&EPA
United States
Environmental Protection
Agency
Industrial Environmental Research
Laboratory
Research Triangle Park NC 27711
EPA-600/7-79-226
September 1979
Interpretation
of Environmental
Assessment Data
Interagency
Energy/Environment
R&D Program Report
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series. Reports in this series result from the
effort funded under the 17-agency Federal Energy/Environment Research and
Development Program. These studies relate to EPA's mission to protect the public
health and welfare from adverse effects of pollutants associated with energy sys-
tems. The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the transport of energy-related pollutants and their health and ecological
effects; assessments of, and development of, control technologies for energy
systems; and integrated assessments of a wide'range of energy-related environ-
mental issues.
EPA REVIEW NOTICE
This report has been reviewed by the participating Federal Agencies, and approved
for publication. Approval does not signify that the contents necessarily reflect
the views and policies of the Government, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/7-79-226
September 1979
Interpretation of Environmental
Assessment Data
by
N. H. Sexton, F. W. Sexton,
L. I. Southerland, and T. D. Hartwell
Research Triangle Institute
P.O. Box 12194
Research Triangle Park, NC 27709
Contract No. 68-02-2156
T. D. No. 22600
Program Element No. EHE624
EPA Project Officer: Larry D. Johnson
Industrial Environmental Research Laboratory
Office of Environmental Engineering and Technology
Research Triangle Park, NC 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
-------
ABSTRACT
Nineteen environmental assessment, projects were performed on a variety
of industrial sources prior to the spring of 1978 and the results were
compiled in Compilation of Level 1 Environmental Assessment Data. EPA-600/
2-78-211. It became apparent during compilation of the data that methods
were needed to organize, summarize, and interpret the bulk of data produced
by environmental assessments. Preliminary attempts to formulate viable
models for interpreting environmental assessment data are presented in this
document. These models are evaluated using data from the four most compre-
hensive environmental assessments: the three pilot studies (the Exxon
fluidized-bed combustor mi nip!ant, the EPA-ATMI study of textile effluents,
and the study of a Chapman low-Btu coal gasifier) and a study to evaluate
the Source Assessment Sampling System (SASS) train at 10 industrial sites.
A format for entering results from environmental assessments on Fortran
computer sheets is presented and more complete data entry sheets, being
developed by the IERL Special Studies staff, are discussed. Various pre-
viously proposed models (Source Severity; Source Assessment Models SAM/I,
SAM/IA, and SAM/IB) are investigated using the data from the four previously
mentioned studies. Linear (Pearson) and rank (Spearman) correlations between
biotests and chemical results, and among biotests, were calculated using the
data from these four studies; a summary of possible rank correlations is
presented. In the study of correlations, the data are examined from each
study individually and from the entire data set taken as a whole. These
results must be viewed as preliminary owing to the limited number of data
pairs available for calculating many of the correlations. However, it is
hoped that the results of the correlation analysis will direct future research
in cause-effect relationships and guide future studies. As more complete
data become available, the procedures described in this document can be
applied to interpret data and develop models.
n
-------
CONTENTS
Page
Abstract ii
Figures v
Tables vi
Acknowledgments viii
1. INTRODUCTION 1
2. DATA IDENTIFICATION 4
2.1 Study 06—Chapman Low-Btu Coal Gasifier 4
2.2 Study 10—Fluidized-Bed Combustor 6
2.3 Study 15--Particulates From Various Industrial
Processes 9
2.4 Study 19—Textile Plants 10
2.5 Description of Sample Data Formats 12
3. QUANTIFICATION OF DATA 19
3.1 Chemical Results—Inorganic 19
3.1.1 Spark Source Mass Spectrometry 19
3.1.2 Anion Analysis 20
3.1.3 Atomic Absorption 20
3.1.4 Chemiluminescence 20
3.1.5 ORSAT 20
3.1.6 Gas Chromatography 20
3.1.7 Aqueous Analysis 20
3.2 Chemical Results—Organic 20
3.3 Bioassay Results 23
3.3.1 Microbial Mutagenicity 26
3.3.2 Rabbit Alveolar Macrophage. ... 27
3.3.3 WI-38 Human Lung Fibroblast 27
3.3.4 Chinese Hamster Ovarian Cell Clonal Assay 27
3.3.5 Rodent Acute Toxicity 27
3.3.6 Freshwater Algal Test 28
3.3.7 Marine Algal Test 28
3.3.8 Daphnia 28
3.3.9 Grass Shrimp 29
3.3.10 Fathead Minnow Test 29
3.3.11 Sheepshead Minnow Test 29
4. DATA ANALYSIS AND INTERPRETATION 36
4.1 Summary Statistics 37
4.2 Correlation Analysis 37
4.2.1 Rank (Spearman) correlation 37
-------
CONTENTS (continued)
Page
4.2.2 Study 06--Coal Gasifier 42
4.2.3 Study 10--Fluidized-Bed Combustor 43
4.2.4 Study 15--SASS Train Evaluation, Various
Industries 43
4.2.5 Study 19--Textile Plants' Effluents 43
4.2.6 All Studies Grouped Together 44
4.2.7 Liquids 47
4.2.8 Solids 53
4.2.9 Summary Correlations 53
4.3 Scatter Plot Analysis 55
4.3.1 Chemical-Biological Test Combinations 60
4.3.2 Biological Test Combinations 66
4.3.3 Summary of "Good" Correlations and
Scatter Plots 67
4.4 Stepwise Regressions 71
4.4.1 Biological Tests V. Chemical Data 71
4.4.2 AMES2 V. Its Component Strains 74
4.4.3 Relationships Between the Biological Tests 74
4.5 Probit Analysis 76
4.6 Analysis of the Data Set Through Use of Models 78
4.7 Other Investigators' Analyses of This Data Set 84
4.8 Engineering Data 87
4.9 Original Results Compared to the Litton Quantification
Scheme 90
4.10 Correlations Among Ames/Salmonella Strains 90
4.11 Comparison of Schemes for Ranking Samples Using Level 1
Data 95
5. CONCLUSIONS 99
5.1 Discussion 99
5.2 Recommendations 103
References 108
Appendixes
A. The Data Base A-l
B. Draft Report From Don Lewis B-l
C. "Good" Correlations Found in This Data Set C-l
D. Battelle Pattern Recognition Report D-l
IV
-------
FIGURES
Number Page
1 Process flow diagram for the coal gasification unit
(study 06), site A, Level 1 sampling locations 5
2 Exxon fluidized-bed combustion miniplant (study 10),
Level 1 sampling locations 8
3 Process flow diagram for the textile plants (study 19),
Level 1 sampling locations 11
4 Plot of TA98P V. AMES2 38
5 Scatter plot showing logarithmic curve pattern 63
6 Scatter plot showing linear relationship 64
7 Scatter plot showing random distribution 65
8 Scatter plot showing negative slope 68
9 Scatter plot showing one of several miscellaneous
patterns 69
-------
TABLES
Number Page
1 Level 1 Biological Analyses 13
2 Level 1 Chemical Analyses 13
3 Summation of Test Results By Study and By Sample 14
4 Parameter Code Sheet for Pilot Studies, Level 1 Environ-
mental Assessments 15
5 Unit Code Sheet for Pilot Studies, Level 1 Environmental
Assessments 17
6 Sample Identification Code 18
7 Organic Categories in the MEG List 22
8 Response Ranges for Ranking of Various Biotests 24
9 Quantification of Biotest Results, Examples 25
10 Quantification of Data Sf'o
11 Example Data for Rank Correlation 40
12 Computations for Rank Correlation 40
13 "Good" Rank Correlations in Study 15 40
14 "Good" Rank Correlations in Study 19 45
15 "Good" Rank Correlations From All Studies Combined 48
16 "Good" Rank Correlations in Liquid Samples 51
17 "Good" Rank Correlations in Solids Samples 54
18 Summary of Chemical and Biological Test Combinations Where
Study Groupings Show at Least Two "Good" Correlations
Including at Least One "Good" Rank Correlation 56
19 Summary of Biological Test Combinations Where Study
Groupings Show at Least Two "Good" Correlations,
Including at Least One "Good" Rank Correlation 59
20 Patterns in Selected Scatter Plots, Chemical Tests V.
Bioassays 61
21 Patterns in Selected Scatter Plots, Bioassays V.
Bioassays 62
22 Summary of Patterns in Selected Scatter Plots, Chemical
Tests v. Bioassays 70
23 Summary of Patterns in Selected Scatter Plots,
Bioassays v. Bioassays 70
VI
-------
TABLES (continued)
Number Page
24 Biological Test Results Using Litton Quantification 73
25 AMES2 and Its Component Strains by Study 75
26 Probit Data Analysis, RAM Cytotoxicity Test 77
27 Toxicity of Filtered and Unfiltered Textile Effluents 79
28 Health Effects: Summary of Chemical and Biological
Test Results 82
29 Ecological Effects: Summary of Chemical and Biological
Test Results 83
30 Associations Between Bioassay Results and Hazard Estima-
tions Based on Chemical Analysis 85
31 Rank Correlations in Nontextile Data 85
32 Rank Correlations in Textile Data 85
33 Effect of Engineering Conditions on Chemical Concen-
trations of Effluents From FBC Miniplant (Classified
by Chemical Category) 88
34 Effect of Engineering Conditions on Chemical Concentra-
tions of Effluents From FBC Miniplant (Classified by
Effluent Stream) 89
35 Comparison of Litton Ranking Scheme and Originally
Reported Values 91
36 Linear Correlations Among the Ames Test Strains 93
37 Goal Gasifier, Cyclone Dust Sample, Strain TA1538 94
38 Fluidized-Bed Combustor, Coal Sample, Strain TA1537 94
39 Various Schemes for Ranking Textile Study Data 96
40 Rank Correlations: Comparisons of Various Ranking
Schemes for Textile Study Data 97
41 Stepwise Regression Analysis of Ranking Schemes for Textile
Study Data 97
42 Guidelines for Bioassay Testing 104
vn
-------
ACKNOWLEDGMENTS
For their generous assistance in sharing their knowledge and informa-
tion and providing guidance in implementing this study, the authors wish to
thank Judi Harris and Philip Levins of Arthur D. Little, Inc.; Beverly
Ausmus, Ken Duke, and Jim Howes of Battelle-Columbus Laboratories; Rod
Parrish of Bionomics; Jim Dorsey, Larry Johnson, Bruce Henschel, Raymond
Merrill, William Rhodes, Max Samfield, and Shabeg Sandhu of the Environmental
Protection Agency, Research Triangle Park; .David Brusick and Ross Hart of
Litton Bionetics; Gary Rawlings of Monsanto Research Corporation; Jim Campbell
and Ned Garrett of Northrop Services, Inc.; Gordon Page of Radian Corporation;
Tom Hughes, Alan Kolber, Linda Little, and Deborah Whitehurst of Research
Triangle Institute; and Vincent Simmon of Stanford Research Institute.
viii
-------
SECTION 1
INTRODUCTION
The Industrial Environmental Research Laboratory (IERL) of the Environ-
mental Protection Agency (EPA) has developed a phased methodology for asses-
sing the relative hazard of an industrial effluent and, where necessary,
developing and evaluating the best available control technology. The IERL
methodology employs both chemical and biological tests in the Level 1 screen-
ing phase of this program. Manuals published on Level 1 methods are avail-
able from IERL.1 2 Level 2 will consist of more specific chemical and/or
biological tests to identify and quantify specific hazardous components of a
sample. Level 3 will consist of two types of studies: long-term studies to
determine changes of an effluent's composition over time and studies to
evaluate the effects of applying control technologies.
To date, 19 IERL environmental assessment (EA) programs have been per-
formed for the Level 1 or screening phase. Several Level 2 efforts are or
soon will be underway. No Level 3 studies have been initiated at this time.
The EA studies have produced many individual sets of information, and a
need has been recognized to consolidate these sets of data into an easily
comprehended model. To achieve this end, a project was undertaken to com-
pile all available data on environmental assessments performed prior to the
spring of 1978. The culmination of that effort was the Compilation of Level 1
Environmental Assessment Data,3 published in October 1978. The resulting
document contains a brief description of each EA study, a schematic repre-
sentation or process flow diagram where available, and a brief summary of
the results. The data from all EAs were made uniform and consistent for
units of measure. Transformation of some results, such as simple ratios or
summations, were performed where appropriate. All such transformations and
units of measure changes are documented in this data interpretation task in
Table 5, page 17.
-------
The purposes of the project were:
1. To identify potential end uses of the EA data, such as prioriti-
zation listings, criteria for transitions from one phase or level
to the next, decision criteria for applying control technologies,
and/or usefulness to EPA's regulatory branches;
2. To develop an appropriate data model for each level;
3. To test this model using data from the three EA pilot studies;
4. To develop standardized reporting formats.
The data used in this task were drawn primarily from the compilation
document and are referred to by study number as listed in that document.
Other sources of information included unpublished (draft) documents and raw
data obtained from the biologists and chemists who performed the testing.
The data used in this task were limited to samples for which Level 1 or
equivalent methods were employed; additional chemical and physical testing
results were not considered. The data were also limited to samples for
which both bioassay and chemical results were available, although the results
for any given sample may not have included the entire complement of Level 1
chemical tests with biological assays as specified in the Level 1 manual.
There were only four Level 1 environmental assessment studies for which
both chemical tests and bioassay results were available; the remaining
studies utilized only chemical tests. The chemical testing protocol had
preceded the bioassay protocol. These four studies were:
Study number Industrial source
06 Coal gasifier
10 Fluidized-bed combustor
15 10 different industrial source types
19 23 different textile plants' effluents
Studies 06, 10, and 19 were the IERL pilot studies in which the Level 1
methodology was tested, evaluated, and improved. Study 15 was performed to
evaluate the SASS train, a Level 1 sampling apparatus for gaseous and partic-
ulate effluents, and involved several kinds of industrial emissions. These
studies are described in more detail in Section 2.
Several facts about the data set should be emphasized. With the previ-
ously mentioned qualifiers, it contains all available Level 1 data for
-------
biological and chemical testing. These data are from the first effort to
test the phased approach methodology and this study is the first attempt to
bring together and interpret all results of such a program. The data set is
small in the statistical sense, but the trends noted in this interpretation
effort should be helpful in guiding the direction of future research efforts
and studies.
The remainder of the document is divided into four sections. Section 2
describes the sources and the individual sampling points, as well as the
format for coding this information for computer use. Section 3 details the
methods of quantifying the biological and chemical testing results. Section 4
describes the analysis of the data and the results of this analysis; the
various models that have been applied to data in the phased environmental
assessment program are discussed. Recommendations based on this study are
found in Section 5. Various appendixes are supplied at the end of the
document for reference and as detailed background information; these include
•,
the original data set and transformations in the SAM/IA model.
-------
SECTION 2
DATA IDENTIFICATION
As previously noted, the raw chemical and biological data interpreted
in this task were generated on four environmental assessment studies: The
Exxon fluidized-bed combustor miniplant,4 a coal gasification process,5 a
series of textile plants,6 and a series of industrial plants.7 Although a
large volume of data was generated from these studies, only Level 1 chemical
and biological results were used in the interpretation task. The Level 1
data available for interpretation were further reduced in quantity due to
the absence of biological results from some of the samples in which chemical
results were generated. Effluent samples in which data were generated by
both analyses are included in the total data population and are identified
by study in the following sections.
2.1 STUDY 06—CHAPMAN LOW-Btu COAL GASIFIER
Study 06 was performed at a full-scale commercial Chapman low-Btu coal
gasification facility. Chemical analysis data were taken .from a letter from
Dr. Gordon Page of Radian Corporation to Dr. William Rhodes, the EPA Project
Officer. The biological data were also taken from unpublished reports sent
to Dr. Rhodes by various contractors.
Figure 1 is a schematic of the atmospheric, fixed-bed, air-blown Chapman
gasifier sampled in study 06. Coal enters the gasifier through a barrel
valve at the top of the gasification chamber, and steam and air are intro-
duced beneath a grating at the bottom of the gasifier. Combustible (product)
gases are volatilized from the coal by the steam and heated air. Ash is
removed by an ash plow from a water-sealed ash pan at the bottom of the
gasifier. The product gas is passed through a cyclone precipitator to
remove suspended particulates, then scrubbed in an aqueous scrubber-quencher
system to facilitate further cleanup. The liquid used to quench the product
gas is directed into a separator where the organic (tar) layer is separated
-------
GASIFICATION
PURIFICATION
LIQUOR SEPARATOR
COAL
DUST
BARREL
VALVE GASES
POKEHOLE LIQUOR TRAP
GASES VAPORS
LIQUOR
SEPARATOR
VAPORS AND STEAM
FUGITIVE
VAPORS
EVAPORATOR
GASES
COAL
PREPARATION
LOW BTU GAS TO
PROCESS FURNACE
COOLING WATER
STEAM
STEAM AIR GASIFIER
ASH
(WET)
COLLECTED
PARTICULATES
(DRY)
BYPRODUCTS
TARS AND OILS
TO PROCESS
FURNACES
QUENCH
LIQUOR
Figure 1. Process flow diagram for the coal gasification unit (study 06),
site A, Level 1 sampling locations.
-------
from the aqueous (liquor) layer. The tars are burned as fuels in a separate
process furnace. Most of the separator liquor is recirculated to the scrub-
bers. When an excess of aqueous separator liquor is accumulated in the
evaporator sump, it is evaporated to the atmosphere.
Figure 1 also shows the sampling points for the samples considered in
this document. These samples are:
Sample number Name Description
02 Separator tar The organic (tar) layer from the
aqueous scrubber-quencher system
of the product gas.
03 Separator liquor The aqueous (liquor) layer from
the scrubber-quencher system.
04 Gasifier ash The coal and ash removed from the
gasification chamber after the coal
was exposed to steam and hot air.
05 Cyclone dust The particulate matter removed
from the product gas by the cyclone
precipitator prior to scrubbing.
2.2 STUDY 10-FLUIDIZED-BED COMBUSTOR
Exxon's fluidized-bed combustion (FBC) miniplant unit was the subject
of study 10. The data on chemical analysis were drawn primarily from a
draft document, Comprehensive Analysis of Emissions from Exxon Fluidized-Bed
Combustion Miniplant Unit, dated September 9, 1977, by J. M. Allen, J. E.
Howes, Jr., S. E. Miller, and K. M. Duke of Battelie-Columbus Laboratories,
and EPA Project Officer D. B. Henschel.4 The data on bioassay testing were
drawn from the investigator's/contractor's internal reports to EPA, as yet
unpublished, as well as from the previously cited document.
A pressurized FBC is a relatively new method for burning sulfur-
containing coal for energy generation without producing high levels of
effluent sulfur gases. The dolomite sorbent and finely powdered coal are
fed into a hot, pressurized chamber over a grating. Air is pumped up through
the grating at a velocity great enough to "fluidize" the coal and sorbent
bed. As the coal is combusted and the sulfur gases produced, these sulfur
gases react with the sorbent to produce nonvolatile sulfides and carbon
dioxide. Although it is possible to regenerate these spent sorbents for
-------
reuse, producing H2S04, S02, or elemental sulfur as byproducts, the regenera-
tion process was not performed in this study.
The 12.5-in. pressurized FBC in this study is located in Linden, New
Jersey. Sampling and analyses were performed by Battelle-Columbus Laborator-
ies. Five runs were sampled; all burned Champion Pittsburgh Seam Coal (2
percent sulfur) with dolomite (CaC03 and MgC03) sorbent injection. As noted
above, sorbent regeneration was not used. Runs 1, 3, 4, and 5 operated at
890°-895° C; Run 2 operated at 805° C. On Runs 2, 4, and 5, a special
glass-lined organic module was used for SASS train collection. A thorough
Level 1 chemical characterization was performed for Runs 2 and 5 and, in
some cases, these values were averaged for comparison with bioassay testing,
which was done exclusively on Run 4. Run 4 chemical data were used when
available.
Figure 2 is a schematic diagram of the Exxon FBC; samples considered in
this project are noted. Other samples were taken in the study but, as
previously mentioned, data on these samples were not used because of a lack
of biological plus chemical characterization. The samples, numbered by
sampling point as in Figure 2 and on the subsequently described coding
sheets, are as follows:
Sample number
01
02
03
04
06
07
Name
Feed coal
Dolomite
Fine particulates
Coarse
particulates
Cyclone dust
Cyclone dust,
leachate
Description
The finely powdered coal fed into
the combustion chamber bed.
The sorbent material fed into the
combustion chamber bed.
The lu cyclone catch (particles >lu
and <3u in size) and the filter catch
(particles 10|j) from the
SASS train.
The particulates removed from the
effluent gas by the second cyclone
precipitator.
An aqueous leachate of the cyclone
dust. For biological tests, the
leaching was done by shaking one
-------
COOLING
WATER
CITY
WATER
^ TO
SCRUBBER
00
COAL
&
, .LIMESTONE
(02) FEED
SUPPLY
AUXILIARY
AIR
COMPRESSOR
FEED
WATER
RESERVOIR
SOLIDS
REJECT
VESSELS
NATURAL GAS
COMPRESSOR
MAIN AIR
COMPRESSOR
(1400 SCFM <5>
150PSIG)
LIQUID FUEL STORAGE
Figure 2. Exxon fluidized-bed combustion miniplant (study 10),
Level 1 sampling locations.
-------
08
09
Bed reject
material
Bed reject
material,
leachate
part solid material with four parts
deionized water for 30 min using
a high-speed shaker, then decanting
and filtering the supernatant. For
chemical tests, there were two dif-
ferent leaching procedures used.
In the first case, 1 g solid material
and 4 ml distilled water were
ultrasonically mixed for 1 hr,
the mixture was centrifuged, the
liquid decanted, another 4 ml
water added, and the process
repeated 10 times; the first and
tenth leachate liquids were
chemically analyzed. In the
second case, a 50-mL buret was
packed with solid material and
distilled water was pumped upward
through the packed column at about
1 drop/15 s; the first 10 ml of
liquid recovered from the top of
the column were chemically analyzed.
The chemical results used in this
study were from the column-leaching
procedure.
The coal ash and spent sorbent
from the fluidized bed.
See explanation for sample 07.
2.3 STUDY 15—PARTICULATES FROM VARIOUS INDUSTRIAL PROCESSES
This study was performed to evaluate the newly developed SASS train's
capacity for sampling at a variety of industrial source sites. The chemical
data were drawn from a report entitled Evaluation of Selected Methods for
Chemical and Biological Testing of Industrial Particulate Emissions by H.
Mahar of the Mitre Corporation, EPA-600/2-76-137, May 1976.7 This study
utilized only two bioassays, the rabbit alveolar macrophage (RAM) test and
the Ames/Salmonella test. The RAM data were obtained from the above-mentioned
report and from Jim Campbell and Ned Garrett of Northrop Services. The Ames
data were obtained from Dr. David Brusick and Dr. Ross Hart of Litton Bionet-
ics. Since the Ames results were obtained after the data analysis had been
performed and since no mutagenic responses were evident in this group of
samples, these data were not entered in the data set.
-------
Only particulate samples were collected in this study, and only RAM
bioassays and spark source mass spectrometry (SSMS) were performed and
entered as part of this study's data base. The particulates were sized via
SASS train collection into two categories: particulates <3u and particulates
3u-10u. The following samples were taken:
Sample number Particle size Source
01 3u-10u Steel plant—open hearth furnace
02 <3p Steel plant—open hearth furnace
03 <3u Steel plant—coke oven heater
04 3u-10u Steel plant—basic oxygen furnace
05 <3u Steel plant—basic oxygen furnace
06 3u-10u Steel plant—iron sintering
07 <3u Steel pi ant--iron sintering
08 <3p Oil-fired power plant
09 <3u Copper smelter
10 3u-10u Copper smelter
11 <3u Aluminum smelter
12 3u-10u Aluminum smelter
13 <3u Ceramics plant
14 3u-10(j Ceramics plant
15 <3u Sludge incinerator
16 3u-10|j Sludge incinerator
2.4 STUDY 19--TEXTILE PLANTS
For Study 19, biological and chemical results were taken primarily from
Source Assessment: Textile Plant Wastewater Toxics Study (preliminary
draft, December 1977) by G. D. Raw!ings of Monsanto Research Corporation,
EPA Project Officer Max Samfield.6 This document covers the sampling efforts
and analytical results for the period from January to December 1977, and is
part of an ongoing study of the aqueous effluents from 23 textile plants.
This environmental assessment study was conducted concurrently with an
effluent guidelines-priority pollutants analysis program and a study by the
American Textile Manufacturers Institute (ATMI) to identify the "best avail-
able (control) technology economically available" (BATEA) for the 23 plants.
The ATMI-BATEA study was already underway when the IERL Level 1 sampling was
begun; therefore, only 15 textile plants were sampled and tested for Level 1
chemistry. All 23 plants were sampled and tested for biological parameters.
The Level 1 samples, except those from Plants R and Y, were 8-hr compos-
ites taken from the textile plants' wastewater treatment facilities at a
location between the clarifier and the chlorine contact basin. Figure 3
shows the sampling locations for these secondary effluent samples.
10
-------
TEXTILE PLANT
RAW
WASTEWATER ^ '
1
o
0
o
0
o
o
AERATION LAGOON
SECONDARY
EFFLUENT
SAMPLE
FHFIER \ . -
CHLORINE
CONTACT
BASIN
1
EFFLUENT
Figure 3. Process flow diagram for the textile plants (study 19),
Level 1 sampling locations.
11
-------
Plant R samples were taken between the aeration lagoon and the settling
basin, and Plant Y samples were taken after the finishing pond.
The textile study's samples then fall into three groups:
Sample numbers Name Description
OA, OB, OC, OE, Textile plants' Aqueous effluent after aerator and
OF, OG, OK, OL, secondary clarifier (full Level 1 analysis,
ON, OS, OT, OU, effluent chemistry and bioassays)
0V, OW, OX
OD, OH, OJ, OM, Textile plants' Aqueous effluent after aerator and
OP, OZ secondary clarifier (Hg, Sb, and As by AA;
effluent and bioassays)
OR, OY As described in (Hg, Sb, and As by AA; and bioassays)
text above
Each of these samples was subjected to the Level 1 chemical analyses as
specified in the first edition of the Level 1 procedures manual8 and to
biological assays as specified in the IERL-RTP Procedures Manual: Level 1
Environmental Assessment Biological Tests for Pilot Studies.2 The new
edition of the biotest manual, scheduled for publication in the summer of
1979, should contain refinements of several procedures.
The types of chemical and biological analyses performed in the four
studies are identified in Tables 1 and 2, and a description of each test's
output is given. The complete battery of tests was not performed on all
samples, so Table 3 lists the individual samples and identifies the analyses
that were performed on each sample.
2.5 DESCRIPTION OF SAMPLE DATA FORMATS
To facilitate entry into the computer, reduced data from each effluent
sample were entered on Fortran coding sheets. Each sheet received data from
only one effluent sample and is therefore identified as a sample data set.
This study compared data from 51 data sets (51 effluent samples). A guide
for entering both chemical and biological data onto the coding sheet used in
this study is given in Table 4. Each of the 31 lines contains an identifier
column, followed by 7 columns of parameters. The identifier numbers, making
up the first column, reference the line of data (first two digits), the
study from which the data are taken (third and fourth digits), and the
12
-------
TABLE 1. LEVEL 1 BIOLOGICAL ANALYSES
Test
Output
HEALTH EFFECTS
Ames/Salmonella (Ames)
Rabbit Alveolar Macrophage (RAM)
WI-38 Human Lung Fibroblast* (WI-38)
Chinese Hamster Ovarian Cell (CHO)
Rodent Acute Toxicity (RAT)
ECOLOGICAL EFFECTS
Freshwater algal (FW ALGAL)
Marine algal (SW ALGAL)
Daphnia
Grass shrimp
Fathead minnow (FW FISH)
Sheepshead minnow (SW FISH)
Mutagenicity to bacterial DNA
Toxicity to mammalian lung cells
Toxicity to human lung cells
Toxicity to mammalian reproductive cells
Lethality to whole animal (mammal) and behavioral and physiological effects
Growth inhibition or stimulation to selected freshwater plants
Growth inhibition or stimulation to selected marine plants
Lethality to a freshwater invertebrate animal
Lethality to a marine invertebrate animal
Lethality to a freshwater fish
Lethality to a marine fish
*The WI-38 Human Lung Fibroblast Test is an optional test in the Level 1 testing protocol. At the present time, the RAM test is used
for paniculate samples and the CHO test for liquids and extracts. Note that FW refers to freshwater test and SW refers to saltwater
test.
TABLE 2. LEVEL 1 CHEMICAL ANALYSES
Tert
Output
Spark Source Ma» Spectrometry (SSMS)
Atomic Absorption Spectrometry (AAS)
Infrared Spectrophotometry (IR)
Low Resolution Man Spectrometry (LRMS)
Liquid Chromatographic Separation (LC)
Gravimetric Analysis (GRAV)
GasChromatography-Flame-lonization
Detection (GC-FID) and Total
Chromatographable Organics Scheme
(TCO)
Ion Chromatography (1C)
Field Gas Chromatography
Test kits, wet-chemical methods
All elements except H and He
Hg, Sb, and As
Organic functional groups
Organic functional groups and some specific compounds
Organic components in 7 separate fractions from the least polar components
in fraction 1 to the most polar in fraction 7
Weight of nonvolatile organic components
Weight of volatile organic components
Identification and quantitation of inorganic anions
Identification and quantitation of very volatile components of gaseous effluent
streams
Identification and quantitation of selected inorganic anions in liquid samples
13
-------
TABLE 3. SUMMATION OF TEST RESULTS BY STUDY AND BY SAMPLE
Analytical test-biological
Ames
RAM
CHO
RAT
FW algal
SW algal
Daphnia
Grass shrimp
FWfish
SWfish
Analytical test-chemical
SSMS
IR
LRMS
LC
GRAV
GC-FID
Anions
Aqueous analysis
Coal gasifier, study 06
o
? 1 1 1
ills
*» »» tZ o
CD •
CO 00 CD CJ
V V V V
V V V V
V V V
V
V
V
V
x/
V
V V x/ v7
V V V V
x/ x/ x/
x/ V V
x/ V x/
V
i
1
u.
V
V
V
x/
x/
V
V
V
Dolomite
Fine particulate
x/ V
V
V X
V
v'
x/
V
V x7
V V
0)
"3
e
CO
§
o
o
V
V
x7
V
x/
V
V
V
v'
FBC,
tt
•o
_o
"3
U
V
V
X/
N/
V
V
V
x/
V
study 10
Cyclone dust
leachate
V
x/
V
V
V
x/
x/
V
f If
S"° ^> 5
ea£
V
V V
V
x/
x/
V x/
^J
\/
V
V
V
x/ V
Mitre Corporation, Textile plants,
study 15 study 19
All samples 1 5 plants*
V V
V
V
V
V
V
x/
x/
x/
V
v'
x/
8 plants*
V
v'
•Plants A, B, C, E, F, G. K, L, N, S, T, U, V, W, X.
tplants 0, H, J, M, P, R, Y, Z.
Note: The WI-38 Human Lung Fibroblast Test is an optional test in Level 1 testing protocol. At the present time, the RAM test is used for paniculate samples and the CHO test
for liquids and extracts.
-------
TABLE 4. PARAMETER CODE SHEET FOR PILOT STUDIES,
LEVEL 1 ENVIRONMENTAL ASSESSMENTS
en
010000
020000
030000
040000
050000
060000
070000
080000
090000
100000
110000
120000
130000
140000
150000
160000
170000
180000
190000
200000
210000
220000
230000
240000
250000
260000
270000
280000
290000
300000
310000
1
SAMPLE ID
VIABILITY INDEX
@ 1000 pg/mL
SOIL
MICROCOSM
DAPHNIA (48)
RAT-% INCREASE IN WT.
U
Pt
Lu
Gd
Ba
Cd
Zr
Ge
Mn
a
f
MEG4
MEG11
MEG18
MEG25
LC6
GC11
°2
CONDUCTIVITY
NH3
TA1535-
TA100+
AMES2
FW FISH
2
RAM AT 1000 (Jg/mL
ATP-(fg/cell)
@1000jjg/mL
FW ALGAL
SW FISH (24)
Th
Ir
Yb
Eu
Ct
Ag
Y
Ga
Cr
S
B
MEGS
MEG12
MEG19
MEG26
S~
LC7
GC12
CO2
Ph
DISSOLVED SOLIDS
CN
TA1535+
TA1538-
RAM
DAPHNIA
3
PROBIT-VIAB.
SW ALGAL
SW FISH (48)
Bi
Os
Tm
Sm
I
Pd
Sr
Zn
V
P
Be
MEG6
MEG13
MEG20
d
LC1
LC8
GC13
S02
ACIDITY
SUSPENDED SOLIDS
SCN"
TA1537-
TA1S38+
WI-38
4
WI-38 AT 600 jiL/mL
PROBIT-ATP
FW FISH (24)
SW FISH (96)
Pb
Re
Er
Nd
Te
Rh
Rb
Cu
Ti
Si
Li
MEG7
MEG14
MEG21
F
LC2
GC7
Hg-AA
CO
ALKALINITY
cr
TA1537+
CHO
5
AMES1
FW FISH (48)
GRASS
SHRIMP (24)
TI
W
Ho
Pr
Sb
Ru
Br
Ni
Sc
Al
MEG1
MEG8
MEG15
MEG22
S04=
LC3
GC8
Sb-AA
H2SCOS
BOD
803=
TA98
RODENT
6
CHO, MEAN X SURVIVAL
RATLDso
FW FISH (96)
GRASS
SHRIMP (48)
Hg
Ta
Dy
Ce
Sn
Mo
S«
Co
Ca
Mg
MEG2
MEG9
MEG16
MEG23
LC4
GC9
At-AA
HCN
COD
H2S
TA98+
FW ALGAL
7
OTHER
RAT TEST
DAPHNIA (24)
•V^^-««H»
GRASS
SHRIMP (96)
Au
Hf
Tb
La
In
Nb
A.
F*
K
Na
MEG3
MEG10
MEG17
MEG24
"&
GC10
NOX
•=2
DO
ORGANICS
TA100-
GRASS SHRIMP
-------
sample number (fifth and sixth digits). For example, 031001 refers to data
line 3 from study 10, sample 01.
The key to study identifiers is:
06: Coal gasification
10: Exxon's FBC mini pi ant
15: Ten industrial sources, SASS evaluation
19: Textile plant effluents.
Each piece of data for a study is identified by its sample code number
and column number. For example, 110605-column 2 identifies the amount of
silver found in the Radian Cyclone Dust sample. To further identify the
data, the units for each parameter are noted in Table 5. Looking at line
11, column 2, (jg/g is the unit to which all data on silver content have been
reduced.
The sample identifier (sample ID) is found on line 1, column 2 of each
coding sheet. It is limited to nine characters. Table 6 provides a descrip-
tion of the coded sample identifiers. Appendix A lists the entire data base
by individual sample data sets.
It should be noted that other, more thorough data formats for the IERL
Environmental Assessment Program are being developed. Mr. Gary Johnson of
the IERL Special Studies Staff is developing detailed data entry forms that
are compatible with the EPA UNIVAC 1100 computer system. This new data
analysis and storage system, the Environmental Assessment Data System (EADS),
will contain much supportive information; e.g., plant conditions at the time
of sampling, atmospheric conditions, date, run number, personnel identifica-
tion, and documentation of any anomalies. At the time of the preparation
and execution of this study, the EPA data systems were completed and opera-
tional only for fine particulates.9 10 As part of this task, the authors
cooperated with Mr. Johnson in the design of input for bioassay data entry
forms and shared information on other sampling/analytical types that were
available from this study.
16
-------
TABLE 5. UNIT CODE SHEET FOR PILOT STUDIES.
LEVEL 1 ENVIRONMENTAL ASSESSMENTS
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
• MQ/mL-
X OF VIA. & ATP
AS A % OF CONTROLS
Jig/mL »
tOFsWOF
POSITIVE PARAMETERS
«-%. 0/100 g -
f g/g •
cg/g-
cs/g
pg/g •
14 DAY
PROBIT 1/LDgQ
ECsflX. 96HH.-
X OF VIA. a ATP
AS A % OF CONTROLS
PROBIT 1/LD.a,
MEAN % SURVIVAL
AT200nL/roL
• g/kg-
LCjoX
OCCURRENCES
EC5o%^
pg/g -
pg/g -
(j»/g -
ig/g -
wi/g -
pg/g -
^g/g -
mg/g-
mg/g-
mg/g-
mg/g-
WT.
M«/9 •
Aig/g •
% •
-»mg/g-
M9/9
-pprn-
*yumhoj. e 25°
-1-14 »«-
g/m3-
MAXIMUM VALUE OF REVERTANTS PER PLATE/REVERTANTS IN CONTROL
FOR ALL AMES' STRAINS W & W/O METABOLIC ACTIVATION *
LITTON/BRUSICK'S INTERPRETATION. WITH NO DETECTABLE TOXICITY - 1. LOW TOXICITY - 2. MODERATE TOXICITY - 3,
HIGH TOXICITY - 4
-------
TABLE 6. SAMPLE IDENTIFICATION CODE
Sample code no.
010602
010603
010604
010605
011001
011002
011003
011004
011006
011007
011008
011009
011501
011502
011503
011504
011505
011506
011507
011508
011509
011510
011511
011512
011513
011514
011515
011516
01190A
01190B
01190C
01190D
01190E
01190F
01190G
01190H
011901
01190J
01190K
01190L
01190M
01190N
011900
01190P
01190Q
01190R
01190S
01190T
01190U
01190V
01190W
01190X
01190Y
01190Z
Sample name
SEPARATAR
SEPARALIQ
GASI FLASH
CYCLODUST
FEEDCOAL
DOLOMITE
FINEPART
CORSEPART
CYCLODUST
CYCLOLEACH
BEDREJECT
BED LEACH
STEEL3MIC
STEEL1MIC
COKE1MICR
STEEL03MI
STEEL01MI
1RUN3MICR
1RUN1MICR
01L1MICRO
COPPER1MI
COPPER3MI
ALSM1MICR
ALSM3MICR
CERAM1MIC
CERAM3MIC
SLUDGE1MI
SLUDGE3MI
PLANT A
PLANT B
PLANT C
PLANT D
PLANT E
PLANT F
PLANT G
PLANT H
PLANT 1
PLANT J
PLANT K
PLANT L
PLANT M
PLANT N
PLANT 0
PLANT P
PLANT Q
PLANT R
PLANTS
PLANT T
PLANT U
PLANT V
PLANT W
PLANT X
PLANT Y
PLANT Z
Description
Coal gasif ier separator tar
Coal qasifier separator liquor
Coal gasifier (unground) ash
Coal gasifier cyclone dust
FBC feed coal
FBC dolomite
FBC fine paniculate* (1jj cyclone and filter)
FBC coarse particulates (3/i and 10ju cyclone)
FBC cyclone dust
FBC cyclone dust leachate
FBC bed reject
FBC bed reject leachate
Steel plant-open hearth furnace (3^-10^ particulates)
Steel plant-open hearth furnace (<3jj particulates)
Steel plant-coke oven heater (<3n particulates)
Steel plant-basic oxygen furnace (VIO/j particulates)
Steel plant-basic oxygen furnace {<3u particulates)
Steel plant-iron sintering (3/1-10/1 particulates)
Steel plant-iron sintering (<3*i particulates)
Oil-fired power plant (<3«i particulates)
Copper smelting (<3*j particulates)
Copper smelting (3M-10M particulates)
Aluminum smelting (<3*t particulates)
Aluminum smelting (3^-10*1 particulates)
Ceramics plant (
-------
SECTION 3
QUANTIFICATION OF DATA
The effluents and feedstocks sampled during the four EA pilot studies
were liquids, solids, and slurries. This variation resulted in analytical
results being reported in different units from study to study and sample to
sample. For example, analytical results of spark source mass spectrometry
(SSMS) generated on solid samples were generally reported in micrograms (ug)
of element per gram (g) of sample. Conversely, SSMS results on liquid
samples were reported in micrograms of element per liter (L) of sample.
This variability in the data units from different analytical tests was resolved
by choosing suitable units in which each test should be reported and then
converting data across all studies into uniform units. Data in consistent
units could then be entered into the program as absolute numbers for compari-
son. Data thus entered include results for SSMS, anion, gas chromatography
(GC), atomic absorption (AA), and aqueous analyses. Results of organic
analysis (liquid chromatography [LC] fraction weights, infrared spectropho-
tometry [IR], and low resolution mass spectrometry [LRMS]) were summarized
by use of a method explained in the subsection of this section on organic
analysis.
Bioassay results were quantified by use of the actual LC50 and LD50
values as well as by a grouping model designed by Litton Bionetics. Compari-
son of the data quantified by these two methods provided a quality assurance
(QA) check to insure that the grouped data truly represented the actual
results.
3.1 CHEMICAL RESULTS—INORGANIC
3.1.1 Spark Source Mass Spectrometry (U to Li)
In a given sample, data for any element present at a level above the
minimum detection limit were converted, where necessary, to micrograms per
gram of effluent sample (ppm by weight). Each elemental concentration less
than the minimum detection limit was recorded as a value equal to one-half
19
-------
the minimum detection limit; these minimum detection limit data were further
coded by the use of a negative sign preceding each value on the data page.
The absolute values of all numbers entered in the data set were used for
computations in this study, but in future studies with larger data bases,
this coding would allow other options, such as eliminating minimum detection
limit values or setting these values equal to zero.
3.1.2 Anion Analysis (Cl". F". S04. NOg. COa. S")
Reported values were converted to weight percent (g/100 g), if necessary,
and entered for each sample in which data were generated.
3.1.3 Atomic Absorption (As. Sb. Hg)
Absolute values were converted to micrograms per gram, if necessary,
and entered for each sample in which data were generated.
3.1.4 Chemiluminescence (NO )
,*\
No data were available for samples in this data set.
3.1.5 ORSAT (02)
No data were available for samples in this data set.
3.1.6 Gas Chromatography (C02. S02. CO. H2S-COa, HCN, F2, C1g)
No data were available for samples in this data set.
3.1.7 Aqueous Analysis
Data were available only on the textile study. Absolute values were
entered in grams per cubic meter (ppm by weight) for each sample.
3.2 CHEMICAL RESULTS—ORGANIC
Liquid chromatographic fraction gravimetric weights were entered directly
into the data set with only unit changes, where needed. Volatile (TCO)
weights were not generated in these studies. Therefore, any subsequent
correlations or other data manipulations consider only the nonvolatile
(gravimetric [GRAV]) portions of these samples. The quantification of
organic results from written form (i.e., alcohols, -CH3) to multimedia
environmental goals (MEG) categories (categories of similar organic com-
pounds such as amines and benzene derivatives) was more complex. All organic
20
-------
analyses results (LC, IR, and LRMS) were used to determine the weights of
MEG categories found in each sample. For reference, Table 7 gives the
chemical categories found in each MEG category.
Three separate methods were used for the quantification of organic
analytical results into MEG categories, depending on the amount of available
information:
I. In cases where only IR and LC/GRAV results were available, the
following procedure was used:
A. List categories that could possibly be present in a given LC
fraction.
B.. Examine IR report for that LC fraction. Examine LC/GRAV report
for that LC fraction.
C. Assign a weight equal to one-half the GRAV weight to each com-
pound indicated as possibly present by IR.
D. Assign a weight equal to one-tenth the GRAV weight to each com-
pound indicated as not present by IR.
E. Total the assigned weights by MEG category.
II. In cases where LRMS and LC/GRAV results are available, this procedure
was used:
A. List the major categories and their reported intensities from
the LRMS report.
B. Total the intensity values.
C. Estimate weights as fractional parts of the total LC/GRAV weight:
Intensity of category X Total GRAV weight _ Estimated weight
Z Intensities of all of fraction of category X
categories in fraction
D. Total the estimated category weights by MEG category.
III. The following procedure was applied in cases where there was a very
small amount of organic material in a given fraction and, as a result,
no LRMS was run, but LRMS spectra were available on adjacent fractions:
A. Compare the IR spectra of adjacent fractions with the IR spectrum
of the fraction under consideration.
B. Match with the most similar fraction.
21
-------
TABLE 7. ORGANIC CATEGORIES IN THE MEG LIST
1 Aliphatic hydrocarbons
2 Halogenated aliphatic hydrocarbons
3 Ethers
4 Halogenated ethers
5 Alcohols
6 Glycolj, epoxides
7 Aldehydes, ketones
8 Carboxylic acids and derivatives
9 Nitriles
10 Amines
11 Azo compounds, hydrazine, and derivatives
12 NKrosamines
13 Mercaptans, sulfides and bisulfides
14 Sulfonic acids, sulfoxides
15 Benzene, substituted benzene hydrocarbons
16 Halogenated aromatic hydrocarbons
17 Aromatic nitro compounds
18 Phenols
19 Halophenols
20 Nitrophenols
21 Fused aromatic hydrocarbons
22 Fused nonalternant polycyclic hydrocarbons
23 Heterocyclic nitrogen compounds
24 Heterocyclic oxygen compounds
25 Heterocyclic sulfur compounds
26 Organometallics
22
-------
C. Assign the same weight distribution (as determined in II, above)
to the fraction under consideration as was calculated for the most
similar adjacent fraction.
