0099-2240/78/0035-0136$02.

00/()
APPLIED AND ENVIRONMENIAl MICROBIOLOGY, Jan. 1978, p.136-141 Vol. 35, No. 1
Copyright © 1978 American Society for Microbiology Printed in U.S.A.

Rapid Enumeration of Fecal Coliforms in Water by a

Colorimetric /3-Galactosidase Assay
L. S. WARREN,t R. E. BENOIT,* AND) J. A. JESSEE
Department of Biology, Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061
Received for publication 3 August 1977

The colorimetric f8-galactosidase assay is based upon the enzymatic hydrolysis

of the substrate o-nitrophenyl-fl-D-galactoside (ONPG) by fecal coliforms. This
technique provides an estimate of the fecal coliform concentration within 8 to 20
h. A 100-ml portion of test sample was passed through a 0.45-j.m membrane
filter. This filter was then incubated at 37°C for 1 h in EC medium followed by
the addition of filter-sterilized ONPG. The incubation was continued at 44.5°C
until a half-maximum absorbance (at 420 nm) was reached. The time between
the start of incubation and the half-maximum absorbance was proportional to
the concentration of fecal coliforms present. Escherichia coli (K-12) was used to
measure the kinetics of substrate hydrolysis and the response time of different
cell concentrations. High cell densities produced an immediate response, whereas
1 cell/ml will produce a response in less than 20 h. In field studies in which
samples were taken from a range of grossly polluted streams to relatively clean
lake water, a linear correlation between ONPG hydrolysis times and fecal coliform
most-probable-number values was established. A total of 302 isolates randomly
selected from positive ONPG-EC media, which were derived from 11 different
habitats, were identified as E. coli (96.69%), Enterobacter cloacae (2.32%),
Klebsiella pneumoniae (0.66%), and Citrobacter freundii (0.33%).
There is often an urgent need to evaluate fecal origin could produce gas in a glucose me-
quickly the bacterial quality of water. Natural dium at 46°C, whereas coliform bacteria of
disasters, malfunctions in potable water distri- nonfecal origin are rarely able to do so (5). The
bution and sanitary waste transmission systems, Eijkman test is positive when gas generated by
or contamination of recreation waters from ur- the formic hydrogenlyase system (10) is de-

Downloaded from https://journals.asm.org/journal/aem on 06 April 2024 by 49.185.208.91.

