United States
Environmental Protection
Agency
Environmental Monitoring
Systems Laboratory
Research Triangle Park NC 27711
Research and Development
EPA/600/S4-87/008 June 1987
Project Summary
Intercomparison of Sampling
Techniques for Toxic Organic
Compounds in Indoor Air
Chester W. Spicer, Michael W. Holdren, Laurence E. Slivon,
Robert W. Coutant, and Douglas S. Shadwick
Because people spend a major frac-
tion of their time indoors, concern
exists over exposure to volatile organic
compounds present in indoor air. This
study was initiated to compare several
VOC sampling techniques in an indoor
environment. The techniques com-
pared include distributive air volume
sampling, high and low rate passive
sampling, and whole air collection in
canisters. The study focussed on ten
target compounds:
chloroform benzene
1,1,1 trichloroethane toluene
tetrachloroethylene styrene
bromodichloro- p-dichloro-
methane benzene
trichloro- hexachloro-
ethylene butadiene.
Altogether, ten separate 12-hour sam-
pling experiments were conducted.
Two experiments sampled the back-
ground air of the residence. For the
other eight experiments, the indoor air
was spiked with the target compounds.
Three different spike levels were uti-
lized to cover a range of target com-
pound concentrations. The nominal
spike concentrations were 3, 9, and 27
ng/l for each of the ten target com-
pounds. Statistical analysis of the sam-
pling results indicates generally high
correlation coefficients (greater than
0.90) between the methods. The most
notable exception was benzene, which
had lower correlation coefficients. In
general, the distributed air volume
sampling technique and the low rate
passive technique measured concentra-
tions less than or equal to the canister
method.
This Project Summary was, devel-
oped by EPA's Environmental Monitor-
ing Systems Laboratory, Research Tri-
angle Park, NC, to announce key
findings of the research project that is
fully documented in a separate report
of the same title (see Project Report
ordering information at back).
Introduction
There is an increasing need to meas-
ure volatile organic compounds in air.
This need has been especially great for
hazardous organic species in indoor air.
A variety of approaches have been uti-
lized for collection and analysis of haz-
ardous organic pollutants in air. One of
the most widely used sampling tech-
niques is collection on Tenax* solid ad-
sorbent. This study employed an active
Tenax sampling technique, in which
four samples are collected simulta-
neously at different flow rates. This
technique is called distributed air vol-
ume sampling.
Collection of whole air samples in
passivated canisters is another tech-
nique which has been employed for
volatile organic species sampling. This
study employed a canister sampling
procedure using passivated 6 liter stain-
less steel sampling canisters and a con-
stant flow sampling system.
Passive collection of volatile organic
chemicals is a very attractive means of
sampling, especially for human expo-
sure studies. Small stainless steel cylin-
ders containing Tenax adsorbent be-
Mention of trade names or commercial products
does not constitute endorsement or recommenda-
tions for use.
-------
hind a diffusion barrier have proven to
be very successful for sampling a wide
range of compounds, when used under
appropriate conditions. Both high-
sampling-rate and low-sampling-rate
devices were employed in the present
study.
All of the techniques noted above
have been employed for ambient air
sampling/and some have been used to
sample indoor air. The study reported
here compared these methods for the
first time in an indoor environment,
with emphasis on a number of haz-
ardous organic chemicals observed in
indoor air.
Overview of Study Design
The intercomparison of volatile or-
ganic compound sampling techniques
was carried out in an unoccupied resi-
dence. The techniques compared in-
cluded distributive air volume sam-
pling, high- and low-rate passive
sampling, and whole air collection in
canisters. Samples were collected in du-
plicate for each technique except the
canister procedures; each canister sam-
ple was analyzed twice, to provide an
estimate of analytical precision for this
sampling medium. All samples were
analyzed within three days of collection
by GC/MS. The study focussed on ten
target compounds:
chloroform benzene
1,1,1 trichloroethane toluene
tetrachloroethylene styrene
bromodichloro- p-dichloro-
methane benzene
trichloro- hexachloro-
ethylene butadiene.
Altogether, ten separate 12-hour sam-
pling experiments were conducted.
Two experiments sampled the back-
ground air of the residence. For the
other eight experiments, the indoor air
was spiked with the target compounds.