The revised Level 1 procedures manual (Chapter 9)1 gives detailed
instructions and examples for the above calculations. This method of sum-
marizing the organic data was originally suggested by Dr. Judi Harris and
Dr. Philip Levins of Arthur D. Little, Inc.
3.3 BIOASSAY RESULTS
Bioassays were quantified by two methods. One method involved an
assignment of "nondetectable," "low," "moderate," or "high" to each biological
response. This method, proposed by Dr. David Brusick of Litton Bionetics,11
is referred to hereafter as the Litton quantification scheme. Levels of
response are assigned to bioassays on the basis of the concentration at
which a biological system is affected by a sample relative to the maximum
applicable dose (MAD) of sample possible for that system. The levels of
toxicity assigned by Litton Bionetics are: (1) no detectable toxicity if no
significant response occurs at the MAD; (2) low toxicity if the response
occurs at a concentration between the MAD and 1/10 MAD; (3) moderate toxicity
if the response occurs at a concentration between 1/10 MAD and 1/100 MAD;
and (4) high toxicity if the response occurs at a concentration less than
1/100 MAD.
The MAD concept must be applied differently to each biological test.
Table 8 gives response ranges and maximum applicable doses for several of
the Level 1 bioassays. Table 9 identifies a variety of possible reporting
units and illustrates how they can be quantified into toxicity levels. In
the Ames test, the criteria suggested by Brusick and used herein for a
positive (mutagenic) response is a ratio of test revertants to control
revertants greater than 3 for strains TA1535, TA1537, or TA1538, or a ratio
greater than 2 for strains TA98 or TA100, plus an increasing dose-response
relationship over three successive dose ranges.
The second method used to quantify the bioassay results was through the
use of the originally reported results on many of the parameters measured in
each bioassay. For example, the RAM test system was tested for four major
parameters: cell count, viability, protein, and ATP at several dosage
23
-------
TABLE 8. RESPONSE RANGES FOR RANKING OF VARIOUS BIOTESTS*
ro
Response ranges
Assay
Health Tests
Ames
RAM, CHO.WI-38
Rodent
Aquatic
Algae
Fish
Invertebrate
Activity measured
Mutagenesis
Lethality (LC50)
Lethality (LDgg)
Growth inhibition ECgrj
Lethality LC$Q
Lethality LCsg
MAD High
5 mg/plate or <0.05mgor
500 ML/plate <5^L
1 ,000 Mg/m L or < 1 0 Mg or
600ML/mL <6ML
lOg/kgor <0.1
IQmL/kg
100 mg/L <1
100 mg/L <1
100 mg/L <1
Moderate
0.05-0.5 mg or
5-50 ML
10-100 M9 or
6-60 ML
0.1-1.0
1-10
1-10
1-10
Low
0.5-5 mg or
50-500 ML
100-1, 000 Mg or
60-600 ML
1-10
10-100
10-100
10-100
Not detectable
ND at>5mg or
NDat>500
LC5fj> 1,000^9 or
LCso>600ML
LD50>10
EC50>100
LC50>100
LC5u>100
MAO = Maximum applicable dose (technical limitations).
LD5Q = Calculated dosage expected to kill 50% of population.
= Calculated concentration expected to kill 50% of population.
= Calculated concentration expected to produce effect in 50% of population.
ND = Not detectable.
'Subsequent to the completion of this work, the response ranges for the aquatic tests were redefined based on recommendations from EPA Geologists.
The new ranges are:
MAD High Moderate Low Not detectable
Algae
Fish
Invertebrate
1,000 mg/L or
100%
1 ,000 mg/L or
100%
1,000 mg/L or
100%
<20% or
<200mg
<20%or
<200 mg
<20%or
<200 mg
20-75% or
200-750 mg
20-75% or
200-750 mg
20-75% or
200-750 mg
75-100% or
750-1,000 mg
75-100% or
750-1, 000 mg
75-100% or
750-1,000 mg
EC5Q>100%or
ECso>1,OOOmg
LC5Q>100%or
LC50>1.000mg
LCso > 100% or
LC5Q> 1,000 mg
-------
TABLE 9. QUANTIFICATION OF BIOTEST RESULTS, EXAMPLES
Biotest
AMES
RAM
Viability Index
ATP
WI38
Viability Index
ATP
RAT
Viability Index
Other Effects
FW ALGAL
OAPHNIA
FW FISH
SW ALGAL
GRASS
SHRIMP
SW FISH
Reported
units
Max. # Revertants, Test
# Control Revertants
EC.0% (24 hr)
EC20% (24 hr)
As % of control
at maximum dose
As % of control
at maximum dose
"so- o'kg
Distended caecum,
weight loss in 5 9
EC20% (14 day)
EC50% (48 hr)
LCSQ% (96 hr)
EC5fl% (96 hr)
IC50% (96 hr)
LC50% (96 hr)
Reported
value(s)*
1.9
13.3
3.8
56.1
11.2
>10
6 Effects
2
0
48.8
2.3
26.3
47.5
Assigned
toxicity
level*
N
M
L
N
M
H
L
M
L
L
FW = Freshwater.
SW = Saltwater (marine).
N = No detectable toxicity.
L = Low toxicity rating.
M = Moderate toxicity rating.
H = High toxicity rating.
"Data in these columns are sample data from Plant N.
25
-------
levels. After discussion with biologists familiar with the system, it was
decided that five values might be of special interest in this data set.
These values were:
1. Viability index at the MAD of 1,000 \ig/mL:
,,• u-1-4. w u-i-4. vx # of cells in test culture
Viability = Viability x # of cells in control culture '
index (as percent-
age of control)
2. ATP at 1,000 ug/mL.
3. The average of 1 and 2, above.
4. The LD50 value, as determined from probit analysis of the dose-
response relationship between pollutant dosage and viability.
Probit analysis was performed via the Statistical Analysis System
(SAS). The methods for probit analysis are detailed in reference
12. The reciprocal of the LD50 value was entered (1/LD50) so that
toxic responses would exhibit a positive slope and a positive
correlation coefficient in the subsequent statistical analysis.
Probit analysis is discussed in more detail in the next section of
this document.
5. 1/LDso when a dose-response relationship between pollutant level
and ATP level was considered as in 4.
Efforts were directed at identifying the presence of any parameter that
would correlate with other bioassays or chemical assays. Quantification of
the individual biotests, using the reported or raw data, is discussed in the
following sections.
3.3.1 Microbial Mutagenicity (Ames)
Two different methods of quantifying Ames test data were used and both
were based on the maximum mutagem'c ratio, herein defined as the ratio of
the maximum average number of revertant colonies in any experimental dose to
the average number of spontaneous revertants in control cultures. The
maximum mutagem'c ratio was determined both with S-9 (rat liver microsomal)
activation and without S-9 activation. Five different strains of Salmonella
typhimurium were used in the test series; therefore 10 ratios were generated.
A second method of interpreting these results was to group results in a
manner slightly different from the Litton grouping noted above. A maximum
26
-------
mutagenic ratio less than or equal to 3 was given a value of 1 in the data
set. A maximum mutagenic ratio greater than 3 but less than or equal to 4
was given a value of 2. A ratio greater than 4 but less than 5 was coded as
3. All ratio values equal to or greater than 5 were quantified as 4.
3.3.2 Rabbit Alveolar Macrophage (RAM)
The quantification methods for this test were described as an example
in Section 3.3.
3.3.3 WI-38 Human Lung Fibroblast (WI-38)
The parameters measured were cell count, viability, protein content,
and ATP content as in the RAM test. Because the WI-38 bioassays were per-
formed on so few samples, only one value was entered in the data set: the
average of (1) the viability index at the MAD (600 pL/mL) and (2) the ATP
content at the MAD. In both the RAM and the WI-38 systems, it was observed
that ATP was a more sensitive parameter than protein; that is, ATP levels
showed a greater amplitude of response than protein levels and ATP responses
were generally noticed at lower dosage levels.
3.3.4 Chinese Hamster Ovarian Cell Clonal Assay (CHO)
The average of replicates of the percent survival figure (number of
live cells as a percentage of the total number of cells) at the MAD (200
uL/mL) was entered into the data set. iv
3.3.5 Rodent Acute Toxicity (RAT)
Three means of quantifying the data from this bioassay were employed.
First, the LD50 value was entered, but because no lethality was observed in
any samples at the maximum dosage of 10 g of sample (by gavage) per kilogram
of rodent body weight, this did not seem to be a good, discriminating quanti-
fication method. A close examination of the raw data (the investigators'
laboratory reports) showed many atypical behavioral and physiological responses
observed in experimental rats fed the various effluent samples. These
atypical responses included reduced activity immediately after dosing, soft
stools, a dark red material around the nose or eyes, eye irritation, rough
appearance of the ventricles of the heart, abnormal breathing sounds after
dosing, abnormal penile discharge, and hair loss in the inguinal region and
the area around the eye.
27
-------
The second parameter entered in the data set for the RAT test was the
'sum of all abnormal responses noted by the investigator for all 10 rats in
the test group; it was possible for this second parameter to exceed 10 if
more than one atypical response was noted in a single test animal.
The third means of quantifying a sample's effect on a rat was based on
body weight. An expected weight gain was observed in the control group,
indicative of the normal rodent growth pattern of continuous weight gain.
For each group of 10 experimental animals (five females, five males), the
percent gain in body weight was calculated:
Final Initial Final _ Initial
weight mate* " weight males weight female» weight female!
Initial weight male** Initial weight femalet v i/vi . «_,____- • ^ ,
_ : ——— A 100 • average percent weight gain.
3.3.6 Freshwater Algal Test (FW Algal)
The original investigators' reported EC20 value (reported as percentage
of effluent sample in the dilution water media) was encoded in the data set.
The effect measured was inhibition of growth, as measured by chlorophyll
content of the experimental culture compared to chlorophyll content of the
control. The EC2o was calculated from chlorophyll measurements taken on the
fourteenth day. A standard ECso value was not used because the very low
e-.
toxicity of many samples from the textile study would have prohibited discrim-
ination between samples using the EC50; that is, all samples with less than
50 percent growth inhibition would appear to be the same numerically.
3.3.7 Marine Algal Test (SW Algal)
Again, growth inhibition was the effect measured. The EC50 value after
4 days of testing was the value entered into the data set.
3.3.8 Daphm'a
Lethality, or apparent death as measured by immobility after gentle
prodding, was the effect of interest. As reported by the original investi-
gator, the EC50 was entered into the data set as percentage of effluent
sample concentration in diluent aqueous medium after 24 hr and again after
48 hr of testing.
28
-------
3.3.9 Grass Shrimp
The original investigators' reported LC50 values (lethality after 24,
48, and 96 hr) were entered into the data set as percentages of effluent
sample concentration in diluent aqueous medium.
3.3.10 Fathead Minnow Test (FW Fish)
The original investigators' reported LC50 values (after 24, 48, and
96 hr) were entered into the data set as percentages of effluent.
3.3.11 Sheepshead Minnow Test (SW Fish)
The original investigators' reported LC50 values (after 24, 48, and 96
hr) were entered into the data set as percentages of effluent.
Table 10 lists the original results from each study used in this inter-
pretation, the quantification method used to reduce all the results into
equivalent units suitable for comparisons, and the coded name assigned to
each value for computer encoding and entry.
29
-------
TABLE 10. QUANTIFICATION OF DATA
Study
co
O
Original
results
RAM Cytotoxicity*
6 Cyclone dust
Gasifier ash (unground)
10 Fine participates, cyclone dust, bed
reject, coarse participates, coal,
dolomite
6 Separator tar
Separator liquor
10 Cyclone dust leachate
Bed reject leachate
19 Plants L, M, R
6 Cyclone dust
Gasifier ash (unground)
10 Fine participates, cyclone dust, bed
reject, coarse particulates, coal.
dolomite
6 Separator tar
Separator liquor
Quantification
Computer-coded
final results
Northrop Data: Viability Index, mean of 6 repetitions @ 1,000 M9/mL
ATP as % control, mean of 6 repetitions
(VIABIND + ATP) ^2
Northrop Data: Viability Index,mean of 6 repetitions @ 1,000 jig/mL
ATP as % control, mean of 6 repetitions
(VIABIND+ ATPK 2
ADL Data: Viability Index, mean of 2 repetitions @ 1.MO
ATP as % control, mean of 2 repetitions
(VIABAND + ATPK2
ADL Data: Viability Index, mean of 2 repetitions @ 600
ATP as % control, mean of 2 repetitions
(VIABAND + ATPK2
Northrop Data: Viability Index, mean of 5 repetitions @ 600 ML/mL
ATP as % control, mean of 5 repetitions
(VIABAND+ATP)42
Northrop Data: Viability Index, mean of 3 repetitions 9 600
ATP as % control, mean of 3 repetitions
(VIABIND + ATP) ^2
Northrop Data: Viability Index, mean of 6 repetitions @ 1.
ATP as % control, mean of 6 repetitions
(VIABINO + ATP) * 2
Northrop Data: Viability Index, mean of 6 repetitions @ 1.000
ATP as % control, mean of 6 repetitions
(VIABIND + ATP) ^2
ADL Data: Viability Index, mean of 2 repetitions @ 1,000
ATP as % control, mean of 2 repetitions
(VIABIND+ ATPH 2
ADL Data: Viability Index, mean of 2 repetitions @ 600 /A/mL
ATP as % control, mean of 2 repetitions
VIABINO
ATP
RAM 1000
VIABINO
ATP
RAM
VIABAND
ATP
RAM 1000
VIABIND
ATP
RAM 1000
VIABIND
ATP
RAM 1000
VIABINO
ATP
RAM 1000
VIABIND
ATP
RAM 1000
VIABINO
ATP
RAM 1000
VIABINO
ATP
RAM 1000
VIABINO
ATP
RAM 1000
•The RAM 1000 parameter includes data on leachates (tested at 600 nil ml) and on solids (tested at 1,000 ig/mL).
(continued)
-------
TABLE 10 (continued)
Original
Study results
RAM Cytotoxicity* (con.)
10 Cyclone dust leachate
Bed reject leachate
19 Plants L, M, R
19 Plant N
15 Viability Index @ 1,000 ng/mL
WI-38
19 WI-38 @ 600 nUml
Plants L, M, N, R
CHO
19 Mean percent survival
Plants L, M, N, R, D, J, H
RAM
6,10,19 Viability Index, mean at variable doses
6,10,19 ATP, mean at variable doses
AMES
6,10,19 Control and revertant colonies at variable
doses across several strains with and
without metabolic activation (S-9)
RAT
6,19 Rodent Acute Toxicity (RAT) LD50's
6,19 Behavioral and physiological determinations
Quantification
Northrop Data: Viability Index, mean of 5 repetitions @ 600 nL/mL
ATP as % control, mean of 5 repetitions
(VIABIND + ATPK2
Northrop Data: Viability Index, mean of 3 repetitions @ 600 /A/mL =
ATP as % control, mean of 3 repetitions =
(VIABIND + ATPK2
Same as Plants L, M, R but only 2 repetitions =
Entered as reported
[Viability Index (mean of 3 repetitions @ 600 ^UmL) plus ATP % (mean of
3 repetitions)] divided by 2
Mean of 3 repetitions
Probit analysis by the Statistical Analysis System (SAS) program.
ECgg values were calculated and their reciprocal values entered
Probit analysis by the Statistical Analysis System (SAS) program.
ECgQ values were calculated and their reciprocal values entered
The largest ratio of revertant to control colonies across all strains
reported! Ratios were grouped such that ratio values <3= 1, values
between 3 and 4 = 2, values between 4 and 5 = 3, and values > 5 = 4.
t
LDgg values entered as reported. If no response @ the maximum dose, =
that dose was entered
Sum of atypical responses, e.g., hairless = 1 =
lethargy = 1
Total = 2 entered
Computer-coded
final results
VIABINO
ATP
RAM 1000
VIABINO
ATP
RAM 1000
VIABINO
Wl 38600
CHOMPS
PROB VI
PROB VI
AMES1
RAT LD50
OTHRAT
(continued)
-------
TABLE 10 (continued)
rs>
Study
6,10,19
6
6,10
6,10,19
6.10
19
6,10,19
6,10,19
6,19
Original
resui-
AL&4.
Liquid sample only -"::::-«:- p- algal
reported values as \ t~ .*" ".at caused
an £€29; Parameter ^AK:.-K was inhibi-
tion of growth as m«ss:."&: : v chlorophyll
content after 14 dav:.
Separator liquor-saiT*rK &>ga! reported
value as % effluent tr-r ::iusfca an ECsg.
Parameter measured «*: T tution to
growth as measured :•• :=• -.•'Dphyll con-
tent after 4 days
FRESHWATH? £tSH
Liquid samples oniv- H- re 48-hr LCso
values reported
Liquid samples oni, -:--:--• . I.^Q
values reported
DAPhlU
Liquid samples onr, -'.i- i;- £B-hr ECjrj
reported; immofa: i-. ->««::. TI^
48-hr ECso reporac
SALT* ATE = J2SH
Liquid samples or;-, -li- t-i :. and 96-hr
LCsg values repc
-------
TABLE 10 (continued)
Study
Original
results
Computer-coded
Quantification final results
SSMS
6,10,15,19
Weight of element per unit of sample
MEG CATEGORIES
Converted all results to ^g/g (g/m^ = mg/L =
Concentrations of organics by MEG category Entered as reported by the contractor
10
IRandLC/GRAVDATA
GJ
CO
19
IR, LRMS.and LC/GRAV DATA
Procedures^ specified in Chapter 9 of the IERL-RTP revised Level 1
procedures manual,were followed:
For each LC fraction:
(1) The IR and LC/GRAV reports were examined
(2) IR functional groups and/or compounds were assigned to
MEG chemical categories
(3) Each functional group and/or compound found by IR was given
50% of the LC/GRAV weight (unless only one compound was
found, in which case it was given 100% of the LC/GRAV weight)
(4) From a table in the procedures manual, a list of possible
compounds eluting in each LC fraction was examined
(5) Possible compounds not found by IR analysis were assigned 10%
of the LC/GRAV weight
This procedure was repeated for each LC fraction, then the MEG
category weights were totalled across fractions and entered as
sample totals in the Fortran format
Procedures, as specified in Chapter 9 of the revised IER L-RTP
Level 1 procedures manual.were followed:
For each LC fraction with LRMSdata:
(1) The LRMS and LC/GRAV reports were examined
(2) Intensity values for major categories were totalled
(3) Weights were estimated as fractional parts of the total LC/G RAV
weight:
LNUto LMLI*
LNMEG 1-LNMEG 26 (mg/g)*
(see Table 7)
LNMEG 1-LNMEG 26 (mg/g)*
(see Table 7)
Intensity of category/compound X
I Intensities of all
category/compounds in fraction
Total
weight -
of
fraction
*,
Estimated
weight of
category/
compound X
*ln showing an element or compound abbreviation, the initial letters "LN" indicate that a logarithmic transformation has been performed on the data.
For ease of identification, "LN" is omitted from the abbreviation after this table.
(continued)
-------
TABLE 10 (continued)
Study
Original
Quantification
Computer-coded
final resultst
19
10.19
10
19
10
6,10,19
10,19
10,19
IR, LRMS, and LC/GRAV DATA (con.)
ANION ANALYSIS (in various units)
LC GRAVIMETRIC (in M9/9 weight* in mg)
LC GRAVIMETRIC
(in n9/g weights per fraction)
GC7-GC13(inM9/g)
Hg by AA, reported in various units
Sb by AA, As by AA
AQUEOUS ANALYSIS of liquid samples
For each LC fraction with only IR data:
(1) IR spectra of adjacent LC fractions that received LRMS analysis
were compared to the spectrum of the fraction under consideration.
(2) The fraction under consideration was matched with the most
similar adjacent fraction.
(3) The assignment of category/compound weight distribution of the
fraction under consideration was made in the same manner as above
(using LRMS data).
These procedures were repeated for each LC fraction, then the MEG
category weights were totalled across fractions and entered as sample
totals in the Fortran format
All results converted to weight % and entered
Entered as reported
All results converted to /ifl/g (conversion factors supplied by EPA
Project Officer, Max Samfield)
Entered as reported
All results converted to MS/9
All results converted
Entered as reported
M= E =
CLM, FM.S04E, N03M,
N02M, C03E, SE
LC1-LC7
LC1-LC7
GC7-GC13
HGAA
SBAA, ASAA
PH, ACIDTY.
ALKNTY, BOD, COD,
DO, CONDTY,
DISSOL, SUSSOL,
CBM, LNS03E*H2S,
ORGNCS, LNNH3*.
LNCNM;SCNM
*ln showing an element or compound abbreviation, the initial letters "LN" indicate that a logarithmic transformation has been performed on the data.
For ease of identification, "LN" is omitted from the abbreviation after this table.
por a given element/compound abbreviation, a final letter "M" shows a negative charge; "£", two negative charges; and P, a positive charge.
(continued)
-------
TABLE 10 (continued)
Study
Original
results
Quantification
6,10,19 AMES-number of revertant colonies
formed relative to dose
CO
en
The largest ratio of revertants to control colonies (with and without
activation) was entered for each strain tested
*See Table 8 for a detailed description of this method.
Computer-coded
final results
TA1535M.TA1535P,
TA1537M.TA1537P,
TA 98M, TA 98P,
TA100M.TA100P,
TA1538M.TA1538P
6,10,19 AMES
RAM
Wl-38
CHO
RAT
FW algal
Grass Shrimp
FW fish
Oaphnia
Litton* Quantification, using
- ATP measurement
- ATP measurement
- Mean % survival (viability)
- Lethality
- Inhibition of growth
- Lethality
- Lethality
- Lethality (immobility)
the following parameters:
AMES 2
RAM
Wl-38
CHO
RODENT
FWAGL
GRSSHP
FWFSH
DAPNA
-------
SECTION 4
DATA ANALYSIS AND INTERPRETATION
Many approaches were instituted in an effort to analyze the data popu-
lation and generate an accurate, detailed interpretation. This section will
detail the approaches, present results of each approach, and define decision
criteria that were used to interpret the results.
The approaches to data analysis are:
1. Basic summary statistics on each sample parameter including maximum
values, minimum values, mean, standard deviation, and the number
of values in each data set (e.g., total number of fluoride deter-
minations, total number of algal tests).
2. Linear (Pearson) and rank (Spearman) correlation analyses of
chemical v. biological parameters and biological v. biological
parameters. Correlations were determined for the entire data
population (all four studies), as well as for each individual
study, and on liquid and solid samples.
3. Scatter plot analysis.
4. Stepwise regression to ascertain antagonistic and/or synergistic
effects.
5. Probit analysis of RAM data to linearize the dose-response rela-
tionship of such data.
6. Analysis using MEG/MATE values, SAM/IA, and Source Severity models.
7. Description of concurrent cooperative work by Battelle-Columbus
Laboratories in evaluating these data using a pattern recognition
program, and further statistical analysis by Don Lewis of EPA/RTP.
8. Analysis of engineering data from the FBC process (study 10).
9. Original results compared to the Litton quantification scheme.
10. Correlations among Ames/Salmonella strains.
11. Comparison of schemes for ranking samples.
36
-------
To show the limited nature of the data being examined, a representative
scatter plot of typical values is provided in Figure 4; these data show a
"good" correlation (correlation >0.50 with a significance factor <0.05)
according to the study's decision criteria.
4.1 SUMMARY STATISTICS
It was determined from examining the summary statistics that there are
much greater ranges of chemical concentration values and biological response
values in the FBC study (study 10), the coal gasification study (study 06),
and the SASS evaluation (study 15) than in the textile study (study 19).
This may be due to the nature of the samples. The textile study data set
consisted of only treated aqueous effluents from textile plants, while the
other three studies' data sets consisted of various aqueous and organic
liquid and solid samples.
4.2 CORRELATION ANALYSIS
The correlation studies were executed utilizing the statistical analysis
system (SAS),12 a packaged computer program. Data from the 51 sample sets
were entered and sorted by study (06, 10, etc.), by sample (cyclone dust,
tar, etc.), by analytical test (Ames, SSMS, etc.), and by chemical (Zn, COa,
etc.); the group of sample data sets presented in Appendix A was the first
output. The basic summary statistics were calculated for each parameter as
discussed in the previous section.
To determine the degree of association between pairs of variables, the
correlation coefficients were calculated. Two types of correlations were
computed: the linear (Pearson Product-Moment) correlation and the rank
(Spearman) correlation. The linear correlation is used in cases where
normality and linearity can be assumed. The rank correlation makes no
assumptions regarding normality or linearity, and seems to be more appropri-
ate for this data set; therefore only rank correlations are discussed.
4.2.1 Rank (Spearman) Correlation (r )
The Spearman correlation is based on rankings and, in contrast to the
linear correlation, makes no assumptions about normality or linearity.
Thus, in the present case where many of the data were discrete (e.g. , the
Ames test results) or highly skewed (e.g., many of the chemical test
37
-------
00
27
24
21
18
15
12
9
LEGEND: A = 1 OBSERVATION
8=2 OBSERVATIONS, ETC.
A
A
C
O
G
A
A
NOTE: 19 OBS HAD MISSING VALUES
AMES2
Figure 4. Plot of TA98P vs. AMES2.
-------
results), it appears that this correlation is more appropriate than the
Pearson correlation. To compute this correlation, complete the following
steps:
1. Rank the observations of the X variable from 1 to N.
2. Rank the observations of the Y variable from 1 to N.
3. List the corresponding N observations for X and Y and beside each
observation give the observation's rank.
4. Determine the value of d. for each observation by subtracting the
Y rank from the X rank. Square this value to determine each
observation's d.2.
5. Sum the d..2's for the N observations to determine
6. Compute the rank correlation by the formula
N
6 £ d.2
r =
s * N(NZ-1)
The above computations can be demonstrated by using the data in Table 11.
Note in Table 12 that when two observations are tied, each receives the
average of the two ranks that would have been assigned had no ties occurred
(e.g., for X = 0 in Table 11, the average of the ranks is 1.5 in Table 12).
From the computations in Table 12, r is computed by equation (1) as
= i 6(109.50) _ n
s 12(144-1) u-
To test whether rg is significantly different from zero, a two-tailed
test of significance can be performed. The test of significance is based on
the Student1s-t distribution with N-2 degrees of freedom. The statistic is
computed as follows:
The value computed by equation (2) is compared with t-tables for N-2 degrees
of freedom. For our example data,
t,n = 0.62 I" ^^m =2.50,
1U Ll-(0.62rJ
39
-------
TABLE 11. EXAMPLE DATA FOR RANK CORRELATION
Observation
1
2
3
4
5
6
X
0
0
1
1
3
4
Y
42
46
39
37
65
88
Observation
7
8
9
10
11
12
X
5
6
7
8
8
12
y
86
56
62
92
54
81
TABLE 12. COMPUTATIONS FOR RANK CORRELATION
Rink
Observations X Y dj dj2
1
2
3
4
5
6
7
8
9
10
11
12
1.5
1.5
3.5
3.5
5
6
7
8
9
10.5
10.5
12
3
4
2
1
8
11
10
6
7
12
5
9
-1.5
-2.5
1.5
2.5
-3.0
-5.0
-3.0
2.0
2.0
-1.5
-5.5
3.0
2.25
6.25
2.25
6.25
9.00
25.00
9.00
4.00
4.00
2.25
30.25
9.00
109.50
TABLE 13. "GOOD" RANK CORRELATIONS IN STUDY 15
RAM RAM
Mo + Be +
V + Mn
Note: The transformation x = log(c + 1), where c = chemical
concentration, was made for each chemical element.
+ = Positive correlation.
-= Negative correlation.
40
-------
which for a two-tailed test is significantly different from zero at the 0.05
level of significance.
In summary, if the absolute value of the correlation coefficient is
close to 1, then a good correlation is said to exist. If examination of the
scatter plot (graph of all data pairs used in a correlation) shows the data
spread along the correlation curve and eliminates the presence of a single
high value influencing a correlation, the statement of association is further
validated. Well-defined dose-response nonlinear curves, as shown by data
plots, are also important to note. The evaluation of scatter plots is
discussed in the next section of this chapter.
In the remaining analyses, no assumptions of the presence or absence of
a linear relationship between chemical and biological results were made; the
effort was directed at identifying trends or relationships. For this purpose,
data from the four studies were analyzed collectively, each study was analyzed
independently of the others, and then liquid and solid samples were analyzed
separately.
Correlations were performed on the following data sets:
1. Original investigator's biotest results v. original investigator's
biotest results to determine interrelationships among the biotests.
2. Original biotest results v. chemical results to determine the
existence of relationships between chemical concentrations and
response in biological systems.
3. Ames test bacterial strains v. Ames test bacterial strains to
determine interrelationships among the test strains (e.g., TA98 v.
TA1538).
4. Biotest results quantified by Litton v. chemical results.
5. Biotest results quantified by Litton v. originally reported biotest
results.
The output of the SAS correlation programs showed the correlation
coefficient, a probability factor, and the size of the correlation data set.
For example, the rank correlation for parameters RAM1000 v. Uranium concen-
tration across all four studies generated the following results:
0.42432
0.1305
14
41
-------
The correlation coefficient of 0.42432 is low so a significant rank
correlation would not be expected. In addition, the probability factor of
0.1305 indicates a significant probability exists that the correlation could
be due to chance (i.e., only when the probability factor is smaller than
0.05 would the correlation be considered significantly different from zero
at the 0.05 level). The number of data pairs is 14; that is, 14 samples
contained data on both uranium concentration and rabbit alveolar macrophage
response.
Decision criteria were established to reduce the number of correlations
considered for more detailed analysis. The decision made was to tabulate
all paired data sets that had a rank correlation coefficient greater than or
equal to 0.50 plus a significance factor less than 0.05. All linear and
rank correlations meeting these criteria were noted as "possible" or "good"
relationships and are reported in Appendix C. Since rank correlations are
considered more appropriate for this data set, the following section discus-
ses each group of "good" rank correlations. The discussion of results in
each study or grouping is organized to include the following:
Chemistry v. Biology (Ecology-related)
Chemistry v. Biology (Health-related)
Biology (Ecology) v. Biology (Ecology)
Biology (Ecology) v. Biology (Health)
Biology (Health) v. Biology (Health)
A positive correlation implies that the value of one parameter (biolog-
ical response, chemical concentration) increases as the value of the other
parameter increases. A negative correlation implies that one parameter
decreases as the other increases. Because LD50, LC50, EC50, and EC20 values
are expressed as percent sample concentration in this data set, a negative
correlation coefficient for these parameters implies a positive relation-
ship; that is, LD50 a I/intensity of biological response.
4.2.2 Study 06—Coal Gasifier
There were only four samples in the data set. This sample size is too
small to use in drawing conclusions, so no correlations are given.
42
-------
4.2.3 Study 10--F1uidized-Bed Combustor
There were only six samples in the data set. This sample size is too
small to use in drawing conclusions, so no correlations are given.
4.2.4 Study 15--SASS Train Evaluation, Various Industries
Sixteen sample were in this group. Only the RAM test results are in
the data set; these "good" rank correlations were shown: RAM cell viability
was positively correlated with Mo, V, and Be, and negatively correlated with
Mn.
Table 13, page 40, summarizes the "good" rank correlations in Study 15.
4.2.5 Study 19—Textile Plants' Effluents
There were 15 samples in this group. This study had the most complete
data set and many "good" rank correlations were found:
The RAM test was positively correlated with Pb, W, Hf, Yb,
Tin, Dy, Tb, Eu, Sm, Nd, Cd, Ag, Nb, Zr, Sr, Rb, Ni, V, NOz,
pH, and NOg.
The RAT parameter, which considered behavioral and physiolog-
ical effects, was positively correlated with Pb, Ta, Yb, Er,
Zr, and Fe, and negatively correlated with MEG1, LC4, and
LC5.
The freshwater algal test was positively correlated with Pb,
Sm, Ce, Cr, and S, and negatively correlated with MEG1.
The saltwater algal test was positively correlated with Ce,
Zr, and V.
The freshwater fish test was positively correlated with U,
Pb, Dy, Sm, Ce, Sn, Ag, Zr, Y, Sr, Rb, Ni, V, Li, NOg, and
total organics, and negatively correlated with Cl, LC6, and
As.
The Daphnia test was positively correlated with Ce, La, Zr,
Mn, Ca, and NOs.
The saltwater fish test was positively correlated with U, Th,
Pb, Tl, Hf, Lu, Yb, Tm, Er, Dy, Gd, Eu, Cd, Ag, Zr, Y, Sr,
Rb, Ga, Sc, Ca, Si, Be, As, and conductivity, and negatively
correlated with MEGS and LC6.
43
-------
The grass shrimp test was positively correlated with U, Th,
Bi, Pb, W, Hf, Lu, Yb, Tm, Er, Dy, Gd, Eu, Nd, Cd, Ag, Y, Sr,
Rb, Ni, Sc, Ca, Si, Be, and conductivity, and negatively
correlated with MEGS and LC6.
Ames strain TA1535 (no S-9) and Ames strain TA98 (no S-9)
were positively correlated with CHO.
Ames strain TA98 (with S-9) was positively correlated with
RAM cell viability.
The freshwater fish test was positively correlated with the
saltwater algal test, the Daphm'a test, the saltwater fish
test, and the grass shrimp test.
The saltwater fish test was positively correlated with the
Daphnia test, RAM cell viability, and RAM ATP level.
The grass shrimp test was positively correlated with the
Daphnia test and the saltwater fish test.
Table 14 summarizes "good" correlations for Study 19.
4.2.6 All Studies Grouped Together
All data (51 samples) from studies 06, 10, 15, and 19 considered together
give these "good" rank correlation.
The RAM test was positively correlated with Cd, Br, S, MEG1,
MEG18, MEG21, MEG23, MEG25, NOg, HQ~2t LC2, LC8, and C12, and
negatively correlated with Ce, Ga, Fe, V, Ti, Cu, Si, Mg, Be,
and Hg.
The Ames test was positively correlated with Bi, Pb, Ga,
MEG2, MEG15, LCI, LC2, LC3, Hg, and As.
The RAT test was positively correlated with Pb, Fe, and MEG1.
The freshwater algal test was positively correlated with Ce,
La, Ba, Sr, Rb, As, Mn, Ti, Ca, K, S, Mg, F, and 504.
The saltwater algal test was positively correlated with Ce,
Zr, and SCv
The freshwater fish test was positively correlated with Pb,
Li, and total organics, and negatively correlated with MEG15.
The Daphnia test was positively correlated with Ca.
44
-------
TABLE 14. "GOOD" RANK CORRELATIONS IN STUDY 19
FW SW FW SW GRASS
RAM CHO RAT ALGAL ALGAL FISH DAPHNIA FISH SHRIMP WI-38 AMES
U 4-4-4-
Th + +
Bi +
Pb + 4-4- 4- 4-4-
W + 4-4-
Ta +
Hf 4- 4-4-
Lu + +
Yb + 4- 4-4-
Tm + 4-4-
Er + 4-4-
Dy + 4-4-4-
Tb +
Gd 4-4-
Eu + 4-4-
Sm + + 4-
Nd + +
Ce 4-4-4-4-
La +
Sn 4-
Cd + 4-4-
Ag + + 4-4-
Nb +
Zr + 4- 4-4-4-4-
Y 4-4-4-
Sr + 4- 4- +
fib + 4-4-4-
As -
Ga 4-
Ni + + 4-
Fe +
Mn +
Cr +
V + 4-4-
Sc 4-4-
Ca 4-4-4-
Cl -
s +
Si 4-4-
Be 4-4-
Li +
See notes at end of table. (continued)
45
-------
TABLE 14. (continued)
FW SW FW SW GRASS
RAM CHO RAT ALGAL ALGAL FISH DAPHN1A FISH SHRIMP WI-38 AMES
MEG1 - —
MEGS - —
NOs 4- + +
N02 +
LC4 -
LC5 —
LC6 — — —
As-AA +
pH +
Conductivity + +
Cr" +
Total organics +
+ = Positive correlation.
— = Negative correlation.
Note: The transformation x = logic + 1), where c = chemical concentration, was made for all chemical elements, MEG1-MEG26
S0|, NH3, and CM".
46
-------
The saltwater fish test was negatively correlated with MEG15
and LC6 and positively correlated with conductivity.
The grass shrimp test was positively correlated with U, W,
Cd, Ag, Sr, Rb, Sc, Ca, Si, and conductivity, and negatively
correlated with MEG15 and LC6.
The RAT test was positively correlated with Sr, Rb, Sc, and
LC3, and negatively correlated with MEG3, MEGS, and DO.
The saltwater algal test was positively correlated with the
freshwater algal test, the freshwater fish test, and the
saltwater fish test.
The freshwater algal test was positively correlated with RAM
ATP and WI-38.
The freshwater fish test was positively correlated with grass
shrimp and RAM ATP.
The Daphnia test was positively correlated with the fresh-
water fish test, saltwater fish test, and grass shrimp test.
The saltwater fish test was positively correlated with the
freshwater fish test, grass shrimp test, RAM ATP, and RAM
cell viability.
The grass shrimp test was positively correlated with RAM ATP
and RAM cell viability.
The RAM test was positively correlated with Ames strain TA98
(with S-9), TA1538 (with S-9), and WI-38.
The CHO test was positively correlated with Ames strains
TA1535 (no S-9) and TA98 (no S-9).
The RAT test was positively correlated with Ames strain
TA1535 (with S-9).
Table 15 summarizes the significant correlations found in the data set
of all studies combined.
4.2.7 Liquids
There were 15 liquid samples from studies 06, 10, and 19. They showed
these "good" rank correlations:
The RAM test was positively correlated with W, Cd, Ag, Rb,
Br, Sc, Si, MEG1, $04, pH, and Cr++, and negatively corre-
47
-------
TABLE 15. "GOOD" RANK CORRELATIONS FROM ALL STUDIES COMBINED
RAM CHO
U
Bi
Pb
W
Ce -
La
Ba
Cd 4-
Ag
Zr
Sr
Rb
Br +
As
Ga -
Fe ~~
Mn
v -
Ti —
Sc
Ca -
K
S +
Si -
Mg —
F
Li
Be ~
ME61 +
MEG2
MEG3
MEG5
MEG15
MEG18 +
MEG21 4-
MEG23 4-
MEG25 +
S04
N03 +
NOz
FW SW FW SW GRASS
RAT ALGAL ALGAL FISH DAPHNIA FISH SHRIMP Wl 38 AMES
+
+
+ + +
+
+ +
+
•f
+
+
4-
+ + +
+ + +
4-
4-
+
4-
4-
+ +
+ + 4-
4-
4-
4-
+
4- :
+
4-
4-
4- 4-
See notes at end of table. (contjnued)
48
-------
TABLE IS. (continued)
FW SW FW SW GRASS
RAM CHO RAT ALGAL ALGAL FISH DAPHNIA FISH SHRIMP WI-38 AMES
LC1 +
LC2 + +
LC3 + +
LC6 - -
LC8 +
Hg-AA — +
As-AA +
pH +
DO -
Conductivity + +
Organics +
+ = Positvie correlation.
—= Negative correlation.
Note: The transformation x = log(c + 1), where c = chemical concentration, was made for all chemical elements, MEG1-MEG26,
S03,NH3,andCN~.
49
-------
lated with U, Th, Bi, Pb, Tl, Au, Pt, Ir, Os, Hf, Lu, Yb, Tm,
Er, Ho, Dy, Tb, Gd, EU, Sm, Nd, Pr, Te, Pd, Rh, Ru, Mo, Ge,
Ga, Ni, Fe, V, Na, Be, and Hg.
The RAT test was positively correlated with Pb, Ta, Hf, Yb,
Er, Mo, Fe, Cs, Rb, and Na, and negatively correlated with
MEG1, MEG18, LC5, and DO.
The freshwater algal test was positively correlated with W,
Ce, La, Bu, Cs, Ag, Nb, Lr, Y, Sr, Rb, As, Mn, Ti, Ca, K, S,
Mg, F, $04, and NOs-
The freshwater fish test was positively correlated with Pb,
Li, and total organics, and negatively correlated with MEG15.
The Daphnia test was positively correlated with Ce.
The saltwater fish test was positively correlated with MEG15
and conductivity, and negatively correlated with LC6.
The grass shrimp test was positively correlated with U, W,
Cd, Ag, Sr, Rb, Sc, Si, and conductivity, and negatively
correlated with MEG15 and LC6.
The freshwater algal test was positively correlated with RAM
ATP and WI-38.
The saltwater algal test was positively correlated with
freshwater algal test and freshwater fish test.
The freshwater fish test was positively correlated with the
saltwater algal test, grass shrimp test, and RAM ATP.
The Daphnia test was positively correlated with the fresh-
water fish test.
The saltwater fish test was positively correlated with salt-
water algal test, freshwater fish test, Daphnia test, grass
shrimp test, RAM cell viability, and RAM ATP.
The grass shrimp test was positively correlated with the fresh-
water fish test, Daphnia test, RAM ATP, and RAM cell viability.
The RAM test was positively correlated with WI-38, RAT, and
Ames strains TA1537 (no S-9) and TA98 (with S-9).
The CHO test was positively correlated with Ames strains
TA1535 (no S-9) and TA98 (no S-9).
Table 16 shows the "good" correlations on the liquid samples.
50
-------
TABLE 16. "GOOD" RANK CORRELATIONS IN LIQUID SAMPLES
FW SW FW SW GRASS
RAM CHO RAT ALGAL ALGAL FISH DAPHNIA FISH SHRIMP WI-38 AMES
U - +
Th -
Si —
Pb - + +
T1 -
Au -
Pt -
Ir -
Os —
W + + +
Ta +
Hf - +
Lu -
Yb +
Tm -
Er +
Ho -
DV
Tb -
Gd -
Eu -
Sm ~~
Nd -
Pr -
Ce +
La +
Ba +
Cs +
Te —
Cd + +
Ag + +
Pd -
Rh -
Ru -
Mo — +
Zr +
Y +
Sr + +
Rb + + + +
Br +
As +
See notes at end of table. (continued)
51
-------
TABLE 16. (continued)
FW SW FW SW GRASS
RAM CHO RAT ALGAL ALGAL FISH DAPHNIA FISH SHRIMP WI-38 AMES
Ge -
Ga -
Ni -
Fe - +
Mn +
V -
Ti +
Sc + +
Ca +
K +
S +
Si + + .
Mg +
Na - +
F - +
Be —
Li +
MEG1 + -
MEG15 - + —
MEG18 -
S04 +
NO* + +
LC5
LC6 ~ —
Hg-AA —
pH +
DO ~
Conductivity + +
Cr" +
Organics +
+ = Positive correlation. .
— = Negative correlation. {
Note: The transformation x = logic-i-1), where c = chemical concentration, was made for ill chemical elements, MEG1-MEG26,
SOj. NH3, and CM".
52
-------
4.2.8 Solids
There were nine samples from studies 06, 10, and 19 used in the correla-
tion matrix; correlations obtained were:
The Ames test was positively correlated with Bi, I, Pb, F,
MEG1, MEG2, MEG15, MEG21, MEG25, LC2, and LC3, and negatively
correlated with Cl .
The RAM test was positively correlated with Bi, I, MEG1,
MEG2, MEG15, MEG21, MEG25, LCI, LC2, LC3, and LC8, and nega-
tively correlated with P.
The Ames test was positively correlated with RAM.
Table 17 summarizes the correlations for solids data.
4.2.9 Summary Correlations
It is readily apparent that there were many potential or possible
relationships based on the data from these four studies. It is also apparent
that some of these "good" correlations are due to the small number of data
points, because the relative number of "good" correlations tends to decrease
when the number of data pairs used in calculating the correlation coefficient
increases. More data, as suggested in the last section of this document,
would produce a data analysis with greater statistical reliability and
validity.
As noted previously, both rank and linear correlations were examined
for studies 06, 10, 15, 19, all studies combined, all liquid samples combined,
and all solid samples combined. Relationships noted below are cases where
the study groupings show at least two "good" correlations including at least
one "good" rank correlation. With these qualifiers in mind, the following
trends were observed from this analysis of the environmental assessment
data:
1. Pb was correlated with the greatest number of bioassays, followed
by Rb. Pb concentration was correlated with seven biological
parameters: Ames, RAM, RAT, freshwater algal, freshwater and
saltwater fish, and grass shrimp. Rb was correlated with six
biological parameters: RAM, RAT, freshwater algal, freshwater and
saltwater fish, and grass shrimp.
53
-------
TABLE 17. "GOOD" RANK CORRELATIONS IN SOLIDS SAMPLES*
RAM AMES RAM AMES
Bi +
Pb
1 +
Cl
D
F
MEG1 +
MEG2 +
+ MEG15 +
+ MEG21 + +
+ MEG23 +
MEG25 + +
LCI +
+ LC2 + +
+ LC3 + +
+ LC8 +
+ = Positive correlation.
— = Negative correlation.
Note: The transformation x = log(c +1), where c = chemical concentration, was made for all
chemical elements, MEG1-MEG26,863, NH3, and CN~.
54
-------
2. The rabbit alveolar macrophage (RAM) test appeared to be more
sensitive for this group of samples, being correlated with
60 chemical parameters, as compared to Daphnia, which was corre-
lated with 1 chemical parameter.
Table 18 summarizes those chemistry v. biology combinations that show
at least two "good" correlations, including at least one "good" rank corre-
lation. As previously stated, the correlations had been run on each study
individually, on all studies together, on liquid samples, and on solid
samples; combinations where the correlation coefficient was greater than
0.50 with a significance level less than 0.05 had been selected as "good".