ban and agricultural wastes in runoff illustrate tected, whereas in this assay the enzymatic hy-
some instances where the 22 or 72 h required drolysis of ONPG is employed. Theoretically,
to complete the fecal coliform analysis is too the time required to achieve detectable ONPG
long to analyze the problem and select the most hydrolysis in the inoculated medium is propor-
reasonable solution. The appearance of sewage tional to the quantity of fecal coliforms in the
#' sludge at Long Island beaches in June 1976 just inoculum.
hours before the summer season began is an
example of such a situation. MATERIALS AND METHODS
The fecal coliform index has proven to be one Escherichia coli strain ATCC 14948 (K-12) was the
of the most important indicators of health haz- principal laboratory culture used in the initial phase
ard due to fecal pollution in water (6). This of this study, although strains ATCC 9637 and 9723
report presents a technique in which the fecal were also used. Fecal coliforms isolated from various
coliform concentration can be estimated within water sources were identified as E. coli by growth
8 to 20 h. The possibility exists that the assay and gas production in EC broth in 24 h at 44.5°C,
characteristic colonies on EMB agar, and IMViC type
time may be reduced by an increase in sample ++--. Fecal coliform isolates that were not IMViC
volume or decrease in incubation temperature. ++-- or which formed uncharacteristic colonies on
The assay makes use of the chromogenic sub- EMB agar were identified by the use of the Analytical
stance o-nitrophenyl-,8-D-galactoside (ONPG) in Profiles Index (API 20E, Analytab Products, Inc.,
conjunction with the 44.5°C temperature re- Plainview, N.Y.).
quirement of the fecal coliform test. EC medium (Difco, 24.67 g/liter) with 0.022 M
The basis for this assay is a modification of ONPG (Sigma) was used throughout the study. EC
Eijkman's observation that coliform bacteria of medium was chosen because: the medium contains
bile salts, which inhibit most spore-forming or gram-
t Present address: Charles Pfizer and Co., Inc., Groton, CT positive bacteria capable of fermenting lactose; it is a
06340. well buffered, complex medium that meets the require-
136
Vol.. 35, 1978 RAPID ENUMERATION OF FECAL COLIFORMS IN WATER 137
ments of f?-galactosidase synthesis; the medium has is shown in Fig. 1. A sigmoid curve was shown
little color interference with the ONPG hydrolysis in all cases except in the inoculum containing
product; and E. coli grows very rapidly in this medium. 106 cells/ml. In these cases the inoculum was
The ONPG was dissolved in distilled water, filter sufficient to permit immediate observable enzy-
sterilized, and added to the medium shortly before matic hydrolysis, which yielded a sigmoid curve
inoculation to minimize nonenzymatic hydrolysis of
the galactoside. All other media were Difco products. after a slowly ascending front edge. Detectable
The hypothesis was tested using a more dilute EC levels of the o-nitrophenyl indicator were pro-
medium than is normally used in fecal coliform mea- duced by the 105, 104, 103, and 102 cells/ml in-
surement (1) to reduce optical interference and possi- ocula at approximately 4.25, 6, 7.75, and 9.25 h,
bly the time required for preincubation. respectively. The time from first detectable color
Portions of 1 ml each of various dilutions of a appearance to complete color development was
washed, 12-h culture of E. coli K-12 were added to 9 approximately 1 h. The curves shown in Fig. 1
ml of the ONPG-EC medium in standard screw-cap (excluding 6.0 x 106 and 3.4 x 106 cells/ml) were
culture tubes. The initial cell density was determined used to generate a standard curve using in SAS
by a spread-plate procedure on plate count agar after
serial dilution in phosphate-buffered water blanks (1). General Linear Models nonlinear regression pro-
All inoculated tubes were incubated in a circulating cedure (3). Since the R-square value of the
water bath with temperature (44.5 ± 0.1°C) calibrated regression analysis was 0.893, it was concluded
against a National Bureau of Standards certified ther- that the rate of ONPG hydrolysis in the ONPG-
mometer. The rate of color production from the ONPG EC medium was consistent once a minimum
substrate was determined by periodically removing number of cells was present.
culture tubes. The enzymatic reaction was stopped by The equation that describes the shape of the
the addition of 3 ml of 1 M Na2CO3, and the absorb- standard curve is:
ance was read on a Bausch and Lomb Spectronic 20
at a wavelength of 420 nm. The 100% transmission 1.3312
value for the spectrophotometer was set with uninoc- absorbance
absorbance = 1 +
e(2.24397-0032l time [h])
ulated ONPG-EC medium, which was incubated at
44.5°C for a period equal to that of the inoculated Membrane filters were very useful in concen-
media to account for spontaneous ONPG hydrolysis. trating fecal coliforms from grab samples from
The Na2CO3 developed the full color of the liberated the different aquatic habitats tested, but the
o-nitrophenol and eliminated any turbidity caused by usual cautions must be applied to prevent dam-
the bacterial cells in the medium. The experiment
was repeated with different concentrations of bacteria age to the cells (9). The relative recovery effec-
to evaluate how time of color production was related tiveness of different types of membrane filters
to inoculum size, which varied from 102 to 106 cells. was not studied. Furthermore, the fecal coli-
To evaluate the ONPG method in natural aquatic forms collected from the different habitats may
systems, grab samples were tested from water sources be in different states of physiological vigor,

Downloaded from https://journals.asm.org/journal/aem on 06 April 2024 by 49.185.208.91.

ranging from a grossly polluted, low-flow stream to which are quite different from that of the fresh,
lake and river water of very high bacteriological qual- mid-log E. coli K-12 cells used to generate the
ity. Three to five replicates of 100-ml replicates of data in Fig. 1. Therefore, a preincubation period
each sample were filtered through 0.45-nm Gelman at 370C (before ONPG addition) was used to
GN-6 filters. The filters were aseptically folded and
inserted into culture tubes containing EC medium. minimize the physiological shock of the medium
The samples were then preincubated at 37°C for 1 h and high temperature. A preincubation period
before beginning the 44.5°C incubation. Filter-steri- of 1 h was more desirable than one of 2, 3, or 4
lized ONPG was added at the end of the preincubation h because there was more statistical variation
period. The incubation at 44.50C was terminated when between repetitions in the longer incubation pe-
the absorbance was judged to be between 0.1 and 1.0.
The ezperiment was complete when: a portion of the
medium was streaked on plate count agar for subse- ,r
quent culture identification; the reaction was stopped;
the length of incubation was recorded; and the absorb-
ance was measured as described above. E 1.2