Three different spike levels were utilized
to cover a range of target compound
concentrations. The nominal spike con-
centrations were 3, 9, and 27 ng/l for
each of the ten target compounds. Spik-
ing was accomplished by vaporizing an
initial charge of the ten-compound mix-
ture into the furnace blower duct to
achieve the approximate spike concen-
tration. Throughout the remainder of
the 12-hour period, the target com-
pounds were continuously introduced
into the duct from a gas cylinder at a
rate calculated to make up for dilution
caused by infiltration. Continuous oper-
ation of the house furnace blower and
several oscillating fans located through-
out the house was used to provide thor-
ough mixing, and also to provide the air
velocity required for efficient operation
of the passive devices.
For the distributive air volume tech-
nique, one blank was run for each sam-
ple (i.e., per set of four tubes). The
blanks transported to the field were not
opened or handled prior to analysis. Du-
plicate samples were collected using a
second distributive air volume collec-
tion system, and both the high- and low-
rate passive samples were collected in
duplicate. Temperature, relative humid-
ity, and air exchange rate were moni-
tored during each experiment.
All samples and blanks were analyzed
on the same GC-MS system to minimize
the uncertainty in the analytical portion
of the measurements so that the com-
parison can focus on the sampling tech-
niques.
Experimental
The residence used for these experi
ments is located in the Upper Arlingtoi
suburb of Columbus, OH, approxi
mately 10 km northwest of downtown
The test house was unoccupied durini
this study, and has been used for thi
past year for heating, ventilation, am
air infiltration experiments. The housi
is a three-bedroom, two-bath rancl
constructed in 1963. All sampling wa
conducted in the living room of the res
idence.
Sampling Methodology for
Target Organic Compounds
Distributed Air Volume
Sampling
Sampling with Tenax adsorben
tubes was accomplished with the sys
tern shown in Figure 1. For experiment:
1 through 4, the mass flow controllers
Adsorbent
Tubes
Figure 1. Diagram of distributive air volume sampler.
-------
were adjusted so that nominal air vol-
jmes of 5, 10, 15, and 20 liters were
collected over the 12-hour sampling pe-
riod. For experiments 5 through 11,
nominal sample volumes of 5, 10, 20,
and 40 liters were collected. Actual volu-
metric flow rates were measured before
and after each experiment. Duplicate
sets of Tenax adsorbent tubes were col-
lected during all ten experiments. These
tubes consisted of packed glass car-
tridges as used with the Volatile Organic
Sampling Train (VOST) System devel-
oped by EPA for stack sampling. Tenax
for the VOST-type cartridges was pur-
chased from Alltech Associates. The
Tenax was extracted in a Soxhlet ap-
paratus with methanol followed by pen-
tane, and dried in a vacuum oven. The
clean material was then transferred to
the individual glass cartridges. Forty-
eight hours before use, the sampling
cartridges were further cleaned by heat-
ing (240°C) under a helium purge (50
ml/minute). When not being used for
sampling or analysis, all Tenax car-
tridges were stored at room tempera-
ture in one-gallon metal cans contain-
ing a layer of charcoal.
A sample blank was carried along
with each set of four Tenax cartridges.
The blank cartridge was not removed
from its storage container; it simply ac-
companied the four adsorbent tubes to
the test site and then back to the labora-
tory for analysis.
Canister Sampling
Six-liter stainless steel canisters (De-
maray Scientific Instrument, Ltd.) were
used for collecting integrated whole air
samples. A stainless steel pump (Metal
Bellows Corp., Model-158) directs flow
to the sample canister. The sampling
rate is controlled with a mass flow con-
troller (Tylan, FC-260). A stainless steel
tube restrictor (0.03-inch inner diameter
by 6-inch length) was positioned up-
stream of the mass flow controller to
prevent pump oscillations from affect-
ing the mass flow. The sample flow was
set so that a final canister pressure of 10
psig was achieved (i.e., 10 cc/min for a
12-hour sample period).
In preparation for sampling, the can-
ister was sequentially filled (15 psig)
and evacuated (25 in. of Hg) five times
using zero air (Aadco, Inc.) as the flush-
ing gas. After the fifth evacuation, the
canister was sealed, transferred to a
higher vacuum system and pumped
down to 0.1 torr. A liquid nitrogen trap
was utilized in this system to prevent
contamination from the pump oil. An
oven (100°C) was also used to bake out
the canisters during evacuation.