Table 19 summarizes the biology v. biology combinations that show at
least two "good" correlations, including at least one "good" rank correla-
tion. In Table 19, it is of interest to note that all the aquatic animal
tests were correlated; that is, all possible combinations involving salt-
water fish, freshwater fish, Daphnia, and grass shrimp were interrelated and
qualify as "good" correlations. The freshwater algal test was the only
ecology-related test that correlated with more than one health-related test,
showing possible relationships with the RAM and WI-38 tests. The RAM test
(health-related) was correlated with the freshwater and saltwater fish and
grass shrimp tests as well as with the freshwater algal test. As expected,
the freshwater and saltwater algal tests were highly correlated, as were the
RAM and WI-38 cytotoxicity tests.
4.3 SCATTER PLOT ANALYSIS
The criteria for discussion of scatter plots were similar to those used
in discussion of "good" correlations. Scatter plots are discussed for cases
where a "good" rank correlation for all studies combined was shown supported
by a "good" correlation for at least one other grouping.
In discussions of the chemical-biological test combinations, it should
be noted that the chemical parameter is considered the independent variable,
on the x-axis, and the biological parameter is considered the dependent
parameter, on the y-axis.
In examining the scatter plots, several patterns in the data become ap-
parent. However, it must be remembered that the data base is statistically
too small, with too few observations and too many missing numbers, for the
55
-------
TABLE 18. SUMMARY OF CHEMICAL AND BIOLOGICAL TEST COMBINATIONS WHERE STUDY
GROUPINGS SHOW AT LEAST TWO "GOOD" CORRELATIONS INCLUDING AT LEAST ONE
"GOOD" RANK CORRELATION
Biotoejcal tests
Chemical
tests
U
Th
Bi
Pb
Tl
Au
Pt
Ir
Os
W
Ta
Hf
Lu
Yb
Tin
Er
Ho
Dy
Tb
Gd
Eu
Sm
Nd
Pr
Ce
La
Ba
Cs
1
Te
Sn
Cd
Ag
Pd
Rh
Ru
Mo
Zr
Y
Sr
Rb
Br
As
Ge
AMES
X*
X
X
X*
X*
X"
X* ,
X*
X*
X"
X*
X*
X*
X*
X*
X*
X"
X*
X*
X"
X"
Health-Related
RAM RAT CHO
X»
X*
X"
X X
X»
X'
X"
X'
X"
X
X*
X
x«
X X
X
X* X
x»
X
X
X*
X
X
X
X"
X
X*
X*
X
X
X*
X"
X*
X X
X X
X
X X
X
X*
EcolOfy-Rtlittd
FW SW FW SW
ALGAL ALGAL FISH FISH DAPHNIA
X
X
X XX
X
X
X
X
X X
X
X X
X
X
X
XXX
X X
X XX
X XX
X
GRASS
SHRIMP
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
(continued)
56
-------
TABLE 18. (continued)
Chemical
tests
Ga
Zn
Ni
Fe
Mn
Cr
V
Ti
Sc
Ca
K
Cl
S
Si
Mg
F
Be
Li
MEG1
MEG2
MEGS
MEG5
MEGS
MEG15
MEG18
MEG21
MEG23
MEG25
F-
SG-4
N02
LCI
LC2
LC3
LC4
LC5
LC6
LC8
AMES
X
X*
X*
X*
X*
x«
X*
X*
X*
X*
X
x'*
X*
x«
X
X
X
Biological tests
Health -related Ecology-related
FW SW FW SW GRASS
RAM RAT CHO ALGAL ALGAL FISH FISH DAPHNIA SHRIMP
X*
X X*
X X
X
X
X
X X
X
X* X XX
X
X' X
X X
X X
X
X
X
X X
x»
X*
X*
X* XX
X* X* X* X»
X
X*
X*
X*
X X
X
X
X*
X
X*
X
X* X
X X
X*
(continued)
57
-------
TABLE 18. leontinutd)
Biological ttstt
Health-Related Ecology-Related
Chemical
tests AMES
Hg-AA X
A$-AA X"
pH
DO
Conductivity
Cr"
Organic*
FW SW FW SW GRASS
RAM RAT CHO ALGAL ALGAL FISH FISH OAPHNIA SHRIMP
X*
X
X
x x
X
X
"Good" correlation is band on fewer than 10 pairs of numbers.
Note: The transformation x = logfc + 1), where c = chemical concentration, was made for all chemical elements, MEG1-MEG26
$03, NH3, and CN". '
58
-------
TABLE 19. SUMMARY OF BIOLOGICAL TEST COMBINATIONS
WHERE STUDY GROUPINGS SHOW AT LEAST TWO "GOOD" CORRELATIONS,
INCLUDING AT LEAST ONE "GOOD" RANK CORRELATION
Health-Related
AMES
RAM
Wl-38
Ecology-Related
FW ALGAL
SW ALGAL
FW FISH
SW FISH
DAPHNIA
GRASS
SHRIMP
Health-Related
FW SW
AMES RAM WI-3B ALGAL ALGAL
X»
X* - X* X*
X* - X*
X* X* X
X
X X
X X
X
Ecology
FW
FISH
X
X
-
X
X
X
-Related
SW GRASS
FISH DAPHNIA SHRIMP
X X
X
XXX
- X X
X - X
X X -
"Correlation is based on fewer than 10 pairs of numbers.
Note: The transformation x = logic + 1), where c = chemical concentration, was made for all chemical elements. MEG 1-MEGZ6,
S03,NH3,andCI\T.
59
-------
results to be considered conclusive. With these qualifiers in mind the fol-
lowing observations are made. Table 20 lists the chemical v. biological
test combinations, grouped by the pattern shown in the scatter plot. Table
21 presents patterns shown among the biological parameters.
4.3.1 Chemical-Biological Test Combinations
Thirteen scatter plots of MEG categories v. RAM parameters were graphed.
These plots involved MEG1, MEG18, MEG23, and MEG25. The seven plots showing
logarithmic curves plotted the MEG values against RAM1000 (rabbit alveolar
macrophage), VIABIND (viability index), and ATP. Figure 5 is a typical plot
from the data set showing a logarithmic curve.
A linear trend is observed in the four plots of MEG values v. PROBVI
(viability index probit analysis). A linear trend is also found in the AMES
plots for Pb, Bi, and As-AA, as well as the RAM plots of Fe, Si, and Mg
values. Figure 6 shows a scatter plot of the linear type.
Scatter diagrams were classified as random observations for many com-
binations. Approximately one-third of the grass shrimp plots were thought
to consist of random observations; chemical parameters involved were U, Si,
LC6, and conductivity. Half of the PROBVI plots show a random pattern (Br,
S, NOa, NOg), as do a few of the AMES plots (Bi, Ga, Hg-AA). Figure 7 shows
a representative random scatter plot.
All freshwater algal samples and almost all saltwater algal samples
showed at least a minimal negative slope; that is, the higher chemical
values were paired with the lower biological values. The chemical parameters
correlating with freshwater algal are Ca, Mn, S, SO^, Ba, Sr, Ca, K, Mg, As,
and Cd. The chemical parameters related to saltwater algal are Zr and Ce.
In contrast, high values for chemical parameters were sometimes paired
with high values for biological parameters, giving a scatter plot with a
positive slope. Scatter plots classified thusly include two-thirds of the
grass shrimp plots, as well as some AMES2 and RAM (Litton transformation)
data. Correlations in this grouping include grass shrimp paired with W, Cd,
Ag, Sr, Rb, Ca, Sc, Sr, and Si; AMES2 paired with Pb, Ga, MEG2, MEG10,
MEG15, LCI, LC2, and LC3; and RAM paired with MEG1, MEG21, and MEG25.
60
-------
TABLE 20. PATTERNS IN SELECTED SCATTER PLOTS, CHEMICAL TESTS V. BIOASSAYS
LOG
MEGIv. RAM1000
*MEG25v. RAM1000
*MEG23v. VIABIND
*MEG25 v. VIABIND
MEGIv. ATP
MEGISv. ATP
•MEG25 v. ATP
cr>
LINEAR
ORGNCSv. FWFSH96
Cd v. GSHMP48
pHv. PROBVI
*MEG25 v. PROBVI
MEGIv. PROBVI
*MEG23v. PROBVI
Pb v. AMES1
AsAA v. AM ESI
Pbv. OTHRAT
Biv. AMES2
AsAA v. AMES2
Fe v. RAM
Si v. RAM
Mgv. RAM
RANDOM
PR.
504 v. SWALGAL
Liv. FWFSH96
Cav. DAPH48
CONDTYv.SWFSH48
LC6 v. SWFSH96
CONDTYv. GSHMP24
Si v. GSHMP24
LC6 v. GSHMP24
CONDTYv. GSHMP48
Sr v. GSHMP48
LC6 v. GSHMP96
CONDTY v. GSHMP96
*MEG5v. RATWT
DO v. RATWT
Uv. GRSSHP
Pb v. FWFSH
Brv. PROBVI
NO 2V. PROBVI
NO^v. PROBVI
Sv. PROBVI
Bi v. AMES1
Gav. AMES1
HgAAv.AMESI
Fev. OTHRAT
NEGATIVE
SLOPE
Cev. FWALGAL
Mnv. FWALGAL
Sv. FWALGAL
S0|v. FWALGAL
Bav. FWALGAL
Sr v. FWALGAL
Cav. FWALGAL
Kv. FWALGAL
Mgv. FWALGAL
As v. FWALGAL
Cd v. PROBVI
MEGIv. OTHRAT
Zrv. SWALGAL
Ce v. SWALGAL
V v. RAM
POSITIVE
SIOPE
POSITIVE
SIDPE
OR
Wv. GSHMP24
Cdv.GSHMP24
Agv. GSHMP24
Sr v. GSHMP24
Rb v. GSHMP24
Ca v. GSHMP24
Sc v. GSHMP24
W v. GSHMP48
Siv. GSHMP48
Rbv. GSHMP48
Cav. GSHMP48
Ag v. GSHMP48
Scv. GSHMP48
Agv. GSHMP96
Rbv. GSHMP96
*MEG3 v. RATWT
LC2v. PROBVI
MEG15v. AMES1
LC2v.AMES1
LC3 v. AMES1
Pbv.AMES2
Gav. AMES2
*MEG2v. AMES2
MEG15v. AMES2
LC1 v. AMES2
LC2 v. AMES2
LC3 v. AMES2
MEG1 v. RAM
*MEG21 v. RAM
•MEG25 v. RAM
"Correlat o~ s rasfid on fe^er than 10 pairs of numbers.
Note: 7-f t-a-sfcrnation >. - logic * 1), where r = chemical concentration, was made for all chemical elements, MEG1-MEG26, SOg, IMHg, and CN~
-------
TABLE 21. PATTERNS IN SELECTED SCATTER PLOTS. BIOASSAYS V. BIOASSAYS
LINEAR
RANDOM
* *
• * " * •
RAMIOOOv. ATP
RAMIOOOv. VIABIND
VIABIND v. ATP
SWFSH24v.SWFSH48
SWFSH24 v. SWFSH96
*SWFSH24v. RAM 1000
*SWFSH24v. VIABIND
SWFSH48 v. SWFSH96
*SWFSH48v. RAM 1000
r$ *SWFSH48 v. VIABIND
• i
AMES1 v. TA98P
FWFSH96 v. DAPH48
*SWFSH96 v. RAM 1000 FWFSH96 v. GRSSHP
•SWFSH96 v. VIABIND DAPH48 v. GSHMP24
GSHMP24 v. GSHMP48 GSHMP48 v. PROBATP
•GSHMP24 v. RAM1000 GSHMP96 v. PROBATP
•GSHMP24 v. VIABIND GRSSHP v. PROBATP
GSHMP48 v. GSHMP96 RAMIOOOv. TA98P
*GSHMP48v. RAM 1000 VIABIND V.TA98P
"GSHMP48 v. VIABIND ATPv.TA1538P
GSHMP96 v. GRSSHP AMES1 v. TA1538P
*GSHMP96v. RAM 1000 PROBVI v. TA98P
*GSHMP96v. VIABINO
POSITIVE
SLOPE
*
*
* *
AMES1 v. AMES2
AMES1 v. TA100M
AMES2v.TA1538P
PROBVI v. RAM
PROBATP v. RAM
OTHRATv.TA1535P
FWALGLv. ATP
SWALGALv. FWALGAL
SWALGALv. FWFSH96
SWALGALv. SWFSH24
SWALGALv. SWFSH96
FWFSH96 v. SWFSH96
FWFSH96 v. SWFSH48
DAPH48 v. SWFSH96
GSHMP24 v. GSHMP96
GRSSHP v.FWFSH
DAPH48 v. SWFSH24
NEGATIVE MISCELLANEOUS
SLOPE
: . I
•
RAMIOOOv. PROBATP SWFSH24v. GRSSHP
RAMIOOOv. PROBVI SWFSH48 v. GRSSHP
RAM1000 v. RAM SWFSH48 v. PROBATP
VIABINO v. PROBVI SWFSH96v. GRSSHP
VIABIND v. RAM GSHMP24 v. GRSSHP
ATP v. PROBATP GSHMP48 v. GRSSHP
ATP v. RAM
*ATP v. WI38
*FWALGLv.WI38
SWALGAL v. FWFSH
FWFSH96 v.FWFSH
DAPH48 v. FWFSH
SWFSH24 v. FWFSH
SWFSH48 v. FWFSH
SWFSH48 v. PROBVI
SWFSH96 v. FWFSH
SWFSH96V. PROBVI
GSHMP24 v. FWFSH
GSHMP24v. PROBVI
GSHMP24 v. PROBATP
GSHMP48 v. FWFSH
GSHMP48v. PROBVI
GSHMP96v. FWFSH
GSHMP96v. PROBVI
'Correlation is based on fewer than 10 pairs of numbers.
Note: The transformation x = logic + 1), where c = chemical concentration, was made for all chemical elements, MEG1-MEG26, SO^,
, and CN~
-------
S
LU
7.2
6.4
5.6
4.8
a 4.0
» 3.2
2.4
1.6
0.8
0.0 -
LEGEND: A = 1 OBSERVATION
B = 2 OBSERVATIONS, ETC.
I l I l
I I l I I l 1 1
A A
I
I I
I I
I
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
RAM 1000
NOTE: 40 OBS HAD MISSING VALUES
Figure 5. Scatter plot showing logarithmic curve pattern.
-------
I
a
i
180
160
140
120
100
80
60
40
20
I
LEGEND: A = 1 OBSERVATION
B = 2 OBSERVATIONS, ETC.
I
I
AA
I
I
I
I
I
I
I
I
I
I
I
I
I
I I
I
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70
Grass Shrimp (48-hr)
NOTE: 34 OBS HAD MISSING VALUES
75 80 85 90 95 100
Figure 6. Scatter plot showing linear relationship.
-------
180
160
140
120
$ 100
(0
g
-5
80
£ eo
40
20 -
LEGEND: A = 1 OBSERVATION
B = 2 OBSERVATIONS, ETC.
-A
I
A A
I I
I 1 I I
I ! I I
I !
I
0 5 10 15 20 25 30
NOTE: 28 OBS HAD MISSING VALUES
35 40 45 50 55 60 65
Daphnia (48-hr)
70 75 80 85 90 95 100
Figure 7. Scatter plot showing random distribution.
-------
4.3.2 Biological Test Combinations
Of the 21 scatter plots that appear to depict linear relationships,
almost all show a positive slope. As expected by their definitions, when
RAM1000, VIABIND, or ATP are paired, the linear relationship and positive
slope are shown.
When any of the saltwater fish LC50 values were correlated with either
RAM1000 or the viability index, a positive linear relationship appeared.
The same type of relationship was shown when these saltwater fish studies
were correlated among themselves. As with saltwater fish, the grass shrimp
studies showed a positive linear relationship to both RAM1000 and viability
index data. When measurements were intercorrelated with grass shrimp studies,
positive linearity was also shown.
Most correlations involving the (AMES) mutagenicity strains (positive
and negative strains of TA1535, TA1537, TA1538, TA98, TA100) show the rela-
tionship to be random; these include pairing with AMES1, RAMIOOO, viability
index, ATP, and viability index probit analysis. The correlation of the
grass shrimp studies with ATP probit analysis also appears to show a ran-
domly distributed set of numbers.
Some tendency toward a positive slope is shown when saltwater algal is
correlated with freshwater fish, freshwater algal, or saltwater fish studies;
that is, higher values on the former biological test are paired with higher
values on the latter. ''When freshwater fish (96 hr) studies are correlated
with the saltwater fish studies, the tendency toward a positive slope is
shown.
When the grass shrimp are correlated with freshwater fish (Litton
quantisation), viability index probit analysis, and ATP probit analysis,
some tendency toward a negative slope is shown. Similarly, some tendency
toward a negative slope is shown when the saltwater fish studies are corre-
lated with freshwater fish (Litton quantisation) and viability index probit
analysis.
In terms of health-related biological tests, the following negative
tendencies are shown:
1. RAMIOOO correlated with RAM (Litton quantisation) and probit
analyses of viability index and ATP.
66
-------
2. Viability index correlated with RAM (Litton quantisation) and
probit analysis of viability index (PROBVI).
3. ATP correlated with probit analysis of ATP, and Litton quantita-
tion of RAM and WI-38 human lung fibroblast studies.
Figure 8 shows a scatter plot from the data set with a negative slope.
When the Litton quantitation of the grass shrimp studies is correlated
with the (LC50) saltwater fish or (LC50) grass shrimp studies, the plots
tend to show a horizontal set of points, followed by a sharp drop (a verti-
cal set of data points). This type of plot shows a high response in the
grass shrimp test with varying response levels in the saltwater fish test
and lesser response levels in the grass shrimp test at the no-response level
in the saltwater fish test. Figure 9 shows the plot of these data.
4.3.3 Summary of "Good" Correlations and Scatter Plots
Table 22 presents a summary of good biological and chemical correla-
tions that also^showed "good" scatter plots. "Good" scatter plots are
defined herein as those showing linear or logarithmic patterns. Table 22 is
in the same format as Table 18 but qualifies the mathematically "good"
correlations in Table 18 by listing only those pairings that showed a good
mathematical correlation plus a scatter plot that was subjectively evaluated
as being "good" (and eliminates those without "good" rank correlations for
all studies combined). Thus, Table 22 presents a summary of pairings that
were qualified as possibly valid by the screening processes discussed in
this section. Eleven of the thirteen combinations showing "good" correla-
tions and scatter plots involved health-related biological tests. Seven of
these combinations involved RAM parameters. The chemical parameters involved
are Fe, Si, Mg, MEG1, MEG23, MEG25, and pH. AMES parameters were in three
of the combinations; the chemical parameters correlated were Bo, Pb, and
As-AA. The RAT, freshwater fish, and grass shrimp tests were each combined
with one chemical test (Pb, total organics, and Cd, respectively).
Pb was the only chemical test correlated with more than one biological
test. As noted above, combinations were Pb v. AMES and Pb v. RAT.
Table 23 presents a similarly qualified summary of "good" correlations
and scatter plots for selected correlations involving bioassay v. bioassay.
The only health-related parameter involved is the set of RAM parameters; it
67
-------
oo
180 -
160
140
120
100
80
60
40
20
- A
A
A
I
LEGEND: A = 1 OBSERVATION
B = 2 OBSERVATIONS, ETC.
I
o.oo 0.02
AA
1
1
1
1
1
o.oe o.os o.io 0.12 0.14 o.ie o.ie 0.20 0.22 0.24 0.26
Probit ATP
0.28 0.30
NOTE: 40OBS - -2 MISSING VALUES
Figure 8. Scatter plot showing negative slope.
-------
UD
180
160
140
120
100
80
60
40
20
0 -
LEGEND: A = 1 OBSERVATION
B = 2 OBSERVATIONS, ETC.
Grass Shrimp
A
A
A
NOTE: 360BS HAD MISSING VALUES
Figure 9. Scatter plot showing one of several miscellaneous patterns.
-------
TABLE 22. SUMMARY OF PATTERNS IN SELECTED SCATTER PLOTS,
CHEMICAL TESTS V. BIOASSAYS
Biological tests
Health-Related Ecology-Related
Chemical
tests AMES
Bi X
Pb X
Cd
Fe
Si
Mg
MEG1
MEG23
MEG25
As-AA X
PH
Organics
RAM
X
X
X
X
X*
X*
X
FW GRASS
RAT FISH SHRIMP
X
X
X
'Correlation is based on fewer than 10 pairs of numbers.
Note: The transformation x = logic + 1), where c = chemical concentration, was
made for all chemical elements, MEG1-MEG26, S0|, NH3, and CN~.
TABLE 23. SUMMARY OF PATTERNS IN SELECTED
SCATTER PLOTS, BIOASSAYS V. BIOASSAYS
GRASS
RAM SW FISH SHRIMP
Health-Related
RAM
Ecology-Related
SWFISH
GRASS
SHRIMP
X* X*
X*
X*
"Correlation is based on fewer than 10 pairs of numbers.
Note: The transformation x = logic + 1), where c = chemical
concentration, was made for all chemical elements,
MEG1-MEG26, S0§, NH3, and CN~.
70
-------
is combined with the ecology- related saltwater fish and grass shrimp parame-
ters.
4.4 STEPWISE REGRESSIONS
4.4.1 Biological Tests V. Chemical Data
Stepwise regression is commonly used by researchers to indicate which
variables out of a large group of variables appear to be the most important
in predicting the data for a given variable (e.g., AMES2). The procedure
assumes that a linear relationship exists between the dependent variable
(AMES2) and the independent variable (e.g., Bi). For example,
AMES = a + B! (Bi.) + B2 (CSp + B3 (N02) + ei
where a, BI, B2, and B3 are unknown parameters and e. is a random error
term.
In brief, the stepwise regression procedure used is as follows. The
stepwise computer program finds the single-variable model (i.e., AMES2
predicted by only one variable) that produces the largest R2 statistic
(where R2 is the square of the multiple correlation coefficient). After
entering the variable with the largest R2, the program uses the partial
correlation coefficients to select the next variable to enter the regres-
sion. That is, the program enters the variable with the highest partial
correlation coefficient with AMES2 (given that the variable with the largest
R2 is already in the model). 'An F test is performed to determine if the
variable to be entered has a probability greater than the specified "signif-
icance level for entry." After a variable is added, the program looks at
all the variables already included in the model and computes a partial
F-statistic to determine if these variables should remain the model. Any
variable not producing a partial F significant at the specified "signif-
icance level for inclusion" is then deleted from the model. The process
then continues by determining if any other variables should be added to the
regression. The process terminates when no variable meets the conditions
for inclusion or when the next variable to be added to the model is one just
previously deleted from it. For the present analysis, all variables in the
final regression model were deemed significant at the 0.10 level of signif-
icance.
71
-------
Stepwise regressions were run to determine which of the chemical test
results (151 variables) would best predict data for the biological tests
AMES2, RAM, RODENT, grass shrimp, and freshwater fish. The Litton-grouped
bioassay results were considered here. It was not possible to run regres-
sions to predict the results of the tests WI-38, CHO, freshwater algal, and
Daphnia because of limited sample sizes. Table 24 presents the biological
test data that were available for these stepwise regressions. It is impor-
tant to note from Table 24 the limited variation in the various biological
tests: (1) AMES2 has only 3 values out of 33 that are not equal to 1; (2)
all RODENT values are equal to 1; and (3) RAM, grass shrimp, and freshwater
fish have only 15 or 16 values each and they vary from 1 to 3. This lack of
variability certainly makes stepwise regression results suspect and of
questionable value in the present context. However, the results are pre-
sented here in the interest of completeness. Also, it is important to note
in running the regressions that missing chemical data were set equal to
zero. This is certainly a questionable procedure but some compromise had to
be made because of the large amount of missing chemical data.
The results of running the stepwise regressions were as follows:
1. AMES2 (sample size = 33)
Best predicted by (the logarithm of) Bi. The regression equation
with Bi had an R2 = 0.96 (R2 = [correlation]2 and indicates the
amount of variation accounted for by the regression equation).
Because Bi accounted for almost all of the variation in AMES2,
other chemical predictor variables were not considered. However,
it is important to note that several other chemical variables by
themselves would also account for over 90 percent of the variation
in AMES2.
2. RAM (sample size = 16)
Order of chemical R2 at
variables selected each step
NH3 .53
Ca .69
GC8 .82
Sussol .87
MEG2 .91
72
-------
TABLE 24. BIOLOGICAL TEST RESULTS USING LITTON QUANTITATION
CO
STUDY
SAMPLE
AMES2
RAM
WI38
CHO
RODENT
FWALGAL
GRASS SHRIMP
FWFISH
DAPHNIA
6
6
6
6
10
10
10
10
10
10
10
10
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
02
03
04
05
09
08
07
06
04
03
02
01
OA
OB
OC
OE
OF
06
OK
OL
ON
OS
OT
OU
0V
OW
OX
OD
OH
OJ
OM
OP
OR
OY
OZ
01
4 2 <
1 3
1
1
•
1
•
1
2
2 2
1 1
1 1
1 •
1 •
1 •
1 •
1 •
1 •
1 •
1 2 2
1 2 2
1 • •
1 • i
1 » •
1 • i
1 • •
1 • •
1 • •
1 • •
1 • •
1 2 1
1 • •
1 1 1
1 • •
1 • •
1 • •
»
3
2
2
«
•
•
•
(
4
•
•
•
1
•
3
•
•
•
•
• • •
1 • •
1 • •
• •
2 1
. .
2 2
• •
• •
• •
• •
• •
• 3
• 1
• 3
• 0
• 1
• 1
• 1
• 1
• 3
• 1
• 3
• 1
• , •
• 3
• 1
•
•
•
•
•
•
•
•
•
. .
• .
, ,
• ,
2 2
• ,
1 1
• «
• •
• •
• •
• •
2 •
1 •
2 •
, .
1 •
1 •
1 .
2 •
3 •
1 •
2 •
1 •
• •
3 •
1 •
•
•
•
•
•
•
•
•
•
• = Missing value.
1 = No detectable response
2 = Low response
3 - Moderate response
4 = High response
-------
(That is, a regression equation with NH3, Ca, C8, Sussol, and MEG2
accounted for 91 percent of the variability in RAM.)
3. RODENT (sample size = 32)
No variables selected because all RODENT values = 1.
4. Grass shrimp (sample size = 15)
Order of chemical R2 at
variables selected each step
Organics .20
Acidity .37
5. Freshwater fish (sample size = 15)
Order of chemical R2 at
variables selected each step
Cr~ .29
MEG18 .54
C7 .72
4.4.2 AMES2 V. Its Component Strains
Table 25 presents a data listing of AMES2 and its component strains.
These data were used to run stepwise regressions designed to predict AMES2
from the strains. In running these regressions, TA1538M and TA1538P were
not considered because of missing data. The results were as follows (sample
size =32):
Order of strains R2 at
selected each step
TA98P .82
TA98M . 91
TA1535M . 94
Thus, three strains accounted for 94 percent of the variation in AMES2. Of
course the caveats listed previously also apply to the present analysis.
Figure 4 (p. 38) presents a plot of TA98P v. AMES2. This plot shows that
the high correlation is essentially based on only one data point.
4.4.3 Relationships Between the Biological Tests
Table 24 indicates the limited number of samples for which data are
available on more than one biological test. Accordingly, stepwise regres-
sions were not considered feasible for the purpose of predicting the results
74
-------
TABLE 25. AMES2 AND ITS COMPONENT STRAINS BY STUDY*
tn
STUDY
6
6
6
6
10
10
10
10
10
10
10
10
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
SAMPLE AMES2 TA1535M
02 4 1.2
03 1 1.1
04 1 1.2
05 1 0.9
09 ••
08 1 1.1
07 ••
06 1 1.1
04 2 0.8
03 2 1.8
02 1 1.7
01 1 1.3
OA 1 1.7
OB 1 1.6
OC 1 2.2
OE
OF
OG
OK
OL
ON
OS
OT
OU
0V
OW
OX
OD
OH
OJ
OM
OP
OR
OY
•
1.7
1.0
2.4
1.5
1.9
1.7
1.7
1.7
1.2
1.3
1.9
5.2
1.5
1.3
1.4
1.7
1.6
2.0
OZ 1 1.2
TA1535P
2.0
1.1
1.4
1.9
•
1.3
•
1.0
1.3
1.5
1.1
1.0
1.5
1.2
1.5
•
1.5
0.7
1.2
1.3
1.5
1.5
1.2
1.4
0.7
1.3
1.5
2.3
1.4
1.4
0.9
1.3
1.4
1.8
0.8
TA1537M
2.2
3.4
1.8
1.6
•
1.2
•
1.2
2.6
1.5
1.5
1.0
1.0
1.6
1.9
•
1.1
1.3
1.5
1.6
1.0
1.3
1.0
1.7
1.3
1.7
1.2
2.6
1.2
1.7
1.4
1.6
1.6
1.6
1.8
TA1537P
17.3
1.5
1.1
1.1
•
1.0
•
0.8
1.6
1.1
2.6
1.2
1.1
1.0
1.2
•
9.1
1.2
2.7
1.5
1.2
1.4
1.2
1.7
1.4
2.2
1.3
3.4
1.4
1.1
1.9
1.4
1.5
1.7
0.8
TA98M
1.5
0.9
1.3
1.4
•
1.1
•
1.1
4.2
2.6
1.0
1.3
1.2
1.2
1.0
•
1.3
1.1
1.0
1.1
1.7
1.7
1.3
1.5
0.8
1.3
1.6
2.6
1.3
0.8
0.9
1.0
1.5
1.5
0.9
TA98P
233
1.4
1.1
0.8
•
1.2
•
0.8
2.1
1.4
1.3
1.3
0.8
1.0
3.2
•
1.1
1.4
0.9
1.1
1.3
2.1
1.1
0.9
1.1
1.3
0.9
2.6
0.9
1.3
1.3
1.0
1.0
1.6
1.8
TA100M
1.4
1.1
1.2
1.1
•
1.3
•
1.3
1.3
1.6
1.3
1.2
1.2
1.1
1.2
•
1.2
1.2
1.0
1.0
1.3
1.3
1.2
1.2
1.2
1.1
1.2
1.6
1.1
1.1
1.1
1.3
1.2
1.1
1.2
TA100P TA1538M TA1538P
4.8 1.5 33.9
1.0 1.4 1.2
1.1 1.5 1.0
1.1 1.2 1.1
• • •
1.0 1.5 1.5
• • •
0.8 0.8 0.8
0.8 3.9 1.5
1.1 6.9 2.7
1.0 1.3 0.8
1.2 2.0 0.8
1.1
1.1
1.0
£
1.3
1.1
0.9
1.1
1.2
1.2
1.0
1.1
1.1
1.3
1.0
1.2
1.0
1.0
1.0
1.2
0.9
1.1
1.1
= Missing value.
1 = No detectable response
2 = Moderate response
4 = High response
'Numbers reported for component strains are maximum mutagenic ratios (see pp. 103-107 for details).
-------
of one biological test from the other tests. Unfortunately, the limited
variation and small sample sizes in the tests make these correlations rather
small except for the grass shrimp-freshwater fish combination.
4.5 PROBIT ANALYSIS
A transformation can be made to turn nonlinear regressions between two
variables into linear regressions. The logarithm of the variable represent-
ing dosage and the probit of the variable representing response are used.
The probit procedure assumes a normal distribution of the data, and
calculates maximum likelihood estimates of the slope, intercept, and natural
(threshold) response rates for biological data. A thorough description of
the probit analysis procedure can be found in Finney, 1952.13 Probit analy-
sis was performed on dose-response data for ATP content and viability in the
RAM test. Data from studies 06, 10, and 19 were considered together. The
average ATP content and average viability at five dosage levels were entered
into the SAS PROBIT procedure.
The SAS PROBIT Procedure was run at Triangle Universities Computation
Center (TUCC) and plots of all data on the same scale were examined and
compared with the probit analysis. A check (V) on Table 26 indicates that a
toxic response was (visually) evident from examination of the log-dose-
response plot; that is, viability or ATP level decreased as dosage of the
sample increased, in comparison to the controls.
The slope of the probit line, as shown in the SAS printout, is given in
Table 26; however, the slope value was not a good indicator for this test
because each sample had a different intercept. For instance, a sample with
high toxicity at the initially low dose could have a small slope and a high
toxicity. A sample that showed a response at a moderate or high dose might
have a greater slope value yet actually be less toxic than the first case.
Where a dose-response plot did not show a toxic response or where
confidence values on the LD50 at the 95 percent level (X ± 2S) showed a very
high or infinite spread, and where the absolute values of slope value were
very low (<0.15), the toxicity value (1/LD50) was entered as zero. The
toxicity value was entered as 1/LD50 so that very low LD50 values (essen-
tially nontoxic by this test) could be entered as zero. By using the recip-
rocal of the LD50 value, it was also possible to obtain a positive value for
76
-------
TABLE 26. PROBIT DATA ANALYSIS, RAM CYTOTOXICITY TEST
Viability
Sample
Separator tar
Separator liquor
Gasifier ash
Cyclone dust
Fine paniculate
Cyclone dust
Cyclone leachate
Bed leachate
Plant A, filtered
Plant B, filtered
Plant C, filtered
Plant D, filtered
Plant D, unfiltered
Plant F, filtered
Plant F, filtered
(retest)
Plant H, filtered
Plant J, filtered
Plant K, filtered
Plant L, filtered
Plant M, filtered
Plant M, unfiltered
Plant N, filtered
Plant N, unfiltered
Plant P, filtered
Plant R, filtered
Plant T, filtered
Plant U, filtered
Plant W. filtered
Plant X, filtered
Plant Y, filtered
Plant Y, unfiltered
Plant Z, filtered
Plant G, filtered
Plant V, filtered
Plant P, unfiltered
Encoded
sample no.
0602
0603
0604
0605
1003
1006
1007
1009
190AF
190BF
190CF
190DF
190DU
190FF
190HF
190HF
190JF
190KF
190LF
190MF
190MU
190NF
190NU
190PF
190RF
190TF
190UF
190WF
190XF
190YF
190YU
190ZF
190GF
190VF
190PU
Slope
- .664
-1.539
-0.042
-0.081
-0.986
-1.606
-0.158
-0.059
-0.028
-0.0406
-0.258
-0.033
-0.007
0.033 •
-0.109
-0.019
-0.107
Plot shows
LD50 toxieity
5.495
2.787
36.930
12.889
7.079
7.264
10.918
33.346
29.410
39.128
5.915
46.960
109.269
-39.710
7.073
63.273
13.050
V
V
—
--
V
V
V
V
—
—
—
0.013-118.194
0.402
-0.219
-0.011
-0.146
-0.396
-0.096
-0.034
-0.097
0.072
-0.181
0.534
-0.228
-0.332
-0.284
~Q
~Q
-o
-1.422
6.016
108.785
12.254
7.481
25.600
53.844
-17.091
-12.568
8.910
-0.364
7.410
8.316
7.619
No probit
No probit
No probit
V
—
V
—
—
—
V
V
V
V
—
—
1/LOM
0.182
0.359
0
0
0.141
0.138
0.092
0
0
0
0.169
0
0
0
ot
ot
0
0
0*
0.166
0
0
0.134
0
0
0
0
0.112
0*
0.135
0.120
0.131
0
0
0
Slope*
-1.216
High value
-0.118
--0
- 0.651
-0.310
-0.199
- 0.252
- 0.207
-0.184
- 0.779
- 0.148
sO
-0.190
- 0.838
- 0.191
afl
- 0.248
- 0.584
- 0.605
- 0.286
- 1.122
- 0.578
= 0
- 0.140
-0.210
0.007
- 0.565
-0.244
-Q
- 0.109
- 0.439
0.164
0.105
-3.44
ATP
Plot shows
LDso toxieity
6.049
No probit
15.481
No probit
6.405
8.089
7.807
8.182
8.631
11.391
6.153
21.401
No probit
8.768
5.586
6.745
No probit
11.859
5.738
8.039
5.944
4.744
4.756
No probit
14.963
7.604
-90.553
6.569
7.597
No probit
9.248
8.754
11.545
1.600
6.187
V
V
—
--
V
N/
v/
V
V
V
V
—
V
V
V
V
V
V
V
V
V
V
—
V
—
V
V
—
V
V
—
V
1/LD50
0.165
N/A
0
0
0.156
0.124
0.128
0.122
0.159
0.088
0.163
0
0
0.114
0.179*
0.148*
0
0.084
0.174
0.124
0.168
0.211
0.210
0
0
0.131
0
0.152
0.132
0
0
0.114
0.087
0
0.161
•If |slope|<0.15 or plot does not show toxic response and confidence limits infinite, entered 1/LD5Q value as0 in Fortran program.
* Entered average of values.
Confidence limits not <*, but values negative at-25 (lower limit).
77
-------
the correlation; that is, as toxicity increases, the numerical value for
toxicity increases.
There had been some speculation as to the effects of field-filtering
samples on the toxicity (biological response) of these samples. From the
data set for the textile effluents, Study 19, there does not seem to be a
pattern of toxicity values showing that unfiltered samples are more toxic
than filtered samples. Table 27 shows the 1/LD50 values for filtered and
unfiltered effluents for the five textile effluents analyzed in this manner.
In Table 27, there is no evident trend showing filtered samples to be less
toxic than unfiltered samples, as indicated by the RAM test.
4.6 ANALYSIS OF THE DATA SET THROUGH USE OF MODELS
Various systems or models have been proposed to aid in the interpreta-
tion and analysis of data from the Environmental Assessment Program. All of
these systems are based on the MEG/MATE values,14 1S on Threshold Limit
Values (TLV's), or on legal emission standards.
Monsanto Research Corporation (MRC), in conjunction with IERL, devel-
oped the Source Severity Model, which ratios levels of pollutants found by
analysis of effluent samples to the acceptable concentrations of these
pollutants. The source severity factor (S) is defined as "the ratio of the
calculated maximum ground level concentrations of the pollutant species to
the level at which a potential environmental hazard exists."16 The maximum
ground level concentration for gaseous emissions from point sources is
calculated by the simple Gaussian Plume equation. The acceptable pollutant
level is estimated as the TLV/300 for those pollutants for which no Federal
regulations exist.
In these cases, S is calculated as:
c _ 5.50 ,
5 ~ (TLV)h2
where
TLV = threshold limit value, g/m3,
h = stack height, meters.
The acceptable pollutant level is the ambient air quality standard for
noncriteria pollutants.
78
-------
TABLE 27. TOXICITY * OF FILTERED AND UNFILTERED
TEXTILE EFFLUENTS
Toxicity unfiltered >
Plant Unfiltered Filtered toxtcity filtered?
Cell viability
D
M
N
Y
P
ATP
D
M
N
Y
P
0
0
.134
.120
N/A
0
.168
.210
0
.161
0
.166
Q
.135
0
0
.124
.211
0
0
SAME
NO
YES
NO
N/A
SAME
YES
NO (wSAME)
SAME
YES
•Toxicity expressed as 1/LD5Q in the RAM test.
79
-------
Equations for calculating S for the five criteria air pollutants are as
follows:
Participates S = 70 Qh"2
SOX S = 50 Qh"2
NOV S = 315 Qh"2'1
-2
Hydrocarbons S = 162.5 Qh
CO S = 0.78 Qh"2
where Q = emission rate, g/s.
For discharges to water, the source severity factor is based on the
drinking water standards and is calculated as follows:
,. VD * Wz
¥ '
where
VQ = discharge flow rate, mVs;
CQ = discharge concentration, g/m3;
Sfi = leachable solid waste generation, g/s;
f, = fraction of the solid waste to water;
fy - fraction of the material in the solid waste;
VR = river flow rate, m3/s;
D = drinking water standard, g/m3.
It can be seen readily that all of the source severity equations follow
the same general concept:
Concentration found in x Dilution
Estimation pf hazard _ effluent or emission factor
or severity "~Estimation of acceptable level
The Source Assessment Models (SAM's) developed for the IERL by the
Acurex Corporation follow this same concept and expand the list of materials
for which estimations of an acceptable environmental discharge level are
available. This expanded list, herein called the MEG/MATE value list,14 is
was prepared for IERL by the Research Triangle Institute. This comprehen-
sive list incorporates drinking water standards, emission standards, TLV
information, and other information available from whole animal tests (L050's)
and other toxicity tests. MATE values are estimations of an acceptable
discharge level and MEG values are estimates of permissible levels for
long-term exposure in the environment. These values are given as concentra-
tions, for health-related and ecology-related concerns, for solid, liquid,
80
-------
and/or gaseous forms of these compounds for which information is available.
The MEG/MATE value list is being revised and expanded as more information
from research efforts becomes available.
The SAM's are comprised of a series of models. SAM/IA,17 the simplest
model, is a ratio of concentration of material found to the appropriate MATE
value for that material. If a MATE value is not available for a certain
material, the lowest MATE value in the MEG category (category of similar
compounds, such as methane, ethane, propane) is used; this is a worst-case
assumption. In the SAM/IA model, these MATE ratios can be summed to give a
potential degree-of-hazard value (PDOH) for any given effluent or emission
stream. Multiplying the PDOH by the discharge rate gives the toxic unit
discharge rate (TUDR), a unitless measure that can be used to compare dis-
similar effluents or emission streams.
SAM/IB, the Source Assessment Model for bioassays, is still being
developed. An early publication on SAM/IB18 described appropriate reporting
formats for many of the bioassays used in Level 1 of the Environmental
Assessment Program.
SAM/I,19 recently circulated in draft form, incorporates dilution
factors appropriate to the source of discharge to the concepts developed in
SAM/IA: MEG ratios, PDOH's and TUDR's. SAM/I is appropriate for summariz-
ing Level 1 or Level 2 environmental assessment data. The SAM models are
intended as uniform protocols for data interpretation and can be used to
rank individual discharges, establish priorities, identify problem areas,
and provide a means of comparing control options.
The data set of samples from the three pilot studies was entered in the
SAM/IA model. Data from Study 15, the SASS evaluation, were not entered
because many chemical parameters were missing. MATE ratios were calculated
for all chemical parameters for which analysis had been performed. A pos-
sible source of error here could be the inclusion of more ratios for some
samples where more analyses were performed; but the establishment of a
uniform protocol for analysis should eliminate this error in future work.
The ratios were summed for each sample to give a PDOH value for that sample.
The health-related and ecology-related PDOH values for each sample are
presented in Tables 28 and 29. These tables also show the bioassay results
in the Litton-grouped forms,11 reported as nondetectable (ND), low (L),
moderate (M), or high (H).
81
-------
TABLE 28. HEALTH EFFECTS: SUMMARY OF CHEMICAL
AND BIOLOGICAL TEST RESULTS*
Study
Coal gasifier
Fluidized-bed combustor
Coal gasifier
Fluidized-bed combustor
Textile effluent
Textile effluent
Textile effluent
Textile effluent
Coal gasifier
Textile effluent
Textile effluent
Fluidized-bed combustor
Fluidized-bed combustor
Textile effluent
Textile effluent
Fluidized-bed combustor
Fluidized-bed combustor
Textile effluent
Fluidized-bed combustor
Textile effluent
Textile effluent
Textile effluent
Textile effluent
Textile effluent
Textile effluent
Fluidized-bed combustor
Coal gasifier
H - HIGH toxicity rating.
M = MODERATE toxicity
L - LOW toxicity rating.
Sample
Separator liquor
Bed reject leachate
Separator tar
Cyclone leachate
Plant L
Plant W
Plant T
Plant N
Gasifier ash
Plant U
Plant S
Fine paniculate*
Feed coal
Plant V
Plant F
Cyclone dust
Coarse particulates
Plant B
Bed reject
Plant K
Plant A
Plant C
Plant G
Plant X
Plant E
Dolomite
Cyclone dust
rating.
N = No detectable toxicity.
la
POOH
6.10 x 107
1.16 x 106
1.01 x 10*
6.66 x 10s
8.36 x 103
4.99 x 103
3.35 x 103
3.28 x 103
1.72x 103
8.86 x 102
8.35 x 102
7.00 x 102
5.33 x 102
5.19 x 102
4.52 x 102
3.80 x 102
3.43 x 102
3.30 X 102
1.58x 102
1.43x 102
1.14 x 102
950 x 101
9.40 x 10l
3.60 x 101
3.30 x 101
2.30 x 101
1.70 x 101
1a=PDOH
1b=PDOH
2 =AMES
3 =RAM-
4 =WI-38
5 -RAT"
1b 2
POOH AMES
7.7854 N
6.0644 t
6.0041 H
5.8238 t
35224 N
3.6984 N
3.5245 N
3.5161 N
3.2343 N
2.9474 N
25217 N
2.8451 L
2.7267 N
2.7152 N
2.6551 N
2.5798 N
2.5353 L
2.5185 N
2.1987 N
2.1553 N
2.0569 N
1.9956 N
15731 N
1.5563 N
1.5185 N
1.3617 N
1.2304 N
= potential degree
= potential degree
3
RAM
M
N
L
N
M
L
M
M
N
N
t
L
N
N
L
N
N
N
N
N
N
L
N
L
N
N
N
of hazard.
of hazard
4
WI-38
t
t
t
t
L
t
t
L
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
(LOG10)
5
RAT
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
» mutagenicity test
• rabbit aveolar macrophage.