Each water sample that was examined by the 6oi.o e3.105 ,sdo 8*2x103
a
12
ONPG method was also tested for fecal coliform den-
sity by traditional most-probable-number (MPN) pro- 3I
cedures (T). A variety of isolates from the confirmed up / 9.106 7.9.
fecal coliform MPN tubes were tested in the ONPG-
EC medium. 2 3 4 5 S 7 * 9 10
RESULTS Ti. 4.-sl
The rate of ONPG hydrolysis after different FIG. 1. Rates of ONPG hydrolysis by varying sized
concentrations of E. coli K-12 were added to inocula of E. coli K-12 in ONPG-EC medium at
the ONPG-EC medium and incubated at 44.50C 44.50C.
138 WARREN, BENOIT, AND JESSEE Ali''l. ENVIInON. MICROBIOI..
riods; there was no evidence that a significantly is

greater quantity of cells was recovered with the 6

longer preincubation period. Without preincu- - .

bation, samples of very low fecal coliform levels -;

xI
14
b.

(1 to 50/100 ml) sometimes yielded negative 6 2 .

h .
results after 24 to 48 h of incubation. Samples
of 100 ml each from these relatively clean habi-
tats always yielded ONPG activity after a long 2 0 215
Log MPN/100 ml
3.0 3-S

lag when the 1-h preincubation step was used.

Preliminary studies showed that ONPG hy- Fict. 2. Regression analysis of fecal coliform MPN
and time required for half-maximum absorbance
drolysis of mixed cultures of fecal coliforms from after preincubation. The 19 samples were taken from
raw sewage and the habitats used in this study 12 different habitats. Each point is an average of
proceeded at rates similar to those of E. coli K- three to five repetitions. R-square = 0.602. Each lower
12. Therefore, to estimate the fecal coliform case letter refers to a sample obtained from habitats
content of a given water sample, the color de- described in footnote b, Table 1.
velopment of the inoculated ONPG-EC medium
was arrested (preferably near the midpoint), the cepted fecal coliform definition. To test this
length of incubation was recorded, and the ab- hypothesis, bacteria that grew in the ONPG-EC
sorbance was measured. This absorbance was medium at 44.5°C were isolated by streak plate
located on the standard curve described above, isolation on plate count agar and identified.
and the difference between it and the half-max- These data (Table 2) indicate that the test is
imum absorbance (0.7) was noted. This differ- relatively specific for E. coli, although some
ence can be translated on the X axis to a time members of the genera Enterobacter, Citrobac-
correction so that all values are compared to ter, and Klebsiella were isolated. To determine
the time required to reach half-maximum ab- if this test was underestimating the fecal coli-
sorbance after preincubation. This correction form concentration, bacteria isolated on plate
permits the investigator to check samples at set count agar from the fecal coliform MPN proce-
intervals, such as every hour, and still obtain dure were examined for their ability to grow
consistent results. and hydrolyze ONPG in the ONPG-EC medium
To test the hypothesis that the time required at 44.50C. Of the 656 isolates that produced gas
to reach half-maximum absorbance of ONPG in EC broth at 44.50C, 638 grew in the ONPG-
hydrolysis in the ONPG-EC medium is depend- EC medium and hydrolyzed ONPG at a rate
ent on the number of fecal coliforms in a natural comparable to that of E. coli K-12. A total of
sample, a variety of samples were examined and 18 isolates were "atypical" in that only a slight
the results were compared to traditional MPN color production was noted and very little

Downloaded from https://journals.asm.org/journal/aem on 06 April 2024 by 49.185.208.91.

