane and methoxychlor. On the basis of the
NSA findings, neither methoxychlor nor lindane posed health threats in rural water
supplies. Both chemicals are poisons, however, and their presence even in small
amounts clearly is undesirable. Despite the predominantly open-country aspect of
the location of the households with methoxychlor contamination, all were served by
community systems. In contrast, households with lindane contamination were
served by individual, intermediate, or community systems.
— Chlorophenoxy herbicide levels in rural supplies
As was true for the insecticides, the two herbicides studied in the NSA
presented no problem to rural supplies: neither chemical was detected in any of the
NSA subsample specimens. Analytic procedures were designed to detect 2,4-D only
in concentrations exceeding 0.01 micrograms per liter, and to detect 2,4,5-TP only
in concentrations exceeding 0.1 micrograms per liter. Interim primary MCLs for
the chemicals were equivalent to 100 micrograms per liter for 2,4-D and ten
micrograms per liter for 2,4,5-TP; these levels also were the NSA reference values.
Again, it was possible that household supplies had minute concentrations of the
herbicides which were below the laboratory detection levels, but those amounts
would have been far less than what was deemed unacceptable according to the
reference values. On the basis of the NSA findings, then, neither 2,4-D nor 2,^,5-
TP posed a health threat in rural water supplies.
Summary of organic constituent findings
Organic constituents studied in the NSA were limited to four chlorinated
hydrocarbon insecticides and two chlorophenoxy herbicides. Of these six
-------
V - 196
substances, only two were detected at all in NSA households. Those two substances
were lindane—found only in small proportions of households in the West and
South—and methoxychlor—found only in a small proportion of households in the
West. None of the values for the substances exceeded or even closely approached
the respective NSA reference values.
RADIOACTIVITY
With the nuclear age in development, there is an increased likelihood of
exposure to radioactivity from various military, industrial, medical, and pharma-
ceutical sources. Radiation from these sources is added to that from natural
sources to increase our total exposure.
The potential hazard from drinking water, except in special situations,
appears to be slight, however. According to the NRC: "The radiation associated
with most water supplies is such a small proportion of the normal background to
which all human beings are exposed that it is difficult, if not impossible, to
measure any adverse health effects with certainty. In a few water supplies,
however, radium can reach concentrations that pose a higher risk of bone cancer
for the people exposed."
For many of the other contaminants studied in the NSA, there are levels of
exposure which normally pose no health hazard. Indeed, in proper amounts, and in
correct chemical composition, some constituents found in water may have health
benefits. Radiation, however, has no known threshold below which health risks
disappear. With small doses, however, the estimated health risks for normal
individuals becomes very low.
The EPA has established maximum allowable contamination levels for two
broad categories of radioactive substances: those of natural origin and those most
likely created by man's activities. In the former category, there are MCLs for
gross alpha particle activity, for radium-226, and for radium-228. In the latter
-------
V - 197
category, limits are placed on average annual exposure to beta particle and photon
radioactivity from artificially created radionuclides such as tritium, strontium-90,
and iodine-131.
While both naturally occurring and man-made radionuclides have alpha,
beta, and gamma emissions, the former are generally more important because of
their alpha activity, and the latter for their beta or beta-gamma activity. Hence
the MCL for groundwater, where naturally occurring radionuclides are expected to
predominate, refers to alpha activity. Surface waters, on the other hand, can be
affected by all types of radioactivity and are subject, under certain conditions, to
both gross alpha and gross beta MCLs.
The contributors to alpha activity in groundwater are expected to be
natural deposits of the uranium decay series (beginning with uranium-238) or the
natural thorium decay series (beginning with thorium-232). In the United States,
uranium deposits are generally more prevalent than thorium sources, especially in
the South and West. As a result, more radium-226 than radium-228 is usually found
in US groundwaters. Radium-228, of the thorium series, may be more prevalent in
other nations, however, since thorium tends to be more common than uranium
throughout the world. Furthermore, radium-228 is included in the gross alpha
standard, even though it is more important as a beta emitter. The reason for
including radium-228 is that its radioactive decay daughters are alpha emitters.
Chemically, they are similar to calcium so, when ingested, they tend to accumulate
in bone tissue.
Interpretation of the federal interim primary standard for gross alpha
radiation can best be understood by referring to Figure V-28, which was prepared
by the EPA Environmental Monitoring and Support Laboratory in Las Vegas. At the
right of the flow chart, one sees that a specimen exceeds the MCL-based NSA
reference value if gross alpha radiation is greater than fifteen picocuries per liter,
providing that the radioactive contributions of uranium and radon are excluded. In
-------
V - 198
Figure V-28
Flow Chart Describing Gross Alpha Radiation Reference Values
MEASURE
{OSS ALPHA
ALPHA
>5 pCi/
ALPHA
NO
NO
>15
YES
YES
measure;
RADON a URANIUM
MEASURE
/ALPHA \
/ MINUS \
ADON a URANIUM
ALPHA /
\ >15 pCj/1 /
226
NO
NO
YES
MEASURE
Ra-228
YES
PLUS
RA-228
NO
YES
WITHIN
VALUE
VALUE
Adapted from a flow chart for determining gross alpha MCL compliance produced by the EPA Enviromentai
Monitoring Systems Laboratory of Las Vegas, Nevada.
-------
V - 199
the NSA, however, specimen preparation involved evaporation. As a result, radon,
a gas, escaped before its radiation could be measured. With sufficient time, radon
could "grow" back as a natural result of radiation decay processes, but the analyses
were not delayed and the presence of radon radiation thus was not allowed for.
Nevertheless, this shortcoming was not expected to create a serious bias in the
findings. Therefore, any NSA specimen which produced a reading of more than
fifteen picocuries per liter, after discounting uranium radioactivity, was deemed to
be above the MCL-based NSA reference value. In addition (see left portion of flow
chart), if a specimen had a gross alpha reading which was less than fifteen
picocuries per liter but more than five picocuries, it still exceeded the NSA
reference value if the combined radiation values for radium-226 and radium-228
exceeded five picocuries per liter.
Although the NSA reference value was based on the federal interim
primary MCLs, it is necessary to recall that, as explained earlier in this chapter,
there is a distinction between the requirements for determining compliance with
the federal regulations and the procedures followed in the NSA. In particular,
compliance with the regulations must be determined on the basis of the average of
several findings, not just on one as in the NSA. Specifically, compliance with
radioactivity standards must be based on the analysis of an annual composite of
four consecutive quarterly samples, or on the analysis of the average of results for
four samples obtained at quarterly intervals.
Another qualification is that low-level radiation is difficult to measure
. reliably. For example, there are a variety of factors which can interfere with
measurement of low-level gross alpha radiation. Most notably, large concentra-
tions of total dissolved solids in water specimens can absorb or otherwise alter the
radiation and thus influence the readings. The effect on the emission energy levels
can also make it difficult to distinguish alpha from beta emission.
-------
V - 200
The standard for beta emissions generally relates to man-made radio-
nuclides. Atmospheric nuclear-weapon testing has produced ubiquitous background
radiation which will continue even though atmospheric testing has been reduced in
the world. In addition, water from nuclear power plants, or from manufacturing
plants using radioactive materials, occasionally contributes to the total dosage in
some surface water sources. Surface water sources thus are more likely than
groundwater to contain manufactured radionuclides. The federal monitoring
criterion for gross beta radioactivity focuses on water sources designated as
contaminated by nuclear facilities and on surface water sources which, though not
designated as contaminated by effluents from nuclear facilities, serve 100,000 or
more people. Each state has the right to require more stringent monitoring.
There were very few NSA specimens which were from water sources that
were subject to the gross beta monitoring criterion. Therefore, an assessment of
compliance was not justified and the particular radionuclides contributing to gross
beta radiation were not generally measured. Nevertheless, gross beta readings
were taken and will be reported in order to provide a baseline measure of the
conditions in rural America.
Since much of the water tested in the NSA was from groundwater sources,
the gross beta levels reported did not necessarily reflect only emissions from man-
made radionuclides. For example, naturally occurring radiation from the uranium
and thorium decay series in groundwater can influence beta radiation. However,
the cost of determining the particular source of the beta radiation put this beyond
the scope of the NSA. Hence, the reported beta emissions in the NSA were caused
by an unknown mix of naturally occurring and man-made beta emitters.
Measures of radioactivity were taken only on specimens from the Group II
NSA subsample; 50 picocuries per liter was the screening level used when initiating
a test for gross beta MCL compliance.
-------
V - 201
— Gross alpha radiation levels in rural supplies
As expected, low levels of gross alpha radiation were found in all rural
supplies, indicating the anticipated effect of ubiquitous, natural radiation sources
in the environment. The levels almost always were within the reference value. In
fact, only 0.5 percent of all rural households (an estimated 115,000) surpassed the
reference value, all of which were in the South and located in other rural areas.
Problem supplies were almost equally divided between SMSA and nonSMSA
households, and equally divided between households with individual and community
systems.
Additionally, OA percent of all rural households (an estimated 96,000) had
readings which were inconclusive but questionable. For these households, total
compliance with the NSA reference value was not demonstrated.
Another 1.5 percent of all rural households (an estimated 319,000) had
readings which were suspicious: readings may have exceeded the NSA reference
value, but the levels of total dissolved solids in these specimens were so great that
the gross alpha readings were unreliable. For these households, noncompliance
with the NSA reference value was not adequately demonstrated.
In summary, 97.5 percent of all rural households were within the gross
alpha radiation reference value. Of the remaining 2.5 percent, only 0.5 percent
clearly exceeded the reference value.
— Gross beta radiation levels in rural supplies
All households in the NSA had gross beta readings below the NSA reference
value of 50 picocuries per liter. The highest reading was 28 picocuries per liter.
The median level for rural America was only 5.36 picocuries per liter, slightly more
than one-tenth of the reference value. Although the data indicated that beta
emissions throughout rural America were well within the reference value, it is
important to recall that about 72 percent of the major household supplies in rural
-------
V - 202
America had groundwater sources, and those sources would not generally be prone
to contamination by man-made radionuclides. Furthermore, the NSA survey was
designed to select representative rural households, not likely sites of contamina-
tion, so that the survey would not necessarily have detected rural supplies which
used surface water containing man-made radionuclides.
Median values for gross beta radiation were quite near the national median
of 5.36 picocuries per liter regardless of households' regional location, SMSA
status, or size-of-place location. Median values also were at the national level
regardless of the size of system serving the households.
— Other radionuclides studied in the NSA
In conjunction with the gross alpha and gross beta screening tests, certain
other radionuclides were to be studied in the NSA, either when warranted by the
screening test results, or as desirable on an occasional basis.
The study list included strontium-S9, strontium-90, cesium-134, tritium,
and iodine-131. However, the substances were assessed so infrequently that no
generalizations about the results were possible.
Summary of radioactivity findings
Rural household supplies showed low levels of both gross alpha and gross
beta radiation. The presence of background gross alpha radiation, in particular,
was not surprising, since it is produced by natural sources commonly found in
groundwater. Despite the prevalence of the radiation, levels of radioactivity were
so low in the NSA that radiation in drinking water was not shown to be a national
problem. These estimates do not rule out potential serious localized problems,
however.
Local geological makeup is probably the best guide for indicating whether a
gross alpha test is justified. Areas with known deposits of radioactive minerals
-------
V - 203
should be particularly monitored. Areas downwind from weapons testing facilities,
or downstream from nuclear facilities, are likely candidates for gross beta
examination.
SUMMARY OF WATER QUALITY STATUS
The material in the preceding sections of Chapter V described the quality
of the water conveyed by rural water supplies in terms of selected constituents.
These substances, considered individually, were also the primary focus of the
summaries that were prepared for each category of water quality properties
(bacterial content, physical or chemical properties, inorganic constituents, organic
constituents, and radioactivity). In this summary, the emphasis shifts away from
the constituents as separate aspects of water quality, and toward the implications
of the results for the entire set of substances.
The first part of this summary, which presents tabulations of the propor-
tions of households exceeding each NSA reference value, provides information on
the total number of constituents with reference values that were exceeded, and
also the number of reference values that were exceeded by various proportions of
rural households. Besides estimating the incidence of constituents that appeared in
rural water supplies in concentrations over the relevant reference values for the
nation, this portion of the summary .also facilitates general comparisons between
(1) regions of the country, (2) SMSA/nonSMSA households, (3) places of different
sizes, and (4) water supply systems of different sizes. The second part of this
summary, in contrast, is restricted to those substances for which primary MCLs
have been established. More specifically, the second section aggregates house-
holds with supplies having constituents in concentrations greater than the level
prescribed by the MCL, and compiles this information for the nation and for each
analytical grouping (region, SMSA/nonSMSA, size of place, and size of system). In
addition to presenting results, each component includes a brief discussion of the
-------
V - 204
rationale used in selecting that particular approach to summarizing the water
quality data, while the latter portion of the section addresses some limitations of
these summaries.
Proportions of households exceeding NSA reference values: Approach 1
As indicated earlier in this chapter, 43 constituents were studied in the
NSA (see Table V-l). Of these substances, only 28 were assigned NSA reference
values (see Table V-2). (The reasons for not assigning reference values for certain
constituents were detailed in the separate sections on the respective constituents
in this chapter.) This portion of the summary is confined to the set of 28
constituents for which NSA reference values were defined.
While entire distributions of constituent concentrations were examined in
the previous analyses, and NSA reference values were used only as descriptive
tools, the focus here is exclusively on the proportions of households exceeding the
NSA reference values. The reference values provide important benchmarks
regarding health, aesthetic, and economic implications associated with the relevant
concentrations of the constituents. Despite these different implications, there is
no attempt in this first summary approach to distinguish among reference values.
That is, no allowance is made for the possibility that exceeding a health-related
reference value may have more serious public consequences than exceeding an
aesthetic-related reference value. In the second approach, presented later,
incidence rates are summarized for households exceeding only health-related
reference values, in order to focus on that one aspect of water quality. Even then,
however, there are various restrictions that must be established for the results to
be interpreted usefully.
-------
V - 205
Incidence of households exceeding NSA reference values
One dimension of the overall national and subnational rural water quality
situation is summarized in Table V-4. In that table are the percentages of rural
households which exceeded the NSA reference values for the relevant constituents.
All of the data were presented in the separate individual-constituent analyses in
preceding sections of this chapter. However, the data previously were not
displayed together as they are in Table V-4.
According to information presented in previous sections of this chapter, a
total of twenty constituents were detected in concentrations that exceeded the
reference values. The constituents which were never found to exceed the
respective reference values at any household were chromium, all six organic
pesticides, and gross beta radioactivity.
In examining Table V-4, it should be recalled that for many of the twenty
constituents, results were available for only a 10 percent subsample of the NSA
sample. Specifically, the following constituents (referred to as Group II constitu-
ents) were analyzed for the subsample only: arsenic, selenium, fluoride, cadmium,
mercury, chromium, barium, silver, endrin, lindane, methoxychlor, toxaphene, 2,4-
D, 2,4,5-TP, gross alpha radiation, and gross beta radiation. All other constituent
results in Table V-4 are based on the full NSA sample.
For the nation as a whole, Table V-4 indicates that the most frequent
problems were caused by bacterial contamination, iron, lead, cadmium, and
mercury. The NSA reference values were exceeded for each of these constituents
in 15.0 percent or more of all rural households.
It is apparent from Table V-4 that there were distinct regional differences
in the results for the twenty constituents found in concentrations over the
reference values. Households in the Northeast exceeded reference values for the
smallest number of substances. Specifically, Northeast households exceeded the
reference values for fourteen of the twenty constituents; in contrast, households in
-------
V - 206
Table V-4
Percentage of Rural US Households Exceeding NSA Reference Values
Total Fecal Standard Color
Coliform Coliform Plate Count (15 standard
(1/100 ml) (0/100 ml) (500/ml) color units)
NATION
28.9
12.2
19.3
2.3
REGION
Northeast
28.3
14.0
10.2
0.5
North Central
24.4
8.0
17.1
3.4
South
31.7
13.9
22.8
2.6
West
30.6
13.6
24.8
1.6
SMSA/NonSMSA
SMSA
18.3
6.8
13.8
1.4
NonSMSA
33.9
14.7
22.0
2.8
SIZE OF PLACE
Large rural
comm unities
17.7
4.9
15.0
1.5
Small rural
communities
19.6
4.0
11.7
5.4
Other rural
areas
31.2
13.8
20.5
2.2
SIZE OF SYSTEM
Community
15.5
4.5
13.9
1.9
Intermediate
43.3
20.2
17.8
1.9
Individual
42.1
19.8
26.6
3.0
-------
V - 207
Table V-4 continued
Estimated
Total
Dissolved
Solids Magnesium Nitrate-N Sulfates
(500 mg/1) (125 mg/1) (10 mg/1) (250 mg/1)
NATION
14.7
0.1
2.7
4.0
REGION
Northeast
5.0
0.0
0.3
0.5
North Central
23.9
0.1
5.8
7.4
South
10.2
0.0
1.3
0.7
West
22.2
0.5
4.0
11.7
SMSA/NonSMSA
SMSA
15.1
0.1
1.7
2.2
NonSMSA
14.5
0.1
3.2
4.8
SIZE OF PLACE
Large rural
communities
15.8
0.0
4.2
2.6
Small rural
communities
17.7
0.0
4.7
7.5
Other rural
areas
14.3
0.1
2.4
3.8
SIZE OF SYSTEM
Community
15.0
0.0
1.6
4.2
Intermediate
13.4
0.4
3.0
1.7
Individual
14.7
0.1
4.1
4.2
-------
V - 208
Table V-4 continued
Iron Manganese Sodium Lead Arsenic*
(0.3mg/l) (0.05 mg/1) (lOOmg/l) (0.05 mg/1) (0.05 mg/1)
NATION
18.7
14.2
14.3
16.6
0.8
REGION
Northeast
North Central
South
West
16.0
28.2
17.0
7.0
16.9
19.9
12.3
4.7
6.0
19.2
14.1
15.0
9.6
10.8
23.1
16.9
0.0
1.8
0.0
2.1
SMSA/NonSMSA
SMS A
NonSMSA
13.8
21.0
9.9
16.3
14.9
13.9
12.9
18.3
0.0
1.2
SIZE OF PLACE
Large rural
communities
9.4
11.4
15.7
15.1
0.0
Small rural
communities
23.3
21.7
17.0
18.1
6.6
Other rural
areas
19.5
14.0
13.8
16.6
0.4
SIZE OF SYSTEM
Community
Intermediate
Individual
7.7
28.7
29.9
7.2
23.3
20.7
15.8
10.3
13.3
17.7
20.5
14.1
0.9
0.0
0.8
-------
V - 209
Table V-4 continued
Selenium* Fluoride* Cadmium* Mercury* Chromium*
(0.01 mg/1) (1.4mg/l) (0.01 mg/1) (0.002 mg/1) (0.05 mg/1)
NATION
13.7
2.5
16.8
24.1
0.0
REGION
Northeast
North Central
South
West
0.0
25.7
2.1
41.3
0.0
1.8
2.7
6.2
1.6
20.7
17.3
27.1
22.0
31.8
25.0
10.4
0.0
0.0
0.0
0.0
SMSA/NonSMSA
SMSA
NonSMSA
14.4
13.3
1.6
2.9
21.4
14.3
21.5
25.5
0.0
0.0
SIZE OF PLACE
Large rural
communities
16.5
2.9
19.8
16.2
0.0
Small rural
communities
6.6
6.6
7.3
27.6
0.0
Other rural
areas
14.0
2.1
17.3
24.6
0.0
SIZE OF SYSTEM
Community
Intermediate
Individual
12.5
21.7
13.6
2.9
6.7
0.8
21.2
26.9
7.9
23.3
36.0
22.3
0.0
0.0
0.0
-------
V - 210
Table V-4 continued
Barium* Silver* Endrin* Lindane* Methoxychlor*
(1.0 (0.05 (0.0002 (0.004 (0.1
mg/1) mg/1) mg/1) mg/1) mg/1)
NATION 0.3 4.1 0.0 0.0 0.0
REGION
Northeast
0.0
4.8
0.0
0.0
0.0
North Central
0.0
3.7
0.0
0.0
0.0
South
0.7
4.8
0.0
0.0
0.0
West
•
0.0
2.1
0.0
0.0
0.0
SMSA/NonSMSA
SMSA
0.0
5.1
0.0
0.0
0.0
NonSMSA
0.4
3.6
0.0
0.0
0.0
SIZE OF PLACE
Large rural
•
communities
0.0
3.3
0.0
0.0
0.0
Small rural
communities
0.0
3.6
0.0
0.0
0.0
Other rural
areas
0.3
4.2
0.0
0.0
0.0
SIZE OF SYSTEM
Community
0.0
2.1
0.0
0.0
0.0
Intermediate
0.0
3.4
0.0
0.0
0.0
Individual
0.8
7.1
0.0
0.0
0.0
-------
V - 211
Table V-4 continued
Toxaphene* 2,4-D* 2,4,5-TP* Gross Alpha*Gross Beta*
(0.005 mg/1) (0.1 mg/1) (0.01 mg/1) (see Fig.V-28) (50 pCi/1)
NATION
0.0
0.0
0.0
0.5
0.0
REGION
Northeast
0.0
0.0
0.0
0.0
0.0
North Central
0.0
0.0
0.0
0.0
0.0
South
0.0
0.0
0.0
1.3
0.0
West
0.0
0.0
0.0
0.0
0.0
SMSA/NonSMSA
SMSA
0.0
0.0
0.0
0.0
0.0
NonSMSA
0.0
0.0
0.0
0.8
0.0
SIZE OF PLACE
Large rural
communities
0.0
0.0
0.0
0.0
0.0
Small rural
communities
0.0
0.0
0.0
0.0
0.0
Other rural
areas
0.0
0.0
0.0
0.6
0.0
SIZE OF SYSTEM
Community
0.0
0.0
0.0
1.0
0.0
Intermediate
0.0
0.0
0.0
0.0
0.0
Individual
0.0
0.0
0.0
0.0
0.0
~Constituent analyzed for only the 10 percent NSA subsample.
-------
V - 212
each of the other three regions surpassed the reference values for eighteen of the
twenty constituents.
The largest proportions of households above specific reference values
occurred in the North Central. In particular, North Central households had the
largest proportion of households above the reference value for seven constituents:
nitrate-N, iron, manganese, sodium, color, estimated total dissolved solids, and
mercury. Households in the West had the highest rates for six constituents:
sulfates, magnesium, arsenic, cadmium, selenium, and fluoride. Households in the
South had the largest percentages above reference values for four constituents:
total coliform bacteria, lead, barium, and gross alpha radiation. Northeast
households had the greatest rates for only fecal coliforms. Equal proportions of
households in the South and Northeast—4.8 percent—were above the reference
value for silver.
In terms of the general water quality implications of the data in Table V-4,
households in the Northeast clearly were least likely to have problems. Households
in the North Central generally were most likely to have potential water quality
problems—whether the problems were related to health, aesthetic, or economic
effects. However, potential bacteriological problems, as indicated by the findings
for total coliform and fecal coliform, were most likely to occur in the South and
the West. As reported previously in this chapter, the NSA findings suggested a
particularly serious potential health problem involving the presence of two metals:
lead and mercury. Levels of a third metal—cadmium—also caused concern.
Although higher-than-reference-value concentrations of these metals were dis-
covered in every region of the US, the rates differed substantially from one region
to another. Households in the South were most likely to exceed the reference
value for lead; households in the North Central were most likely to be above the
reference value for mercury; households in the West were most likely to surpass
the reference value for cadmium.
-------
V - 213
In terms of the magnitude of the potential for water quality problems in
the rural US, the situation was most serious in the North Central and South. In
these two regions combined, households had the largest percentages over the
reference values for twelve out of the twenty constituents which were found in
concentrations exceeding the NSA reference values. The relative magnitude of the
potential problem was enhanced because such a large proportion of rural households
were located in these two regions. Specifically, of all rural households in the
nation, as based on the entire NSA sample, 70.1 percent were located in the North
Central and South. Of ail rural households in the nation, as based on the 10 percent
NSA subsample, 67.3 percent were located in the North Central and South.
NonSMSA households had a higher potential for water quality problems
than did SMSA households. As Table V-4 shows, more than 15.0 percent of SMSA
households exceeded the reference values for total coliform bacteria, mercury,
total dissolved solids, and cadmium. Among nonSMSA households, more than 15.0
percent were over NSA reference values for total coliform, mercury, iron,
manganese, and lead. In all, nonSMSA households had higher percentages above the
reference values for fourteen of the twenty constituents. In contrast, SMSA
households had higher percentages above the reference values for only four of the
twenty constituents. The situation among nonSMSA households was particularly
serious because of the numbers of households involved: about two-thirds of the
total US rural population lived in nonSMSA areas.
Values for total coliform bacteria, lead, and mercury exceeded the respec-
tive reference values in over 15.0 percent of the households in every size-of-place
category. Iron problems tended to be more prominent in small communities and in
other rural areas. Total dissolved solids and sodium problems were slightly higher
in large and small rural communities than in other rural areas. Manganese
problems seemed to be concentrated in small rural communities, while selenium
-------
V - 214
and cadmium problems occurred most frequently in other rural areas and large
rural communities.
Among all categories of system size, values for mercury and total coliform
bacteria exceeded the reference values in over 15.0 percent of the households.
Sodium appeared over the reference value most often among households served by
community systems. Selenium was prominent among households served by inter-
mediate systems. Five constituents were found in concentrations exceeding the
reference values in more than 15.0 percent of rural households served by
community water systems or by individual systems. This situation was the same
for seven constituents among households served by intermediate systems. More
than 15.0 percent of rural households exceeded the reference value for total
coliform, regardless of the size of supply system serving the household.
Summary of health-related reference values: Approach 2
The second approach to analyzing the implications of the NSA laboratory
findings is presented in this section. Whereas the first summary approach focused
on comparisons across specific constituents, this approach assessed each household
in terms of the number of constituents which exceeded MCL-based reference
values. This assessment resulted in a tabulation of the number and percentage of
households exceeding no reference value, one reference value, two reference
values, and so forth. This second approach considered only those constituents for
which the NSA reference values were based on primary MCLs. Since primary
MCLs were based predominantly on health implications, this aggregation has health
risk as a principal concern. However, these results are based on the 10 percent
subsample of households, not the full NSA sample, since only the subsample
included all the constituents for which primary MCLs were established.
In addition to the fact that these results have a reduced base, there are
important limitations to the interpretability of a simple count of households
-------
V - 215
exceeding a given number of reference values. Some of these limitations will be
discussed after presenting the summary results; other limitations will be addressed
in Chapter VI.
Number of reference values exceeded among households in the rural US
At the national level, it can be seen from Table V-5 that almost two-thirds
of all rural US households (63.6 percent, or 14.0 million) exceeded one or more
primary MCL-based reference values. Furthermore, almost one-third (31.7 per-
cent) of all households were above two or more reference values and 9.9 percent
exceeded three or more MCL-based reference values. Just over one-third of all
households (36.4 percent) exceeded no reference value.
The results showed distinctive patterns in the different NSA groupings. As
to regional differences, the rate of households above various reference values was
greatest in the West, where about three-quarters of the households (75.4 percent)
exceeded one or more reference values. The rates in the South (66.6 percent) and
North Central (64.8 percent) also were high. Though lower than in the West, these
rates were substantially higher than in the Northeast (45.2 percent). Even though
the proportion of households exceeding at least one MCL-based reference value
was 30 percentage points lower in the Northeast than in the West, still nearly half
of the households in the Northeast exceeded at least one reference value.
The regional pattern of households above two or more MCL-based refer-
ence values was somewhat different from the overall rates. In order of magnitude,
the specific rates for exceeding two or more reference values were 44.1 percent in
the West, 37.8 percent in the North Central, 33.5 percent in the South, and 8.3
percent in the Northeast. In other words, although North Central households
ranked third in terms of the overall regional rate for exceeding one or more
reference values, those households ranked second in exceeding two or more
reference values. Furthermore, the greatest number of reference values surpassed
-------
V - 216
Table V-5
Rural Households Where Major Water Supplies Had Constituent Values
in Excess of Primary MCL-Based Reference Values
MCL-Based Number and Percent of Households
Reference Values North
Exceeded Nation Northeast Central South West
Exceeded 0 8,005,000 2,024,000 2,187,000 3,105,000 682,000
36.4% 54.8% 35.2% 33.4% 24.6%
Exceeded 1 7,007,000 1,362,000 1,676,000 3,074,000 870,000
31.9% 36.9% 27.0% 33.1% 31.3%
Exceeded 2 4,772,000 243,000. 1,475,000 .2,601,000 531,000
21.7% 6.6% 23.7% 28.0% 19.1%
Exceeded 3 1,613,000 64,000 470,000 447,000 578,000
7.3% 1.7% 7.6% 4.8% 20.8%
Exceeded 4 449,000 0 270,000 64,000 116,000
2.0% 0.0% 4.3% 0.7% 4.2%
Exceeded 5 128,000 0 136,000 0 0
0.6% 0.0% 2.2% 0.0% 0.0%
Total exceeding 13,969,000 1,669,000 4,026,000 6,185,000 2,095,000
one or more
reference values 63.6% 45.2% 64.8% 66.6% 75.4%
Total 21,974,000 3,693,000 6,213,000 9,291,000 2,777,000
Households
100.0% 100.0% 100.0% 100.0% 100.0%
-------
V - 217
Table V-5 continued
MCL-Based Number and Percent
Reference of Households
Values Exceeded SMSA NonSMSA
Exceeded 0 3,156,000 4,764,000
44.8% 31.9%
Exceeded 1 1,718,000 5,365,000
24.4% 35.9%
Exceeded 2 1,395,000 3,396,000
19.8% 22.7%
Exceeded 3 624,000 974,000
8.9% 6.5%
Exceeded 4 147,000 301,000
2.1% 2.0%
Exceeded 5 0 133,000
0.0% 0.9%
Total exceeding 3,884,000 10,170,000
one or more
reference values 55.2% 68.1%
Total Households 7,040,000 14,934,000
100.0% 100.0%
-------
V - 218
Table V-5 continued
MCL -Based
Reference Values
Exceeded
Number and Percent of Households
Large Small Other
Rural Rural Rural
Communities Communities Areas
Exceeded 0
Exceeded 1
Exceeded 2
Exceeded 3
Exceeded 4
Exceeded 5
1,084,000
45.7%
678,000
28.6%
608,000
25.6%
0
0.0%
0
0.0%
0
0.0%
510,000
33.8%
636,000
42.2%
263,000
17.5%
99,000
6.6%
0
0.0%
0
0.0%
6,459,000
35.7%
5,673,000
31.4%
3,922,000
21.7%
1,477,000
8.2%
440,000
2.4%
125,000
0.7%
Total exceeding
one or more
reference values
1,286,000
54.3%
999,000
66.2%
11,637,000
64.3%
Total Households 2,369,000 1,509,000 18,095,000
100.0% 100.0% 100.0%
-------
V - 219
Table V-5 continued
MCL -Based
Number and Percent of Households
Reference Values
Exceeded
Individual
System
Intermediate Comminity
System System
Exceeded 0
2,734,000
31.2%
580,000
26.0%
4,589,000
41.8%
Exceeded 1
3,421,000
39.0%
693,000
31.1%
2,978,000
27.1%
Exceeded 2
1,896,000
21.6%
350,000
15.7%
2,503,000
22.8%
Exceeded 3
642,000
7.3%
285,000
12.8%
706,000
6.4%
Exceeded 4
67,000
0.8%
179,000
8.0%
205,000
1.9%
Exceeded 5
0
0.0%
141,000
6.3%
0
0.0%
Total exceeding
one or more
reference values
6,031,000
68.8%
1,648,000
74.0%
6,392,000
58.2%
Total Households
8,765,000
2,228,000
10,981,000
100.0%
100.0%
100.0%
Each figure in the tahie is rounded independently; numbers
and percentages may not equal rounded totals.
-------
V - 220
per household was four in all regions except the North Central, where the
maximum was five.
Among SMSA and nonSMSA households, the overall proportion of house-
holds above one or more reference values was about thirteen percentage points
higher among nonSMSA households than among SMSA households (68.1 percent,
compared to 55.2 percent). On the other hand, the proportion of SMSA households
over two or more reference values was 30.8 percent, a proportion which was close
to the comparable percentage among nonSMSA households (32.2 percent).
The size-of-place comparison showed that households in small rural com-
munities fared poorest overall in exceeding reference values but, as seen in Table
V-5, small rural communities also had the smallest proportion of households above
the reference values for two or more constituents. The largest proportion of
households exceeding two or more reference values occurred in other rural areas.
Specifically, the rates for surpassing two or more reference values were 32.9
percent in other rural areas, 25.6 percent in large rural communities, and 24.0
percent in small rural communities.
Rates were substantially higher among households served by intermediate
systems than among those served by individual or community systems. As seen in
Table V-5, nearly three-quarters of all households using intermediate systems
exceeded one or more reference values. This was about sixteen percentage points
higher than the rate among households served by community systems. Further-
more, 42.9 percent of households served by intermediate systems exceeded two or
more reference values; 6.3 percent exceeded five reference values. In contrast,
27.8 percent of individual-system households and 31.1 percent of community-
system households exceeded two or more reference values, and none exceeded
more than four. Overall, households served by community systems clearly fared
the best in terms of this measure of water quality. Despite this, the advantage was
only a relative one: 58.2 percent of households served by community systems
-------
V - 221
exceeded one or more reference values. Furthermore, when rates for exceeding
more than one MCL-based reference value are considered, individual systems
showed lower problem rates than community systems.
Limitations of approach
This summary of counts of households exceeding MCL-based reference
values needs to be qualified carefully. One complication results from an important
interpretative distinction between the MCLs themselves and the MCL-based
reference values used in the NSA. The primary MCLs are specific concentrations
of constituents which, because they pose possible health threats, cannot be
exceeded in public drinking water supplies. Thus, the analysis here does not
provide information about households failing to comply with primary MCLs in the
technical sense as spelled out in federal regulations. However, the analysis does
provide information about households surpassing NSA reference values which are
based on MCLs and which thus provide health-related measurements.
A second limitation of this approach is that each constituent is treated as
though it were independent of others included in the index. That is, the summary
simply adds the number of MCL-based reference values exceeded, and makes no
attempt to consider their relative importance or possible interactions. The focus
of the analysis so far has been on the concentrations of individual constituents in
rural households. This perspective has provided insight into the relative magnitude
of potential difficulties posed by different constituents. However, this analysis
was not intended .to provide information about the possible interrelated effects of
levels of several constituents occurring simultaneously in one household.
The water-quality implications posed by multiple constituents" in one supply
can be assessed only in a limited fashion in the NSA. Physical and chemical
interactions may occur among simultaneously present constituents, but the almost
-------
V - 222
infinite combinations of constituents and concentrations cannot possibly be quan-
tified in a general survey like the NSA.
The focus of the NSA on the individual, unique characteristics of each
constituent is consistent with the current approach to water quality control,
however. That is, since so many factors—ranging from the acidity of water to the
presence of other compounds—can influence the effect of the constituents, it
generally has been practical only to establish criteria and standards for the
constituents independently. (There are, of course, exceptions such as the tempera-
ture-related MCLs for fluorides and the complex, interrelated MCLs for radio-
active materials.)