= human embryonic lung cells.
• rodent acute toxicity.
•Subsequent to the completion of this work, the term "Potential Degree of Hazard" or POOH was changed to
"Discharge Severity" or DS.
tData not available.
82
-------
TABLE 29. ECOLOGICAL EFFECTS: SUMMARY OF CHEMICAL
AND BIOLOGICAL TEST RESULTS*
6a
Study
Fluidized-bed combustor
Coal gasifier
Fluidized-bed combustor
Coal gasifier
Fluidized-bed combustor
Textile effluent
Fluidized-bed combustor
Coal gasifier
Fluidized-bed combustor
Textile effluent
Textile effluent
Textile effluent
Textile effluent
Fluidized-bed combustor
Textile effluent
Textile effluent
Textile effluent
Textile effluent
Textile effluent
Textile effluent
Textile effluent
Textile effluent
Coal gasifier
Textile effluent
Fluidized-bed combustor
Fluidized-bed combustor
Textile effluent
H = HIGH toxicity rating.
M= MODERATE toxicity
L * LOW toxicity rating.
N - No detectable toxicity
Sample
Bed reject leachate
Separator liquor
Fine particulates
Separator tar
Feed coal
Plant N
Coarse particulates
Cyclone dust
Cyclone leachate
Plant L
Plant C
Plant B
Plant E
Cyclone dust
Plant V
Plant F
Plant T
Plant U
Plant A
Plant G
Plant K
Plant S
Gasifier ash
Plant W
Bed reject
Dolomite
Plant X
rating.
.
POOH
1.50
6.94
1.01
2.90
2.67
1.51
8.93
8.06
6.20
6.07
3.62
3.05
2.91
2.59
X
X
X
X
X
X
X
X
X
X
X
X
X
X
2.54 x
2.23
X
1.56x
1.46
1.36
1.15
X
X
X
1.Q6X
9.48
9.30
5.62
2.68
4.78
X
X
X
X
X
3.43 X
10*
107
10*
105
10s
10s
104
104
104
104
104
104
104
104
104
104
10"
104
ID4
104
104
103
103
103
103
102
102
6b
POOH
8.1763
7.8415
6.0055
5.4624
5.4267
5.1806
4.9513
4.9065
4.7922
4.7830
4.5589
4.4836
4.4638
4.4137
4.4051
4.3482
4.1939
4.1646
4.1345
4.0618
4.0249
35770
3.9683
3.7494
3.4275
2.6794
2.5353
7
FW
8
FW
9 10
SW
ALGAL FISH DAPHNIA ALGAL
L
H
t
t
t
M
t
t
L
N
N
L
M
t
N
N
N
N
N
N
N
N
t
N
t
t
N
6a =
6b =
7 =
8 =
9 =
10 =
11 =
L
H
t
t
t
L
t
t
N
L
L
N
N
t
L
N
L
N
L
L
N
N
t
L
t
t
N
PDOH =
PDOH =
L
H
t
t
t
H
t
t
N
L
L
N
M
t
M
L
N
L
M
L
N
t
t
M
t
t
N
t
M
t
t
t
M
t
t
t
M
L
N
L
t
L
L
L
N
t
L
L
N
t
L
t
t
N
11 12
SW GRASS
FISH SHRIMP
N
H
t
t
t
L
T
t
N
N
L
N
N
t
t
N
L
N
L
N
N
N
t
L
t
t
N
N
H
t
t
t
L
t
t
L
N
L
N
N
t
t
N
L
N
L
N
N
N
t
L
t
t
N
potential degree of hazard.
potential degree of hezard
(LOG10).
FW ALGAL = fresh water algal.
FW FISH = fathead minnow.
FW DAPHNIA = fresh water invertebrate.
SW ALGAL* marine algal.
SW FISH = marine sheepshead minnow.
12 = SW GRASS SHRIMP = saltwater shrimp.
•Subsequent to the completion of this work, the term "Potential Degree of Hazard" or POOH was changed to "Discharge Severity" or
OS and the response ranges for aquatic ecological tests were redefined based on recommendations from EPA ecologies.
not available.
83
-------
Several observations can be made from Tables 28 and 29. There is
generally good agreement between health-related PDOH values in Table 28 and
health-related biotest results. Such an agreement is also noted for the
ecology-related tests in Table 29 but there are notable exceptions to this
pattern of responses; for example, Plant W, Table 29, shows positive bio-
assay responses for five of the six ecological biotests but its POOH value
is the fourth from the lowest in a series of 27 samples. For the purpose of
comparing these two methods (the chemically based PDOH values and the actual
bioassay results) to determine hazard, linear regression analyses were
performed to compare each bioassay with the corresponding PDOH value. The
correlations are presented in Table 30.
It can be observed from Tables 28 and 29 that neither chemistry alone
nor bioassay alone can predict the overall hazard of an effluent or emis-
sion, but that these two test systems, taken together, can thoroughly evaluate
a material both for content of known hazardous chemicals and for effect of a
variety of biological receptors. It should be noted that each bioassay was
designed to measure a different endpoint, so it is not appropriate to sum
the various response levels across bioassays.
4.7 OTHER INVESTIGATORS' ANALYSES OF THIS DATA SET
Two other organizations investigated these data from the four IERL
environmental assessments for patterns and for correlations among the vari-
ous parameters. Dr. James R. Hoyland of Battelle-Columbus Laboratories
performed a computerized pattern recognition analysis on the data and Don
Lewis of EPA-RTP analyzed the data for nonparametric correlations.
A preliminary report from Dr. Hoyland on his work, presented in Appen-
dix D, deals with the analysis of this data set via a computer program for
recognition of patterns and relationships. This pattern recognition program
performs computations similar to classical multivariate analysis on grouped
or "discretized" data, determining whether one or more groups of parameters
are correlated with one or several other groups of parameters. Dr. Hoyland
found that this data set was not sufficiently large or, in many cases,
sufficiently complete to apply his program for pattern recognition and gain
statistically significant results. As discussed on page D-6 of this document,
the following relationships were very tentatively suggested by the data:
84
-------
TABLE 30. ASSOCIATIONS BETWEEN BIOASSAY RESULTS AND
HAZARD ESTIMATIONS BASED ON CHEMICAL ANALYSIS
Correlation coefficients
Health-related Ecology-related
(PDOH v. bioassay*) (POOH v. bioassay*)
AMES .456 (n=25) FW ALGAL .609 (n=18)
RAM .427 (n=26) FW FISH .599 (n=18)
WI-38 NA (n=2) DAPHNIA .485 (n=17)
RAT NA (y-1) SW ALGAL .710
-------
N03 and Co related to RAM.
LC8, Ca, and Co related to RAM-ATP.
S related to freshwater algae.
Mr. Don Lewis, a biostatistician at EPA-RTP, has also examined this
data set for possible relationships between chemical concentrations and
biological responses. Mr. Lewis suggested in his report, which is presented
in Appendix B, that the rank (Spearman) correlation was more appropriate for
analysis of this data set because the assumptions of normal distribution and
linearity inherent in linear (Pearson) correlation could not be made for
environmental data or for biological response. Therefore, he used only the
rank (Spearman) correlation in his analysis of the data.
Mr. Lewis grouped the data as textile and nontextile, and he observed
substantially lower concentrations of the measured chemical parameters in
the textile samples. He stated that the data show "high degrees of vari-
ability" even when grouped in this manner. From his correlation analysis,
Mr. Lewis concluded that there was no strong association between chemical
and biological results. He theorized that this might be due to the statis-
tically small sample set or to the low concentrations of many chemicals
(below the biological response level). With these qualifiers, the associa-
tions described in Table 31 (rank [Spearman] correlations) were found in the
nontextile data. Table 32 describes relationships found in the textile
data.
Mr. Lewis examined the relationships among the various strains of
Salmonella bacteria used in the Ames test, and he observed that some muta-
genic ratios decreased as chemical concentrations increased. This would
tend to support the hypothesis that a toxic effect was caused by a chemical
constituent(s) as discussed in the section of this report on interpretation
of the Ames test. Mr. Lewis performed multiple regression analysis on the
Ames strain ratios and found that the best pair of predictors was TA98
(without activation) and TA1538 (with activation), explaining 90 percent of
the variation.
In examining the rank correlations among the other biotests, Mr. Lewis
found a positive association between freshwater fish and grass shrimp, and a
negative correlation between freshwater fish and saltwater algae. As men-
tioned previously, a more thorough discussion of his work can be found in
Appendix B.
86
-------
4.8 ENGINEERING DATA
Engineering data were available for chemical concentrations of effluent
from the FBC mini pi ant. Although a large amount of data has not been amassed,
enough information has been gathered to begin statistical analysis. Out of
163 chemical-source combinations, 128 (79 percent) were measured in Runs 2,
4, and 5. A t-statistic was utilized to determine whether there was a
significant difference in concentrations of chemicals in the effluent gases
when coal was burned at different temperatures. Run 2 was made at 805° C;
Runs 4 and 5 were made at 890° C and 895° C, respectively.
Although conclusions are tentative, a trend in the data has appeared.
Significant differences are shown in 24 of the 128 chemical-source combina-
tions examined (Tables 33 and 34). In 20 of these cases (83 percent), a
significantly larger amount of a given chemical compound was found at the
lower temperature. In the remaining four cases (17 percent), a significantly
greater quantity of a particular material was found at the higher temperature.
Data for an additional 10 chemical-source combinations appear to show a
significant difference by temperature; however, a t-statistic-cannot be
calculated since the estimate of variance was zero. These combinations are
noted in Tables 33 and 34 by a dagger.
Analysis of the data showed that the most significant difference between
the high- and low-temperature runs occurred in the GC10 measurement for
particulates >3p; the lower temperature produced more particulates (0.99
confidence level). In terms of classification by chemical, the chloride
anion measurement showed several significant differences. The measurements
for bed reject leachate and particulates >3u showed differences at the 0.95
confidence level; measurements for cyclone dust and particulates <3u showed
differences at the 0.90 confidence level. In terms of effluent stream,
particulates >3u showed the most significant difference by temperature. In
addition to the GC10 measurement noted above (0.99 level of confidence), the
following were significant at the 0.95 level of confidence: GC8, GC9, GC13,
and chlorine anion. The LC6 measurement was significant at the 0.90 confi-
dence level. The quality of the data would be improved by recording addi-
tional runs, both at the lower and higher temperatures mentioned above.
87
-------
TABLE 33. EFFECT OF ENGINEERING CONDITIONS IN CHEMICAL CONCENTRATIONS* OF
EFFLUENTS FROM FBC MINIPLANT (CLASSIFIED BY CHEMICAL CATEGORY)
Chemical
CO
S02
°2
C02
NOX
GC7
GC8
GC9
GC10
GC13
LC2
UC3
LC5
LC6
LC7
cr
f~
S04
SOs
NOa
S
Source
Continuous monitor
Continuous monitor
Continuous monitor
Continuous monitor
Continuous monitor
Cyclone dust
Part iculate> 3^t
Participate > 3/J
Paniculate >SM
Bed reject
Particul8te>3j/
Cyclone dust
Cyclone dust
Paniculate <3p
Paniculate >3/j
SASSwesh
Bed reject
Paniculate <3p
Paniculate >3^
SASSwash
Bed reject leachate
Paniculate > 3/J
Cyclone dust
Paniculate <3/u
Paniculate <3ji
Bed reject leachate
Bad reject
Bed reject leachate
Cyclone dust
Cyclone dust leachate
Bed reject leachate
Cyclone dust leachate
Paniculate 3/u
Cyclone dust
Run 2
805° C
130
151
7.6%
9.8%
136
117
340
193
415
13
265
' 73
29
122
238
710
70
139
926
426
176
970
240
1,000
150
4.0
323,000
1,500
111,000
900
3.1
2.3
40
100
Run 4
890° C
56
41
6.0%
12.5%
124
15
4
12
0
0
1
12
0
0
0
143
0
0
0
0
640
50
300
20
320
2.1
261,000
200
142,000
200
1.5
1.5
10
30
RunS
895° C
53
29
5.5%
13.1%
119
41
30
17
5
4
6
0
0
0
35
0
0
0
0
0
620
70
310
110
310
1.7
277,000
200
142,000
200
1.3
1 :i
1(1
Ml
Level of
confidence
.95
.90
.80
.80
.80
.80
.95
.95
.99
.80
35
.90
t
t
.90
. .80
t
t
t
t
.95
.95
.90
30
.95
.80
.80
t
t
t
911
HII
t
HII
* Units ere in ppm unless otherwise noted.
t Estimate of variance is 0, so t-statistic cannot be utilized.
88
-------
TABLE 34. EFFECT OF ENGINEERING CONDITIONS IN CHEMICAL CONCENTRATIONS* OF
EFFLUENTS FROM FBC MINIPLANT (CLASSIFIED BY EFFLUENT STREAM)
Source
Bed reject
Bed reject leachate
Cyclone dust
Cyclone dust leachate
Paniculate <3ju
Part icul ate -3iJ
SASS wash
Continuous monitor
Chemical
GC10
S04
LC7
cr
503
F
SO 4
LC2
cr
GC7
S"
LC3
804
503
S04
f~
cr
LC5
LC7
N0§
GC10
GC8
GC9
GC13
cr
LC6
LC7
LC6
LC7
CO
S02
02
C02
Nl)v
Run 2
805° C
13
323,000
70
176
3.1
4.0
1,500
73
240
117
100
29
111,000
2.3
900
150
1,000
122
139
40
415
340
193
265
970
238
926
710
426
130
151
7.6%
9.8%
136
Run 4
890° C
0
261,000
0
640
1.5
2.1
200
12
300
15
30
0
142,000
1.5
200
320
20
0
0
10
0
4
12
1
50
0
0
143
10
56
41
6.0%
12.5%
124
Run 5
895° C
4
277,000
0
620
1.3
1.7
200
0
310
41
50
0
142,000
1.3
200
310
110
0
0
10
5
30
17
6
70
35
0
0
0
53
29
5.5%
13.1%
11!)
Level of
confidence
.80
.80
t
.95
.90
.80
t
.90
.90
.80
.80
t
t
.80
t
.95
.90
t
t
t
.99
.95
.95
.95
.95
.90
t
.80
t
.95
,.90
.80
.80
III)
"Unitsan.' in ppin unless otliiirwisu
^Estimate of variance isO, so t-statistic cannot b« calculated
89
-------
4.9 ORIGINAL RESULTS COMPARED TO THE LITTON QUANTIFICATION SCHEME
The originally reported results or simple transformations of the raw
data, such as sums or ratios, were compared with the grouped values, herein
called the Litton scheme.11 The Litton scheme groups responses from bio-
tests and places these responses in one of four categories: nondetectable,
low, moderate, or high. The four response categories were entered into the
data set as 1, 2, 3, and 4, respectively. For the purpose of comparing
these two methods of quantifying the bioassay results, all data from all
studies were considered. Even better correlations would be expected if the
samples were grouped by study; such samples could be expected to be more
similar than samples from diverse industrial sources. Table 35 shows the
data pairs (correlations), the linear correlation coefficient, the signifi-
cance factor, and the number of data pairs used in the correlation.
Because there were no positive results on the RAT lethality test, and
since there were only two data pairs for the freshwater fish 24-hr, the
freshwater fish 48-hr, and the Daphm'a tests, these results were not appro-
priate for correlation. All other correlation results except the CHO test
showed "good" correlations by the criteria defined for the rest of the
statistical analysis in this study (correlation coefficient >0.50, signifi-
cance factor <0.05); the CHO test had only four data pairs so these results
should not be considered indicative of, or conclusive of, a lack of correla-
tion. Generally, these results show good agreement between the originally
reported values such as LD50, EC20, etc., and the summary-grouping scheme
proposed by Litton.
4.10 CORRELATIONS AMONG AMES/SALMONELLA STRAINS
The five strains of Salmonella typhimurium used in the Ames test were
considered; these are TA1535, TA1537, TA1538, TA98, and TA100. The maximum
mutagenic ratio (ratio of maximum revertants to control) for each strain,
with and without metabolic activation, was entered in the data set. These
numbers were then compared using the linear and rank correlation coefficients
There are two groups of Ames S. typhimurium bacteria detecting two kinds of
mutations: TA1535 and TA100 are sensitive to base-pair substitution and
TA1537, TA1538, and TA98 are sensitive to frameshift mutation.20 The data
were also examined for these expected patterns of response. Data from all
studies were considered together in this effort.
90
-------
TABLE 35. COMPARISON OF LITTON RANKING SCHEME AND
ORIGINALLY REPORTED VALUES
Correlation
AMES 2 v. AMES 1
RAM v. RAM1000
RAMv.VIABINO
RAM v. ATP
RAMv. PROBVI
RAMv. PROBATP
WI-38v.WI-38600
CHO v. CHOMPS
RODENT v. RATLD50
RODENT v.OTHRAT
RODENT v.RATWT
FWALGALv.FWALGAL
GRSSHPv. GSHMP24
GRSSHPv. GSHMP48
GRSSHPv. GSHMP96
FWFSH v. FWFSH24
FWFSH v. FWFSH48
FWFSH v. FWFSH96
DAPNAv.DAPH24
DAPNAv. DAPH48
Linear
correlation
coefficient
.8417
- .81886
- .74085
- .74613
.69660
.71964
-.98012
- .87315
- .0000
- .0000
- .0000
- .0000
- .83622
- .84142
- .99027
-1.0000
-1.0000
- .74629
-1.0000
-1.0000
Significance
factor
.0001
.0001
.0010
.0009
.0118
.0125
.0199
.1269
1.000
1.000
1.000
—
.0001
.0001
.0001
—
—
.0022
—
—
Number of
data pairs
23
16
16
16
12
11
4
4
25
25
26
2
15
15
15
2
2
14
2
2
Comment
No positive results
No positive results
No positive results
n<3
n<3'
n<3
n<3
n<3
91
-------
Table 36 shows the linear correlations found when all the Ames test
bacterial strains were compared. The checks (V) indicate "good" correla-
tions as defined earlier in the study. The boxes (n) in the upper right
corner of each matrix space show where "good" correlations could be expected
based on groupings for base-pair substitutions or frameshift mutations. The
first number listed in each matrix space is the correlation coefficient, the
second number is the significance factor, and the third is the number of
data pairs available for this correlation (n). From the table, four "good"
correlations that were anticipated were not found; these were 1538M v. 1537M,
98M v. 1537M, 100M v. 1535M, and 100P v. 1535P. (The number designates
Ames' strain, P indicates plus S-9 microsomal preparation, and M indicates
minus S-9.) A number of other unexpected "good" correlations were found.
It should be remembered in analyzing these data that, of the 32 samples
tested, only 3 showed a mutagenic response. Of these three samples, two
were classified as low mutagenicity and only one as high mutagenicity.
Therefore, the actual data set considers 4 points, 3 positives, and 29
samples where the mutagenic ratio was too low to indicate mutagenicity (by
the criteria of a threefold increase in revertants over controls).
The actual maximum mutagenic ratios were used here rather than some
interpretation of what constitutes a mutagenic response. Although Ames
showed a high correlation between positives in his Salmonella assay for
mutagenicity and positives in classical whole animal tests for carcinogeni-
city, his data involve single pure compounds rather than the mixtures dis-
cussed above. Because more detailed laboratory studies are needed to define
and accurately interpret these relationships for complex mixtures, mutagenic
ratios were used as a numerical representation of mutagenic response for the
mixtures.
These Ames test results are not corrected for toxicity to the bacterial
culture, apparently causing anomalies in the interpretation of some raw
data. A higher mutagenic ratio (i.e., >3) is noted at a lower dosage but
the mutagenic ratio decreases as dosage increases. This could be an actual
valid observation and interpretation, or it could mean that the sample is
toxic to the bacterial culture, killing bacteria before they can mutate and
form colonies. Without toxicity data, it is impossible to know which is the
case. (See Tables 37 and 38.) Suggestions for further investigation are
discussed in Chapter 5.
92
-------
TABLE 36. LINEAR CORRELATIONS AMONG THE AMES TEST STRAINS*
<£>
co
1535M
1535P
1537M
1537P
1538M
1538P
98M
98P
100M
100P
1535M
-
.54
.0012
32
.20
.28
32
.04
.83
32
.43
.21
10
-.0029
.99
10
.22
.22
32
-.042
.82
32
42 D
'.02
32
-.057
.76
32
1535P
-
—
.28
.12
32
.40
.02
32
.10
.78
10
.65 V
.04
10
.40
.023
32
.37
.036
32
.37
.037
32
.38 a
.033
32
1537M
-
—
_
.20
.27
32
.055o
.88
10
.24
.50
10
.37 n
.040
32
.30
.096
32
.16
.37
32
.20
.26
32
1537P
-
—
—
—
-.14
.70
10
.99 [7]
.0001
10
.050
.79
32
.88 [7]
.0001
32
.24
.19
32
.90
.0001
32
1538M
-
—
_
_
—
-.083
.82
10
.700
.026
10
-.11
.77
10
.73 V
.017
10
-.12
.74
10
1538P 98M 98P 100M 100P
- - - -
_ _ _ _ _
_ _ _ _ _
_ _ _ _ _
_ _ _ _ _
_ _ _ _ _
-.019
.96
10
.998 0 -075
.0001 .68 - -
10 32
.33 .54 v/ .29
.34 .0015 .11
10 32 32
.99 .0035 .98 v -26
.0001 .98 .0001 .15
10 32 32 32
*First number is correlation coefficient; second number is significance factor; third number isN.
P = With (+) activation; M = Without (-) activation.
v/= Good correlations; ID = Expected to find good correlation.
-------
TABLE 37. COAL GASIFIER, CYCLONE DUST SAMPLE,
STRAIN TA1538
Average Average
/-tg of Compound revertants per rcvertants per
per plate plate without S-9 plate with S-9
0 (OMSO/Control) 13 21
10 ' 19 20
50 18 17
100 13 22
500 10 18
1000 13 20
5000 6 16
TABLE 38. FLUIDIZED-BED COMBUSTOR.
COAL SAMPLE. STRAIN TA1537
/jg of Compound
per plate
Average
revertants per
plate without S-9
0 (DMSO/Control) 6
10 6
50 5
100 4
500 4
1000 2
5000 1
Average
revertants per
plate with S-9
5
6
5
5
5
4
4
94
-------
4.11 COMPARISON OF SCHEMES FOR RANKING SAMPLES USING LEVEL 1 DATA
For the textile effluent study's bioassay program evaluation, an expert
panel approach was used. The expert panel was the Bioassay Subcommittee of
the Environmental Assessment Steering Committee. Table 39 shows the textile
effluent study's samples as they were ranked:
1. By the expert panel,
2. By descending ecology-related PDOH values,
3. By descending health-related PDOH values,
4. By increasing weight gain in the rodent acute toxicity test,
5. By increasing 14-day EC2o values in the freshwater algal system,
6. By increasing 96-hr LCSO values in the freshwater fish system
(fathead minnow),
7. By increasing 48-hr EC50 values in the Daphnia system,
8. By increasing 96-hr LC50 values in the saltwater fish system
(sheepshead minnow),
9. By increasing 96-hr LC50 values in the grass shrimp system,
10. By increasing 96-hr EC50 values in the saltwater algal system.
So that there is a complete set of information for each sample, only
the 15 samples that were analyzed by the Level 1 chemistry are considered
here. Several biotests were not considered in this scheme because only a
few (or no) samples were tested by these bioassays (RAM, WI-38, CHO) or
because the bioassay did not give a positive response for any sample (Ames,
RAT-LDso). Five values were missing in the parameters considered in Table 39,
and these were assigned a value equal to the mean of the values in the
column where the missing value appeared.
For analysis of the data, the experts' rankings were quantified in two
different ways. First, each plant was given a numerical rank (denoted as
PLANTRNK in Table 40) with a ranking of 1 as most toxic, etc. Second, the
three categories assigned to groups of plants by the experts were ranked:
most toxic = 2, least toxic = 1, no measurable toxicity = 0. This ranking
scheme was denoted as TOXIC in Table 39. A rank correlation was performed
on the various other methods of ranking listed above (numbers 2-10) versus
TOXIC and versus PLANTRNK. Table 40 presents rank correlations (asterisks
denote significance at the 0.05 level) showing a high degree of association
95
-------
TABLE 39. VARIOUS SCHEMES FOR RANKING TEXTILE STUDY DATA
PLANT
N
A
W
C
T
V
L
S
B
E
F
G
K
U
X
PLANTRNK
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
11.5
11.5
11.5
11.5
11.5
11.5
11.5
TOXIC
2
2
2
2
2
2
2
1
0
0
0
0
0
0
0
ECOLPDOH
5.18
4.13
3.74
4.55
4.19
4.40
4.78
357
4.48
4.46
4.34
4.06
4.02
4.16
2.53
HEALPDOH
3.51
2.05
3.69
1.99
3.52
2.71
3.92
2.92
2.51
1.51
2.65
1.97
2.15
2.94
1.55
RATWTGN
40.7
26.2
10.3
13.6
43.9
43.4
9.4
10.7
19.9
48.5
32.3
41.4
39.5
18.5
25.1
FWALGAL
2
76
94
100
100
100
42
100
30
2
100
100
100
100
100
FWFISH
48.8
19.0
55.2
46.5
46.5
36.0
23.5
100.0
100.0
100.0
100.0
64.7
100.0
100.0
100.0
DAPHNIA
1.0
9.0
6.3
41.0
100.0
9.4
28.0
47.1
100.0
7.8
81.7
62.4
100.0
12.1
100.0
SWFISH
47.5
62.0
37.5
69.5
68.0
84.6
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
GRSSHMP
26.3
21.2
19.6
12.8
34.5
72.5
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
SWALGAL
2.3
71.5
50.0
90.0
70.0
94.0
1.7
100.0
100.0
71.5
85.0
59.0
77.0
100.0
100.0
-------
TABLE 40. RANK CORRELATIONS:
COMPARISONS OF VARIOUS RANKING
SCHEMES FOR TEXTILE STUDY DATA
Expert ranking
Bioassay PLANTRNK TOXIC
ECOL-PDOH
HEALTH-PDOH
RAT
FW ALGAL
FW FISH
RAPHNIA
SW FISH
GRASS SHRIMP
SW ALGAL
-.24
-.45
.14
.27
.78*
.58*
.92*
.89*
.47
.30
.54*
-.18
-.18
-.88*
-.48
-.81*
-.81*
-.48
NOTE: Numbers are correlation coefficients; asterisks indi-
cate significance at the 0.05 level.
TABLE 41. STEPWISE REGRESSION ANALYSIS
OF RANKING SCHEMES FOR TEXTILE
STUDY DATA
PLANTRNK
TOXIC
Order of bioassays
selected
GRASS SHRIMP
FW FISH
HEALTH-PDOH
FW FISH
SW FISH
R2*at
each step
.82
.89
.92
.77
.85
R2* = Coefficient of determination (ratio of explained vari-
ation/total variation).
97
-------
between the experts' rankings and freshwater fish, saltwater fish, and grass
shrimp.
To further characterize these relationships, stepwise regression was
run for PLANTRNK v. other ranking schemes and TOXIC v. other ranking schemes.
Table 41 presents results of this analysis. It is noted that grass shrimp
and saltwater fish are highly correlated, since only one of these two tests
is selected each time in a stepwise regression. Three bioassays (freshwater
fish, saltwater fish, and grass shrimp) predicted 92 percent of the variance
in plant ranking by an expert panel and 85 percent of the variance if grouped
rankings were considered.
98
-------
SECTION 5
CONCLUSIONS
5.1 DISCUSSION
In discussions with IERL project officers and discussions at workshops
and symposia on the environmental assessment program, a diversity of needs
was noted, including:
1. A method for evaluating the overall multimedia effects of a control
technology to support process/control developers and users in the
most cost-effective manner.
2. A means of ranking sources for their overall severity or potential
for environmental damage in order to more effectively support and
direct research on controls.
3. A way to relate the results of the IERL environmental assessment
program to the needs of various governmental regulatory agencies
and laboratories to facilitate standards setting and enforcement.
4. The establishment of a data base characterizing the various waste
streams or numerous industrial types to better support regulatory
offices and the environmental sciences community.
There is an urgent need to determine where research efforts should be
focused, and also a need to evaluate control devices for their overal1 effi-
ciency. The Level 1 bioassays and the Level 1 chemistry, as interpreted
through the SAM/IA model, are useful tools in answering these needs; the
SAM/IA model, now in draft form, uses dispersion modeling to rank various
effluents and emissions. Tables 28 and 29 in the previous section rank all
samples from the four pilot studies by potential degree of hazard, based on
chemical analysis for their relative hazards. These tables also show the
various bioassay results as interpreted by the Litton grouping method. It
is important to note here that each bioassay measures a unique biological
effect; therefore, it is not appropriate to sum the results of a sample's
bioassays to obtain a total "biologically based degree-of-hazard." For each
99
-------
effluent or emission, the bioassay results must be considered in light of
the receptor ecosystem. When more data become available in an expanded data
base, such factors as population density and pollutant dispersion should be
included in a model dealing with environmental assessment data.
To best meet the needs of control technology evaluations and source
rankings, the total multimedia effects should be considered. For example,
if a mutagenic fly ash is efficiently removed from a gaseous emission by a
newly developed control device, then a suitable method must be specified for
disposing of, or rendering harmless, the fly ash collected by the control
device (e.g., extraction and incineration of the extract). In supporting
the developers and users of control technologies, the EA program should also
consider the cost and feasibility of controls.
When all factors relevant to an intelligent decision on the acceptabil-
ity of a waste stream—chemical test results, bioassay results, potential
receptors, costs, and availability of controls--are considered, there appears
to be no mathematical model now conceivable that will encompass all of them.
The SAM models are useful for summarizing the many numbers generated by the
environmental assessment chemical analysis, but the data on which the SAM's
are based, the MEG/MATE values, must be constantly updated. When compared
to the need, there is a paucity of information correlating results of the
various environmental assessment bioassays with human health and other
environmental effects outside of laboratory conditions. An enormous amount
of information, an expanded data base, is being generated that should provide
answers to many questions about the hazards of discharging industrial process
waste streams into the environment and exposing humans to possible risks.
Although there are similarities of chemical composition and biological
effects with waste streams from comparable industries, the differences in
geographical location make each industrial site and each waste stream a
unique case. Although the environmental assessment program incorporates the
state-of-the-art tests in these fields, expert judgment is needed to evaluate
the results of an environmental assessment and to make,decisions concerning
applicable control technology. Therefore, it appears that the best means of
answering the question of potential hazard would be the use of a multi'disci-
plinary. expert panel composed of members knowledgeable in human health
effects, other ecological effects, chemical hazards, control technology
100
-------
developments, socioeconomic factors, and statistics. It is proposed that the
expert panel be composed of the environmental assessment EPA Project Officer,
a representative of the primary contracting group, a member of the IERL
Process Measurements Branch, and at least one specialist in each of the six
disciplines mentioned above.
To meet the third and fourth needs mentioned in this evaluation—the
support of standard setting and enforcement measures—it seems logical that
the Level 2 and/or Level 3 environmental assessment efforts be identical to
or easily correlated with the EPA reference and equivalent methods specified
for standard setting and enforcement in the Federal Register and other
officially and legally recognized publications. The philosophy of Level 2
is to include any pertinent Federal Register or other officially recognized
standard methods as part of the Level 2 protocol; however, there very often
are no standard (Federal Register) methods for pollutants of interest in
Level 2 efforts. Level 1 of the environmental assessment scheme is designed
as a comprehensive, cost-effective survey technique and, as such, could not
easily be made compatible with the compound-specific Federal Register tech-
niques for sampling and analysis. The Level 1 data, however, could be very
valuable in assisting the regulatory agencies in focusing their initial
efforts on the most hazardous sources, providing a description of the engi-
neering and chemical processes involved at a source, and providing a broad
base of information on both the chemical makeup and biological effects of
waste streams at a source.
In a discussion of the interpretation of data from the IERL environmen-
tal assessment program, it should be mentioned that several earlier reports
have dealt with this topic. An evaluation of the SASS train, executed and
reported by the Mitre Corporation,7 had as one goal a determination of
correlations between the biological activity and the chemical composition of
samples. Using the logarithm of concentration from analysis of SSMS or
GC-MS data, the percent viability from the RAM cytotoxicity test results,
and linear regression analysis, these investigators were unable to establish
strong correlations between individual elemental concentrations and observed
cytotoxicity in a given sample. It was suggested that this lack of strong
correlation might be due to a lack of information on the biological avail-
ability/chemical form of the various elements.
101
-------
In another approach, each element was assigned a toxicity rating from 1
(least toxic) to 10 (most toxic), based on most probable chemical form and
most comparable exposure route. A chemical toxicity index was calculated by
multiplying the toxicity rating by the concentration using SSMS or GC-MS
analysis. Using linear regression analysis, the common logarithm of the
chemical toxicity index was compared with the percent viability in the RAM
test at maximum dosage (1,000 ug/mL). It was found that 20 elements (Sb,
As, Be, Bi, Cd, Cr, Co, Cu, Pb, Mn, Mo, Ni, Se, Ag, Te, Tl, Sn, Ti, V, and
Zn) were most influential in predicting the toxicity of a given particulate
sample. This method of analysis of the data assumes linear relationships
and does not account for synergistic or antagonistic effects.
In her presentation at a recent symposium on bioassays, Dr. Judi Harris
of Arthur 0. Little, Inc., also examined the relationship between chemical
and biological data in a paper entitled "Comparison of Chemical and Biological
Data in Level 1 Environmental Assessments."21 In her paper, she suggested
the following:
The uniformly positive response obtained in the initial Level 1 bioassays
would imply that further biological testing of this sample/source is
warranted. In determining the appropriate direction of Level 2 biotests,
it is constructive to compare the Level 1 results of bioassays with
those of the chemical analysis, even though no cause-effect correlation
could be made on the basis of the available Level 1 data. The chemical
analysis showed the presence of a variety of different classes of
organic compounds, each of which contains at least some individual
species known or suspected to be biologically active. It seems quite
plausible to hypothesize that the variety of cytotoxic and mutagenic
activities found in Level I bioassay of the whole sample, are, in fact,
attributable to different chemical components of the sample.
Comparison of the Level 1 chemical and biological data therefore provides
valuable direction for the Level 2 studies. Monitoring of chemically
fractionated samples in Level 2 by the most sensitive bioassay might
lead to separability of the cytotoxic and mutagenic effects of the
material.
Other probably useful confirmatory biotests, on either whole or frac-
tionated samples, might include:
Other mutagenicity tests in microbial and mammalian cells
In-vitro transformation systems
Teratogenicity and/or chromosome damage analyses
102
-------
5.2 RECOMMENDATIONS
The following suggestions are made as a result of a thorough examina-
tion of the data from the environmental assessment programs to date. It is
understood that some of the proposed changes or additions may already be in
the process of being implemented.
1. A uniform protocol for both chemical and biological analysis of sample
types (solids, liquids, gases) should be implemented throughout the
IERL. Such a protocol would allow for better statistical evaluation of
results by ensuring a complete and consistent data set on each sample
and eliminating missing values. It is essential that complete sets of
information be obtained on each type of sample for a viable data inter-
pretation model to be developed. Table 42 shows a sampling protocol22
that IERL has proposed for use in environmental assessment projects.
The SAM model series assumes that such a uniform protocol (including
standard number of analyses) has been followed; the SAM/I and SAM/IA
models total the ratio of each chemical's measured concentration to its
MEG or MATE value. If more chemical analyses have been done on one
sample than on another, it is likely that the sum of ratios (the PDOH
value) would be greater for the first sample than the second, even
though the second might have the same or greater potential hazard
value.
Sampling and analysis procedures should also be made uniform. In
the pilot studies, different leaching procedures were used to generate
samples from bulk solids for chemical analysis and for bioassay, so it
was impossible to tell whether the chemical and biological systems were
measuring solutions of similar or different compositions. The revised
Level 1 manual1 specifies a suitable aqueous leaching procedure that
should be used with all chemical and biological tests on Teachable
materials.
2. More validation studies for Level 1 of the environmental assessment
program should be planned and executed using the proposed uniform
sampling protocol. These studies would increase the size of the avail-
able data base and they could be performed using the revised Level I
procedures. In the pilot studies, some replicate samples should be
103
-------
TABLE 42. GUIDELINES FOR BIO ASSAY TESTING
Sample
Liquids
Solids
Solidsleachates
SASS participate
SASS organics
Gases
Muta-
genicity
A
A
A
A*
A*
Cyto-
toxicity
A
A
B
A*
A*
Whole
animal
A
A
B
^^
Aquatic Plant
A
A
B
C
Soil
C
C
C
A Required test.
8 Required when solid handling or use may result in water runoff.
C Terrestrial ecological tests (stress ethylene and soil microcosm) are in a developmental stage.
Their use is recommended; however, they are not required on a laboratory-wide basis at this
time.
"Additional SASS samples (1-2 extra runs) may be necessary to acquire adequate quantity of sample
for biological tests.
104
-------
analyzed so that a determination of variability can be made. The
precision of analytical methods can also be determined by interlabora-
tory testing programs.
3. The continuation of basic research into cause-effect relationships
should be supported to expand the MEG/MATE list. This information is
essential to the determination of hazardous effects for which suitable
bioassays have not been developed. For example, there is no Level 1
bioassay that will react to the mutagenic potential of beryllium.
Therefore, the chemistry and biology analyses need to be run simultane-
ously and the basic knowledge on the biological effects of pure com-
pounds needs to be well defined.
4. The terrestrial tests need further development with an emphasis on the
interpretation of results. Even though these tests (soil microcosm and
stress ethylene) yield much information, a clearly defined relationship
is needed between terrestrial bioassay results and potential hazardous-
ness of the sample; this is especially important in the soil microcosm
test where response values at increasing dosages may increase, then
decrease, for the same sample. In addition, the available data from
the terrestrial bioassays show such a large degree of background varia-
bility that low toxic effects may not be detected. Measures need to be
implemented to reduce the differences among both replicate samples and
replicate controls; that is, to reduce the background. Work is cur-
rently underway to evaluate and incorporate revised soil and plant
stress ethylene tests and a new insect test into the terrestrial bioassay
group.
5. To ensure the completeness of the historical record of test results
from an environmental assessment, it is recommended that the IR and
LRMS spectra be included as an appendix to the Environmental Assessment
Report. These spectra could be a valuable record in later checks for
the presence or absence of some currently unidentified compound. The
spectra would also aid in the interpretation of organic results and
classification of weights of organics into MEG categories: If an LRMS
spectrum is not available for a given fraction, the Level 1 manual
suggests that the weight cf materials in that fraction be assigned as
they were in the adjacent fraction with the most similar spectrum.
105
-------
6. A scale needs to be developed for quantifying responses other than
lethality in the rodent acute toxicity test. For the group of samples
considered in this study, it was impossible to discriminate or rank
samples based on LD50 values because there were no lethal effects.
However, various physiological and behavioral effects were observed in
the samples. In an attempt to correlate the sublethal effects with
chemical concentrations, the number of sublethal effects was simply
totaled, with all effects considered of equal importance. The Level 1
bioassay manual2 lists various possible sublethal effects; numerical
weights could be assigned to these effects by experts in the field of
whole animal toxicity.
7. In some test results from the aquatic ecological series, response
values were reported as EC50 (concentration, expressed as percent
sample in an aqueous diluent, affecting 50 percent of the population)
or as LD50 (dose, expressed as percent sample in an aqueous diluent,
that is lethal to 50 percent of the population). An examination of the
raw data showed, in many cases, that a significant response occurred
only at a level affecting less than 50 percent of the population, so
EC50 values could not be computed for these samples. In order to show
the numerical differences in the magnitude of responses, some values
were computed as EC20 (concentration affecting 20 percent of the popula-
tion). Realizing this variation exists, it is suggested that data
collection and summary forms for the aquatic ecological series include
entry spaces for EC20, EC50, and EC80 values (or LD20, LD50, and LD80,
if appropriate for the sample set), with the stipulation that all
values be computed. This computation can be performed by a variety of
readily available methods, such as the SAS PROBIT program12 or graphical
interpolation.
8. A number of those with whom this study was discussed expressed concern
about the use of the Ames test on complex mixtures. Almost all of the
validation and sensitivity studies correlating the Ames test with whole
animal tumor studies have involved pure compounds. Although all types
of complex mixtures found in environmental samples could not be vali-
dated through whole animal testing, some representative samples (e.g.,
106
-------
automobile exhaust, fly ash from various coal-to-energy technologies)
might be tested as both whole and fractionated samples in the Ames
system and in the whole animal system.
9. One of the important types of information sought in an environmental
assessment study is the effect on human health of various industrial
waste streams. If pertinent epidemiological or other data are avail-
able, they should be assembled and studied; such data have been amassed
for coking operations and for asbestos-handling operations.
10. Decision criteria for proceeding from Level 1 to Level 2 should be
based on all data, the bioassay results as well as the chemistry. At a
recent meeting, the Bioassay Subcommittee of the Environmental Assess-
ment Steering Committee suggested that the following criteria be used
for bioassays:
a. Proceed to Level 2 if Level 1 samples show a mutagenic response in
the Ames test.
b. Proceed to Level 2 if Level 1 bioassay, other than the Ames test,
shows a trend toward a toxic response. A significant positive
response in more than one bioassay was suggested as indicating a
trend in toxicity for the sample in question.
When the PDOH values and the bioassays are compared, PDOH values greater
than ~1,000 for either health-related or ecology-related calculations
correspond with a trend toward toxic responses in the bioassays; more
data are needed to further define the significance of the PDOH value
limits. An indication of the presence of a significant quantity of any
known carcinogen or any known severe acute toxin, as shown by Level 1
chemistry, would certainly dictate proceeding to Level 2 validation
chemistry.
107
-------
REFERENCES
1. D. E. Lentzen, D. E. Wagoner, E. D. Estes, and W. F. Gutknecht, IERL-RTP
Procedures Manual; Level 1 Environmental Assessment (Second Edition),
EPA-600/7-78-201, U.S. Environmental Protection Agency, Research Triangle
Park, NC, October 1978.
2. K. M. Duke, M. E. Davis, and A. J. Dennis, IERL-RTP Procedures Manual:
Level 1 Environmental Assessment Biological Tests for Pilot Studies^
EPA-600/7-77-043, U.S. Environmental Protection Agency, Research Trianqle
Park, NC, April 1977.
3. N. H. Gaskins and F. W. Sexton, Compilation of Level 1 Environmental
Assessment Data, EPA-600/2-78-211, U.S. Environmental Protection Agency
Research Triangle Park, NC, October 1978. '
4. J. M. Allen, J. E. Howes, Jr., S. E. Miller, and K. M. Duke, Comprehen-
sive Analysis of Emissions from Exxon Fluidized-Bed Combustion Mini plant.
UnTf, Draft Report, September 9, 1977, EPA Project Officer D. B. Henschil
U.S. Environmental Protection Agency, Research Triangle Park, NC. '
5. G. C. Page, Environmental Assessment: Source Test and Evaluation Report--
Chapman Low-Btu Gasification, EPA-600/7-78-202, U.S. Environmental Pro-
tection Agency, Research Triangle Park, NC, October 1978.
6. G. D. Raw!ings, Source Assessment: Textile Plant Wastewater Toxics
Study. Draft Report, December 1977, EPA Project Officer Max Samfield,
U.S. Environmental Protection Agency, Research Triangle Park, NC. (Now
available as EPA-600/2-78-004h from the U.S. Environmental Protection
Agency, Research Triangle Park, NC.)
7. H. Mahar, Evaluation of Selected Methods for Chemical and Biological
Testing of Industrial Particulate Emissions. EPA-60Qyz-76-137. U.S.
Environmental Protection Agency, Research Triangle Park, NC, May 1976.
8. J. W. Hamersma, S. L. Reynolds, and R. F. Maddalone, IERL-RTP Procedures
Manual: Level 1 Environmental Assessment. EPA-600/2-76-160a, U.S.'
Environmental Protection Agency, Research Triangle Park, NC, June 1976.
9. M. P. Schrag, A. K. Rao, G. S. McMahon, and G. L. Johnson, Fine Particle
Emissions Information System Reference Manual, EPA-600/2-76-173, U.S.
Environmental Protection Agency, Research Triangle Park, NC, June 1976.
10. M. P. Schrag, A. K. Rao, G. S. McMahon, and G. L. Johnson, fine Particle
Emissions Information System User Guide. EPA-600/2-76-172, U.S. Environ^
mental Protection Agency, Research Triangle Park, NC, June 1976.
108
-------
11. D. Brusick and R. Hart, internal communication, Litton Bionetics,
August 1978.
12. A. J. Barr, J. H. Goodnight, J. P. Sail, and J. T. Helwig, A User's
Guide to SAS 76. SAS Institute, February 1977.
13. D. J. Finney, Probit Analysis, 2nd ed., Cambridge: Cambridge University
Press, 1952.
14. J. G. Cleland and G. L. Kingsbury, Multimedia Environmental Goals and
Environmental Assessment. Volume 1. EPA-600/7-77-136a. U.S. Environmen-
tal Protection Agency, Research Triangle Park, NC, November 1977.
15. J. G. Cleland and G. L. Kingsbury, Multimedia Environmental Goals and
Environmental Assessment, Volume 2, EPA-600/7-77-136b. U.S. Environmen-
tal Protection Agency, Research Triangle Park, NC, November 1977.
