EPA-600/1-77-004
January 1977
AN ASSESSMENT OF THE CHESS SULFATE AND NITRATE DATA
during the period RETA performed the chemical analysis
by
Leo T. Heiderscheit
Statistics and Data Management Office
and
Marvin B. Hertz
Population Studies Division
Health Effects Research Laboratory
Research Triangle Park, N.C. 27711
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
HEALTH EFFECTS RESEARCH LABORATORY
RESEARCH TRIANGLE PARK, N.C. 27711
-------
DISCLAIMER
This report has been reviewed by the Health Effects Research Laboratory,
U.S. Environmental Protection Agency, and approved for publication. Mention
of trade names or commercial products does not constitute endorsement or
recommendation for use.
-------
FOREWORD
The many benefits of our modern, developing, industrial society are
accompanied by certain hazards. Careful assessment of the relative risk
of existing and new man-made environmental hazards is necessary for the
establishment of sound regulatory policy. These regulations serve to
enhance the quality of our environment in order to promote the public
health and welfare and the productive capacity of our Nation's population.
The Health Effects Research Laboratory, Research Triangle Park
conducts a coordinated environmental health research program in toxicology,
epidemiology, and clinical studies using human volunteer subjects. These
studies address problems in air pollution, non-ionizing radiation,
environmental carcinogenesis and the toxicology of pesticides as well as
other chemical pollutants. The Laboratory develops and revises air quality
criteria documents on pollutants for which national ambient air quality
standards exist or are proposed, provides the data for registration of new
pesticides or proposed suspension of those already in use, conducts research
on hazardous and toxic materials, and is preparing the health basis for
non-ionizing radiation standards. Direct support to the regulatory function
of the Agency is provided in the form of expert testimony and preparation of
affidavits as well as expert advice to the Administrator to assure the
adequacy of health care and surveillance of persons having suffered imminent
and substantial endangerment of their health.
This study addresses a concern regarding some of the air pollution
measurements collected in the CHESS air monitoring system. Severe manpower
limitations forced the Health Effects Research Laboratory to contract for
the ctremical analysis of high-volume filter strips. Some measurement
effects resulted, and this paper investigates the nature of the differences
and recommends an appropriate procedure.
H. Knelson, M.D.
>Director,
Health Effects Research Laboratory
-------
ABSTRACT
In the early 1970s certain filters from the CHESS network were collected
and sent to the Human Studies Laboratory Bioenvironmental Laboratory Branch
(BELB) for sulfate and nitrate analyses. These analyses were interrupted on
October 1, 1972 and subsequently continued under contract signed March 1973
with Rickman, Edgerley, Tomlinson, and Associates (RETA). Many of the filters
were not analyzed until RETA became fully operational in May 1973, whereupon
their measurements began to be inexplicably and consistently low.
The disparity between HSL results and RETA's findings engendered an
investigation involving reanalyses to verify the apparent disparity, and then
to determine a statistical adjustment factor to correct for anomalies. In
November 1974 Rockwell International undertook the reanalyses of these CHESS
filters.
In Rockwell's reanalyses, the difference in RETA's sulfate data showed
a mean ratio of 51%, but the.results on nitrates were inconclusive. The
effects of time and handling were not conclusively determined, altlrotit)ti"tfef't -
was an indication that nitrates decompose with timelaps-e more significantly
than sulfates. Based on the results in this report^ it was recommended that
CHESS measurements for sulfates be increased by 51% for all sites for -the
entire period RETA performed these analyses.
IV
-------
CONTENTS
Page
Foreword i i i
Abstract iv
Figures vi
Tables vii
1. Introduction . . . 1
2. Other Studies Related to the Problem ... 3
3. Rockwell's Reanalysis of CHESS Filters . . 4
4. Conclusions 8
5. Appendix 27
-------
FIGURES
Number Page
1. November 1974 - Rockwell Reanalyzed Latter 1973 Filters
for S04 24
2. November 1974 - Rockwell Reanalyzed Nine Bad Months
for S04 25
3. November 1974 - Rockwell Reanalyzed Early 1974 Filters
for SO4 26
-------
TABLES
Number Page
1. Ratio of Sulfate Quarterly Arithmetic Mean to 1973
Quarterly Arithmetic Mean — First and Second Quarters 9
2. Ratio of Quarterly Arithmetic Mean to 1973 Quarterly
Arithmetic Mean -- Third and Fourth Quarters 11
3. Ratio of Nitrate Quarterly Arithmetic Mean to 1973
Quarterly Arithmetic Mean -- First and Second Quarters 13
4. Ratio of Nitrate Quarterly Arithmetic Mean to 1973
Arithmetic Mean -- Third and Fourth Quarters 15
5. Analysis of Variance for Rockwell's Reanalysis 17
6. November 1974 - Rockwell Reanalyzed 200 RETA Filters 18
7. November 1974 - Rockwell Reanalyzed "Glassine" Filters 19
8. November 1974 - Rockwell Reanalyzed HSL's Filters 20
9. November 1974 - Rockwell Reanalyzed Nine Bad Months 21
10. November 1974 - Rockwell Reanalyzed Latter 1973 Filters 22
11. November 1974 - Rockwell Reanalyzed Early 1974 Filters 23
A-l. April 1973 - RETA Reanalyzed HSL's September 1972 Filters.. 32
A-2. June 1974 - RETA Reanalyzed RETA's October 1972 filters 33
A-3. June 1974 - RETA Reanalyzed HSL's September 1972 Filters... 34
A-4. April-May 1974 - Laboratory Comparison 35
A-5. June 1974 - Laboratory Comparison 36
A-6. July 1974 - Lab Comparison for Side-by-Side CHESS Sites 37
A-7. May 1973 - Quality Control Spiked Solutions for Sulfates
Mean Response and 95% Confidence Interval per
Laboratory per Level 38
A-8. December 1973 - RETA Analysis of Spiked Solutions 39
A-9. NYCDAR-CHES5 N03 Discrepancy Study - RETA and QAEML
Analysis for N03 in Solutions 40
A-10. July 1974 - Analysis of Unexposed Spiked Filters Mean
Responses and 95% Confidence Intervals at Each Level 41
VI 1
-------
SECTION 1
INTRODUCTION
The sulfate and nitrate fractions of total suspended participate Community
Health Environmental Surveilance Systems (CHESS) data showed a dramatic and
sustained decrease beginning in September and October of 1972. A review of
the data in Tables 1 through 4 indicates that the sudden decrease was not an
isolated incident, but that a peculiarity existed across CHESS sites and the
drop could not be attributed solely to seasonal variation.
Tables 1 and 2 show the sulfate ratios of the quarterly arithmetic means
to the respective 1973 quarterly arithmetic means by site. The sulfate ratios
for the first three quarters are generally similar through 1972, but drop in
1973. However, the fourth quarter ratios show the decrease dn 1972, and it
is sustained in 1973.
The nitrate ratios displayed in Tables 3 and 4 show similar decreases,
although they also show a difference between 1970 and 1971 that has never
been explained.