A third complication of this summary approach results from the fact that
some constituents with primary MCLs were studied in specimens from all NSA
sample households, but some were studied only in a subsample of those households.
Since all of the constituents could be assessed here only in the smaller Group II
subsample, only findings referring to those 267 households are considered in this
second summary approach. As a result of the smaller sample size, the confidence
in the number of households estimated is lower than if the full sample had been
involved. Also, subnational differences must be fairly substantial before there can
be statistical confidence that the differences are real.
A final limitation of this summation of households exceeding MCL-based
reference values is its inherent insensitivity. It is possible, for example, for a
household to have a water supply which is relatively free from any of the measured
contaminants. However, on the particular day when the water was taken for the
NSA survey, a few coliform bacteria may have been in the water supply of .that
household. Such a household would be recorded as having exceeded one reference
value. On the other hand, another household could have a water supply for which
many or all of the contaminants measured in the NSA were present in concentra-
tions very close to, but not exceeding the reference values. This household would
-------
V - 223
show up as having surpassed no reference values and would therefore appear to
have better water quality than the household previously described, a very dubious
conclusion. This problem results from having a single threshold (the MCL-based
reference value) for each constituent. While the concentration of a contaminant
varies across a wide range, a count of over-reference-value cases implies that
those above the reference value have poor water quality while those below have
good water quality. This inherent insensitivity has other important undesirable
implications: a household water supply which barely exceeds a reference value is
judged equal to one which exceeds the reference value by a factor of thousands.
While these limitations must be kept in mind in evaluating the summary of
rural US households exceeding MCL-based reference values, they should not be
taken as a repudiation of the summary. The importance of the summary lies in
showing the remarkable number of households across the US with water supplies
which exceeded MCL-based reference values.
The count of over-reference-value households is a powerful summary
indicating the widespread nature of water quality problems. Part of the effort in
Chapter VI will be to develop summarizing indices which avoid some of the
limitations outlined here. Further, Chapter VI will present indices which are more
appropriate for the purposes of analyses in later chapters.
PERCEPTION OF QUALITY
So far in this chapter, water quality has been described in terms of specific
substances that can be detected and measured by laboratory analysis. Another
approach to portraying household water quality is to focus on people's perceptions
—their subjective judgments about their water supplies. Though not necessarily as
reliable as laboratory measurements, people's perceptions of the quality of their
household water supply provided an important supplement to laboratory data in the
NSA. First, they served as an independent source of information about household
-------
V - 224
water quality; second, they reflected prevailing conditions of the household water
supply, while the laboratory data reflected the particular conditions that existed
the day that the NSA specimens were collected.
There was an additional reason for the NSA to assess people's subjective
judgments about their household water supplies—the potential influence such judg-
ments have on household decision-making. In that sense, NSA perceptual informa-
tion may be as important as NSA laboratory data for understanding rural water
conditions. People's perceptions about the quality of the household water supply
probably had largely determined past household decisions about using or improving
the water supply. Moreover, people's perceptions of the quality of their water
supplies would have a direct bearing on any future governmental efforts to improve
household water conditions.
Some simple examples illustrate the influence of perceptions of water
quality on people's behavior. Most importantly, many problems of water quality
cannot be discerned by drinking the water. Consequently, a person may have a
firm belief that certain water is good, pure drinking water when in fact it may
exceed federal MCLs for one or more constituents. In the absence of any
perceptual indication of a problem in water quality, a household would continue to
rely on such water instead of seeking ways to improve it.
As another example, to someone tasting it for the first time, water with a
high sulfate content is usually thought to have an unpleasant odor or taste. Such
water may or may not exceed federal MCLs. But since people can acquire a
102
physiological tolerance to water with a high sulfate content, people who are
used to drinking it may believe that it is pleasant enough to drink and that it has no
health effects of any consequence, whether or not it exceeds federal MCLs. These
people would be unlikely to support governmental proposals to limit sulfates in
drinking water, especially if meeting proposed regulations would entail a direct
cost to their household.
-------
V - 225
In addition, people's perceptions are sometimes influenced by special
circumstances. For example, if a person had a substandard dug well but none of his
neighbors had adequate drilled wells, he might be content with his supply. Again, if
a person were pleased with other physical characteristics of his housing, he might
have a positive attitude toward a water supply which he knew had certain problems
or inconveniences.
Still other factors may determine perceptions. If a person were satisfied
with the operation and management of the system supplying the household water
and, in addition, believed that the water was reasonably priced, he might overlook
its inferior quality. Likewise, if the water were inexpensive because of an
unlimited quantity being available from the supply, a person might ignore inferior
quality.
Perceptions can also work the other way and influence people to make
unnecessary improvements to a water supply. The opinion that some aesthetic
aspect of a water supply—color, for example—is objectionable or even physically
harmful could provide the impetus for expensive improvements at the household or
for regulations more stringent than theoretically necessary to maintain public
health.
Given the potential impact of people's perceptions of water quality on
household water use and on public support for proposals that would affect
household water conditions, it was important for the NSA to assess perceptions
about prevailing water conditions at rural households. The inquiry included a series
of questions related to perceived quality. Questions were asked about the odor,
taste, clarity, color, sediment content, and temperature of the major household
water supply. For each of the characteristics, respondents first were asked
whether the condition was present and, if so, to what degree. Next, they were
asked for a description of the condition and whether any changes occurred in it, as
-------
V - 226
well as about the duration, possible reasons, seasonal variation, and agreeability of
the condition.
The results, both national and subnational, are presented in sequence for
each of the six perceptual characteristics related to quality (odor, taste, cloudi-
ness, color, sediment, and temperature). At the national level, results are
compiled for the intensity and duration of the perceived condition, whether or not
changes occurred in it, the reported reason for the condition, and its agreeability.
Subnational results are restricted to the condition's intensity and duration and any
associated changes or fluctuations in it. Because they are best discussed
separately, the results on seasonal variation are presented in the last part of this
section. (In reading this section, it should be kept in mind that NSA perceptual
data are based on the judgments of only one individual in each household—the
household head or another carefully chosen adult.)
Odor
Three-quarters of all rural households (16.4 million) reported that there
was never any odor present in the water supply (see Figure V-29). At another one
million households, the water supply generally had no odor, while 2.2 million
households reported that slight odors occurred occasionally. Another two million
households reported a prevalent odor—either a slight odor that was present most or
all of the time, or a strong odor that was present only some of the time. Less than
400,000 households reported that a strong odor was generally or constantly present
in the water. In a related finding, though water supply conditions might be subject
to fluctuations from time to time, only 17.1. percent of all households reported
changes in the odor of the water supply.
Odors that were reported included some that respondents could identify as
specific substances—chlorine and sulfur, for example—and others that had a less
definite origin, such as "iron," and swampy or putrid odors. Although water supply
-------
V - 227
Figure V-29
Intensity and Duration of Perceived Odor
in Rural Household Water Supplies
never any
odor
generally
no odor
occasional
slight odor
prevalent
odor
generally
strong odor
constant
strong odor
I
Number of Households (in millions)
2.2 4.4 6.6 8.8 11.0 13.2 15,4 17.6 19£ 2 2D
0 20 30 40 50 60 70 80
Percent of Households
90 100
-------
V - 228
odors were expected to be described in many different ways, the most frequent
responses reflected only five specific odors (see Table V-6). In particular, about
1.1 million households reported a chlorine odor, while another 0.8 million supplies
were reported to smell swampy. Other frequently reported odors were sulfur (0.7
million supplies) and iron (0.3 million supplies). Some odors besides those listed
individually in Table V-6 also were reported, but each was mentioned at fewer than
1 percent of the households reporting odors. These odors, which are subsumed in
the "other miscellaneous" category, were detected in about 488,000 water supplies.
Reasons for water supply odor
Reasons cited for perceived water supply conditions sometimes were
related to deliberate or planned activities such as chlorination and maintenance
practices. Others were related to supply technology (inadequacy of physical
facilities, and breakdowns of facilities), and to mismanagement of the system. The
category, "inadequacy of the physical facilities," reflected the general condition of
the water supply's physical components. It included responses such as (1) "the
cistern does not filter out debris"; (2) "the pipes are too small"; (3) "the storage
capacity is inadequate"; and W "the well is not deep enough." The category, "a
problem within the house," on the other hand, referred to minor, short-term
difficulties that were more closely associated with water supply facilities situated
within the dwelling unit. Some specific responses in this category were (1) "the
pipes broke," and (2) "the softening system gave out." The category, "a breakdown
in the physical facilities of the water system outside the house," referred to
problems such as broken pipes that occurred on the household property but outside
the dwelling unit itself.
Other reasons for perceived water supply characteristics were related to
natural conditions. With regard to odor, for instance, "the mineral content of the
water" referred to dissolved minerals or gaseous odors. A less specific category,
-------
V - 229
Table V-6
Perceived Odors in Rural Household Water Supplies
Nation
Perceived Odor
Number of
Households
Percent of
Households
Chlorine
1,128,000
28.0
Swampy
815,000
20.2
Sulfur (rotten eggs)
692,000
17.2
Iron
334,000
8.3
' Putrid (sewage odor)
264,000
6.6
Other miscellaneous
488,000
12.1.
Don't know
311,000
7.7
*Total
4,032,000
100.1
*Table includes all households which reported a preva-
lent, generally strong, or constant strong odor, and
some households that reported an occasional slight
odor.
Each figure in the table is rounded independently;
numbers and percentages may not equal rounded
totals.
-------
V - 230
"the natural odor of the source," was used when a respondent was vague about the
reasons for the water supply odor.
The categories "weather conditions" and "seasonal factors" were differ-
entiated by their predictability. For example, if the respondent attributed a
condition to spring rains, the category "seasonal factors" was used. If heavy rains
were blamed for a condition, but with no mention of seasonality, the category
"weather conditions" was used.
A problem attributed to a known source, but not conveniently or easily
remedied, was categorized as a "situation beyond household or water system
control." A case in point concerned a community experiencing odor problems
attributed to a specific constituent in the water. Since the cost of removing the
constituent from the water was beyond the users' ability to pay, the constituent
was not removed. In another case, a disturbance of the aquifer was attributed to
construction work in the area.
With regard to water supply odors (see Table V-7), by far the most
prevalent reason cited for the condition was deliberate or planned activities (31.3
percent of households reporting water supply odors). Specifically, though not
reflected in the table, nine out of ten households which reported a chlorine odor
attributed it to deliberate or planned activities. A few of these households
attributed the chlorine odor to mismanagement of the system.
Altogether, 23.8 percent of households reporting water supply odors
attributed the odor to natural factors such as the mineral content of the water, the
natural odor of the source, weather conditions, and seasonal factors.
Also mentioned relatively often as the cause of water supply odors were
factors related to supply technology and management, such as inadequacy of
physical facilities, breakdowns, and mismanagement (16.0 percent, combined).
-------
V - 231
Table V-7
Reported Reasons for Perceived Odor in Rural
Household Water Supplies
Nation
Number of Percent of
Reason Households Households
Deliberate or planned activities
(additives, etc.) 1,746,000 31.3
Mineral content of the water 684,000 12.2
Inadequacy of the physical facilities
of the system (pipes too small,
storage capacity too small,
inadequate filter, etc. 428,000 7.7
Natural odor of the source 300,000 5.4
Mismanagement of the system (ill-
trained operators, inattention, etc.) 261,000 4.7
Weather conditions 256,000 4.6
A breakdown in the physical facilities
of the water system outside the house 203,000 3.6
A problem within the house 201,000 3.6
Situation beyond household or water
system control 139,000 2.5
Seasonal factors 90,000 1.6
Water stagnation 55,000 1.0
Other miscellaneous 126,000 2.3
Don't know 1,097,000 19.6
~Total 5,585,000 100.1
~Table includes only those households which reported an odor in their
water supplies.
Each figure in the table is rounded independently; numbers and
percentages may not equal rounded totals.
-------
V - 232
Agreeability of water supply odor
Using a five-point attitudinai scale, the NSA sought to assess how
agreeable the water supply seemed in all rural households as to odor or the lack of
odor. At 2.2 million households (9.5 percent), the water's odor was deemed
disagreeable, while at 12.3 million households (56.1 percent), it was considered
pleasant. Specifically, the water's odor was strongly disliked at 635,000 rural
households (2.9 percent), moderately disliked at 1.5 million (6.6 percent), moder-
ately liked at 8.0 million (36.3 percent), and very well liked at 4 A million (19.8
percent). At about 7.5 million households (34.3 percent), the water was not noticed
or thought about very much in regard to odor.
Major subnational patterns in water supply odor
Little variation was detected among the four regions of the United States
and between SMSA and nonSMSA households in either the intensity and duration of
water supply odors or in changes from the usual condition. In the size-of-place
comparison, however, reported odor conditions were substantially different in other
rural areas from what they were in large and small rural communities (see Figure
V-29a). Specifically, more households in other rural areas reported there was never
any odor in the water supply (76.5 percent, compared to 64A percent for large
rural communities and 66.8 percent for small rural communities). Also, as place
size increased, so did the prominence of water supply odors. That is, a greater
proportion of households in large rural communities reported prevalent odors,
generally strong odors, and constant strong odors (18.0 percent, combined) than in
small rural communities (12A percent, combined) or other rural areas (9.7 percent,
combined). Also noticed in'the NSA data, though not depicted in Figure V-29a, was
a stronger tendency for households in large and small rural communities to report
changes in the usual condition of the water supply with regard to odor (about 27
-------
V - 233
Figure V-29a
Size-of-Place Variation in Perceived Odor
in Rural Household Water Supplies
0
r-
10
20
I—
30 40 50 60 70
80
-1—
90
100
—i
1 1
l t
i I
COMMUNITIES
12,369,000
1 ,i, (. „
lllr
Hr
. i
[ 1 I
LARGE RURAL
: 1 1 . 1
J-r
b
223"
SMALL RURAL COMMUNITIES
11,509,000
1 !
OTHER RURAL AREAS
118,095,000
10 20 30 40 50 60 70 80 90 100
Percent of Households
KEY:
never any odor
generally no odor
occasional slight odor
prevalent odor
generally strong odor
constant strong odor
-------
V - 234
percent of households in both large and small communities, compared to about 15
percent of households in other rural areas).
A comparison of households with regard to size of supply system also
showed substantial differences (see Figure V-29b). A much smaller proportion of
households served by community water systems reported there was never any odor
in the supply (67.7 percent, compared to 80.2 percent for intermediate systems and
81.7 percent for individual systems). Not reflected in the graph is the fact that a
greater proportion of households served by community water systems reported
changes in the water supply with regard to odor (24.1 percent, compared to 11.3
percent for intermediate-system households and 9.7 percent for individual-system
households). This finding is consistent with changes being reported more often by
households in large and small rural communities, where community water systems
are more common than in other rural areas.
Taste
Although perceptions of odor and taste generally would be expected to
occur together, water supply tastes were reported more often than were odors.
In all, 5.6 million households (25.5 percent) reported water supply odors; 7.8 million
households (35.5 percent) reported tastes. Some of the reported tastes were
pleasant, however, in contrast to all of the reported odors. (For some respon-
dents, the water tasted sweet, or simply "like good, clean, fresh water.")
Graphs depicting subnational variations in perceived characteristics of
water supplies reflect the number of households in each subnational
category. In a size-of-place comparison, for example, the size of the
other-rural-areas portion of the graph reflects the fact that the bulk of
the rural population lived in other rural areas.
-------
V - 235
Figure V-29b
Size-of-System Variation in Perceived Odor
in Rural Household Water Supplies
0 10 20 30 40 50 60 70 80 90 100
1 M ! I | 1 [— 1 1 [ 1 1 1 1
INTERMEDIATE
2,228,000
COMMUNITY
t0;98L000
i , i
i i
Percent of Households
never any odor
generally no odor
occasional slight odor
prevalent odor
generally strong odor
constant strong odor
-------
V - 236
At 796,000 households (3.6 percent), tastes were reported to occur infre-
quently; generally there was no taste (see Figure V-30). At 2.1 million households
(9.7 percent), a slight taste occurred occasionally. At 4.1 million (18.6 percent),
the water supply had a prevalent taste—either a slight taste that was present
most or all of the time, or a strong taste that was present only some of the time.
Relatively few households reported that the water supply generally had a strong
taste (1.0 percent, or 218,000) or that it had a strong taste that was present
constantly (2.5 percent, or 553,000).
Compared with the number of households reporting specific intensities of
water supply odors, slightly fewer households reported that the water supply had
generally no taste or an occasional slight taste (2.9 million versus 3.2 million,
combined). Twice as many households reported prevalent tastes, generally strong
tastes, and constant strong tastes (4.9 million versus 2.4 million, combined). These
differences may stem from the fact that people are more aware of tastes than of
odors.
Changes in taste were about as frequent as changes in odor: 15.4 percent
or 3.4 million households reported changes in taste, while 17.1 percent or 3.8
million households reported changes in odor.
Strictly speaking, human beings can perceive only four tastes—sweet, sour,
salty, and bitter—but these characterized only about 12 percent of the tastes
noticed in household water (see Table V-8). At 1.9 million households (27.3 percent
of all households where a taste was perceived), minerals were tasted in the water,
while a chlorine taste was perceived at 1.1 million households. At 0.8 million
households (11.0 percent), the taste was not specific enough to identify. At other
households, the taste was described as "swampy" or like "good, clean, fresh water."
Although it could be argued that "swampy" actually referred to a perceived odor
rather than a taste, and that "good, clean, fresh water" was neither a taste nor an
odor, these answers occurred frequently and were tabulated so as to preserve all
-------
V - 237
Figure V-30
Intensity and Duration of Perceived Taste
in Rural Household Water Supplies
never any
taste
generally
no taste
occasional
slight taste
prevalent
taste
generally
strong taste
constant
strong taste
Number of Households (in millions)
£2
—i—
4.4 6.6 8.8
0 13.2 15.4 17.6 19.8 22.0
10 20 30 40 50 60 70 80 90 100
Percent of Households
-------
V - 238
Table V-8
Perceived Tastes in Rural Household Water Supplies
Nation
Number of Percent of
Perceived Taste
Households
Households
Mineral water
1,927,000
27.3
Chlorine
1,134,000
16.1
Taste not specific enough
to identify
776,000
11.0
Swampy
590,000
8.6
Good, clean, fresh water
521,000
7 A
Sweet
377,000
5.3
Salty
258,000
3.6
Bitter
219,000
3.1
Sour
25,000
OA
Other miscellaneous
516,000
7.3
Don't know
716,000
10.1
Total
7,058,000
100.2
~Table includes all households which reported a prev-
alent, generally strong, or constant strong taste, and
some households that reported an occasional slight
taste.
Each figure in the table is rounded independently;
numbers and percentages may not equal rounded
totals.
-------
V - 239
information obtained. As shown in the table, a swampy taste was reported in about
0.6 million households, and a good, clean taste in about 0.5 million households.
Reasons for tastes
As with odors, water supply tastes were most often attributed to planned
activities. This reason was cited in 2.2 million households. In close correspondence
with the frequency of minerals as a perceived taste, the presence of minerals was
cited as the reason for water supply tastes in 1.2 million households, while the
natural condition of the source was mentioned at 0.9 million households. Tastes
were attributed to supply technology (inadequacy of facilities, or breakdowns) or to
system management at 13.4 percent, combined, of the households reporting tastes.
Other reasons also were given, as shown in Table V-9.
Agreeability of tastes
Household water was thought to have an agreeable taste in 13.2 million, or
59.9 percent of all rural households. In addition, in 28.4 percent of all rural
households (6.4 million), the water's taste was thought to be all right, or was not
thought about very much. On the other hand, the taste was felt to be disagreeable
in 2.4 million households—or 11.1 percent of all rural households.
Major subnational patterns in tastes
Very little regional variation was evident in the intensity and duration of
water supply tastes. The largest variation—a difference of only six percentage
points—occurred in the proportions of households reporting that there was never
any taste in the water supply: the proportions were 60.8 percent in the Northeast,
64.6 percent in the South, 65.5 percent in the North Central, and 67.0 percent in
the West. This difference was offset, however, by the proportions reporting that
the water supply generally had no taste, which ranged from 7.0 percent in the
-------
V - 240
Table V-9
Reported Reasons for Perceived Taste in Rural
Household Water Supplies
Nation
Reason
Number of
Households
Percent of
Households
Deliberate or planned activities
2,237,000
27.8
Mineral content of the water
1,190,000
14.8
Natural taste of the source
866,000
10.8
Inadequacy of the physical
facilities of the system
456,000
5.7
Mismanagement of the system
373,000
4.6
A problem within the house
342,000
4.3
A breakdown in the physical facilities
of the water system outside the house
247,000
3.1
Weather conditions
220,000
2.7
Seasonal factors
56,000
0.7
Other miscellaneous
337,000
4.2
Don't know
1,713,000
21.3
* Total
8,041,000
100.0
~Table includes only those households which reported a taste in their
water supplies.
Each figure in the table is rounded independently; numbers and
percentages may not equal rounded totals.
-------
V - 241
Northeast to 1.2 percent in the West. Otherwise, the occurrence of reported tastes
held closely to the national pattern. Changes in tastes of water supplies were
reported at only 11.8 percent of rural households in the North Central, compared to
16.4 percent in both the South and West and 17.9 percent in the Northeast.
No substantive differences were found between SMSA and nonSMSA
households. in the occurrence of tastes in the water supply, or with regard to
changes in them.
Variations among the three size-of-place categories followed somewhat the
same pattern as seen for water supply odors. As shown in Figure V-30a, more
households in other rural areas reported that the water supply never had a taste
(66.1 percent, compared to 54.6 percent of households in large rural communities
and 61.2 percent in small rural communities). Further, prevalent tastes were
reported in 21.3 percent of households in large rural communities, compared to
16.0 percent in small rural communities and 18.5 percent in other rural areas.
However, households reported generally strong tastes and constant strong tastes in
about equal proportions: 3.9 percent in large communities, 3.7 percent in small
communities, and 3.4 percent in other rural areas. Changes in the taste of the
water supply were reported at 25.0 percent of households in large rural communi-
ties and 23.0 percent of households in small rural communities, but at only 13.7
percent of households in other rural areas.
In a comparison of households served by supply systems of different sizes,
only 57.6 percent of households served by community systems reported that the
water supply never had a taste, compared to 72.2 percent of households served by
individual systems and 67.9 percent of households served by intermediate systems
(see Figure V-30b). Occasional slight tastes were- far more frequently reported
among households served by community systems, and there was little variation in
the proportions of households reporting prevalent tastes, generally strong tastes,
and constant strong tastes.
-------
V - 242
Figure V-30a
Size-of-Place Variation in Perceived Taste
in Rurai Household Water Supplies
0 10 20 30 40 50 60 70 80 90 100
11 M 1 I r
T
l
i r
k
¦ r
\
LARGE RURAL COMMUNITIES
! [2,369,000
iffi
IU,
1: 1 1 1
SMALL RURAL COMMUNITIES
f | i 1,509,000
' • L \ 1 A
OTHER RURAL AREAS
8,095,000
0 10 20 30 40 50 60 70 80 90 ICO
Percent of Households
*
KEY:
never any taste
generally no taste
occasional slight taste
prevalent taste
generally strong taste
constant strong taste
-------
V - 243
Figure V-30b
Size-of-System Variation in Perceived Taste
in Rural Household Water Supplies
0 10 20 30 40 50 60 70 80 90 100
11 11 11 1 1 1 1 1 1 1 1 1
INDIVIDUAL
8,765,000
i
i 1
INTERMEDIATE
2,228,000
* 1 *. r
r-
COMMUNITY
10,981,000
i i i 11
_L
_L
_L
J
0 10 20 30 40 50 60 70 80 90 100
Percent of Households
KEY:
never any taste
generally no taste
occasional slight taste
prevalent taste
generally strong taste
constant strong taste
-------
V - 2W
Changes in the taste of the water supply were noticed in approximately 9
percent of households served by individual and intermediate systems, but in about
22 percent of households served by community systems.
Cloudiness
Water's cloudiness, color, and sediment content have potential implications
for water quality. These three physical attributes were included in the NSA survey
because they are conditions that are easily observable and because they have been
103
known to play a part in people's attitudes towards their water supplies.
At 73.1 percent of rural households, no cloudiness was ever noticed in the
water supply (see Figure V-3'l). At another 9.9 percent, there was generally no
cloudiness present. However, 10.0 percent or 2.2 million households reported an
occasional slight cloudiness, and 6.1 percent or 1.3 million households reported a
prevalent cloudy condition. Less than 1 percent of rural households reported that
the water supply was generally or constantly very cloudy (0.5 percent and 0.3
percent, respectively). Changes in the water supply with respect to cloudiness
were reported at 23.3 percent of rural households.
Reasons for water supply cloudiness
Cloudiness was attributed to supply technology (inadequacy of facilities or
breakdowns) and system management at 29.6 percent, combined, of the households
reporting cloudiness. At another one million households, the condition was blamed
on deliberate or planned activities. Natural forces—the weather, seasonal factors,
the water's mineral content, and the "natural cloudiness of the source"—were
thought to account for the condition at 23.7 percent of the households reporting
cloudiness. Other reasons were also given, as shown in Table V-10. However, at a
large proportion of households where cloudiness was noted in the water (19.2
percent), no reason for the condition could be cited.
-------
V - 2H5
Figure V-31
Intensity and Duration of Perceived Cloudiness
in Rural Household Water Supplies
never any
cloudiness
generally
cloudiness
occasional
sljght
cloudiness
prevalent
cloudiness
generally
very cloudy
constantly
very cloudy
Number of Households (in millions)
2.2 4.4 6.6 8.8 11.0 13.2 15.4 17.6 19.8 22.0
10 20 30 40 50 60 70 80 90 100
Percent of Households
-------
V - 246
Table V-10
Reported Reasons for Perceived Cloudiness in Rural
Household Water Supplies
• Nation
Reason
Number of
Households
Percent of
Households
A breakdown in the physical facilities
of the water system outside the house
969,000
16.2
Deliberate or planned activities
965,000
16.2
Weather conditions
703,000
11.8
Inadequacy of the physical facilities
of the system
515,000
8.6
Mineral content of the water
421,000
7.0
A problem within the house
361,000
6.0
Mismanagement of the system
284,000
4.8
Natural cloudiness of the source
184,000
3.1
Seasonal factors
108,000
1.8
Water stagnation
49,000
0.8
Other miscellaneous
266,000
4.5
Don't know
1,146,000
19.2
Total
5,971,000
100.0
~Table includes only those households which reported cloudiness in
their water supplies.
Each figure in the table is rounded independently; numbers and
percentages may not equal rounded totals.
-------
V - 247
Agreeability of water supply cloudiness
At fourteen million rural households (63.3 percent), the degree of clarity in
the water supply was found pleasing; at 6.5 million (29.8 percent), feelings were
neutral; but at 1.5 million households (6.9 percent), the condition was displeasing.
Major subnational patterns in cloudiness
Problems with cloudiness in the water supply were least often reported
among households in the Northeast, where 77.4 percent of households reported
there was never any cloudiness and only 4.0 percent reported cloudiness that was
prevalent or worse (see Figure V-31a). Although the South had the smallest
proportion of households reporting there was never any cloudiness in the water
supply (70.1 percent), 24.2 percent in the South (combined) reported there was
generally no cloudiness or that there was only occasional slight cloudiness. The
North Central and West had the highest proportions of households where cloudiness
was prevalent or worse—9.6 percent and 9.0 percent, respectively. There was also
regional variation in the proportions of households reporting changes in the water
supply with respect to cloudiness: changes were noticed in 26.9 percent of
households in the South, 23.5 percent in the West, 21.0 percent in the North
Central, and 18.1 percent in the Northeast.
The SMSA/nonSMSA comparison showed that proportionately fewer non-
SMSA households reported cloudiness never being present in the water supply (71.1
percent, compared to 77.3 percent), but a higher proportion of nonSMSA households
reported there was generally no cloudiness or that occasional slight cloudiness
occurred (22.0 percent, compared to 15.9 percent, combined). An equal proportion
of SMSA and nonSMSA households-1—about 7 percent—reported prevalent cloudiness
or a very cloudy condition that was generally or constantly present. Changes in the
usual condition were reported by 18.7 percent of SMSA households and 25.5 percent
of nonSMSA households.
-------
V - 248
Figure V-31a
Regional Variation in Perceived Cloudiness
in Rural Household Water Supplies
9-
10
11 1' I
20 30 40 50 60 70 80 90 100
NORTHEAST
3,693,000
NORTH
CENTRAL
6,213*000
SOUTH
< t t
1 • 1:
1 1
0
/Wl
WEST
2,777,00
• A.: .... , 1
if
1 1 1 1 1 1 1 1
0 10 20 30 40 50 60 70 80 90
Percent of Households
100
KEY:
never any cloudiness
generally no cloudiness
occasional slight cloudiness
prevalent cloudiness
generally very cloudy
constantly very cloudy
-------
V - 249
Household water supplies in other rural areas were reported free of any
cloudiness more often than those in large or small rural communities (see Figure V-
31b). Specifically, 74.3 percent of households in other rural areas reported there
was never any cloudiness present in the water supply, compared to 69.7 percent in
small communities and 65.9 percent in large communities. Also, only 6.2 percent
of households in other rural areas reported a cloudy condition that was prevalent or
worse, compared to 10.4 percent of households in small communities and 11.1
percent in large communities. Large rural communities had the greatest propor-
tion of households reporting an occasional slight cloudiness—15.9 percent, com-
pared to 9.9 percent for small rural communities and 9.3 percent for other rural
areas. Changes in the water supply with respect to cloudiness were reported in
29.9 percent of households in large rural communities, 24.0 percent of households
in small rural communities, and 22.4 percent of households in other rural areas.
Consistent with the variations in the size-of-place comparison, individual
systems—which were most common in other rural areas—were most often free of
any cloudiness. Specifically, 78.4 percent of households served by individual
systems reported there was never any cloudiness in the water supply, compared to
73.3 percent of those served by intermediate systems and 68.8 percent of those
served by community systems (see Figure V-31c). Water from community systems
tended to be more cloudy: 9.2 percent of households served by community systems
reported cloudiness that was prevalent or worse, compared to 4.4 percent of
individual-system households and 5.6 percent of intermediate-system households.
(The worst condition reported among intermediate-system households was preva-
lent cloudiness.) Changes in the water supply with respect to cloudiness were
reported least frequently among households served by individual systems (18.8
percent, compared to 23.3 percent of households served by intermediate systems
and 26.9 percent of households served by community systems).
-------
V - 250
Figure V-31b
Size-of-Place Variation in Perceived Cloudiness
in Rural Household Water Supplies
0
10 20 30 40 50 60 70 80 90 100
i ii ii i r
T
1
sA/ll
LARGE RURAL COMMUNITIES
1
1
1 1
2,369,000
ggj
SSjSF?
SMALL RURAL COMMUNITIES
I I 11,509,000
S3:
P
OTHER RURAL AREAS
18,095,000
I i i i i I
_L
I
JL
_L
_L
0
10 20
KEY:
30 40 50 60 70
Percent of Households
ne ver any cloudiness
generally no cloudiness
occasional slight cloudiness
prevalent cloudiness
generally very cloudy
constantly very cloudy
80 90 100
-------
V - 251
Figure V-31c
Size-of-System Variation in Perceived Cloudiness
in Rural Household Water Supplies
0 10 20 30 40 50 60 70 80 90 100
1 | ii i i 1 1 1 1 1 1 1 1 1
INDIVIDUAL
8,765,000
m
[¦ r
'km
n i !
INTERMEDIATE
2,228,000
COMMUNITY
10,981,000
I *
i i i i i i i i i i i i i i i
0 10 20 30 40 50 60 70 80 90 100
Percent of Households
never any cloudiness
generally no cloudiness
occasional slight cloudiness
prevalent cloudiness
generally very cloudy
constantly very cloudy
-------
V - 252
Color
Almost 80 percent of rural households—17.5 million—reported that the
water supply never had any color, while 1.3 million reported there was generally no
color. An occasional slight color was noticed in 1.5 million water supplies. At
another 1.4 million households, there was a prevalent color in the water supply.
Worse conditions with respect to color were relatively rare: 0.9 percent or 191,000
households reported that generally the water supply was very colored, and 0.3
percent or 68,000 reported that it was very colored all the time (see Figure V-32).
Changes in the water supply with respect to color were reported at 3.3 million
households (14.9 percent).
The predominant colors reported were gray, yellow, white, and brown,
though many others were also mentioned (see Table V-ll). Gray water was
reported in 0.5 million households, yellow water in 0.5 million, white water in OA
million, and brown water in slightly less than 0.4 million households.
Reasons for water supply color
Color was most frequently attributed to factors of supply technology
(inadequacy or breakdowns of facilities) and management; such factors were
blamed for the condition at 28.5 percent, combined, of households reporting a color
in the water supply. At other households, color was blamed on the mineral content
of the water (13.9 percent) or on deliberate or planned activities (13A percent).
Other reasons that were reported are shown in Table V-12.
Agreeability of water supply color
Overwhelmingly—in almost 94 percent of all rural households—the water
supply seemed agreeable as to color (27.7 percent of households reported neutral
feelings). This finding corresponds with the fact that almost 80 percent of
households reported there was never any color in the water. A mild dislike for the
-------
V - 253
Figure V-32
Intensity and Duration of Perceived Color
in Rural Household Water Supplies
never any
color
generally
no color
occasional
slight color
prevalent
color
generally
very colored
constantly
very colored
Number of Households (in millions)
2.2 4.4 6.6 8.8 11.0 13.2 15.4 17.6 19.8 22.0-
I I 1 i i i
i i
0 20 30 40 50 60 70 80 90 100
Percent of Households
-------
V - 254
Table V-ll
Perceived Colors in Rural Household
Water Supplies
Nation
Colors
Number of
Households
Percent of
Households
Gray
462,000
19.2
Yellow
' 459,000
19.1
White
408,000
17.0
Brown
377,000
15.7
Red
290,000
12.1
Orange
154,000
6.4
Blue
76,000
3.1
Green
26,000
1.1
Other miscellaneous
118,000
4.9
Don't know
34,000
1.4
Total
2,402,000
100.0
* Table includes all households which reported a preva-
lent color, a generally very colored condition, or a
constantly very colored condition, as well as some
households which reported an occasional slight color.
Each figure in the table is rounded independently;
numbers and percentages may not equal rounded
totals.
-------
V - 255
Table V-12
Reported Reasons for Perceived Color in Rural
Household Water Supplies
Nation
Number of Percent of
Reason Households Households
A breakdown in the physical facilities
of the water system outside the house 757,000 16.8
Mineral content of the water 628,000 13.9
Deliberate or planned activities 603,000 13.4
A problem within the house 425,000 9.4
Weather conditions 380,000 8.4
Inadequacy of the physical
facilities of the water system 370,000 8.2
Natural color of the source 172,000 3.8
Mismanagement of the system 158,000 3.5
Seasonal factors 69,000 1.5
Other miscellaneous 157,000 3.5
Don't know 789,000 17.5
~Total 4,508,000 99.9
*Table includes only those households which reported color in their
water supplies.