16. Environmental Protection Agency, Emissions Assessment of Conventional
Combustion Systems. Vol. 1: Gas-Fired and Oil-Fired Residential Heating
System Source Categories (Draft). EPA Contract 68-02-2197. Project
Officer Dr. Ronald A. Venezia, U.S. Environmental Protection Agency,
Research Triangle Park, NC. (Now available as EPA-600/7-79-029b from
the U.S. Environmental Protection Agency, Research Triangle Park, NC.).
17. L. M. Schalit and K. J. Wolfe, SAM/IA: A Rapid Screening Method for
Environmental Assessment of Fossil Energy Process Effluents. EPA-600/7-
78-015, U.S. Environmental Protection Agency, Research Triangle Park,
NC, February 1978.
18. M. A. Herther, L. R. Waterland, and R. J. Milligan, SAM IB: A Rapid
Screening Method Incorporating Biological Hazard Evaluation for Environ-
mental Assessment of Fossil Energy Process Effluents (Draft). EPA
Project Officer J. S. Bowen, U.S. Environmental Protection Agency,
Research Triangle Park, NC, August 1978.
19. L. A. Anderson, M. A. Herther, and R. J. Milligan, SAM I: An Interme-
diate Screening Method for Environmental Assessment of Fossil Energy
Process Effluents. Acurex Report TR-79-154. EPA Project Officer. J7s.
Bowen, U.S. Environmental Protection Agency, Research Triangle Park,
NC, December 1978.
20. B. Ames, J. McCann, and E. Yamasaki, Methods for Detecting Carcinogens
and Mutagens with the Salmonella/Mammalian-Microsome Mutagenicity Test,
Mutation Res., 31:347-364, 1975.
21. M. D. Waters, S. Nesnow, J. L. Huisingh, S. S. Sandhu, and L. Claxton,
Application of Short-term Bioassays in the Fractionation and Analysis
of Complex Environmental Mixtures. EPA-600/9-78-027. U.S. Environmental
Protection Agency, Research Triangle Park, NC, September 1978.
22. J. A. Dorsey, internal communication with J. K. Burchard, August 1978.
109
-------
APPENDIX A
THE DATA BASE
The 51 Fortran-encoded sample sets are presented as they were numbered
in this study. These sample sets include all appropriate chemical and bio-
logical testing results from the four studies described in this document.
Table 4 in the text of the document shows the location of parameters on
these data printouts. Table 5 in the text indicates the unit of measure in
which each parameter is expressed. Negative signs preceding data identify
results that were less than the minimum detectable limit of the analytical
procedure; in such cases a value equal to half the minimum detectable limit
was entered. See subsection 3.1.1.
A-l
-------
DATA BASE FOR STUDY 06
010602SLPARATAR
2,6
020602
030602*
040602
050602
060602
070602
080602
090602
100602
110602
120602
130602
140602
150602
160602
170602
180602
190602
200602
210602
220602
230602
240602
250602
260602
270602
280602
290602
300602
310602
3.5
0.
o.
0.
0.
3.
2.5
0.
0.
0.
0.
0,
80.
100.
293370.
1.2
4.8
4.
010603SEPARALIQ
020603
030603
040603
050603
060603
070603
080603
090603
100603
110603
120603
130603
140603
150603
160603
170603
180603
190603
0.
0.11
23.2
0.
0.
0.
0.
300.
5.
10.
0.
1000.
0.
0.
300.
1.7
o.
o.
0.
0.
0.
0.
1.
9.
10.
2000.
1.
0.
0.
0.
o.
27880.
2.0
1.5
2.
0.
o.
0.5
.23
0.
0.
o.
0.
1.
2.
5.
0.
o.
6000.
9000.
0.
0.
0.
0.182
5.
o.
0.
0.
5.
0.
20.
0.
0.5
0.
0.
0.
0.
0.
0.
1490.
41630.
2.2
33.9
0.359
1.9
,20
0.
0.
0.
0.
300.
0.
60.
0.
0.
20000.
0.
100.
500.
0.
0.165
50,
0.
0.
0.
0.
0.
0.5
300.
0.
0.
0,
0.
0.
300.
0.
10950.
0.06
17.5
0.
.16
0.
0.
0.
0.
0.
0,
10,
10.
100.
2000.
3.
0.
0.
100.
4.
0.
0.
0.
o.
80.
0.
0.
-.25
0.
100.
40.
40.
0.
96800.
1.5
1.
.03
.41
0.
10.
0.
0.
70.
0.
300.
0.
2.
30.
100.
0.
0,
0.
o.
0.
5,
0.
0.
0.03
0.
0.
50.
40.
0.
0.
200.
47000.
23,9
10.
.02
.41
0.
0.
0.
3.
30.
0.
2000.
0,
20000.
2000,
0.
0.
0.
500,
o.
0.
o.
5,
0.
5.
-•I
0.
7000.
o.
0,
60.
0.
100.
41030.
1.4
o.
0.51
.25
o.
0.
o.
7.
0.
*,
800.
o.
20000.
»
0.
0.
o.
0.
o.
A-2
-------
STUDY 06 (continued)
50.
0.2
0.10
200603
210603
220603
230603
240603
250603
260603
270603
280603
290603
300603
310603
026604
030604
040604
050604
060604
070604
080604
090604
100604
110604
120604
130604
140604
150604
160604
170604
180604
190604
200604
210604
220604
230604
240604
250604
260604
270604
280604
290604
300604
310604
010605CYCLODUST
020605 89.2
030605
040605
050605 17.
060605 0.
070605 0.
0.
0.3
1.2
1.1
1.
0.02
-0.
5000.
1.1
1.0
1.
FLASH
87.9
1000.
1.1
1.4
3.
98.7
109.4
0.07
3.0
1.2
3.
0.
1.5
0.
0.9
1.
1.
1.4
10.
-.15
1.4
1.5
1.
94.9
100.5
0.
0.
1.8
1.0
0.
0.
0.
1.1
0.
60.
0.
1.3
1.
U
0.
o.
1.1
10.
0.
0.
1.
1.1
0.
19.7
400.
0.
0.
0.
2000.
-4.5
90.
0.
- 0.
0.
0.
0.
0.
0.
0.
0.
0.
, o.
0.
-10.
50.
30.
600.
20.
0.
o.
0.
0.
0.
0.
0.
0.
0.
0.
2000.
0.
30.
800.
10000.
0.
0.
0.
0.
20.
0.
0.
0.
0.
0.
20000.
1000.
3000,
7000.
70.
0,
0.
0.
-0.
0.
0.
0.
0.
200.
0.
0.
0.
50.
3000.
0.
0.
0.
0.
0.
0.
0.
90.
300.
0.
30.
50.
50000.
6000.
0,
0.
0.
0.
0.
0.
0.
100.
0.
20.
-.2
10000.
10000.
0.
0.
0.
0.
0.
1.2
33,
0,
0.
A-3
-------
STUDY 06 (continued)
080605
0^0605
100605
110605
120605
130605
140605
150605
160605
170605
180605
190605
200605
210605
220605
230605
240605
250605
260605
270605
280605
290605
300605
310605
0.
0.
500.
0,
30.
0.
200.
0.
0.
0.
0.
113.
0.9
1.1
1.
0.
0.
o.
0.
10.
-5.
30.
300.
7.
0.
0.
0.
0.
19.
0.
0.
0.
0.
80.
0.
20.
8000.
0.
0.
0.
0.
109.
44.
0.
0.
0.
0.
3.
900.
200.
2000.
2.
0.
0.
0.
0.
18.
1.9
1.2
1.
1.6
1.1
1.1
0.
0.
100,
0.
o.
100.
2.
100.
0.
0.
0.
0.
0.
20.
o.
0.
0.
3.
2000.
500.
0.
0.
0.
0.
0.
20.
0,
0.
».a
1000.
1000.
0.
o.
o.
o*
o.
1.
3.
0,8
1,1
1,
A-4
-------
DATA BASE FOR STUDY 10
011009BED LEACH
021009 99.4
031009
041009
051009
061009
071009
081009
091009
101009
111009
121009
131009
141009
40.9
-3000.
-2500.
-500.
-2000.
SOOOOO.
-1000.
3000.
-2000.
10000.
15*100920000000.
161009
171009
181009
191009
201009
211009
221009
231009
241009
251009
261009
271009
281009
291009
301009
311009
-2000.
2.
011008BEDREJECT
021008
031008
041008
051008
061008
071008
081008
091008
101008
1J1008
121008
13.1008
141008
151008
161008
171008
181008
191008
61.2
.2
-.15
-.05
1.
100.
-.25
20.
7.
100.
40.
.3
0.
0.
0.
84.0
68.5
30.
100.
-3000.
-1000.
-2000.
-2000.
7000.
-500.
-300.
3000.
30000.
150000000.
200000.
1.
2.
56.6
52.0
1.
-.1
-.15
.5
2.
-.05
5.
15.
30.
15000.
100.
0.
0.
0.
0.
100.
-500.
-2500.
-500,
-3000.
600.
-1000.
700000.
-50000.
5000.
150000.
-30.
64000.
1.
-.05
-.15
-.05
1.5
-.035
-.05
300.
5.
30.
100.
1.
0.
0.
0.
0.122
31.8
100.
15000.
-1000.
-2000,
-1000.
-2000.
-1000,
150000.
30000.
300000.
400000,
200000.
2100,
-2.5
1.
-.1
-.15
4.
-.15
-.2
40.
15.
400.
70000.
100.
0.
0.
0.
25.3
100.
-2000.
5000.
-500.
-500.
-350.
-7500.
70000.
300000.
100.
-2500.
-50000.
-2000.
-500.
100000.
70000.
15000.
-5000.
-1500.500000000.
150000.
200000.
-10.
1500000.
1.
1.
2.
-.15
-.05
.5
.4
-.5
1.
150.
3.
50000.
0.
0.
0.
0.
200000.
-150.
50.
2.
-.15
-.5
,6
20.
1.
4.
1.5
2.
150000.
100000.
0.
0.
0.
0.
42,3
100.
-2500.
-2000.
-500.
70000.
-1000.
-300.
150000.
2000000.
10000000.
2000000.
800.
1.
-.05
-,15
.5
6.
-.3
1.5
15.
30000.
20000.
50.
0.
0.
0.
0.
A-5
-------
STUDY 10 (continued)
201008
211008
221008
231008
241008
251008
261008
271008
281008
291008
301006
311006
0.
8.7
0.
0.
1.1
1.0
1.
011007CYLOLEACH
021007
031007
041007
051007
061007
071007
081007
091007
101007
111007
121007
131007
141007
77.3
91.6
-1500.
-1000.
-250.
-1000.
100000.
-500.
3000.
-1000.
-2500.
15100770000000.
161007
100000.
0.
.005
0.
0.
1.3
1.5
1.
67.6
57.9
11.
100.
-1500.
-500.
-100.
•1000.
20000.
500.
-150.
2000.
50000.
250000.
-15.
.025
0.
0.
0.
1.2
1.5
0.092
100.
-250.
-1000.
-250.
-1500.
2000.
-500.
300000.
-250.
20000.
5000000.
1000000,
.01
0.
0.
-.01
1.0
0.128
100.
100.
7000.
-500.
-1000.
1000.
-1000.
-500.
300000.
50000.
150000.
700000.
26.1
0.
0.
2.
60.
1.1
1.
100.
75.
-500.
1500.
-250.
-250.
•150.
-1500.
7000.
30000.
500000000.
1500000.
.002
0.
0.
35.
1.2
100.
75.
-1000.
-7500.
-1000.
1000.
3000.
30000.
20000.
-2500.
150000000.
15000000.
171007
181007
191007
201007
211007
221007
231007
241007
251007
261007
271007
261007
291007
301007
311007 1.
011006CYCLODUST
021006 81.7
031006
041006
051006
061006 2.
071006 -.2
53000.
400. 200000.
-2.5
-10.
1500000.
-150.
SO.
1.
1.
69.9
58.1
4.
.1
1.
0.138
.3
-.2
2.
0.124
20.
-.1
1.5
-.25
'.2
-.5
-.0005
0.
0.
1.3
100.
72.5
-1000.
-1000.
-250.
10000.
"500.
-150,
15000.
200000.
300000.
2500.
49000.
-.05
•1.
A-6
-------
STUDY 10 (continued)
081006
091006
101006
111006
131006
131006
141006
151006
161006
171006
181006
1*1006
201006
2U006
221006
251006
241006
251006
261006
271006
261006
291006
301006
311006
.3
5.
600,
1.5
100.
6.
100.
10.
6.
0.
0.
.002
.004
2.3
0.
a.
1.1
0.8
1.
011004CORSEPART
021004
031004
041004
051004
061004
071004
061004
091004
101004
111004
121004
131004
£ (
-------
STUDY 10 (continued)
271004
261004
291004
301004
311004
011003F
021003
031003
041003
051003
061003
071003
081003
091003
101003
111003
121003
131003
141003
151003
161003
171003
181003
191003
201003
211003
221003
231003
241003
251003
261003
271003
281003
291003
301003
311003
011002
021002
031002
041002
051002
061002
071002
081002
091002
101002
111002
121002
131002
141002
0.8
0.8
2.
INE PART
58.3
7.
-.2
.4
6.
2000.
-1.5
100.
10.
200.
a.
15.
0.
0.
0.
.053
-.1
0.
118.
1.8
1.1
2.
DOLOMITE
101.8
.3
-.2
-.05
-.15
4,
-.25
.45
.45
75.
1.3
3.9
1.
47.5
36.8
15,
-.1
6.
3.
40.
1.
30.
70.
200.
4000.
100.
0.
0.
0.
.003
-.015
0.
88.
1.5
6,9
2.
97.1
95.
3.
-.1
-.15
-.075
-0268
-.05
.3
.2
5.
2.6
1.5
0.141
.4
-.2
0.3
7.
.4
-1.
700.
100.
200.
300.
4.
.003
0.
0.
.003
767.
562.
81.
1.5
2.7
-.075
-.2
-.05
-.25
-.035
-.1
40.
3.5
1.5
1.6
0.156
100.
-.1
3.
40.
-.15
-.1
600.
40.
4000.
100000.
500.
.003
0.
.05
.032
494,
659.
.16
1.1
1.1
-.1
-.15
-.25
-.1
-,05
4.5
2.25
25.
4.2
4.
4.
1.
1.
10.
1.
-.5
4.
200.
20.
100000.
.38
.0
.1
.05
10.8
461.
1038.
5.3
20.
2.6
1.
-.1
-.2
-.05
0.15
-.125
-.3
4.5
70.
-.15
2.1
-.2
-1.
5.
70.
3.
5.
10.
0.
30000.
20000.
.13
.003
.1
.003
.001
28.
266.
85.
1.4
-.2
-1.
-.15
.5
-.1
-.75
.as
-.05
225000.
1.3
-.05
3.
U
30.
-.3
5,.
40.
30000.
15000.
2000.
.003
0.
.003
•.0005
0.
*5.
1.6
-.05
-.15
-.05
,4
-.15
-.048
13.
* ** •
800.
5000.
A-8
-------
STUDY 10 (continued)
151002
161002
171002
181002
191002
201002
211002
221002
231002
241002
251002
261002
2*1002
281002
291002
301002
31'1002
235.
10.
63. a
1.7
1.0
1.
011001FEEDCOAL
021001
031001
041001
051001
061001
071001
081001
091001
101001
111001
121001
131001
141001
151001
161001
171001
181001
191001
201001
211001
221001
231001
241001
251001
261001
271001
281001
291001
301001
311001
72.8
.6
-.03
-.15
.1
180.
.05
4.5
.3
6.5
1335.
.2
0
0
.214
.014
.10
320.5
94.
1.3
1.2
1.
300.
10.
.068
1.1
1.3
1.
71.
69.2
5.
-.015
.2
.08
.5
.03
0.5
3.
30.
2000.
5.
.434
0
0
.094
1.065
542.5
76.5
1.0
2.0
1.
2.
.0125
.052
1.5
0.8
-.01
-.03
-.015
.8
-.015
-.015
19.
32.
16.
32.
.1
.014
.042
0
.49
121.5
0.
313.5
1.0
0.8
1500.
12.5
-.006
2.6
28
-.015
.15
1.
-.015
-.015
21.
16.
600,
48000.
8.
.228
.122
.05
.011
95.
345.5
1.2
650.
.6
30.
1.0
1.
1.
-.045
-.75
.04
.5
1.5
-.03
16.
11.
1.5
17000.
.11
.462
.14
.05
.03
187.5
349.5
1.3
1.
175000.
-.0005
1.3
-.1
-1.5
.3
8.
.15
2.
4.5
4,5
1500.
3520.
.02
.014
.03
.094
-.0005
93.
191.5
1.3
750,
-.0012
1.3
".Ol
1.
-.032
4.
-.05
.2
8.
6400.
2000,
200.
.014
,064
.044
.03
-.00075
42.5
65.
1.2
A-9
-------
DATA BASE FOR STUDY 15
011501STEEL3MIC
021501
031501
041501
051501
061501
071501
081501
091501
101501
111501
131501
131501
141501
151501
161501
171501
181501
191501
301501
211501
221501
231501
241501
251501
261501
271501
281501
291501
301501
311501
60.3
.33
-.05
-.05
-.05
6.3
5.4
.64
7.1
5200.
270.
790.
-.115
-.05
-.05
-.05
.26
41.
.78
90.
880.
20000.
47.
11.
-.05
-.05
-.05
1.7
-.05
4.8
20000.
290.
5000.
-.12
900.
-.05
.68
.15
-.05
32.
3600.
33.
8900.
59.
011502STEEL1MIC
021502
031502
041502
051502
061502
071502
081502
091502
101502
111502
121502
131502
141502
15150
161502
171502
181502
191502
65.3
1.5
-.05
-.05
-.05
12.
44.
1.4
18.
4400.
260.
760.
-.11
-.05
-.05
-,05
2.5
65.
.51
87.
60.
40000.
31.
21.
-.05
-.05
-.05
3.9
-.05
11.
20000.
330.
4900.
.34
1900.
-.05
.65
.40
-.05
95.
2900.
15.
4300.
110.
.53
11.
-.05
.13
27.
-.05
10.
490,
-.08
68.
1.5
10.
-.05
.13
70.
-.05
43.
470.
-.08
140.
-.05
.82
-.05
1.2
340.
35.
2,7
120.
20000.
10000.
-.05
.79
-.05
1.1
610.
79.
2.3
110.
8300.
10000.
-.05
'.05
-.05
.21
.93
190.
400000,
6900,
20000,
-.05
-.05
-.05
.20
0.9
480.
400000.
30000.
80000.
A-10
-------
STUDY 15 (continued)
201502
311502
221502
231502
241502
251502
261502
271502
281502
291502
301502
3U502
011503COKE1MICR
021503 73.2
031503
041503
051503
061503
071503
061503
091503
101503
111503
121503
131503
141503
151503
161503
171503
161503
191503
201503
211503
221503
231503
241503
251503
261503
271503
281503
291503
301503
31,1503
OU504STEEL03MI
021504 85.9
031504
041504
051504
061504 6.2
071504 -.05
3.4
-.05
-.05
-.05
1.0
9.4
.66
.98
250.
17.
220.
.24
-.05
-.05
-.05
.53
4.2
.08
4.0
94.
10000.
3.3
48.
-.05
-.05
-.05
1.6
-.05
1.2
1100.
38.
41.
.25
940,
-.05
-.05
.85
-.05
21.
170.
8.0
200.
6.1
7.5
.87
-.05
-.05
3.7
-.05
1.1
51.
10000.
5.3
-.05
.18
-.05
.10
66.
17.
9.8
2.6
1100.
170.
-.05
-.05
-.05
.03
-.05
3.9
10000.
1700.
2200.
'0.125
-.05
5.0
',05
170.
.78
12.
',05
.44
-.05
-.Ob
A-11
-------
STUDY 15 (continued)
081504
091504
101504
111504
121504
131504
141504
151504
161504
171504
181504
191504
201504
211504
221504
231504
241504
251504
261504.
271504
281504
291504
301504
311504
-.05
-.05
68.
9.7
.69
10.
10000.
30000.
50000.
-.05
-.05
2,8
440.
.56
97.
220.
3500.
130.
-.05
-.05
19.
-.05
43.
5800.
40.
3500.
-.13
-.05
.73
.15
-.05
63.
1100.
.44
10000.
27.
-.05
0.15
17.
-.05
110.
93.
-.09
73.
-.05
5.9
140.
38.
.SI
130.
40000.
20000.
-.05
2.2
2,3
67.
800000,
40000,
90000.
011505STEEL01MI
021505
031505
041505
051505
061505
071505
081505
091505
101505
111505
121505
131505
141505
151505
161505
171505
181505
191505
201505
211505
221505
231505
241505
251505
261505
90.0
-.07
-.05
-.05
-.05
40.
12.
.40
6.0
7700.
30000.
60000.
-.07
-.05
-.05
-.05
1.2
43.
.71
57.
170.
50000.
50.
2.9
-.05
-.05
-,05
2.6
-.05
H.
4600.
23.
1200.
-.075
210.
-.05
.18
-.04
-.05
74.
800.
9,6
5600.
16.
.97
6.8
-.05
.04
9.7
-.05
65.
46.
-.05
21.
-.05
.52
-.05
.73
40.
10.
.40
57,
20000.
2500.
-.05
-.05
-.05
.28
2.7
53.
400000*
20000,
770,
^ w
A-12
-------
STUDY 15 (continued)
271505
281505
291505
301505
311505
0115061RUN3MICR
021506
031506
041506
051506
061506
071506
081506
091506
101506
111506
121506
131506
141506
151506
161506
171506
181506
191506
201506
211506
221506
231506
24J506
251506
261506
271506
281506
291506
301506
311506
111.4
2.3
-.05
.06
.41
60.
14.
10.
2.9
640.
3300.
10000.
2.3
-.05
.57
.98
18.
6.1
65.
6.3
76.
3100.
61.
5.7
-.05
.03
1.1
29.
-.05
160.
760.
48.
1300.
.57
2100.
.53
9.5
.58
-.05
180.
1000.
290,
30000,
18.
11.
6.8
.13
.95
2.5
-.05
150.
26.
1.1
1300.
-.05
.29
1.8
38,
4.4
11.
89.
86.
400000.
60000.
-.05
.63
.10
7.3
6.5
35.
60000.
10000.
30000.
0115071RUN1MICR
021507
03)1507
041507
051507
064507
071507
081507
091507
101507
111507
121507
131507
141507
46.0
3.7
-.05
,24
.28
140.
22.
14.
9.2
321.
3.7
-.05
.21
.67
25.
35.
25.
4.4
52.
100.
-.05
.09
-.05
69.
-.05
110.
1100.
33.
8200.
.37
6.5
2.0
-.05
950.
2200.
370.
35.
3.6
.09
2.8
3.5
-.05
270.
64.
1.1
-.05
.20
1.3
46.
6,1
17.
310.
59.
300000.
-.05
.44
^ — » »
.15
10.
4.5
^ v ^
120,
20000.
30000.
A-13
-------
STUDY 15 (continued)
151507
161507
171507
181507
191507
201507
211507
221507
331507
241507
251507
261507
271507
281507
291507
301507
311507
0115080
021508
031508
041508
051508
061508
071508
081508
091508
101508
111506
121508
131508
141508
151508
161508
171508
181508
191508
201508
211508
221506
231508
241508
251506
261508
271508
281506
291508
301508
311508
10000.
7000.
1L1MICRO
15.8
1.3
-.05
-.05
.10
280.
.97
2.2
4.7
220.
33.
31.
10000.
76.
1.3
-.05
-.05
.33
4.6
4.4
13.
190.
1000.
70000.
16.
890. 20000.
.39 28.
1.1 450.
-.05
-.Ob -.05
-.05 0,68
.61 .12
-.05 -.05
57. 29.
900. 300.
20000. 210.
1600. 4500.
8.1 13.
011509COPPER1MI
021509
38.0
890.
40000. 50000,
.78
2.4
-.05
.14
7.8
-.05
10.
5.2
-.04
3.40
-.05
.41
.31
14.
14.
814.
63.
220.
10000.
5900.
-.05
-.05
.04
.52
• 22
10000.
3500.
200000,
A-14
-------
STUDY 15 (continued)
031509
041509
051509
061509
071509
081509
091509
101509
111509
121509
131509
141509
151509
161509
171509
1*81509
191509
201509
211509
221509
231509
241509
251509
261509
271509
281509
291509
301509
311509
5.6
-.05
-.05
-.05
170,
4900.
11.
140.
310.
610.
210.
.56
-.05
-.05
-.05
1.9
1100,
19.
6.5
190.
20000.
11.
20000.
-.05
-.05
-.05
160.
-.05
29.
10000.
49.
1300.
5.9
350000.
-.05
-.05
9.8
3600.
-.05
140.
50000.
560.
10000.
43.
3900.
180.
-.05
2.0
80000.
-.05
750.
1300.
.78
250.
1.3
-.05
3.9
40000.
5100.
5500.
180.
7500.
20000.
25.
-.05
-.05
4.1
31.
30000.
30000.
100000.
30000.
0-11510COPPER3MI
021510
031510
041510
051510
061510
071510
081510
091510
101510
111510
121510
131510
141510
151510
161510
171510
181510
191510
201510
211510
32.6
2.4
-.05
-.05
.31
81.
1400.
11.
24.
880.
1100.
370.
1.1
-.05
-.05
-.05
3.7
2200.
8.1
13.
79.
20000.
23.
10000.
-.Ob
-,0b
3.6
160.
-.05
82.
7000.
98.
1300.
59.
40000.
-.05
9.8
1300,
-.05
43.
200000.
1100.
30000,
18.
1100.
13.
-.05
2.0
30000.
-.05
750.
760.
1.7
2500.
79.
-.05
20.
4600.
2500.
3400.
460,
40000.
10000.
-.05
-.05
-.05
7.9
6.7
6.0
200000.
10000.
70000.
A-15
-------
STUDY 15 (continued)
221510
231510
241510
251510
261510
271510
281510
291510
301510
311510
011511ALSM1MICR
021511 37.1
031511
041511
051511
061511 1.3
071511 -,05
081511 -.05
091511 -.05
101511 5.2
111511 16.
121511 5.3
131511 3.4
141511 20.
151511 690.
161511 500000.
171511
181511
191511
201511
211511
221511
231511
241511
251511
261511
271511
281511
291511
301511
311511
011512ALSM3MICR
021512 52.4
031512
041512
051512
061512 .40
071512 -.05
061512 -.05
091512 -.05
-.12
-.05
-.05
-.05
7.5
2.0
2.2
900.
90.
30000.
72.
-.06
-.05
-.05
-.05
270.
-.05
-.05
-.05
34.
-.05
9.4
160.
5100.
7600.
2.9
56.
-.05
-.05
-.05
330.
-.05
1.1
26.
-.05
98.
2200.
130.
740.
9.8
78.
-,05
-.05
3.0
21.
-.05
.08
130.
-.05
86.
2800,
-.085
20000.
.19
20.
-.05
.03
-.05
1.5
-.05
.45
100.
67.
22.
43.
1800.
330.
-.05
.91
-.05
.14
-.05
-.05
-.05
,37
1.8
4100,
40000.
4100.
80000.
-.05
-.05
-.05
.05
A-16
-------
STUDY 15 (continued)
101512
111512
121512
131512
141512
151512
161512
171512
181512
191512
201512
411512
221512
$31512
241512
251512
261512
271512
281512
291512
301512
311512
25.
l.»
1.5
2.1
6.3
140.
200000.
.27
1.1
1.4
200.
56.
6000.
27.
16.
-.05
50.
56.
1500.
2000.
1.8
4.3
-.05
6.1
4600.
140.
230.
13.
66,
-.05
23.
3800.
-.04
3500.
33.
51.
14.
27.
20000.
870.
.46
700.
20000.
840.
27.
011513CERAM1MIC
021513
031513
041513
051513
061513
071513
081513
091513
101513
111513
121513
131513
141513
151513
161513
171513
181513
191513
201513
211513
221513
231513
241513
251513
261513
271513
281513
54.1
10.
-.05
.32
.89
480.
13.
35.
6.3
1600.
330.
90000.
10.
-.05
1.1
1.3
6.
.57
61.
60.
150.
3800.
100.
140.
-.05
.12
2.5
11.
-.05
1200.
360.
540.
610000.
2,3
60.
,50
45
2.7
-.05
460.
210.
510000.
400000,
390.
10.
17.
.27
13
10.
-.05
14.
88.
15.
30000.
-.05
.54
1.7
180
36.
23.
42.
3.4
150000.
100000.
-.OS
.99
.36
37
29,
890.
60000.
40000.
200000.
»
A-17
-------
STUDY 15 (continued)
291513
301513
311513
011514CERAM3MIC
021514
031514
041514
051514
061514
071514
081514
091514
101514
111514
121514
131514
141514
151514
161514
171514
181514
191514
201514
211514
221514
231514
241514
251514
261514
271514
281514
291514
301514
311514
72.0
5.6
-.05
.36
.99
290.
6.6
38.
1.4
1300.
200.
10000.
11.
-.05
9.2
1.8
6.6
.64
38.
36.
120.
4700.
230.
79.
-.05
.14
3.6
5.9
-.05
1400.
220.
210.
6700.
1.2
33.
1.
49
1.3
-.05
370.
93.
6700.
300000.
430.
5.3
3.7
.3
9.9
5.3
-.05
7.6
36.
40.
30000.
-.05
.60
4.5
200.
9,2
12.
20,
18.
150000.
60000.
-.05
1,3
.53
27,
32.
110.
60000.
40000.
150000,
011515SLUDGE1MI
021515
031515
041515
051515
061515
071515
081515
091515
101515
111515
121515
131515
141515
151515
161515
43.2
11.
-.05
.72
1.5
1700.
660.
210.
2.8
850.
4600.
4300.
2.3
-.05
.29
.97
18.
600.
18.
130.
4000.
8000.
82.
1500,
-.05
.10
1.1
280.
-.05
1300.
5600.
94.
100000.
2.4
1300.
.26
47.
.29
-.05
22.
30000.
5300.
100000.
50.
2.5
35.
.13
9.4
2.5
-.05
720.
4500.
3.8
9000.
-.05
1.4
1.8
190.
10000.
110.
73.
850.
10000.
60000.
84.
•
2.7
• 21
39,
^ w
30.
w w
100.
70000.
10000,
100000.
w w •
A-18
-------
STUDY 15 (continued)
171515
181515
191515
201515
211515
221515
231515
241515
251515
261515
271515
281515
291515
301515
3U515
oil516SLUDGE3MI
021516
031516
041516
051516
061516
071516
081516
091516
101516
111516
121516
131516
141516
161516
161516
171516
181516
191516
201516
211516
221516
231516
241516
251516
261516
271516
281516
291516
301516
311516
65.3
31.
-.05
.37
.55
2000.
790.
120.
1.7
1000,
2700.
2500.
3.1
-.05
.37
1.3
11.
500.
21.
74.
2300.
5700.
65.
1900.
-.05
.06
2.6
330.
16.
1500.
5600.
55,
100000.
14.
2200.
.31
55,
.34
-.05
24.
30000.
3800.
80000.
85.
5.3
44.
,13
11.
30.
-.05
420.
3000.
4,4
6900.
-.05
3.4
2.5
220.
5000.
33.
52.
1000.
200000.
70000.
120.
3.7
.30
42.
31.
43.
60000.
1000.
80000.
A-19
-------
DATA BASE FOR STUDY 19
01190APLANT A
02190A
03190A
04190A
05190A
06190A
07190A
08190A
09190A
10190A
11190A
12190A
13190A
14190A
1519QA
16190A
17190A
18190A
1919QA
20190A
21190A
22190A
23190A
2ai90A
25190A
26190A
27190A
28190A
29190A
30190A
31190A
9.0
26.2
-0.0015
-.0005
-.001
-.0015
.51
.005
.031
-.005
.48
11.
1.1
.0000788
1.02
1500.
12.8
1.7
1.1
1.
2.
76.
75.
-0.0015
-.0005
-.0035
-.001
.003
.001
.003
-.005
1.4
130.
.0000398
-.005
7.3
1725.
1.5
0.
69.5
-0.003
-.0005
-.001
-.001
.011
-.0005
.70
13.
2.5
6.
-.0015
-.005
2.1
0.
234.
1.0
0,159
62.
.38
-.0035
-.002
-.0005
-.0005
.19
,14
.087
18.
.12
1.61
-.00025
100.
.18
1.
1.1
1.
37.6
-.002
-.004
-.0005
.009
,095
-.0005
.47
1.
-.003
6.4
.001873
.0000788
.001085
.00085
.53
.03
168.
8.5
1.2
1.
10.
19.
24.
-.0005
.002
.10
.043
.003
.021
170.
10.
.00019
-.005
-.0025
1652.
4.
.8
10,
w •
21.2
".0005
-.005
".0005
• 00<|
".0005
» ~ w- ^r ^
-.0075
^ * +f
6.1
^ • •
8.8
9 ^J
180,
.0000398
.000006
~ %p
"•005
» » *F ^
•'
5,5
W IV
63.7
** w f
1.2
* w C»
\
•* •
01190BPLANT B
02190B
03190B
04190B
05190B
06190B
07190B
081908
0919QB
10190B
11190B
12190B
13190B
14190B
15190B
161908
1719QB
1819QB
19190B
100.
19.9
-.0015
-.0005
-.0005
-.0015
.2
.006
.004
-.0005
.41
2.1
3.7
.00139
30.
100.
-.0015
-.0005
-.003
-.001
.001
,001
.002
-.0005
.24
300.
0.
100.
too.
-.0025
-.0005
-.001
-.001
.004
-.0005
.22
1.2
.005
45.
-.0015
.000005
0.088
100.
.17
-.0015
-.0015
-.0005
-.0005
,25
.16
.079
8.
.032
1.
100.
-.001
-.0035
-.0005
-.001
.019
-.0005
.26
.017
-.0015
.69
.000934
.000515
10.
100.
100.
-.0005
-.0015
.017
-.0035
.006
.008
21.
14.
6
w»
106
* V V ,
".0005
^* V ^P
"•0045
w ^ » ~ J
-•0005
"•OOl
• " '• *
-.0005
w ^ w w ^p
• 2a
• *»^
2.4
*• • f
79
* * •
190.
' W «
A-2G
-------
STUDY 19 (continued)
20190B
2h90B
22190B
23190B
24190B
25190B
26190B 1
27190B
28190B
29190B
30190B
3H90B
OH90CPLANT
02190C
02&190C
04190C
05190C 1
06190C
07190C -.
08190C
09190C
10190C
11190C
12190C
13190C
14190C
15190C
16190C
17190C
18190C
19190C
20190C
21190C
22190C
23190C
24190C
25190C
1.39
200.
2.5
1.6
1.1
1.
1.
C
41,
3.6
0025
0005
0015
.002
1.1
.03
.009
.004
.18
2.3
12.
26190C 2400.
27190C
28190C
2*190C
30190C
3tl90C
OJ190EPLANT
02190E
03190E
04190E
05190E
06190E
07190E
3.4
2.2
1.0
1.
2.
E
7.8
48.5
0005
0005
-.005
7.5
1681.
1.2
too.
100.
-.0025
-.0005
-.0055
-.0015
.001
10.
.001
-.0005
.053
340.
10.
2924.
1.5
2.
100.
-.0005
-.0005
-.005
1.5
0.
7,
1.6
0.169
90.
72,
-.005
-.0005
-.0015
-.003
.024
-.0005
1.1
1.2
4.4
20.
-.0025
0.
24.
1.9
100.
-.001
-.0005
1.03
.0006
5,5
.004
7.3
1.0
0.163
69.5
-.125
-.0035
-.0025
-.001
-.0005
.38
.63
.71
27,
.044
.0007
8.3
.031
1.08
1.2
100.
.063
.0368
-.005
-.00025
1.2
1.
1.
17.7
-.0035
-.0065
-.001
.004
.62
-.0005
13.
.45
-.0085
.99
.004
.004
25.
1.0
1.
1.
100.
-.001
-.0015
.0000002-. 00000025
.38
-.0025
99.
20.
1.0
10.
46.5
15.5
-.0015
-.0015
.082
.011
.27
.018
41.
.18
.00233
.0025
396.
5.
3.2
10.
100.
100.
-.005
7,
3.18
1.1
1.
0.
" •
12.8
-.0005
-.0085
-.001
-.002
.004
-.0125
2.1
7.7
370.
.000464
6.
28.2
1.2
3.
0,
^ w
100.
•fc V V •
-.0005
^ W V V J
-.0015
A-21
-------
STUDY 19 (continued)
08190E
09190E
10190E
1119QE
12190E
13190E
14190E
15190E
16190E
17190E
18190E
19190E
2019QE
21190E
2219QE
23190E
24190E
25190E
26190E
27190E
2819QE
29190E
30190E
31190E
-.0005
-.0005
.29
.004
.009
-.0005
.035
1.3
13.
310.
3.4
1.
-.001
-.0005
-.0005
.001
-.0005
-.0005
.043
290.
7.5
13120.
-.0005
-.001
.011
-.0005
.094
.76
.018
9.2
-.0005
0.
10.
-.001
.007
-.0005
-.0005
.071
.10
.13
2.9
.014
-.00025
35.
,004
2.0
-.0005
.004
.13
-.0005
.92
.036
-.0005
.39
.0012
.0008
-2.5
1,
-.0005
.022
.006
.004
.002
.001
36.
2.1
.00792
-.0025
78.
.1
".0005
.019
-.0005
.035
V v «* rf
.86
• ww
26.
«• V .
70.
r v •
.0000016
8.5
** • j
3. h
•J • O
01190FPLANT F
02190F
03190F
04190F
05190F
06190F
07190F
08190F
09190F
10190F
11190F
1219QF
13190F
14190F
15190F
16190F
1719QF
18190F
19190F
20190F
21190F
22190F
23190F
24190F
25190F
26190F
8i,7
32.3
-.0015
-.0005
-.001
-.0015
.16
.004
.011
-.005
.13
56.
,001992
1.86
1900.
100.
100.
-.002
-.0005
-.0035
-.001
-.0005
-.0005
-.0005
.001
.015
38.
-.0015
5.17
7.4
2006.
0.
85.
100.
-.0035
-.0005
-.001
-.002
.002
-.0005
.28
1.5
.022
48.
.004
.94
4.27
0.
10.
0.114
100.
.033
-.0025
-.002
-.001
-.0005
.054
.49
.1
4.6
.03
.0009
2.4
.004
1.
100.
1.
-.0025
-.0045
-.0005
.002
.19
-.0005
.16
.024
27.
15.
.036983
.02753
.0021446
.001
.19
.0003
96,
10.
100.
100.
10.
-.001
.003
.004
.005
.006
.021
10.
490.
0.
.33
-.0025
276.
.1
«
100
• V v .
Q
-.0005
-,0055
".0005
.002
".0005
-.0085
3 7
c . r
.83
p Q
»• . T
•0000043
.in
« * H
•
16.
A-22
-------
STUDY 19 (continued)
2T190F
28J90F
29190F
30190F
31190F
1.54
1.7
1.3
1.
1.
1.5
1.1
.56
9.1
1.3
1.
1.1
1.2
1.
0tl90GPLANT G
02190G
03190G
04190G
OS190G
06190G
07190G
08190G
09J90G
J0190G
11190G
12190G
13J90G
J4190G
15190G
16190G
17190G
J8190G
19190G
20190G
211*06
221*06
23190G
20190G
25190G
26190G
27190G
28190G
29190G
30190G
311*06
62.4
41.4
-.0005
-.0005
-.0005
-.0005
,23
.002
.011
-.0005
.17
1.1
3.2
,0002118
2.9
155.
72,5
1.0
1.1
1.
1.
100.
100.
-.0005
-.0005
-.0005
-.0005
.001
.003
-.0005
-.0005
.016
43.
.01062
1.39
7.5
276.
0.7
0.
59.
100.
-.0005
-.0005
-.0005
-.0005
.007
-.0005
.069
.64
.12
16.
-.0005
3.18
1.55
0.
5.
1.3
0.087
100.
.10
-.0005
-.0005
-.0005
-.0005
.017
.11
.11
8.7
.32
.72
-.00025
30.
.003
4.
1.2
100.
-.0005
-.0005
-.0005
-.0005
1.2
-.0005
.036
.038
-.0005
2,2
,0554699
.0085216
.0056
.23
.011
42.
1.1
1.
64.7
100.
-.0005
-.0005
.004
,032
.006
.001
.13
16.
1.4
.000132
.09
.0025
502.
-1.
1.4
100.
-.0005
-.0005
-.0005
.004
-.0005
.014
1.6
20.
48,
.01062
.0000076
.03
8.
27.2
1.2
1.
0U90KPLANT K
02190K
03190K
04190K
00190K
06J90K
0,7190K
08190K
09190K
10190K
11190K
12190K
13190K
i«!90K
100.
39.5
-.001
-.0005
-.0005
-.001
.19
.003
.005
.01
.008
100.
100.
-.001
.,0005
-.002
-.001
-.0005
-.0005
-.0005
.001
.09
0.
77.
100.
-.002
-.0005
-.0005
-.001
23.
-.0005
.19
.77
.015
0.084
100.
.017
-.0015
-.001
-.0005
-.0005
.027
.11
.016
1.
100.
-.0005
-.0025
-.0005
-.0005
.048
-.0005
2.9
.014
-.001
10.
100.
100.
-.0005
-.0005
.011
.006
-.0005
.012
24.
1.
100.
-.0005
-.003
-.0005
-.001
-.0005
.32
.72
26,
A-23
-------
STUDY 19 (continued)
15190K
16190K
17190K
18190K
19190K
20190K
21190K
22190K
23190K
24190K
25190K
26190K
27190K
28190K
29190K
30190K
31190K
01190LPLANT
02190L 1
03190L
04190L
05190L
06190L
07190L
08190L
09190L
10190L
1U90L
12190L
13190L
14190L
15190L
16190L
17190L
18190L
650.
87.
875.
3.1
2.4
0.9
1.
1.
L
00.0
28.0
9.4
0005
0005
0005
0005
.37
.003
.026
0005
.27
2.1
2.5
105.
7,2
1256.
1.2
67.1
34,1
42.
100.
-.0005
-.0005
-.0015
-.0005
.007
.001
.001
-.0005
330.
6.8
-.001
0.
14.
1.5
0.
1.7
100.
-.001
-.0005
-.0005
-.0035
.005
-.0005
.99
2.4
.23
10.
-.0005
120.
.005
-.00025
710.
.004
2.5
2.7
47.5
0.174
100.
.14
.001
-.0005
-.0005
-.0005
.2
.54
.24
15.
2.1
19190L .04113
20190L
21190L
22190L
23190L
24190L
25190L
26190L
27190L
2S190L
29190L
30190L
31190L
OU90NPLANT
02190N
5.51
555.
.5
1.5
1.1
1.
2.
N
61,7
11.7
5.8
725.
1.3
2.
34.2
6.6
1.27
1.81
0.
42.
1.6
2.
0.
.22
-.00025
30,
.03
.88
1.5
2.
33.6
.211
.71
-.00005
.0008
-2.5
1.0
1.
1.
100.
-.0005
-.0015
-.0005
-.0005
.3
-.0005
.51
.25
-.0005
.44
.02308
.04113
.00325
.046
.86
.003
13.
1.1
1.
1.
13.
.00044
-.0025
131.
1.
0,9
15.8
10.
23.5
100.
-.0005
.011
.046
.015
.002
.033
110.
4.6
.00135
.86
.0025
234.
3.
1.1
1.1
10.
120,
.0000056
5.
2.75
1.0
1,
s.
w
100.
-.0005
-»ooa
-.0005
.005
.005
.031
^ » ** ^
4.5
* W «P
5»5
7e.
^* 9
.0000864
,97
* w •
a
ia.i
• ° . j
l.o
* • w
t
« *
fc.
A-24
-------
STUDY 19 (continued)
03190N
04190N
05190N
06190N
07190N
08190N
09190N
10190N
11190N
12190N
13190N
I4190N
J5190N
I6190N
J7190N
J8190N
J9190N
20190N
21190N
22190N
23190N
24190N
25190N
26190N
27190N
28190N
29190N
30190N
JJ190N
01190S
021903
031903
041903
05190S
061903
07190S
081903
09190S
101903
111903
12190S
131903
141903
151903
161903
171903
181903
191903
20190S
21190S
1.
40.7
-.002
-.0005
-.001
-.0015
1.3
.004
.054
.005
27.
1.1
41,
.00231
2.05
990,
12.8
1.9
1.2
1.
3.
PLANT S
10.7
-.001
-.0005
-.0005
-.0005
,097
.008
.016
-.0005
.1
2.5
1.1
.00391
2.
75.
-,002
-,0005
-.004
-.0015
.001
.002
.017
.002
44.
1400.
.89
7.
1352.
1.5
2.
100.
100.
-.001
-.0005
-.0015
-,0005
.002
.001
-.0005
-.0005
.016
24.
2.3
75.
-.004
-.0005
-.0015
-.0025
.12
-.0005
2.1
580.
.033
110.
-.002
4.45
1.66
20.
13.
1.0
2.
100.
100.
.012
-.0005
-.001
-.0005
.017
-.0005
.027
.29
.006
15.