It is unlikely that this across-the-board decrease could have been
caused by a decrease in actual pollutant levels. An investigation by
Environmental Monitoring and Support Laboratory (EMSL) could uncover no
significant Federal or State sulfur regulations that were implemented for
the first time during the period in question.
The most obvious source of error was a change in the laboratories
responsible for the sulfate and nitrate analyses. On October 1, 1972 the
-------
Bioenvironmental Laboratory Branch (BELB) of the Human Studies Laboratory
(HSL), now Health Effects Research Laboratory (HERL), ceased analysis of
sulfates and nitrates, and exposed filters received after that date were
stored for future analysis. A contract was signed in March 1973, turning
over the sulfate and nitrate analyses to Ryckman, Edgerley, Tomlinson, and
Associates (RETA). The same analytical methods used by HSL, BELB, were
also used by RETA. RETA was not fully operational until May 1973 and had
the difficult responsibility of overcoming the backlog of filters — which
were not stored and shipped under the most desirable conditions. Therefore,
many of the filters were not actually analyzed until May and June of 1973.
The effects of this excessive handling and delays in analyzing the
filters cannot be determined. However, it was felt that the decrease in
pollutant levels was so large, and was sustained for so long, that the
problem entailed more than excessive handling and delays in analyzing the
filters. Therefore, the purpose of this report is to show whether or not
RETA's laboratory procedures produced erroneous results, and if so, how the
existing data base can be adjusted to account for the errors.
Quality control efforts were minimal during the period in which RETA
was performing the analysis. However, our investigation did turn up several
sets of data which were related to the problem.
The Human Studies Laboratory conducted an experiment in November 1974
in which Rockwell International (the current contracted laboratory) reanalyzed
some 219 filters which had been analyzed initially by RETA and HSL. This
experiment provides the most direct approach to the problem, and our con-
clusions and recommendations are based primarily on it.
-------
SECTION 2
OTHER STUDIES RELATED TO THE PROBLEM
Scattered attempts were made by various groups to check the quality of
RETA's output during the period in question. However, a unified approach
to the problem was sorely lacking.
In these experiments (see Appendix), the problem was approached in four
ways: 1) a second laboratory "reanalyzed" exposed filters which had been
analyzed some time earlier, 2) two laboratories analyzed the "same" exposed
filters simultaneously, 3) one or more laboratories analyzed the "same"
solutions of known concentrations, and 4) two laboratories analyzed the
"same" spiked filters of known concentrations.
RETA showed good reproducibility of HSL analyses in April 1973, just
before RETA became operational. However, in June 1974, RETA's reanalysis
of filters was different from earlier analyses by both RETA and HSL.
When RETA and QAEML analyzed the "same" filters simultaneously in
April through June 1974, RETA's measurements were significantly lower.
In July 1974, RETA consistently understated concentrations of spiked
filters.
-------
SECTION 3
ROCKWELL'S REANALYSIS OF CHESS FILTERS
GENERAL PROCEDURES
In November 1974, 219 filters from the CHESS network were retrieved,
and new strips were cut from them and shipped to Rockwell for reanalysis
for sulfates and nitrates. The original design called for one filter per
month per 12 stations from July 1972 through June 1974, but almost one-
third of the filters could not be reanalyzed, either because they were
in poor condition or they could not be located.
Twenty-two of these filters had been initially analyzed by HSL and the
other 197 by RETA. They were selected at random from high-exposure and low-
exposure sites in Charlotte, Birmingham, Chattanooga, New York, Utah, and
California.
In addition to these 219 filters, 62 other filters were sent to
Rockwell for reanalysis. These filters were originally analyzed by RETA,
then set aside for quality control purposes in June 1973 and never
utilized. Since these filters were stored in glassine envelopes, it
was felt that they were better preserved than the others.
Statistical Analysis
An analysis of variance was performed on the 219 filter results to test
for possible effects of sites, days between exposure and reanalysis, and the
laboratory which did the first analysis. The first concentration was
considered one of the independent variables and the second concentration
was analyzed as the dependent variable.
-------
The concentrations, unless otherwise stated, are in micrograms
per cubic meter (yg/m3).
The results in Table 5 show no effects of time or sites, but show
strong influences from the first concentration and from the laboratory
which did the first analysis.
To make a valid comparison between the laboratories we must have them
both operating on the same experimental material. Thus, we must make three
assumptions:
• First, since a second strip was cut from the filter for the reanalysis,
we must assume the pollutant concentration is uniform across the filter.
However, we are not aware of any documentation on this point.
• Second, we assume no change in pollutant amount between analyses due
to the handling of the filters.
• Third, the natural pollutant decomposition over time was assumed to
be negligible. It was generally felt that this is a valid assumption for
sulfates, but some scientists feel that nitrates decompose by a significant
amount.
Subject to the above three assumptions, the data set can be thought of
as n pairs of realizations of the log-normally distributed random variables,
x and y. Since the filters were pulled from a population at random, we can
assume that the random variables X = In x and Y = In y have a bivariate
normal distribution with means y2. ancj y2 variances a2 and a2 and correlation
x y x y
coefficent p.
Then the random variable Z = X - Y is normally distributed with mean
yz = yx " yy and variance °z = CTx + °y " 2pCTxay' The fact tnat exP(z) = X/Y
enables us to make inferences about the ratio of the random variables x and y.
It was felt that gross violations of the first two assumptions would be
-------
manifest as outliers. Therefore, the criterion for labeling an observation
as an outlier and excluding it from the analysis was based on the ratio.
If we represent the sample arithmetic mean by I and the sample variance
by S2, we know that the statistic t=(Z"-yJ/(S//rvT) is distributed as Student's
t with n-1 degrees of freedom. Then t is used to test the hypothesis that
y =0, or equivalently that x=y. Confidence intervals are constructed for
y , which is the geometric mean of the ratio x/y.
The computer printouts show the descriptive statistics for x and y (the
numerator in the ratio is listed first, then the denominator in the second
column) along with the value of t and the 95% confidence interval for y .
The value of t is significant at the 0.05 level if and only if the confidence
interval includes one.
Table 6 shows the results of Rockwell's reanalysis of the 197 filters
originally analyzed by RETA. RETA is 43-60% low on sulfates, with a mean
ratio of 51%, but there is no significant difference on nitrates.
The data from the 62 filters which were stored in glassine envelopes
are analyzed in Table 7. RETA comes up 22-35% low on sulfates, but 31-55%
high on nitrates.
The 22 filters which were first analyzed by HSL are analyzed in Table 8.
There is no significant difference on sulfates, but HSL is 18-66% high on
nitrates.
The filters initially analyzed by RETA were divided into three groups
to see if this breakdown would show an effect that the analysis of variance
did not show. The first group, filters exposed between September 1972, and
June 1973, were the backlogged filters which were more physically abused.
The second group, filters exposed between July 1973, and December 1973, were
-------
analyzed on a more timely basis. The other group of filters exposed in 1974
were analyzed on RETA's regular schedule -- within three weeks of exposure.
A comparison of the results in Tables 9-11 with Table 6 shows no group
effect. The sulfate intervals are 36-60%, 32-70%, and 48-72% respectively,
all of which contain 51%. None of the three groups show a significant
difference for nitrates.