Passive Device Sampling
Procedure
Two types of passive sampling de-
vices (PSDs) were utilized in this study.
Both types are stainless steel cylinders
filled with 0.4 g Tenax GC. The device
shown schematically in Figure 2 was de-
signed to have a relatively high sam-
pling rate for organic compounds. This
device employs a 200-mesh"stainless
steel screen as the diffusion barrier.
The second type of device is similar to
that shown in Figure 2, except that the
200-mesh screen is replaced by a stain-
less steel plate with a single 0.5 mm
hole in the center. This device was de-
signed to have a reduced sampling rate.
The sampling rates for PSDs employ-
ing reversible adsorption are dependent
on the specific design of the device as
well as the retention volume of each or-
ganic compound with respect to Tenax.
Based on the design specifications of
these devices and the 12-hour exposure
time used in this study, the effective
sample volumes for each target com-
pound are shown in Table 1. A time-
weighted average sampling rate was
used to compute the volumes for the
high rate PSD, because the rate de-
creases with time for some compounds
over the 12-hour sampling period.
Stain/ess Steel
Perforated Plate
Area
Sorbent
Containment
Area
Before sampling, the PSDs were
cleaned by baking at 200°C for at least
two hours in an oven designed to hold
ten PSDs. During bakeout, the PSDs
were exposed to a flow of 100 ml/min of
hydrocarbon-free N2. Following
cleanup, the PSDs were stored in a
stainless steel cylinder which was
purged with hydrocarbon-free N2 and
pressurized through a quick-connect fit-
ting. With this device, the PSDs were
transported to and from the testing res-
idence. The blank PSDs were kept in the
cylinder during sampling at the resi-
dence.
Three high-rate and three low-rate
PSDs were employed for each experi-
ment. Two devices of each type were
analy/ed; the third was collected as a
backup. One of each PSD type was
transported to the residence for use as a
blank. These PSDs were not removed
from the transfer cylinder.
Efficient operation of the PSDs re-
quires movement of air around the de-
vice to replenish the boundary layer,
which is depleted by sampling. The fur-
nace blower and three oscillating fans
were operated in the living room during
each experiment, providing linear ve-
locities of 25-35 ft-min"1 at the location
of the PSDs.
Additional measurements which
were made to complement the organic
species sampling intercomparison in-
Stainless Steel
Retainer Ring
200-Mesh
Stainless Steel
Diffusion Screen
Stainless Steel
Body
Figure 2.
Thermally desorbably passive sampling device.
3
-------
Table 1. 12-Hour Sample Volumes for High Rate and Low Rate Passive Sampling Devices
Target Compound
chloroform
1, 1, l-fricri/oroethane
tetrachloroethylene
bromodichloromethane
trichloroeth ylene
benzene
toluene
styrene
p-dichlorobenzene
hexachlorobutadiene
High Rate
PSD Volume (1)
8.28
5.19
36.2
28 -
16.6
17.9
40.3
46
41
31.4
Low Rate
PSD Volume (1)
1.80
1.53
1.86
1.70
1.89
2.01
1.92
1.81
1.53
1.25
eluded temperature, relative humidity,
NO/NOX, and air infiltration rate. These
measurements were made in the living
room simultaneously with the organic
species sampling. Temperature and rel-
ative humidity were monitored with an
EG&G Model 911 unit. The concentra-
tions of NO and NOX were monitored
with a CSI Model 2200 portable chemi-
luminescence instrument. Readout
from both of these instruments was re-
corded on two Weather/Measure dual
channel recorders.
Air infiltration rate was measured by
the decay method, using SF6 as the inert
tracer. The concentration of SF6 was
measured every 30 minutes throughout
each 12-hour experiment with a Hewlett
Packard Model 5790 electron capture
gas chromatograph with a 1 cc sam-
pling loop and automated gas sampling
valve.
Analysis Methods
The instrumentation used in this ef-
fort consisted of an Extranuclear Simul-
scan quadrupole mass spectrometer in-
terfaced to a Hewlett Packard 5710A gas
chromatograph. Data acquisition and
reduction was performed with an on-
line Finnigan INCOS 2300 data system.