-.001
1.23
47.5
.95
-.002
-.002
-.0005
-.0005
.51
.11
.089
54.
.033
.10
-.00025
0.
1.8
11.2
1.2
2.
100.
.085
-,0015
-.002
-,0005
-.0005
.81
.28
.01
18.
.011
.47
43.5
.001
-.005
-.0005
.002
.12
-.0005
.19
.39
-.002
110.
.006774
.00673
.00365
.0646
.39
.002
36.
1.7
1.
1.
100.
-.005
-.0025
-.0005
-.0025
.84
-.0005
13.
.005
-.001
18.
.03682
.00825
.03868
.015
1.97
48.8
37.4
-.0055
-.002
.005
.008
,030
,063
.46
570.
42.
.00055
.21
-.0025
266.
.1
1.3
10.
100.
100.
-.006
-.0005
-.0025
.024
.021
.001
-.0005
11.
2.8
.00044
1.11
26.3
-.0005
-,006
-.0005
.006
.001
.40
80.
58.
150.
.00146
.0000003
.10
9.
9.24
1.3
3.
5.
100.
-.0005
-.002
-.001
-.0035
.002
,064
1,
37.
95.
' ~ 9
.0000033
.66
A-25
-------
STUDY 19 (continued)
22190$
23190S
24190S
25190S
26190S
27190S
2819QS
29190S
30190S
31190S
2.68
640.
72.5
1.7
1.2
1.
1.
.65
7.8
692.
1.5
1.36
0.
349.
1.3
-,00025
130.
0.
72.
1.4
.074
59.
1.7
1.
-.0025
1035.
.3
2.1
7t
5.40
** • ~ V
1.1
* w «•
t
* *
01190TPLANT T
021 90T
03190T
04190T
05190T
06190T
07190T
08190T
09190T
10190T
11190T
12190T
13190T
14190T
15190T
16190T
17190T
18190T
19190T
20190T
21190T
22190T
23190T
24190T
25190T
26190T
27190T
28190T
29190T
30190T
31190T
01190U
02190U
03190U
04190U
05190U
06190U
07190U
0819QU
0919QU
100.
43.9
.002
-.0005
-.0005
-.0005
.022
.001
.006
-.0005
.059
.51
.95
.0166
1.74
460.
13.6
1.7
1.0
1.
2.
PLANT U
12.1
18.5
-.0005
-.0005
-.0005
-.0005
100.
100.
-.0005
-.0005
-.001
-.0005
.001
.002
,002
-.0005
.058
.70
2.92
7.4
660.
1.2
100.
100.
-.0005
-.0005
-.0005
-.0005
0.
70.
100.
-.0005
-.0005
-.0005
-.0005
.002
-.0005
.03
.29
.014
.79
-.0005
1.85
1.71
0.
44.
1.0
0.
100.
100.
-.001
-.0005
-.0005
-.0005
0.131
68.
.042
-.001
-.0005
-.0005
-.0005
.13
.04
.019
1.9
.008
.26
-.00025
300.
0.
6.4
1.2
0.
100.
.006
-.0005
-.0005
1.
100.
-.0005
-.001
-.001
.002
.009
-.0005
.13
.045
-.0005
3.3
.05196
.02277
.00287
.01
.67
32.
1.3
1,
1.
100.
-.0005
-.001
-.0005
-.013
10.
46.5
100.
-.0005
-.0005
.002
.005
.005
.005
.002
5.2
1.4
.00008
1.29
-.0025
414.
6.
1.1
10.
100.
100.
-.0005
-.0005
-.0005
0.
*
3
-------
STUDY 19 (continued)
10190U
11190U
12190U
13190U
14190U
15190U
16190U
17190U
18190U
19190U
20190U
21190U
22190U
23190U
2al90U
25190U
26190U
27190U
28190U
29190U
J0190U
J1190U
.16
.003
.002
-.0005
.53
170.
2.4
.00383
7.3
770.
5.44
1.7
1.1
1.
1.
.ooe
-.0005
.001
-.0005
.005
16.
3.83
7.3
1331.
1.4
.076
-.0005
.32
16.
.002
6,4
-.0005
19.7
7.93
0.
111.
1.7
-.0005
-.0005
.043
.099
.024
14.
.023
2.7
-.00025
120.
.014
2,96
1.7
.19
-.0005
.55
.058
.003
.24
.09303
.01888
,00405
0.
3.67
.001
34.
1.5
1.
.003
.007
.018
.1
180.
11.
.00008-.
1.47
,0025
748.
3.5
0.9
-.0005
.14
.12
37.
38.
,0015
00000025
1.77
9.
14.6
1.2
1.
01190VPLANT V
02190V
03190V
04190V
05190V
06190V
07190V
08190V
09190V
10190V
11190V
12190V
13190V
14190V
15190V
16190V
17190V
18190V
19190V
20190V
21190V
22190V
23190V
24190V
25190V
26190V
27190V
28190V
9.4
43.4
-.0005
-.0005
-.0005
-.0005
.36
.011
.002
-.0005
.31
1.
1.4
,00237
.39
360.
2.5
1.2
100.
-.0005
-.0005
-.0005
-.0005
-.0005
-.0005
.001
-.0005
.066
45.
.00538
2.26
7.1
0.7
0.
94,
.003
-.0005
-.0005
-.0005
-.0015
-.0005
.19
3.1
.012
4.2
-.0005
1.53
1.98
0.
1.3
A- 27
0.
.089
-.001
-.001
-.0005
-.0005
.022
2.
.12
9.5
.041
.49
-.00025
.4
.003
1.7
1.4
1.
-.0005
-.001
-.0005
-.001
.01
-.0005
1.1
.01
-.0005
4.
.01863
.03109
.00002
.0057
.38
.004
-2.5
0.8
10.
36.
-,0005
-,0005
,005
.009
.005
-.0005
.073
43.
7.4
.000088
.55
-.0025
128.
.5
1.1
0.
-.0005
-.001
-.0005
.003
-.0005
.012
4.7
2.
24.
.00538
.0000264
1.16
9.
1.2
-------
STUDY 19 (continued)
29190V
3019QV
31190V
1.1
1.
1.
01190W PLANT W
02190W
03190W
04190W
05190W
06190W -.
07190W
08190W
09190W
10190W
11190W
12190W
13190W
14190W
15190W
16190W
17190W
18190W
19190W
20190W
21190W
2219QW
23190W
24190W
25190W
6,3
10.3
0015
000$
0005
.001
0005
.006
.023
.002
.22
1.2
11.
0248
.42
26190W 1250,
27190W
28190W
29190W
30190W
31190W
01190XPLANT
02190X
0319QX
04190X
05190X
06190X
0719QX -.
08190X -.
0919QX
10190X
11190X
12190X
13190X
1419QX
15190X
1619QX
.38
1.3
1.3
1.
3.
X
100.
25.1
0005
0005
0005
0005
.13
0005
0005
.005
.028
130.
11.
94.
42.
-.0015
-,0005
-.003
-.002
.003
.003
.006
.002
.028
16.
1.41
e.i
1648.
1.3
100.
100.
-.0005
-.0005
-.001
-.0005
-.0005
-.0005
.007
.001
.027
2.9
0.112
50.
41.
-.0025
-.0005
-.001
T.0015
-.0065
-.0005
.31
.06
.011
.12
-.0015
2.09
1.17
0.
217.
1.7
0.
100.
100.
-.0005
-.0005
-.0005
-.0005
.037
-.0005
.14
.14
.026
29.
-.0005
0.152
37.5
.17
-.0015
-.002
-.0005
-.0005
2.
.072
.079
30.
.054
.34
.0005
950.
.003
.075
2.2
0.132
100.
.03
-.0005
-.0005
-.0005
-.0005
.023
.87
.018
18.
.052
1.
37.5
-.001
-.0035
-.0005
-.002
-.0065
-.0005
.55
.024
9.001
7,3
.06161
.00576
.00016
0.
1.32
-.00025
84.
1.3
1.
1.
100.
-.0005
-.001
-.0005
-.0005
1.9
-.0005
2.5
.006
-.001
8.2
10.
55.2
24.8
-.005
-.005
.011
-.0035
.018
.048
94.
12.
.00123
.85
.004
837.
.1
1.3
10.
100.
100.
-.0005
-.0005
-.0005
.004
.005
-.0005
.12
23.
4.6
0.
19.6
-.0005
-.0045
-.0005
-.003
-.0005
• 033
• 67
660,
8.3
.0000145
• 57
5.
— »
15.
w
1.1
™ w •
3.
™ •
0.
w •
100.
w w •
"•0005
••001
^ m
••0005
— ™ W ff
.001
^ •
••OOOS
— -V V ,^
• 022
**^»
2,5
9.
wl
^ 9
A-28
-------
STUDY 19 (continued)
17190X
18190X
19190X
20190X
21190X
22190X
23190X
24190X
25190X 7.2
26190X 285. 437.
27190X ,05
28190X 1.9 1.5
29190X 1.0
2?0190X 1.
31190X 1,
01190DPLANT D
021^00
031900 100.
041900 100.
0'5190D 46.7
06190D
07190D
061900
091900
101900
111900
121900
131900
141900
151900
161900
171900
181900
191900
201900
211900
221900
231900
241900
251900
261900
27190D
281900 5.2 2.3
391900 1.2
301900 1.
311900
01190HPLANT H
02190H
03190H 96.
04190H 100.
.0001 .0000033 ,000044
.0009 .0009 -.0025
°» HO. IS. 256. 7.2
1.3 .039 ,ot 13.5
0.
1-2 1.3 1.6 0.9 1.2
1. 1.
°«
o. o. 10. 6.
100.
•.00025 .002 .006
2.6 3.4 2.6 2.6 1.6
1.
28.
0. 0.164 lo, 5t
A-29
-------
STUDY 19 (continued)
0519QH 31.3
06190H
07190H
08190H
09190H
1019QH
11190H
12190H
13190H
14190H
15190H
16190H
17190H
18190H
19190H
20190H
21190H
22190H
23190H
24190H
25190H
26190H
27190H
28190H 1.5
29190H 1.0
30190H 1.
3U9QH
01190JPLANT J
02190J
03190J
04190J 100.
05190J 39.1
06190J
07190J
08190J
09190J
10190J
11190J
12190J
13190J
14190J
15190J
16190J
17190J
18190J
19190J
20190J
2U90J
22190J
23190J
1.4
1.2
-.00025
1.4
.006
1.3
1.
0.
0.
100.
.0025
0.9
48.
10.
100.
1.1
10.
-.00025 -.00025
.0025
A-30
-------
STUDY 19 (continued)
24190J
25190J
26190J
27190J
28190J 1.3
29190J 1.0
30190J 1,
3U90J
01190MPLANT M
02190M
&3J90M
04190M
&5190M
06190M
57190M
08190M
09190M
10190M
Ul^OM
12190M
13190M
14190M
15190M
16190M
17190M
18190M
19190M
20190M
21190M
22190M
23190M
26.3
60.
22.8
25190M
26190M
27J90M
28190M
29190M
30190M
31190M
01190PPLANT
1.0
1.
Q3190P
05190P
06190P
07190P
08190P
Q9190P
10190P
I1190P
100.
19.0
l'a
52.6
78.8
100.
l-7
0.166
90.8
0.124
.8 1.3
1.
31.5
10.
1.1
1.
-.00025
1.9
.OOa -.0025
2.
1.
0.9
1.
1.3
1.1
10.
100,
1.
A-31
-------
STUDY 19 (continued)
12190P
13190P
14190P
15190P
16190P
17190P
18190P
19190P
20190P
21190P
22190P
23190P
24190P
25190P
26190P
27190P
2.8190P
29190P
30190P
31190P
01190RPLANT
02190R
03190R
04190R
05190R
06190R
07190R
08190R
09190R
10190R
11190R
12190R
13190R
1.7
1.2
1.
93.2
8.0
26.7
1.3
90.2
87.1
93.
1,6
-.00025 -.00025 .0025
1.4 1.0 1.0
1.
97.2
0. 0.
0.
10.
16.5
1.3
5.
15190R
16190R
17190R
18190R
19190R
20190R
21190R
22190R
23190R
24190R
25190R
26190R
27190R
28190R
29190R
30190R
-.00025 -.00025
.0025
1.6
0.9
1.
1.4
1.
1.6
1.
1.5
3.
1.5
1.
1.0
1.2
A-32
-------
STUDY 19 (continued)
31190R
01190YPLANT Y
02190Y 0.135 0. 10. 10.
03190Y 100. 100.
04190Y 100.
05190Y 33.2
06190Y
07190Y
08190Y
09190Y
10190Y
1J190Y
12190Y
14190Y
13190Y
16190Y
17190Y
18190Y
19190Y
20190Y
2I190Y
22190Y
23190Y ..00085 .003 -.0025
24190Y
25190Y
26190Y
27190Y
28190Y 2.0 1.6 1.6 1.7 1.5 1.6 1.1
29190Y 1.1
30190Y 1. !.
31190Y
01190ZPLANT Z
02190Z 0.131 0.114 10. 1.
03190Z 18. 100.
04190Z 42.6
05190Z 18.6
06190Z
07190Z
06190Z
09190Z
101902
H190Z
12190Z
13190Z
1S190Z
16190Z
17190Z
A-33
-------
STUDY 19 (continued)
19190Z
20190Z
21190Z
22190Z
23190Z -.00025 .012 .0025
24190Z
25190Z
26190Z
27190Z
28190Z 1.2 0.8 1,8 0.8 0.9 1.8
29190Z 1.1
30190Z 1. 1.
3119QZ
A-34
-------
APPENDIX B
DRAFT REPORT FROM DON LEWIS
Don Lewis, a biostatistician employed in the Process Measurements
Branch, IERL/EPA, examined the data from the three pilot studies. His draft
report follows.
B-l
-------
Introduction
The Process Measurements Branch, IERL/RTP has developed a phased sampling
and analytical strategy for environmental assessment programs, the first phase
of which is directed at the quantification of mass emissions for inorganic
elements and organic classes (1,2). Included in this phase, Level 1, is a
set of biological tests designed to determine the biological effects, both
ecological and health-related, of the samples collected.
The purpose of this report is to describe the apparent relationships
between the biological test results and the chemical analyses of the samples,
as well as provide a descriptive analysis of the data used to quantify these
relationships. Discussion of the results and suggestions for future work will
follow thereafter.
The data represent biological and chemical analyses from pilot studies of
four pollution sources: (1) secondary wastewater from U.S. textile plants
(textile study), (2) air emissions from a pressurized fluidized-bed combustion
unit (Exxon study), (3) air emissions from a low-BTU gasification facility
(Radian study), and (4) gaseous effluents from a collection of different
industrial sites (Mitre study).
A total of 51 samples were collected. In most cases only one sample was
taken per plant; in the case of the Exxon study all of the samples were collected
at the same plant but under varied conditions.
Of the 51 samples 23 were collected from the textile plants, eight of which
had little or no chemical analysis performed, rendering them useless for the
purposes of this report.
B-2
-------
Twelve samples were analyzed from the Exxon and Radian studies and 16 from
the Mitre study. Complete data are not available from any of the studies; hence,
the statistical analyses represent a "piecemeal" approach, the results of which
should pertain only to the corresponding source of data.
The data were compiled by N. H. Gaskins and F. W. Sexton, Systems Measurement
Division, Research Triangle Institute, Research Triangle Park, North Carolina from
references 3-7 and by personal contact with the contractors who collected the data.
References 1-7 should be consulted for specific information about the nature of
the data, how it was collected, by whom, etc.
The Data
One hundred and ninety-one variables were measured in the study, 150 of which
are chemical analyses. The chemical analyses represent concentrations of the 74
primary elements, the 26 multi-media environmental goal (MEG) values, the eight
liquid chromatography (LC) fractions, the seven gas chromatography (GC) values,
and measurements of 35 other physical and chemical parameters of interest. A
list of these parameters is given in Table 1.
The forty-one biological parameters represent 31 tests for mutagenicity,
cytotoxicity, lethality to selected aquatic organisms and rats, and growth
inhibition of selected algal species, plus the Salmonella test strain revertent
ratios that form the basis of the Ames test for mutagenicity. Reference 8 can
be consulted for details regarding the Ames test and related information. The
individual tests are listed in Table 2 with a brief description included.
A severe limitation in this study is the amount of data available for analysis.
More than 80% of the biological data was not collected, or the tests were not per-
formed. Table 3 gives a breakdown by study of the biological parameters and the
data that is available. Note that even when data are available for a certain
B-3
-------
parameter, there may not be a distribution of the responses, eliminating the
possibility of evaluating the relationship of that variable to the chemical
factors.
The body of data do not represent a homogenous sample which can be considered
in total. The analyses of the textile samples were made on secondary effluent
rather than untreated effluent. The non-textile samples represent effluent from
dissimilar pollution sources to that of the textile plants; hence, the samples
are not particularly comparable, at least chemically. This will be discussed
further in the Results section. Henceforth, all results will be reported by
study-type.
An additional problem with much of the chemical data is the presence of
missing values. Since much data were missing, it was necessary to determine
whether the data were in fact not collected or whether the result fell below a
"minimum detectable limit (MDL)", and therefore was not reported. All possible
missing values which were not reported due to the latter case were identified
and replaced by one-half of the MDL. The decision was made to substitute these
values in order to minimize any possible bias (negative or positive) in the
results. Therefore, all remaining missing values represent data which were
not collected or which were not analyzed.
The Mitre study is included in the data set although the only analyses that
were completed were the 74 elemental values and the RVI biological parameter.
Since the goal of the study was to quantitate the biological/chemical relation-
ship in the data, the Mitre study data was not included in the descriptive
section on the chemical analyses in this report.
B-4
-------
Methods
The statistical methods utilized are grouped according to the two-fold
purpose of this report: descriptive analysis of the parameters and statistical
analysis of the relationship between the parameters.
The descriptive portion of the study utilized typical summary statistics:
sample mean (X), sample standard deviation (A), minimum (min), maximum (max),
and coefficient of variation (CV).
To quantify relationships between parameters both Pearson's product-moment
correlation and Spearman's rank correlation were employed as well as multiple
regression techniques (9). Few multivariate methods could be utilized due to
the severe restrictions missing data places on these methods. Spearman's rank
correlation was preferred to the product-moment correlation because fewer
assumptions are required of the data. In particular Pearson's correlation
presumes that the two variables follow the bivariate normal distribution, a
very tenuous assumption to make about any of these variables. Generally, the
chemical data are highly skewed upward while concentrated near the censored
value or MDL below.
When either Pearson's correlation or multiple regression is used, natural-
log transformations of the data are made to ease this difficulty.
Results
The results are presented in two sections: (a) descriptive analysis and
(b) analysis of biological/chemical associations.
(a) Descriptive Analysis. Descriptive statistics of the chemical parameters
are listed by study in Tables 4 and 5 (SSMS - elemental analysis) and Tables 6 and
7 (other physical and chemical measurements). The non-textile plants are consid-
B-5
-------
ered a homogenous sample for these purposes (Mitre study samples not included in
these results).
In general the data show high degrees of variability even within the afore-
mentioned categories. Most of the parameters have a lower range of zero or the
MDL. The CV's of these parameters generally exceed 1.0 and others exceed 2.0,
implying high variability of the sample values relative to their average concen-
trations. CV's exceeding 2.0 usually indicate a sample mean which is dominated
by a maximum value entirely out of range of the remainder of the data. In these
cases the sample mean is not a good measure of central tendency and should be
disregarded. If it can be determined, the sample median is a better statistic
to consider.
Clearly the textile data exhibit substantially lower concentrations and levels
of the chemical parameters than do the non-textile data, due to the difference in
the nature of the two sources, as mentioned earlier.
The biological tests for which enough data were available for correlation
analysis are noted by a * in Table 2. Descriptive statistics for the continuous-
type variables and distributions of the discrete variables are given in Tables
8 and 9.
In general the test results do not exhibit a wide range of response to the
pollutant samples. There was no distribution of response to the textile effluent
in the Ames test with continuous data (AMES) and the Ames test with discrete data
(AMES2). In fact AMES and AMES2 in each case registered no mutagenicity. The
non-textile data exhibit a somewhat wider response with three of eight samples
showing high mutagenicity according to AMES and three of nine samples exhibiting
at least slight mutagenicity according to AMES2. The Rabbit Alveolar Macrophage-
B-6
-------
test with discrete data (RAM2) exhibited similar distribution in the non-textile
data. Data were available on the Grass Shrimp (GS) test and the Fresh Water Fish
(FWF) test only for the textile plants; however, these data were distributed
across the range of responses.
These variables represent the data available for statistical anal/sis among
those variables measured. Other variables where data were available but were
entirely or predominately constant in value are RLD50, FWA, FWF96, DAPH48, PRAM,
PATP, SWF at 24, 48, and 96 hours and GS at 24, 48 and 96 hours. The remaining
biological variables were predominately missing values or were not measured.
Also compiled as a part of the data set but not summarized here are the ten
variables representing the revertant ratios, from which AMES is quantitated.
Data are available on these variables from the textile, Exxon and Radian studies.
(b) Statistical Analysis. Where data were available, each of the biological
variables was correlated with each of the chemical variables by computing the
Spearman's rank correlation. The coefficients were then tested under the hypothesis
of no association, or equivalently, that the true correlation is equal to zero. The
significance level of each correlation coefficient was then calculated, and those
whose two-sided p-values fell below 0.05 were judged to exhibit "significant"
association. Rather than listing all of the correlations computed (over 2000),
only those which were significant according to this criterion are listed and are
found in Tables 10, 11 and 12.
In general there is not a great deal of evidence showing strong associations
between the chemical tests and biological results. The number of significant
correlations cited does not exceed the number that should be expected given a
significance level of 0.05 and the assumption that there are no underlying
B-7
-------
relationships being measured. No particular pattern emerges from the correlations
which are significant, i.e., the biological tests do not agree to a large extent
in the "choices" of chemicals to which they are sensitive.
To qualify the above remarks, the following points must be made:
(1) The sample sizes involved in the bulk of the correlations are
less than 15. With such small sample sizes very large correlations must
result before significance is reached. In other words there may be under-
lying associations that were not identified due to the paucity of data.
In addition, much of the data is missing and possible relationships
could not be measured.
(2) Many of the chemical concentrations may not be in the range
of response of the biological parameter. Because a large number of the
chemicals were scattered near the MDL, no range in the biological tests
was observed; hence, no correlation resulted.
(3) The biological result represents a response to a host of
chemical substances and to possible interactions of those substances.
A response can't be linked directly to a specific chemical in the
presence of others. One can control for the effects of extraneous
variables in the analysis; however, to do so without knowing which
to control for (out of a multitude to choose from) is a huge task
which these data do not warrant.
With that in mind each biological test will now be discussed in regard to
the chemical association observed. The following tests involve the non-textile
data only.
B-8
-------
AMES was found to be positively correlated with the elements Bi, Pb, La,
and MEG categories 2 and 15. (Recall that AMES did not vary in the textile data.)
AMES2 was found to correlate positively with the same chemical constituents as AMES
with the exception of Lu, denoting general agreement between the two. In addition
AMES2 correlated negatively with Cl, C03~, and S03" and positively with F. Note
the very small sample sizes involved here.
RAM2 was found to correlate positively with I, MEG categories 1, 21, and 25
and Hg-AA (non-textile data only). RATP was found to correlate negatively with
MEG categories 1, 21, and 25. Not surprisingly RAM correlates with the same
variables, since RAM is a function of RATP (RAM = (RATP + RVI)/2). The RVI test
did not correlate with any of the chemical variables, despite the largest sample
size (n=28) available to any of the biological variables.
The following correlations were based upon the textile data only. RT was
found to be positively correlated with Pb, Zr, and Fe while negatively correlated
with Ta, Yb, MEG category 1 and LC fraction 4. SWA exhibited positive correlations
with Sm and LC fraction 2 and negative correlations with La, Zr, V, and NO ~. GS
exhibited positive associations with Pr, Ag, Y, Se, and Ni and negative correla-
tions with Dy, Eu, Be, and LC fraction 6. The FWF test correlates positively
with Ag, Zr, Y, Sr, Rb, Ni, and NO,". Negative correlations are found with Dy,
ij
Sm, and Cl.
The Salmonella strain ratios were also correlated with the chemical parameters.
TA1535- was found to correlate with 18 variables, and 13 of these correlations
were negative. Hence, in'general this ratio decreases with increasing chemical
concentrations. TA1535+ correlates with eight chemical parameters, six of which
are also negative. TA1537- correlated with only three chemicals. TA1537+ was
sensitive to 13 chemical parameters, and eight of these correlations were
B-9
-------
negative. TA98- was positively correlated with four chemical variables and TA98+
with 12, all of which were positive. The most sensitive of the strains to the
chemical constitution of the samples was TA100-, correlating positively with 38
parameters, all of which were primary elements. TA100+ related inversely with
six of seven chemical parameters. TA1538- correlated only with Hf, Si, and CO-
O
TA1538+ correlated positively with eight of ten variables, the others negative.
Table 12 lists all of the significant correlations by strain. These results
indicate that a greater degree of association between the chemical and the
revertent ratios is apparent than indicated by the correlations between the
chemicals and AMES.
Evidence for an association between a chemical parameter and the biological
response to a pollutant is of greater interest if more than one biological test
is associated with that chemical. Table 13 lists the twelve chemical constituents
which were found to be associated with at least two biological tests. Of the
primary elements, Bi, Pb, Cl, Zr, Ag, Dy, and Y are all significantly related to
at least two tests. In addition MEG categories 1, 15, 21, and 25 as well as NO ~
are associated with at least two tests. The significant associations between
these MEG categories and the tests are based upon the non-textile data results.
It should be noted that much of the MEG data from the textile plants is
missing, or to be more precise, was not found by the chemical analysis. Only
MEG categories 1, 8, 15, and 18 exhibit 8 or more data points. In comparison
data from all of the 26 categories were present from the non-textile plants.
Hence, an explanation for the lack of correlation between the biological tests
and the MEG categories in the textile data is the absence of these constituents
in the samples. Had these MEG categories been present, more correlations could
B-10
-------
have been observed. The necessary point to be made is that the lack of observed
correlation is not based upon statistical testing but the result of an absence
of data.
As a measure of agreement the rank correlations between each pair of the
revertent ratios were computed and are given in Table 14. Note that none of the
non-activated strains are associated with their respective activated strains.
The overall agreement among the strains is not particularly good, there being
only seven significant correlations out of a total of 45 intercorrelations.
A question could be raised here: which of these strains "contributes" the
most to the variation of the Ames indicator variable? One possible answer might
be found by performing a multiple regression of the strain ratios on AMES. This
was accomplished after first logging the values of the ratios.
With a sample size of only nine, it was found that the best pair of pre-
dictors of AMES were TA98- and TA1538+, explaining 90% of the variation in AMES.
TA98- was most highly correlated (unconditionally) with AMES followed by TA1538-
and TA100-. However, both of the latter were found to correlate with TA98-
(Table 14); hence, their effects are eliminated after consideration of TA98-.
These results should be interpreted with caution since they are based on only
nine observations.
An additional indicator of agreement between the biological tests would
be the rank correlations among the tests. Table 15 gives those correlations by
data source. The results do not reveal a great deal of agreement or association;
the significant correlations occur between tests which are either functions of
or slightly different versions of one another. Worthy of note is the high
correlation between GS and FWF and the inverse correlation between FWF and SWA,
all ecologically-oriented tests. The RVI-RATP correlation indicates that the
two types of tests involving the rabbit alveolar macrophage are in good agreement.
B-ll
-------
Probably of greatest interest in this analysis are results aimed at the
simultaneous consideration of the chemical variables in relation to one or more
biological parameters. This can be accomplished statistically through multi-
variable analyses by determining the combination of variables and their inter-
actions which explain significantly the variations in the biological parameters.
A group of eleven elemental variables identified by IERL chemist, Frank Briden,
as potentially toxic were used as the independent variables in regression
analyses modelling the biological parameters. The group consisted of Be, Cd,
Ti, As, Sb, Pb, Hg, Cr, Se, V, and Te.
For each biological variable, where data were available, a step-wise
regression was performed, and independent variables which were found to enter
the regression at the 0.05 level of significance were identified. If two or
more chemical variables were found to enter the regression, then the respective
interaction terms were also tested. Due to the nature of the data, the trans-
formation X = log (c+1), where C = chemical concentration, was made on each
chemical.
The difficulties with missing data arise here to complicate this analysis.
When considering a group of variables simultaneously, only those cases (samples)
which have complete data on these variables can be used in the analysis. As a
result the sample sizes are often reduced drastically, to the point that the
statistical results may not have meaning. In addition, the sample size may
change with the addition of any new variable to the regression, and the results
may change accordingly as cases are added or deleted. Nevertheless, regressions
were performed on ten of the biological variables and all of the Salmonella
strains. A similar logarithmic transformation was made on the ratios to bring
the data in line with the usual regression assumptions of normality in distribu-
tion of the dependent variable.
B-12
-------
For nine of the ten biological tests none of the eleven elements were found
to explain a significant amount of variation in the test. Hence, no relationship
between these chemicals and the biological response is apparent. Only for SWA
was a variable found to be significant in the regression (the element, CR).
However, this relationship should be suspect due to two outliers in the data
and the failure of the rank correlation to corroborate this finding.
The ten Salmonella strain ratios were analyzed after first controlling
for the effects of the data source (study-type) in the analysis by use of an
indicator variable. TA1537- was found to be associated significantly with Cd
(a=0.01). Note, however, that the rank correlation between these two variables
does not substantiate this. None of the remaining strains were found to associate
with any of the eleven elements after controlling for study-type. Although the
indicator variable for study-type was not generally a significant regressor in
those analyses, more often than not the variable explained more variation in the
log-ratios than any of the elements.
Additional regression analyses were performed using the MEG categories as
regressors. The focus of these investigations was the identification of the sets
of MEG values which explain significant variation in the biological variables.
One series of regressions was performed for each study-type. Quite strong
restrictions on data availability, as mentioned earlier, reduced the choice of
variables in each case. For the textile study only MEG categories 1, 8, 15,
and 18 were considered. For the non-textile study only those MEG categories
which exhibited significant rank correlations with at least one biological
test were considered. This group consisted of MEG's 1, 2, 15, 21, 23, and 25.
B-13
-------
The results of the analyses for the textile plants were not especially
revealing. The only significant relationship between the MEG categories and
the biological tests was that between RT and MEG1, corresponding to the signifi-
cant rank correlation found earlier. All other regressions did not exhibit
statistically significant relationship. As a reminder, only MEG categories 1, 8,
15, and 18 were investigated.
For the non-textile plants no relationships were found between the MEG cate-
gories and AMES, contrary to the correlation results (Table 10). A strong relation-
ship between AMES2 and MEG2 was observed, accounting for 80% of the variability in
AMES2. MEG1 was found to explain a significant portion of the variation in both
RAM and RATP. Finally, RVI exhibited a strong association with MEG23. The
following comments pertain to these results:
(1) Except for the AMES analysis, they agree with the rank
correlation analysis results;
(2) In each case only one MEG category from the group that
was considered was found to be significant in the regression.
After adjusting for this variable in the regression analysis,
none of the others were found to explain a significant amount
of the remaining variability. The MEG categories are themselves
correlated; therefore, adjusting for the effects of a particular
MEG variable will likely rule out the residual effects of the
others.
(3) Performing linear regressions on essentially categorical
variables using skewed variables as independent variables in the
regression is at best a tricky proposition, and results should
be interpreted with caution.
B-14
-------
Discussion
In a general sense these data do not reveal a large degree of association
between the chemical variables and the biological tests. Approximately 2000
correlations were computed and 20 regression analyses performed, and the number
of statistically significant results observed did not exceed the number that
would be expected if the data were truly random in distribution. As summarized
by Table 13, however, the constituents listed there seem to be related to the
biological response of the sample effluents. These results are not conclusive
but provide evidence for possible associations.
An appropriate questions arises: Why are more associations not found in
these data? A number of factors can be identified which may have prevented the
discovery of additional relationships.
1) The quality of the data. For the purposes of these analyses the
quality of the data was very poor. Missing data reduced the sample -sizes avail-
able for many of the variables and prevented analyses of others. The multivariable
analyses such as regression were severely restricted by this problem; hence, the
question of interaction or synergy remains unanswered by this report.
Also, the data were generally skewed upward and clustered about zero and the
MDL. Since few variables were distributed evenly across the range of values,
standard statistical procedures involving the assumption of normality are not
useful in this setting. One should also ask: do the data span the range of
toxicity? Apparently the body of the chemical data does not fall in the range
of response for most of the biological tests, especially the textile data. Few
high correlations are observed possibly for this reason alone.
B-15
-------
2) Sample size. The sample sizes were generally too small to elicit a
significant result even if an association were present. No general rule exists
concerning minimum sample size requirements; however, pairwise comparisons
involving fewer than 15 observations will not be significant at the 0.05 level
unless the absolute value of the rank correlation coefficient is greater than
0.455 (10). The fact that the analyses were performed by study-type rather
than over the entire body of data further reduced the sample sizes.
3) The nature of the sample medium. The nature of the sample medium pre-
vented a realistic assessment of the association between any of its particular
constituents and a biological test. The test result probably represented a
response to the entire medium, all of its constituents, not to an isolated
constituent. The effects of one constituent should only be assessed after
controlling for the effects of certain others in the analysis. Therefore,
there arises a real need to focus on a collection of variables rather than
individual variables. This presents the dilemma of multivariable methods,
such as regression analysis, which require complete data. Furthermore, when
evaluating observational data which do not arise from a designed study or an
experiment, the results may not be valid due to confounding of the effects,
that is, an artificial relationship between a biological response and a
chemical parameter resulting purely from relationships to a third variable.
Regardless, such methods were only used in a limited fashion in this analysis.
4) There are no relationships here. At least at the level of the chemical
constituents found in these samples, there may be no relationships to be measured.
Assuredly, there are associations between some of these tests and the chemical
parameters; however, the chemical data were simply not in the range of toxicity
of the biological tests to elicit enough gradation of response to measure an
association.
B-16
-------
Suggestions for future work would include the following:
(a) collection of additional, more complete data;
(b) collection of better quality data; and
(c) laboratory verification of the associations.
If the goal of a future study were to determine the relationships between
the chemical makeup of and biological responses to an effluent, then additional,
more complete data is needed. Data should be collected at fewer sites with more
replications per site. In the setting of the present study, it is difficult to
identify any semblance of a random sample. The sampling units are plants, all
of which are unique. Each datum or sample unit arose from a unique source, and
thus random variation has no meaning unless it can be assumed that all of the
data sources are in some sense homogenous.
«
Even if that assumption can be made, the question arises: Are other sources
of variation (which are measurable), such as day-to-day and within-plant variation
controlled for? If these sources of variation cannot be controlled for in the
sampling, they can be controlled for in the analysis, provided pertinent data
are collected.
More complete data should be collected so that missing entries can be elim-
inated. Efforts should be made to ensure that all of the parameters are measured
from each sampling unit. In addition, there are process variables such as tempera-
ture or humidity which ought to be collected but were not. Perhaps a protocol
which would spell out clear instructions for data collection would be useful
for these purposes.
Finally, samples need to be collected where there will be a range in the
chemical response and the biological response. This goal may operationally
contradict some of the above recommendations, especially if there is little
B-17
-------
within-plant variability. However, when the collection of samples yields no
biological response or no range of response, no knowledge is gained with regard
to chemical/biological associations. This leads to a final recommendation.
Even if the above measures are taken, there remains a need for controlled
laboratory investigation of these relationships, for two reasons:
(i) Too many uncontrolled factors remain to statistically describe
the association between the chemical factors and the biological effects.
From these data it is not possible to know the cause of a high biological
response. One chemical alone, two or more in combination, or two or more
interacting may all be the cause of such a response. But the answer
probably won't be ascertained without a carefully controlled experiment
in which all extraneous factors can be kept constant.
(ii) In a laboratory setting the chemical composition which is
subjected to biological testing can be simulated to induce a range of
biological response, such that the main effects of each constituent
and the corresponding interactions can be measured and statistically
analyzed. Expensive field sampling and chemical analyses can be
avoided, as well.
Summary
The purpose of this report is to describe the relationships between the
chemical analyses and biological tests associated with samples from four
pollution sources. A total of 191 variables were measured, 150 of which were
chemical variables. A severe limitation in this endeavor was the amount and
quality of the data available for statistical analysis, much of which was
missing or essentially contant in value.
B-18
-------
In general the results do not reveal a strong degree of association between
the chemical analyses and biological tests. A few of the chemical variables were
found to correlate highly with more than one biological test. Although not con-
clusive, these results provide evidence for possible associations to further
investigate.
Of primary interest is the identification and measurement of interactions
among the chemical constituents. These issues could not be studied in this
report. Since multivariable methods are involved in such analyses, the need
for higher quality, more complete data is apparent. The control of extraneous
factors and identification of probable sources of variation is also an important
effort. Ideally the experimental setting would produce the most useful data for
these purposes.
B-19
-------
References
1. Lentzen, D. E., Wagoner, D. E., Estes, E. D., and Gutknecht, tf. F.,
IERL-RTP Procedures Manual: Level 1 Environmental Assessment, Second
Edition, (Draft), September 1978.
2. Dorsey, J. A., Johnson, L. D., Statnick, R. M., and Lochmuller, C. H.,
Environmental Assessment Sampling and Analysis; Phased Approach and
Techniques for Level 1. EPA-600/2-77-115, June 1977.
3. Brusick, D., and Hart, R., Internal Communication, Litton Bionetics,
August 1978.
4. Allen, J. M., Howes, J. E., Jr., Miller, S. E., and Duke, K. M., Com-
prehensive Analysis of Emissions from Exxon Fluidized-Bed Combustion
Miniplant Unit (Draft), Battelle Columbus Laboratories, September 1977.
5. Unknown, Environmental Assessment; Source Test and Evaluation Report
for a Commercial Chapman Low-BTU Gasification Facility, (Draft), Radian
Corporation, Austin, Texas, EPA Contract 68-02-2147.
7. Mahar, H., Evaluation of Selected Methods for Chemical and Biological
Testing of Industrial Particulate Emissions, EPA-600/2-76-137, May 1976.
8. Ames, B., McCann, J., and Yamasaki, E., "Methods for Detecting Carcinogens
and Mutagens with the Salmonella/Mammalian-Microsome Mutagenicity Test",
Mutation Res., Vol. 31. 1975, pp 347-364.
9. Neter, J., and Wasserman, W., Applied Linear Statistical Models,
Richard D. Irwin, Inc., 1974.
10. Snedecor, G. W., and Cochran, W. G., Statistical Methods, Sixth Edition,
Iowa State Univ. Press. Ames, Iowa, 1967, p 557.
B-20
-------
TABLE 1
ADDITIONAL PHYSICAL AND CHEMICAL PARAMETERS
Parameter
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
Cl"
F"
so4=
N03~
N02"
s
s=
Cr"
S°3=
CN"
SCN"
N0x
°2
co2
so2
CO
H2S-COS
HCN
Description-Units
gms/100
gms/100
gms/100
gms/100
gms/100
gms/100
gms/100
gm/M3
gm/M3
gm/M3
gm/M3
ppm
%
%
ppm
ppm
ppm
ppm
gms of sample
gms of sample
gms of sample
gms of sample
gms of sample
gms of sample
gms of sample
Parameter
19. F2
20. C12
21. H2S
22. NH3
23. Hg-AA
24. Sb-AA
25. As-AA
26. Ph
27. Acidity gm/M
Description-Units
ppm
ppm
gm/M
gm/M3
ugm/gm of sample
ygm/gm of sample
ugm/gm of sample
1-14 scale
3
28. Alkalinity gm/M'
29. BOD — "'3
30. COD
31. DO
gm/M~
gm/M2
gm/M*
32. Conductivity ymhos @ 25 C
33. Dissolved Solids gm/M
34. Suspended Solids gm/M
35. Organics gm/M
B-21
-------
TABLE 2
BIOLOGICAL RESPONSE VARIABLES
Test
1. Ames test for
jnutagenicity*
2. Rabbit Alveolar
Macrophage
Viability Index*
3. Rabbit Alveolar
Macrophage ATP
production*
4. Combination of
RVI* and RATP
5. Rat Test*
6. Rat LD
SO
7. Salt Water Algae*
8. Fresh Water Algae
9. Human Lung
Fibroblast
10. RAM - probit
11. ATP - probit
12-14. Grass Shrimp
@ 24, 48, and 96 hrs.
15-17. Fresh Water Fish
e 24, 48, and 96 hrs.
Variable Name
AMES
RVI
RATP
RAM
RT
RLD50
SWA
FWA
WI38
PRAM
PATP
GS24, GS48,
GS96
FWF24, FWF48,
FWF96
Description
categorical index of
mutagenicity
# of cells that survive
as % of control
10" g/cell 8 dosage of
1000 yg per ml of culture
medium (expressed as %
of control)
RAM = 1/2 (RVI + RATP)
Sum of all atypical
behavioral and physiological
responses
gms/Kg of rat body weight
at which 50% lethality is
estimated
EC50 measured @ 96 hours
as a % of control
EC20 measured §14 days
as a % of control
mean of viability index
and ATP production for
HLF, as % of control
inverse of LD50 from
probit analysis
RAM system - inverse of
LD50 based upon probit
analysis
LC50 as a % of aqueous
effluent sample
LC50 as a % of aqueous
effluent sample
B-22
-------
Test
Variable Name
Description
18,19. Daphnia
and 48 hrs.
@ 24
20-22. Salt Water
Fish 8 24, 48,
and 96 hrs.
DAPH24, DAPH48"
SWF24, SWF48
SWF96
LC50 as a % of aqueous
effluent sample
LC50 as a % of aqueous
effluent sample
The following are versions of the above tests developed by D. Brusick. All are
categorical variables ranging from 1 to 4.
23. Ames test for
mutagenicity
(Brusick)*
24. RAM test (Brusick)*
25. Rat Test
26. Grass Shrimp*
27. Daphnia
28. WI38-2
29. Fresh Water Algae
30. Fresh Water Fish*
AMES 2
RAM2
RODENT
GS
DAPH
WI38-2
FWA2
FWF
based upon max. applicable
dose
corresponds to RAM
categorized version of RT
categorized GS96
categorized DAPH48
categorized WI38
categorized version of FWA
categorized FWF96
1. See Reference 3.
2. 1 a not significantly different from control
2 a up to 1/10 of max. applicable dose
3 = 1/10 to 1/100 of max. applicable dose
4 = less than 1/100 of max applicable dose
Tests which are starred by an asterisk are those which have enough data for
correlation analysis.
B-23
-------
TABLE 3
AVAILABLE DATA FOR ANALYSIS
BIOLOGICAL VARIABLES
Variable
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
*
**
1
2
3
AMES
RVI
RATP
RAM
RT
RLD50
SWA
FWA
WI38
PRAM
PATP
GS24
GS48.
GS96
FWF24
FWF48
FWF96
DAPH24
DAPH48
SWF24
SWF48
SWF96
AMES2
RAM2
WI38-2
RODENT
FWA2
GS
FWF
DAPH
- data are
all data
- all data
Radian Exxon
(n*4) (n=8)
4
4
4
4
3
r
31
1
1
0
4*
3*
1
1
1
1
1
1
1
1
1
1
1
4
4
1
3**
0
0
0
0
predominately O's
are 1's
are 10 's
6
8
8
8
0
i
81
1
1
0
4*
4*
0
0
0
4
3
3
4
4
0
0
0
6
8
0
6**
4
0
4
4
Textiles
(n=23)
14**
4
4
4
22
1
22 x
222
13
4
21*
21*
143
143
143
0
0
0
0
22
143
14
14
23**
4
4
23**
0
13
13
0
Mitre
(n-16)
0
16
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
- 13 are 100' s
- more than
50% are 100 's
B-24
-------
TABLE 4
DESCRIPTIVE STATISTICS: ELEMENTS
(Non-textiles, n=12)
Element
U
Th
Bi
Pb
Tl
Hg
Al
Pt
Ir
Os
Re
W
Ta
Hf
Lu
Yb
Tm
Er
Ho
Dy
Tb
Gd
Eu
Sm
Nd
Pr
Ce
La
Ba
Cs
I
N
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
X
409.3
377.9
63.0
1861.6
209.3
291.7
291.7
291.7
125.1
291.7
125.1
542.8
4792.1
250.7
62.6
176.1
62.6
250.7
62.7
250.8
62.7
250.8
250.7
376.9
177.1
65.3
158.4
6687.7
50557
2256.2
242.2
S
341.6
888.7
148.4
4393.9
560.8
721.1
721.1
721.1
296.8
721.1
296.8
1412.8
13957.
592.7
148.4
560
148.4
592.7
148.4
592.7
148.4
592.7
592.7
888.7
362.2
146.8
283.2
19535
139617
5699.7
570.9
MIN
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0.5
0.4
3
0
0
MAX
3000
3000
500
1500
2000
2500
2500
2500
1000
1000
1000
5000
50000
2000
500
2000
500
2000
500
2000
500
2000
2000
3000
1000
500
1000
70000
500000
20000
2000
c.v.