The ratio of Rockwell to RETA is plotted against the RETA measurement
for each time group in Figures 1-3. The plots indicate that the ratio does
not depend on pollutant level.
-------
SECTION 4
CONCLUSIONS
We recommend that sulfates be increased by 51% for all CHESS sites
for the entire time period RETA performed the analyses.
We believe there is a decomposition effect with nitrates, as evidenced
by the Tables 7 and 8. Since we could not detect a difference between RETA
and Rockwell on the 197 filters, we cannot recommend a nitrate adjustment.
However, if decomposition prevented a valid comparison, an independent
study to quantify that effect could enable us to formulate a nitrate adjust-
ment factor.
8
-------
TABLE 1. RATIO OF SULFATE QUARTERLY ARITHMETIC MEAN TO
1973 QUARTERLY ARITHMETIC MEAN - - FIRST
AND SECOND QUARTERS
Site
0211
0221
0232
0312
0323
0331
0411
0421
0431
0511
0521
0531
0541
0621
0622
0631
0632
0633
0634
0635
0641
0642
1st Quarter
70 71
2.36 1.62
1.61 1.24
2.74 1.95
1.33
2.26
1.78 1.50
2.54
2.25
2.03
0.32
0.62
0.61
0.80
1.19
1.43
1.53
1.50
1.55
1.11
1.74
1.27
1.29
72
1.76
1.34
2.10
1.80
2.89
1.86
2.80
2.31
1.92
1.16
0.87
0.70
0.84
1.60
2.10
1.85
1.56
2.14
1.48
1.98
1.54
1.88
73
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00-
1.00
1.00
1.00
1.00
1.00
1.00
1.00
2nd Quarter
70 71
1.90 1.51
1.83 1.58
2.09 1.54
1.63
1.31
2.27 1.75
1.10
1.26
1.38
1.57
1.79
l.BtJ
1.51
1.40
1.38
0.81
1.61
1.41
1.37
1.43
1.32
1.28
72
1.41
1.41
1.65
1.89
1 .33
1.67
1.51
1.41
1.49
2.23
2.54
1.69
•1.49
1.56
1.47
0.89
1.51
1.47
1.48
1.35
1.34
1.50
73
1.00
1.00
1 .00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1 .00
1.00
1.00
1 .00
1.00
1.00
1.00
1.00
(continued)
-------
TABLE 1 (continued)
Site
70
1st Quarter
71 72
73
70
2nd Quarter
71 72
73
0711
0721
0722
0732
2.27 1.00
1.82 1.00
1.79 1.00
1.90 1.00
0.87 1.72 1.00
0.76 1.58 1.00
0.72 1.48 1.00
1.71 1.00
0811
0821
0822
0831
0832
0841
0842
3.97
5.61
5.80
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.79
1.15
1.18
1.26
1.17
1.23
1.23
1.00
1.00
1 .00
1.00
1.00
1.00
1.00
10
-------
Site
TABLE 2. RATIO OF SULFATE QUARTERLY ARITHMETIC MEAN TO
1973 QUARTERLY ARITHMETIC MEAN - - THIRD AND
FOURTH QUARTERS
3rd Quarter
4th Quarter
0211
0221
0232
0312
0322
0331
0411
0421
0431
0511
0521
053T
0541
0621
0622
0631
0632
0633
0634
0635
0641
0642
70
1.16
1.09
1.47
1.52
1.41
1.59
1.45
1.52
1.64
1.69
1.45
1.52
1.23
71
0.95
1.03
1.22
1.41
1.16
1.30
1.28
1.23
1.28
1.52
1.43
1.35
1.10
1.10
1.21
1.19
1.29
1.43
1.27
1.17
1.10
1.00
72
1.14
1.09
1.24
1.61
1.45
1.40
1.35
1.16
1.18
1.12
1.37
1.18
. 1.01
1.35
1.43
1.39
1.26
1.80
1.64
1.35
1.47
1.30
73
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
70
1.82
1.49
1.96
1.64
1.79
2.22
3.13
2.27
2.33
1.67
1,67
1.72
1.47
1.89
1.75
1.92
1.82
1.75
71
1.98
1.90
2.08
1.57
1.63
2.07
2.91
2.11
2.44
3.45
2.86
2.44
1.49
1.73
1.60
1.60
1.53
1.91
1.82
1.92
1.89
1.75
72
1.20
0.67
1.06
1.05
1.14
1.44
0.94
0.86
1.00
1.72
1.71
1.68
1.07
1.15
0.97
1.21
1.01
1.23
1.39
1.15
1.13
1.18
73
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
(continued)
11
-------
TABLE 2 (continued)
Site
0711*
0721*
0722*
0732*
0811
0821
0822
0831
0832
0841
0842
3rd Quarter
4th Quarter
70
71
1.18
1.08
1.11
72
1.00
1.00
1.00
1.00
0.84
0.79
0.82
0.75
0.81
0.85
0.77
73
1.00
1.00
1.00
1.00
1.00
1.00
1.00
70
71
2.56
2.50
2.04
72
1.00
1.00
1.00
1.00
0.72
0.73
0.63
0.57
0.57
0.55
0.71
73
1.00
1.00
1.00
1.00
1.00
1.00
1 .00
NEW JERSEY DENOMINATORS ARE 1972 QUARTERLY MEANS
12
-------
TABLE 3. RATIO OF NITRATE QUARTERLY ARITHMETIC MEAN
TO 1973 QUARTERLY ARITHMETIC MEAN - - FIRST
AND SECOND QUARTERS
Site
0211
0221
0232
0312
0323
0331
0411
0421
0431
0511
0521
0531
0541
0621
0622
0631
0632
0633
0634
0635
0641
0642
1st
70 71
1.50 1.00
1.00 0.69
2.00 1.22
1.23
1 .36
1.56 1.22
1.18
1.21
1.00
0.50
0.71
0.80
1.21
2.40
1.62
19.75
3.60
6.12
5.50
2.40
3.14
1.60
Quarter
72
2.17
1.23
2.00
1.30
1.29
1.28
2.36
2.86
1.95
1.00
1.58
2.40
4.86
3.00
2.46
11.00
3.40
5.62
5.67
2.70
2.71
1.90
73
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1 ;--00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
2nd
70 71
1.50 3.00
1 . 50 3.17
1.57 3.00
2.89
2.21
1.64 2.50
1.25
1.86
2.33
3.20
3.14
2.29
3.00
5.14
4.57
3.83
5.33
7.00
7.14
6.00
4.00
3.43
Quarter
72
1.50
1.67
1.57
1 .56
1.21
1.43
1.75
2.14
1.83
2.80
2.14
1.57
2.00
2.14
1.86
1.38
2.78
3.13
3.29
3.00
1 .57
1.71
73
1.00
1.00
1 .00
1.00
1.00
1.00
1 .00
1.00
1.00
1 .00
1.00
~1 .00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1 .00
1.00
1.00
(continued)
13
-------
TABLE 3 (continued)
Site
70
1st Quarter
71 72
0711
0721
0722
0732
0811
0821
0822
0831
0832
0841
0842
4.36
3.89
7.79
73
2.23 1.00
1.94 1.00
2.00 1.00
2.00 1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
2nd Quarter
70 71 72 73
9.00 3.00 1.00
5.63 2.63 1.00
3.20 1.50 1.00
2.25 1.00
2.17
1.41
2.08
1.95
1.71
3.27
2.60
1.00
1.00
1.00
1.00
1.00
1.00
1.00
14
-------
Site
TABLE 4. RATIO OF NITRATE QUARTERLY ARITHMETIC MEAN
TO 1973 QUARTERLY ARITHMETIC MEAN -- THIRD
AND FOURTH QUARTERS
3rd Quarter
4th Quarter
0211
0221
0232
0312
032?