fecal coliform values obtained with the same growth was observed. Of these 18 atypical iso-
samples. These data are shown in Fig. 2. Each lates, 11 were identified as K. pneumoniae, 2 as
point represents the average of three to five E. coli, and the remaining 5 were not identified.
replicates. Table 1 briefly describes the habitats Some strains of E. coli such as ATCC 9723 and
from which the samples were collected and pre- 9637 responded to the ONPG-EC medium in
sents the numerical data from which Fig. 2 was the same manner as the K. pneumoniae isolates.
constructed. The linear equation:
DISCUSSION
time (h) of half-maximum ONPG absorbance
= 17.468 - 2.341 logio fecal coliform MPN The hypothesis that a short-time fecal coli-
form test could be developed based on the rate
was determined by an SAS General Linear of hydrolysis of ONPG was established with the
Models linear regression procedure (3). An R- laboratory data derived from cultures of E. coli
square value of 0.602 was computed, and the K-12 and field data from water samples taken
95% confidence limits of the regression line are from various habitats. The test was selective for
shown. Data points n and s were deleted from the detection of E. coli, and an estimate of the
the regression analysis for reasons to be dis- fecal coliforms population could be obtained in
cussed. Those data points, which depart from 8 to 20 h at densities frequently encountered in
the linear function, reflect the strengths and natural samples. Very consistent results were
weaknesses of the two methods used and will obtained with replicate runs of E. coli K-12,
be discussed in that light. indicating that the experimental procedure was
The usefulness of the ONPG method is based effective for detecting very high and low num-
on the premise that the test is relatively specific bers of bacteria in log growth phase.
yet does not exclude bacteria that fit the ac- In the regression analysis of field samples (Fig.
VOL. 35, 1978 RAPID ENUMERATION OF FECAL COLIFORMS IN WATER 139
2), the values that fall outside the 95% confi- TABLE 2. Indentification of randomly picked
dence limits of the regression line reflect some isolates from ONPG-EC medium by the IMViC
of the differences in physiological conditions and procedure and Analytical Profiles Index (API 20E)
genetic diversity within the group of bacteria Species No. isolated % of total
that qualify as fecal coliforms. Some of the hab- E. coli 292 96.69
itats (c, e) had higher fecal coliform MPN values E. cloacae 7 2.32
than would be predicted by the ONPG hydrol- K pneumoniae 2 0.66
ysis time. These habitats share the common C. freundii 1 0.33
characteristic of being distant from the most
likely sources of fecal contamination. Further- which have been designated as critical areas in
more, the water at these locations was well aer- the Potomac River Basin because of high total
ated and low in temperature (less than 5°C) and and fecal coliform levels (15), also had higher
organic nutrients. Several other habitats (i, o), MPN values than predicted by the ONPG hy-
drolysis time. Both of these habitats were lo-
TABLE 1. Description of sites examined in field cated a short distance below sewage treatment
study plants discharging chlorinated effluent. It is pos-
Half-maximum sible that the fecal coliforms from these sites
were damaged by chlorination or some toxic
Location" Habi,I Fecal col-
tat iforms/
ONPGactivity
Stan- agent.
100 ml Time dard de- Two samples (f, q) were obtained from habi-
(h) viation tats that had a water temperature of approxi-
a North River 1 5 16.93 2.360 mately 30°C and a high organic and inorganic
b Claytor Lake 2 11 13.15 1.084 nutrient status. The fecal coliforms in these sam-
c Back Creek 4 13 17.28 3.060 ples were apparently near their optimum phys-
d Lewis Creek 3 17 13.52 0.295 iological state.
e Shenandoah 4 20 16.67 3.396 The data points n and s were deleted from
River the regression analysis because of the unusual
f Factory 5 33 11.67 0.059 microbial population at this site. These samples
g Claytor Lake 2 49 12.17 0.234 were taken from a foul-smelling, anoxic stream
h Hawksbill 6 110 11.02 0.308 receiving cannery wastes. K. pneumoniae was
Creek found to dominate the fecal coliform MPN cul-
i South River 1 200 14.20 0.488
ture tubes. No K. pneumoniae strains were iso-
j Duck Pond 7 200 12.07 0.313
lated from the ONPG-EC tubes from the same
k Duck Pond 7 200 11.92 0.133
1 Claytor Lake 2 240 11.48 0.359 site. In these tubes E. coli was the only organism

Downloaded from https://journals.asm.org/journal/aem on 06 April 2024 by 49.185.208.91.