Each figure in the table is rounded independently; numbers and
percentages may not equal rounded totals.
-------
V - 256
color of the water was reported at k.k percent of rural households, while a strong
dislike for it was reported at 1.9 percent, or OA million households.
Major subnational patterns in color
About equal proportions of households in each of the four regions reported
that there was never any color present in the water supply, but some variation
could be seen in the relative severity of the condition among the regions (see
Figure V-32a). That is, though no households in the West reported that the water
supply was always very colored, still a higher proportion of households there
reported a prevalent colored condition or a generally very colored condition (12.8
percent, combined). In the North Central, 8.6 percent of households in all
reported either a prevalent colored condition or a very colored condition that was
present generally or constantly. By comparison, about 6 percent of households in
the Northeast and South reported these conditions. (As in the West, a constantly
very colored condition was never reported in the Northeast; such a condition was
reported at 0.6 percent of households in the North Central and 0.4 percent in the
South.)
Proportionately more households in the Northeast and South reported less
severe conditions. Combining households that reported that the water supply
generally had no color with those reporting that a slight color occurred occasion-
ally, these less severe conditions were reported at 17.0 percent of households in
the Northeast, 14.3 percent in the South, 11.2 percent in the North Central, and
7.3 percent in the West. The proportions of households reporting changes in the
water with respect to color were 12.1 percent in the West, 13A percent in the
North Central, 15.5 percent in the South, and 17.8 percent in the Northeast.
No substantive variation (less than four percentage points) was seen
between SMSA households and nonSMSA households with respect to water supply
color or changes in the water supply color.
-------
V - 257
Figure V-32a
Regional Variation in Perceived Color
in Rural Household Water Supplies
0 10 20 30 40 50 60 70 80 90 100
T'' i 11 1 1 1 n 1 1 1 1 1
r11 1 11 i i i i i i i i
1
tm
NORTHEAST
3,693,000
• >
NORTH
CENTRAL
6,213,000
O
<3
F m
O
r
!V ^:
I
SOU-
9,2
: .. '¦
WEST
2,777,000
i i i i i i i i i i i i i i i
0 10 20 30 40 50 60 70 80 90 100
Percent of Households
never any color
generally no color
occasional slight color
prevalent color
generally very colored
constantly very colored
-------
V -258
Compared with large and small rural communities, a smaller proportion of
households in other rural areas reported color in the household water supply (see
Figure V-32b). In other rural areas, 81.4 percent of households reported that
there was never any color present in the water supply, compared to 68.9 percent
in large rural communities and 73.3 percent in small rural communities. The
water supply exhibited a color that was prevalent or even more pervasive at 7.0
percent of households in other rural areas, 9.0 percent of households in small
communities, and 10.2 percent of households in large communities.
Changes in the usual condition of the water supply with respect to color
were noticed by 25.0 percent of households in large communities, 21.8 percent of
households in small communities, and by only 13.0 percent oi households in other
rural areas.
Comparing households served by water supply systems of different sizes,
about equal proportions of those served by intermediate systems and community
systems reported color in the household water supply: 24A percent for inter-
mediate systems, and 23A percent for community systems. In contrast, only 16.0
percent of households served by individual systems reported water supply color.
Despite this difference, which reflected the proportion of households reporting
that color never occurred in the water supply, there were only slight deviations
from national estimates in the proportions of households reporting various other
degrees of the condition.
Consistent with the pattern seen in the size-of-place comparison, changes
in the usual condition of the water supply with respect to color were reported by
about 17 percent of households served either by community systems or inter-
mediate systems, but by only 10.9 percent of households served by individual
systems.
-------
V - 259
Figure V-32b
Size-of-Place Variation in Perceived Color
in Rural Household Water Supplies
0 10 20 30 40 50 60 70 80 90 100
i11 1 n i i i n i i i i i
LARGE
RURAL
COMMUNITIES
2,369.000
1 1 1
SMALL RURAL COMMUNITIES
j j j 1,509,000
OTHER RURAL AREAS
[ j j t8t09St000
11 i i 11
_l
_L
0 10 20 30 40 50 60 70 80 90 100
Percent of Households
KEY:
never any color
generally no color
occasional slight color
prevalent color
generally very colored
constantly very colored
-------
V - 260
Sediment
While 14.7 million rural households (66.8 percent) reported there was
never any sediment in the water supply, 7.3 million households (33.2 percent)
experienced some degree of sedimentation (see Figure V-33). For most households
reporting sediment (4.6 million, or 20.9 percent), the condition was an occasional
problem, and did not involve heavy sediment. In 491,000 households (2.2 percent),
the water supply generally had no sediment. A prevalent sediment condition
—heavy sediment occurring some of the time or a milder condition that was
present most or all of the time—was reported in two million households (9.0
percent). Heavy sediment characterized as either generally or constantly present
was reported at a total of 247,000 households (1.1 percent). Changes from the
usual condition of the water supply with respect to sediment were noticed at 16.8
percent of all households.
Reasons for water supply sediment
Sediment was most frequently attributed to factors of supply technology
(inadequacy or breakdowns of facilities) and management; altogether, such factors
were cited at 32.5 percent of rural households reporting sediment (see Table V-13).
Various natural conditions—the mineral content of the water, the presence of
sediment in the source, weather conditions, and seasonal factors—were cited at
25.9 percent. A large proportion of households (20.0 percent) could not give a
reason for the sediment condition.
Agreeability of water supply sediment
An estimated 59.5 percent of rural households liked the water supply with
respect to sediment, 14.3 percent disliked it, and 26.2 percent had neutral feelings.
As would be expected, conditions of heavy sediment and no sediment both
generated strong feelings. The absence of sediment was very well liked in eight
-------
V - 261
Figure V-33
Intensity and Duration of Perceived Sediment
in Rural Household Water Supplies
Number of Households (in millions)
.0 13.2 15.4 17.6 19.8 22.0
2.2 4.4 6.6 8.8
_ J ! ! L_
never any
sediment
generally
no sediment
occasional
sediment
prevalent
sediment
generally
heavy sediment
constant
heavy sediment
0 20 30 40 50 60 70 80 90 100
Percent of Households
-------
V - 262
Table V-13
Reported Reasons for Perceived Sediment in Rural
Household Water Supplies
Nation
Number of Percent of
Reason Households Households
A breakdown in the physical facilities
of the water system outside the house 1,052,000 14.4
Inadequacy of the physical
facilities of the system 1,017,000 13.9
Mineral content of the water 850,000 11.6
Deliberate or planned activities 687,000 9.4
A problem within the house 634,000 8.7
Source contains sediment
which does not settle out 499,000 6.8
Weather conditions 490,000 6.7
Mismanagement of the system 307,000 4.2
Seasonal factors 62,000 0.8
Other miscellaneous 260,000 3.6
Don't know 1,465,000 20.0
*Total 7,324,000 100.1
* Table includes only those households which reported sediment in
their water supplies.
Each figure in the table is rounded independently; numbers and
percentages may not equal rounded totals.
-------
V - 263
out of ten rural households; on the other hand, heavy sediment was strongly disliked
whenever it occurred. In contrast to the strong feelings aroused by heavy sediment
and a total lack of sediment, neutral feelings were expressed more frequently when
the supply provided water having a moderate level of sediment.
Major subnational patterns in sediment
Except for regional differences, subnational comparisons showed no appre-
ciable variation in the frequency and intensity of sediment conditions. The chief
regional difference was the much higher proportion of households in the West that
reported sediment in the water supply (see Figure V-33a). In the West, only 56.5
percent of households reported that there was never any sediment in the water,
compared to 66.9 percent in the Northeast, 65.8 percent in the North Central, and
70.5 percent in the South. A small proportion in each region reported that the
water supply generally had no sediment, and proportions reporting occasional
sediment were about equal in the four regions. Prevalent or even more pervasive
sediment conditions were reported at 16.3 percent of households in the West,
combined, compared to 9.3 percent in the Northeast, 11.1 percent in the North
Central, and 7.8 percent in the South. Thus, sediment was reported most
frequently in the West and least frequently in the South.
Changes from the household water supply's usual condition with respect to
sediment were noticed most often in the West (20.5 percent of households,
compared to 18.1 percent in the Northeast, 16.5 percent in the North Central, and
15.5 percent in the South). Other subnational comparisons showed no variation
$
from national estimates with respect to changes from the usual sediment condition.
Temperature
At almost all rural households, the temperature of the water supply was
perceived as usually cool or cold. The water was described as cool at 56.0 percent
-------
V - 264
Figure V-33a
Regional Variation in Perceived Sediment
in Rural Household Water Supplies
0 10 20 30 40 50 60 70 80 90 100
1 1 1 1 1 i 1 1 1 i 1 1 1 1 1
1
m
1- ¦¦¦:-,¦
i«i.
NORTHEAST
3,693,000
. t ¦ ¦
NORTH
CENTRAL
6,213,000
SOUTH
WEST
2,777,000
I i i i i I 1 I I I I I L_ I I
0 10 20 30 40 50 60 70 80. 90 100
Percent of Households
never any sediment
generally no sediment
occasional sediment
prevalent sediment
generally heavy sediment
constant heavy sediment
-------
V - 265
of rural households (12.3 million), and as cold at 37.9 percent (8.3 million). The
water was depicted as warm at only 5.9 percent of rural households (1.3 million).
No changes in the usual water temperature were noticed at 75.0 percent of
households, but changes were reported at 5.5 million of them. Further, for the vast
majority of those 5.5 million households—92 percent—the water temperature was
fairly constant; for 1.5 million, however, fluctuations in water temperature
occurred all the time. Fluctuations were more common in warmer supplies than in
colder ones.
Reasons for water supply temperature
At almost 61 percent of households, the temperature of the water supply
was attributed to the natural temperature of the source (see Table V-H). Seasonal
variation—another natural occurrence—was the next most frequently cited reason
for the temperature of the water supply, but it was mentioned at only 12.5 percent
of households. Other reasons, such as various possible features of the water supply
system, were mentioned only infrequently. No reason could be cited at 17.7
percent of households.
The reasons given in Table V-14 for the temperature of the water supply
are not distinguished by the specific usual temperature reported for the household
supply. When the reasons were considered in light of the usual temperature of the
water supply, some striking differences became apparent. At 92.6 percent of
households where the water supply was cold, the temperature was attributed to
natural conditions. In contrast, natural conditions were judged to be the cause of
the water temperature at only 66.1 percent of households where the water supply
was cool. Seasonal factors rarely were cited as the reason for cold temperatures
(in 5.7 percent of rural households reporting cold water), but more often as the
reason for cool temperatures (in 21.2 percent of households reporting cool water).
In contrast to the reasons given for cold and cool water, the reasons for warm
-------
V - 266
Table V-14
Reported Reasons for Perceived Temperature of Rural
Household Water Supplies
Reason
Nation
Number of Percent of
Households Households
Natural temperature of the water
13,389,000
60.9
Seasonal factors
2,754,000
12.5
Pipes are too close to the surface
of the ground
835,000
3.8
Pipes pass too close to a heating
device
424,000
1.9
Storage tank is in direct sunlight
or otherwise heated
313,000
1.4
Other miscellaneous
367,000
1.7
Don't know
3,889,000
17.7
Total
21,974,000
99.9
Each figure in the table is rounded independently; numbers and
percentages may not equal rounded totals.
-------
V - 267
water most often were related to the physical structure of the supply. That is,
pipes were thought to pass too closely to heating devices in 18.9 percent of
households reporting warm water, and pipes were thought to be laid too closely to
the surface of the ground in 30.1 percent of those households. In further contrast,
seasonal factors were cited more often at households reporting warm water
temperatures (25.0 percent of these households).
Agreeability of water supply temperature
In general, the temperature of the water supply was found to be agreeable.
It was described as satisfactory or better at fully 96.0 percent of rural households.
A moderate dislike of the water temperature was reported at 3.0 percent of house-
holds, and a strong dislike for it was stated at only 0.9 percent.
Clear preferences about water temperature were also indicated. As would
be expected, a water supply that was usually cool or cold was unquestionably
preferred, with cold water preferred more strongly. A moderate to strong dislike
for warm water also was observed frequently.
Major subnational patterns in temperature
Regional variations in reported household water temperatures seemed to be
related to climatic differences (see Figure V-34a). Water supplies in the Northeast
were predominantly cold (63.3 percent of households), with only 2.2 percent
reported to be warm. The North Central showed much the same pattern, though a
smaller 'proportion of households reported that the supply was usually cold (53.8
percent). The South had the smallest proportion of households having cold supplies
(17.6 percent). Overall, warm water supplies were infrequent, but they were
reported by about equal proportions of households in the South and West (8.2
percent and 8.8 percent, respectively). Most households in the South and West
reported cool water (74.1 percent in the South and 54.8 percent in the West).
-------
V - 268
Figure V-34a
Regional Variation in Perceived Temperature
of Rural Household Water Supplies
0 10 20 30 40 50 60 70 80 90 100
11 1 1 l 1 1 1 1 1 1 1 1 1
NORTHEAST
NORTH
CENTRAL
SOUTH
WEST
2,777,000
i i
In.,! I l I l i i J J i
0 10 20 30 40 50 60 70 80 90 100
Percent of Households
KEY:
cold
cool
warm
-------
V - 269
Changes in water temperature were reported by 33.6 percent of households
in the West, 28.9 percent in the South, 21.8 percent in the Northeast, and 17.1
percent in the North Central.
A comparison of SMSA and nonSMSA households showed a variation of
three percentage points or less from national estimates of the usual temperature of
rural household water supplies. Likewise, the same proportions of SMSA and
nonSMSA households reported changes in the usual temperature of the supply.
Slight differences were observed when households in large rural communi-
ties, small rural communities, and other rural areas were compared. With
decreasing place size, the proportion of cold supplies increased and the proportion
of warm supplies decreased. For example, in large communities, 30.5 percent of
household supplies were described as cold and 11.4 percent as warm. Among
households in small communities, 35.4 percent were described as cold and 6.4
percent as warm. In other rural areas, 39.2 percent were described as cold and 5.2
percent as warm- Changes in the usual temperature of the water supply were more
common in large rural communities, where one out of every three households
reported temperature fluctuations, compared to one in four in both small rural
communities and other rural areas.
Striking differences were observed in the usual temperature of the water
supply among households served by systems of different sizes (see Figure V-34b).
Among households served by individual systems, 53.8 percent reported cold water,
44.0 percent reported cool water, and 2.1 percent reported warm water. Among
households served by intermediate systems, the proportions were 44.2 percent cold,
50.9 percent cool, and 4.9 percent warm. Households served by community systems
showed a very different pattern, however: only 24.1 percent reported cold
temperatures, 66.8 percent reported cool temperatures, and 9.2 percent reported
warm temperatures.
-------
V - 270
Figure V-34b
Size-of-System Variation in Perceived Temperature
of Rural Household Water Supplies
0 10 20 30 40 50 60 70 80 90 100
1 11 1 1 I 1 1 1 ! 1 1 1 I 1
INTERMEDIATE
2,228,000
uY<\ A'.\WaW>\WAY- >s< W
COMMUNITY
10.981,000
0 10. 20 30 40 50 60 70 80 .90
Percent of Households
100
KEY:
cold
cool
warm
-------
V - 271
The proportions of households reporting changes in the usual temperature
of the water supply increased as system size increased. This finding was consistent
with the size-of-system differences in the usual temperature of the water supply.
That is, a large proportion of households served by community systems (one out of
every three) reported changes in water temperatures. On the other hand, water
temperature changes were reported much less frequently among households served
by individual systems (17.1 percent) and intermediate systems (19.8 percent).
Seasonal variation in perceived water quality
Household respondents were asked if they noticed seasonal variation in
their water supplies with regaad to odor, taste, cloudiness, color, sediment, or
temperature. If any seasonal variation was identified, respondents were then asked
to describe what the usual condition was during each of the four seasons. The
possible responses corresponded with those characterizing respondents' initial
evaluations of usual water supply conditions—for example, no taste, slight taste, or
strong taste.
In general, very little seasonal variation was noticed for any of the
characteristics, with the exception of temperature. As to the water's odor, taste,
color, cloudiness, and sediment content, seasonal variation in each characteristic
was noticed in only about one million rural households. In contrast, 7.6 million, or
about one-third of all rural households, reported temperature fluctuations resulting
from seasonal factors.
• Strong odors occurred more frequently during the summer season, as did
strong tastes. Strong odors were perceived in 0.3 million water supplies during the
summer months, and strong tastes in OA million. In contrast, strong odors and
tastes were perceived in only about 0.1 million water supplies during the winter
months. Slight odors and tastes were noticed with equal frequency during each of
the four seasons.
-------
V - 272
Seasonal variations in the temperature of household water supplies were
just as one would expect: cold water temperatures were reported most often in the
winter months, slightly less frequently during the spring, and so forth.
Slight or heavy cloudiness was perceived most often during the winter and
spring. Slight cloudiness was reported in OA million rural water supplies during the
winter and 0.3 million during the spring. Very cloudy conditions were reported in
approximately 0.1 million water supplies during both the winter and spring.
Color was present in water supplies relatively infrequently during the
winter months. A very colored condition was perceived in only 24,000 water
supplies in the winter, compared to 0.2 million in the summer. Slight color was
perceived in 0.2 million water supplies in winter, but in approximately 0.3 million
during spring, summer, and fall.
Sediment occurred most frequently in summer and winter. In summer, 0.2
million supplies had heavy sediment and OA million had moderate sediment. In
winter, 38,000 supplies had heavy sediment and 0.3 million had moderate sediment.
Subnational comparisons of seasonal variation were not drawn because of
the small number of households reporting seasonal variation in water conditions.
QUANTITY
Beginning with this section, the focus of Chapter V shifts away from water
quality to other dimensions of household water conditions—quantity, availability,
cost, affordability, and health effects. As to quantity, competing demands on
water resources, diminishing underground water reserves, and occasional localized
droughts have prompted new interest in the demand for water. Domestic
consumption, though it constitutes only a small portion of the total demand for
water, is a focus of attention because it requires water of especially high quality.
In fact, the quantity of easily obtained drinking water is limited, particularly in
certain geographical areas of the US. Less abundant sources in those areas can
-------
V - 273
cause water supply problems for some rural households, no matter what type of
water supply is used. Households which are dependent upon one supply of water for
all uses (both indoor and outdoor) generally run the greatest risk of experiencing
quantity problems. People who supply their own water sometimes need elaborate
pumping and storage equipment to obtain a reliable supply; for these people, the
equipment itself imposes mechanical limitations on water quantity.
NSA investigators focused on these considerations by exploring domestic
water use, supply capacity, and users' subjective judgments about the amount of
water available. Potential water use by households using individual supply systems
was estimated according to the supply's pump capacity, the effective volume of
pressurized storage tanks, or the capacity of other storage tanks. As to households
served by community systems (fifteen or more connections), a check of the billed
meter readings provided an indication of water consumption. But not all systems
were metered, and for those which were, it was not always possible to obtain
sufficiently detailed information about quantity from the bills. Potential water use
by persons who used hauled water or purchased bottled water as their major supply
was estimated by the amount of water hauled or purchased (as reported in Chapter
IV). Users' judgments about the quantity of their supplies were explored in a series
of direct questions (see "Perceived quantity," below).
RECORDED QUANTITY
About 4.8 million households received bills which reported the total volume
of water delivered. However, since the billing periods varied, the average daily
consumption at each household had to be computed, and only 4.5 million households
had bills with sufficient information for this computation. This meant that fairly
exact measurements of water consumption were available for roughly 20 percent of
all rural households. Furthermore, for some households with more than one supply,
-------
V - 274
consumption was slightly underestimated since estimates did not include the
quantity of water used from supplemental supplies.
The range of average daily consumption in households served by community
systems was striking. Some households averaged as little as twelve liters (three
gallons) per day, others as much as 5,123 liters (1,352 gallons) per day. The mean
daily consumption per household was 829 liters (219 gallons). The median was 664
liters (175 gallons) per day.
Taking into account the number of people residing in the household, daily
per capita consumption ranged from a low of twelve liters (three gallons) to a high
of 2,562 liters (676 gallons). (Such very high consumption figures may have been
caused by the fact that many households had only one supply of water, which they
used for both indoor and outdoor purposes. Some households also used the
household supply for agricultural uses such as irrigation and watering livestock.)
The mean consumption rate was 285 liters (75 gallons), and the median 227 liters
(60 gallons), per person per day. These consumption figures were based on water
bills for all four seasons, but since most of the interviews were conducted during
the summer, estimates adjusted for seasonal differences would probably be lower.
The majority of rural households were not connected to metered systems,
and for them, the NSA had no direct measure of water quantity. For these
households, quantity could' be described only in regard to devices which had a
bearing on the volume of water available to the household. Capacities of pressure
tanks, storage tanks, and pumps all were considered to affect both the quantity and
availability of water. That is, they influenced the total volume of water a
household could obtain, as well as the reliability of the supply.
There were roughly 9.8 million rural households which had on-premises
water pumps for their water supplies. The average pump capacity was 41 liters
(eleven gallons) per minute. The median value was 22 liters (six gallons) per
minute. Most of these households (9.5 million) also had at least one pressure tank
-------
V - 275
(about 1 percent had more than one pressure tank). Pressure tank capacities
ranged from roughly two liters (one-half gallon) to over 10,000 liters (over 2,600
gallons). Tanks held an average (mean) of 219 liters (58 gallons); the median
capacity was 114 liters (30 gallons).
An attempt also was made to measure the effective volume of pressure
tanks. This involved running the water until the pump started to operate. All of
the water outlets then were closed, and the pressure tank was allowed to fill.
After the pump turned off, indicating the pressure tank was fully charged, one tap
was opened and the water volume was measured until the pump came on again.
This volume of water was taken to be the effective volume of the pressure tank.
Unfortunately, the construction and layout of some systems precluded the mea-
surement since there was no reasonable way to monitor the pump operation.
Among the approximately nine million households where the effective volume of
the pressure tank could be determined, the average effective volume was 51 liters
(fourteen gallons) while the median was thirteen liters (three gallons). About 4
percent of these tanks were found to be completely waterlogged. That is, the
cushion of compressed air in the tank which was supposed to provide the pressure
was entirely dissipated. In these households, the pump went on every time a tap
was opened. The largest effective volume encountered was 800 liters (211 gallons).
Auxiliary storage tanks (not pressure tanks) were relatively rare. Only 4.3
percent of all rural households had storage tanks. When they were present, they
tended to be large. The average size was about 3,500 liters (925 gallons); the
median was about 760 liters (200 gallons).
Pressure tanks are a more or less standard feature of many household
supplies, but storage tanks are a different matter. If a household has an on-
premises storage tank, it is probably because the system does not provide an
adequate quantity of water on demand. Storage tanks in general represent an
attempt to ensure sufficient quantities of water when needed. The fact that 4.3
-------
V - 276
percent of rural households had such devices does not mean that only 4.3 percent
had a water quantity or reliability problem. However, it can be taken as an
indication that those households with storage tanks could afford this fairly
expensive method of supply stabilization. The sections later in this chapter on the
perceptions of rural people regarding quantity and availability provide other
indications of the extent of water quantity problems throughout the US.
Subnational variation in recorded quantity
There were large differences in consumption patterns from one region to
another. The median daily per capita consumption among households which were
metered and billed for their water was 188 liters (50 gallons) in the North Central,
212 liters (56 gallons) in the Northeast, and about 234 liters (62 gallons) in both the
South and West. A comparison of means (rather than medians) showed the West
with the highest average per capita daily consumption (307 liters, or 81 gallons).
Average figures in other regions were 290 liters in the South, 275 liters in the
Northeast, and 250 liters in the North Central. The larger average consumption
figure in the West reflected the greater frequency, relative to the other regions, of
households using very large quantities of water. In turn, the use of large quantities
of water was related to several factors: lower levels of precipitation and
coincident supplemental watering of lawns and gardens, the use of swamp coolers,
and the general unavailability of supplemental supplies.
There was little difference between SMSA and nonSMSA households in daily
per capita consumption, or among households in large communities, small commun-
9
ities, and other rural areas: median daily per capita usage was uniformly about 230
liters (61 gallons). No size-of-system comparison could be made, since only
households served by community systems could provide billing information which
included consumption figures.
-------
V - 277
Two-thirds of households in the West were connected to community
supplies, but among those that used wells or surface supplies, the potential for
greater per capita consumption still was evident—at least on the basis of pump
capacity. The median pump capacity in the West was 48 liters (thirteen gallons)
per minute. The other regions, by contrast, were fairly uniform at roughly twenty
liters (five gallons) per minute. Pressure tanks in the West had nearly double the
effective volume of that in other regions (23 liters). Storage tanks also tended to
be large in the West, with a median size of 3,777 liters (997 gallons), compared to
229 liters (61 gallons) in the Northeast and 568 liters (150 gallons) in the North
Central. The South, however, actually had the largest median storage tank
size—3,791 liters (1,000 gallons). In the other three regions, slightly more than 5
percent of the households had storage tanks, compared to about 3 percent in the
South.
There were no substantial differences between SiMSA and nonSMSA house-
holds regarding pump capacity or with respect to pressure tanks and storage tanks.
In the size-of-place comparison, however, it was found that households in other
rural areas were more likely to have these devices. (The majority of households in
other rural areas had individual wells which use such devices.) The most notable
other difference was that storage tanks were used in fewer than 2 percent of
households in small communities, but almost 5 percent of households in other rural
areas had them. Also, in other rural areas, the median storage tank size was 758
liters (200 gallons), about twice that in small rural communities. (There were too
few storage tanks in large-rural-community households to permit a comparison.)
9
Comparing households served by systems of different sizes, only 0.3
percent of households served by community systems had supplementary devices at
the household such as pumps and storage tanks. Between households using
individual systems and those using intermediate systems, the major differences
pertained to the size of the devices that were used. Among households using
-------
V - 278
intermediate systems, pumps had a larger median capacity (26 liters per minute,
compared to twenty liters per minute), a larger median size of pressure tank (152
liters, compared to about 114 liters), and substantially larger storage tanks. About
8 percent of households using individual systems had a storage tank; the median
capacity was 758 liters (200 gallons). Among households using intermediate
systems, 9.4 percent had a storage tank; the median capacity was 1,895 liters (500
gallons).
PERCEIVED QUANTITY
The volume of water used in a household is inextricably tied to people's
beliefs about the ability of the water supply to provide enough water. Mistaken
beliefs could lead the members of a household to restrict water usage unnecessar-
ily, or to continue normal usage when the supply is actually low and requires
conservation.
In the NSA, most rural households reported having ample water supplies
(see Figure V-35). Of all rural households, 17.7 million (80.7 percent) reported that
the major household supply completely satisfied their water requirements. In
another 1.1 million households (5.0 percent), the supply almost always provided
enough water. Further, almost 2.5 million households (11.3 percent) reported that
the supply usually provided as much water as wanted. However, in some
households, the supply usually did not provide as much water as people wanted
(470,000 households, or 2.1. percent), and supplies at some households never
provided an acceptable quantity of water (206,000, or 0.9 percent).
Little fluctuation was noticed in the amount of water readily available
from household supplies. At 85.5 percent of households, no change was reported.
However, some change in water quantity was noticed in 3.2 million households (14.4
percent).
-------
V - 279
Figure V-35
Perceived Quantity of Rural Household Water Supplies
all the water
/ ever want
almost always all
the water / want
usually as much
water as / want
usually not as much
water as / want
never as much
water as / want
Number of Households (in millions)
2.2 4.4 6.6 8.8 11.0 13:2 15.4 17.6 19
! -I !- — 1
10
T
.8 22.0.
T
20 30 40 50 60 70 80 90 100
Percent of Households
-------
V - 280
Reasons for insufficient quantity
Of those households that reported receiving less water than they needed, or
that reported changes occurring in the amount usually provided by the supply, 39.0
percent or 1.6 million households attributed the problem to inadequacy of the
physical facilities of their supply. That is, the pipes, the storage tank, or some
other feature of the supply was inadequate. Another 16.1 percent (680,000
households) attributed the problem to a breakdown in the system. Weather
conditions reportedly caused the problem at 22.6 percent of the households, or
951,000. Interestingly, though one might expect quantity problems to be a result of
an insufficient aquifer or other source, the problem was blamed on an inadequate
supply at only 6.3 percent of the households (265,000). Other reported reasons for
the condition are given in Table V-15.
Agreeability of water supply quantity
The quantity of the household water supply generally was seen as satisfac-
tory. The amount of water obtained throughout the year was liked in 46.6 percent
of all rural households (10.2 million); it was liked very much in 31.2 percent (6.9
million); and it was considered all right, or not thought about much, in 17.8 percent
of rural households (3.9 million). The quantity of water available to the household
was disliked in only one million rural households (4A percent).
Major subnational patterns in perceived water quantity
Although all four regions showed about equal proportions of households
reporting that the water supply never provided enough water, the West stood out as
having more frequent problems with quantity than the other three regions (see
Figure V-35a). Proportionately, many fewer households in the West reported
always getting enough water from the household supply (73.8 percent, compared to
83.4 percent in the Northeast, 82.9 percent in the North Central, and 80.1 percent
-------
V - 281
Table V-l5
Reported Reasons for Perceived Insufficiency in Quantity of Water
Provided by Rural Household Water Supplies
Nation
Reason
Number of
Households
Percent of
Households
Inadequacy of the physical
facilities of the system
1,643,000
39.0
Weather conditions
951,000
22.6
A breakdown in the physical facilities
of the water system outside the house
680,000
16.1
Inadequate supply
265,000
6.3
Seasonal factors
151,000
3.6
Deliberate or planned activities
145,000
3.4
A problem within the house
72,000
1.7
Mismanagement of the system
60,000
1.4
Other miscellaneous
140,000
3.3
Don't know
107,000
2.5
~Total
4,214,000
99.9
~Table includes only those households which reported that the
water supply provided an insufficient quantity of water.
Each figure in the table is rounded independently; numbers and
percentages may not equal rounded totals.
-------
V - 282
Figure V-35a
Regional Variation in Perceived Quantity
of Rural Household Water Supplies
0 10 20 30 40 50 60 70 80 90 100
11 1 11 i 1 1 1 1 1 1 1 1 1
NORTHEAST
3,693,000
NORTH
CENTRAL
6,213,000
¦¦ • • >: • '
SOUTH
^ 1
WEST
2,777,0
0 10 20 30 40 50 60 70 80 90 100.
Percent of Households
KEY:
all the water / ever want
almost always all the water / want
usually as much water as / want
usually not as much water as / want
never as much water as / want
-------
V -283
in the South). The proportions reporting they almost always got enough water were
bigger in the South and West (6.0 percent and 7.5 percent, respectively) than in the
Northeast and North Central (4.4 percent and 2.6 percent, respectively). Supplies
that usually provided enough water occurred about evenly in the four regions (from
9.8 percent in the Northeast to 12.5 percent in the North Central). In the West, 7.2
percent (200,000 households) reported either that the water supply usually did not
provide enough water or that it never did, compared to 2.4 percent in the
Northeast, 1.9 percent in the North Central, and 2.8 percent in the South. (The
greater frequency of perceived quantity problems in the West may have been
related to the greater demand for water among households in the West (see
Chapter III, "Regional variation in uses of water").
Little or no variation from national estimates of perceived quantity was
apparent between households located within and outside of SMSAs, or among
households served by systems of different sizes. SMSA/nonSMSA variation
amounted to less than two percentage points; the size-of-system comparison
showed at most a difference of three percentage points in any particular category.
In the size-of-place comparison as well, there was little variation from
national estimates, except that in large rural communities, a larger proportion (9.2
percent, compared to the national estimate of 5.0 percent) reported almost always
getting enough water, and a smaller proportion (7.1 percent, compared to the
national estimate of 11.3 percent) reported usually getting enough water from the
household supply.
As to changes in the water supply with respect to quantity, a higher
proportion of households in the West (20.5 percent) reported variations in the
amount of water the supply provided. In the Northeast and North Central regions,
changes in the quantity of water provided by the household supply were reported in
roughly 11 percent of rural households. In the South, changes were noticed in 15.7
percent of all rural households.
-------
V - 284
No other subnational comparison showed substantive variations from
national estimates with respect to changes in the usual quantity provided by the
water supply.
Seasonal variation in perceived water quantity
Although seasonal factors rarely were cited as reasons for an insufficient
water supply, seasonal variations were perceived in approximately two million rural
households across the country.
It is clear from Table V-16 that quantity problems were noticed much more
frequently during the summer months. Of those households where seasonal
variation occurred, 36A percent—a total of 736,000 households—reported that
during the summer, the water supply usually or never provided as much water as
wanted. On the other hand, water was in abundance during the other seasons:
supplies were reported to be at least usually ample in more than 89 percent of the
households reporting seasonal variation.
A V AIL ABILITY
To completely satisfy household needs, a water supply must be capable of
providing a sufficient volume of water on a continuous basis. If enough water
cannot be obtained, even temporarily, the supply's availability is limited. Such
limitations, whether they occur on an intermittent or protracted basis, may have
significant consequences for the members of a household. Many sorts of adjust-
ments may have to be made, such as adapting normal patterns of water use and
consumption to the availability of the supply, or acquiring water from an alternate
supply when the major supply is not providing enough. Thus, availability is a
separate issue that must be addressed when evaluating rural water conditions,
independent of water quality and distinct from the quantity of water produced by a
supply.
-------
V - 285
Table V-16
Seasonal Variation in Perceived Quantity of Rural
Household Water Supplies
Percent of Households
Quantity Provided
Spring
Summer
Fall
Winter
All the water ever wanted
54.7
10.2
49.1
66.1
Usually as much water as wanted
39.6
53.4
40.2
23.2
Usually not as much water
as wanted
4.4
29.5
7.3
6.9
Never as much water as wanted
1.3
6.9
3.0
3.5
Don't know
0.0
0.0
0.3
0.3
*Total
100.0
100.0
99.9
100.0
~Table includes only those households which reported seasonal variation in the
quantity of water provided by their water supplies.
-------
V - 286
Conceptually, availability has two components. First, availability may be
defined in terms of reliability, a common measure of supply performance. Specific
to reliability are various kinds of supply interruptions—those that are the result of
intentional actions such as maintenance work, and those that result from break-
downs. The second component of availability pertains to the supply's accessibility,
or the difficulty of obtaining water from the supply. Each of the two components
reflects the ability of a supply to provide water whenever it is needed, which is the
central concern in any definition of availability.
RELIABILITY
In the NSA, reliability was measured with reference to both intentional
supply interruptions and breakdowns. An interruption was defined as any tempo-
rary loss of a supply, whether it occurred because of scheduled activities such as
routine maintenance and repair operations or because of some other factor. A
breakdown, on the other hand, referred to a reduction or loss of the water supply
which was caused by some unanticipated event, such as equipment failure, operator
error, a flood, or an earthquake. According to these definitions, interruptions and
breakdowns are analogous to power outages, a concept that is employed in the
electrical industry to indicate service quality. However, rather than expressing
reliability as a ratio of the number of hours of unsatisfactory service to the total
number of hours during a given period, as electrical outages are commonly
reported, the NSA simply enumerated breakdowns and other supply interruptions
which were reported at the household. The information on breakdowns was
compiled for all households, while inquiries about interruptions were restricted to
households at which the supply loss was serious or frequent enough to prompt the
installation or use of another supply.