0331
0411
0421
0431
0511
0521
0531
0541
0621
0622
0631
0632
0633
0634
0635
0641
0642
70
1.00
2.50
2.00
1.45
1.49
1.00
3.57
2.63
4.00
2.78
2.38
5.56
2.00
1.49
1.00
71
4.50
8.50
9.50
3.85
2.23
2.07
3.45
4.55
4.00
3.23
3.23
2.33
3.45
4.96
7.89
6.08
6.94
5.76
18.67
7.26
4.73
4.00
72
2.00
2.00
2.50
2.23
1.57
1.24
1.48
1.88
2.00
1.61
1.32
1.33
1.83
2.00
2.32
7.40
3.97
3.57
9.60
3.76
1.49
1.50
73
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
70
0.71
0.70
0.80
0.85
0.67
0.65
1.27
1.00
0.94
3.70
1.67
1.16
1.16
1.89
2.22
4.00
1.69
2.33
3.85
2.27
1.41
1.56
71
2.86
2.50
2.80
2.08
1.67
1.65
2.80
2.89
3.01
2.81
3.00
3.57
2.57
3.87
4.67
3.76
3.54
4.42
9.39
3.20
2.32
2.67
72
0.86
0.80
0.80
1.00
0.80
1.00
0.81
0.61
0.59
1.00
1.17
0.83
0.50
1.09
1.00
1.48
1.00
0.77
0.85
0.68
0.66
0.67
73
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
OD
1.00
1,00
1 .00
1.00
1.00
1.00
1.00
1.00
1.00
(continued)
15
-------
TABLE 4 (continued)
Site
0711*
0721*
0722*
0732*
0811
0821
0822
0831
0832
0841
0842
3rd Quarter
4th Quarter
70
71
2.08
3.45
2.72
72
1.00
1.00
1.00
1.00
2.70
1.85
3.85
3.13
2.56
3.23
3.23
73
1.00
1.00
1.00
1.00
1.00
1.00
1.00
70
71
5.00
5.26
4.17
3.03
72
1.00
1.00
1.00
1.00
1.03
0.91
0.67
0.42
0.85
0.81
0.62
73
1.00
1.00
1.00
1.00
1.00
1.00
1.00
"NEW JERSEY DENOMINATORS ARE 1972 QUARTERLY MEANS
16
-------
TABLE 5. ANALYSIS OF VARIANCE FOR ROCKWELL'S REANALYSIS
Factor
Initial Concentration
Days Between Exposure and Reanalysis
Sites
Laboratories
Factor
Initial Concentration
Days Between Exposure and Reanalysis
Sites
Laboratories
Sulfates
d.f.
1
1
1
1
Nitrates
d.f.
1
1
1
1
Partial S.S,
26280.2
83.3
124.0
401.4
Partial S.S
868.9
11.6
5.8
33.2
F value p
644.85 0.000
2.04 0.154
3.04 0.083
9.85 0.002
F value p
248.40 0.000
3.32 0.070
1.66 0.199
9.49 0.002
17
-------
LU
oo o o o en
CM co o en t—
CM en r— o LO
C\J
CO
CO
CM
OO
o:
LU
CD
ro
10
CO
en
LO
CM
O
« OO
LO O
CO
o
ID
CO
cu
2
o
o
oo
o
co
CO
LO
CO
, —
0
o
CM
id
1 —
o
0
o
00
o
<—
CO
p^.
CO
CM
O
O
CM
Q
LU
C£
_l
_J
O
O
o;
t/)
-
en
o
o
CO
i-O
o
o
O r— CM
O CM O
en o i—
o
o
i— CO
en
CM
CM r—
00 «*
LO r^
• •
CM t—
en
OL
LU
CD
CQ
CO
O
4J
•r-
LO
IO
-P
03
o;
eu
o
•r—
S-
cu
o
eu
(O 4->
> fO
s- cc:
eu
+J c
C ro
1—1 QJ
s
cu
o o
C -i-
cu s-
T3 +->
•i- CU
t- E
c o
o cu
O CD
s« s-
LO O
en 14-
10
CD
O
C
cu
CO
O
O
c
o
ro
t—
eu
S-
S-
O
CJ>
CO
c
(J •!-
(O
>
S-
CO
CO
JD
o
i.
eu
rO
CD
o
-M
•r~
s-
ro
eu
(J
si
4->
cu
o
cu
cs
c
o
X C
ro -i-
o
+->
ra
CD
Q
-o
ro
-»-> -0
ro C
•i- ro
> 4->
CO 00
Q
O
T3 -i-
s- s-
rO *->
-O CO
o
eu
ro
+->
oo
18
-------
oo
O)
-M
03
S-
10
LO
in
10
co
to
o
CM
1
0
01
CM
1
'oj
o
o
01
0
•-'
LO
CM
en
o
LO
o
o
0
<3-
0
0
LO
O
CTl
O
o
o
CTi
O
0
LO
to
0
CM
CM
-"
LO
CM
LO
CM
CM
LO
^D
CO
00
oo
=
Q
LiJ
M
_J
c£
•^
<£
UJ
rv*
1
_l
UJ
(__ J
O
•— t-
to
CD
•4_)
re
M-
i^—
•^
OO
^J
oo
• ^d~
CM i — i —
CO LO
CM "CTl
Q
r— CM O
CM
•—
CM
<£.
t
or
o
LO
co'
0
,—
,—
0)
•5.
o
o
ce:
CO
o
i —
CM
CM
CD
CO
r^
LO
"~
r^
co
CM
o
CM
0
O
CO
LO
o
o
to
00
LO
o
o
1^-
CM
o
o
LO
LO
•vf
l^~
CD
"~
O
0
LO
0
r—
CM
CM
CO
r-'
CTl
UO
to
r-T
CTl
UJ
CO
to
CD
o
(/)
O
•r—
t^
• i—
4J
(O
+J
C/7
O
+J
Qi
C
rO
0)
E
o
QJ
C3
Interval
O)
0
c
O)
E
o
CD
CJ3
i_
0
fficient of Li
OJ
o
E
O
•r—
-l->
ro
P^
0)
S-
s_
o
o
vations
s_
OJ
to
o
M-
O
^,
<1J
^>
E
•z.