0.83
2.35
2.35
2.36
2.68
2.47
2.47
2.47
2.37
2.47
2.37
2.60
2.93
2.36
2.37
3.18
2.37
2.36
2.37
2.36
2,37
2.36
2.36
2.36
2.04
2,25
1.79
2.92
2.76
2.53
2.36
B-25
-------
Element
Te
Sb
Sn
In
Cd
Ag
Pd
Rh
Ru
Mo
Nb
Zr
Y
Sr
Kb
Br
Se
As
Ge
Ga
Zn
Cu
Ni
Co
Fe
Mn
Cr
V
Ti
Sc
Ca
K
Cl
S
P
N
11
12
12
11
11
12
12
11
12
12
12
12
12
12
12
11
6
12
11
12
12
12
12
0
12
12
12
12
12
12
12
12
8
12
12
X
250.7
79.6
8611.5
125.1
126.5
83.7
125.3
125.3
750.1
8335. A
42.1
554.6
48.0
83786
39260
7030. 7
0.2
13828
252.8
439.3
4210.9
6867
27586
-
203183
1198.5
6723.2
2124.9
39134
41660803
54214875
869417
203.9
12525767
431669
S.
592.7
106.0
28095
296.8
296.8
184.3
296.8
296.8
2097.2
20336.4
87.9
1078.7
84.6
203199
89118
20511. 7
0
41906
591.9
935.8
14081
15365
83780
—
549305
2785
15338
5593
88077
141018954
141294420
2805617
464
42293741
1404728
B-26
MIN
0
0.125
0
0
0
0
0
0
0
0
0
0
0.15
19
0.5
0
0
0.1
0
0
0
2.25
0
—
0
0
0
0
0
0.15
0
300
4
250
0
MAX
2000
350
100000
1000
1000
500
1000
1000
7500
70000
300
3000
300
700000
300000
70000
1
150000
2000
3000
50000
50000
300000
_
2xl06
10000
50000
20000
300000
5xl08
5xl08
107
700000
l.SxlO8
5xl06
c.v.
2.36
1.33
3.26
2.37
2.35
2.20
2.37
2.37
2.80
2.44
2.09
1.95
1.76
2.43
2.27
2.92
0
3.03
2.34
2.13
3.34
2.24
3.04
_
2.70
2.32
2.28
2.63
2.25
3.38
2.61
3.22
2.27
3.38
3.25
-------
Element
Si
Al
Mg
Na
F
B
Be
Li
N
MIN
MAX
C.V.
12
11
12
12
8
12
12
11
125875
173396
1296423
167417
12756
17449
84170
19151
192961
423003
4208895
563714
35254
56442
282048
60251
0
0
50
0
0.2
0.02
0
0
70000
1.5xl06
l.SxlO7
2xl06
100000
200000
106
200000
1.53
2.44
3.25
3.37
2.76
3.23
3.35
3.15
B-27
-------
TABLE 5
DESCRIPTIVE STATISTICS: ELEMENTS
(Textiles, n*23)
Element
U
Th
Bi
Pb
Tl
Hg
Al*
Pt*
Ir*
Os*
Re
W
Ta
Hf
Lu
Yb
Tm
Ev
Ho
Dy
Tb
Gd
Eu
Sm
Nd
Pv
Ce
La
Ba
Cs
I
Te
N
15
15
15
15
15
0
15
15
15
15
0
15
7
15
IS
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
X ,
(xlO )
1.17
1.1
2.8
160.
1.37
-
0.5
0.5
0.5
0.5
-
2.63
2.0
3.13
0.67
2.13
0.8
1.53
0.57
1.0
0.57
0.93
0.87
1.30
1.70
2.97
4.4
6.87
341.3
1.63
155.5
0.57
CxlO'3)
0.67
0.68
2.9
237.4
1.34
-
0
0
0
0
-
1.80
2.57
2.39
0.31
1.53
0.37
0.97
0.18
1.19
0.18
0.53
0.48
1.00
1.63
3.54
5.57
10.32
375.2
1.72
593.3
0.18
MIN.
CxlO"5)
0.5
0.5
0.5
6
0.5
-
0.5
0.5
0.5
0.5
-
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
1.0
O.S
0.5
1.5
0.5
MAX-
Cxio"-5)
2.5
2.5
12
950
5
-
0.5
0.5
0.5
0.5
-
6.5
6.0
8.5
1.5
5.5
1.5
3.5
1.0
5.0
1.0
2.0
2.0
3.5
7.0
13.0
22
40.5
1300
7.0
2300
1.0
c.v.
0.57
0.61
1.03
1.48
0.97
-
0
0
0
0
„
0.68
1.28
0.76
0.46
0.71
0.46
0.63
0.31
1.19
0.31
0.56
0.55
0.76
0.9S
1.19
1.26
i.so
1.09
1.05
3.81
0.31
B-28
-------
Element
Sb
Sn
In
Cd
Ag
Pd*
Rh
Ru
MO
Nb
Zr
Y
Sv
Rb
Br
Se
As
Ber
Gia
Zn
Cu
Ni
Cob
Fe
Mn
Cr
V
Ti
Sc
Ca
K
Cl
S
P
Si
Al
Mg
N
15
15
0
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
X-3
(xlO •>)
378.5
24.13
-
6.03
667.8
0.5
0.5
0.5
11.33
1.17
13.97
2.90
452
315
2418
26.4
92.1
2.67
1.1
41437
0.383
0,160
0.700
7.37
1.99
3.29
0.494
0.122
2.40
90.28
67.51
68.81
205.4
21.77
23.31
11.86
41.10
(xlO'3)
547.5
29.86
-
7.15
2582
-
-
-
11.56
1.42
14.40
4.40
564
517
4377
69.3
131.4
2.87
1.2
149067
0.512
0.274
0.117
20.17
6.92
11.72
1.254
0.173
7.19
144.4
165.4
169.1
354.6
28.50
29.82
27.69
124.6
B-29
MIN
(xlO'3)
6.5
3.0
-
0.5
0.5
0.5
0.5
0.5
3.5
0.5
0.5
0.5
27
17
36
0.5
1.5
0.5
0.5
60
0.04
0.005
SxlO"4
0.12
0.008
0.005
0.002
0.01
5xlO"4
5.2
0.83
0.51
0.70
0.12
1.9
0.24
0.18
MAX
(xlO'3)
1900
100
-
30
10000
0.5
0.5
0.5
43
5.0
54
17
2100
2000
13000
270
400
10
5.0
580000
2
1
0.46
80
27
44
4.4
0.71
27
570
660
650
1400
110
120
110
490
C.V.
1.44
1.23
-
1.18
3.86
0
0
0
1.02
1.21
1.03
1.51
1.24
1.64
1.81
2.62
1.42
1.07
1.09
3.59
1.33
1.71
0.16
2.76
3.47
3.56
2.53
1.41
2.99
1.50
2.45
2.45
1.72
1.30
1.27
2.33
3.03
-------
Element
Na
F
B
Be
Li
N
15
14
1
15
14
X -
CxlO""*)
96.48
13.67
.
0.001
0.204
(xlO~3)
96.28
23.58
-
0.001
0.552
MIN,
2.9
0.95
-
5xlO"4
0.005
MAX
CxlO'3)
370
87
-
0.004
2.1
c.v.
0.99
1.72
_
1.00
2.74
*all values = 1/2 MDL
B-30
-------
TABLE 6
DESCRIPTIVE STATISTICS: PHYSICAL AND CHEMICAL PARAMETERS
(Textiles, n=15)
Chemical
Cl"
F"
so4a
NO/
N02~
co3=
s=
LCI
LC2
LC3
LC4
LC5
LC6
LC7
LC8
GC7
GC8
GC9
GC10
GC11
GC12
GC13
Hg-AA
Sb-AA
As-AA
N
0
0
IS
IS
15
0
0
11
11
11
11
11
11
11
11
0
0
0
0
0
0
0
15
14
15
X S MIN MAX C.V.
127.3 200.4 0 646 1.57
0.001 0.002 0 0.008 2.00
M) M) 0 0.0005
3.290 5.59 0.005 19.7 1.69
0.724 0.799 0.03 2.7 1.10
0.929 1.071 0.005 3.67 1.15
0.649 0.499 0.005 1.47 0.76
0.556 0.577 0.005 1.77 1.03
2.478 2.137 0.390 7,30 0.86
2.748 3.376 0.005 11.7 1.22
2.458 1.995 1.17 7.93 0.81
4xlO"4 2.5xlO"4 2.5xlO"4 9xlO"4 0.62
0.009 0.020 2.22
B-31
-------
Chemical
NO
X
°2
co2
so2
CO
H2S-COS
HCN
F2
ci2
Ph
ACIDITY
ALKALINITY
BOD
COD
DO
COND
DISSOLVED
SOLIDS
SUSPENDED
SOLIDS
cv"
so3=
H2S
ORGANICS
NH3
CN"
SCN"
N
0
0
0
0
0
0
0
0
0
15
15
15
14
15
15
15
14
14
15
0
15
14
15
0
0
X S MIN MAX C.V.
7.473 0.856 5.8 10.0 0.11
1.333 5.164 0 20 3.87
170.8 283.2 0 950 1.6S
42.96 46.27 2.5 168 1.07
471.6 433.7 78 1652 0.91
6.88 1.70 4 9 0.24
910 643.6 155 2400 0.70
2310.9 3240.8 276 13120 1.52
77.23 110.02 1.3 349 1.42
0.141 0.461 0 1.8 3.26
2.981 5.122 0.01 20 l.7i
17.03 15.70 2.73 63.70 0.92
13.80 24.27 0.05 72.5 1.75
B-32
-------
Chemical
As-AA
N0x
°2
co2
so2
CO
H2S-COS
HCN
F2
ci2
Ph
ACIDITY
ALKALINITY
BOV
COD
DO
COND
DISSOLVED
SOLIDS
SUSPENDED
SOLIDS
cv"
S03=
H2S
ORGAN I CS
NH.
CN"
SCN"
N X. S_ MIN MAX C.V.
6 52.83 19.43 33 85 0.36
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
7 428611 731898 20 l.SxlO6 1.70
0
0
0
0
0
B-33
-------
TABLE 7
DESCRIPTIVE STATISTICS: PHYSICAL AND CHEMICAL PARAMETERS
(Non-textiles, n=12)
Chemical
Cl"
F"
*>/
N03~
NO.,"
co3-
s~
LCI
LC2
LC3
LC4
LC5
LC6
LC7
LC8
GC7
GC8
GC9
GC10
GC11
GC12
GC13
Hg-AA
Sb-AA
N
11
12
8
8
8
7
6
5
5
5
5
S
5
5
5
5
5
5
5
5
5
5
9
6
x:
10636.5
208.3
50007
37.50
6225
10.70
0.195
180.9
125.6
138.5
28.60
13.16
64.1
108.5
112.4
217.7
282.1
95.9
26.6
43.2
33.5
79.7
0.604
5.433
S
23791.9
606.7
92577
69.44
17286
23.45
0.429
331.5
209.2
190.0
38.17
19.2.2
143.3
242.6
251.3
283.6
447.8
124.2
35.1
58.0
44.7
135.2
1.076
3.738
MIN
0
0.006
'X/Q
*\*o
'"v/O
0.1
0.003
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
2.0
MAX
64000
2100
200000
150
49000
63.4
1.065
767
494
461
93
42.5
320.5
542.5
562
659
1038
266
65
118
88
313.5
2.5
10
C.U.
2.23
2.91
1.85
1.85
2.77
2.19
2.20
1.84
1.66
1.41
1.33
1.46
2.23
2.23
2.23
1.30
1.58
1.29
1.31
1.34
1.33
1.69
1.78
0.68
B-34
-------
TABLE 8
DESCRIPTIVE STATISTICS: BIOLOGICAL TESTS
Continuous Variables
Test
RVI
RVI
RATP
RAM
RT
SWA
DAPH48
Study
non-textiles
Mitre
non-textiles
non-textiles
textiles
textiles
textiles
N
12
16
12
12
22
13
22
X
67.19
58.91
58.51
62.87
3.45
71.46
57.70
S_
33.34
24.29
34.19
32.93
3.56
34.87
41.23
MIN
0
15.8
0
0
0
1.7
1
MAX
102
111.4
109.4
98.7
10
100
100
C.U.
0.50
0.41
0.58
0.52
1.03
0.49
0.71
Note: Variables cited here are only those which have sufficient data for
correlation analysis.
B-35
-------
TABLE 9
DESCRIPTIVE STATISTICS: BIOLOGICAL TESTS
Discrete Variables
Variable
AMES
AMES2
RAM2
GS
FWF
Study-Type
Textiles
Non-Textiles
Textiles
Non-Textiles
Textiles
Non-Textiles
Textiles
Non-Textiles
Textiles
Non-Textiles
Response
!_
14
5
23
7
1
0
8
0
7
2
2_
0
0
0
2
3
2
0
0
4
1
3_
0
0
0
0
0
1
5
0
2
1
4
0
3
0
1
0
0
0
0
0
0
Tota:
""•^•••••i*
14
8
23
9
4
12
13
0
13
4
B-36
-------
TABLE 10
STATISTICALLY SIGNIFICANT RANK CORRELATIONS
(Non-textile data)
Test
AMES
AMES 2
RAM
RAM2
RATP
RVI2
Chem.
Farm.
Bi
Pb
Lu
MEG2
MEG 15
Bi
Pb
Cl
F
MEG2
MEG15
I
COD
MEG1
MEG21
MEG25
I
COD
MEG1
MEG21
MEG25
I
COD
MEG1
MEG21
MEG25
I
Br
Se
Ca
MEG1
MEG21
MEG23
MEG25
NO ~
N
10
10
10
9
9
10
10
6
6
9
9
12
12
9
9
9
12
12
9
9
9
12
12
9
9
9
28
27
28
28
9
9
9
9
8
r
0.808
0.688
0.631
0.769
0.691
0.819
0.644
-0.828
0.840
0.791
0.707
-0.733
0.627
-0.746
-0.732
-0.758
0.661
-0.751
0.821
0.783
0.824
-0.797
0.641
-0.746
-0.732
-0.758
-0.438
-0.429
-0.461
0.410
-0.797
-0.800
-0.673
-0.810
0.708
p-value
<.01
.03
.05
.02
.04
<.01
.04
.04
.04
.01
.03
<.01
.03
.02
.03
.02
.02
<.01
<.01
.01
<.01
<.01
.03
.02
.03
.02
.,02
.'03
.01
.03
.01
.01
.05
.01
.05
two-sided p-value
includes Mitre data
B-37
-------
TABLE 11
STATISTICALLY SIGNIFICANT RANK CORRELATIONS
(Textile data)
Test
GS
FWF
RT
*
SWA
DAPH48
Chem.
Farm.
Dy
Eu
Pv
Ag
Y
Se
Ni
U
Pb
LC6
Dy
Ag
Y
Ni
U
Pb
Sm
Zr
Sr
Rb
Cl
Ce
N03V
Pb
Ta
Yb
Ev
Zr
Fe
MEG1
LC4
LC5
Sm
Zr
V
Ce
N03~
LC2
Ce
La
Zr
Mn
Ca
N03-
N
13
13
13
13
13
13
13
13
13
10
13
13
13
13
13
13
13
13
13
13
13
13
13
14
6
14
14
14
14
10
10
10
13
13
13
13
13
10
14
14
14
14
14
14
r
0.568
0.588
0.560
0.654
0.583
0.574
0.614
0.785
0.571
-0.640
0.613
0.632
0.613
0.632
0.579
0.693
0.563
0.667
0.596
0.617
-0.573
0.554
0.648
0.555
0.920
0.535
0.539
0.582
0.593
-0.812
-0.680
-0.767
-0.552
-0,742
-0.554
-0.687
-0.550
0.632
-0.647
-0.596
-0.557
-0.553
-0.771
-0.589
p-value
.04
.04
.05
.02
.04
.04
.03
<.01
.04
.05
.03
.02
.03
.02
.04
<.01
.05
.01
.03
.03
.04
.05
.02
.04
<.01
.05
.05
.03
.03
<.01
.03
,01
.05
<.01
.05
<.01
.05
.05
.01
.03
.04
.04
<.01
.03
1
two-sided p-value
B-38
-------
TABLE 12
STATISTICALLY SIGNIFICANT RANK CORRELATIONS
STRAIN REVERTANT RATIOS
(Textile and Non-textile data)
Strain
TA153S-
TA1535+
TA1537-
TA1537+
Chem.
Farm.
Hg
Re
Ta
Ce
La
Ba
Sb
In
Zr
Y
Sr
Cu
Co
Ti
K
S
Li
LC8
F
Cl"
Cr"
MEG3
MEGS
MEG6
MEG13
MEG 14
Cd
Pd
F
U
Fe
V
Zr
Ti
P
H2S
Sb-AA
LC7
MEG1
MEGS
MEG18
MEG21
N
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
16
24
9
14
12
12
10
9
9
24
21
24
24
24
24
24
24
24
14
17
16
20
20
20
9
r
0.576
0.566
0.479
-0.657
-0.565
-0.442
-0.402
0.580
-0.422
-0.450
-0.416
-0.502
-0.437
-0.489
-0.470
-0.411
-0.409
0.662
0.407
-0.890
0.569
-0.694
-0.741
-0.754
-0.773
-0.674
0.409
-0.561
0.519
-0.393
-0.542
-0.437
-0.420
-0.434
-0.481
-0.518
-0.475
0.591
0.589
0.634
0.680
0.710
p-value
<.01
<.01
.03
<.01
<.01
.03
.05
<.01
.04
.03
.04
.01
.03
.02
.02
.05
.05
<.01
.05
<.01
.03
.01
<.01
.01
.02
.05
.05
<.01
<.01
.05
<.01
,03
.04
.03
' .02
.05
.05
.02
<.01
<.01
<.01
.03
B-39
-------
Chem.
Strain Farm.
TA78- Ga
LCI
LC3
MEG15
TA98+ Bi
Cd
Ag
Nb
Br
Se
N03~
LCI
MEG1
MEG15
MEG21
MEG25
TA100- U
Th
Bi
Pb
Tl
Ah
Pt
Os
Lu
Yb
Tm
Er
Ho
Dy
Tb
Gd
Eu
Nd
Pv
Ce
La
Te
Cd
Rh
Ru
Mo
Nb
Y
Rb
Se
Ge
Ga
N
24
16
16
17
24
24
24
24
24
24
20
16
20
17
9
9
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
r
0.404
0.600
0.693
0.670
0.444
0.499
0.466
0.513
0.512
0.431
0.553
0.524
0.513
0.518
0.786
0.743
0.453
0.457
0.767
0.484
0.543
0.478
0.479
0.479
0.518
0.423
0.531
0.461
0.480
0.473
0.535
0.433
0.401
0.504
0.553
0.403
0.448
0.474
0.469
0.474
0.481
0.542"
0.501
0.401
0.400
0.422
0.434
0.535
p-value
.05
.01
<.01
<.01
.03
.01
.02
.01
.01
.04
.01
.04
.02
.03
.01
.02
.04
.03
<.01
.02
<.01
.02
.02
.02
.01
.04
< .01
.02
.02
.02
<.01
.04
.05
.01
<.01
.05
.03
.02
.02
.02
.02
<.01
.01
.05
.05
.04
.03
< .01
B-40
-------
Strain
TA100-
(cont.)
TA100+
TA1538-
Chem.
Parm.
Ir
V
K
Al
Be
Zn
Hg-M
Sb-AA
As-AA
LC7
MEG 2
MEG 15
Do
Ag
B
Ca
Ho
Si
Sn
LC6
LC7
BOD
Hf
Si
C03"
N
24
24
24
24
24
24
21
17
18
16
9
17
14
24
24
24
24
24
24
16
16
13
9
9
7
r
0.478
0.474
0.412
0.479
0.443
0.452
0.548
0.771
0.554
-0.650
0.728
0.590
0.628
-0.420
-0.424
-0.403
-0.443
-0.402
-0.410
0.506
0.527
0.642
0.751
0.622
-0.778
p-value
.02
.02
.05
.02
.03
.03
.01
<.01
.02
<.01
.03
.01
.02
.04
.04
.05
.03
.05
.05
.05
.04
.02
.02
.05
• .04
TA1538+
Cl
-0.667
.05
1
two-sided p-value
B-41
-------
TABLE 13
CHEMICAL PARAMETERS ASSOCIATED WITH
MULTIPLE BIOLOGICAL TESTS
Parameter
Biological Tests
Ag
Bi
Ca
Ce
Cl
Co
Dy
I
Pb
Se
Y
Zr
GS, FWF
AMES, AMES2
DAPH48, RVI
SWA, FWF, DAPH48
AMES2, FWF
RAM, RAM2, RATP
GS, FWF
RAM, RAM2, RATP, RVI
AMES, AMES2, RT, GS, FWF
GS, RVI
GS, FWF
DAPH48, RT, SWA
MEG1
MEG 15
MEG21
MEG25
DAPH48, FWF, SWA
RAM, RAM2, RATP, RVI, RT
AMES, AMES2
RAM, RAM2, RATP
RAM, RAM2, RATP, RVI
B-42
-------
TABLE 14
RANK CORRELATIONS AMONG STRAINS
(Textile and Non-textile data)
oo
i
CO
TA1535-
TA1535+
TA1537-
TA1537+
TA98-
TA98+
TA100-
TA100+
TA1538-
TA1535+ T1537-
0.324 -0.291
0.110
TA1537+
0.180
0.062
0.318
TA98- TA98+
0.093 -0.074
0.615** 0.066
-0.044 0.363
0.010 0.307
0.150
TA100-
-0.007
0.225
-0.180
-0.106
0.431*
0.434*
TA100+
0.050
0.375
-0.106
0.214
0.355
0.211
0.028
TA1538-
0.293
0.223
0.056
0.217
0.630*
0.625*
0.421
0.284
TA1538-t-
-0.100
0.733*
0.474
0.231
0.554
0.646*
0.492
0.225
0.523
* two-sided p-value <.05
** two-sided p-value -.01
-------
TABLE 15
RANK CORRELATIONS AMONG BIOLOGICAL TEST
Textile Data
FWF
GS 0.866**
FWF
RT
SWA
Non-Textile Data
AMES
AMES 0.902**
AMES2
RAM
RAM2
RVI
RT SWA DAPH48
-0.051 -0.417 -0.498
0.052 -0.702** -0.680*
-0.1S6 0.209
0.398
RAM RAM2 RVI RATP
-0.596 0.412 -0.466 -0.596
-0.518 0.458 -0.450 -0.517
-0.760** 0.909** 0.986**
-0.760** -0.760**
0.853**
* two-sided p-value <.05
** two-sided p-value <.01
B-44
-------
APPENDIX C
"GOOD" CORRELATIONS FOUND IN THIS DATA SET
Matrices are presented showing all correlations among data pairs where
the correlation coefficient is greater than 0.50 and the significance factor
is less than 0.05 (95 percent level of significance). Such "good" correla-
tions are presented for each study individually, all studies considered
together, all liquid samples, and then all solid samples.
C-l
-------
TABLE C-1. CHEMICAL AND HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS 2:0.50 AND SIGNIFICANCE LEVELS -0.05
STUDY 6, PEARSON CORRELATIONS (CORRELATION,NUMBER OF OBSERVATIONS)
Chemical
tests*
Ba*
Cd*
Br*
Cu*
Fe*
Cr*
V*
Ca*
S*
p.
Si*
NIEG1*
MEGS*
MEG18*
MEG21*
MEG23*
MEG25*
RAM1000
-.99 (3)
.99 (4)
.99 (4)
-.97 (4)
-.99 (4)
-.99 (4)
-.99 (4)
-.99 (4)
-.99 (4)
VIABIND
-.99 (3)
.98 (4)
.99 (4)
-.97 (4)
-.99 (4)
-.99 (4)
-.98 (4)
-.99 (4)
-.99 (4)
Biological tests, health-related
ATP PROBVI AMES1 QTHRAT AMES2 RAM
-.95 (4) -.95 (4)
-.99 (3)
-.99 (3)
-.97 (4) -.97 (4
.99 (4)
-.98 (4) -.98 (4
.99 (4)
-.96 (4) -.96 (4)
.98 (4) .98 (4]
-.95 (4) -.95 (4)
-.99 (4) -.99 (4)
-.97 (4)
-.99 (4) .96 (4) .96 (4!
-.98 (4) .97 (4) .97 (4;
-.99 (4)
-.99 (4)
-.99 (4)
"The transformation x = log (c) + 1, where c = chemical concentration, was made for each chemical test.
TABLE C 2. HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS SHOWING CORRELATIONS :>0.50
AND SIGNIFICANCE LEVELS <0.05
STUDY 6, PEARSON CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Biological test,
heilth-relited
VIABIND
ATP
PROBATP
TA1537M
TA1537P
TA98P
TA100P
TA1538P
RAM
R AMI 000 ATP
.99 (4) .99 (4)
.99 (4)
-.99 (3) -.99 (3)
Biological test.
PROBVI PROBATP
-.99 (3)
.98 (4)
.99 (3)
.99 (4)
health-related
AMES1 RATWT
.99 (4)
.99 (4) .99 (3)
.99 (4)
.99 (4)
AMES2 RAM
.98 (4)
.99 (4)
.99 (4)
.99 (4)
.99 (4)
C-2
-------
TABLE C-3. CHEMICAL AND HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS ^0.50 AND SIGNIFICANCE LEVELS <0.05
STUDY 10, PEARSON CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Chemical
tests RAM 1000
U*
Th*
Bi*
W*
Hf*
Lu*
Yb*
Er*
Ho*
Cs*
Ag»
As*
Zn*
Fe*
Cr*
K*
S*
Na*
Be*
MEG1*
MEG2*
MEG16*
MEG24*
MEG25*
FM
C03E .83 (6)
LC1
LC2
LC3
GC8
Sb-AA
Al-AA
Biological tests, health-related
ATP PROBVI AMES1
.86 (6)
.84 (6)
.87 (6)
-.98(4)
.84 (6)
.85 (6)
.97 (6)
.91(6)
-.98 (4)
-.97 (4)
-.96 (4)
-.99 (4)
-.95 (4)
.92 (6)
.97 (4)
-.99 (4) .92 (6)
.83 (6)
.96 (4)
.99 (4)
AMES2
.85 (6)
.82 (6)
.90 (6)
.81 (6)
.81 (6)
.86 (6)
.84 (6)
.99 (6)
.96 (6)
.86 (6)
.95 (6)
RAM
.95 (6)
.95 (5)
.99 (5)
.95 (5)
.97 (5)
.99 (5)
.98 (5)
.92 (5)
.94 (5)
'The transformation x = log (c) + 1, where c - chemical concentration, was made for each chemical test.
C-3
-------
TABLE C-4. HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS SHOWING CORRELATIONS
iJB.50 AND SIGNIFICANCE LEVELS
-------
TABLE C-5. CHEMICAL AND HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS iO.50 AND SIGNIFICANCE LEVELS <0.05
STUDY 10, SPEARMAN CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Chemical
tests
Th*
Bi*
Pb*
Tl*
W*
HP
Lu»
Er*
Ho*
Dy*
Tb»
Gd*
Eu*
Pr»
Ba*
Cs*
Ag*
Mo*
Y*
Rb*
As*
Ge*
Ga*
Zn*
Ni*
Mn*
Cr*
Sc*
Cl*
Si*
Al*
Na*
F*
Be*
Li*
MEGS*
MEG7*
MEGS*
CIM
FM
N02M
LC5
Biological tests, health-related
RAM1000 VIABIND ATP AMES1
.85 (6)
.85(6)
.82 (6)
.82 (6)
.85 (6)
.84 (6)
.88 (6)
.84 (6)
.82 (6)
.82 (6)
.82 (6)
.85 (6)
.84 (6)
.85 (6)
.82 (6)
.82 (6)
.84 (6)
-.85 (6)
.82 (6)
.82 (6)
.82 (6)
.86 (6)
.90 (5) .90 (5)
.90 (5) .90 (5)
.95 (5) .95 (5)
.74 (8) -.85 (6)
.82 (6)
.71 (8)
.89 (5) .89 (5)
AMES2
.83 (6)
.83 (6)
.83 (6)
.84 (6)
.84 (6)
.87 (6)
.85 (6)
.89 (6)
.85 (6)
.83 (6)
.84 (6)
.84 (6)
.84 (6)
.84 (6)
.83 (6)
.83 (6)
.85 (6)
.83 (6)
.83 (6)
.83 (6)
.83(6)
.84 (6)
+.83 (6)
.84 (6)
.83 (6)
.85 (6)
.84 (6)
.83 (6)
-.83(6)
.83 (6)
.83 (6)
.84 (6)
.84 (6)
.84 (6)
.84 (6)
-.83 (6)
.84 (6)
'The transformation x = log (c) + 1, where c = chemical concentration, was made for each chemical test.
C-5
-------
TABLE C-6. HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS SHOWING CORRELATIONS
<>0.50 AND SIGNIFICANCE LEVELS <0.05
STUDY 10, SPEARMAN CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Biological tests,
health-related
VIABIND
ATP
TA98M
TA98P
TA1538M
TA1538P
AMES2
RAM1000
.86 (8)
.98 (8)
-.93 (6)
Biological tests, health-related
VIABIND ATP AMES1
.76 (8)
.85 (6)
-.93 (6) .82 (6)
.98 (6)
AMES2
.84 (6)
.84 (6)
.83 (6)
TABLE C-7. CHEMICAL AND HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
STUDY 15, PEARSON CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Chemical
test
Biological test,
health-related
VIABIND
Chemical
test
Biological test,
health-related
VIABIND
Chemical
test
Biological test,
health-related
VIABIND
Mo*
-.70 (16)
-.68(16)
Be*
-.58 (15)
* The transformation x = log (c) + 1, where c = chemical concentration, was made for each chemical test.
TABLE C-8. CHEMICAL AND HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >O.SO AND SIGNIFICANCE LEVELS <0.05
STUDY 15, SPEARMAN CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Chemical
test
Mo*
Biological test,
health-related
VIABIND
-.78 (16)
Chemical
test
Mn*
Biological test,
health-related
VIABIND
.53 (16)
Chemical
test
V*
Biological test
health-related
VIABIND
-.65 (16)
Chemical
test
Be*
Biological test,
health-related
VIABIND
.69 (15)
*The transformation x = log (c) + 1, where c = chemical concentration, was made for each chemical test.
C-6
-------
TABLE C 9. CHEMICAL AND HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
STUDY 19, PEARSON CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Chemical
tests
U*
Th*
Pb*
Tl*
W*
Ta*
HP
Yb*
Tm*
Dy*
Tb*
Eu*
Sm*
Nd*
Cs*
Cd*
Ag*
Mo*
Nb*
Zr*
Sr*
Rb*
Br*
Ga*
Ni*
Fe*
V*
Ti*
Li*
MEG1
N03M
N02M
LC4
LC5
Sb-AA
Ph
BOO
DO
Conductivity
Dissolved
solids
PROBVI
.57 (13)
.63 (13)
58 (13)
.61 (13)
.60 (13)
56(13)
.62(13)
.82 (13)
.68 (13)
.59(13)
.81 (13)
.82(13)
.62 (13)
.67 (13)
.60 (13)
.75(13)
-.62(12)
.85 (13)
.80(13)
.85 (13)
.63 (13)
.66(12)
Biological tests, health-related
PROBATP OTHRAT RATWT CHO
.57(13)
.60(13) .58(14)
.99 ( 6)
.56(13)
.63(13)
.59 (13)
.65(13)
.61 (15)
.79 (14)
.56 (15)
.72(13) .62(14)
.64 (13)
.55 (15)
.62 (14)
.61 (14)
.54 (14)
.65(13)
-.74(10)
-.70(10)
-.69(10)
-.96(4)
.69 (13)
.52(15)
•The transformation x = log (c) + 1, where c = chemical concentration, was made for each chemical test.
C-7
-------
o
I
CO
TABLE CIO. CHEMICAL AND ECOLOGY-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS ;>0.50 AND SIGNIFICANCE LEVELS <0.05
STUDY 19, PEARSON CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Chemical
tests
U*
Th»
Pb*
W*
Hf*
Lu*
Yb*
Tm*
Er»
Ho*
Oy*
Gd«
Eu»
Sm*
Nd*
Ce*
Bt"
Cs»
Sn»
Cd*
Ag*
Zr*
Y»
Sr*
Rb»
Biological tests, ecology-related
FWALGAL SWALGAL FWFISH96 OAPH48 SWFISH24 SWFISH48
•
-.55 (14)
-.58(15) -.66(13)
-.57 (14)
-.59 (14)
-.60 (14)
-.59 (14)
-.88 (14) -.87 (14)
-.76(14) -.87(14)
-.64 (13)
-.55 (15)
-.65 (15) -.78 (13)
-.59 (13)
-.59 (15)
-.84 (13) -.58 (14)
-.68 (13) -.53 (15)
-.79 (14) -.77 (14)
SWFISH96
-.68 (14)
-.63 (14)
-.55(14)
-.58 (14)
-.75(14)
-.77 (14)
-.63(13)
-.61 (14)
-.66(14)
GSHP24
-.68 (14)
-.73(14)
-.58 (14)
-.76 (14)
-.78 (14)
-.73 (14)
-.79 (14)
-.77 (14)
-.73 (14)
-.61 (14)
-.69 (14)
-.81 (14)
-.64 (14)
-.59 (14)
-.63 (14)
-.58 (14)
-.55 (14)
-.66 (14)
-.54 (14)
GSHP48
-.65(14)
-.70 (14)
-.59 (14)
-.74 (14)
-.75 (14)
-.69 (14)
-.77 (14)
-.74 (14)
-.72(14)
-.63 (14)
-.67 (14)
-.81 (14)
-.60 (14)
-.59 (14)
-.58(14)
-.57(14)
-.64 (14)
-.56(14)
GSHP96
-.79 (14)
-.57 (14)
-.54(14)
-.59 (14)
-.61 (14)
-.60 (14)
-.64(14)
-.62 (14)
-.61 (14)
-.54 (14)
-.57 (14)
-.54 (14)
-.69 (14)
GRSSHP FWFISH
.78(13)
.68 (13)
.57 (13)
.56(13)
.72(13)
.63 (13) .68 (13)
.75(13)
.63(13)
.66 (13)
.63(13)
The transformation x = log (c) + 1, where c = chemical concentration, was made for each chemical test.
-------
TABLE C 10 (continued)
o
Chemical
tests
Ga*
Ni*
Fe*
Mr*
Cr*
V*
Ti*
Ca*
K*
Cl*
S*
Li*
MEG1*
S04E
N03M
N02M
As-AA
Ph
Acidity
Alkalinity
Conductivity
Dissolved
solids
Organic*
FWALGAL
-.53(15)
-.55(15)
-.61 (14)
-.71 (15)
-.65(11)
-.68 (15)
-.53 (15)
-.55(15)
-.54 (14)
-.55(15)
SWALGAL
-.61 (13)
-.59 (13)
-.80 (12)
-.57 (13)
-.62 (12)
-.69 (13)
-.60 (13)
-.60 (13)
Biological tests, ecology-related
FWFISH96 DAPH48 SWFISH24 SWFISH48 SWFISH96 GSHP24 GSHP48 GSHP96 GRSSHP FWFISH
-.61(14) -.59(14) -.56(14) -.58(14) -.63(14) -.55(14)
-.65(15) -.71(14) -.72(14) -.63(14) .59(13)
.59 (12)
-.72(14) -.69(14) -.59(14)
-.56 (14)
-.79(14) -.55(14) -.56(14) .59(13)
-.62 (14)
-.52(15)
-.55(14)
-.85 (14) -.77 (14) -.60 (14)
-.56(14)
-.59 (14)
-.66(14) -.64(14)
.56(13)
-.68 (14) -.54 (14)
'The transformation x = log (c) + 1, where c = chemical concentration, was made for each chemical test.
-------
TABLE C-11. HEALTH AND ECOLOGY-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
STUDY 19, PEARSON CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Biological tests,
ecology-related
Biological tests, health-related
W138-600
VIABIND
ATP
PROBVI
FWALGAL
SWALGAL
FWFSH96
SWFSH48
SWFSH96
GRSHMP24
GRSHMP48
GRSHMP96
GRSSHP
FWFSH
.97 (4)
.99 (4)
-.98 (4)
-.67(12)
-.72(12)
-.69(12)
-.60(12)
PROBATP
.67(12)
-.62(12)
-.59(12)
-.59(12)
-.62(12)
.62(12)
.75(12)
C-10
-------
o
I
TABLE C 12. ECOLOGY-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
STUDY 19, PEARSON CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Biological tests,
ecology-related
FWALGAL
FWFSH96
DAPH48
SWFSH48
SWFSH96
GRSHMP24
GRSHMP48
GRSHMP96
GRSSHP
FWFSH
Biological tests, ecology-
SWALG
.63(13)
.64(13)
-.71(12)
FWFSH96 DAPH48
.60(21)
.56(13)
.63(14)
.60(14)
.61(14) .56(13)
.70(14)
-.69(13)
-.73(13) -.66(12)
SWFSH24
.91 (14)
.84(14)
.66(14)
.71 (14)
.64(14)
-.62(13)
-.77(13)
SWFSH48
.89(14)
.89 (14)
.91 (14)
.82(14)
-.77 (13)
-.80 (13)
related
SWFSH96
.83(14)
.85(14)
.94(14)
-.94(13)
-.92(13)
GRSHMP24
.99(14)
.89 (14)
-.83(13)
-.74(13)
GRSHMP48 GRSHMP96
-.84(13) -.99(13)
-.76(13) -.83(13)
-------
TABLE C 13. HEALTH RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS J).50 AND SIGNIFICANCE LEVELS '0.05
STUDY 19, PEARSON CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Biological tests,
health-related
Biological tests, health-related
WI38-600
CHOMPS
ATP
PROBVI
PROBATP
WI38-600
TA98M
TA98P
TA100P
WI38
.78 (7)
.96 (4)
.98 (4)
.99 (4)
-.97 (4)
-.95 (4)
.56(21)
TABLE C-14. CHEMICAL AND HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS ;>0.50 AND SIGNIFICANCE LEVELS <0.05
STUDY 19, SPEARMAN CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Biological tests, health-related
Chemical
tests
Pb*
W*
Ta*
Hf*
Yb*
Tm*
Er*
Dy*
Tb*
Eu"
Sm*
Nd»
Cd*
Ag*
PROBVI
.68(13)
.74(13)
.63(13)
.58(13)
.55(13)
.63 (13)
PROBATP
.67 (13)
.60(13)
.60 (13)
.69 (13)
.58(13)
.72(13)
Chemical
OTHRAT tests
.55(14) Mb*
Zr*
.92 (6) Sr*
Rb*
.54 (14) Ni*
Fe*
.54(14) V*
MEG1*
N03M
LC4
LC5
Ph
Cr++
Biological tests, health-related
PROBVI PROBATP
.72(13)
.71 (13)
.62 (13)
.56(13) .72(13)
.60 (13)
.65 (13)
.58(13) .56(13)
.63(13)
.58(13)
OTHRAT
.58 (14)
.59 (14)
-.81 (10)
.68(10)
.77 (10)
'The transformation x = tog (c) + 1, where c = chemical concentration, was made for each chemical test.
-------
o
I—•
CO
TABLE C 15. CHEMICAL AND ECOLOGY-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
STUDY 19, SPEARMAN CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Chemical
tests
U*
Th*
Bi*
Pb*
W*
HP
Lu*
Yb*
Tm*
Er*
Dy«
Gd*
Eu*
Sm*
Nd*
Ce*
La'
Sn»
Cd*
Ag*
Zr*
Y*
Sr*
Rb*
As*
Biological tests, ecology-related
FWALGAL SWALGAL FWFISH96 DAPH48 SWFISH24 SWFISH48
-.58 (14)
-.63 (14)
-.66(15) -.59(15) -.67(14) -.67(14)
-.64 (14)
-.64 (14)
-.58 (14)
-.66 (14)
-.68 (14)
-.63 (14)
-.57 (14) -.69 (14)
-.61 (14)
-.65 (14) -.78 (14)
-.52 (15)
-.53 (15) -.69 (13) -.65 (14)
-.60 (14)
-.53 (15)
-.54 (14)
-.56 (14)
-.74(13) -.56(14) -.65(14) -.55(14)
-.64 (14) -.55 (14)
-.61 (14)
-.59 (14)
.52(15)
SWFISH96
-.71 (14)
-.61 (14)
-.55 (14)
-.64 (14)
-.65 (14)
-.62 (14)
-.67 (14)
-.58 (14)
GSHP24
-.63 (14)
-.68 (14)
-.63 (14)
-.68 (14)
-.68 (14)
-.65 (14)
-.70 (14)
-.72 (14)
-.66 (14)
-.72 (14)
-.66(14)
-.79 (14)
-.55 (14)
-.58(14)
-.61(14)
-.64 (14)
-.59 (14)
GSHP48
-.63(14)
-.68 (14)
-.54 (14)
-.64 (14)
-.69 (14)
-.69 (14)
-.70 (14)
-.71 (14)
-.72 (14)
-.68 (14)
-.66 (14)
-.68 (14)
-.76(14)
-.55 (14)
-.57 (14)
-.57 (14)
-.65 (14)
-.55 (14)
GSHP96
-.76 (14)
-.57 (14)
-.57 (14)
-.56(14)
-.56 (14)
-.59 (14)
-.61 (14)
-.62 (14)
-.59 (14)
-.65(14)
-.56(14)
-.69 (14)
-.68 (14)
-.55 (14)
-.57 (14)
GRSSHP FWFISH
.78(13) .58(13)
.57(13) .69(13)
.57(13) .61(13)
.59 (13)
.56(13)
.55(13)
.65(13) .63(13)
.67(13)
.58(13) .61(13)
.60(13)
.62(13)
*The transformation x = log (c) + 1, where c = chemical concentration, was made for each chemical test.
-------
TABLE C-15 (continued)
o
i
Chemical
tests FWALGAL SWALGAL
Ga'
Mi-
Mr"
Cr' -.55(14)
V* -.55(13)
Sc"
Ca-
d-
s' .61 (15)
Si'
Be'
Li'
MEGT .64(11)
MEGS'
N03M
N02M
LC6
AsAA
Conductivity
Organics
Biological tests, ecology-related
FWFISH96 DAPH48 SWFISH24 SWFISH48 SWFISH96
-.78(14) -.61(14) -.56(14)
-.59 (15)
-.55 (14)
-.53 (15)
-.60 (14)
-.77(14) -.56(14) -.55(14)
-.57 (14)
-.58(14)
-.54 (14)
.65 (10)
-.59 (14)
-.52 (15)
.63(10) .65(10)
-.62 (14) -.56 (14)
-.63 (14)
-.76 (14)
GSHP24
-.55 (14)
-.60 (14)
-.57 (14)
-.61 (14)
.65 (10)
.63(10)
-.66 (14)
GSHP48 GSHP96
-.59(14) -.58(14)
-.59(14)
-.54(14)
-.56 (14)
-.61 (14)
.68 (10)
.70 (10)
-.66(14) -.56(14)
GRSSHP FWFISH
.61(13) .63(13)
-.57(13)
.65(13)
-.64 (10)
"The transformation x = log (c) + 1, where c = chemical concentration, was made for each chemical test.
-------
TABLE C-16. HEALTH-AND ECOLOGY-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
STUDY 19, SPEARMAN CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Biological tests. Biological tests, health-related Biological tests, Biological tests, health-related
ecology- ecology-
re)ated PROBVI PROBATP I8|ated PROBVI PROBATP
SWFSH48 -.68(12) -.59(12) GRSHMP24 -.77(12) -.62(12)
SWFSH96 -.60(12) GRSHMP48 -.70(12) -.63(12)
GRSHMP96 -.73(12) -.62(12) GRSSHP .61(12)
FWFSH .81 (12)
C-15
-------
TABLE C-17. ECOLOGY-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
STUDY 19. SPEARMAN CORRELATIONS (CORRELATION. NUMBER OF OBSERVATIONS)
Biological tests.
FWFSH96
DAPH48
SWFSH24
SWFSH48
SWFSH96
GRSHMP24
GRSHMP48
GRSSHP
FWFSM
Biological tests, ecology-related
SWALGAL
.60(13)
.70(12)
FWFSH96
EC 21)
SC -14)
57 '14)
.50(14)
.63i14)
-71 '13)
- 30-13)
OAPH48
.69(13)
.62(13)
.56(13)
-.68 (12)
SWFSH24
.84(14)
.84 (14)
.74 (14)
.73 (14)
-.69 (13)
-.74 (13)
SWFSH48
.87 (14)
.98 (14)
.97 (14)
-.83 113)
-.77 (13)
SWFSH96 GRSHMP24 GRSHMP48
.83(14)
.82 (14) .99 (14)
-.96 (13) -.83 (13) -.83 (13)
-.90(13) -.75(13) -.73(13)
GRSHMP96 GRSSHP
-.96(13)
-.83(13) .87(13)
Biological tests,
health-related
TABLE C-18. HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
STUDY 19, SPEARMAN CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Biological tests, health-related
CHOMPS
PROBVI
PROBATP
TA1535W
TA98V
TA98P
.93 (7)
.90 (7)
.61(21)
-------
o
i
TABLE C-19. CHEMICAL AND HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
ALL STUDIES, PEARSON CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Chemical
tests RAM1000 VIABIND
Th*
Bi"
Pb*
Tl*
Hf*
Lu*
Yb*
Tm*
Er*
Ho*
Dy*
Tb*
Gd*
Eu*
Sm*
Nd*
Pr*
Ce*
Cs*
!*
Sb*
s-*
C::*
J:*
«.;-
*7 . *
. -
Biological tests, health-related
ATP PROBVI AMES1 OTHRAT RATWT AMES2 RAM(2) WI-38(2) CHO
.52 (24)
.81 (24) .98 (25)
.67(24) .71(17) .56(25)
.58 (24)
.62 (24)
.57 (24)
.61 (24)
.56 (24)
.60 (24)
.63 (24)
.57 (24)
.56 (24)
.56 (24)
.59 (24)
.52 (24)
.53 (24)
.58 (24)
.99 (3)
.61 (24) -99 (3)
.99 (3)
.99 (3)
.99 (3)
.51 (24)
.55 (24)
.52 (24)
.99 (3)
.51 (24) -99 (3)
:n x = log (c) + 1,
c = chemical concentration, was
for each chemical test.