With respect to breakdowns, about 16.3 million households in the rural US
(74.2 percent) reported that there had been no breakdowns in their water supplies
-------
V - 287
during the year prior to the NSA (see Table V-17). However, approximately 5.6
million supplies, or 25.7 percent, did break down at least once. There had been a
single breakdown in 15.1 percent of all rural supplies (3.3 million), while two or
more occurred in 10.6 percent (2.3 million). Approximately 1.5 percent of supplies
(330,000) broke down six times or more. The maximum number of breakdowns
reported was 52, or an average of one per week.
When supply breakdowns occurred, they tended to be lengthy enough to be
considered severe. In the NSA, a severe breakdown was defined as a loss of supply
which lasted more than six consecutive hours during the previous year. More than
3.2 million households —15 percent of all rural households, or slightly more than 57
percent of those that reported supply breakdowns—indicated that one or more of
the breakdowns had been severe. (For the vast majority of these households,
however, only a single such incident had occurred.) Fewer than 161,000 supplies
had had four or more severe breakdowns.
Generally, breakdowns did not occur within the housing structure. At
about 83 percent of the 5.6 million households where supply breakdowns occurred
(4.6 million), the problem arose outside of the building; at 17 percent (952,000), the
problem arose inside. The pattern was the same for those 3.2 million households
where severe supply breakdowns occurred: about 81 percent of these households
(2.6 million) reported that the breakdowns occurred outside the structure, and 19
percent (600,000) reported they happened inside. Therefore, whether severe or not,
breakdowns were substantially more often attributed to some aspect of the supply
external to the household rather than to the household plumbing and treatment
facilities.
Rural households reacted to water supply breakdowns in different ways,
depending upon when the breakdown occurred, its duration, and other factors. The
most common response was to simply wait until the supply was restored, which was
what about 48 percent of the 5.6 million households (2.7 million) did. Slightly more
-------
V - 288
Table V-17
Frequency of Rural Water Supply Breakdowns
During Previous Year
Nation
Frequency of
Breakdowns
Number of
Supplies
Percent of
Supplies
None
16,299,000
74.2
1
3,324,000
15.1
2
1,121,000
5.1
3
520,000
2.4
4
251,000
1.1
5
106,000
0.5
6 or more
324,000
1.5
Unspecified
29,000
0.1
Total
21,974,000
100.0
Each figure in the table is rounded independently;
numbers and percentages may not equal rounded totals.
-------
V - 289
than 32 percent (1.8 million) obtained water by borrowing it temporarily from
friends, neighbors, or relatives, while 16.0 percent (900,000) performed the
necessary repairs to restore the supply. About 5 percent of the households in which
breakdowns occurred (580,000) were able to acquire water from another supply on
the premises, and approximately 3 percent hauled water from an off-premise
supply (168,000). The least common response to a breakdown was to purchase
water from a commercial establishment; less than 2 percent of the households did
this (112,000). Although most of these households had a single, characteristic
reaction to a supply breakdown, certain households reported doing various things
(which caused the percentages to exceed 100 percent).
Although severe breakdowns occurred at 3.2 million households, substan-
tially fewer households had gone to the trouble and expense of installing a more
reliable supply. In fact, across the nation, only about 1 percent of all households
used another water supply because the major supply was frequently or chronically
interrupted. While comparatively small, this proportion represented about 221,000
rural households.
To summarize, 5.6 million of the supplies from which rural households
obtained their water broke down one or more times during the year before the
NSA. Slightly more than 3.2 million rural households (about 15 percent) reported
breakdowns that were considered severe by the NSA definition. Independent of
severity, the vast majority of the 5.6 million households that reported breakdowns
(about 83 percent, or 4.6 million) attributed them to problems outside of the
household. When a breakdown occurred, most households (4.3 million) waited until
the supply was repaired or borrowed water from relatives, friends, or neighbors.
Regional variation in reliability
According to regional compilations, there were substantial differences in
the proportions of households where water supply breakdowns occurred, and in the
-------
V - 290
number of breakdowns typically reported (see Table V-18). In the West, for
example, 20.5 percent of all rural households experienced one or more water supply
breakdowns, while in the South 29.2 percent did. Further, about 14 percent of the
households in the South reported two or more breakdowns, compared to only 6.7
percent of the households in the West. In addition, household supplies in the South
tended to break down more often than supplies in the other regions. This tendency
was observed in spite of the fact that the supplies of a very small proportion of the
households in the North Central broke down as many as 52 times, while the
maximum number of supply breakdowns reported by households in the South was 25.
Compared to the number of households that noted supply breakdowns and
compared to the number of rural households in each region, the number of
households that reported severe breakdowns showed appreciable regional variation.
Approximately 59.3 percent of the households in the North Central where break-
downs occurred reported the breakdowns to be severe, compared to 54.6 percent in
the West, 51.0 percent in the South, and 49.8 percent in the Northeast. Alterna-
tively, counting all rural households in each region, 17.1 percent of supplies in the
South sustained at least one severe breakdown, while severe breakdowns were
limited to 10.8 percent of the Northeast's rural households. (Severe breakdowns
were experienced by 11.1 percent of households in the West and 15.0 percent in the
North Central.) This finding suggests that household supplies in the South were
more likely to sustain severe breakdowns than those in the Northeast, North
Central, and West.
There was considerable regional variation in the point of origin of the
supply breakdowns—that is, whether they arose within or outside the housing
structure. Considering all rural households that reported breakdowns, whether
severe or not, slightly more than 30 percent in the Northeast indicated that the
breakdowns occurred within the dwelling unit, while about 70 percent reported that
the breakdowns took place outside. This was in sharp contrast to the West, where
-------
V -291
Table V-18
Regional Variation in Frequency of Riral Water Supply
Breakdowns During Previous Year
Percent of Supplies
Frequency of
Breakdowns
Northeast
North
Central
South
West
None
78.2
74.6
70.6
79.4
1
14.6
15.4
15.6
13.8
2 or more
7.2
9.9
13.6
6.7
Unspecified
0.0
0.1
0.2
0.1
Total Percent
100.0
100.0
100.0
100.0
Total Supplies
3,693,000 6,
213,000
9,291,000
2,777,000
-------
V - 292
about 2 percent reported breakdowns inside the dwelling unit and roughly 98
percent reported breakdowns outside the dwelling unit. When only severe break-
downs were considered, the same pattern emerged, except that instead of the
Northeast showing the highest proportion of breakdowns inside the house, the North
Central did. In the North Central, 32.5 percent of the households where
breakdowns occurred reported they originated inside the house.
Finally, there were noticeable regional differences in households' reactions
to supply breakdowns. About 58 percent of the households in the West that
reported supply breakdowns simply waited until the water supply resumed opera-
tion, compared to M.O percent of the households in the Northeast. In the West,
only 6.4 percent of the households that reported breakdowns repaired the supply,
while 28.3 percent of those in the Northeast did. Although borrowing water was a
common response in all regions, it was most prevalent in the South, where 35.7
percent of the households with inoperable supplies borrowed water temporarily.
Using an alternate supply was most common in the West, where that action was
taken by 9.9 percent of the households with supply breakdowns. In the Northeast,
6.2 percent of the households that reported breakdowns hauled water when their
major supply was unavailable.
Only slight regional variations were detected in the proportions of rural
households that had installed an alternate supply because the major supply was
frequently or chronically subject to breakdowns. The proportions ranged from 0.5
percent in the South to 2.2 percent in the West.
In summary, household supplies in the South and North Central tended to be
less reliable in terms of breakdowns than supplies in the other two regions. About
29 percent of the households in the South, and 25.3 percent of those in the North
Central, reported supply breakdowns, compared to the national total of 25.7
percent. Severe breakdowns also tended to be more prevalent in the South and
North Central, where they were reported by 17.1 percent and 15.0 percent of the
-------
V - 293
households, respectively. While the majority of households in each region indicated
that supply breakdowns originated outside of the structure, a much larger propor-
tion of the households in the West (about 98 percent) attributed breakdowns to that
source. Households in the West also reflected more of a tendency to either wait
until the water supply was restored (approximately 58 percent) or to obtain water
from an alternate supply (9.9 percent), compared to households in the other
regions. For all regions, however, waiting was the most common response to a
supply breakdown, and borrowing was the second most common.
SMSA/nonSMSA variation in reliability
Relative to the regional differences that were identified, the variation in
SMSA and nonSMSA households with respect to supply breakdowns was insignifi-
cant. At least one supply breakdown was reported at 24.2 percent of SMSA
households and at 26.4 percent of nonSMSA households. Likewise, the proportions
of SMSA households that had one supply breakdown and those with two or more
breakdowns (13.5 percent and 10.7 percent, respectively) were approximately the
same as the proportions for nonSMSA households (15.9 percent and 10.5 percent).
The largest difference between SMSA and nonSMSA households was observed in the
maximum number of breakdowns reported, which was 52 for SMSA households and
25 for nonSMSA households.
Severe supply breakdowns occurred at 12.4 percent of all SMSA households
and 15.7 percent of all nonSMSA households. These proportions represented 51.3
percent of the SMSA households that reported supply breakdowns and 59.6 percent
of nonSMSA households that reported breakdowns. With respect to the origin of
breakdowns inside or outside the house, there was a difference between SMSA
households and nonSMSA households of less than three percentage points.
As for responses to breakdowns, no appreciable differences could be seen
between SMSA and nonSMSA households except that a larger proportion of SMSA
-------
V - 294
households waited until the water supply became available (53.5 percent, compared
to 45.4 percent) and a greater proportion of nonSMSA households made repairs to
the water supply (17.6 percent, compared to 11.0 percent). That is, approximately
the same proportions of SMSA and nonSMSA households that reported supply
breakdowns either borrowed water, purchased water, hauled water, used an
alternate water supply, or obtained water by some other means.
Size-of-place variation in reliability
Generally, households in small rural communities reported more supply
breakdowns than households in large rural communities or other rural areas (see
Table V-19). Slightly less than 30 percent of households in small communities
reported one or more breakdowns, compared to 22.2 percent of households in large
communities. Additionally, 13.7 percent of households in small communities
reported two or more breakdowns, compared to 11.5 percent in large communities
and 10.2 percent in other rural areas. In large communities, the maximum number
of breakdowns reported was eighteen; in small communities, it was eleven; in other
rural areas, it was 52.
The data on severe breakdowns reflected roughly the same pattern of
variation. Specifically, only 8.1 percent of all rural households in large communi-
ties reported severe breakdowns, compared to 16.6 percent of the households in
small communities and 15.4 percent in other rural areas. These figures represented
36.4 percent of the households in large rural communities with supply malfunctions,
55.7 percent in small communities, and 59.5 percent in other rural areas. As with
supply breakdowns in general, severe breakdowns occurred more frequently in small
communities.
The origin of breakdowns inside or outside the house varied substantially
according to the size of the place where households were situated. For example,
5.7 percent of those households in large communities where breakdowns occurred
-------
V - 295
Table V-19
Size-of-Place Variation in Frequency of Rural Water Supply
Breakdowns During Previous Year
Percent of Supplies
Frequency of
Large Rural
Small Rural
Other Rural
Breakdowns
Communities Communities
Areas
None
77.8
70.3
74.0
1
10.7
16.0
15.6
2 or more
11.5
13.7
10.2
Unspecified
0.0
0.0
0.2
Total Percent
100.0
100.0
100.0
Total Supplies
2,369,000
1,509,000 18
,095,000
-------
V - 296
reported that the breakdowns originated inside the dwelling unit, while 94.3
percent reported that they originated outside. In small communities, 13.1 percent
originated inside the dwelling and 86.9 percent originated outside. Among
households in other rural areas, the comparable percentages were 19 percent inside
and 81 percent outside. When only households with severe breakdowns were
considered, the percentages for households in other rural areas did not change
appreciably, but among households in large communities, those originating inside
increased to 12.6 percent, and those originating outside decreased to 87.4 percent.
Among households in small communities, the percentages changed similarly; the
change amounted to six percentage points rather than seven.
Although waiting was the most frequent response to a supply breakdown
among households in rural communities and in other rural areas, the proportions
differed considerably. Slightly less than 72 percent of the households in large
communities that reported breakdowns simply waited until the supply became
available again, compared to 61.5 percent of the households in small communities
and about 44 percent of the households in other rural areas. Likewise, the
proportions of households in other rural areas that repaired the supply or borrowed
water were significantly higher than in large or small communities. More
specifically, the percentage of households that made repairs to a supply or
borrowed water in response to a breakdown varied from 51.0 percent of households
in other rural areas to around 35 percent of the households in small communities
and 30 percent of those in large communities. No major differences besides these
could be discerned in reactions to supply breakdowns at households in large
communities, small communities, and other rural areas.
In summary, the supplies of households in small rural communities tended
to break down more frequently than supplies providing water to households in large
communities or other rural areas. About 30 percent of the households in small
rural communities reported one or more supply breakdowns, compared to 26.0
-------
V - 297
percent in other rural areas and 22.2 percent in large rural communities.
Consistent with this pattern, households in small rural communities also reported
an appreciably higher incidence of severe supply breakdowns. Additionally, a much
larger proportion of households in large rural communities simply waited until the
supply was repaired, while proportionately more households in other rural areas
either initiated repair operations or borrowed water.
Size-of-system variation in reliability
In relative terms, the supplies of households that obtained water from
intermediate systems tended to be less reliable than the supplies of households
attached to individual or community systems (see Table V-20). Approximately 33
percent of households served by intermediate systems reported at least one
breakdown, compared to 26.4 percent of households served by community systems
and 22.8 percent of households using individual systems. Additionally, 21.4 percent
of households on intermediate systems reported severe supply breakdowns, a
proportion substantially greater than the 16.0 percent of households on individual
systems and the 12.3 percent of households on community systems. Consistent
with these differences, the supplies of households on intermediate systems broke
down more frequently in general than supplies of other households, although the
maximum number of breakdowns reported—52—occurred among households on
community systems. Finally, while breakdowns were least prevalent among house-
holds on individual systems, these breakdowns tended to be of longer duration.
That is, breakdowns were severe 70.0 percent of the time among households served
by individual systems, compared to 65.0 percent of the time among households
served by intermediate systems and 46A percent of the time among households
served by community systems.
Although breakdowns generally originated outside of the dwelling unit, the
predominance of such breakdowns varied by the size of the water supply system
-------
V -298
Table V-20
Size-of-System Variation in Frequency of Rural Water Supply
Breakdowns During Previous Year
Percent of Supplies
Frequency of
Individual
Intermediate
Community
Breakdowns
System
System
System
None
77.2
66.9
73.3
1
15.9
23.0
12.9
2 or more
6.9
10.1
13.5
Unspecified
0.0
0.0
0.3
Total Percent
100.0
100.0
100.0
Total Supplies
8,765,000
2,228,000 10
,981,000
-------
V - 299
serving the household. Only 7.1 percent of the households on community systems
that reported breakdowns said the breakdowns originated within the house, for
example, while the remaining 92.9 percent noted that the breakdowns originated
outside the house. Among households served by intermediate systems, the
proportions were 19.3 percent inside and 80.7 percent outside; among households
served by individual systems, the proportions were 30.6 percent inside and 63.4
percent outside. Similar differences were observed in the data on severe
breakdowns, but they were not quite as pronounced.
Interestingly, household reactions to breakdowns differed considerably
depending on the size of the system serving the household. Only about 24 percent
of the households served by individual systems waited until the supply was restored,
compared to 34.4 percent of the households served by intermediate systems and
67.6 percent of the households served by community systems. Conversely, 44
percent of the households served by individual systems borrowed water, compared
to 36.4 percent of households served by intermediate systems and 23 percent of
households served by community systems. (These are reasonable findings, since
maintenance and repairs of individual systems are the responsibility of the
individual household, and since breakdowns in an intermediate or community
system would most often affect neighbors' supplies as well, unlike breakdowns in an
individual system.) In addition, as would be expected in light of the large
proportion of breakdowns originating outside of the house, only 5.0 percent of the
households that were served by community systems and that reported breakdowns
took action to repair the supply, while 29.2 percent of the households served by
individual systems did. Finally, a much larger proportion of the households on
intermediate systems—about 10 percent—hauled water when their principal supply
was unavailable.
To summarize, compared with households on individual or community
systems, a larger proportion of households on intermediate systems reported one or
-------
V - 300
more supply breakdowns. Also, a greater proportion of households on intermediate
systems reported breakdowns that were severe. While severe breakdowns were
least common for households on individual systems, the duration of those break-
downs tended to be longer. Comparing responses to breakdowns, households on
community systems generally waited until service was restored, while those on
intermediate systems borrowed water or waited, and those on individual systems
borrowed water or took action to repair the supply.
ACCESSIBILITY
The second dimension of availability is accessibility—the ease or difficulty
of obtaining water from the supply. A perfectly accessible supply provides water
when water is needed and does not pose an inconvenience to the user in either its
location or the operation of its equipment. Further, it provides water at a pressure
that accommodates the household's uses of water. Supplies that are inconvenient
or which do not generate sufficient pressure to satisfy household needs are failing
to some extent in providing water. Since the water flow they provide is
intermittent or unpredictable, they are not unlike supplies that break down
frequently. Both conditions require users to make adjustments that would be
considered unusual by households that have properly designed and functioning water
supplies.
In the NSA, accessibility was measured by several indicators. The concept
of convenience was approached first by estimating the distance between the point
at which water entered the dwelling unit and the point at which it was withdrawn
from the source. Second, the number of inconvenient supplies was estimated on
the basis of the number of household representatives who said they particularly
disliked the water supply because of its inconvenience. (This measure may have
included only extremely inconvenient supplies, however.) Another indicator of
accessibility was water pressure, which was measured with a pressure gauge;
-------
V - 301
household representatives also reported on the water supply's usual pressure, any
changes in pressure, and reasons for the pressure being too high or too low,
fluctuating, or changing. Household representatives also reported on seasonal
variation in pressure, and how much they liked or disliked the water supply's
pressure.
The data on the distance between the dwelling unit and the source of the
water supply applied to 10.6 million households whose water supplies consisted of
wells, springs, surface water, and cisterns. (For households served by community
water systems, the point of withdrawal from the source was defined as being
located on the premises of the dwelling unit. Also, for hauled and purchased
bottled supplies, the distance could not be determined because the source was not
specified; these supplies are discussed separately in Chapter IV.) Among the 10.6
million water supplies from which data were obtained, there was a great deal of
variation in supply accessibility as measured by this indicator. As shown in Table
V-21, almost 52 percent of supplies (5.5 million) had the point of withdrawal either
on the premises of the household (in the basement, for example), or within ten
meters (33 feet) of the household structure. For another 31.9 percent (3.4 million),
the point of withdrawal was eleven to 50 meters away (36 to 164 feet). Altogether,
90.2 percent of supplies (9.6 million) obtained water from a point 100 meters (328
feet) or less away from the dwelling unit. For 0.6 percent, or 63,000 supplies,
however, the point of withdrawal was more than 1,000 meters away from the
dwelling unit (more than half a mile). The maximum distance reported was 9,900
meters, or slightly more than six miles, and the median was nine meters (about 30
feet).
The major water supply at 2.0 percent of all rural households (442,000) was
disliked specifically for its inconvenience. The NSA did not quantify the
inconvenience or the magnitude of its effect on the households, but it can be
assumed that for these households it was necessary to spend abnormal amounts of
-------
V -302
Table V-21
Distance Between Rural Dwelling Unit and
Point of Withdrawal from Source
Nation
Distance
Number of
Supplies
Percent of
Supplies
On premises
643,000
6.1
1-10 meters
4,818,000
45.5
11-50 meters
3,373,000
31.9
51 - 100 meters
711,000
6.7
101 - 1,000 meters
642,000
6.1
More than 1,000 meters
63,000
0.6
Unspecified
332,000
3.1
* Total
10,582,000
100.0
~Table excludes community water supplies, hauled
supplies, and purchased bottled sipplies.
Each figure in the table is rounded independently;
numbers and percentages may not equal rounded
totals.
-------
V - 303
time and effort in obtaining the water supply, or that whatever efforts they could
expend did not result in a supply that was convenient or fully accessible.
The pressure of water supplies, as measured by a pressure gauge, ranged
from a minimum of two pounds per square inch (psi) to a maximum of 190 psi. As
indicated in Table V-22, about 7.5 percent of the 18.7 million supplies for which
measurements were recorded (1.4 million) had a pressure of twenty psi or less; 44.5
percent (8.3 million) had a pressure of 21 through 40 psi; 31.9 percent (six million)
had a pressure of 41 through 60 psi. Another 13.7 percent of supplies (2.4 million)
had a pressure of 61 through 100 psi, and the remaining 2.4 percent (455,000) had a
pressure higher than 100 psi. The median water pressure was approximately 39 psi.
(For various reasons, the pressure measurement was not performed at all house-
holds, and 490,000 households did not have pressurized water supplies, but pressure
was measured for 18.7 million households.)
The usual pressure of rural water supplies was generally felt to be
satisfactory. About 83 percent of the 21.5 million rural households with pressur-
ized supplies (17.8 million) reported that the pressure was usually about right, while
7.3 percent (1.6 million) said that it was too low and a small number of households
(1 percent or 215,000) complained that the pressure was too high. Although only
8.5 percent (1.8 million) reported that fluctuating pressure was the usual condition
of the supply, 22.7 percent (4.9 million) reported that the pressure changed from
time to time.
Of the 6.6 million households that reported high, low, or fluctuating
pressure conditions, or that reported the occurrence of pressure changes, 35.3
percent (2.3 million) indicated that the condition was fairly constant. For 64.1
percent (4.2 million), the condition was present or changed only "some of the time"
or "hardly ever."
As to reasons for the water supply's pressure, 50.8 percent (3.4 million) of
the 6.6 million households asked (all those where the pressure was not always
-------
V -304
Table V-22
Rural Water Supply Pressure
Nation
Pressure (Pounds
Number of
Percent of
per Square Inch)
Supplies
Supplies
1-20 psi
1,402,000
7.5
21 - 40 psi
8,301,000
44.5
41 - 60 psi
5,956,000
31.9
61 - 100 psi
2,353,000
13.7
Greater than 100 psi
455,000
2.4
*Total
18,668,000
100.0
~Pressure measurements apply to 18.7 million house-
hold supplies. Pressure could not be measured for 2.7
million households, and another 490,000 supplies were
not pressurized.
Each figure in the table is rounded independently;
numbers and percentages may not equal rounded
totals.
-------
V - 305
"about right" or that noted changes in the water pressure) attributed the condition
to a basic inadequacy of the system's physical facilities. System breakdowns
outside the dwelling unit reportedly caused the pressure condition at 12.3 percent
of the affected households (812,000), and another 9.7 percent of households
(640,000) associated the pressure condition with problems inside the house. Of the
remaining households, 6.6 percent (436,000) attributed the condition to deliberate
or planned activities or to mismanagement, and 2.9 percent (191,000) related the
condition to seasonal factors. Other reasons were suggested at 5.8 percent
(383,000), and 11.9 percent (785,000) could not venture a reason for the pressure
condition.
Almost one-fourth of the 6.6 million households where pressure was high,
low, or usually fluctuating (23.2 percent, or 1.6 million) reported that they noticed
seasonal variations in the water pressure. Seasonal variation most often consisted
of pressure being too low or fluctuating too much in the summer (reported in 1.3
million households). In winter, by contrast, when demand for water is generally
lowest, 1.3 million, or 80.9 percent, reported that the pressure was about right.
Although 30.7 percent of the 21.5 million rural households with pressurized
water supplies (6.6 million) reported that the pressure was too high or too low, or
that it usually fluctuated, only 7.1 percent (1.5 million) said that they disliked the
pressure or that they disliked it very much. For 25.3 percent of households (5.4
million), the pressure was considered all right or was not thought about very much.
On the other hand, 67.4 percent of the households (14.5 million) reported that they
liked it or that they liked it very much.
As indicated by several factors, water supplies for rural households
appeared generally to be accessible. About 52 percent of the supplies for which
the necessary distance measurement was obtained (wells, springs, surface 'vater
supplies, and cisterns) withdrew water from a point that was within ten meters (33
feet) of the household structure. Water pressure, as determined by a pressure
-------
V - 306
gauge and summarized by a median of 39 psi, was also generally sufficient. On a
less positive note, however, 1.4 million households, or 7.5 percent of the households
for which measurements were recorded, had supplies with a pressure of less than
twenty psi. This corresponded almost exactly to the proportion of all rural
households that perceived the usual water supply pressure to be too low, which was
7.3 percent. An additional 8.3 percent of the rural households with pressurized
supplies reported either that the pressure was too high or that it fluctuated.
Although there were other reasons, pressure difficulties were most often attributed
to some unspecified inadequacy of the water supply's physical facilities. Also,
since water demands are typically greater in the summer, a larger proportion of
households reported seasonal pressure variations during that period in comparison
to the fall, winter, or spring. Finally, while 7.1 percent of rural households with
pressurized supplies expressed a moderate or strong dislike for the water pressure,
67.4 percent liked it.
Regional variation in accessibility
There was a great deal of variation from region to region in the distance
between the dwelling unit and the point where the water was withdrawn from the
source (see Table V-23). For the Northeast and North Central, distances were
quite similar. In the Northeast, 57.7 percent of supplies had the point of
withdrawal on the household premises or within ten meters of the structure, and
this was true for 61 percent of supplies in the North Central. As measured by this
indicator, accessibility was lower in the South and West. In the South, only 43.9
percent of supplies had the point of withdrawal on the household premises or within
ten meters of the structure; in the West, the percentage was 31.0 percent. The
point of withdrawal was eleven to 50 meters away from the structure for 26.0
percent of supplies in the Northeast, 27.3 percent in the North Central, 37.9
percent in the South, and 38.4 percent in the West.
-------
V -307
Table V-23
Regional Variation in the Distance Between Rural Dwelling Unit
and Point of Withdrawal from Source
Percent of Supplies
Distance
Northeast
North
Centred
South
West
On premises
4.4
9.2
4.6
2.9
1-10 meters
53.3
51.8
39.3
28.1
11-50 meters
26.0
27.3
37.9
38.4
51 - 100 meters
5.5
4.0
7.7
16.5
101 - 1,000 meters
7.1
2.6
8.2
10.4
More than 1,000 meters
0.5
0.2
0.7
2.2
Unspecified
3.2
4.9
1.7
1.6
Total Percent
100.0
100.0
100.1
100.1
~Total Supplies 2,034,000 3,825,000 3,851,000 872,000
~Table excludes community water supplies, hauled supplies, and
purchased bottled supplies.
-------
V - 308
Altogether, the point of withdrawal was within 100 meters of the structure
for 89.2 percent of supplies in the Northeast, 92.3 percent in the North Central,
89.5 percent in the South, and 85.9 percent in the West. The maximum distance
was 3,116 meters (1.9 miles) in the Northeast; 1,840 meters (1.1 miles) in the North
Central; 9,930 meters (6.2 miles) in the South; and 8,050 meters (five miles) in the
West.
Relatively few households in each region reported that they disliked the
water supply particularly for its inconvenience. The proportions ranged from 1
percent in the West to 2.7 percent in the South.
Water supply pressure, as measured by a pressure gauge, varied consider-
ably from region to region (see Table V-24). Pressure between one and twenty psi
was reported least often in the Northeast (4.3 percent of supplies) and most often
in the North Central (10.2 percent of supplies). The bulk of household supplies in
all regions had pressure readings from 21 through 60 psi. Pressure between 61 and
100 psi was found most often in the West (24.6 percent of supplies) and least often
in the North Central (6.5 percent).
In all regions, the great bulk of supplies were reported to have satisfactory
water pressure, according to household representatives. The proportions of
households where the pressure was too high were about equal, ranging from 0.4
percent in the Northeast to 1.6 percent in the South. Pressure was reported to be
too low at 8.4 percent of households in the Northeast, 6.6 percent in the North
Central, 8.2 percent in the South, and 4.6 percent in the West. Fluctuating
pressures were reported at 7.5 percent of households in the North Central and
South, 9.6 percent in the Northeast, and 12.3 percent in the West. Changes from
the usual condition of the supply with regard to pressure were reported at 30.9
percent of households in the West, compared to 23 percent in the Northeast, 20.8
percent in the North Central, and 21.3 percent in the South.
-------
V -309
Table V-24
Regional Variation in Measured Pressure
of Rural Water Supplies
Peroent of Supplies
Pressure
Northeast
North
Central
South
West
1-20 psi
4.3
10.2
7.2
6.5
21-40 psi
4 8.8
55.0
38.3
25.2
41 - 60 psi
33.0
28.0
38.0
42.8
61 - 100 psi
11.8
6.5
14.5
24.6
Greater than 100 psi
2.2
0.1
2.0
0.8
Total Percent
100.1
99.8
100.0
99.9
Total Supplies
3,335,000 5,
275,000 7,
876,000
2,707,000
~Table includes only supplies for which pressure measurements
were obtained.
-------
V - 310
Another indication that changes often occurred in pressure was that the
pressure characterized as usual was reportedly present in the water supply most or
all of the time at only 42.8 percent of households in the Northeast, 38.2 percent in
the South, 29.4 percent in the North Central, and 28.5 percent in the West.
The proportions of households assigning various reasons to the water supply
pressure also varied from region to region (see Table V-25). Of all households
queried (those where the pressure was not always about right, or that reported
changes in the pressure), inadequacy of the system was blamed much more often
than anything else. About two-thirds of the households in the West and about half
of the households in the Northeast, North Central, and South gave this as the
reason for the condition. The condition was attributed to a breakdown outside the
house at 20.1 percent of households in the North Central, but at only 6.5 percent in
the West. Problems inside the house caused the condition at 18.4 percent of
households in the Northeast, but at only 3.4 percent of households in the West.
Seasonal variation in water supply pressure occurred most often in the
West, where it was reported at 14.8 percent of the households that had pressurized
water supplies. By comparison, seasonal variation was reported at 7.3 percent of
such households in the South, 6.4 percent in the North Central, and 4 percent in the
Northeast. Seasonal variation also was most dramatic in the West, where over 90
percent of responding households reported that the pressure was about right in fall,
winter, and spring, but only 9.1 percent reported it was about right in summer. In
the Northeast, the pressure was about right at 69.5 percent of responding
households in the fall, 79.7 percent in the winter, and 66.7 percent in the spring,
but at only 12 percent in the summer. In the North Central, it was about right at
81 percent of responding households in the fall, 88.2 percent in the winter, and 79.9
percent in the spring, but at only 18.6 percent in the summer. The South showed
the least dramatic pattern of seasonal differences: about 64.4 percent of
responding households reported the pressure was about right in the fall, compared
-------
V -311
Table V-25
Regional Variation in Reported Reasons for Perceived Pressure
Conditions in Rural Water Supplies
Percent of Households
North
Reason
Northeast
Central
South
West
Inadequacy of the
system's physical
facilities
49.3
52.0
45.1
66.1
A breakdown in the
physical facilities
outside the house
11.1
20.1
10.1
6.5
A problem within
the house
18.4
8.5
8.9
3.4
Deliberate or planned
activities
2.0
6.2
5.4
0.6
Mismanagement of
the system
0.0
1.3
3.7
2.9
Seasonal factors
1.7
2.0
5.0
0.0
Other miscellaneous
5.8
2.5
8.0
5.9
Don't know
11.7
7.3
13.9
14.7
Total Percent
100.0
99.9
100.1
100.1
~Total Households 1,198,000 1,719,000 2,697,000 990,000
~Table includes only households that reported high, low, or fluctuat-
ing supply pressure, or that reported pressure changes.
-------
V - 312
to 69.1 percent in the winter, 68.5 percent in the spring, and 26.5 percent in the
summer.
In the summer, pressure was too low at 35.3 percent of the households
reporting seasonal variation in the Northeast, compared to 33.8 percent in the
North Central, 26.4 percent in the South, and 37.9 percent in the West. It
fluctuated at 48.9 percent of the responding households in the Northeast, 47.6
percent in the North Central, 40.7 percent in the South, and 51.3 percent in the
West.
The agreeability of the water supply was about equal for all regions. In the
West, 6.3 percent of all rural households reported they disliked the pressure, or
disliked it very much; the highest proportion of households responding this way was
in the South, where 7.8 percent of all rural households disliked the pressure.
SMSA/nonSMSA variation in accessibility
Distances between the dwelling unit and the point where water was
withdrawn from the source tended to be greater for nonSMSA households than for
households located within SMSAs. Among nonSMSA households, 49.2 percent of
supplies had the point of withdrawal either on the premises or within ten meters of
the dwelling unit, compared to 58.4 percent of supplies among SMSA households.
The point of withdrawal was eleven to 50 meters away for 33.1 percent of
nonSMSA household supplies, compared to 28.4 percent of SMSA household supplies.
Distances of 51 to 100 meters occurred in about equal proportions: 7.1 percent for
nonSMSA supplies and 5.7 percent for SMSA supplies. For 7.1 percent of nonSMSA
supplies, the point of withdrawal was 101 to 1,000 meters away, compared to 3.0
percent of SMSA supplies. Less than 1 percent of both SMSA and nonSMSA supplies
had the point of withdrawal more than 1,000 meters away from the dwelling unit.
-------
V - 313
No substantive difference was found between SMSA and nonSMSA house-
holds in the proportions that disliked the water supply particularly for its
inconvenience.
Water pressure tended to be lower among nonSMSA households. Specific-
ally, 9.0 percent of nonSMSA households had a water pressure of from one to
twenty psi, compared to 4.4 percent of SMSA households. Also, a pressure of 21 to
40 psi was measured at 44.2 percent of nonSMSA households, compared to 40.3
percent of SMSA households. A pressure of 41 to 60 psi was recorded at 38.7
percent of SMSA households and 33.2 percent of nonSMSA households. Fifteen
percent of SMSA supplies and 12.4 percent of nonSMSA supplies had a pressure of
61 to 100 psi. Only a few supplies—1.5 percent of SMSA supplies and 1.2 percent
of nonSMSA supplies—had pressure readings above 100 psi.
Seasonal variation in water pressure was reported at 7.6 percent of SMSA
households and 7.3 percent of nonSMSA households. Likewise, differences of
roughly three percentage points or less were seen in the proportions of SMSA and
nonSMSA households reporting various conditions of water supply pressure and in
the proportions reporting that changes occurred in those conditions. However,
among those households where pressure was not always about right or that reported
changes in the usual condition, there was considerable variation in the reasons
associated with the condition. Overwhelmingly, inadequacy of the system was
blamed. (This was the reason given at 53.9 percent of SMSA households and 49.1
percent of nonSMSA households.) A problem inside the house was the cause at 11.3
percent of SMSA households and 8.8 percent of nonSMSA households. Breakdowns
outside the house were blamed at 9.4 percent of SMSA households and 13.9 percent
of nonSMSA households. Deliberate actions were thought to be the cause at 5.9
percent of nonSMSA households, but at only 1.3 percent of SMSA households.