E
ro
0)
s:
o
• r—
^_)
O)
E
-i->
•r—
S-
c£
E
rO
OJ
s:
o
•r-
s-
^-^
ro
•»—
>
Ol
Q
TD
E i-
3 ro
^E "^3
•r- C
C fO
•r- -(->
S OO
E
O
•r—
-1-)
ro
O)
TD
S_
T3
E
ro
•P
00
O
•i—
S-
4-)
rjj
E
0
OJ
o
19
-------
CU
•P
(O
O
CO
00
a:
CM
CM
CO
oo
•
O
VO
00
CTl CM O O CM
O 00 O CT> O1
VO ^f O O i—
oo i— en o LO
10
CT)
C\J
CM
CM
£Z
CU
2
o
o
o;
LO
oo
i —
CM
LO
00
r-«
0
O
0
CM
01
o
O1
0
o
LO
•S3-
CM
co
oo
oo
LO r- o o o
UD l£> O O LO
r**- r^ r^*- CM LO
*3- i— i— r~~ CM
CM CM LO i—
r— OO
oo
*• CTl
^3- 0
O1
•
O
O
CM
,
f^
CU
S
o
o
LO
^«O
•S3"
LO
CM
00
o
LO
CM
CM
O
0
•S3-
CM
LO
O
o
LO
cri
00
CO
CM
CM
i —
«d-
r---
CTl
o
o;
CTl
t/i
en
o
oo
CO
•X.
(/>
O
4->
0)
E
o
(1)
fO
>
i.
CU
C;
1— 1
O)
o
c:
CU
•a
•r-
M-
C
0
0
^5
LO
cr>
4_>
(C
Di
to
CU
2:
u
^.
-p
CU
E
o
CU
CD
S-
o
4-
^
CU
o
t£
Lf_
OJ
o
o
c
o
• r—
-p
(O
r^
CU
S-
s-
o
o
in
C
o
•r-
[ ^
fO
>
S-
cu
t/1
J3
0
4-
0
S-
cu
_a
E
^
CJ3
X C
tO '!-
c
o
CU
o
TD
S-
fO
+J
oo
CU
T3
S-
ra
-o
c
(0
-p
oo
u
o
a>
20
-------
UJ
CT>
oo
CO
o
(XI
o
o
LT>
o -=j- oo
CTi O i—
o •=*• r-~
o oo cvj
oo
n:
a
«C
03
to
a>
rO
S-
CT>
CTi
O
CTi
r—
CO
o
un
oo
'oj
3
u
o
o
CVl
oo
•
oo
ID
to
r—
•
CM
O
o
^j-
•
oo
o
o
CM
•
O
CVJ
LO
o
•
oo
to
CvJ
Q
UJ
UJ
CJ
o
Di
O)
OO
o
to
CM
LO
OO
U1
LD
LT>
CXD
a;
j*:
o
o
a:
CM
o
<=!-
tn
cr>
o
o
CO
O
O
o
o
O
O
oo
LO
CTl
CM
LT)
CO
•
CXI
10
en
ca
CD
O
c
o
CTi
CQ
o
(O
i^ C£
cu
(U
to
c
o
.)_>
(/)
• r-
+->
03
-|_3
00
03
Di
f~
(0
cu
s:
o
tf—
s_
-I_J
cu
E
o
cu
CD
c
t—i
cu
o
c:
-o
• r—
tf—
C"
0
O
ci^i
tn
CT.
03
CU
^~
(_)
• f^
i_
-(->
O)
E
0
cu
CD
i_
0
H—
4-
M-
CU
O
t_5
c
o
• I—
4^
03
r—
CU
5-
t_
O
0
03
>
S-
o>
to
f^i
o
M-
O
^.
CU
o
E
3
^
(T3
CU
o
cu
c
(O
cu
o
si
cu
o
cu
CD
1
X
03
c
o
03
-o
"3
03
4-)
00
O3
•r—
OJ
T3
S-
re>
T3
c
03
4->
OO
cu
o
CO
CD
21
-------
"vt- OO O O *3-
ID i— O CT> CM
CM 00 O O «d-
• • • • »
OO i— «J- O OO
oo
oo
c£
LU
OO
1^
CT>
O)
CO
CTi
•— CT>
LO
•> r»»
CT>
•— o
10
CU
^<:
u
o
OL
•— CO
LT> CO
VO •*
• •
CM i—
O O
O CTi
CM O
VO O
CO
oo
•a-
CO
CO
C£.
LU
I—
<
—I
LU
£
_J
oo
CM
o
CM
CM
LO
O
to
i —
o
0
CM
CTi
o
0
*±
CM
^f-
0
r^
i —
CO
CM
CTl
,-J
CTl
OL
LU
CQ
CD
o
CO
re
in
o
•1 —
•»->
I/)
4_>
re
•P
OO
0
•r—
+->
re
cc
C"
re
0)
2:
o
•r—
c^
4_>
GJ
^
O
OJ
E
0
O)
CD
5-
O
4-
efficien
o
o
c
o
•r—
4->
re
1—
CJ
i.
i.
o
0
rvations
O)
10
o
o
4-
0
i^
Ol
o
E
3
•z.
C.
to
cu
S
o
• ^
4_^
O)
F
c~
4-J
•r—
s_
«c
c
re
cu
2:
o
il
i ^
Cl)
^
O
CD
E
3
E
X
s:
pr
3
E
C
•r—
51
c
0
(O
•I—
>
OJ
t 1
TD
i-
re
-a
c
re
4^
oo
0)
o
TJ
5-
re
c
re
4_>
OO
U
•r-
s_
40
ai
E
O
a>
22
-------
to
CO
00
a:
r--.
CTi
>-
_1
eC
UJ
Q
UJ
a:
o
§
CTi
UJ
CQ
CO
•
O
o
o
I—.
CM
CO
ID
in
CQ
to
03
t(-
^^
~^
00
LO
CTi
LO
•
r—
o
to
O
1 i
10
•1—
4->
10
4->
CD
E
0
cu
CD
,
>
S-
cu
4^
^
O)
o
C"
cu
T3
•t—
t^ —
C
o
o
^^
LO
CTl
^
n
CM
CO
—
0
to
f~
fO
r—
OJ
s-
i_
0
o
VD
LO
C
o
«l —
4-3
«3
S_
O)
(/)
o
O
(4 —
o
£_
cu
.0
E
z
i —
•
r
r
r
*>^
^
•
OO CM
LO
CM
O
LO
LO
CM
LO
ya
r
r
^
r
^^
-^.
O
CM
C
NJ
LO
r
.
'
C
0
p
(O
cu
Q
c
(O
O)
s:
o
•r—
4-3
O)
E
4->
•i —
S-
cC
c
(Q
01
s:
0
•1 —
^_
+J
cu
f~
o
cu
CJ3
E
E
X
(O
s:
E
i
•r—
C
• i —
s:
c
0
'43
«3
• r—
>
CU
Q
-o
S-
T3
C
(^
4-)
00
T3
TD
f~
(O
4->
00
O
T—
S-
4->
CU
E
0
O)
CD
23
-------
c
e
u
E
P
fe
(*-
L
a:
...L
1^ !.
o: IK
r-
i in
;
IM
JT
-}
ir
I'
'1 !