m Duck Pond 7 500 10.13 0.223 isolated. K. pneumoniae has been found in high
n Front Royal 8 700 15.33 0.559 numbers in forest-related samples and fresh pro-
o North Fork of 1 918 13.57 0.687 duce (4). Strains of K. pneumoniae, which were
Shenandoah of human or paper mill effluent origin were not
River tested to determine if they were ONPG positive
p North Fork of 1 1,300 10.40 0.209 at 44.5°C. A more extensive testing program
Shenandoah with field samples and pure cultures of known
River
q Factory 5 2,300 8.28 0.119 history must be examined before the specificity
r Hawksbill 6 3,300 9.23 0.243 of this method can be evaluated.
Creek The possibility exists that the relationship
s Front Royal 8 130,000 9.13 0.533 between ONPG hydrolysis time and the real
a Locations a, c, d,e, f, h, i, n, o, p, q, r, and s are in number of fecal coliforms present (approximated
the Shenandoah River Basin of Virginia. Locations b, by the MPN) is nonlinear, but a far greater
g, j, k, 1, and m are in or near Blacksburg, Va. number of samples covering a broader range of
b Habitat codes: (1) Well aerated, downstream of fecal coliform values than was possible in this
chlorinated sewage treatment plant effluent; (2) water investigation are needed to prove or disprove
supply and hydroelectric reservoir; (3) moderate flow this. However, analysis of the available data
stream, receives agricultural runoff and urban sewage indicates that the relationship is linear (F test
discharge; (4) fast flowing, well aerated stream, far fails to reject linearity at a = 0.50). A polynomial
from any known point source; (5) textile plant effluent, regression using a x2 term failed to significantly
warm temperature, high nutrient; (6) moderate flow improve the R-square term.
stream, low dissolved oxygen, high levels of tannery
wastes; (7) small, highly eutrophic pond, receives ex- Geldreich et al. (8) have shown that 96.4% of
tensive urban and rural runoff; (8) low flow, foul- the human fecal coliform strains they examined
smelling stream receiving cannery wastes. were detected by EC broth when incubated at
140 WARREN, BENOIT, AND JESSEE Ai.,i,i,. ENVIRON. MICROBIOL.

44.5°C. Our data (Table 2) are in agreement high-volume sampling apparatus such as that
with Geldreich's findings and suggest that the described by Levin et al. (12) should give more
ONPG test is specific for E. coli yet does not rapid results with no decrease in sensitivity.
exclude a significant portion of the fecal coliform Reasoner and Geldreich (16) have stated that
population. The non-E. coli isolates are not the cost per test for rapid bacteriological assays
likely to impair the usefulness of the test since of water may necessarily be higher than those
their growth in the ONPG-EC medium was slow for conventional methods. This would certainly
and their ONPG hydrolysis was limited. be the case for radiometric methods using la-
The autocytotoxic effects of some 83-D-galac- beled substrates (2), the glutamate decarboxyl-
tosides have been described (11; F. Whitehouse ase method (17), or the gas chromatographic
and H. Proctor, Bacteriol. Proc., p. 148, 1969). presumptive test for coliforms (14). The ONPG
We observed the autocytotoxic phenomenon method is not only highly specific but, since a
with E. coli ATCC strains 9723 and 9637 as well relatively small amount of media is required and
as with the K. pneumoniae isolates. The effect the cost of ONPG is negligible, its cost may be
was not noted with E. coli K-12, and very few equal to or lower than that of conventional
field isolates exhibited autocytotoxicity. Most of methods.
the E. coli isolates demonstrated growth pat- The ONPG method described shows promise
terns similar to that shown by E. coli K-12. E. as a rapid, highly specific test for fecal contam-
coli ATCC strains 9723 and 9637 did produce ination in water. It is especially promising for
luxuriant growth and ONPG hydrolysis at tem- use when a specific fecal coliform limit has been
peratures of up to 43°C, but not at 44.5°C. Van established. For example, if the limit is to be
Donsel et al. (D. J. Van Donsel, R. M. Twedt, 200 fecal coliforms or less per 100 ml, no detect-
and E. E. Geldreich, Bacteriol. Proc., p. 25, 1969) able enzymatic hydrolysis of the ONPG should
have shown the optimum temperature for be observed in less than 11 h (using the protocol
growth of fecal coliforms to be between 40 and of this investigation). The test is a departure
44°C. Our preliminary experiments with E. coli from most recent rapid methods in that simplic-
K-12 were conducted at 43.5°C, and we observed ity, low cost, and specificity have been retained
significantly faster rates of ONPG hydrolysis at while providing an assessment of the bacterio-
that temperature. This study was done at 44.5°C logical quality of water in a shorter time period
because we wished to retain maximum specific- than is possible with conventional methods.
ity, but a lower temperature is clearly more LITERATURE CITED
desirable if specificity is not reduced. It may be 1. American Public Health Association. 1974. Standard
possible to gain a 10 to 15% decrease in the methods for the examination of water and wastewater,
incubation time by lowering the temperature 1 14th ed. American Public Health Association, Inc., New
to 20C.