Several other reasons were given in about equal proportions among SMSA and
nonSMSA households.
-------
V - 314
As for seasonal variation in water supply pressure, a higher proportion of
nonSMSA households reported the pressure was about right in each season. The
household water supply pressure in general was liked about equally at SMSA and
nonSMSA households; only 6.9 percent of nonSMSA households and 7.6 percent of
SMSA households reported either a moderate or strong dislike for the water supply
pressure.
Slze-of-place variation in accessibility
A comparison of households located in large rural communities, small rural
communities, and other rural areas showed a great deal of variation in accessibility
as measured by the distance between the household and the point where the water
supply was withdrawn from its source. Specifically, 61.6 percent of household
supplies in large rural communities had the point of withdrawal either on the
premises or within ten meters of the dwelling unit, compared to 77 percent of
supplies in small rural communities and only 50.8 percent in other rural areas. The
point of withdrawal was eleven to 50 meters away for 24.6 percent of households in
large rural communities, 14.1 percent in small communities, and 32.4 percent in
other rural areas.
In other rural areas, the point of withdrawal was 51 to 100 meters away
from the dwelling unit for 6.9 percent of households, and from 101 to 1,000 meters
away for another 6.2 percent. Distances of more than 50 meters were reported for
only 5.5 percent of supplies in large communities and 3.3 percent in small
communities. Likewise, the only water supplies that withdrew water from a source
over 1,000 meters away from the household were located in other rural areas (0.6
percent of the supplies in other rural areas). In small rural communities, the
maximum distance reported was 75 meters (246 feet); in large rural communities,
it was 200 meters (657 feet); and in other rural areas, it was 9,930 meters (6.2
miles).
-------
V - 315
Consistent with these figures was the finding that other rural areas had the
greatest proportion of households reporting that the water supply was disliked
particularly for its inconvenience (2.3 percent, compared to 0.5 percent of
households in both large and small rural communities).
Higher pressure readings were more common in large rural communities
than in small rural communities or other rural areas. In large communities, only
3.1 percent of supplies had pressure readings between one and twenty psi,
compared to 5.7 percent of supplies in small rural communities and 8.2 percent of
supplies in other rural areas. Similarly, 26.3 percent of supplies in large rural
communities had pressure readings between 21 and 40 psi, compared to 37.7
percent in small rural communities and 45.4 percent in other rural areas. Readings
between 41 and 60 psi were found in 44.4 percent of supplies in large rural
communities and 41.8 percent of supplies in small rural communities, but in only
33.3 percent of supplies in other rural areas. Higher pressures were more common
in large communities, where 22.4 percent of supplies had a pressure of 61 to 100
psi, and 3.7 percent had a pressure of more than 100 psi. Readings between 61 and
100 psi were found in 14.9 percent of supplies in small rural communities and 12.0
percent of supplies in other rural areas. Pressure readings higher than 100 psi were
recorded for 1.1 percent of supplies in other rural areas; no supplies in small
communities had pressure readings over 100 psi.
Water pressure was reported to be "about right" less often among house-
holds in large rural communities. There, 79.3 percent of households reported that
the pressure was about right, compared to 82.7 percent in small rural communities
and 83.8 percent in other rural areas. Otherwise, there was no substantive
deviation from national estimates regarding pressure conditions, except that
changes in the usual pressure were noticed more frequently at households in large
communities (27.8 percent) compared to households in small rural communities
(21.9 percent) and other rural areas (22.1 percent).
-------
V - 316
Reasons given for the usual condition of the water supply with respect to
pressure varied considerably, however, depending on the size of place where the
household was located (see Table V-26). Problems with pressure were most
frequently attributed to Inadequacy of the system regardless of where the
household was located, but households in small rural communities in particular
attributed problems to this cause (64.1 percent, compared to roughly 50 percent
elsewhere). Breakdowns outside the house were seen as the cause of pressure
problems most frequently in other rural areas. Deliberate actions were mentioned
at 7.5 percent of households in large rural communities, compared to 1.8 percent of
households in small communities and 4 percent of households in other rural areas.
Seasonal variation in water pressure was reported at 10.9 percent of
households in small rural communities, 10.8 percent of households in large rural
communities, and 6.6 percent of households in other rural areas. For the most
part, the variation reflected difficulties during the summertime. Large rural
communities showed the highest proportion of households with unsatisfactory
pressure in the summer; only 15 percent of households where seasonal variation
occurred said the pressure was about right in the summer, compared to 24.8
percent of households that reported seasonal variation in small communities and
18.7 percent of households that reported seasonal variation in other rural areas.
Of the households reporting that seasonal variation occurred, small rural
communities had the highest proportion in every season that said the pressure was
about right. In spring and fall, the lowest proportion reporting that the pressure
was about right was found in other rural areas (74.7 percent in the fall and 73.5
percent in the spring).
Summertime difficulties were more often related to fluctuating water
pressure than to pressure that was too low. In large and small communities both,
slightly more than 50 percent reported that pressure fluctuated. This was almost
twice the proportion that reported low pressure. In other rural areas, 43.4 percent
-------
V -317
Table V-26
Size-of-Place Variation in Reported Reasons for Perceived Pressure
Conditions in Rural Water Supplies
Peroent of Households
Reason
Large Rural
Communities
Small Rural
Communities
Other Rural
Areas
Inadequacy of the
system's physical
facilities
50.6
64.1
49.7
A breakdown in the
physical facilities
outside the house
6.4
8.5
13.6
A problem within
the house
10.3
5.9
9.9
Deliberate or planned
activities
7.5
1.8
4.0
Mismanagement of
the system
2 .4
5.0
2.0
Seasonal factors
3.9
0.0
3.0
Other miscellaneous
2.0
5.3
6.5
Don't know
17.0
9.3
11.3
Total Percent 100.1 99.9 100.0
~Total Households 8,660,000 448,000 5,296,000
~Table includes only households that reported high, low, or fluctuat-
ing supply pressure, or that reported pressure changes.
-------
V -318
of households where seasonal variation occurred reported fluctuating pressure and
34.2 percent said the pressure was too low. The decreased pressure of summertime
may have represented an improvement to some households: in large communities,
3.8 percent of households where seasonal variation occurred reported the pressure
was too high for three seasons out of the year, but no households reported that the
pressure was too high during the summer.
Households in large rural communities were in general less satisfied with
the water pressure than households in small communities or other rural areas. Only
59 percent of households in large rural communities reported that they liked the
pressure or that they liked it a great deal, compared to 69.1 percent of households
in small communities and 68.5 percent of households in other rural areas. On the
other hand, 10.9 percent of households in large communities expressed a moderate
or strong dislike for the water supply pressure, compared to 5.6 percent of
households in small communities and 6.7 percent of households in other rural areas.
The other households reported that the pressure was all right, or that they didn't
give it much thought.
Size-of-system variation in accessibility
Distances from the dwelling unit to the point where the water was
withdrawn from its source were measured for supplies of households on individual
and intermediate systems only. (By definition, the supplies of households served by
community systems had the point of withdrawal on the premises.) As would be
expected, supplies for households on intermediate systems showed greater dis-
tances between the point of withdrawal from the source and the dwelling unit.
Among households served by individual systems, 57.8 percent of supplies had the
point of withdrawal either on the premises or within ten meters of the structure,
while this was true for only 27.0 percent of supplies at intermediate-system
households. Distances between eleven and 50 meters were seen about equally
-------
V - 319
among households on individual and intermediate systems: in 31.3 percent of
individual-system households and 34 percent of intermediate-system households.
Greater distances were more common for supplies of households on intermediate
systems, as would be expected. Distances between 51 and 100 meters were found
in 5.4 percent of households on individual systems but in 11.9 percent of households
on intermediate systems. Distances between 101 and 1,000 meters were found in
3.0 percent of individual-system households but in 18.1 percent of intermediate-
system households. Relatively few households had supplies which extracted water
from sources more than 1,000 meters distant: 0.4 percent of households on
individual systems and 1.4 percent of those on intermediate systems.
Households served by individual systems more frequently reported that the
water supply was disliked particularly for its inconvenience. This situation was
found at 4.3 percent of households served by individual systems (381,000 house-
holds), but at only 1.2 percent of households served by intermediate systems
(27,000) and 0.3 percent of households served by community systems (34,000).
As expected, pressure readings of supplies varied quite a bit depending on
the size of the water system (see Table V-27). Among supplies for households on
individual and intermediate systems, pressure was virtually never over 60 psi, and
for sizable proportions of supplies, the water pressure was between one and twenty
psi. Only 2.2 percent of supplies among households served by community systems
showed pressure readings between one and twenty psi, and almost one-quarter of
these supplies had pressure readings between 61 and 100 psi. The maximum
pressure was 76 psi among supplies of households on individual systems, 81 psi
among those on intermediate systems, and 190 psi among those on community
systems.
Surprisingly, although the pressure readings taken at rural households
indicated that community systems supplied water at higher pressures than individ-
ual or intermediate systems, more households served by community systems
-------
V -320
Table V-27
Size-of-System Variation in Measured Pressure
of Rural Water Supplies
Percent of Supplies
Individual Intermediate Community
Pressure System System System
1-20 psi
11.9
17.1
2.2
21 - 40 psi
63.8
62.4
23.3
41 - 60 psi
23.6
16.9
47.1
61 - 100 psi
0.7
3.5
24.7
Greater than 100 psi
0.0
0.0
2.6
Total Percent 100.0 99.9 99.9
~Total Supplies 7,593,000 1,825,000 9,775,000
~Table includes only those supplies for which pressure measure-
ments were obtained.
-------
V - 321
expressed dissatisfaction with the supply's water pressure. Only 80.8 percent of
these households reported that the usual pressure was about right, compared to
85.2 percent of households served by individual systems and 87.9 percent served by
intermediate systems. One would have expected that the lower pressures of
individual and intermediate systems would prompt complaints more often than the
typically higher pressures supplied by community systems.
Complaints about pressure being too high occurred infrequently, but were
not restricted to households served by community systems: pressure was considered
too high at 0.2 percent of households served by individual systems, 0.3 percent of
households served by intermediate systems, and 1.8 percent of households served by
community systems. Pressure was reportedly too low at 7.6 percent of households
served by individual systems, UA percent of households served by intermediate
systems, and 7.7 percent of households served by community systems. Fluctuating
pressure was reported at 7 percent of households served by individual systems, 7 A
percent served by intermediate systems, and 9.7 percent served by community
systems.
In addition to the fact that more households served by community systems
reported fluctuating water pressure, more of these households reported that
changes occurred in the usual pressure—28.2 percent, compared to 15.9 percent of
households served by individual systems and 20.6 percent served by intermediate
systems.
Inadequacy of the system was most often seen as the cause of the various
pressure conditions that were reported, regardless of the size of the system serving
the household. However, the prominence of this response was most marked among
households served by intermediate systems: 67A percent of the households where
pressure was not always "about right" and which were served by intermediate
systems attributed the condition to system inadequacy, compared to 52.9 percent
-------
V - 322
of the households served by community systems and 42.3 percent of the households
served by individual systems.
Among households served by individual systems, only two reasons besides
inadequacy of the system were given by substantial proportions of responding
households. These were problems inside the house (19.4 percent) and external
breakdowns (18.6 percent).
Among households served by intermediate and community systems, no
other reason besides inadequacy of the system was mentioned by more than 10.0
percent of responding households. External breakdowns were the next most
frequently mentioned reason for the pressure condition (9.6 percent of responding
households served by community systems, 8.7 percent of responding households
served by intermediate systems). Problems inside the house were mentioned by 7.5
percent of households served by intermediate systems and by 5 percent of
responding households served by community systems. The conditions were attri-
buted to deliberate actions at 6.8 percent of the households served by community
systems, but at none of those served by intermediate systems.
Although the actual number of households involved was small (65,000), the
2.9 percent of households which were served by intermediate systems and which
reported seasonal variation in water pressure showed the most dramatic pattern.
While these households had the highest proportion reporting that pressure was
about right during fall, winter, and spring, they had the lowest proportion—in fact,
no households at all—reporting that pressure was about right in summer. For
almost two-thirds of the 65,000 households, pressure fluctuated during summer; for
one-third, pressure was too low. For the other three seasons, data for these
households were difficult to interpret. During the spring, 86.7 percent of the
affected households said the pressure was about right and 13.3 percent said it
fluctuated. During the fall, 88.8 percent said it was about right and 11.2 percent
-------
V - 323
said it was too low. During the winter, 94.4 percent said the pressure was about
right and 5.6 percent said it was too high.
Seasonal variation also was reported infrequently among households served
by individual systems (227,000, or 2.7 percent of such households). Most seasonal
pressure problems arose during summer, when only 23.1 percent of the households
noticing seasonal variation said that the pressure was about right, compared to 67.5
percent in the fall, 71.7 percent in the spring, and 72 percent in the winter. During
the summer, fluctuating pressure and low pressure were reported about equally
(39.3 percent and 35.1 percent of the 227,000 affected households, respectively).
Surprisingly, seasonal variation was most common among households served
by community systems, being reported at 1.3 million of these households, or 11.8
percent. Summer was again the season when most problems occurred, with 46.1
percent of the affected households reporting fluctuating pressure, 31.2 percent
reporting low pressure, and 18.9 percent reporting pressure that was about right.
In the other three seasons, roughly equal proportions reported pressure that was
about right (77.5 percent in both spring and fall, and 81.8 percent in winter).
Overall, the water pressure supplied by community systems was liked least
often. Of all rural households served by community systems, 63.5 percent reported
that they liked the pressure conditions or liked them a great deal, compared to 70.9
percent of households served by individual systems and 74.2 percent served by
intermediate systems. The supply was all right, or not thought about very much
with regard to pressure, by 27.5 percent of households served by community
systems, 23.3 percent served by individual systems, and 22 percent served by
intermediate systems. The pressure was disliked, or disliked very much, at 8.8
percent of households served by community systems, 5.8 percent served by
individual systems, and 3.9 percent served by intermediate systems.
-------
V - 324
EFFECTS OF QUALITY, QUANTITY, AND AVAILABILITY CONDITIONS
Inadequacies of water supplies in terms of quality, quantity, and availabil-
ity can have a variety of consequences for households. Some of these effects are
naturally more serious than others, although those that impinge upon the health and
physical well-being of household residents are probably the most critical. Incon-
veniences associated with particular supplies, while not directly related to health,
can also disrupt a household's pattern of living, sometimes seriously. The least
severe water supply problems—those that interfere only minimally with the routine
of the household and those that are transitory—nonetheless can become a source of
irritation, frustration, or discontent if they cannot be rectified.
In the NSA interviews, residents reported a variety of problems which they
associated with inadequacies of various kinds of their water supplies. In addition to
being asked about general problems or inconveniences they may have experienced,
residents were asked specifically whether anyone living in the household or any
visitors to the household had become ill from drinking the household water.
Although there are obvious problems of attribution regarding illnesses, it was
important to find out the extent to which rural residents believed that their water
supplies caused health problems. Also, this information was a valuable supplement
to NSA laboratory data on household water quality.
Here, the issue of health effects is addressed with reference to reported
illnesses. This orientation is substantially different from that in the portion of this
chapter devoted to laboratory-measured water quality, where the concentrations of
health-related constituents were examined specifically in regard to their potential
health threat. Following the section on reported illnesses, there is a discussion of
other specific problems that residents reported occurring as a consequence of
perceived water supply inadequacies.
-------
V - 325
REPORTED ILLNESSES
On the basis of NSA results, illnesses among rural residents during the
preceding year were thought to be associated with the water supply in 1.7 percent
of all rural households (374,000 households). The possibility that household visitors
had experienced water-related illnesses during the preceding year was reported in
fewer households (0.9 percent or 198,000). In total, 2.3 percent of all rural
households reported occasions when either residents or visitors became ill; in 0.3
percent of these households, both resident and visitor illnesses were reported.
Among those households where illnesses were attributed to the water
supply, diarrhea was the most common malady, mentioned roughly one-third of the
time. Abdominal pains were experienced about half as frequently as diarrhea. In
the remaining households, a variety of other illnesses were reported.
Subnational variation in reported illnesses
The proportion of rural households which associated illnesses with the
water supply was so small that, for the most part, it was impossible to examine the
data for variations by region, SMSA/nonSMSA, size of place, or size of system. In
general, however, the distribution of households reporting illnesses was relatively
uniform across these subnational categories. The only significant deviation from
the pattern was observed among households in large rural communities. Illnesses
were reported at approximately 5.4 percent of these households, which was more
than double the national rate. This finding is inconsistent with water quality
measurements reported earlier in this chapter which indicated that, overall, water
supplies of households in large rural communities had substantially lower concen-
trations of certain contaminants, including microorganisms, than household supplies
in the other two size-of-place categories. One possible interpretation of this
finding is that household residents in large rural communities have a greater
tendency than other rural residents to attribute illnesses to the water supply.
-------
V - 326
Unfortunately, verifying this interpretation is impossible, since the NSA did not
acquire data on the actual incidence of water-related diseases in rural households.
Further, the total number of households involved (75,000) was not large enough to
analyze in detail.
SPECIFIC SUPPLY PROBLEMS AND INCONVENIENCES
The condition of the water supply can have specific and troublesome
implications for household residents besides those related to health. In the NSA,
respondents were asked to identify household problems or inconveniences which
were directly attributable to the water supply. In about twenty million households
(91 percent), no specific problems or inconveniences were mentioned. Among the
approximately two million households which noted problems or inconveniences,
the water was reported to leave deposits in sinks, pipes, and on kitchenware in
659,000 households. Some households (318,000) reported that the water affected
their laundry, while 132,000 households indicated that it altered the flavor or
appearance of food and drink. An additional 625,000 rural households reported
various combinations of the preceding problems and inconveniences, the most
common of which involved the deposits and effects on laundry. Although some
other individual problems were noted besides these, they were not very frequent.
For example, about 79,000 households reported that the water was too costly to
treat.
Subnational variation in specific supply problems and inconveniences
Some regional variation could be detected in the proportions of households
reporting specific supply problems and inconveniences. Proportionally, more rural
households in the Northeast were affected than in any other region (12A percent);
problems mentioned most frequently there were that the water left deposits,
affected laundry, and affected the flavor or appearance of food and drink. These
-------
V - 327
effects were attributed to the high mineral content of water supplies in the
Northeast. The South, on the other hand, had the smallest proportion of rural
households reporting specific supply problems and inconveniences (7.3 percent).
No substantive variation could be seen in the SMSA/nonSMSA, size-of-
place, and size-of-system comparisons.
COST
RECORDED COST OF HOUSEHOLD WATER,
MODE OF PAYMENT, AND BILLING INTERVALS
The cost of household water is determined by a number of factors,
including capital expenditures for equipment as well as ongoing costs of treatment,
maintenance, and labor. Though the NSA focused on the household, there was no
reliable way to collect such cost data for individual supply systems (self-suppliers).
For instance, initial equipment investment—for a water pump and piping, for
example—generally had been financed by a previous owner. Furthermore, it was
not expected that there would be reliable records of current household costs for
operating and maintaining individual supplies.
In light of these considerations, general utility bills offered the only
reliable means to assess household water costs. This meant, however, that the NSA
cost analysis was restricted to households served by community systems.
For the nation as a whole, 9.2 million rural households (43.7 percent) were
billed regularly for their water supply. A similar number—9.1 million, or 43.2
percent—paid for water indirectly. Indirect payments were related to the
operation and maintenance of self-supply equipment, and included costs of elec-
tricity, pump and pipe replacement, and storage tank maintenance, for example. In
another 2.5 million households (12.0 percent), the cost of water was included in
-------
V - 328
rent payments. Among the remaining 1.1 percent of rural households (approxi-
mately 236,000 households), either the cost was included in the mortgage payment
or the method of payment could not be determined.
For those households which received regular water bills (9.2 million), 61.4
percent were billed monthly, 15.7 percent once every two months, 19.2 percent
once every three months, and only 2.5 percent once every six months. At some
households, bills were received once every four months, or even once a year (1.2
percent, combined).
Although many rural households regularly received water bills, the bills
were not always available at the time of the NSA survey. In other cases, the bill
was available, but it lacked major components—the amount charged or the gallons
consumed. Moreover, the amount paid for water was sometimes subsumed in a
total utility bill (one including sewer costs, for example). Data for households for
which water costs were subsumed in a total utility bill were analyzed separately to
prevent overestimation of water costs. Because of these adjustments, detailed
NSA cost analysis could be projected to only 4.6 million households, essentially
one-fifth of the nation's rural households.
To maintain comparability between the NSA estimates and other resources
on household water cost, two approaches were used in analyzing the cost data. The
first (the same approach used in the Temple, Barker, & Sloane study to be discussed
shortly) involved calculating a unit cost per thousand gallons. Unfortunately, this
estimate was not always reliable. This was because some bills lacked key
t
information needed to calculate consumption. The attempt was made to obtain the
most reasonable estimate, but if the daily per capita estimate seemed too high or
too low (taking into account both inside and outside uses of water), the household
was excluded from the cost analysis.
On the basis of the per-thousand-gallon calculation, the median water cost
was $1.35 nationally. The range was from $.08 through $23.41 per thousand
-------
V - 329
gallons. Because of some extremely high values, the mean was $2.55 per thousand
gallons (see Table V-28).
The range for monthly household water costs was greater than that
computed for costs per thousand gallons. This was expected since the per-unit cost
did not take into account the monthly amount of water consumed, which varied by
as much as 30,000 gallons. Water costs for the nation ranged from $1.00 to $58.00
per month. Both the mode and median were $7.00; the average monthly cost per
household was $8.33.
A survey of financial characteristics of community water systems, under-
taken by the consulting firm of Temple, Barker, <5c Sloane (TBS),^®^ provides data
that can be compared with NSA cost results. For the nation, NSA data showed a
median cost of $1.35 per thousand gallons. In contrast, for multiple-connection
systems studied in the TBS survey, average rates ranged from $.32 to $.86 per
thousand gallons, depending upon the size of the population served.
The lack of correspondence between cost results for the Temple, Barker, &
Sloane and NSA studies may be a result of differences in sample designs. Whereas
the NSA collected cost data at the rural household, the TBS survey collected cost
information from systems, which they chose from a nationwide EPA Inventory of
Water Systems. Thus, NSA cost estimates were based on actual charges recorded
on water bills; in the TBS study, system representatives were asked how much a
typical residential customer would pay for 100,000 gallons of water, based on the
system's current rate structure. Left unclear was whether the TBS estimates
included costs of operation and maintenance, debt retirement, and connection
charges, or whether they simply represented a unit cost for the water itself. If the
cost was only for water and did not represent complete costs, the large discrepancy
between the two studies would be explained, since the NSA billing amount
reflected complete charges to the water user. (The only potential additional costs
would have been initial, one-time-only connection fees and separately itemized
-------
V -330
Table V-28
Regional Variation in Recorded Water Costs
per Thousand Gallons (1978)
Dollar Estimates
North
Statistic
Nation
Northeast
Central
South
West
Median
1.35
1.34
1.50
1.33
2.00
Mode
1.33
1.00
1.14
1.33
3.19
Mean
2.55
2.34
2.51
1.99
2.80
Minimum
0.08
0.46
0.33
0.26
0.08
Maximum
23.41
16.10
16.58
23.41
17.24
~Total Households 4,413,000 520,000 762,000 2,594,000 537,000
*Table includes those households where itemized utility bills were available.
-------
V - 331
service charges.) Also, part of the discrepancy can be attributed to inflation, since
the TBS data were collected in the spring of 1976 and the NSA data more than two
years later. Finally, probably as a result of disproportionate sampling of very large
water systems in the TBS survey, average system charges were very much lower
than NSA estimates.
Recorded costs at households receiving nonitemized utility bills
For the most part, charges for water were stated explicitly in water bills.
Some households, however, received utility bills that included charges for water
without separating them from charges for other utilities. These bills included
charges for natural gas, electricity, garbage disposal, and sewage disposal in
addition to water charges. Of those households receiving such nonitemized bills,
about 80 percent (237,000 households) were billed for one other utility service,
while roughly 20 percent (56,000 households) were billed for two other utility
services in addition to water service. Sewage and garbage disposal charges were by
far the most frequently included charges in these bills. Charges for electricity
were least frequently included.
As would be expected, total charges shown on these nonitemized utility
bills were higher than recorded costs for water at households billed specifically for
water. The average household cost recorded on nonitemized bills was $15.11,
compared to $8.33 per month for households being charged separately for water.
Regional variation in mode of payment, billing intervals, and recorded cost
Regional differences were observed in modes of payment for household
water (see Table V-29). In most rural households, payment was made either to a
designated supplier ("receive a regular billing for water"), or else to undesignated
parties (i.e., an electric company, a plumbing supply company, etc.) which provided
for the operation and maintenance of individual supplies. The latter situation was
-------
V -332
Table V-29
Regional Variation in Mode of Payment for Water
Percent of Households
Mode of Payment
Nation
Northeast
North
Central
South
West
Receive a regular
billing for water
43.7
33.7
34.4
51.5
52.8
Operation and
maintenance
(individual systems)
43.2
50.3
57.0
35.8
26.5
Included in rent
payments
12.0
14.8
7.6
11.6
19.3
Other
1.1
1.2
0.9
1.0
1.4
Total Percent
100.0
100.0
99.9
99.9
100.0
* Total Households
21,025,000
3,611,000
6,009,000
8,775,000
2,630,000
*Does not include households which associated no cost with the water supply.
-------
V - 333
most frequent in the Northeast and North Central (50.3 percent and 57.0 percent,
respectively). This was consistent with the finding that the predominant type of
major water supply in these regions was the individual well (see Chapter IV).
Conversely, approximately 52 percent of the households in the South and West
received a regular billing for water. Again, this was consistent with the finding
that community water supplies were the predominant type in the South and West.
Water was included in rent payments in 19.3 percent of households in the West,
compared to only 7.6 percent in the North Central.
As to billing intervals, the percentage of billed households which received
monthly statements varied widely from region to region. However, this difference
may or may not have reflected the true condition in light of the small number of
sampling points in the Northeast. Because of this uncertainty, regional variations
for the billing period cannot be assessed reliably.
Mean costs per thousand gallons ranged from $1.99 in the South to $2.80 in
the West (see Table V-28). Mean costs were comparable in the Northeast and
North Central—$2.34 and $2.51, respectively. Similarly, the median cost was
lowest in the South, $1.33 per thousand gallons, and highest in the West, with half
of the households charged at least $2.00 per thousand gallons. (Again, it should be
kept in mind that these cost estimates are based on information from those rural
households where itemized bills were available.)
The regional variations that were detected in monthly water costs require
additional discussion because factors that affect monthly water cost, such as the
number of people in the household, the number and types of water-using devices,
the amount of water used, and the billing month may also differ by region.
Consequently, an assessment of whether cost variations are more directly influ-
enced by geographical location or by these other variables can only be done by
applying the more sophisticated analytical procedures that will be used later in this
-------
V - 334
report. Specifically, for a more extensive treatment of monthly household water
cost and its determinants, the reader should consult Chapter XII.
SMSA/nonSMSA variation in mode of payment,
billing intervals, and recorded costs
Differences between SMSA and nonSMSA households also were observed in
modes of payment for household water, but they were smaller than regional
differences (see Table V-30). As shown in the table, regular billings for water
predominated in SMSA households (47.4 percent of these households), whereas
payment in the form of operation and maintenance of individual supplies was
prevalent in nonSMSA households (46.4 percent). Water costs were included in rent
payments in 15.2 percent of SMSA households and in 10.4 percent of nonSMSA
households.
The billing period varied considerably between SMSA and nonSMSA house-
holds (see Table V-31). Monthly billings were twice as common for nonSMSA
households (75.3 percent, compared to 36.3 percent of SMSA households). Con-
versely, two- and three-month billings were far more common among SMSA
households. Specifically, bills were received once every two months at 27.0
percent of SMSA households and at 9.5 percent of nonSMSA households. Likewise,
three-month billings appeared at 33.5 percent of households within SMSAs,
compared to 11.3 percent of those outside SMSAs.
In general, households within SMSAs had lower water costs per thousand
gallons. Mean rates were $1.98 for SMSA households and $2.34 for nonSMSA
households. Half of the households within SMSAs were charged $1.08 or less for
each thousand gallons, while half of nonSMSA households were charged a maximum
of $1.62. This result was consistent with Temple, Barker, 6c Sloane findings.*^ In
that study, residential water rates were sharply higher for multiple-connection
systems associated with sparsely populated regions than for systems serving large
-------
V -335
Table V-30
SMSA/NonSMSA Variation in Mode of Payment for Water
Percent of Households
Mode of Payment
Nation
SMSA
NonSMSA
Receive a regular billing
for water
43.7
47.4
41.9
Operation and maintenance
(individual systems)
43.2
36.7
46.4
Included in rent payments
12.0
15.2
10.4
Other
1.1
0.7
1.3
Total Percent 100.0 100.0 100.0
~Total Households 21,025,000 6,292,000 14,103,000
*Does not include households which associated no cost with the water
supply.
-------
V -336
Table V-31
SMSA/NonSMSA Variation in Billing Periods
Percent of Households
Billing Period
Nation
SMSA
NonSMSA
Monthly
61 A
36.3
75.3
Once every two months
15.7
27.0
9.5
Once every three months
19.2
33.5
11.3
Other
3.7
3.2
4.0
Total Percent
100.0
100.0
100.1
•¦Total Households
5,241,000
1,865,000
3,376,000
~Billing periods could be estimated for 5.2 million of the 9.6 million
households that received regular water bills.
-------
V - 337
metropolitan areas. This trend appeared to be related to economies of size
achieved in large community supply systems, a subject investigated in depth later
in this report.
Size-of-place variation in mode of payment,
billing intervals, and recorded costs
In addition to regional and SMSA/nonSMSA differences, substantial varia-
tion in the mode of payment for household water occurred among rural households
according to whether they were located in large rural communities, small rural
communities, or other rural areas (see Table V-32). For example, about four-fifths
of households located in rural communities received regular water billings, whereas
in other rural areas only about one-third did. The most common way of paying for
water in other rural areas was through the operation and maintenance of individual
water supplies (approximately 50 percent of all households). In large communities,
only 6.2 percent of households paid operation and maintenance expenses. In small
communities, nearly 16 percent of rural households paid for water in this fashion.
For those households that received a regular water bill, the billing periods
were about the same in large and small rural communities, but were quite different
in other rural areas (see Table V-33). In each category of size of place, monthly
billings were the most common, but they were reported more often in large and
small communities (more than 70 percent of water bills, compared to about 55
percent in other rural areas). Bimonthly billings were far more common in other
rural areas (21.5 percent of water bills) than in large or small communities (about 6
percent of water bills in each).
The average cost of water per thousand gallons (the mean) was similar for
households located in large and small communities ($2.52 and $2.40, respectively),
but lower in other rural areas ($2.10). Median water costs were lower than the
mean, and showed less variation. In large rural communities, the median cost per
-------
V -338
Table V-32
Size-of-Place Variation in Mode of Payment for Water
Percent of Households
Mode of Large Rural Small Rural Other Rural
Payment Communities Communities Areas
Receive a regular billing
for water
83.5
78.0
35.5
Operation and maintenance
(indivi&ial systems)
6.2
15.7
50.7
Included in rent payments
10.0
5.3
12.8
Other
0.3
1.0
0.9
Total Percent 100.0 100.0 99.9
~Total Households 2,369,000 1,509,000 18,095,000
~Does not include households which associated no cost with the water
supply.
-------
V -339
Table V-33
Size-of-Place Variation in Billing Periods
Percent of Households
Large rural Small rural Other rural
Billing period
communities
comm unties areas
Monthly
72.3
70.7
55.3
Once every two months
5 A
6.0
21.5
Once every three months
17.1
17.2
20.4
Other
5.1
6.1
2.7
Total Percent
99.9
100.0
99.9
~Total Households
1,251,000
607,000
3,383,000
~Billing periods could be estimated for 5.2 million of the 9.6 million
households that received regular water bills.
-------
V - 340
thousand gallons was $1.45; in small communities, it was $1.60; and in other rural
areas, it was $1.33.
Size-of-system variation in mode of payment,
billing intervals, and recorded costs
Almost all rural households served by individual systems paid for their
water by operating and maintaining the supply (about 95 percent). However, for
2.7 percent of households with individual systems, the cost was included in rent
payments (see Table V-34). A small number of households with individual systems
(1.6 percent) received a regular billing for water. (These were households that
purchased water on a regular basis; recall that the NSA defined purchased bottled
water as an individual supply even though the distributor may have bought the
water from a community system.)
For households served by intermediate systems as well, the most common
way of paying for water was in the expense of operating and maintaining the
system (65.0 percent of households). For 27.8 percent served by intermediate
systems, the cost was subsumed in rent payments, and 4.7 percent received regular
billings.
As would be expected, households served by community water systems
predominantly paid for their water by a regular billing (83.0 percent). The second
most common mode was to have the cost included in rent payments (16.4 percent).
Interestingly, about 22,000 rural residents—probably trailer park owners—paid for
household water by operating and maintaining systems of fifteen or more connec-
tions. (Another plausible explanation for this finding may have been cooperative
arrangements whereby users shared in the operation and maintenance of the
system.)
Although small proportions of households served by individual and inter-
mediate systems received regular water billings, they were unable to provide
-------
V -341
Table V-34
Size-of-System Variation in Mode of Payment for Water
Percent of Households
Mode of
Payment
Individual
System
Intermediate
System
Community
System
Receive a regular billing
for water
1.6*
4.7
83.0
Operation and maintenance
(individual systems)
94.7
65.0
0.2
Included in rent payments
2.7
27.8
16.4
Other
1.0
2.5
0.5
Total Percent 100.0 100.0 100.1
**Total Households 8,765,000 2,228,000 10,981,000
~These were households which purchased water on a regular basis and
received a bill for it.
**Does not include households which associated no cost with the water
supply.
-------
V - 342
written bills during the interview. Since billing periods and expenditures for water
could not be determined for these households, no data existed for a comparison of
recorded cost by size of system.
PERCEIVED COST
Almost all recorded NSA cost data pertained to households served by
community systems because the data were compiled only from water bills available
at the households. Perceived cost, on the other hand, was assessed for all rural
households.
On perceived cost, respondents were first asked whether they thought the
water supply was "inexpensive," "reasonable," "expensive," or "very expensive."
The only other possible responses were "dont know" or that the water had "no
cost." The cost of water was felt to be reasonable or inexpensive in the majority
of rural households, a total of 17.3 million (see Figure V-36). Moreover, in slightly
less than one million rural households, no cost was associated with the water
supply. At the other end of the scale, the water was considered expensive or very
expensive in 14.0 percent of all rural households, a total of about three million.
NSA interviewers also asked household representatives whether their water
costs had increased, decreased, or remained the same during the past year. Water
costs were judged to have remained the same in about six out of every ten rural
households, a total of 12.2 million. Costs reportedly increased in 7.9 million
households and decreased in only 248,000. For 631,000 households, it could not be
determined if costs had changed or, if they had, in what direction. (This was often
a result of respondents not living in households long enough to assess cost trends.)