101
Lu 0
IN
IM
f-
IM
rj
|r-
I in
lui
IM I
i i
i
rsl
I
ICT
N
— C 4 * *J ft
24
-------
IT
ii
. i <•
is
0
,t
o
Is!
25
-------
la-
id
IE?
tt
l«r
|>
O
_J
a.
-I-
rg
it"
t •
i
i!
I1"
! 10
i
i 10
03
Cj
f^.
a
-------
APPENDIX
EXPERIMENTS INVOLVING THE REANALYSIS OF EXPOSED FILTERS
The same rationale that was used in the Statistical Analysis of the
Rockwell experiment in the text is used here.
1.A. (Table A-l) In April 1973, several weeks before RETA was fully
operational, 10 filters from the CHESS network which had been analyzed
by HSL in September 1972 were flown to St. Louis to have RETA reanalyze
them for sulfates.
The data are in milligrams per filter, so the values would have to be
approximately halved to convert to micrograms per cubic meter. Table A-ll
contains conversion factors for the different units used in these studies.
The analysis shows a laboratory difference, but it is only 3-15%. In
the absence of any standards to go by, this was, and still is- considered
good agreement.
1.B. (Tables A-2 and A-3) In June 1974, RETA was asked to reanalyze 30
filters exposed in the CHESS network in September and October 1972.
Eighteen of the filters were initially analyzed by RETA, and the remaining
12 were analyzed first by HSL.
Table A-2 shows the results for the 18 filters first analyzed by RETA.
The 1974 analyses were 1-15% higher than 1972 for sulfates, but were 13-32%
higher than 1972 for nitrates.
Table A-3 compares RETA 1974 vs. HSL 1972. RETA's sulfate measurements
were 35~108% lower than HSL's. The RETA nitrate values were also well below
HSL's, but the correlation between nitrate values was a low 0.478.
27
-------
EXPERIMENTS COMPARING LABORATORIES ANALYZING THE SAME FILTERS SIMULTANEOUSLY
In the experiments, two strips were cut from the filter and one given
to each laboratory for analysis. Thus, we need to make the assumption of
uniformity of pollutant concentration across the filter. However, the other
two assumptions are not necessary, as there is no timelag between analyses.
Statistical Analysis
The same approach was used on these data sets that was used on the
experiments involving reanalysis of exposed filters.
2.A. (Table A-4) In April and May 1974, 13 filters were given to RETA and
QAEML for a laboratory comparison for both sulfates and nitrates.
The measurements showed high correlations, but RETA was 23-41% low on
sulfates and 26-46% low on nitrates.
2.B. (Table A-5) In June 1974, another laboratory comparison involving
RETA and QAEML was performed using 16 filters. Only nitrates were analyzed,
as this was part of the New York City DAR-CHESS "N03 Discrepancy Study.
The data are in rr.icrograms per filter*"strip, so the values-would have
to be divided by approximately 200 to convert to micrograms per cubic meter.
RETA was 19-30% low on these nitrate measurements.
2.C. (Table A-6) In July 1974, two CHESS hi-vol instruments were located
in Garden Grove, California, for 21 days and in Thousand Oaks, Calfornia,
for two days. The filters from one instrument were analyzed by RETA and the
other by Rockwell.
These data necessitate the further assumption that the pollutant
concentration is uniform across the CHESS sites. The Garden Grove
instruments were located one mile apart.
28
-------
There was no significant difference between the RETA and Rockwell measure-
ments of sulfates. RETA was low on nitrates, but the correlation coefficient
on nitrates was a very low 0.247.
EXPERIMENTS COMPARING LABORATORIES ANALYZING SPIKED SOLUTIONS
Spiked solutions containing known (unknown to the laboratory) pollutant
concentrations were given to each laboratory for analysis. Several levels
of concentration were used.
The only assumption needed is that the solutions are homogeneous. If
the solution was separated into bottles, aliquots, or otherwise broken down
into smaller containers, it needs to be assumed that the concentration would
remain the same throughout.
All the data are in micrograms per milliliter, so the values would have
to be divided by approximately 3.4 to canvert to micrograms per cubic meter.
Statistical Analysis
We assumed that all solutions were identical except for pollutant content,
and that level was known without error. The mean respefise of each .laboratory
at each level was tabulated, and a 95% confidence interval for that if?.ean-was
calculated when possible. If the laboratory was functioning correctly, we
would expect the confidence interval to include the known level of input.
3.A. (Table A-7) In May 1973, RETA and OAEML were each given five sulfate
solutions at each of three levels to check RETA's quality. Both laboratories
exhibited errors in measurement, but due to ignorance concerning the method
itself, no alarm was sounded.
3.B. (Table A-8) In December 1973, RETA analyzed 16 solutions at two
pollutant levels for sulfates and nitrates. Two solutions at each level
were analyzed on both the morning and afternoon of each of four days.
29
-------
An analysis of variance was performed to test for possible time of day
and data effects. Neither was significant for sulfates, but both were
significant at the 0.05 level for nitrates. The ANOVA table appears with
RETA's estimates of the levels in Table A-8.
RETA's accuracy is poor, but these are very low levels, especially
for nitrates.
3.C. (Table A-9) In June 1974, as part of the New York City DAR-CHESS N03
Discrepancy Study, RETA and QAEML analyzed solutions at six levels of N03.
QAEML provided one measurement at each level, while RETA analyzed three
solutions at each level.
Confidence intervals were not calculated, as RETA's values contained
almost no variability. Although RETA appears to obtain lower measurements
at higher concentrations, we did not evaluate these differences because of
the small sample size and lack of knowledge concerning interlaboratory
variablity.
EXPERIMENT INVOLVING ANftfcYSM-"8F SPIKED FILTERS
In July 1974, Rockwell and RETA analyzed five filter strips of known
concentration at each of five levels for sulfates and at each of four levels
for nitrates.
It has to be assumed that the pollutant concentration was distributed
uniformly across the filters.
The data are in milligrams per filter, so the numbers would have to be
approximately halved to convert to micrograms per cubic meter.
Statistical Analysis
Mean responses and confidence intervals for the mean responses were
tabulated as in Section 3 above.
30
-------
Table A-10 contains the data showing that RETA is consistently low on
both sulfates and nitrates. Rockwell is also low at some levels, but not to
the same extent as RETA. A knowledge of interlaboratory variability would
enable further inferences to be made from this data set.
CONVERSION FACTORS FOR TABLES
Assuming: Time = 1440 minutes, Flow = 50 cubic feet per minute
Micrograms per Cubic Meter = 2.95 x 10"1 Micrograms per Milliliter
= 5.89 x 10"3 Micrograms per Strip
= 4.91 x lO"4 Micrograms per Filter
31
-------
TABLE A-l. APRIL 1973 - RETA REANALYZED HSL'S SEPTEMBER 1972 FILTERS
Sulfates
Geometric Mean Ratio 0.906
95% Confidence Interval
for Geometric Mean Ratio (0.846, 0.970)
Correlation Coefficient of Logs 0.9818
Number of Observations 10
HSL RETA
Arithmetic Mean 33.680 36.360
Geometric Mean 31.424 34.676
Maximum 58.200 57.600
Minimum 20.000 23.400
Standard Deviatton 13.747 12.307
Geometric Standard Deviation 1.472 1.375
32
-------
GO
Di
LU
CM
1-^
CT>
CM
CM
CM
, *
to
r—
00
• UO
i— tO CO
to i—
LU
CtL
>-.