-------
TABLE C 19 (continued)
o
i
00
Chemical
tests
Br*
As*
Ge*
Ga*
Cu*
Ni*
Fe*
Cr*
Y*
Ti*
Si'
Al*
Mg*
MEG1*
MEG2*
MEGS*
MEG5*
MEG6*
MEGS*
MEG10*
MEG13*
MEG15*
MEG18*
MEG21*
MEG23*
MEG24*
MEG25*
LCI
LC2
LC3
R AMI 00
.62 (14)
-.82(11)
-.82(11)
-.82(11)
-.88(9)
-.88 (9)
-.88 (9)
-.75 (9)
.78 (9)
-.77 (9)
Biological tests, health-related
VIABIND ATP PROBVI AMES1 OTHRAT RATWT
.50 (24)
.62 (24)
.52(17)
.95(17)
.57(14)
.53 (23) .55 (16)
.52(17)
-.91(11) -.63(11) .76(16)
.87 (6)
.87 (6)
.81 (7)
-.91(11) -.62(11) .83(16)
.82 (6)
.62(17)
-.91(11) -.62(11) .84(16)
-.94 (9) -.79 (9)
-.95 (9) -.79 (9) .81 (6)
.81 (8)
-.95 (9) -.79 (9)
-.86(9) .73(14) .84(17) .90(11)
-.87(9) .61(14) .64(17) .93(11)
-.87(9) .59(14) .61(17)
AMES2 RAM(2) Wl 38(2) CHO
.99 (3)
.99 (3)
-.55(14)
-.77 (14)
.55 (14)
.99 (3)
-.54 (14)
-.60(13)
-.54(14)
.71(20) .64(11) .99(3)
.91 (9)
.80(9)
.51(20) .71(11) .99(3)
.90(11)
.80 (9)
.91 (17)
.72(11) .99(3)
.78 (9)
.85 (9)
.94 (8)
.79 (9)
.93(18)
.92(18)
.90 (18)
"The transformation x = log (c) + 1, where c = chemical concentration, was made for each chemical test.
(continued)
-------
TABLE C 19 (continued)
o
i
Chemical
tests
R AMI 000 VIABIND
LC4 -.77 (9) -.86 (9)
LC5 .77 (9) .86 (9)
LC6 -.77 (9) -.86 (9)
LC7 -.77 (9) -.86 (9)
LC8 -.77 (9) -.87 (9)
GC8
Hg-AA
Sb-AA
As-AA
Ph
BOD
DO
Conductivity
Dissolved
solids
NH3
ATP PROBVI
.59 (14)
.59 (14)
.59 (14)
.59 (14)
.60 (14)
.85(13)
.63(13)
.66 (12)
.68 (14)
Biological
AMES1
.61(17)
.61(17)
.61 (17)
.61(17)
.62(17)
.55(21)
.90(21)
.90 (18)
tests, health-related
OTHRAT RATWT
.64(11)
.93(11)
.74(11)
.91(11)
.69 (13)
.52 (15)
AMES2 RAM(2) WI-38(2) CHO
.90(18)
.90 (18)
.90 (18)
.90 (18)
.91 (18)
.94 (5)
.88 (26) -.96 (4)
.90 (27)
"The transformation x = log(c) + 1, where c = chemical concentration, was made for each chemical test.
-------
TABLE C 20. CHEMICAL AND ECOLOGY-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS ;>0.50 AND SIGNIFICANCE LEVELS <0.05
ALL STUDIES. PEARSON CORRELATIONS (CORRELATION. NUMBER OF OBSERVATIONS)
o
1
INJ
O
Chemical
tests
Ba*
1
Sb*
Sr*
As*
Mn*
Cr*
Sc*
Ca*
K*
S*
P*
Mg*
Be*
Li*
MEG1*
MEGS*
MEGS*
MEGS*
MEG18*
FWALGAL
-.52(18)
-.50 (18)
-.50(18)
-.50 (18)
-.54 (18)
-.64 (18)
-.51 (17)
-.56 (18)
-.54 (18)
-.65 (18)
-.62(18)
-.51 (18)
Biological tests, ecology-related
SWALGAL FWFSH 24 FWFSH 48 FWFSH 96 DAPH 24 SWFSH 24 SWFSH 48 SWFSH 96 GRSHMP24 GRSHMP48
-.54 (14)
-.63 (14)
-.66 (14)
-.58 (13)
.99 (3)
-.68 (14)
-.62 (14)
.99 (3) .99 (3)
.99 (23) .99 (3)
-.72(13) -.63(15)
-.82(11) -.81(11) -.71(11) -.66(11)
.96 (4)
.96 (4)
-.82(11) -.81(11) -.71(11) -.66(11)
-.82(11) -.81(11) -.72(11) -.66(11)
Ph
Acidity -.55 (15)
Alkalinity
Conductivity
Dissolved
solids -.54 (14)
Cr-*+ -.55(15)
Organ ics
.69 (13)
.60(13)
-.59 (14)
-.56(14)
-.66(14)
-.60 (13)
-.68 (14)
-.64(14)
.54 (14)
-.60 (15) -.57 (15) -.56 (15)
'The transformation x = log (c) + 1, where c = chemical concentration, was made for each chemical test.
-------
o
ro
TABLE C-21. HEALTH- AND ECOLOGY RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
ALL STUDIES COMBINED, PEARSON CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Biological tests,
ecology-related
FWALGAL
SWALGAL
SWFSH 24
SWFSH 48
SWFSH 96
GRSHMP24
GBSHMP48
GRSHMP96
GRSSHP
RAM 1000 WI-38600
.97 (4)
.94 (5)
.94 (5)
.98 (5)
.97 (5)
.97 (5)
.96 (5)
VIABIND
.97 (5)
.97 (5)
.96 (5)
.98(5)
.97 (5)
.94(5)
-.99 (4)
Biological tests, health-related
ATP PROBVI
.80(7)
-.79(15)
-.86(15)
-.75(15)
-.78(15)
-.73(15)
-.64(15)
PROBATP
-.67 (12)
-.59(14)
-.59(14)
.59(14)
-.61(14)
.60(14)
TA1537M
-.68(14)
.72(14)
-.54(14)
-.54(14)
WI-38
-.91 (5)
-------
TABLE C-22. ECOLOGY-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS
1
ro
ro
Biological tests,
ecology-related
FWALGAL
SWALGAL
FWFSH 24
FWFSH 48
FWFSH 96
DAPH 24
DAPH 48
SWFSH 24
SWFSH 48
SWFSH 96
GSHMP 24
GSHMP48
GSHMP 96
GRSSHP
FWFSH
Biological tests, ecology-related
SWALGAL FWFSH 24
.73(14)
—
—
.99 (3)
.73 (14)
.60 (13)
.62 (13)
-.71 (12)
FWFSH 96
.63(16)
.69 (16)
.76(16)
.70 (16)
.70(16)
.75(16)
-.65 (14)
-.75(14)
OAPH 48
.65 (23)
.99 (3)
.55 (16)
.57 (16)
.54(16)
.56 (16)
.58 (16)
-.69 (14)
SWFSH 24
.97(17)
.92 (17)
.77(17)
.77(17)
.70(17)
-.61 (15)
-.74(15)
SWFSH 48
.69 (16)
.57 (16)
.94 (17)
.89 (17)
.87(17)
.80(17)
-.75 (15)
-.77(15)
SWFSH 96 GSHMP 24 GSHMP 48 GSHMP 96
.62(13)
.76(16)
.54(16)
.92(17)
.94 (17)
.87(17) .99(17) .91(17)
.87(17) .92(17)
.90 (17)
-.91(15) -.84(15) -.84(15) -.99(15)
-.87(15) -.67(15) -.69(15) -.76(15)
FWFSH
-.75(15)
-------
o
ro
co
TABLE C-23. HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS
-------
TABLE C-24. CHEMICAL AND HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS -0.50 AND SIGNIFICANCE LEVELS 0.05
ALL STUDIES, SPEARMAN CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Chemical
tests
Bi*
Pb*
Ce*
Cd*
Sr*
Rb*
Br*
Ga*
Fe*
Y*
Ti*
Sc*
Ca*
S*
o Si*
i
£ Mg*
Be*
MEG1*
MEG2*
MEGS*
MEG5*
MEG15*
MEG18*
MEG21*
MEG23*
MEG25*
NOj
N02
LC1
LC2
LC3
LC8
Hg-AA
As-AA
Ph
DO
Biological tests, health-related
RAM 1000 VIABIND ATP PROBVI PROBATP AMES1 OTHRAT RATWT AMES2
.57 (24) .57 (25)
.55(24) .52(17) .53(25)
.59(21)
-.51 (19)
-.63 (19)
-.62 (20)
.52 (24) .52 (25)
.52(17)
-.54 (19)
.51 (21)
-.62(11) -.68(11) .64(16) .63(13)
.80 (9) .79 (9)
.88 (6)
.88 (6)
.58(17) .58(17)
-.60(11)
-.57 3 .94(6) .89(5)
-.76(9) -.81 3> -.76(9) .94(6)
.54 (17)
.73(10) .77(10) .59(17)
.57(18)
.58(14) .64(17) .62(18)
.64(17) -.58(12) .62(18)
.56(21) .54(30)
.69 (18) .62 (27)
.63 (13)
.57(15)
RAM
-.59(14)
.54(14)
-.69(14)
.61 (14)
-.59(14)
.65(14)
.53(14)
.64(14)
-.55(14)
.72(11)
.78(9)
.82(9)
-.73(10)
.72(9)
-.51 (13)
The
.1 -;-£ : = chemical concentration, was made for each chemical test.
-------
TABLE C 25. CHEMICAL AND ECOLOGY-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
ALL STUDIES, SPEARMAN CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Chemical
tests
U*
Pb*
W*
Ce*
La*
Ba*
Cd*
Ag*
Zr*
Sr*
Rb*
As*
^ Mn*
on Ti*
Sc*
Ca*
K*
S*
Si*
Mg*
F*
Li*
MEG15*
SO*
LC6
Conductivity
Organics
FWALGAL SWALGAL FWFSH 96
-.69 (18) -.75 (14)
-.57 (18)
-.63 (18)
-.79 (14)
-.58 (18)
-.63 (18)
-.60 (18)
-.58 (18)
-.55(18)
-.63 (18)
-.56(18)
-.73 (18)
-.52(18)
-.51 (16)
-.63(15)
.73 (9)
-.51(17) -.55(17)
.76(14)
Biological tests, ecology-related
DAPH48 SWFSH48 SWFSH 96 GSHMP24
-.59(17)
-.52 (17)
-.54 (17)
-.57 (17)
.55(17)
-.52(17)
-.58(17) -.50(17)
-.54(17)
.81(8) .76(8) .81(8)
.63(10) .65(10) .83.(10)
-,63 (14) -.66 (14)
GSHMP48 GSHMP96 GRSSHP FWFSH
.56(15)
.53(15)
^.60(17)
-.52(17)
-.53(17) -.56(17)
-.57 (17)
-.53 (17) -.50 (17)
-.51 (17)
-.51 (17)
-.53 (17)
.79 (8) .83 (8)
.70(10) -.64(10)
-.66(14) -.56(14)
"The transformation x = log(c) + 1, where c = chemical concentration, was made for each chemical test.
-------
TABLE C 26. HEALTH AND ECOLOGY-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
ALL STUDIES COMBINED, SPEARMAN CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Biological tests,
ecology-related
SWFSH 24
SWFSH 48
SWFSH 96
GRSHMP24
GRSHMP48
GRSHMP96
FWALG
GRSSHP
FWFSH
Biological tests, health-related
RAM 1000
.89 (5)
.89 (5)
.89 (5)
.89 (5)
.89 (5)
.89 (5)
VIABIND
.89 (5)
.89(5)
.89 (5)
.97 (5)
.97(5)
.97(5)
ATP
.89 (5)
.89(5)
.89(5)
.86(7)
PROBVI
-.67(15)
-.53(15)
-.82(15)
-.78(15)
-.76(15)
.53(14)
PROBATP
-.62(14)
-.64(14)
-.62(14)
-.63(14)
-.65(14)
.65(14)
.77(14)
WI-38
-.95 (5)
o
I
ro
en
TABLE C-27. ECOLOGY RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS ^0.50 AND SIGNIFICANCE LEVELS <0.05
ALL STUDIES COMBINED, SPEARMAN CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Biological tests,
ecology-related
FWALGAL
FWFSH 96
SWFSH 24
SWFSH 48
SWFSH 96
GSHMP24
GSHMP48
GSHMP96
GRSSHP
FWFSH
Biological tests, ecology-related
SWALGAL FWFSH 96
.55(14)
.68(14)
.60(13)
.56 (13)
-.66(14)
.70(12) -.82(14)
DAPH 48
.63 (23)
.73(16)
.78(16)
.58(16)
.58 (16)
.57(16)
.67(14)
SWFSH 24 SWFSH 48
.61(16) .70(16)
.89(17)
.66(15) -.79(15)
-.69(15) -.72(15)
SWFSH 96
.76(16)
.87(17)
.91 (17)
-.92(15)
.83 (15)
GSHMP24
.62(16)
.78(17)
.93(17)
.82(17)
-.84(15)
-.61(15)
GSHMP48
.64(16)
.77(17)
.93(17)
.82 (17)
.99(17)
.93(17)
-.84(15)
-.59(15)
GSHMP 96
.73(16)
.74(17)
.89(17)
.92(17)
.94(17)
.98(15)
-.71 (15)
GRSSHP
.73(15)
-------
o
I
ro
TABLE C-28. HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
ALL STUDIES COMBINED, SPEARMAN CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Biological test,
health-related
VIABIND
ATP
PROBVI
PROBATP
TA1535M
TA1535P
TA98M
TA98P
TA100M
TA1538P
AMES2
RAM (2)
WI-38
RAM 1000 CHO
.83(16)
.89 (16)
-.71 (12)
-.89(11)
.93 (7)
-.90(17)
-.66 (14)
-.71 (10)
-.80 (16)
VIABIND
.58(16)
-.87(12)
-.64(14)
-.66(10)
-.75(16)
Biological test, health-related
ATP PROBVI PROBATP
-.95(11)
-.58(14) .56(27)
-.68(10)
-.70(16) .59(12) .79(11)
-.95 (5)
AMES
.53(18)
.51 (23)
.56 (23)
.79 (10)
.99 (24)
OTHRAT AMES2
.50 (24)
.79 (10)
-------
TABLE C-29. CHEMICAL AND HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS kO.50 AND SIGNIFICANCE LEVELS <0.05
ALL LIQUID SAMPLES COMBINED, PEARSON CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Chemical
tests RAM1000
U*
Th*
Bi*
Ph*
TP
Hg* .99 (3)
Au*
Pt*
Ir*
Os*
W*
Ta*
Hf*
Lu*
Yb*
Tm*
Er*
Ho*
Dy*
Tb*
Gd*
Eu*
Sm*
Nd*
Pr*
Ce*
La*
Cs*
1*
Te*
Sb*
Sn*
In*
Cd«
Ag*
Pd*
Rh*
Ru*
Mo*
Nb*
Zr*
Y*
VIABIND ATP
.90 (5)
.90 (5)
.90 (5)
.90 (5)
.90 (5)
.99 (5)
.90 (5)
.90 (5)
.90 (5)
.90 (5)
.99 (4)
.90 (5)
.90 (5)
.89 (5)
.90 (5)
.90 (5)
.90 (5)
.90 (5)
.90 (5)
.90 (5)
.90 (5)
.90 (5)
.90 (5)
.90 (5)
.99 (3)
.90 (5)
.90 (5)
.90 (5)
.90 (5)
Biological tests, health-related
PROBVI OTHRAT RAM WI-38
.60 (15) -.89 (5)
-.99 (3) -.99 (3)
-.99 (3) -.99 (3)
-.99 (3) -.99 (3)
-.99 (3) -.99 (3)
-.99 (3) .99 (3)
.98 (7)
.52(15)
-.99 (3)
-.99 (3)
-.99(3)
-.99 (3)
-.99 (3)
-.99 (3)
-.99 (3)
.99 (3)
.99 (3)
.99 (3)
.99 (3)
.89 (5) .99 (3)
.99 (3)
.99 (3)
.80 (15)
.99 (3)
.99 (3)
.99 (3)
CHO
*The transformation x = log (c) + 1, where c = chemical concentration, was made for each chemical test.
C-28
-------
TABLE C-29 (continued)
Biological tests, health-related
Chemical ——- _ - _._ . ,
tests RAM1000 VIABINO ATP PROBVI OTHRAT RAM WI-38 CHO
Br* .99 (3)
As* .99 (3)
Ge« .90 (5)
Ga* .90(5) .64(15)
Ni* .90(5) .63(15)
Fe* .88(5) .57(15) -.93(5)
Cr* .97 (4) .96 (4) -.99 (4)
Ti* .99 (3)
So* .99 (3)
Cl* .96 (4)
Na* -.89 (5)
MEG1* .95(11) .99(3) .99(3)
MEGS* .95(11) .99(3) .99(3)
MEG10* -.991 3)
MEG18* .95(11) 99(3) .99(3)
Cl- .99 (3) .99 (3) .99 (3)
S04 -.99 (4)
LC4 -.70 (10)
LC5 -.70 (10)
SbAA .96 (4)
pH .84(13)
BOD .69 (13)
Conductivity .63 (13)
Dissolved
solids .66 (12)
NH3* .67(14)
C-29
-------
TABLE C-30. CHEMICAL AND ECOLOGY RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS ,0.50 AND SIGNIFICANCE LEVELS .0.05
ALL LIQUID SAMPLES COMBINED, PEARSON CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Biological tests, ecology-related
\»1ICIII lt*«l
tests
Ba*
1*
Sb*
Sr*
As*
Ni*
Mn*
Cr*
Sc*
Ca*
K*
S*
P*
Mg*
F*
Be*
Li*
MEG1*
MEG3*
MEG5*
MEGS*
MEG18*
S04E
PH
Acidty
Alknty
Condty
Oissol
CrM
Orgncs
NH3*
FWALGAL
-.52 (18)
-.50(18)
-.50 (18)
-.50(18)
-.53 (18)
-.64 (18)
-.51 (17)
-.56(18)
-.54 (18)
-.64 (18)
-61 (18)
-51 (18)
-.52(16)
-.54(15)
-.54(14)
-.55(15)
SWALGAL FWFSH24 FWFSH48 FWFSH96 DAPH24 SWFSH24 SWFSH48 SWFSH96 GSHMP24 GSHMP48 FWFSH
-.54(14)
-.63 (14)
.99 (3) .99 (3)
-.66(14)
-.58(13)
.99 (3)
-.68(14)
-.62(14)
.99 (3)
.99 (3)
.72(13) -.63(15)
-82(11) -.81(11) -.71(11) -.66(11)
.96 ( 4)
.96 ( 4)
-.82(11) -.81(11) -.71(11) -.66(11)
-.82(11) -.81(11) -.72(11) -.66(11)
-.69(13)
-.56 (14)
-.59(15)
-.59 (14)
-.66(14) -.64(14)
-.60(13) .56(13)
-.68(14) -.54(14)
-.60(15) -.57(15) -.56(15)
*The transformation x = log (c) + 1, where c = chemical concentration, was made for each chemical test.
-------
o
I
CO
TABLE C-31. HEALTH- AND ECOLOGY-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
ALL LIQUID SAMPLES COMBINED, PEARSON CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Biology tests,
ecology-related
FWALGAL
SWALGAL
SWFSH24
SWFSH48
SWFSH96
GRSHMP24
GRSHMP48
GRSHMP96
GRSSHP
FWFSH
Biological tests, health-related
R AMI 000 WI-38600
.97 (4)
.94 (5)
.94 (5)
.97 (5)
.97 (5)
.97 (5)
.96 (5)
VIABIND ATP
.80 (7)
.97 (5)
.94 (5)
.96 (5)
.98 (5)
.97 (5)
.94 (5)
-.99 (4)
PROBVI
-.79 (15)
-.86(15)
.75(15)
-.78(15)
-.73 (15)
-.64(15)
PROBATP WI-38
-.91 (5)
-.67(12)
-.59 (14)
-.59 (14)
-.59 (14)
-.61 (14)
.60 (14)
.72(14)
-------
TABLE C-32. ECOLOGY-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS 0.50 AND SIGNIFICANCE LEVELS <0.05
ALL LIQUID SAMPLES COMBINED, PEARSON CORRELATIONS
(CORRELATION, NUMBER OF OBSERVATIONS)
Biological tests,
health-related
VIABIND
ATP
PROBVI
PROBATP
RAM
WI-38
Biological tests, health-related
RAM1000 WI38600
.86 (7)
.83 (7) .99 (4)
-.78 (7)
-.91 (7)
-.98 (4)
VIABIND
-.91 (7)
-.86 (6)
-.79 (7)
ATP
-.93 (5)
-------
TABLE C-34. CHEMICAL AND HEALTH- RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
ALL LIQUID SAMPLES COMBINED, SPEARMAN CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Chemical
tests
U*
Th*
Bi*
Pb*
Tl*
Au*
Pt"
Ir*
Os*
W*
Ta*
Hf*
Lu*
Yb*
Tm*
Er*
Ho*
Dy*
Tb*
Gd*
Eu*
Sm*
Nd*
Pr*
Cs*
Te*
Cd*
Ag*
Pd*
Rh*
Ru*
Mo*
Rb*
Br*
Ge*
Ga*
Ni*
Fe*
V*
Sc*
Si*
Na*
Be*
MEGT
MEG18*
N03
LC5
HgAA
pH
DO
Cr++
RAM1000
.90 (5)
.90 (5)
.90(5)
.90 (5)
.90 (5)
.97 (5)
.97(5)
.97 (5)
.97 (5)
.90 (5)
.90 (5)
.90 (5)
.90 (5)
.90 (5)
.97 (5)
.90 (5)
.97 (5)
.90 (5)
.90 (5)
.90 (5)
.97 (5)
.97 (5)
.97 (5)
.97 (5)
.90 (5)
.90 (5)
.90 (5)
.90 (5)
.90 (5)
.90 (5)
.90 (5)
ATP
.90 (5)
.90 (5)
.90(5)
.90 (5)
.90(5)
.97(5)
.97(5)
.97(5)
.97 (5)
.90 (5)
.90 (5)
.90 (5)
.90(5)
.90 (5)
.97(5)
.90 (5)
.97 (5)
.90 (5)
.90 (5)
.90(5)
.97 (5)
.97(5)
.97 (5)
.97 (5)
.90(5)
.90 (5)
.90 (5)
.90 (5)
.90(5)
.90 (5)
Biological tests, health-related
PROBVI PROBATP OTHRAT
.58(15)
.52(16)
.84 ( 7)
.56(15)
.57(15)
.57(15)
.57(16)
.62 (16)
.54(15)
.60(16) .54(15)
.51 (16)
.62(15)
.53 (16)
.51 (16)
.56(15)
.61(11) -.82(11)
-.63(11)
.55(15)
-.77(10)
.63 (13)
.58(13)
RATWT RAM
-.95(5)
-.95(5)
-.95(5)
-.95(5)
-.95(5)
-.97 (5)
.97 (5)
-.97(5)
-.97 (5)
.95 (5)
-.95(5)
-.95(5)
-.95(5)
.95(5)
-.97 (5)
.95 (5)
-.97 (5)
-.95(5)
-.95(5)
-.95 (5)
-.97 (5)
-.97(5)
-.64 (16)
-.97 (5)
-.97 (5)
-.97(5)
-.97 (5)
-.95(5)
-.54(16)
-.95 (5)
-.95 (5)
-.95(5)
-.95(4)
-.95(5)
-.95(5)
.95(5)
.83 (7)
.57(15)
The transformation x = log (c) + 1, where c = chemical concentration, was made for each chemical test.
C-33
-------
-iBLE C-35. CHEMICAL AND ECOLOGY-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
ALL L: 1. L SAMPLES COMBINED, SPEARMAN CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Chemical
tests
U*
Pb*
W*
Ce*
La*
Ba*
Cd*
Ag*
Zr*
Y*
Sr*
Rb*
As*
Mn*
Ti*
Sc*
Ca*
K*
S*
Si*
Mg*
F*
Li*
MEG15*
804
NOj
LC6
Conduc-
tivity
Organics
FWALGAL
-.57 (18)
-.68 (18)
-.57 (18)
-.62 (18)
-.64 (18)
-.55 (18)
-.58(18)
-.63 (18)
-.60 (18)
-.58 (18)
-.55(18)
-.63 (18)
-.56 (18)
-.73 (18)
-.52 (18)
-.51 (16)
-.51 (17)
-.55(17)
Biological tests, ecology-related
RV'S-H DAPH48 SWFSH24 SWFSH48 SWFSH96 GSHMP24 GSHMP48 GSHMP96 GRSSHP FWFSH
.56(15)
.53(15)
-.59(17) -.60(17)
-.52(17) -.52(17)
-.54(17) -.53(17) -.56(17)
-.57 (17) -.58 (17)
-.55(17) -.53(17) -.50(17)
-.52(17)
-.54(17) -.53(17)
- ; ~ • i
-: ': -.78 (8) -.81 (8) -.76 (8) .81 ( 8) .79 ( 8) .83 ( 8)
.63(10) .65(10) .63(10) .70(10) -.64(10)
-.63(14) -.66(14) -.66(14) -.56(14)
- • " * ' L
*The transformation x = eg
'••ft : = chemical concentration, was made for each chemical test.
-------
TABLE C-3a HEALTH- AND ECOLOGY-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
ALL LIQUID SAMPLES COMBINED, SPEARMAN CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Biology tests,
ecology-related
FWALGAL
SWFSH24
SWFSH48
SWFSH96
GRSHMP24
GRSHMP48
GRSHMP96
GRSSHP
FWFSH
Biological tests, health-related
RAM1000
.89(5)
.89 (5)
.89 (5)
VIABIND
.89 (5)
.89 (5)
.89 (5)
.97 (5)
.97 (5)
.97 (5)
ATP
.86 (7)
.89 (5)
.89 (5)
.89 (5)
PROBVI
-.66(15)
-.53 (15)
-.82 (15)
-.78 (15)
.76(15)
.53 (14)
PROBATP WI-38
-.94 (5)
-.62 (14)
-.64 (14)
-.62 (14)
-.63 (14)
-.64 (14)
.65 (14)
.77 (14)
o
I
CO
en
TABLE C 37. ECOLOGY-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
ALL LIQUID SAMPLES COMBINED, SPEARMAN CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Biological tests,
ecology-related
FWALGAL
SWALGAL
FWFSH96
DAPH48
SWFSH24
SWFSH48
SWFSH96
GRSHMP24
GRSHMP48
GRSSHP
FWFSH
Biological tests, ecology-related
SWALGAL FWFSH96 DAPH48 SWFSH24
.55(14)
.68 (14) .60 (13)
.63 (23) .61 (16)
.73 (16)
-.65(14) -.66(15)
-.70(12) -.82(14) -.67(14) -.69(15)
SWFSH48
.70 (16)
.68116)
.89 (17)
-.79(15)
-.72(14)
SWFSH96
.56(13)
.76 (16)
.58(16)
.87 (17)
.91 (17)
-.92(15)
-.83 (15)
GSHMP24
.62 (16)
.58 (16)
.78(17)
.93(17)
.82(17)
-.84 (15)
-.61 (15)
GSHMP48
.64 (16)
.57(16)
.77(17)
.93(17)
.82 (17)
.99(17)
-.84(15)
-59 (15)
GSHMP96
.73 (16)
.74(17)
.89(17)
.92(17)
.94(17)
.93 (17)
-.98(15)
-.71(15)
GRSSHP
.73(15)
-------
TABLE C 38. HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
ALL LIQUID SAMPLES COMBINED, SPEARMAN CORRELATIONS (CORRELATION, NUMBER OF OBSERVATIONS)
Biological tests,
health-related
VIABIND
ATP
PROBATP
TA1535M
TA1537M
TA98M
TA98P
RAM
WI-38
RAM1000
.79 (7)
-.83 (6)
-.97 (5)
-.93 (7)
Biological tests, health-related
ATP PROBVI
-.81 ( 7)
.51 (22)
.63 (22)
.95(5)
PROBATP CHO
-.94 (6)
-.93 (7)
-.90 (7)
TABLE C-39. CHEMICAL AND HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS VO.SO AND SIGNIFICANCE LEVELS <0.05
ALL SOLID SAMPLES COMBINED, PEARSON CORRELATIONS
(CORRELATION, NUMBER OF OBSERVATIONS)
Chemical
tests
Bi*
1*
Br*
Fe*
Ca*
Si*
MEG1*
MEG2*
MEGS*
MEG10*
MEG15*
MEG18*
MEG21*
MEG23*
MEG24*
MEG25*
cog
LC1
LC2
LC3
LC4
LC5
LC6
LC7
LC8
GC8
Sb-AA
As-AA
RAM 1000
-.87(9)
-.86(9)
- .70 (9)
.69 (9)
- .84 (8)
- .82 (8)
-.81(8)
-.81(8)
- .82 (8)
-.81(8)
- .82 (8)
-.81(8)
- .87 (7)
- .82 (8)
.83 (6)
- .87 (7)
- .87 (7)
- .86 (7)
- .86 (7)
.86 (7)
.86 (7)
.86 (7)
.86 (7)
VIABIND
- .89 (9)
-.91 (9)
.79 (9)
-.93(8)
-.92(8)
-.91 (8)
-.91(8)
- .92 (8)
-.91 (8)
-.91(8)
-.91(8)
- -94 (7)
- .92 (8)
-.91(7)
- .93 (7)
- .93 (7)
.93 (7)
.92 (7)
.92 (7)
.92 (7)
.93(7)
ATP
-.80(9)
-.72(8)
- .71 (8)
- .77 (7)
.83 (6)
- -81 (7)
- .77 (7)
- .76 (7)
- .76 (7)
.76 (7)
.76 (7)
.76(7)
-70 (7)
Biological tests, health-related
PROBVI PROBATP AMES1 AMES2 RAM
.75(9) .97(9) .71(9)
.69 (9) .94 (9) .78 (9)
.99 (4) .97 (4)
-.79(9)
- .69 (9)
.91 (8)
.91 (8)
.89 (8)
.90 (8)
.91 (8)
.90 (8)
.91 (8)
.90 (8)
.80 (7) .94 (7) .99 (7)
.91 (8)
.79 (7) .91 (7) .93 (7)
.92 (7)
.91 (7)
.91 (7)
.91 (7)
.ill (/)
.!)"(/)
-!M (/)
'M (!.)
.96 (4)
.99 (4)
"The transformation x = log (c) + 1, where c = chemical concentration, was made for each chemical test.
C-36
-------
TABLE C-40. HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
ALL SOLID SAMPLES COMBINED, PEARSON CORRELATIONS
(CORRELATION, NUMBER OF OBSERVATIONS)
Biological test, health-related
health-related
RAM 1000
VIABIND
ATP
PROBVI
TA1537P
TA98P
TA100M
TA100P
TA1538M
TA1538P
AMES2
RAM
RAM 1000
-.77(9)
-.81 (9)
-.78(9)
-.82(9)
-.87(9)
-.78(9)
VIABIND
.97 (9)
-.84(9)
-.88(9)
-.87(9)
-.89(9)
-.90(9)
-.78(9)
ATP
.98 (9)
.90(9)
-.70(9)
-.70(9)
-.71(9)
-.81 (9)
-.75(9)
PROBVI PROBATP AMES1
-.89(5) -.76(9)
- .73 (9)
-.96(5) -.95(5) -.77(9)
.99 (5)
.78 (9)
.69 (9)
.86 (9)
.88 (9)
AMES2
.90 (9)
.92 (9)
.89 (9)
.92 (9)
.81 (9)
RAM
.80 (9)
.68 (9)
.70 (9)
.81 (9)
TABLE C-41. CHEMICAL AND HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS >0.50 AND SIGNIFICANCE LEVELS <0.05
ALL SOLID SAMPLES COMBINED, SPEARMAN CORRELATIONS
(CORRELATION, NUMBER OF OBSERVATIONS)
tests RAM 1000
Bi* -.77(9)
Pb*
I* -.83(9)
Cl*
P*
F*
MEG1*
MEG2*
MEG15*
MEG21*
MEG25*
LCI
LC2
LC3
LC8
VIABIND ATP
-.83(9)
-.88(9)
-.73(8)
-.76(8)
- .71 (8)
-.74(8)
- .75 (8)
Biological tests, health-related
PROBVI PROBATP AMES1
.81 (9)
.68 (9)
.77(9)
-.85 (6)
-.95(5) -.95(5)
.82 (6)
.72 (8)
.80 (8)
.74 (8)
.76 (8)
.80 (7)
.80 (7)
AMES2
.84 (9)
.76 (9)
-.83 (6)
.84 (6)
.79 (8)
.74 (8)
.74 (8)
.78 (7)
-78(7)
RAM
.71(9)
.77 (8)
.77 (8)
.71 (8)
.78 (8)
.80 (7)
.79 (7)
.80(7)
.88 (7)
*The transformation x = log(c) + 1, where c = chemical concentration, was made for each chemical test.
C-37
-------
TABLE C-42. HEALTH-RELATED BIOLOGICAL TEST COMBINATIONS
SHOWING CORRELATIONS ^0.50 AND SIGNIFICANCE LEVELS 0.05
ALL SOLID SAMPLES COMBINED, SPEARMAN CORRELATIONS
(CORRELATION, NUMBER OF OBSERVATIONS)
Biological tests, Biological test, health-related
health related RAM 1000 VIABIND ATP PROBVI PROBATP AMES1 AMES2 RAM
RAM 1000
VIABIND
ATP
PROBVI
PROBATP
TA98M .78 (9)
TA98P .84 (9)
TA100M .72(9) .76(9)
TA1538P .81(9) .74(9)
AMES2 -.76(9) -.68(9) -.76(9) .92(5) .92(5) .98(9) .80(9)
RAM - .72 (9) - .72 (9) - .72 (9) .89 (5) .89 (5) .87 (9) .80 (9)
0 VIABIND ATP
.90 (9) .98 (9)
.88 (9)
PROBVI
-.97(5)
-.97 (5)
-.97 (5)
PROBATP
- .97 (5)
- .97 (5)
- .97 (5)
AMES1 AN
-.77(9)
-.70(9)
-.77(9)
.89 (5)
.89 (5)
C-38
-------
APPENDIX 0
BATTELLE PATTERN RECOGNITION REPORT
Dr. James Hoyland of Battelle-Columbus Laboratories (BCL) examined an
earlier version of this data set using a pattern recognition program spe-
cially developed at BCL for chemical data. A few minor changes were made to
correct the data set but the authors did not consider them significant in
influencing Dr. Hoyland1s conclusions. His preliminary report follows.
D-l
-------
PRELIMINARY REPORT
on
DATA INTERPRETATION BY PATTERN RECOGNITION
to
ENVIRONMENTAL PROTECTION AGENCY
BATTELLE
Columbus Laboratories
Contract 68-02-2686
Directive 105
D-2
-------
INTRODUCTION
This report deals with an initial cursory analysis of biological and
chemical data from 51 samples. There are 48 possible biological properties
and 150 chemical properties, although most data sets are extremely incomplete.
The objective of this preliminary study is to assess the probability that
pattern recognition studies may be of value in determining causal relation-
ships between the chemical data and measured biological activity.
DATA SETS
The data sets used for the preliminary analysis were obtained from
Ms. Nancy Gaskins of Research Triangle Institute. These data have been stored
on disk files at Battelle in such a way that missing items have been flagged.
All mathematical analysis of this data is carried out on a PDP-11/34
minicomputer equipped with a floating point processor and located in the
investigator's laboratory.
PRELIMINARY STUDIES
The first study carried out was a completeness analysis of the
chemical data. This is done by printing a 150 x 51 matrix of Ts or o's.
A 1 indicates that a value for that particular item v/as given, whereas an o
indicates the contrary. The print-out of this analysis is included and is
labeled "logical Analysis of Chemical Data". Variables 1-74 (spark source
elemental analysis) are by far the most complete. In contrast, there is
a marked paucity of organic data, which is extremely distressing in that
it is expected that the latter will be very important for determining
biological acitivty. These data are items 75-100 (MEG 1-MEG 26), 108-115
(LCI - LC8), and 116-122 (GC7-GC13).
D-3
-------
It is of paramount importance in pattern recognition analysis to be
abfe to form a large consistent data set made up of the variable to be classi-
fied (predicted) and the implied causal parameters. To this end, a second .
study was made in which several biological parameters were chosen, the samples
for which these variables are known are selected, and a completeness analysis
of the chemical parameters for this subset is made. The results of this
analysis for the following biological parameters are shown in an accompaning
printout:
Parameter Number Parameter
1 RAM
4 Viability
5 ATP
6 Probit-Viability
7 Probit-ATP
10 Rat-Other
12 Fresh Water Algae
16 96-Hour Fresh Water Fish
18 48-Hour Daphnia
21 96-Hour Salt Water Fish
24 96-Hour Shrimp
25 Rat Weight
A perusal of this analysis is discouraging. Data sets of, at best,
marginal size due to.lack of biological data are further decimated by holes
in the chemical data. This makes accurate assessment of important variables
by pattern recognition techniques both statistically suspect and difficult to
carry out.
The usual way of assessing the importance of variables is to logically
separate the data set into a few discrete classes and then examine the ratio
of the inter- to intra- class variance of the causal parameters to decide
which are most important for inducing class separation. These ratios, however,
are dependent on the number of members in each class and the total number of
samples so that in order to generate results which can be interpreted it 1s
mandatory that the same independent variables be known for each sample. It
is obvious that this is not true in the present case. Another variable
assessment scheme was therefore used which is a measure of the uniqueness
D-4
-------
of ^the range of a given independent variable within classes. This is in
keeping with the fundamental assumptions of pattern recognition that at least
some variables or combination thereof must be within unique ranges within
given classes or no separation is possible. The uniqueness varies between
o (all classes are spanned by the same range of the independent variable) and
1 (all intra- class ranges are unique and there is no overlap).
In order to carry out uniqueness studies as simply as possible,
the completeness analysis results by biological parameter were carefully
considered. For each such biological variable, several uniqueness calcula-
tions were carried out by selecting a set of chemical parameters all of which
are known for several samples, arbitrarily partioning the biological data into
two classes (too little data are available to consider more classes than this),
computing the uniqueness of the selected chemical parameters, and noting only
those which have a uniqueness of at least 0.5.
The computer printout of the results is included. The program is
used interactively. The first entry required is the biological variable
and a template. The latter term refers to which chemical parameter is utilized
to serve as a model for determining which other chemical parameters are also
tested. For example, using Bio 1 and Chemical 1 as a template, the computer
automatically picks all chemical parameters for which values are known for
the first 14 samples listed in the completeness studies by biological para-
meter. These would be chemical parameters 1-5, 6-10, 12, 14-34, 36-46, 48-57,
59-63, 65-67, 69, 70, 72, and 73. The next entry required is the numerical
splitting value. Samples with the biological variable less than this value
are placed in Class 1 and those with a biological variable greater than this
value are placed in Class 2.
It is extremely difficult to assess the significance of the results
of this study due to the very small sample size. We therefore view these
results as suggestive rather than definitive in anyway. Further pattern
recognition v/ork could be carried out using this study as a basis for choosing
chemical variables to consider, but it is felt that such studies will in all
probability lead to many false conclusions due to the sparsity of data (i.e.,
they will not be statistically significant).
D-5
-------
Chemical parameter 55 (cobalt concentration) appears to perhaps
be Delated to RAM results (biological variable 1) since it appears ubiqui-
tously for many template values. Parameter 104 (NO^) appears also in all
templates capable of extracting it implying some relationship is possible.
The small data set obviates attempts to make any further conclusions.
RAM viability (Biological variable 4) is known for 32 of the 51 samples,
thereby making it the best variable for study in a statistical sense. Most
spark source data is available for 30 of these samples, representing again
the best possible chance of producing a statistically sound uniqueness study
on these latter parameters. Unfortuantely, no good parameters were found
although several partioning schemes were tried. Too little organic data
are known to do any meaningful assessment. The study using parameter 76
as a template is considered to be statistically worthless since only two members
can logically be placed in Class 1.
Biological variable 5 (RAM-ATP) is known for only 16 samples. Again,
cobalt (55) is implied to be related to ATP, as is calcium (62) since these
species appear with all templates. It is also tempting to suggest some
relationship may exist with 115 (LC8), but this must be confirmed with a much
larger data set.
Analysis of the probit data (Biological variables 6 and 7) is not
suggestive of any meaningful results. A large number of parameters move in and
out of the lists at random as the template is changed implying a lack of
sufficient data. The same conclusion is reached for Biological variable
10 (rat-other).
Sulfur (chemical parameter 65) apperas to be related to survival
of fresh-water algae (Biological variable 12). All other parameters which
might appear at first sight to be important occur randomly depending on
template except perhaps for barrium (29) and sodium (70).
Studies on other biological variables (96-hour fresh water fish,
48-hour daphnia, 96-hour salt-water fish, and rat weight) did not lead to
any immediate conclusions concerning chemical parameters which might be
important for class separation.
D-6
-------
PRELIMINARY CONCLUSIONS
A few possibly important chemical parameters based on a uniqueness
criterion have been found which may be related to three types of biological
activity. Normally, a full pattern recognition analysis would now be carried
out to determine the predictive power of these chemical parameters. It is
the investigator's opinion, however, that the current data base is too sparse
to draw any meaningful conclusions from such studies since it is not possible
to mask artifacts which can lead to false deduction concerning the parameters
and their values which appear to separate the classes. Further, such studies
would not have any degree of statistical significance.
A further problem which must be addressed is the fact that liquid
effluents should be studied apart from solids. Different biological uptake
methods may apply which can lead to parameter randomization and false con-
clusions. This was not attempted in this preliminary study since it would
further reduce the size of data sets available.
It is recommended that EPA consider the possibility of delaying
further work on this project if more data are to be known in the near future.
If no new data will be available, two options exist. The first is to continue
the current studies even though it is extremely unlikely that any meaningful
results will be obtained. The investigator does not recommend this alterna-
tive since it would appear to be wasteful of funds which may be used to
better purpose elsewhere and from the purely scientific view that generation
and publication of such data is distasteful. The second alternative is to
terminate this task at this point and recommit the remaining funds to other
projects.
D-7
-------
TECHNICAL REPORT DATA
(Please read Inttructions on the reverse before completing)
1. REPORT NO.
EPA-600/7-79-226
2.
3. RECIPIENT'S ACCESSION NO.
d. TITLE ANDSUBTITLE
Interpretation of Environmental Assessment Data
5. REPORT DATE
September 1979
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
N.H. Sexton, F.W. Sexton, L.I. Southerland, and T.D.
Hartwell
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Research Triangle Institute
P.O. Box 12194
Research Triangle Park, NC 27709
10. PROGRAM ELEMENT NO.
RHF.
HE 6?A
PRACT/GF
11. CONTRACT/GRANT NO.
68-02-2156, T.D. 22600
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final: 3/78 - 9/79
14. SPONSORING AGENCY CODE
EPA/600/13
15. SUPPLEMENTARY NOTES IERL-RTP project officer is Larry D.
541-2557.
Johnson, Mail Drop 62, 919/
16. ABSTRACT
repOrt describes preliminary attempts to formulate viable models for
interpreting environmental assessment data. The models are evaluated using data froir
the four most comprehensive environmental assessments. A format for entering
environmental assessment results on FORTRAN computer sheets is presented and more
complete data entry sheets, being developed by IERL-RTP, are discussed. Various
previously proposed models (Source Severity; Source Assessment Models SAM/I,
SAM/IA, and SAM/IB) are investigated using the data from the four previously
mentioned studies. Correlations between biotests and chemical results, and
between biotests, were calculated using the data from these four studies; a
summary of possible correlations is presented. In the study of correlations,
the data are examined from each study individually and from the entire data set
taken as a whole. Generally, results do not reveal a strong degree of association
between the chemical analyses and the biological tests. When all factors
relevant to an intelligent decision on the acceptability of a waste stream are
considered, there appears to be no mathematical model that will encompass all of
them. The models are useful but must still be utilized with considerable care.
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
COS AT i Field/Group
Pollution
Assessments
Mathematical Models
Chemical Analysis
Bioassay
Pollution Control
Stationary Sources
Environmental Assessment
14B
12A
07D
06A
8. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
241
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA form 2220-1 (9-73)
------- |