An additional 0.9 million households associated no cost with the water supply.
-------
V - 343
Figure V-36
Perceived Cost of Rural Household Water Supplies
Number of Households (in millions)
22 4.4 6.6 8.8 II.0 13.2 15.4 17.6 19.8 22.0
no cost
inexpensive
reasonable
expensive
very expensive
10 20 30 40 50 60 .70 80 90 10'
Percent of Housi^oids'
' V'
'¦ i
:V,
-------
V - 344
Subnational variation in perceived cost
The cost of water was thought to be reasonable or inexpensive at the great
majority of rural households in all four regions—at more than 86 percent of
households in the Northeast and North Central, and at more than 75 percent of
households in the South and West (see Figure V-36a). In addition, 6.6 percent of
households in the West said there was no cost associated with the water supply,
compared to 3.4 percent in the Northeast, 3.8 percent in the South, and 1.5 percent
in the North Central. Water was reportedly expensive or very expensive for a
greater proportion of households in the South and West—for 17.1 percent in the
South and 18.9 percent in the West.
In a related finding, the cost of water was reported to have increased in a
larger proportion of households in the West (42.6 percent, compared to about 35
percent of households in the other three regions). For many households in each
region (ranging from a low of 48.2 percent in the West to a high of 63.3 percent in
the Northeast), water costs appeared to have remained the same over the past
year. As might be expected, cost increases were far more common than decreases.
Decreases were infrequent, at best: they were reported at 3.5 percent or less of
rural households in each region.
In general, perceived cost was higher in households located within SMSAs.
The most noticeable differences in the SMSA/nonSMSA comparison were that only
1.8 percent of SMSA households reported that there was no cost associated with
the water supply, compared to 5.7 percent of nonSMSA households, and that 19.1
percent of SMSA households reported that the water supply was expensive or very
expensive, compared to 12.3 percent of nonSMSA households.
The size-of-place comparison showed that, in general, perceived water
costs tended to rise with population size (see Figure V-36b). For example, in other
rural areas, 5.1 percent of households reported that the water supply had no cost,
compared to only 0.9 percent in large rural communities and 2.3 percent in small
-------
V - 31*5
Figure V-36a
Regional Variation in Perceived Cost
of Rural Household Water Supplies
0 10 20 30 40 50 60 70 80 90 100
frr-rr-i 1 1 1 1 1 1 1 1 ,
NORTH
CENTRAL
6,213,000
SOU-
9,2
rH
&!r00
0
1
Percent of Households
KEY:
no cost
inexpensive
reasonable
expensive
very expensive
-------
V - 346
Figure V-36b
Size-of-Place Variation in Perceived Cost
of Rural Household Water Supplies
0
r^
10
"n-
20
I
30 40
50
~~r~
60 70 80
90
—i—
100
» r ) i ¦ " i
LARGE RURAL £0MMUN!Tf£S
til
SMALL RURAL COMMUNITIES
j 1,509,00 0
J 1 I
OTHER RURAL AREAS
18,095,000
¦ i
0 10 20 30 40 50 60 70 80 90 100
Percent of Households
KEY:
no cost
inexpensive
reasonable
expensive
very expensive
-------
V -347
rural communities. At the other extreme, about 14 percent of households in small
rural communities and other rural areas reported that the water supply was
expensive or very expensive, compared to 21.6 percent of households in large rural
communities.
In addition, proportionately more households in large rural communities
reported cost increases during the preceding year. Specifically, 44.4 percent of
these households reported increases, compared to 38.7 percent in other rural areas
and 31.8 percent in small rural communities. Cost decreases were equally rare in
all three size-of-place categories, being reported in about 1 percent of households
in each category.
Just as perceived costs were highest at households in large rural commun-
ities, they were also highest at households served by community water systems
(see Figure V-36c). This is not surprising, since 93.5 percent of households in
large rural communities were served by community systems (see Chapter IV).
On the other hand, less than 2 percent of households served by community
systems reported that the water supply cost nothing, compared to 5.3 percent of
households served by individual systems. Surprisingly, 16.5 percent of households
served by intermediate systems associated no cost with the water supply. Water
supplies were felt to be expensive or very expensive at about 22 percent of
households served by community systems, compared to about 7 percent of
households served either by individual or intermediate systems.
Cost increases during the preceding year were reported most often by
households served by individual systems (43.7 percent). However, large propor-
tions of other households also reported cost increases—36.6 percent of households
served by community systems, and 29.6 percent of households served by inter-
mediate systems.
-------
V - 348
Figure V-36c
Size-of-System Variation in Perceived Cost
of Rural Household Water Supplies
10
II I I I—
20
30
"~r~
40
50 60
7 0 80 9 0 100
INDIVIDUAL
8,765,000
INTERMEDIATE
•2,2£8,&q0
COMMUNITY
10,981,000
L_J
'ill'
30 40 50 60 70
Percent of Households
100
KEY:
no cost
inexpensive
reasonable
expensive
very expensive
-------
V - 349
AFFORDABILITY
Obviously, the cost of water supplies is a factor in the status of rural water
conditions. Cost, the availability of domestic water, and the demand for water all
have widespread economic consequences. To the individual, however, the cost is a
more personal, immediate concern. The individual's financial well-being, his
feelings, and his motivation for acquiring water of good quality are all involved.
Economic demand curves can provide accurate descriptions of the general response
of users to cost and availability of water, but they do not necessarily describe that
response in relation to other considerations. The curves suggest that at a certain
cost level, people will tend to reduce the quantity of water they use rather than
pay more money. However, the curves do not evaluate that inflection point in
relation to other household expenditures. Nor do the curves tell us much about
individual variability in willingness to pay more for higher quality or for larger
quantities of water.
Traditionally, affordability has been defined as the ability to meet expense
without detriment to one's overall financial condition. In the NSA, the concept of
affordability was approached using both recorded and perceived indicators. The
recorded indicator of affordability was the percentage of income used for domestic
water. Perceived indicators were the reported reasonableness of the cost (see
Perceived Cost section, above) and the respondent's disposition to pay more, less,
or the same for a different water supply.
RECORDED INDICATORS OF AFFORDABILITY
The recorded measure of affordability which could be obtained from NSA
data was the ratio of billed cost to total household income. This ratio was then
multiplied by 100 to obtain a percentage of household income paid for water. Since
the billed or recorded cost, as opposed to perceived cost, was available only from
rural households which had sufficiently itemized water bills, the discussion of
-------
V - 350
recorded indicators of affordability was limited to about 4.6 million households
served by community water systems.
At these households, the proportion of household income paid for water
ranged from 0.04 percent through 15.60 percent (see Figure V-37). One-quarter of
the households paid 0.30 percent or less of household income for water. Half paid
0.60 percent or less. Three-quarters paid 0.99 percent or less, and one-quarter paid
1 percent or more.
The cost-to-income ratio represents one attempt to quantify affordability,
but it in no way takes into account other household expenses. In other words,
without additional information, there is no way to assess a particular household's
ability to pay for its water supply. Implied, however, is that as the ratio increases,
the probability of meeting the expense without detriment to the household's overall
financial condition decreases.
Subnational variation of recorded indicators of affordability
With respect to regional differences, households in the Northeast paid a
smaller proportion of income for water than those in the South, West, or North
Central (see Figure V-37a). Specifically, the median cost-to-income ratio in the
Northeast was 0.38, compared to about 0.60 in the South and North Central, and
0.73 in the West. Looking at the data differently, one-fourth of the evaluated
households in the South, West, and North Central paid more than 1 percent of their
income for water. This amounted to 622,000 households in the South, 237,000 in
the North Central, and 171,000 in the West. In contrast, only 13.0 percent of
households in the Northeast (73,000 households) paid that much (not reflected in
Figure V-37a).
The SMSA/nonSMSA comparison showed a more striking difference in the
cost of water in relation to household income, the ratio being much higher for
households located outside of SMSAs. In fact, the median ratio for nonSMSA
-------
V - 351
Figure V-37
Percentage of Total Household Income Paid for Water (1978)
0.30
0.60
0.99
15.60
Billed Charges as Percentage of Total Household Income
KEY:
1,140,000 households per quartile
first quartile
second quartiie
third quartile
fourth quartile
NOTE: Ratio could be computed only for households served by community systems.
-------
V - 352
Figure V-37a
Regional Variation in Percentage of Total Household Income
Paid for Water (1978)
40 3.10
NORTHEAST
(141,000 households per quartile)
0.320 60 1.10 5 45
NORTH
CENTRAL
(213,000 households per quartile)
0.29 0.61 1.01 6.73
SOUTH
(617,000 households per quartile)
0.40
WEST
KEY:
first quartile
second quartiie
third quartiie
fourth quartile
15.60
(168,000 households per quartile)
Billed Charges as Percentage of Total Household Income
NOTE: Ratio could be computed only for households served by community systems.
-------
V - 353
households was twice as high as it was for households located within SMSAs (0.72
versus 0.36). As shown in Figure V-37b, three-quarters of nonSMSA households paid
a maximum of 1.23 percent of their income for water, compared to three-quarters
of households within SMSAs paying a maximum of 0.84 percent. Finally, the
maximum cost-to-income ratio within SMSAs was 6.13, compared to 15.60 outside
of SMSAs.
Two factors caused the large difference in cost-to-income ratios of SMSA
households and nonSMSA households. First, the numerator of the ratio, the amount
paid for water, was considerably higher among households located outside of
SMSAs. Second, total household income (the denominator) was significantly lower
outside of SMSAs. In fact, there was a difference of $6,000 between SMSA and
nonSMSA median household income. With the two factors operating simultan-
eously, it is not surprising that the nonSMSA household ratios were so much higher
than those of SMSA households.
Cost-to-income ratios also varied considerably depending on whether
households were located in large rural communities, small rural communities, or
other rural areas (see Figure V-37c). For three-quarters of the households in other
rural areas, the highest proportion of household income spent for water was 0.86
percent, compared to 1.31 percent in small communities and 1.50 percent in large
communities. Median cost-to-income ratios were about equal for large and small
rural communities (0.82 and 0.76, respectively), but were much lower for other
rural areas (0.52). In summary, although the very highest cost-to-income ratio
occurred in other rural areas (where a single household in the NSA sample paid
15.60 percent of its income for water), costs generally were lower for households in
other rural areas.
Since the a/ailable billing information pertained exclusively to households
served by community water systems, affordability could not be compared by the
size of the supply system serving the household.
-------
V - 35 U
Figure V-37b
SMSA/NonSMSA Variation in Percentage of Total Household Income
Paid for Water (1978)
0.22 0.36 0.84 6.P3
SMSA
(407,000 households per quartile)
0.40
0.72
nonSMSA
1.23
15.60
(732,000 households per quartile)
Billed Charges as Percentage of Total Household Income
KEY:
1 \J
first quartile
second quartile
third quartile
fourth quartile
NOTE: Ratio could be computed only for households served by community systems.
-------
V - 355
Figure V-37c
Size-of-Place Variation in Percentage of Total Household Income
Paid for Water (1978)
0.82 3.64
0.40
LARGE RURAL
COMMUNITIES
(227,000 households per quartile)
n 7R
SMALL RURAL
COMMUNITIES
5.45
(127,000 households per quartile)
0.28 0.52
OTHER RURAL
AREAS
KEY:
first quartile
second quartile
third quartile
fourth quartile
0.86
15.60
(764,000 households per quartile)
Billed Charges as Percentage of Total Household Income
NOTE: Ratio could be computed only for households served by community systems.
-------
V - 356
PERCEIVED AFFORDABILITY
Perceptions of cost have been discussed elsewhere in this chapter (see
Figure V-36). Those same perceptions also shed some light on affordability. Recall
that water was perceived to be inexpensive or absolutely without cost at 30.2
percent of all rural households (6.6 million); it was seen as reasonable at 52.6
percent of rural households (11.6 million), and either expensive or very expensive at
another 14.0 percent (3.1 million households). Interestingly, a similar pattern of
responses occurred when household representatives were asked if they would be
willing to pay more or to pay the same amount for an "ideal" supply—one that was
good to drink and provided as much water as needed. A willingness to pay more for
the ideal supply was reported in 32.9 percent of all rural households (7.2 million),
while 62.9 percent of household representatives (13.8 million) said they would only
be willing to pay the same as they did for the current household supply. A
willingness to pay less was expressed at 3.6 percent of rural households (0.8
million). Given the similarity in response patterns, one might postulate that
household representatives reporting a water supply that was inexpensive or without
cost were the same ones who reported they would be willing to pay more for an
ideal water supply. A perception that the water was reasonably priced might be
assumed to indicate a willingness to pay the same amount for a different supply,
while a perception that the water supply was expensive might be equated with a
desire to pay less for water. When these hypotheses were tested, however, no
correlation was found between the perceived reasonableness of cost and willingness
to pay for a different supply.
In summary, household water costs were considered expensive in 14.0
percent of all rural households—fully three million households. However, expen-
sive water was not necessarily seen as unaffordable, but expensive in comparison to
other costs such as electricity or sewer charges. The disposition to pay less for an
ideal water supply was considered a better indicator of perceived affordability than
Reproduced from
best available copy.
-------
V - 357
the perceived expensiveness of the household supply, although it was a more
conservative measure—that is, it included fewer households. Thus, judging by this
indicator, there were less than one million rural households that had difficulty
bearing the current cost of water, though four times as many households reported
that the water supply was expensive or very expensive.
Subnational variation in perceived affordability
Since subnational variation in the reasonableness of perceived cost was
discussed earlier (see "Subnational variation in perceived cost," above), the
information will not be duplicated here with respect to the concept of afford-
ability. The only other measure of a household's ability to bear the expense of
water was the reported willingness to pay more or less than the current expendi-
ture for an ideal water supply.
Recall that for the nation as a whole, the majority of households (62.9
percent) were willing to pay the same for the ideal water supply as they paid for
their present one. As shown in Table V-35, this situation was slightly less common
in the Northeast (55.9 percent of households), while a willingness to pay more for
an ideal water supply was found in a higher percentage of households there (41.4
percent) than in the other three regions. In fact, a difference of eight percentage
points or more was detected when comparing the Northeast with the other regions
with respect to paying more for an ideal water supply. This disposition to pay more
may have been related to the substantially higher median household income
reported in the Northeast. Discussed in detail in Chapter III, median nonfarm
income in the Northeast was about $15,500, compared to median incomes of about
$12,000 in the North Central and South. The median in the West, $15,000, also was
slightly less than the median reported for the Northeast. In contrast, a desire to
pay less for water was expressed at 5.0 percent of households in the South and at
-------
V -358
Table V-35
Regional Variation in Willingness to Pay for an Ideal Water Supply
Percent of Households
Willingness
to Pay
Nation
Northeast
North
Central
South
West
Willing to pay the
same
62.9
55.9
64.1
64.8
63.3
Willing to pay
less
3.6
1.6
2.3
5.0
4.2
Willing to pay
more
32.9
41.4
33.1
29.7
31.4
Don't know
0.6
1.1
0.5
0.5
1.1
Total Percent
100.0
100.0
100.0
100.0
100.0
Total Households
21,971,000
3,693,000 6
,213,000
9,290,000
2,777,000
-------
V - 359
only 1.6 percent of households in the Northeast. Again, this may have been the
result of lower household income in the South.
Although less variation was observed between SMSA and nonSMSA house-
holds, proportionally more SMSA households than nonSMSA households were willing
to pay more for water (35.8 percent, compared to 31.4 percent). As suggested
earlier, this disposition to pay more may be a function of SMSA households having
larger disposable incomes (see Chapter III).
With respect to size-of-place and size-of-system comparisons, no variation
was found in willingness to pay for a different water supply.
-------
V - 360
REFERENCES
General references are:
Allen, Martin 3., and Geldreich, Edwin E. "Bacteriological Criteria for
Ground-Water Quality." Ground Water 13 (January-February 1975).
McCabe, Leland 3.; Symons, James M.; Lee, Roger D.; and Robeck, Gordon G.
"Survey of Community Water Supply Systems." Journal of the American
Water Works Association (November 1970): 670-687.
McCaull, Julian, and Crossland, Janice. Water Pollution. New York:
Harcourt Brace Jovanovich, 1974.
National Technical Advisory Committee to the Secretary of the Interior.
Water Quality Criteria. Washington, DC: US Government Printing Office,
1968.
Russell, Clifford S., editor. Safe Drinking Water: Current and Future
Problems, Proceedings of a National Conference in Washington, DC, 1978.
The Safe Drinking Water Act (Serial No. 95-10). Washington, DC: US
Government Printing Office, 1977.
US Environmental Protection Agency. Manual of Individual Water Supply
Systems. 2nd rev. ed. Raleigh, NC: US Environmental Protection Agency,
1974.
US Environmental Protection Agency. "National Secondary Drinking Water
Regulations," Federal Register 42: No. 62 (March 31, 1977): 17144-17146.
US Environmental Protection Agency. "National Interim Primary Drinking
Water Regulations." Federal Register 40: No. 248 (December 24, 1975):
59566-59574.
US Environmental Protection Agency. State of the Art of Small Water
Treatment Systems. Washington, DC: US Environmental Protection Agency,
1977.
2
Board of Directors, American Water Works Association. Quality Goals for
Potable Water. January 28, 1968.
3
US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations (EPA-570/0-76-003). Washington, DC: US Government Printing
Office.
-------
V - 361
References continued
4
US Environmental Protection Agency. "National Secondary Drinking Water
Regulations."
^National Academy of Sciences. Drinking Water and Health. Washington, DC:
National Academy of Sciences, 1977.
6Ibid.
Drexel University and National Science Foundation. Water Quality and Health
Significance of Bacterial Indicators of Pollution—Workshop Proceedings. Wesley
O. Pipes, ed. Philadelphia: Drexel University, 1978.
g
National Academy of Sciences, 1977, p. 72.
^Ibid., p. 84.
10Ibid, p. 83.
^Stevenson, Albert H. "Studies of Bathing Water Quality and Health." American
Journal of Public Health 43:529-538 (May 1953).
12
National Academy of Sciences, 1977, p. 83. First-phase assessment of Stevenson
study: Cabelli, Victor 3., and McCabe, Leland 3. "Recreational Water Quality
Criteria," News of Environmental Research in Cincinnati. Cincinnati: National
Environmental Research Center, November 11, 1974.
13
Whitsell, Wilbur 3.; Hutchinson, Gary D.; and Taylor, Donald H. "Problems of
Individual Supplies." The State of America's Drinking Water. Raleigh: North
Carolina State University, 1975.
14
McKee, 3ack Edward, and Wolf, Harold W., eds. Water Quality Criteria, 2nd ed.
Sacramento: State of California State Water Quality Control Board, 1963.
-------
V - 362
References continued
^Geldreich, Edwin E. "Applying Bacteriological Parameters to Recreational
Water Quality." Journal of the American Water Works Association (February
1970): 113-120.
^US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations, p. 30. #
^National Academy of Sciences, 1977, p. 75.
18
Geldreich, Edwin E. "Fecal Coliform and Fecal Streptococcus Density Relation-
ships in Waste Discharges and Receiving Waters." CRC Critical Reviews in
Environmental Control (October 1976):349-368.
19
Geldreich, Edwin E. "Applying Bacteriological Parameters to Recreational
Water Quality," p. 114.
20
National Academy of Sciences, 1977, pp. 86-87.
21
US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations, p. 38.
22
Glaze, William H.; Saleh, Farida Y.; and Kinstley, Warren. "Characterization of
Nonvolatile Halogenated Compounds Formed During Water Chlorination." In
Water Chlorination: Environmental Impact and Health Effects, Vol. 3, edited by
Robert L. Jolley, William A. Brungs, Robert B. Cummings, and Vivian A. Jacobs,
pp. 99-108. Ann Arbor: Ann Arbor Science Publishers, Inc., 1980.
23
Levin, Morris A. "Significance of Wastewater Disinfection to Health Effects
Observed in Swimmers." In Water Chlorination: Environmental Impact and
Health Effects, pp. 11-25.
2k
American Public Health Association, American Water Works Association, and
Water Pollution Control Federation. Standard Methods for the Examination of
Water and Wastewater, 14th ed. Washington, DC: American Public Health
Association, 1975, p. 132.
-------
V - 363
References continued
25
US Environmental Protection Agency. "National Secondary Drinking Water
Regulations."
^Drexel University, 1978, p. 98.
27
US Environmental Protection Agency. Manual of Individual Water Supply
Systems (EPA-430/9-74-007). Washington, DC: US Environmental Protection
Agency, 1974, p. 7.
28McKee, 1963, p. 283.
29
Hem, John D. Study and Interpretation of the Chemical Characteristics of
Natural Water. Washington, DC: US Government Printing Office, 1970.
30McKee, 1963, p. 273.
31
US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations, pp. 122, 123.
32
National Academy of Sciences, 1977, p. 443.
33
National Academy of Sciences, 1977, p. 440.
^Hem, 1970, p. 225.
35
American Public Health Association, p. 200.
36Hem, 1970.
-------
V - 364
References continued
US Environmental Protection Agency. "National Secondary Drinking Water
Regulations." A similar statement is available in American Heart Association.
"Impact of the Environment on Cardiovascular Disease: Report of the American
Heart Association Task Force on Environment and Cardiovascular System."
1980.
38
US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations, pp. 43, 45.
39McKee, 1963, p. 151.
40McKee, 1963, p. 151.
41
World Health Organization. International Standards for Drinking-Water.
Geneva: World Health Organization, 1958, p. 29.
42
Shearer, Lois Ann; Goldsmith, John R.; Young, Clarence; Kearns, Owen A.; and
Tamplin, Benjamin R. "Methemoglobin Levels in Infants in an Area with High
Nitrate Water Supply." American Journal of Public Health 62 (September 1972):
1174-1180.
43
National Academy of Sciences, 1977, p. 424.
44
McCaull and Crossland, 1974.
• .
Ibid.
^National Academy of Sciences, 1977, pp. 412-414.
47
US Environmental Protection Agency. "National Secondary Drinking Water
Regulations."
-------
V - 365
References continued
48
National Research Council, Subcommittee on Iron of the Committee on Medical
and Biological Effects of Environmental Pollutants, Division of Medical Sciences
Assembly of Life Sciences. Iron, p. 112. Baltimore, MD: University Park Press,
1979.
49
McKee, 1963, p. 214.
^National Academy of Sciences, 1977, p. 265.
^National Academy of Sciences, 1977, p. 402.
52
National Academy of Sciences, 1977, p. 410.
53
US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations, p. 124.
54
National Academy of Sciences, 1977, pp. 400-411.
^Federal Register, Vol. 45, No. 168, August 27, 1980, p. 57336.
^National Academy of Sciences, 1977, pp. 400-411.
^National Academy of Sciences, 1977, pp. 310-311.
58
US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations, p. 72.
59
Patterson, Clair C., and Settle, Dorothy M. "Lead in Albacore: Guide to Lead
Pollution in Americans." Science 207 (March 14, 1980):1167-1176. AND
National Academy of Sciences, 1977, pp. 255-256.
-------
V - 366
References continued
^°US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations, pp. 48-50.
61McKee, 1963, p. 206.
^Patterson, Clair C., 1980.
63US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations, pp. 71,72,115.
^National Academy of Sciences, 1977, p. 420.
^National Academy of Sciences, 1977, p. 362.
^National Academy of Sciences, 1977, pp. 344-345.
^National Academy of Sciences, 1977, p. 369.
68
US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations, pp. 113-116.
69
US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations, pp. 48-50.
^National Academy of Sciences, 1977, pp. 390-399.
71US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations, p. 66.
72
Prasad, Ananda S. Trace Elements and Iron in Human Metabolism, pp. 55-60.
New York 6c London: Plenum Medical Book Co., 1978.
-------
V - 367
References continued
73
McCaull, Julian. "Building a Shorter Life." Environment 13:7 (September 1971):
p. 3 ff.
74,, ..
Ibid.
75US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations, p. 59.
^McCaull, "Building a Shorter Life," 1971.
^^McKetta, John J., and Cunningham, William A. Encyclopedia of Chemical
Processing and Design, Vol. 5, pp. 406-421. New York: Marcel Dekker, Inc.,
1977.
Ibid.
79ik-,i
Ibid.
^McCaull, "Building a Shorter Life," 1971.
81
US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations, p. 61.
x?
McCaull, "Building a Shorter Life," 1971.
83
Jonasson, I. R., and Boyle, R. W. "Geochemistry of Mercury." Special
Symposium on Mercury in Man's Environment, Ottawa, Canada, February 1971.
^Prasad, 1978, p. 373.
85
US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations, p. 61.
-------
V - 368
References continued
86
US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations, p. 79.
07
Grant, Neville. "Mercury in Man." Environment 13:4(May 1971):2-15, AND
Wood, John M. "A Progress Report on Mercury," 14:l(January/February 1972):
33-39.
88
National Academy of Sciences, 1977, p. 277.
89
US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations, pp. 48-50.
90
National Academy of Sciences, 1977, p. 246.
91
US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations, p. 58.
^National Academy of Sciences, 1977, p. 230.
^National Academy of Sciences, 1977, pp. 314-315.
94
National Academy of Sciences, 1977, pp. 289, 290.
95Ibid.
96
National Academy of Sciences, 1977, p. 291.
97
World Health Organization. European Standards for Drinking-Water. Geneva:
World Health Organization, 1970, p. 33.
98
US Environmental Protection Agency. "Interim Primary Drinking Water Regula-
tions—Control of Organic Contaminants in Drinking Water." Federal Register
43: No.28 (February 9, 1978):5756-5780.
-------
V -369
References continued
99
US Environmental Protection Agency. National Interim Primary Drinking Water
Regulations, p. 106.
^^National Academy of Sciences, 1977, p. 898.
*®*US Environmental Protection Agency. "Drinking Water Regulations: Radio-
nuclides." Federal Register »1; No. 133 (July 9, 1976):28402-28405.
102
US Environmental Protection Agency. "National Secondary Drinking Water
Regulations."
103
Bruvold, William H. "Consumer Attitudes Toward Taste and Odor in Water."
Journal of the American Water Works Association (October 1977):562-56^.
10^
Temple, Barker <5c Sloane, Inc. Survey of Operating and Financial Characteristics
of Community Water Systems. Submitted to US Environmental Protection
Agency, Washington, DC. Wellesley Hills: Temple, Barker & Sloane, Inc., 1977.
^^Temple, Barker & Sloane, Inc., 1977.
-------
VI
Composites ol
US Rural Water Supply Status
In the preceding chapter, the status of US rural water supply was
characterized principally in terms of quality, quantity, availability, cost, and
affordability. Depending upon their complexity and the conventions for represent-
ing them empirically, these factors were originally expressed as different sets of
variables, or indicators, which reflected specific properties of each status com-
ponent. Quality, for example, was delineated by a large number of discrete
chemicals,- bacteriological contaminants, and other waterborne substances which
were detected through laboratory assays. A second dimension of quality consisted
of tastes, odors, colors, and other aesthetic aspects which may have been noticed
by household members.
Availability, while not as conceptually elaborate as quality, was also
described with respect to a set of unique measures, including indicators of
accessibility and reliability. Quantity, cost, and affordability were not measured
as comprehensively, however, in part because of certain practical obstacles which
could not be eliminated within the constraints of the NSA. An accurate measure of
water quantity, in particular, would have required continued monitoring of house-
hold water use over an extended period of time. At many households, in fact, a
V1-1
-------
VI - 2
precise estimate of water consumption could not have beer, obtained without
installing some type of metering device. Likewise, detailed assessments of cost
and affordabilitv could ha-e been accomplished oniy through a prolonged investiga-
tion of household financial resources, income, and expenditure patterns. The da:a
on quantity, cost, and affordabilitv, which were restricted to one or two indicators
for each factor, made it possible to explicate those status components in
Chapter V. However, the information on quality and availability was much too
extensive to allow a generalized interpretation. Consequently, composite measures
were devised for quality and availability, and are presented in Chapter VI. Thus,
although Chapter VI is actually an extension of Chapter V's description of the
status of US rural water supply, it includes only two of the five factors that were
used in that description.
There were a number of compelling reasons for constructing summary or
composite measures for quality and availability as an alternative to employing the
more specific raw data. Some reasons were peculiar to the NSA, which was
designed to be an overview of rural water conditions, while others were related to
dictates of the measurement process. First, although they typically did not
encompass all the relevant information, the composites allowed general patterns to
be established more easily than could have been done by an examination of
individual parameters. Second, the emphasis on general patterns, rather than on
details, was more compatible with the NSA's purpose as an overview. Third, the
use of composites reduced the volume of material that could have been included in
the analytical portions of this report to a manageable level. This last consideration
was especially critical in the NSA because, unless the composites were formulated,
separate models eventually would have had to be specified for each variable and its
postulated determinants. The effort to develop the composites that are presented
in this chapter was also consistent with the ultimate objective of producing any
-------
VI - 3
summary measure, which is to make the data more comprehensible vnlie having a
minimal effect on their original properties.
As a general procedure, the formation of composites involves synthesizing
variables according to a prescribed set of principles. These principles usually
differ as a function of several conditions, including the diversity or the data
collected for each variable, the units in which the variables are expressed, and the
substantive meaning to be conveyed by the composite, to name only a few. The
present chapter was organized so as to maintain the distinctions between the
principles within the context of each factor and its indicators. Since water quality
is of considerable importance in determining the status of rural water supplies, and
because its set of indicators was the most extensive, the composites for this factor
are presented first. This section of Chapter VI is further subdivided into two parts
to acknowledge the fact that quality was measured in terms of information
produced by laboratories and in terms of perceptions of the rural people who used
• the water. The other major section of Chapter VI, which is also divided into two
parts, is arranged according to the summary measures of availability.
Except for the indices that were generated from the physical constituents
of water quality, which evolved from a common strategy, each composite pre-
sented in the chapter is considered independently. Specifically, the purpose of
each index is established, the appropriate indicators are specified, and the
composite's structure is described. Following the development of the index, its
distribution of scores is compiled for the nation and for the four subnational
groupings (region, SMSA/nonSMSA, size of place, and size of system).
WATER QUALITY COMPOSITES
Water quality was broadly defined in the NSA as "the adequacy of water
for human use." Although the major emphasis of this definition was on water's
effects on human health and well-being, its aesthetic and economic implications
-------
VI - 4
were also recognized. It was realized, of course, that v/ater could be suitable for.
consumption in this narrow sense and still be of little use to human beings if it
were not available in sufficient cia.i titles at reasonable prices. These litter
considerations were addressed separately in the NSA, however, as explained
elsewhere in this report.
It was further acknowledged that domestic water quality could be judged
adequate or inadequate by some particular standard based on physical and chemical
measurements, yet still be considered inadequate by water users. This determina-
tion could be made on the basis of personal experiences and observations or
information from other sources. Regardless of its origin, the decision that water
quality is unsatisfactory could have numerous consequences—reduced consumption,
use of alternative supplies, demands for additional water treatment, and pressure
for regulatory change, to name some of the more significant. The user's perception
of water quality as measured in the NSA was described in Chapter V. Also, the
relationships between those perceptions and the measured values of water quality
will be explored in a subsequent chapter. Here, however, the immediate focus is on
the composite indices developed by NSA investigators to represent biological,
chemical, and physical characteristics of water. Later in this chapter, judgments
about tangible attributes (taste, odor, color, and so forth) will be integrated into a
composite that is designed to reflect water quality strictly in terms of its
perceptual aspects.
'NDICES OF PHYSICAL MEASURES
The households included in the NSA could be contrasted in terms of any
single physical or chemical characteristic without appreciable difficulty. Evidence
for this is provided by the portion of Chapter V where such comparisons were
accomplished and by innumerable other studies on the subject of water and its
constituents. However, it was decidedly futile to compare households on the basis
-------
VI - 5
of several of the over 40 physical and chemical characteristics measured in the
NSA at once. For instance, how would one evaluate the differences in water
quaiitv between two groups of households, oie with supplies that had high
ccnce itrations of sulfates and manganese but with low levels of iron and turbidity,
the other having supplies with an abundance of iron and turbidity but with small
amounts of sulfates and manganese? Unless one employed a composite approach,
any relative assessment of water quality would be restricted to making observa-
tions about each individual constituent.
The NSA indices of physical measures provided the mechanism for combin-
ing a large amount of information in a reduced, more intelligible form. The
objective was a set of summary measures of water quality, based on the
constituents measured in the survey, which would facilitate an understanding of
overall water quality and allow comparisons among groups of rural households.
These indices would also be used as analytical devices in a subsequent effort to
specify factors that affect the quality of water produced by rural water supplies
(see Chapter X).
In reviewing advantages and disadvantages of indexing, NSA researchers
discovered that there had been a substantial professional interest in developing
indices to describe water quality. The most fundamental argument in favor of
indexing has been the ability of an index to characterize generally a complex
situation involving a number of diverse water quality attributes. When constructed
properly, an index enhances the ability to make sense of an otherwise incompre-
hensible assortment of facts. The major argument against an index has been that
the simplification entails an unacceptable loss of specificity and an uninterpretable
distortion of the original data. One particular difficulty has been that the level of
precision associated with measuring any given constituent cannot be conveyed well
in an index. However, this objection presumes that the units in which the data are
expressed must be retained to establish differences, which is not necessarily
-------
VI - 6
correct. Also, as -vii 1 be demonstrated I;i this chapter, transformations are
available that preserve at least the order of the raw data, which satisfies the "".est
critical requirement of index construction.
Background
The principles used to construct the water quality indices in the NSA were
based to a large extent on precedents that emerged from other efforts to develop
summary measures of water quality. Since 1365, when the first formal index
appeared, there have been numerous attempts to represent gradations in water
quality on a composite numerical scaled Some of these indices, such as those
2 3 4 5
formulated by Horton, the National Sanitation Foundation, Prati, McDuffie,
and Dinius,^ were designed to measure the quality of surface waters. In contrast,
7 8 9 10
composites proposed by O'Connor, Deininger, Walski and Parker, Stoner, and
Nemerow and Sumitomo,^ were structured to reflect the different sets of uses
that water can support. Another class of indices, which included composites
12 13 15
devised by the MITRE Corporation, ' Dee, Inhaber, Johanson and
Johnson,^ and Zoeteman,^ was designed specifically for administrative decision-
18 19
making. Finally, indices developed by Harkins, and Schaeffer and Janardan,
were distinguished by their orientation towards established statistical approaches.
These indices in particular, along with several others that have been proposed, are
20
summarized more thoroughly by Ott.
Although the water quality composites referred to in the preceding
paragraph were too diverse to compare systematically, each one evolved from the
same general process. This process, expressed as a series of distinct stages,
involved selecting an underlying purpose for the index, designating a set of
indicator constituents, determining the importance of each indicator or parameter,
and specifying a method for combining the constituents into a single structure. In
-------
VI - 7
each case, the outcome of applying the process was a single value which was
employed to differentiate bodies of water in terms of their quality.