71
to
LU
CM
oo
r~_
•
CM
CO
i —
CT>
•
f—
O
o
CM
•
*3-
o
o
i —
•
0
^~
r^.
en
•
o
oo
uo
to
•
,—
00
r^- i— o
CM «d- en
en oo 01
5* uo
o o
o uo
Di
CO
O
o
o
00
1
LU
(y^
0
LU
>-
CM
CT)
CO
et
I—
LU
C£.
00 O
oo to
O OO
CM
no
O
o
CO
o
o
i— i— cr>
o
o
CO
o
o
uo
CM
oo
oo
oo
LO
in
o
Qi
I
CTi
oo
C7>
o
c
o
CM
I
CQ
I/)
O
•r"
-M
(/)
•r—
-4_>
tO
•p
OO
o
•r—
^_>
fC
o:
c:
(O
cu
s:
o
si
4_3
cu
E
o
cu
cs
«
>
CU
-M
C
t—i
CU
o
c
T3
•r—
M-
C
a
, )
^5
UO
en
+->
n3
C£.
c
s
o
t-
4->
(y
E
O
CD
s_
0
c
cu
•r-
o
14
'I
cu
0
o
c
o
• r-
-(->
fO
r-.
cu
S-
s_
o
0
00
c:
o
•r—
4->
(O
s_
cu
00
-Q
O
M-
o
S-
cu
-Q
2:
c
(O
a>
o
a)
CU S-
E 4->
^ cu
-P E
••- O
s- cu
< C5
E
E
CU
Q
•a
s-
(O
T3
oo
cu
-o
-o
c
-------
00
o:
1 i 1
ULJ
1—
_l
t— 1
U_
CM
r--.
en
UJ
CO
s:
LtJ
Q.
UJ
oo
CO
—
1
CO
n:
o
UJ
>_
i
^^
^^
^
UJ
fV
=t
UJ
o:
• — "• ^c
co h-
LO UJ
to en VQ c£
CU • Is-
4-> O «3- Is- CM
i- CM •
4-> • en o
•r- CM LO
•z. o
_l
s— ' "T~
^-~x ^£
r— (—
CO UJ
tO Or- OL
CU • ^3"
4~> ^ CM LO i —
03 i — en i —
l^ t^ «s .
r- O O
3 i— in
CO CO
_i
r- CO
%fc— •* ~T*
in us o i —
en o o co
CM r~- co o
.
i— O CO O
o oo o o
m •— o o
O lf> ^" ^~
• • • •
CM r — «d" O
r^ uo c^ f~>
CM CM O O
CO r— IX) r—
• • • •
en r- co CM
r—
CO «3- O O
•— «3- O O
co cr> oo en
• • • •
CO r— «3- <*
i— r— CM
^_
O
CM
r>—
LO
LO
CO
•
1
to
CM
CM
•
IO
LO
r—
oo
•
p^.
CM
<«O
en
CO
en
CO
, —
•
CM
CM
en
CM
•
CM
CO
en
p— ,
•
r«
en
to
Oi
o
CO
CO
10
u
•4->
IO
•r—
-4->
n3
i ^
CO
o
•r-
4_>
03
o:
c
fO
cu
s:
u
•r-
i.
4-)
QJ
E
0
0)
03
>
S-
OJ
E:
»— i
(U
o
c
0)
•r—
4-
c
0
CJ
s*^
LO
en
4J
n3
OL
C.
(U
E
O
OJ
f n
i.
O
C
(U
•r~
O
L^.
n-
O)
o
c
o
•4_)
fl3
^~
(U
S-
s-
o
o
(/)
c
o
•p-
4->
ro
>
i-
Ol
(/)
.0
0
M-
O
s^.
(U
.0
F^
3
z:
c
(O
0)
u
• r—
4»>
QJ
E
4^)
.i—
Sta.
CU
Q
T3
S-
(O
T3
C
fl3
4>J
CO
>
OJ
r~\
•o
s-
n3
T3
C
n3
4->
CO
O
^.
4->
CU
E
O
QJ
CD
34
-------
o
oo
I—I
a:
d.
O
o
^
o
o
CO
en
>-
i
oc.
a_
et
CO
*£.
in
o;
TO
LT>
oo
O)
3
OO
oo
(SI
o
to
-t->
oo
o
•r~
to
c
O)
o
•r—
O)
E
o
O)
CD
oo
CM
C ro
i—i ID
OJ
O O
O) S-
-D -(->
••- O)
M- E
c o
O O)
CJ CD
^ s-
LT) O
en 4-
CO
tn
en
CO
CM
cn
o
> (O QJ
S- C£. -r-
5-
S-
o
o
CM
oo
UJ
OJ
o
O
O
CO
OO
O
O
CO
O
o
CTi
oo
ID
^—
ID
CO
0
1 —
UD
CO
o
en
co'
c:
(0
d)
s:
o
• r—
+->
(!)
E
+J
•r—
^.
(1)
E
O
CD
CD
O
0
CM
en
O
o
en
o
o
o
CO
un
E
3
E
X
(O
^
o en
0 0
UD CM
CM CM
o en
O «*
OO 1 —
OO CM
O en
O Lf>
CO CO
OO 00
c
0
•r-
.)_)
(O
•r—
>
O)
Q
•o
E S-
3 (O
E -0
•r- C
c to
si oo
CO
CM
ID
en
^~
LD
1
o
o
0
o
c
o
4->
fO
O)
Q
-a
S-
ro
-o
c~
(O
+J
oo
o
tl
4-)
O)
OJ
CD
35
-------
TABLE A-5. JUNE 1974 - LABORATORY COMPARISON
Nitrates
Geometric Mean Ratio 1.242
95% Confidence Interval
for Geometric Mean Ratio (1.189, 1.297)
Correlation Coefficient of Logs 0.9968
Number of Observations 15
EMSL RETA
Arithmetic Mean 558.800 519.120
Geometric Mean 479.062 385.780
Maximum 1762.000 1375.100
Minimum 94.000 79.200
Standard Deviation 114.869 89.003
Geometric Standard Deviation 1.218 1.215
36
-------
f, s,
CO
CO
O)
GO ro
UJ *~-
\ — -t->
K-f •! —
oo 2:
LO
CO
(O OJ ID OO
CO o
o
OO i —
oo • — '
UJ
o
U-1
0
GO
1
CO
1
UJ
Q *-^
I— i OO
GO «=J-
co
fV d)
O -*->
U- ro
LI
' i
2: r—
O 3
oo oo
CM CM
UO OO
r- r- "* CM
i— CTl
"~I OO O
i— CTl
CTi
1 — 1 •
en O
O_
0
O
•=£
UJ
Di
«^j- r^. o
r~- oo o
^O ^J" r~~
• • •
oo CM r-».