Sample analyses were performed by
thermal desorption of the adsorbent
(passive and distributive air volume
samplers) or direct sampling (canister)
into a cryogenic trap. Distributive air
samples were desorbed at 200°C for ten
minutes while purging with helium at 80
ml/min. Passive monitors were de-
sorbed at 150°C for 15 minutes with an
equivalent helium flow. The lower de-
sorption temperature for the passive
monitors was selected to minimize ther-
mal degradation of sensitive com-
pounds due to the stainless steel pas-
sive monitor housing. The canister was
sampled through approximately 1
meter of Nafion tubing resulting in a dry
sampling volume of 1.0 liter.
The cryogenic trap consists of a 20-
cm loop of 1/8 in. OD stainless steel
tubing packed with 60-80 mesh
silanized glass beads. The trap was
maintained at 87°K with liquid Argon
during the cryofocussing step. The
cryotrap was then heated to 160°C dur-
ing back flushing with 3 ml/min of he-
lium for a period of seven minutes. The
cryotrap eluent was routed,to a Hewlett
Packard 50 m crosslinked SE 30 wide
bore thick film fused silica capillary
column using a 6-port valve maintained
at 100°C. The initial column temperature
was -20°C.
Following cryotrap desorption, the
column was temperature programmed
at 8°C/min to 200°C. GC/MS acquisition
was initiated after a six minute delay.
Electron impact ionization was used
with the instrument scanning from m/z
46-270 with a 0.5 second cycle time.
Data reduction was performed auto-
matically using in-house developed
software. This involved retention driven
reverse search for the target analytes
followed by integration of characteristic
extracted ion current profiles. Quantifi-
cation was performed by comparison to
the chromatographic peak areas of a
known standard. Calibration and per-
formance check analyses were done by
cryotrap sampling a known volume
from an aluminum compressed gas
cylinder containing 1.0 ppm in each o
the target analytes. The concentratioi
of benzene in the calibration cylinde
was found to be within 6% of the calcu
lated value based on direct comparisoi
with an NBS primary standard cylindei
Results
The dates of the ten experiments am
the nominal spike levels are shown ii
Table 2. Some of the target compound
were present in the background air i
the residence, contributing significant!
to the total residence concentration dui
ing the spiking experiments.
An example of the results from the 1
intercompanson experiments is prc
vided in Table 3. The table lists the e>
periment number, date, spike leve
measured air exchange rate, averag
temperature, average relative humidity
and the mean NO and NO2 concentre
tions over the 12-hour sampling perioc
Table 3 also reports the sample vo
umes, blank-corrected concentration;
and blank values for the various sarr
pling techniques.
Analysis of Results
The statistical methods were take
from linear regression analysis. Th
canister method was chosen as the re
erence method throughout. This mean
that the results from the three Tena
methods (DAV, high rate PSD, and lo\
rate PSD) were used as the depender
variable and the canister results as th
independent variable in the regressior
The canister and Tenax means wer
paired by experiment for each con-
pound. The experimental design for th
DAV Tenax was complicated by a nurr
ber of factors and various types of avei
aging were done before performing th
regression analysis. Details and justif
cation for the regression analysis ar
given in the Project Report.
Table 2. Nominal Spike
Experiment No.
1
2
3
4
5
6
7
9
10
11
Levels for Indoor Intercomparison
Date
June 12, 1985
June 18, 1985
June 24, 1985
June 26, 1985
July 1, 1985
JulyS, 1985
July 10, 1985
August 20, 1985
September 23, 1985
September 30, 1985
Experiments
Nominal Spike Levi
(ng/l)
background
3
3
9
9
27
27
background
3
9
-------
Table 3.