The purpose that a composite is designed to satisfy is an important
consideration in all IVex construction efforts. In t.Ifect, the purpose establishes
the basic components and the limitations of the indices. Previous indices have
been intended to reflect such varied concerns as the hygienic quality of water, the
cost or effectiveness of pollution abatement activities, the aesthetic condition of
water, water's suitability for specific uses, and other aspects of quality. The
purpose of the NSA indices, as suggested in an earlier section of this chapter, was
to measure the quality of the water provided by rural household supplies.
Indicator constituents are selected according to the information an index is
designed to convey. In previous indices, constituents relevant to the substantive
emphasis of the index generally were identified by literature searches or by
soliciting the judgments of experts. As indicated below, the NSA used a variant of
this approach to compile its set of water quality indicators.
While some indices assumed equal importance among constituents, it was
more common to assign them weighting factors. Generally, this was accomplished
by asking knowledgeable people to evaluate the relative significance of each
indicator, given the composite's purpose. The resultant weights usually took into
account the differential effect of constituents according to their potential
concentrations. As noted later in the chapter, the weights for the NSA indices
were based on the current water quality standards or the distributional properties
of the constituents. These techniques provided the advantage of being compara-
tively more objective than the procedures used to produce weights for many of the
>
other indices.
Independent of the three other stages, the last part of the index construc-
tion process is to combine the indicators into a single' entity. Most often, this has
been done by devising explicit mathematical functions for subindices and
-------
VI - s
aggregating these into a composite. In the NSA, this approach vas discarded ir.
favor of a strategy that had a more formal statistical basis. This strategy is
considered in more detail in one of the foiio-.ving sections, as are the .ncicators for"
the composites and the development of the .veighung .'actors.
The initial set of indicator constituents
Because of a variety of unique circumstances, the NSA indices did not
present some of the problems which hindered the efforts of other investigators. A
major distinction was that the NSA indices were developed to reflect the quality of
domestic water, while other indices were devised for water that was used as a
drinking water source, for recreation, or for some other purpose. Consequently,
the characteristics incorporated into the latter indices tended to be fairly general
ones which indicated the broadest aspects of water quality. Constituents examined
in the NSA, on the other hand, were usually of direct significance for the quality of
domestic water.
The development of the NSA indices was augmented by substantial
information that was refined even before the survey was undertaken. Constituents
measured in the NSA originally were considered in conjunction with the effort to
promulgate federal drinking water regulations. These regulations were codified as
interim maximum contaminant levels (MCLs) in accordance with provisions of the
Safe Drinking Water Act of 1974. The predominant underlying concern of this
legislation and the MCLs was with the healthfulness of drinking water. As part of
the process by which the regulations evolved, the EPA, with input from other
sources, produced a preliminary list of water contaminants that posed a possible
21 22
health threat. ' Other considerations that influenced the decision to include a
particular constituent in this list were the economic effects of properties such as
water hardness and the aesthetic implications of characteristics such as color.
-------
VI - 9
Also considered were the expense of controlling the concentra: ens o: ::-e /*r:ous
constituents and the expected public exposure to the constituents.
Additional deliberations by relevant federal agencie" and public officials
resulted in a final list of substances which were related to the broad'/ defined
concept of water quality. Finally, interim primary MCLs were established for
more than twenty contaminants deemed to be potential health threats or to
indicate such threats. Public water systems with fifteen or more service
connections or which served 25 or more people for more than 60 days each year
were required to provide water that did not have constituent levels exceeding the
MCLs.
With respect to the NSA, the MCL was selected as the principal criterion
for determining which constituents should be included. A total of twenty
contaminants with primary MCLs were tested for in the water specimens obtained
from households in the NSA (Table VI-1). Radioactivity was measured in terms of
eight specific substances and the total amount of radiation, provided that certain
screening levels were exceeded. As a consequence of applying this selection
principle, all of the substances for which there were official interim primary MCLs
were measured directly or indirectly in the NSA survey, although those that were
expected to be less prevalent were measured only in a 10 percent subsample.
A number of other constituents on the preliminary federal review list were
23
determined to be more closely related to economic or aesthetic effects. Some of
these (a total of thirteen) were assigned secondary MCLs which, while not federally
enforceable, were intended as guidelines for state officials. The NSA examined
five of the thirteen constituents with secondary MCLs. Other substances on the
preliminary list exerted effects that were so ambiguous, indeterminate, or complex
that they precluded assignment of an MCL. Some of these—sodium, for example
—were designated for further attention or research. Also included in this group of
substances were constituents that had standards which were judged not to be
-------
VI - 10
Table VI-1
Constituents Studied in NSA Survey
Constituent Primary MCL Secondary '-'CL No MCL
Assayed in All '.Vater Specimens:
Total colilorm X
Fecal coliform X
r-ecai su'epio^o^cus X
Standard plate .count X
Nitrate-N X
Sulfates X
Calcium X
Magnesium X
Iron X
Manganese X
Sodium X
Lead X
Turbidity X*
Color X
Specific conductance X
Total dissolved solids X
Assayed in 10 Percent of the Specimens (the NSA Group II Subsample):
Arsenic
X
Barium
X
Cadmium
X
Chromium
X
Mercury
X
Selenium
X
Silver
X
Fluoride
X
Gross Alpha
X
Gross Beta
X*
Endrin
X
Lindane
X
Methoxychlor
X
Toxaphene
X
2,4-D
X
2,4,.5-TP
X
Also Measured in NSA Group II Subsample if Warranted by Screening Tests:
Radium-226
X
Radium-228
X
Uranium
X
Strontium-89
X
Strontium-90
X
Cesium-134
X
Tritium
X
Iodine-131
X
Total MREM
X
~Denotes constituent with primary MCL that was not applicable for the NSA.
-------
VI - 11
applicable to the NSA. Gross beta radiation, for instance, had an V,C. that was
employed only in conjunction with monitoring of large community systems or
surface water sources that were located adjacent to nuclear generating facilities.
Since water for rural domestic supplies was extracted so infrequently from these
latter sources, only the gross beta screening test was used. Therefore, gross beta
was treated as if no MCL had been established, even though one is stipulated in the
regulations. Turbidity was likewise excluded because it is not specific and can be a
double count of other constituents with MCLs. By virtue of their presence on the
list of substances for which the NSA compiled data, all of the preceding 41
substances were considered viable candidates for the NSA indexing effort.
The revised set of indicators
In theory, any of the substances included in Table VI-1 could have been
incorporated into a water quality index. However," since several indicators were
too interdependent to make a unique contribution to a composite, some constitu-
ents were eliminated. First of all, fecal coliform, fecal streptococcus, and the
standard plate count were dropped from the list. Although values of those items
were of considerable supplemental help in describing the implications for total
coliform counts, as discussed in Chapter V, the measurements were often duplicate
counts of more specific bacteria types that had already been enumerated by the
total coliform test—which did have an MCL—and therefore introduced the
probability of statistical redundancy (confounding). As a result, the primary
indicator of bacteriological quality in the NSA was the total coliform count. This
decision was consistent with the statistical requirements of the NSA, and with the
professional judgment that this count was the best of the available measures of
bacteriological quality (see Chapter V).
Despite this rationale, one should be aware of objections to relying
exclusively on the total coliform count as the indicator of bacteriological water
-------
VI - 12
quality. In particular, the test seems less reliable for untreated vvarer supplies
than for those producing finished water. To illustrate this more specifically,
untreated rura.1 supplies, which typically were wells, might have certa;:. cor.ci ::ons
that could reduce the sensitivity of the total co!i.f<->rrn test to coliform bacteria.
The effect may be to underestimate the number of ^o!iir^rms because they couid
-?u.
not be detected or distinguisned from other organisms.
Although the preceding objection may be legitimate, the total coliform
count has continued to receive professional support because nothing better has
25
been discovered to replace it. This was the most important reason for including
it in the NSA indexing effort. Another related reason for retaining, the total
coliform count rather than the other indicators of biological contamination was
that it was the only one of the four to have been assigned an MCL. Finally, it is
important to recall that the primary purpose of the composites was to rank rural
households in comparison with each other. Since the MCL was the acknowledged
standard, the use of the total coliform count allowed the NSA to specify the
bacterial quality of water relative to the level prescribed by federal regulations-
Besides these bacteriological constituents, specific conductance and total
dissolved solids also were interdependent. In particular, specific conductance was
converted to an estimate of total dissolved solids which are considered to have
economic and aesthetic effects. Primarily for that reason and because total
dissolved solids does have a secondary MCL in the federal regulations, specific
conductance was not considered further.
Redundancy between indicators was additionally implicated in the decision
to delete uranium, radium-226, and radium-228 from the list of potential index
constituents. The specific reason was that the values for gross alpha radiation
were included as a screening test for uranium and radium-226. It was assumed by
the EPA that the test was also sufficient for detection of radium-228 since its
daughters are alpha emitters even though radium-228 itself is considered a beta
-------
VI - 13
emitter. Radium-228, of the thorium series, is usually of geologic origin and thus
more likely to be important in groundwater rather than surface water. Alpha
radiation is generally of greiter concern than beta radiation in groundwater. Since
the gross beta standard applies to surface water, radium-22S was included under
the gross aipha screen. To avoid double counting, a choice had to be made between
values for either gross alpha or uranium, radium-226, and radium-228. Values for
gross alpha were the more general measure. The gross alpha values, unlike the
specific radionuclides, were measured at all subsample households (see Chapter V)
and were retained for the water quality composites.
Similarly, gross beta radiation was included in the NSA index list rather
than the specific beta-emitting radionuclides. Beta activity, similar to alpha
activity, was monitored in all NSA subsample specimens, but the specific radio-
nuclides were measured only when the gross beta screening value was sufficiently
high, and were therefore unavailable for most households. These substances, which
included strontium-89, strontium-90, cesium-134, tritium, iodine-131, arid a calcul-
ation of total millirems, were omitted from the NSA set of water quality indicators
along with radium-226 and radium-228.
Other constituents, although legitimate candidates for a water quality
composite, were excluded from the index list because they were present only in
concentrations smaller than the pre-established minimum detection levels specified
by the EPA. For all of these substances, which consisted of herbicides and
pesticides, the detection levels were substantially lower than their respective
MCLs. Since the exact quantities of these constituents were below the detection
limits, each one was assigned a value equal to the particular level of detection,
which was constant for all households in the subsample. This absence of variation,
which was associated with endrin, toxaphene, 2,4-D, and 2,4,5-TP, eliminated any
possibility that these constituents might contribute to a composite. This was the
-------
Reproduced from
best available copy
VI - 14
principal reason for discarding them from the effort to develop tr.e water quality
indices.
Two other constituents, calcium and magnesium, were or.iy of very minor
Significance to water quality when considered individual y. out levels of the two
substances together represented the degree of hardness in water (see Chapter V).
Hardness is widely regarded as an important component of water quality, and was
26
considered for inclusion in the federal regulations. The characteristic was
finally excluded from the regulations because of insufficient information about its
significance, but it was included in the NSA as a water quality indicator with
economic consequences. The list of constituents compiled for the composites
included "hardness" as calculated from calcium and magnesium rather than
incorporating the two elements independently.
In summary, the original collection of 41 water quality indicators was
shortened to 23 constituents (Table VI-2). Those substances with primary MCLs
decreased to sixteen items, while all five constituents with secondary MCLs were
retained. Proportionately, the largest reduction occurred to the set of nonMCL
constituents, which was diminished to two: hardness and sodium. Each of the 23
substances represented a potential component of any water quality composite
developed for the NSA.
Effect of sample design
In addition to the original indicators and their properties, the sampling
approach of the NSA imposed another constraint on the development of the water
quality indices. The differential prevalence of constituents suggested that some
substances would probably not be present in sufficient concentrations to justify
laboratory assay for them in all sets of water specimens. Certain of these •
substances were monitored in the NSA, however, because of their inherent toxicity
and the possibility that they were more widely dispersed than expected. Also,
-------
VI - 15
Table VI-2
Constituents Selected for the V/ater Quality Composites
Constituent
Primary MCL Secondar
-o MCL
Assaved in All Water Specimens:
Total coliform
X
Nitrate-N
X
Sulfates
X
Hardness
Iron
X
Manganese
X
Sodium
Lead
X
Turbidity
X
Color
X
Total dissolved solids
X
Assayed in 10 Percent of the Specimens (the
NSA Group II Subsample):
Arsenic
X
Barium
X
Cadmium
X
Chromium
X
Mercury
X
Selenium
X
Silver
X
Fluoride
X
Gross Alpha
X
Gross Beta
X
Lindane
X
Methoxychlor
X
X
X
-------
VI - 16
measurements were compiled on these particular substances to generate a baseline
with which the results of subsequent studies on water quality could be compared.
Therefore, as indicated in Table VI-2, althoigh specimens were ob ained from, each
household selected for the survey, only elevsn index constituents '/ere assessed for
all 2,654. Levels of the remaining twelve substances were determined for only 267
households, which represented about 10 percent of the entire household sample.
The decision to assay some constituents at a subsampie of households had
some major implications for the effort to construct the water quality composites.
First, since contemporary measurement theory assumes that data are attached to
common observational units, one constituent could not be combined with another
unless both pertained to the same household. Obviously, this required that any
prospective combinations of constituents be aligned with either of the two
household groupings stipulated by the sample design (full sample or subsampie). To
illustrate, total coliform and nitrate could hypothetically appear in a composite
which reflected some aspect of water quality for all households in the survey.
Moreover, the constituents of arsenic and barium could be integrated into a
separate index, but only for the 267 households in the subsampie. If a composite
were to be formed from these four indicators, it would be relevant only to the
subsampie of households. Therefore, the primary effect of the sample design was
to restrict how variables could be combined. Ultimately, then, the sample design
impinged upon the NSA's ability to maximize the information conveyed by the
water quality indicators, since certain combinations of constituents could not be
used for a particular set of households.
The sample design employed in the NSA also affected the reliability of any
composites formed from the constituents. Since the statistical precision of an
estimate tends to increase with the size of the sample, composites involving only
constituents examined at all households in the survey would have a higher
probability of being accurate than indices that included any constituent assayed at
-------
VI - 17
the subsample of households. vVhile any composite c: uld be used to make
projections to all rural households or to selected suopopulations, there would be
less confidence in an estimate derived from the subsamoxe because of the smaller
base (10 percent of the full sample). Therefore, a sum:, ary measure incorporating
only constituents surveyed in all specimens would have maximum reliability, while
one including any of the remaining substances would be considerably less reliable.
Further, this aspect of the NSA sample design meant that, beyond a point,
the information contained in a composite could not be increased without precipi-
tating a reduction in its level of precision. For example, if one assumed that there
were no physical interdependencies among the 23 constituents in Table VI-2, the
composite that would be maximally informative would have included all indicators
in the set. However, since that index would have embodied some substances that
were examined only in the subsample, it would not have been as reliable as a
composite that consisted exclusively of indicators measured at the full sample of
households. The effort to construct water quality indices in the NSA, therefore,
explicitly recognized the necessity for making an optimum choice given these two
considerations.
General development strategy
The development of the NSA water quality composites proceeded according
to a general strategy that evolved from the NSA's research requirements. This
strategy suggested the number of indices to be formulated, their purposes, the
constituents that each index was to encompass, and a method for combining the
substances. The set of decisions arrived at in the process of applying the strategy
represented a compromise between the constraints imposed by the NSA's design,
the revised set of indicators, and their properties.
Since water quality is a complex phenomenon consisting of several distinct
but interrelated aspects, NSA researchers determined that it could not be
-------
VI - 13
measured effectively with a single composite. The preference that evolves ••••as to
summarize water quality in terms of two specialized indices and a single
comorer.ensive or ^enerai measure. Each of the three indices fulfilled its o; ln:i;al
i O
purpose, which was to facilitate the intelligibility aiw' interpretation of tn<. cat - on
the physical indicators of water quality.
The first of the specialized water quality indices was based on those
constituents that had a definite and demonstrable effect on human health (see
Table V1-3). Operationally, this index included only indicators that were considered
important enough to have been assigned primary MCLs in the interim federal
regulations, excluding three substances which were prominent enough to be
analyzed separately (total coliform, nitrates, and lead), and two constituents
without applicable MCLs (turbidity and gross beta). The second specialized index,
which was composed of the five substances with secondary MCLs, was confined
generally to constituents that had aesthetic or economic implications. Finally, the
third proposed index was designed to be more comprehensive by acknowledging the
entire spectrum of effects associated with the water quality indicators. This index
synthesized all 23 constituents, including those that did not have an MCL.
Compared to one another, the health and general indices contained more informa-
tion but were less reliable than the aesthetic/economic index. (The aesthe-
tic/economic index was based on water quality data fr'om the full NSA sample of
households, while the health and general indices were based on the subsample. As
with other material presented in this report, this distinction is important because
the sample size affects the statistical precision of the estimates.)
Given the disparity of constituents and the units in which they were
expressed, the three indices could not be developed by directly combining their
respective sets of indicators. Consequently, NSA researchers decided to formulate
the indices by employing a collection of procedures which is referred to as derived
scoring. The essential advantage derived scoring has over other techniques is that
-------
VI - 19
Table VI-3
Components of Water Quality Indices
Index
Constituent
L.tiect
1 - Health Risk Index (267 households)
Arsenic
Barium
Cadmium
Chromium
Mercury
Selenium
Silver
Fluoride
Gross Alpha
Lindane
Methoxychlor
Health
Health
Health
Health
Health
Health
Health
Health
Health
Health
Health
2 - Aesthetic/Economic Index (2,654 households)
Sulfates
Iron
Manganese
Color
Total dissolved solids
Aesthetic, health
Aesthetic
Economic, aesthetic
Aesthetic
Economic, aesthetic
3 - General Index (267 households)
Total coliform
Hardness
Nitrate-N
Sulfates
Iron
Manganese
Sodium
Lead
Turbidity
Color
Total dissolved solids
Infectious disease
Economic
Health
Aesthetic, health
Aesthetic
Economic, aesthetic
Health
Health
Aesthetic, health
Aesthetic
Economic, aesthetic
-------
VI - 20
Table VI-3 continued
Index Constituent
3 - General Index continued:
Arsenic Health
3arium Health
Cadmium Health
Chromium Health
Mercury Health
Selenium Health
Silver Health
Fluoride Health
Gross Alpha Health
Gross Beta Health
Lindane Health
Methoxychlor Health
Reproduced from
best available copy
-------
VI - 21
it permits the researcher to establish the relative position oi an observational unit
or household with respect to a standardized referent. This referent can be
internally generated, assuming that the sample oi observations is representative, or
it can be externally obtained from another sample of observations, in both cases,
however, the position of the unit is determined by converting raw data into a
relative measure.
Of the various methods that could be used for the conversion, NSA
researchers chose to use standard scoring. This common method relies upon linear
or nonlinear transformations of the raw data to produce the derived scores. These
standard scores express a unit's distance from some point in a distribution, usually
the mean, in terms of the standardizing criterion, typically the distribution's
standard deviation. When computed with a linear transformation, the standard
scores retain the exact numerical relations that were present in the original data;
also, their distributional properties are identical to the properties of the original
distribution. Since the properties of the two distributions correspond, any
manipulation of the raw data can also be done to the standard scores without any
distortion of results. Distortions become more severe with nonlinear transforma-
tions, although the order of the observational units in the distribution of standard-
ized scores relative to the original distribution can be preserved. As will be
discussed in the following segments of this chapter, each of the three water quality
composites in the NSA was computed by applying a version of the standard scoring
technique.
Although the methods used to construct them were substantially similar,
the three indices conveyed different levels of information with varying degrees of
statistical precision. When considered from this perspective, each composite was
somewhat distinctive and fulfilled a function that could not be performed by the
other two. This uniqueness was manifested in the structures of the composites and
the patterns of variation that they were designed to disclose.
-------
VI - 22
Lie ex of health risk
The health risk index, one of the t'.vo specialized composites, consisted of
z>e eleven water quality indicators with primary MCLs. Relative measures i ;r the
index were derived for each indicator by using its primary MCL as a standardizing
device. The MCL was selected, rather than some alternative referent, because
each of the constituents was directly comparable at those specific concentrations.
For each substance, this comparability was established by the MCL, which
implicitly equates values above the standard as unacceptable. Another reason for
incorporating the MCL into the index was that the relative importance—or
weighting—of the various contaminants was specified by the levels at which the
MCLs were set. Consequently, the NSA researchers did not have to attempt the
difficult task of assigning separate weighting factors to the substances when
combining them into the index. In other words, the specific values stipulated in the
regulations reflected weights that had evolved from public health experience as
well as from the process of debate and review that preceded the promulgation of
the standards. To illustrate, distinct criteria and widely differing safety factors
were devised to designate the potential risks posed by the various substances. The
built-in safety factor for cadmium, for example, was only four, while the safety
27
factor for endrin and others of the chlorinated hydrocarbon insecticides was 500.
Safety factors were set primarily in relation to the nature of the health
hazard, but also in relation to the extent of scientific knowledge about the
constituent, the techniques for measuring it, and the expense of controlling it. The
federal government's attitude was summarized in its response to national com-
ments on the proposed MCLs: "A question was raised about the fact that different
safety factors are contained in various maximum contaminant levels. The levels
are not. intended to have a uniform safety factor, at least partly because the
knowledge of and the nature of the health risks of the various contaminants vary
widely. The levels set are the result of experience, evaluation of the available
-------
VI - 23
?3
data, and prot355iona! jucgement.'1'" Althougn tne use 01 r\e MCl .-encored tr.e
substances tor each of the specialized indices comparable for measurement
purposes, it ?ho recognized the considerable differences am or. j them.
The primary MCLs for the eleven constituents usually consisted of a single
value which could be incorporated directly into the index. In one instance,
however, a value had to oe selected from a range of concentrations. Specifically,
the limits for fluoride ranged from 1.4 through 2.4 milligrams of fluoride per liter
of water, depending on climatic temperature (under the assumption that the
consumption of water, and therefore the total intake of fluoride, is greater in
warmer climates). To resolve this difficulty, the lowest of the range of values was
chosen for use in the composite. This decision was consistent with the underlying
assumption about the weighting of constituents. That is, the MCLs were
considered to be self-weighted, and NSA investigators avoided making further
technical judgments about a substance's importance. The choice, then, was to
assume that the lowest MCL value was more stringent, and that it should be used
as the standardizing criterion.
The derivation of scores for each of the eleven constituents included in the
health risk index was performed in a series of stages. First, the raw values were
placed in ratio to the MCL and multiplied by an arbitrary constant, which was 10.
The purpose of this procedure was to convert the original values iftto some relative
equivalent or standard unit, which also altered their distributions by dispersing the
observations. Then, if this ratio exceeded the constant, another step translated
those observations with values greater than the MCL closer to the point at which
the ratio and constant were equal. This was accomplished by applying a
logarithmic transformation to the raw values and adding the arbitrary constant to
the result. Besides normalizing the distributions, this procedure also prohibited any
one constituent from artificially dominating the index. The values of the primary
-------
VI - 2k
MCLs ior these eleven Indicators are given in Table VI-- aLor.j with other relevant
information about all 23 constituents.
Zxpressed more formally, the scores for each inlioator included in the
health . isk composite were derived with the following set 01 operations:
Xi
R' - hcrer' 10
where
R[ = the ratio of the constituent's raw value to
its MCL for the ith household
Xj = the raw value of the constituent for the it'1
household
MCL = the primary maximum contaminant level for
the constituent.
Further, if Rj < 10,
Si = Ri
where
Sj = the derived score for the i^ household
R[ = the ratio of the constituent's raw value to
its MCL for the i^ household
or, if Rj > 10,
Sj = In (Ri) + 10
where
Si = the derived score for the i^ household
In = the natural logarithm
. Ri = the ratio of the constituent's raw value to
its MCL for the i*h household.
Reproduced from
best available copy
-------
VI - 25
Table VI4
Primary MCLs and Relevant Summary Statistics
for 'IVa-er Quaiity Indicators
Primary
Summarv
Statistics 'L
index
¦"^"^vicuen:
MCLs
Utilized
Mean
r «-x i j _ . „
_ u ric-j.1,
Aesthetic/
:ii ELcuivjmic
General
Total coiifor:n
*
(14157.7)
(1174S2.5)
X
Hardness
—
(165.9)
(202.4)
X
Nitrate-N
—
(1.5)
(2.5)
X
Sulfates
—
56 A
(51.2)
108.0
(105.7)
X
X
Iron
—
0.40
(0.35)
1.30
(1.20)
X
X
Manganese
—
0.060
(0.072)
0.234
(0.201)
X
X
Sodium
—
(37.9)
(105.7)
X
Lead
—
(0.029)
(0.045)
X
Turbidity
—
(1.1)
(4.4)
X
Color
—
4.7
(4.3)
5.6
(3.9)
X
X
Total dissolved
solids
—
473.2
(433.7)
482.9
(369-.0)
X
X
Arsenic
0.05 mg/1
(0.007)
(0.012)
X
X
Barium
1 mg/1
(0.2)
(0.1)
X
X
Cadmium
0.01 mg/1
(0.006)
(0.007)
X
X
-------
VI - 26
Table VI-4 continued
Primary
Surrmarv
Statistics -
ir -ox
Constituent
MCLs
Utilized
Mean
SD
Aes V;e:ic/
Heai:n economic
General
Chromium
0.05 m ! 12
(0.005)
( 0.001)
X
\/
„ \
Mercury
0.002 mg/1
(0.003)
(0.016)
X
X
Selenium
0.01 mg/1
(0.008)
(0.010)
X
X
Silver
0.05 mg/1
(0.030)
(0.011)
X
X
Fluoride
1.4 mg/1
(0.40)
(0.45)
X
X
Gross Alpha
15 pCi/13
(3.3)
(2.6)
X
X
Gross Beta
—
{5.5)
(1.9)
X
Lindane
0.004 mg/1
(0.0023)
(0.0041)
X
X
Methoxychlor
0.1 mg/1
(0.02)
(0.01)
X
X
^Statistics in parentheses were computed for the subsample of households. All
other entries are for the entire sample.
^The specified level is for total chromium.
^Picocuries per liter.
Reproduced from
best available copy
-------
VI - 27
The application of these computational procedures generated a set of
derived scores for each constituent. With mis particular standardizing inetiiod, the
scores were distributed around the point at which the raw value and the MCL v-ere
cvial, which was in turn fixed by the arbitrary constant. The resulting distr ou-
tions of derived scores were approximately symmetric and at least tended towards
normality. Additionally, the order of the households was maintained, although the
differences between units were no longer meaningful.
Structure of the health risk index
The health risk composite was formed by aggregating the derived scores
across the eleven constituents studied in the N5A for which primary MCLs had
been established. More formally, the general expression for this composite was
11
HRi = I Sij
j = 1
where
HRj = the value or score for the household on
the health risk index
Sij = the derived score for the i^ household on
the constituent
11=3 = the number of constituents that were incor-
porated into the composite.
The resultant index was self-weighting and the magnitude of any one indicator's
contribution was determined by the relative stringency of its particular MCL.
The health risk composite was designed to be employed for two kinds of
comparisons. One was the comparison between index values assigned to individual
households or some particular subset of households. The second was the direct
comparison of the household's index value to the reference level fixed by the
-------
VI - 28
primary MCLs (equivalent to the arbitrary constant multiplied by eleven. which
was the number of constituents in the index).
The first comparison specified the position of different house.:old'.) relative
to each other in terms of the health risk associated with their wat-ir f Voiles. This
comparison provided a sensitive indication of the differences among households, as
single entities or as elements of some agg. ^gate (211 r'j- h?'.!conoid5 that obtained
water from individual systems, for example). Consequently, it was particularly
useful for determining variation among households of different types or in different
geographical locations.
Consistent with the structure of the index, a household with a larger value
was assumed to be exposed to more of a health risk than a household with a smaller
value. However, since the magnitude of the differences among households had no
interpretable meaning, no judgment could be made about the degree to which the
health risk associated with one household's water supply exceeded that of another.
In this sense, the index was ordinal, and simply ranked the households on the
continuum it represented.
The second comparison, by contrast, examined households in relation to a
fixed referent rather than in relation to other households. Specifically, the index
quantified the likely health risk associated with particular households' water
supplies, and allowed a comparison of these levels with acceptable levels as
reflected in federal regulations. For each constituent, the referent in the
composite was the point at which the concentration of the constituent was equal to
the NSA reference value, which was derived from the MCL (see Chapter V). When
compiled for the entire set of constituents, the combined referent or composite
reference value was 110.
In interpreting any of the values for the health risk composite, it must be
emphasized that the index was not designed to accommodate absolute judgments.
If one household's index value is twice as large as another's, the household cannot
-------
VI - 29
be said to Have water of half the quality. Likewise, it cannot be said that the
health risk at the first household is f.vice as great. However, the difference in
values certainly indicates that tne first household's water suppiy is lower in quality
and less i.ealthful than the second household's. Furthermore, greater differences
imply greater potential health risk, although tne extent of the risk is uncertain.
National estimates of health risk
The set of values compiled for the health risk composite displayed only
modest variation across all rural households, considering that it represented eleven
physical constituents. For example, index scores had a maximum of 60 and a
minimum of 17, compared to a theoretical maximum in excess of 120 and a
minimum of 0. Figure VI-1, which presents the distribution of values on the health
risk index, suggests that the scores tended to be low. Approximately 29 percent of
the households had scores of 25 or less, while about 42 percent had scores of 26
through 35. None of the households had a score greater than the reference value of
110, and the average index score was 31.7. (Generally the median is presented in
the tables as a measure of central tendency. The mean was used for the health risk
index since the distribution was normal enough to make the mean and median
almost equivalent.)
Subnational variation in health risk
— Regional variation
According to distributions compiled for households in each region, there
were several distinct differences reflected by the health risk composite., Most
prominent was the finding that households in the North Central and West tended to
have water supplies with higher levels of potential health risk. The average score
on the health risk index was 35.6 for the North Central and 35.2 for the West. By
comparison, the averages for households in the South and Northeast were 30.2 and
-------
VI - 30
Figure VI-1
Distribution of Values on the Health Risk Index
for Rural US Households
I5-,
Lowest value:
Highest value:
Median:
% above referent (110):
17
60
31.7
0.0
Number of households: 21,97^,000
10-
5-
16
20
30
40
Index Value
50
60
-------
VI - 31
26.6, respectively. As ~i;;ure VI-2 suggests, these variations were gentia iy
consistent with other aspects of the distributions, including the range of scores and
the proportions of households below a certain value. For example, while aoou* 73
percent of the housenoias in the Northeast had scores of 33 or leso, only
approximately 23 percen: of the households in the 'Vest had comparable scores,
likewise, although the maximum index value ot 60 was detected for households in
the West and North Central, the respective maximums for households in the South
and Northeast were 52 and 47. Combined, these results indicated that the supplies
of households in the Northeast were associated with the lowest levels of health risk
relative to the supplies of households in the other three regions.
— SMSA/NonSMSA variation
There were no appreciable differences between SMS A and nonSMSA
households on the health risk composite. The magnitude of the variation that was
observed is perhaps best summarized by the mean score obtained for each grouping,
which was 32.7 for SMSA households and 31.2 for nonSMSA households.
— Size-of-place variation
Differences according to size of place were also negligible. The means
varied from 33.2 for households in large communities to 31.6 for households in
small communities or other rural areas.
— Size-of-system variation
Pronounced differences on the health risk composite were observed be-
tween households according to the size of system from which they obtained water.
Households on intermediate systems on the average tended to be exposed to a
higher level of potential health risk than households on community or individual
systems. Figure VI-3, which provides the distributions of values on the health risk
-------
Reproduced from
best available copy
-------
VI - 33
Figure VI-3
Size-of-System Distributions of Values on the Health Risk Index
Individual ^vs+ems
J
O-J , , p.
30 40
n
-------
VI - 34
ird;x icr households on community, intermediate, and individual systems. also
indicates that households on individual systems generally had lower scores on the
health r;3k composite. This pattern can be summoned by the means, .vhich were
33.6, intermediate; 32.6, community; and 29.4, individual.
Aesthetic/economic index
The aesthetic/economic index included the five water quality indicators
with secondary MCLs. Since these constituents had aesthetic or economic
implications but did not affect human health, they could be distinguished qualita-
tively from the substances in the other specialized index. This difference was
reflected in the legal status of the secondary MCLs which, unlike the primary
MCLs, denoted recommended concentrations rather than levels that were legally
enforceable. The unique status of the secondary MCLs was the principal reason for
constructing the aesthetic/economic index with a different set of principles from
those used in constructing the health risk index, although both indices were
represented by linear combinations of derived scores.
In contrast to the method described previously for obtaining a set of
relative measures, conventional standard scores were derived for the constituents
in the aesthetic/economic index. To compute these scores, which were referred to
as "z scores," the mean value of a constituent was subtracted from the raw value
and the difference was divided by the distribution's standard deviation. The mean
and standard deviation for each constituent are provided in Table VI-4. With this
transformation, raw values that were equal to the mean received a standard score
of zero, while those above the mean were positive and those below were negative.
After completing this operation, a second transformation was used to convert the
original scores to a more convenient form. Although others are available, the one
performed on these scores involved adding a constant to each one.
-------
VI - 35
The formal expression for this method of deriving standard scores was
Zi = xi " x
5D
where
Z; = the standard score for the itl1 household
Xj = the value on the constituent for the ltr[
household
X = the mean value of the constituent
SD = the standard deviation of the constituent.
Structure of the aesthetic/economic index
The aesthetic/economic index was constructed by additively combining the
translated linearly derived or standard scores across the five constituents. This
index had the general expression
5
AEi-= I Tjj
j = 1
where
AEj = the value or score for the i^ household on
the aesthetic/economic index
Tjj = the translated standard score for the it^1
household on the jt^ constituent T-. = 2.. +
constant ^ ^
3 = five, the number of constituents that were
incorporated into the index.
Even though a different method was used to produce the derived scores
that composed the aesthetic/economic index, its constituents were also self-
weighting. In contrast to the health risk index, however, the relative contribution
-------
VI - 36
of the indicators was determined by their original distributional properties rather
than by the stringency of their MCLs.
The aesthetic/economic index permitted the same comparisons as the other
specialized composite, with one modification. Instead of co nparir.g household
water quality on the basis of a reference value derived from federal regulations,
tha . rferc-ce value for tne aesthetic/economic index the ievei of water
quality at the average rural household. Again, the index was ordinal and simply
ranked households in terms of the aesthetic or economic effects of their water
supplies. Given the direction of the composite, these effects were more pro-
nounced at households with higher index values.
National estimates of aesthetic/economic effects
Nationally, the values on the aesthetic/economic index varied from a low
of 0 to a high of 49. Despite this large range, the vast majority of rural households
tended to cluster at a few specific values, as the information in Figure VI-4
suggests. For example, about 76 percent of the households had scores of 0, 1, or 2,
but only approximately 2 percent had values in excess of 10. This asymmetry also
was reflected in the median score which, at 1.4, was much closer to the
distribution's minimum. Because the median is a preferred measure of central
tendency for distributions that are excessively nonnormal, it is reported rather
than the mean or average. About 24 percent of the households had index scores
greater than 2, which is also the index reference value.