^~
^-
r—
OJ
s
o
o
CsL
cn en o
oo o o
CO r— O
• • •
*sa~ ^a~ oo
•~
<£.
h-
UJ
Di
CO O O
r^ oj o
LO OJ OO
• • •
LO «3- LO
1 —
r—
r—
o>
3
U
0
C£.
CO O 0
^J" CJ> O
O ID V£>
• • •
vo «^- , —
OJ
o
o
00
•
o
o
CD
oo
•
1 —
0
o
CM
•
1 —
0
o
CO
•
1—
oo
0
co
•
oo
o
^o
co
•
OJ
CO
oo
r—
•
«3"
o
CM
CO
•
^~
*d~
0*1
LT>
•
oo
1 —
f—
00
•
r—
LO
r—
OO
•
OJ
^~
LO
O
•
OJ
63
c/i
CT)
o
r-.
CTl
>-
_i
ZD
T
O
1 LO
4_>
UJ ro
_J -P
CQ OO
O
rO
o:
c:
ro
(U
s:
ff.
1— O
il
4_)
rjj
E
O
O)
CD
Interval
0)
o
c
a>
-o
•r-
i^
c
o
^5
LO
CTl
O
•r"
C
cu
u
•i —
i.
QJ
E
O
CD
i
O
— 1
<4-
o
c:
OJ
•i —
u
4-
O)
O
O
c:
o
•r-
4_>
rO
QJ
S-
S-
o
rvations
OJ
U1
o
o
4-
O
S-
O)
JD
E
~Z.
ro
O)
s:
tj
•i —
4->
O)
E
jc
-(->
• p-
^_
<=C
c:
ro
QJ
s:
o
•r—
S-
4^
O)
E
0
Ol
CD
pn
3
E
x
(O
•SL
E
3
_E
c
•i —
•Si
o
4->
ro
•r—
>
O)
Q
-o
S-
ro
-o
C
ro
-(->
OO
dard Deviation
c
ro
+-3
00
o
•1 —
S-
4->
O)
E
O
OJ
CD
37
-------
TABLE A-7. MAY 1973 - QUALITY CONTROL SPIKED SOLUTIONS FOR SULFATES
MEAN RESPONSE AND 95% CONFIDENCE INTERVAL PER LABORATORY
PER LEVEL
Concentration Level
(yg/ml) RETA QAEML # of Obs.
0 0.0 -- 2.4 (2.1, 2.6) 5
30 24.8 (20.2, 29.4) 28.5 (27.6, 29.4) 5
60 47.0 (41.7, 52.3) 61.1 (52.1, 70.1) 5
38
-------
rABLE A-8. DECEMBER 1973 - RETA ANALYSIS OF SPIKED SOLUTIONS
Factor S.S.
Total 15977.9
Day 101.3
Level 15487.9
Time 26.2
Error 362.3
Level (pg/ml)
10
60
Factor S.S.
Total 34.98
Day 0.16
Level 34.27
Time 0.10
Error 0.44
Level (yg/ml)
1
10
Sul fates
ANOVA Table
d.f. M.S.
31
3 33.7
1 15487.9
1 26.2
26 13.9
Mean Responses
8,47 (7.41, 9.53)
52.47 (49.62, 55.32)
Nitrates
ANOVA Table
d.f. M.S.
31
3 0.05
1 34.27
1 0.10
26 0.01
Mean Responses
0.195 (0.178, 0.212)
2.265 (2.151, 2.379)
F P
2.42 0.087
1111.34 0.000
1.88 0.178
# of Obs.
16
16
F " P
3.23 0.037
2012.44 0.000
5.94 0.020
# of Obs.
16
16
39
-------
TABLE A-9. NYCDAR-CHESS N03 DISCREPANCY STUDY
RETA AND QAEML ANALYSIS FOR N03 IN SOLUTIONS
Concentration Level RETA
(yg/ml) QAEML (average of three)
4 3.75 4.20
6 5.90 5.90
9 9.00 8.50
11 11.00 10.60
14 14.10 13.47
16 16.10 14.20
40
-------
TABLE A-10. JULY 1974 - ANALYSIS OF UNEXPOSED SPIKED FILTERS
MEAN RESPONSES AND 95% CONFIDENCE INTERVALS AT
EACH LEVEL
Sulfates
Concentration Level
(yg/ml) RETA Rockwell # of Obs,
3.00 1.84 (0.51, 3.17) 2.88 (2.55, 3.21) 5
11.26 9.24 (8.08, 10.40) 11.16 (10.53, 11.79) 5
15.18 12.64 (10.98, 14.30) 15.26 (13.71, 16.81) 5
27.76 22.16 (21.24, 23.08) 25.78 (25.35, 26.21) 5
36.00 24.90 (22.34, 27.46) 34.00 (32.96, 35.04) 5
Nitrates
3.05 2.40 (2.15, 2.65) 2.64 (2.53, 2.75) 5
6.04 4.80 (3.99, 5.61) 5.52 (5.38, 5.66) 5
9.01 7.95 (7.09, 8.81) 8.88 (8.74, 9.02) 5
12.00 8.94 (7.59, 10.29) 10.96 (10.82, 11.10) 5
41
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/1-77-004
2.
3. RECIPIENT'S ACCESSI Of* NO.
4. TITLE AND SUBTITLE
AN ASSESSMENT OF THE CHESS SULFATE AND NITRATE DATA
during the period RETA performed the chemical analysis
5 REPORT DATE
January 1977
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Leo T. Heiderscheit and Marvin B. Hertz
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Health Effects Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
10. PROGRAM ELEMENT NO.
1EA615
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
Health Effects Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
13. TYPE OF REPORT AND PERIOD COVERED
In-house
14. SPONSORING AGENCY CODE
EPA-ORD
15. SUPPLEMENTARY NOTES
16. ABSTRACT
In the early 1970s certain filters from the CHESS network were collected and
sent to the Human Studies Laboratory (HSL) Bioenvironmental Laboratory Branch (BELB)
for sulfate and nitrate analyses, These analyses were interrupted on October 1,
1972 and subsequently continued under contract signed March 1973 with Rickman,
Edgerley, Tomlinson, and Associates (RETA). Many of the filters were not analyzed
until RETA became fully operational in May 1973, whereupon their measurements began
to be inexplicably and consistently low. The disparity between HSL results and
RETA's findings engendered an investigation involving reanalyses to verify the
apparent disparity, and then to determine a statistical adjustment factor to correct
for anomalies. In November 1974 Rockwell International undertook the reanalyses of
these CHESS filters.
In Rockwell's reanalyses, the difference in RETA's sulfate data showed a mean
ratio of 51%, but the results on nitrates were inconclusive. The effects of time
and handling were not conclusively determined, although there was an indication that
nitrates decompose with timelapse more significantly than sulfates. Based on the
results of this report, it was recommended that CHESS measurements for sulfates be
increased by 51% for all sites for the entire period RETA performed these analyses.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
sulfates
inorganic nitrates
chemical analysis
CHESS
07 B
13. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19 SECURITY CLASS (This Report)
UNCLASSI
21. NO. OF PAGES
20. SECURITY CLASS (This page)
UNCLASSIFIED
22
EPA Form 2220-1 (9-73)
42
------- |