Downloaded from https://journals.asm.org/journal/aem on 06 April 2024 by 49.185.208.91.

York.
In addition to modifying the temperature of 2. Bachrach, U., and Z. Bachrach. 1974. Radiometric
method for the detection of coliform organisms in water.
incubation, this method may be improved by Appl. Microbiol. 28:169-171.
modifying the composition of the medium or 3. Barr, A. J., J. H. Goodnight, J. P. Sall, and J. T.
making the medium less selective during the Helwig. 1976. A users guide to SAS 76. SAS Institute,
preincubation phase. The preincubation proce- Inc., Raleigh, N. C.
4. Duncan, D. W., and W. E. Razzell. 1972. Klebsiella
dure used in this study appeared to be adequate biotypes among coliforms isolated from forest environ-
for the recovery of fecal coliforms from most ments and farm produce. Appl. Microbiol. 24:933-938.
water samples. However, the data obtained from 5. Eijkman, C. 1904. Die Garungsprobe bei 46° als Hilfs-
samples below sewage treatment plants, where mittel bei der Trinkwasseruntersuchung. Zentralbl.
cells had experienced possible chlorine damage, Bakteriol. Parasitenkd. Infektionskr. Hyg. Abt. 1 Orig.
34:742-748.
indicated that growth and ONPG hydrolysis 6. Geldreich, E. E. 1967. Fecal coliform concepts in stream
were delayed because some cells were physiolog- pollution. Water & Sewage Works. 114:R98-110.
ically debilitated. The procedure proposed by 7. Geldreich, E. E. 1970. Applying bacteriological parame-
Lin (13) to improve membrane filter recovery ters to recreational water quality. J. Am. Water Works
Assoc. 62:113-120.
of fecal coliforms from chlorinated sewage ef- 8. Geldreich, E. E., R. H. Bordner, C. B. Huff, H. F.
fluents could be adapted to the ONPG method. Clark, and P. W. Kabler. 1962. Type distribution of
Simply withholding bile salts from the EC me- coliform bacteria in the feces of warm-blooded animals.
dium until after the preincubation may allow J. Water Pollut. Control Fed. 34:295-301.
9. Geldreich, E. E., H. L. Jeter, and J. A. Winter. 1976.
greater recovery of damaged cells. It should also Technical considerations in applying the membrane
be possible to achieve more rapid results by filter procedure. Health Lab. Sci. 4:113-125.
increasing the quantity of water filtered per test. 10. Hendricks, C. W. 1970. Formic hydrogenlyase induction
In this study, 100-ml amounts were filtered with as a basis for the Eijkman fecal coliform concept. Appl.
Microbiol. 19:441-445.
little or no difficulty, but some samples were 11. Johnston, M. A., and H. Pivnick. 1970. Use of autocy-
too turbid to permit a larger sample size. A totoxic fl-D-galactosides for selective growth of Salmo-
Vol. 35, 1978 RAPID ENUMERATION OF FECAL COLIFORMS IN WATER 141
nella typhimurium in the presence of coliforms. Can. Appl. Microbiol. 32:547-552.
J. Microbiol. 16:83-89. 15. Palmer, R. N. 1976. Critical areas in the Potomac River
12. Levin, M. A., J. R. Fischer, and V. J. Cabelli. 1974. basin. Interstate Commission on the Potomac River
Quantitative large-volume sampling technique. Appl. Basin. ICPRB Tech. Pub. 76-4.
Microbiol. 28:515-517. 16. Reasoner, D. J., and E. E. Geldreich. 1974. Rapid
13. Un, S. D. 1976. Membrane filter method for recovery of bacteriological methods. Proc. 2nd Am. Water Works
fecal coliforms in chlorinated sewage effuents. Appl. Assoc. Tech. Conf., Dallas, Tex.
Environ. Microbiol. 32:547-552. 17. Trinel, P. A., and H. Leclerc. 1972. Automatisation de
14. Newman, J. S., and R. T. O'Brien. 1975. Gas chromato- l'analyse bacteriologique de l'eau-I. Water Res.
graphic presumptive test for coliform bacteria in water. 6:1445-1458.

Downloaded from https://journals.asm.org/journal/aem on 06 April 2024 by 49.185.208.91.