Subnational variation in aesthetic/economic effects
— Regional variation
The regional differences on the aesthetic/economic index were consistent
with the pattern of regional variation on the health risk composite. More
specifically, households in the North Central tended to have the highest index
-------
VI - 37
Figure VI-4
Distribution of Values on the Aesthetic/Economic Index
for Rural US Households
Lowest value:
Highest value:
Median:
% above referent (2):
Number of households:
0
49
1.4
23.8
21,974,000
n—i—i—i—i—i—i—i i M I I I I I—i—i—m—i—i—r~i—1
0 10 20 30 40 50
Index Value
-------
VI - 38
values on the average, while the !ojv#-st scores were associated with households in
the Northeast. The respective medians for the four regions were 3.9 for Households
in the Northeast, 1.1 for househo'do in the South, 1.3 for households in the West,
and 2.2 for households in the Nort., Central. Likewise, as Figure VI-o indicates, the
North Central also had the largest proportion of households with scores that
exceeded the index reference value. These results suggested that rural households
in the North Central had supplies which produced water that was least suitable
from an economic or aesthetic standpoint.
— SMSA/nonSMSA variation
There were virtually no differences on the aesthetic/economic index with
respect to the SMSA/nonSMSA categorization. The medians for both groupings of
rural households were 1.4, and there was about a 1 percent variation in the
proportions of households with scores greater than the reference value. These
percentages were 23A for SMSA households and 24.6 for nonSMSA households.
— Size-of-place variation
Differences on the aesthetic/economic index according to size of place
were only slightly more pronounced. With a median of 1.8, households in small
rural communities tended to have slightly higher values than households in large
communities or other rural areas. The medians ior the other two categories of
households were 1.4 and 1.3, respectively. Compared to the other two categories,
a larger proportion of households in small rural communities also had scores that
exceeded the reference value.
— Size-of-system variation
Except for some minor inconsistencies, there were no appreciable differ-
ences between rural households on the aesthetic/economic index when grouped by
-------
VI - 39
Figure VI-5
Regional Distributions of Values on the Aesthetic/Economic Index
j:
Northeast
North Central
J !
"i
30-
20-
10-
Lowest -'diur:
Honest value:
Median: 0.1
% aDove relerent (2): 7.5
Number of housenolds: 3.o93,CQ0
H-fin rr
10
1—i—n—i—ra\ i—i—i—i i i—r
20 30 40
index Volue
50
40
30-
20-
10-
Lowest -ai'je:
'-honest vaiue:
Median:
•"b above referent (2):
Number of households:
30
2.2
<*0. I
6,213,000
Q i i i i
i t i il iiii n i i
10 20 30
l 1 I I I I
40 50
70 n
South
West
60 -
40-
20 -
10-
lowest value:
Highest value:
Median:
% aoove referent (2):
Number of households:
0
*9
16.1
9,291,300
TT
20
index Value
! ! i
30
i—r—r
40
-m
50
50-,
40-
30-
20-
10-
Lowe-it value:
0
Hignesr value:
19
Median:
1.8
•*b above referent (2):
Number of households:
2,777,000
[ i i r
10
i i i i—m—n—i—i—i—i—m
30 40 50
20
Index Value
-------
VI - w
the size of the system from which they obtained water. The medians, ::r example,
were IA for households either on individual or intermediate systems, and 1.3 for
households on community systems. Similarly, the proportions of households -.vlth
scores greater than the reference value varied only by about 5 percent. The
respective percentages were 21.5 for households on community systems, 26.1 for
households on intermediate systems, and 26.0 for households on individual systems.
General water quality index
The general water quality index developed for the NSA, which was designed
to reflect health, economic, and aesthetic aspects, consisted of all 23 constitu-
ents, including those substances with no MCLs. This index, similar to the
aesthetic/economic index, was formulated by adding together a set of linearly
derived translated standard scores. These scores were derived for each constituent
in the index using the linear transformations described previously. Again, the
standard scores were relative measures that were computed from the means and
standard deviations of the relevant water quality indicators.
Structure of the general water quality index
The formal equation for the general water quality composite was
23
GWQi = I Tij
j = 1
where
GWQj = the value or score for the i^1 household on
the general water quality index
Ty = the translated standard score for the i*h
household on the i*'"1 constituent T.. = Z.. +
* - i] M
constant ; '
3
= 23, the number of constituents that were
incorporated into the composite.
-------
VI - 41
Since it was constructed with a similar set of principles, the general water
quality index was similar in many respects to the aesthetic/economic index. In
particular, the constituents were self-weighting and their relative contributions
were contingent upon t;,eir distributional attributes. Also, the index was ordinal
and its reference point was the level of water quality at the average rural
household. Finally, the index was deigned so that higher values signified lower
levels of water quality.
National estimates of general water quality
For the nation, the distribution of values on the general water quality index
again displayed a tendency for a disproportionately large number of households to
have a limited range of values. As can be observed from Figure VI-6, which
presents the distribution of values for all rural households, about 73 percent of
households had scores ranging from 1 through 10. Also, about- 10 percent had
values in excess of 20, and fewer than 1 percent had a score of 0. While the
maximum value was 36, the average score was 9, which was also the reference
value. Finally, about 68 percent of the households had scores that were below the
reference value, and the median score was 7.1.
Subnational variation in general water quality
— Regional variation
The distributions of values on the general water quality index in the four
regions of the US reflected a pattern of variation that was similar to those
observed for the other two indices. As indicated in Figure V1-7, households in the
North Central had the highest median as well as the largest proportion of
households with scores above the index reference value. Once again, households in
the Northeast had scores that tended to be lower than the reference value; also,
the Northeast had the lowest median. Additionally, households in the West had the
-------
VI - 42
Figure VI-6
Distribution of Values on the General Water Quality Index
for Rural LIS Households
2
o
o
X
20-i
15-
10-
5-
Lowest value:
Highest value:
Median:
% above referent (9):
Number of households:
0
36
7.1
32.2
21,97
-------
.->.rr ...:
VI - ^3
Figure VI-7
Regional Distributions of Values on the General Water Quality Index
Northeast
i i
50 -
. n.,nnn
0 ;0 20 30 *0
*>«•> VOtrt
South
North Central
pi
•• r~i^»rf»t Ch W . "
>1 -...'il.'.OO
wa«i vaiu*
West
-------
VI - w
second highest median, while households in the South had the thirc .vghest. The
medians for each region were 9.8 for the North Central, 9.7 for the '-'est, 5.1 for
the SoJth, and 4.6 for the Northeast. These results suggested that er supplies
in t..c North Central and West produced water of lesser quality eiative to the
supplies of households in the other regions.
— SMSA/nonSMSA variation
SMSA/nonSMSA differences were not appreciable except for the propor-
tions of households with scores that exceeded the index reference value. Among
nonSMSA households, 35.5 percent had scores over the reference value, compared
to 25.3 percent among SMSA households. By comparison, the medians for the two
groupings of households were 7.0 (nonSMSA) and 7.2 (SMSA).
— Size-of-place variation
Variations in general water quality according to size of place were only
slightly more pronounced. The median values were 8.1 for households in small rural
communities, 7.0 for households in other rural areas, and 6.2 for households in large
'ural communities. In contrast to these median differences, households in large
•ural communities were the most likely to have scores over the index reference
/alue of 9. The proportions of households with scores greater than the reference
/alue were 41.2 percent in large rural communities, 33.5 percent in small rural
rommunities, and 30.9 percent in other rural areas.
— Size-of-system variation
As suggested by the information in Figure VI-8, households on intermediate
ystems tended to have higher scores on the general water quality index than
louseholds on community systems or individual systems. This pattern was
eflected in the medians compiled for each of the three groupings as well as in the
-------
VI - 45
Figure VI-8
Size-of-System Distributions of Values on the General Water Quality Index
Individual Systems
vo»j«
Intermediate Systems
J ,oH
Ln
II
Community Systems
5°1
Reproduced from
best available copy
-------
VI - It6
proportions of households with scores in excess of the reference valje. Yedians
were 10.1 for households on intermediate systems, 7.0 for households on inc!iv:djai
systems, and 6.7 for households on community systems. Likewise, the pe'csnta^'js
of households with scores greater than 9 were 53.5, 31.1, ana 2S.S, respectively.
These differences, which indicated that households on intermediate systems had
supplies that produced water of lesser overall quality, were consistent with the
results on the two specialized water quality indices.
Summary, Indices of physical measures
In this section of Chapter VI, water quality was measured by a set of
indices developed from laboratory data on 23 physical constituents. Each of the
three indices encompassed information about a particular combination of sub-
stances and was structured to represent a different aspect of water quality. The
first index, which consisted exclusively of constituents with primary MCLs, was
designed to reflect the potential health risks to which rural households may have
been exposed from their water supplies. The second index, by contrast, incor-
porated only substances with secondary MCLs and summarized water quality in
terms of possible aesthetic and economic effects. The entire complement of
substances, including those without MCLs, composed the third index, which
provided a generalized expression of water quality.
Although two of the three water quality indices were mutually exclusive,
all of them displayed similar patterns of differences between the household
groupings. Households in the North Central were inclined to have higher values on
all three indices, while" households in the Northeast typically had lower scores.
These results suggested that households in the North Central were more likely to
be exposed to a health risk and were more subject to aesthetic and economic water
quality effects than households in other regions. Another implication of these
findings was that supplies of households in the North Central and West provided
-------
VI - 47
water that tended to be of lower overall quality than the national average. Man:ed
variations among households also were detected on two of the three composites
according to the size of system serving the household. Ones again, although thore
were some discrepancies, the patterns were largely parallel. r-!ou5.:ho'd.> on
intermediate systems had higher values on the health risk and general vvater quality
indices, winie the scores of on individual systems or community systems
were lower. This evidence strongly indicated that households on intermediate
systems received water of lower quality than households served by individual or
community systems. Except for differences on the general water quality index,
there were only negligible variations on the indices in the SMSA/nonSMSA and size-
of-place groupings. Comparisons of scores on the general water quality index
indicated that households in small rural communities were receiving lower quality
water than households in large rural communities or other rural areas.
INDEX OF PERCEIVED MEASURES
Another component of the NSA's effort to develop summary measures of
water quality involved constructing a composite that synthesized information on
rural residents' perceptions of the water provided by their major supplies. In
particular, the aesthetic properties inquired about at the rural households in the
NSA sample included odor, taste, clarity, color, and sediment. Information was
also obtained on perceived temperature, but it did not vary enough to make any
noticeable contribution to a summary index. As individual aspects of perceived
water quality, the data corresponding to these indicators were presented in
Chapter V. Once again, however,, as with the physical constituents, it was
necessary to formulate a composite from these attributes, rather than analyzing
them separately. This section of Chapter VI specifies the methods and procedures
that were used to incorporate these five variables into a composite index of
perceived water quality.
-------
VI - 48
Besides the constraint imposed by the particular combination of :nci;ators
that were selected for consideration in the NSA, the only ether restriction on the
indexing process was the structure of the data. For example, one index o."
perceived measures was proposed, as opposed to an entire set, because information
was compiled for each attribute at all households in the NSA sample. In addition to
this symmetry, the information on each indicator consisted of a combination of
responses which reflected both magnitude and persistence. .Since this was true for
all five perceived indicators and because they were already expressed in ordinal
units, each indicator had approximately the same form. Therefore, the principal
task in developing the composite was to prescribe some technique for combining
the data across the indicators.
Structure of the perceived water quality index
After extensive deliberations, NSA researchers determined that perceived
water quality could be represented best by a simple additive measurement model.
The principal motivation behind choosing this device was the lack of any indication
that perceived water quality was more complex than a simple linear function of the
intensity and duration of taste, odor, color, sediment, and clarity. Formally, this
composite was expressed as
PWQi = Tj + Oj + Cj + Si + Cli
where
PWQi = the value or score on the perceived water
quality index for the i*h household
Ti = the value on perceived taste for the i^
household
Oj = the value on perceived odor for the i^
household
-------
VI - 49
Cj = the value on perceived color for the i:'n
housenoid
Sj = the value on perceived sediment for the i*-1"1
household
Clj = the value on perceived clarity for the itn
household.
Given tue minimum and maximum of each of the five components—0 and
5—the preceding formula theoretically could have produced a vector of scores
ranging from 0 to 25. The measure was ordinal in that it simply ranked households
in relation to each other, and the magnitude of the differences had no significance.
Households with higher index values were judged to have supplies that provided
water of lesser perceived quality relative to households that were assigned lower
values.
National estimates of perceived water quality
According to data compiled on the national level, the values on the
perceived water quality index were distributed somewhat unevenly. Figure VI-9,
which provides -the distribution of scores on the perceived water quality index,
indicates a strong tendency toward low index values. For example, about 71
percent of households had scores of 4 or less, while approximately 20 percent had
scores from 5 through 9. Thus, about 91 percent of all rural households had scores
of less than 10 on the perceived water quality index, compared to a theoretical
maximum of 25. The median score was 2.3.
Subnational variation in perceived water quality
Regional variation
Data on regional variation in the perceived water quality index, which are
presented in Figure VI-10, displayed a very uniform pattern. The distributions of
index values were very similar for households in the South, Northeast, and North
-------
VI - 50
Figure VI-9
Distribution of Values on the Perceived Water Quality Index
for Rural US Households
40—1
35
30-
Lowest value:
Highest value:
Median:
Number of households:
0
21
2.3
21,97^,000
15-
10-
5-
0
I I I
I I I I
i—i—r
15
=h-n
20
10
Index Value
-------
VI - 51
Figure VI-10
Regional Distributions of Values on the Perceived Water Quality Index
Northeast North Central
J I
10 -
South
West
40-,
35 -
30-
Lowest value:
Highest value:
Median:
Number of nousenoids:
21
2.2
S,2*l,0C0
Lowest v^iue:
Higneit value:
Median:
Nuftiber Jl households:
20
2.7
2,777.000
~i—I—i—i—r
5
n
M S—TT-!
T 1—PHl
.20
Reproduced from
best available copy
-------
VI - 52
Central, while those for households in the West were somewhat dltf On the
average, households in the South 'ended to have somewhat lower scores than
households in the Northeast and North Central, and households in the ¦•vest tended
to have higher scores. In summary, households in the South appa-e.-uiy perceived
the quality of their water supplies to be slightly better than households in other
regions did; households in the West pet reived the quality ot their water supplies to
be lower than did households m other regions.
SMSA/nonSMSA variation
The distributions of values on the perceived water quality index were also
similar for SMSA and nonSMSA households. Both SMSA and nonSMSA households
had about the same proportions of households in each interval, and both had median
index values which were approximately the same. According to these distributional
properties, there were no appreciable differences between SMSA and nonSMSA
households on the perceived water quality composite.
Size-of-place variation
In comparison to the preceding results, the size-of-place grouping showed
more discernible variations. For example, households in other rural areas per-
ceived the quality of their water supplies to be better than did households in small
and large rural communities. This pattern was reflected in the entire distribution
of values for each set of households as well as in the median index scores. The
proportion of households in other rural areas with an index value of 4 or less was
73.0 percent, which was substantially higher than the proportion of households in
either small or large communities (approximately 65 percent and 64 percent,
respectively). Additionally, the medians for large and small communities were 2:6
and 2.7, both of which were significantly larger than the median for households
located in other rural areas, which was 2.2.
Reproduced from
best available copy
-------
VI - 53
Size-of-system variation
While there were at least detectable differences between rural households
when they were grouped according to region, SMSA/nonS-MSA, and size of place,
none of these variations was as pronoun-®d as those in the size-oi-system
comparison. Figure Vi-u, which provides the distributions of scores on the
perceived wa*or qua!:./ :::de.\ for households on individual, intermediate, sr.d
community water systems, suggests that the values for households using individual
systems tended to be appreciably lower than those of households served by
intermediate or community systems. Likewise, index values for households
supplied by community systems were substantially higher. Only about 67 percent
of the rural households on community systems had scores of 4 or less, and the
median value for this household group was 2.8. In contrast, however, 76.0 percent
of the households using individual systems had scores of 4 or less, while the median
value for this group of rural households was only 1.6. Finally, the proportion of
rural households served by intermediate systems which had an index value of 4 or
less was around 72 percent, while the median score was 2.3. The direction and
magnitude of these differences showed a tendency for rural households on
individual systems to perceive that the quality of their water supplies was
considerably better than households served by intermediate or community systems
perceived their water supplies to be.
AVAILABILITY COMPOSITES
Availability in the NSA was defined as the ability of a supply to provide a
sufficient quantity of water on a continuous basis. More specifically, the notion of
supply availability was represented by two dimensions, reliability and accessibility.
Reliability was defined with respect to supply interruptions and breakdowns, while
accessibility was considered to reflect the ease with which water could be obtained
from a supply. The data compiled for these indicators as individual aspects of
-------
VI - 54
Figure VI-11
Size-of-System Distributions of Values on the Perceived Water Quality index
4'"^ Individual Systems
U
5 i0 -5 20
noti value
h Intermediate Systems
Community Systems
:nd«* vatut
Reproduced from
best available copy
-------
VI - 55
availaoiiity are presented in Chapter V. Consequently, in this section of Chapter
VI, our concern is with the principles and procedures that were employed to
synthesize the information on reliability and accessibility into a corrcspor.di'i'; set
of summary measures.
INDEX OF RELIABILITY
The original data on reliability whicn could theoretically be incorporated
into an index consisted of reported frequencies for two types of breakdowns, minor
and severe. Minor breakdowns were supply malfunctions of six hours or less in
duration, while severe breakdowns were those that persisted longer than six hours.
Supply breakdowns were tabulated for the year prior to the NSA .'interview.
Therefore, the reliability index was derived from these two indicators, the number
of severe and minor supply breakdowns that occurred in the year before the NSA.
The index was designed to reflect both the frequency of breakdowns and
their length. This was accomplished by arbitrarily weighting severe breakdowns so
that they would make a greater contribution to the index than the same number of
minor breakdowns. While a minor breakdown received a weight of one, all major
breakdowns were weighted by a factor of two. Therefore, a household that
reported three severe supply breakdowns would have received a score on the
reliability index that was twice as great as the value assigned to a household that
reported the same number of minor breakdowns. This weighting technique was
predicated on the assumption that a single breakdown which lasted for eight hours
would have approximately the same consequences for households as two break-
downs which were of four hours' duration. Without the application of some type of
weighting, only the frequency of breakdowns would have been permitted to
influence the index.
-------
VI - 56
Structure of the reliability index
The reliability composite that was developed simply consisted of an
arbitrarily weighted combination of variables that conveyed information on the
incidence of minor and severe water supply breakdowns. Furthermore, :he?e
variables were synthesized by incorporating them into an additive formula The
formal expression for the reliability index was
R-i = + 2 (Si)
where
Rj = _ the value or score on the reliability index
for the ith household
Mi = the frequency of minor water supply break-
downs for the ith household
Sj = the frequency of severe water supply break-
downs for the ith household.
The vector of scores that was generated by applying this measurement
procedure theoretically could range from 0 to a value that was constrained by the
maximum number of breakdowns- Also, as with the other composites that were
constructed, the reliability index had ordinal properties. Therefore, it could be
employed to rank households in comparison to each other, although no particular
meaning could be ascribed to the size of the differences. Households with larger
values on the index were judged to have supplies that were less reliable than those
that were assigned smaller, values.
National estimates of reliability
According to Figure VI-12, which presents the distribution of values on the
reliability index, rural water supplies tended to be very reliable. Approximately 75
percent of all rural households had a value of 0 on the reliability composite, while
approximately 22 percent had scores of 1 through 4. Only about 1 percent of all
-------
VI - 37
Figure VI-12
Dir-tribution of Values on the Reliability Index
for Rural US Households
70-
£ 10-
Lowest value:
Highest value:
Median:
Number of households:
0
52
0.2
21,971,000
5-
I L
0
10
Index Value
-------
VI - 58
rural household across the nation had a score of 10 or greater-, a-,d the maximum
value computed for the index was 52. The median value of the index vas 0.2.
Subnational v:; nation in reliability
Regional variation
However small, there was some slight variation in the reliability index
according to the region in which the supply was located. Generally, as the
distributions in Figure VI-13 suggest, supplies of households in the South were less
reliable than supplies in other regions. Only about 71 percent of the households in
the South had a reliability score of 0, compared to about 80 percent in the West,
approximately 79 percent in the Northeast, and roughly 75 percent in the North
Central. Likewise, the median scores ranged from a high of 0.2 for households in
the South to a low of 0.1 for households in the West. Also, a much larger
proportion of households in the South (5.3 percent) had index values in excess of 5.
SMSA/nonSMSA variation
Variations in the reliability index were also small when SMSA and non-
SMSA households were compared. Medians were the same, and the exact same
proportions of households had values of ^ or less on the index. The major
differences between SMSA and nonSMSA households on the index were in the
proportions with a score of 0 (about 78 percent of SMSA households and roughly 73
percent of nonSMSA households), and in the proportions that had scores from 1
through 4 (approximately 18 percent of SMSA households and about 23 percent of
nonSMSA households).
Size-of-place variation
Compared to the preceding differences, the variations according to size of
place were somewhat more substantial. The largest contrasts were between
-------
VI - 59
Figure VI—13
Regional Distributions of Values on the Reliability Index
Northeast
North Central
sJ
i LT
U
4=L
3 i0
IA4«« VQIw*
South
West
'w>it#no
-------
VI - 60
households in small rural communities and those in either large rural communities
or other rural areas. For example, only about 69 percent of the households In small
rural communities has _n index value of 0, compared to roughly 73 percent and
76.0 percent of the households in other rural areas and iarge rural communities,
respectively. A much larger proportion of households in smaii .• urai communities;
on the other hand, had an index value ranging from 1 through k. Even thougn the
medians were uniformly 0.2, these differences suggested that the supplies of
households located in small rural communities were slightly less reliable than the
supplies of households in other rural areas or large rural communities.
Size-of-system variation
A distinct pattern of variation in index scores also emerged for households
using water systems of different sizes. The data indicated that the supplies of
households on intermediate systems generally were not as reliable as those which
provided water to individual-system or community-system households. Only about
70 percent of the households on intermediate systems received an index value of 0,
while approximately 74 percent of the households on community systems and
roughly 76 percent of the households on individual systems had the same composite
score. However, the median score for households on intermediate systems (0.2)
was the same as the medians for other households. Although households on
community systems showed the same median value, a much larger proportion had a
reliability index value in excess of 5. More specifically, only 2A percent of the
households on either individual systems or intermediate systems had scores of 5 or
more, while 4.9 percent of the households on community systems were assigned a
value in this range.
Reproduced from
best available copy
-------
VI - 61
INDEX OF ACCESSIBILITY
Data on supply accessibility, which consisted of information about recorded
pressure and the distance between the extraction point and the household, were too
diverse to be incorporated directly into a composite. The former indicator, for
example, was measured in pounds per square inch, while the latter variable was
expressed in meters. Another incompatibility between the two indicators was that
they were inversely related to each other in terms of supply accessibility. In other
words, supplies with higher pressure were taken to be more accessible than supplies
with lower pressure, but supplies which conveyed water over greater distances
were considered less accessible than those extracting water from a point close to
the household. An additional constraint was imposed by the fact that households on
community supplies were not even administered the question on distance because it
was originally presumed that the question had relevance only for wells, springs,
cisterns, and surface water supplies. Consequently, a measurement procedure had
to be selected which would facilitate the resolution of these particular difficulties.
First, the pressure variable was inverted by simply subtracting each value
in the vector from the maximum value. After this operation was completed, the
order of the households was consistent for both pressure and distance. Next, the
information on distance was expanded by assigning a value of zero meters to all
households that were supplied by community systems. Implicit in this decision was
an assumption that the connection to a community system was the equivalent of
having a water source on the premises. Both procedures were accomplished as
preliminary steps to combining the accessibility indicators, which also had to be
converted into different units. To accomplish this, the derived score technique was
employed to produce standard scores for both variables. These standard scores, to
reiterate, were generated by subtracting a variable's mean from each of its values,
and dividing the difference by the standard deviation. The final operation before
actually developing the composite involved translating the standard scores by
-------
VI - 62
adding the appropriate constant to each one. Since the formal expressions for
these procedures were presented in regard to the water quality composites, they
will not be provided here.
Structure of the accessibility index
The accessibility Ij-.uc* was imposed of an additive combination of the
translated standard scores that were derived from the variables for distance and
pressure. Expressed more formally, the measurement model for this index was
Aj = TZRj + TZDi
where
Aj = the value or score on the accessibility index
for the i*h household
TZPj = the translated standard score on pressure
for the it^ household •
TZDi = the translated standard score on distance
for the fth household.
This procedure generated a set of values that ranged from 0 to 47. The
index was ordered such that households which received greater values were
considered to have less accessible water supplies than households with lesser
values. Also, since the composite was ordinal and simply provided a relative
ranking of households in terms of supply accessibility, the magnitude of any
differences had no particular significance. The only other limitation on the
composite was that it did not include information on households which did not have
pressurized water supplies, such as many households that obtained water by
purchasing or hauling, and some households with wells. Since these supplies were
perhaps the most inaccessible, the composite tended to slightly underestimate
accessibility in the aggregate. Because it was distributed over the entire sample of
households, this bias was judged to be negligible.
-------
VI - 63
National estimates of accessibility
Figure VI-14 presents the distribution of values on the accessibility
composite for all rural households except those without pressjrized supplies. As
seen in the iig':re, the scores on the index tended to cluster in a fe-v specific
intervals of the composite. About 55 percent of the households had scores of S or
9, and about 40 percent rn-i of 6 or 7. Thus, about 95 percent of the
households had scores from 6 through 9. Extreme values, both high, ana low,
occurred very infrequently. The median score compiled for the index was 7.6,
while the reference value was 7.5. Again, since the accessibility index was
formulated from standard scores, the reference value simply designated the level
of supply accessibility at the average rural household. However,- the referent was
not as useful in the context of this section's results because the proportion
exceeding it could not be determined with a sufficient degree of accuracy.
Therefore, the referent is not employed in the following subnational comparisons.
Subnational variation in accessibility
Regional variation
Accessibility index scores showed substantial disparities from one region to
another, as shown in Figure VI-15. Most prominent among these were generally
higher values for households in the North Central and Northeast, and lower values
for households in the South and West. For example, around 24 percent of the
households in the West had scores of 6 or less, compared to about 7 percent of
North Central households. Similarly, only about 36 percent of the households in the
West received accessibility scores of 8 or 9, compared to approximately 69 percent
of households in the North Central. These differences were also manifested in the
medians for the regions, which ranged from a maximum of 7.9 for households in the
North Central to a minimum of 7.2 for households in the West. Generally, then,
Reproduced from
best available copy
-------
VI - 64
Figure VI-14
Distribution of Values on the Accessibility Index
for Rural US Households
40-
35-
30-
*7
15-
10-
5-
0
0
Lowest value:
Highest value:
Median:
Number of households:
0
47
7.6
21,290,000
¦n r~T
n i i—i—i—r>—
15
20
Index Value
-------
VI - 65
Figure VI-15
Regiona; DistriLut ons of Values on the Accessibility Index
Northeast North Central
n
u
LO
i i i i i H i i n i
5 10 15
index Value
o4_
r1. ¦,. h
South
West
40-j
s
u
ru
i
n
Lo»e>t .aluc:
-litest .alue:
\leaian:
Numoer Jl nouSe-nola^;
-I
Hignest w.ue: ^
Mtraia:i: 7.2
\unOcr 'louse'ioiCS; Z.^V.jZQ
S 1
5 1
a ,o J
Q- .0 -
~th—i—rn—r
0
20
0-R-
1 1"
!0
i n i i i i i
:5 2 0
Reproduced from
best available copy
-------
VI - 66
the supplies cl households in the West were nost accessible, and supplies In tho
North Central were least accessible.
SMSA/nonS.'/^A variation
In contrast to the preceding dissimilarities, there were only trivial varia-
tions between SMS A and nonSMSA households on the accessibility index. The
distributions for the two groupings had similar configurations and the medians were
approximately equal. In short, there was no conclusive evidence that the supplies
of SMSA households were more or less accessible than those of nonSMSA
households.
Slze-of-place variation
From the information on differences in supply accessibility according to
size of place, it was apparent that the supplies of households in large rural
communities were more accessible than the supplies of households located in small
rural communities or other rural areas. This pattern of variation was manifested in
the proportions of households-at various intervals of the composite as well as in the
medians that were compiled. For example, about 12 percent of the households in
other rural areas had an accessibility score of 6 or less, compared to about 14
percent of the households in small rural communities and approximately 23 percent
of the households in large rural communities. Conversely, about 38 percent of the
households in large rural communities had a score of 8 or more, which differed
substantially from the roughly 47 percent of households in small rural communities
and 59 percent of households in other rural areas with the same scores. Since they
were sensitive to percentage fluctuations of this magnitude, the medians of 7.2 for
households in large rural communities, 7A for households in small rural communi-
ties, and 7.7 for households in other rural areas reflected a similar progression.
-------
VI - 67
Size-of-system variation
As suggested by the information in Figure VI-i6, the accessibility of
supplies varied even more noticeably by the size of system which provided water to
the household. Households on community systems tended to c-. assigned lower
values on the accessibility composite, as was indicated by the large proportion
(about 70 percent) of those households that received scores of 7 or less. In
contrast, households on individual systems and those on intermediate systems were
considerably more likely to be assigned higher values on the accessibility compos-
ite. For example, approximately 88 percent of the households on intermediate
systems had a score of 8 or more, compared to about 83 percent of households on
individual systems. The medians, which were 7.5 for households orv community
systems, 8.0 for households on individual systems, and 8.3 for households on
intermediate systems, also signified that the supplies of households on intermediate
systems were the least accessible. Similarly, households on community systems
were substantially more accessible.
Reproduced from
best available copy
-------
VI - 68
Figure VI— 16
Size-of-S>stem Distributions of Values on the Accessibility Index
Individual Systems
m
^ae» vc'ue
Intermediate Systems
50q
H
1
45 -
K
33-
2
©
£
J 30-
X
o
5 io -
n
XI
r
j
NumtHfr oJ rxxjiefioldj: '.»03,u00
-f=H-
J
J
4C-1
i
30i
i
25-!
J«>J
Community Systems
vjiue: ^
Higrt«jc vjiue: :0
\ie<3un: .'.5
Numoer 0! .^Ou5«t'0)0S: IZ . 17r). COCi
.M
0 -f7~
10 15
mo«x votut
Reproduced from
best available copy
-------
VI - 69
REFERENCES
Ott, Wayne R. Environmental Indices. Ann Arbor: Ann Arbor Science, i97S.
ZHorton, Robert K. "An Index-Number System for Rating Water Quality."
Journal, Water Pollution Control Federation 37 (March 1965):300-306.
^Brown, Robert M.; McClelland, .Nina I.; Deininger, Rolf A.; and Tozer, Ronald G.
"A Water Quality Index—Do We Dare?" Water and Sewage Works 117 (October
1970):339-343.
4
Prati, L.; Pavanello, R.; and Pesarin, F. "Assessment of Surface Water Quality
by a Single Index of Pollution." Water Research 5 (1971):741-751.
McDuffie, Bruce and Haney, Jonathon T. "A Proposed River Pollution Index."
Paper presented at the spring 1973 meeting of the American Chemical Society,
Division of Water, Air, and Waste Chemistry, New York (April 1973).
^Dinius, S. H. "Social Accounting System for Evaluating Water Resources."
Water Resources Research 8(5)(October 1972): 1 159-1177.
O'Connor, Michael F. "The Application of Multi-Attribute Scaling Procedures to
the Development of Indices of Water Quality." Ph.D. Dissertation, University of
Michigan, University Microfilms No. 72-29, 161 (1972).
g
Deininger, Rolf A. and Landwehr, Jarate M. "A Water Quality Index for Public
Water Supplies." Unpublished report, Department of Environmental and Indus-
trial Health, School of Public Health, University of Michigan, Ann Arbor,
Michigan (July 1971).
9
Walski, Thomas M. and Parker, Frank L. "Consumers Water Quality Index."
Journal of the Environmental Engineering Division ASCE (June 1974):.593-611.
^Stoner, Jerry D. "Water Quality Indices for Specific Water Uses." US Geological
Survey, Reston, Virginia, Circular No. 770 (1978).
Reproduced from
best available copy
-------
VI - 70
References continued
1 Nemerow, Nelson !. and Sumitomo, Hisashi. "'3ene*i :s "Varer Qua!'.:/
Enhancement." Sy. reuse University, Syracuse. Ne .v V-iri 10
DAJ, prepared for the US Environmental Protection Agency (December 1970).
I 2
Truett, J. 3.; Johnson, A.C.; Rows, \v. D.; Feigner, K. 0.; ar,c .Manning, L. J.
"Development of Water Quality Management Indices." '>Va:er Resources bulletin
1 l(3)(June 1?75)-A36-MS.
^Greely, R. S.; Johnson, A. J.; Rowe, W. D.; and Truett, J. B. "Water Quality
Indices." Report No. M72-54, MITRE Corporation, McLean, Virginia (April 1972).
1 £f
Dee, Norbert; Baker, Janet K.; Drobny, Neil L.; Duke, Kenneth M.; and
Fahringer, David C. "Environmental Evaluation System for Water Resource
Planning." Report No. PB 208 822 Battelle Columbus Laboratories, Columbus,
Ohio (January 1972).
^Inhaber, H. "Environmental Quality: Outline for a National Index for Canada."
Science 186 (November 1974):798-805.
Johanson, Edward E. and Johnson, Jaret C. "Identifying and Prioritizing
Locations for the Removal of In-Place Pollutants." Prepared for the US
Environmental Protection Agency, Washington, DC, Contract No. 68-01-2920
(May 1976).
Zoeteman, B. C. "The Potential Pollution Index as a Tool for River Quality
Management." Technical Paper No. 6, Government Institute for Water Supply,
"World Health Organization, International Reference Centre for Community
Water Supply, The Hague, The Netherlands (August 1973).
18
Harkins, Ralph D. "An Objective Water Quality Index." Journal of the Water
Pollution Control Federation 46 (March 1974): 588-591.
19
Schaeffer, David J. and Janardan, Konanur G. "Communicating Environmental
Information to the Public: A New Water Quality Index." Journal of Environ-
mental Education 8 (Summer 1977): 1S-26.
Reproduced from
best available copy
-------
VI - 71
References continued
2°Ott, 1973.
2 1
US Environmental Protection Agency. "National Interim Primary Drinking Water
Regulations." Federal Register 40: No. 248 (December 24, I975):59566-5957^.
22
US Environmental Protection Agency. "National Secondary Drinking Water
Regulations." Federal Register 42: No. 62 (March 31, 1977):i71k4-17146.
23
Calabrese, Edward J. Methodological Approaches to Deriving Environmental
and Occupational Health Standards. New York: John Wiley & Sons, 1978.
24
Allen, Martin 3., and Geldreich, Edwin E. . "Bacteriological Criteria for Ground-
water Quality." Ground Water 13 •(1975):45-52.
25
National Research Council. Drinking Water and Hea.'th'. Washington, DC:
National Academy of Sciences, 1977.
26
US Environmental Protection Agency. "National Secondary Drinking Water
Regulations."
^Calabrese, 1978.
28
US Environmental Protection Agency. "National Interim Primary Drinking
Water Regulations."
------- |