Results from Indoor Intercomparison Study, Experiment 6
Experiment No. 6
Date: July 8, 1985
Organic Spike Level: High
Measured Air Exchange Rate: 0.045 hr
Average Temperature, °C 27°
Average Relative Humidity, percent 37
Average [NO], ppb NA
Average [NO21, ppb NA
Target Chemicals
Sampling Techniques
Integrated Canister Sample 1
Integrated Canister Sample 2
Distributed Ai Volume System 7
Distributed Ai Volume System 1
Distributed Ai Volume System 1
Distributed Ai Volume System 7
Distributed Ai Volume System 2
Distributed Ai Volume System 2
Distributed Ai Volume System 2
Distributed Air Volume System 2
DAV Blank
DAV Blank
High Rate Passive Sampler
High Rate Passive Sampler
High Rate Passive Sampler
HRPS Blank
Low Rate Passive Sampler
Low Rate Passive Sampler
Low Rate Passive Sampler
LRPS Blank
NA Not available
NR Not reported due to interfer-
ence or high blank
Sample
No
C-3
C-3
T-20
T-19
T-13
T-14
T-l
T-4
T-8
T-7
MCT-1
MCT-2
87
90
93
B208
B206
B277
Sample Volume, 1
1 0
1 0
522
1020
1960
3990
478
9 W
1890
3980
*Sampling rate for
PSDs varies with
compound Sam-
ple volume for 12
hour collection
period, in liters
Reported
Units
ng 1
ng 1
ng 1
ng 1
ng 1
ng 1
ng 1
ng 1
ng 1
ng 1
ng
ng
ng 1
ng 1
ng 1
ng
ng 1
ng 1
ng 1
ng
High Rate
PSD
Low Rate
PSD
hloroform
U
23 1
23.8
21 3
22.8
189
11 9
21 5
177
186
100
759
75 7
14 7
152
0 1
83
1 80
Q
1,1-Trichlor
hane
r~~ QJ
21 0
207
21.8
19.8
174
150
186
160
17 1
15 1
03
03
576
567
268
155
155
68
52
1 53
Xrachloro-
hylene
t~- 0)
295
303
31 6
29.4
224
276
287
244
257
28 1
592
430
748
759
02
362
1 86
Z
romodichlo
ethane
tfi E
757
759
762
15.9
129
140
146
11 1
726
734
733
732
77 3
778
0 7
28
: 70
-ichloro-
hylene
£ Q;
757
760
766
75,5
733
73 7
154
118
134
149
790
203
09
99
70 7
1 0
166
1 89
snzene
oa
264
279
28 7
263
24 7
255
245
21 6
25 1
25 1
3 1
05
789
27 0
08
779
788
72
779
207
01
c
Q>
j;
"Q
£
323
337
385
366
30 7
34 7
347
29 7
326
31 7
02
0 1
473
505
34
148
159
20
403
1 92
03
c
QJ
772
77 6
737
734
772
720
12.7
113
11 7
12 1
0 7
769
775
0.3
66
7 1
02
46
1 81
Dichloro-
inzene
Q..Q
773
778
735
728
707
77 6
72.4
772
77 2
779
06
78.5
794
73
66
0 7
47
7 53
exachloro-
Jtadiene
I 5
787
770
774
78.3
759
780
75.6
76.5
767
786
24.6
37.4
77.2
< 7. 7
78
374
725
The two statistics of greatest interest
from the regression analysis are the
slope of the Tenax vs. canister regres-
sion line and the correlation coefficient
between the canister and Tenax means.
The slope is interpreted as a measure of
the agreement between the two meth-
ods in any comparison. If the intercept
is 0, a slope less (greater) than one indi-
cates that the Tenax mean is, generally,
lower (higher) than the canister mean. A
correlation coefficient close to one indi-
cates that the canister and Tenax means
are approximately linearly related over
all experiments. That is, the resultant
means from the measurements by the
two methods show the same relative
peaks and valleys over all experiments.
Figures 3, 5, and 6 are summaries of
the slope (agreement) for a'l three
Tenax methods, while Figure 4 shows
the correlation coefficients for DAV
Tenax for all compounds. The com-
pound numbers on the horizontal axis
refer to the compounds as listed in the
'Overview of Study Design' section of
the Project Report. The chromato-
graphic retention of the compounds in-
creases from left to right: Several con-
clusions have been drawn from the
statistical analysis illustrated in Figures
3, 4, 5, and 6.
1) The slope (agreement) results for
DAV are shown in Figure 3. Note that
± one standard error bars are drawn
on the figure.
a) (9) p-dichlorobenzene and (10)
hexachlorobutadiene had slope
estimates less than one and the
estimates were within two stand-
ard errors of one.
b) (3) benzene, (6) toluene and (7) te-
trachloroethylene had slope esti-
mates greater than one and the
estimates were within two stand-
ard errors of one.
c) The five other compounds had
slope estimates less than one and
the estimates were more than
two standard errors from one.
d) (8) styrene was the only one of
the ten compounds with an inter-
cept estimate (0.79 ng/l) more
than two standard errors from
zero.
e) The more detailed statistical anal-
ysis indicated that for the DAV
Tenax, the slopes for (1) chloro-
form, (2) 1,1,1-trichloroethane,
and (10) hexachlorobutadiene
shown in Figure 3 do not ade-
quately represent the experiment
and these results should be quali-
fied.
2) The low rate PSD slope estimates
(Figure 5) show that all slope esti-
mates were less than one and all in-
tercept estimates were within two
standard errors of zero. With the ex-
ception of (3) benzene, the slope esti-
mates for all compounds were not
within two standard errors of one.
3) The high rate PSD slope estimates
(Figure 6) are as follows:
a) (3) benzene had a slope estimate
less than one and within two
standard errors of one.
b) (1) chloroform and (4) bromo-
-------
dichloromethane had slope esti-
mates less than one but these es-
timates were not within two
standard errors of one.
c) The seven remaining compounds
had slope estimates greater than
one and were not within two
standard errors of one.
d) (10) hexachlorobutadienewasthe
only compound with an intercept
estimate (-1.53 ng/l) more than
two standard errors from zero.
4) The correlation between Tenax
means and canister means was gen-
erally large (>0.90) for all Tenax
methods. The most notable excep-
tion was (3) benzene, which had rela-
tively small correlation coefficients
for all Tenax methods.
Precision estimates were obtained for
the three Tenax methods by pooling the
variance of replicates over all experi-
ments for PSD and DAV and over all
nominal volumes for DAV. The preci-
sion estimate is taken as the square root
of the pooled variance estimate. The
compounds that showed precision esti-
mates greater than 2.0 ng/l were:
(3) benzene for the DAV and low rate
PSD Tenax
(6) toluene for all three Tenax meth-
ods
(7) tetrachloroethylene for the DAV
and high rate PSD Tenax
(10) Hexachlorobutadiene for the low-
and high-rate PSD
Precision estimates for the duplicate
canister sample analyses were less than
1.2 ng/l for all compounds.
2.0
1.5
1.0
0.5
----- Mean Over All Values
456
Compound Number
10
Figure 3.
AGREEMENT. Tenax distributed air volume and canister slope plus and minus
one standard error from the linear regression of Tenax mean vs canister mean.
1.0
.3
2 0.9
0.8
-o 2 -O 0.
o
Mean Over All Values
J I I I J I I I I I
456
Compound Number
8
10
Figure 4.
CORRELA TION: Tenax distributed air volume and canister correlation of Tenax
mean and canister mean.
2.0
1.5
J"
to
0.5
Mean Over All Values
....,._...,. ,....!....,.
3456
Compound Number
1O
Figure 5. AGREEMENT: Tenax low rate personal sampling and canister slope plus and minus
one standard error from the linear regression of Tenax mean vs canister mean.
-------
2.0
1.5
* 1.0
§
«o
0.5
rt
I 26±0.2 T
J T -f
1 i
r i
Mean Over All Values
i i 1 i 1 1 1 1 1
I
1
Figure 6.
1
a
10
Compound Number
AGREEMENT. Tenax high rate personal sampling device and canister slope plus
and minus one standard error from the linear regression of Tenax mean vs canister
mean.
Chester W. Spicer, Michael W. Holdren, Laurence E. Slivon, and Robert W.
Coutant are with Battelle Columbus Division. Columbus, OH 43201; and
Douglas S. Shadwick is with Northrop Services, Research Triangle Park, NC
27709.
James D. Mulik and William A. McClenny are the EPA Project Officers (see
below).
The complete report, entitled "Intercomparison of Sampling Techniques for Toxic
Organic Compounds in Indoor Air," (Order No. PB 87-165 262/AS; Cost:
$18.95, subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officers can be contacted at:
Environmental Monitoring Systems Laboratory
U.S. Environmental Protection Agency
Research Triangle Park. NC27711
-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use S300
EPA/600/S4-87/008
*r
i ' * r- p .> -» T * ^"*"ll'ti
.'^ ->r^. rv I U-0-HU^lMvS," j*
0000329 PS
U S ENVIR PROTECTION AGENCY
REGION 5 LIBRARY
230 S DEARBORN STREET
CHICAGO IL 60604
------- |