SYMPOSIUM
on the RECOVERY of
INDICATOR
ORGANISMS
employing membrane filters
33
\,
LU
C3
Environmental Monitoring and Support Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
-------
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 document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/9-77-024
PROCEEDINGS OF THE SYMPOSIUM ON THE
RECOVERY OF
INDICATOR ORGANISMS
EMPLOYING MEMDRANE FILTERS
EDITED BY
ROBERT H. BORDNER
CLIFFORD F. FRITH
JOHN A. WINTER
September 1977
Co-sponsored by:
The American Society for Testing and Material
The Environmental Protection Agency
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
ENVIRONMENTAL MONITORING AND SUPPORT LABORATORY
CINCINNATI, OHIO 45268
-------
DISCLAIMER
This report has been reviewed by the Environmental Monitoring and Support Laboratory-Cincinnati,
U.S. Environmental Protection Agency, and approved for publication. Mention of trade names or com-
mercial products does not constitute endorsement or recommendation for use.
-------
FOREWORD
Environmental measurements are required to
determine the quality of ambient waters and the
character of waste effluents. The Environmental
Monitoring and Support Laboratory-Cincinnati
conducts research to:
* Develop and evaluate techniques to
measure the presence and concentration
of physical, chemical, and radiological
pollutants in water, wastewater, bottom
sediments, and solid wastes.
* Investigate methods for the concentra-
tion, recovery, and identification of
viruses, bacteria and other microorgan-
isms in water. Conduct studies to deter-
mine the responses of aquatic organisms
to water quality.
* Conduct an Agency-wide quality assur-
ance program to assure standardization
and quality control of systems for
monitoring water and wastewater.
This publication of the Environmental Moni-
toring and Support Laboratory, Cincinnati, en-
titled:
Symposium on the Recovery of Indicator
Organisms Employing Membrane Filters, re-
ports the proceedings of meetings co-spon-
sored by ASTM and EPA for the specific
purpose of solving current analytical problems
in microbiology. Such meetings involving the
combined expertise of government, academic
and industrial laboratories working with the
manufacturers have the greatest chance for
solutions that will be acceptable to everyone
involved in monitoring and controlling pol-
lution in the environment.
Dwight G. Ballinger
Director, EMSL - Cincinnati
-------
ABSTRACT
The Symposium on the Recovery of Indicator Organisms Employing Membrane Filters sponsored
jointly by the United States Environmental Protection Agency and ^he American Society for Testing and
Materials (Committee D-19 on Water) brought together users, manufacturers, research scientists and repre-
sentatives of government agencies to exchange technical information and review the performance of mem-
brane filters. Problems had been reported with the recovery of bacterial indicators from water and waste-
waters by the membrane filter procedures. They were most pronounced in the fecal coliform test. A key
question was whether the cause was differences in sample types, membrane filters or the test method
employed.
Professionals experienced in water analysis presented relevant field experiences, laboratory data and
research findings and discussed problems concerning recovery of organisms stressed or injured by environ-
mental factors. Media, transport phenomena, physical and chemical characteristics of membranes, mem-
brane sterilization methods, incubation temperatures, techniques for comparison of methods, data analysis,
and the status of the proposed ASTM methods for evaluating membrane filters were discussed.
Solutions suggested at the Symposium included use of two-step incubation, overlay and/or enrichment
techniques and modification of membrane filter structures. Recommendations were made to manufac-
turers and to users to develop and improve intralaboratory quality control programs, to standardize inter-
laboratory testing procedures, to participate in these collaborative studies and to generally improve com-
munications among users, manufacturers and standard-setting organizations.
IV
-------
CONTENTS
Foreword iii
Abstract . . iv
Acknowledgement vii
List o.f Attendees viii
Color Plates 1
Welcome 5
Summary 6
Recommendations 7
Session I Uniform Procedures and Quality Control
Session Chairman Warren Litsky
The Membrane Filter Dilemma 8
Robert H. Bordner
Performance Variability of Membrane Filter Procedures 12
Edwin E. Geldreich
Quality Control of Membrane Filter Media 20
David Power
Statistical Interpretation of Membrane Filter Bacteria Counts 26
Karl J. Sladek* Clifford F. Frith and Richard A. Cotton
Effects of Injury on the Recovery of Indicator Organisms on Membrane Filters 34
Alfred W. Hoadley
Effect-of Temperature on the Recovery of Fecal Coliforms 42
James B. Hufham
Optimum Membrane Structures for Growth of Fecal Coliform Organisms 46
Karl J. Sladek,* Robert V. Suslavich, Bernard I. Sohn and Fred W. Dawson
Session II Comparision Studies of Membrane Filters
Session Chairman Phillip E. Greeson
A Comparison of Membrane Filters and Media Used to Recover Coliforms from Water 58
Michael H. Brodsky* and Donald A. Schiemann
-------
Comparison of Membrane Filters in Recovery of Naturally Injured Coliforms 64
John E. Schillinger, Gordon A. McFeters and David G. Stuart*
Efficiency of Coliform Recovery Using Two Brands of Membrane Filters . 67
Frederick A. Harris* and Carl A. Bailey
Comparison of Membrane Filter Brands for the Recovery of the Coliform Group 73
Alfred P. Dufour* and Victor J. Cabelli
A Comparison of Membrane Filters, Culture Media, Incubation Temperatures,
Polluted Water and Escherichia coli Strains in the Fecal Coliform Test 82
Paul J. Glantz**
Session III Modifications to Improve Recovery
Session Chairman Phillip E. Greeson and Robert H. Bordner
Recovery Characteristics of Bacteria Injured in the Natural Aquatic Environment 98
Gary H. Bissonnette, James J. Jezeski, Gordon A. McFeters* and David G. Stuart
A Layered Membrane Filter Medium for Improved Recovery of Stressed Fecal Coliforms ... 101
Robert E. Rose, Edwin E. Geldreich* and Warren Litsky
Measurement of Fecal Coliform in Estaurine Water 109
Alanson P. Stevens, Rosario J. Grasso* and John E. Delaney
An Evaluation of Methods for Detecting Coliforms and Fecal Streptococci
in Chlorinated Sewage Effluents 113
Shundar D. Lin
The ASTM Proposed Membrane Filter Test Procedure for the
Recovery of Fecal Coliforms 133
Don W. Davis* Margareta Jackson and George R. Kinser
Critique on ASTM Test for Recovery of Fecal Coliforms and
Proposal for Modified Method 153
Norman H. Goddard
Summary of Symposium 157
Francis Brezenski and John Winter
Final Discussion 162
Final Remarks 175
Clifford F. Frith
APPENDIX
Comparison of Membrane Filter Counts and Plate Counts on Heterotrophic and
Oil Agar Used to Estimate Populations of Yeast, Fungi and Bacteria 178
J. D. Walker, B.F. Conrad, P.A. Sessman, and R.R. Colwell
*
**
Speaker
Paper summarized by Bernard J. Dutka, CCIW, Hamilton, Ontario
VI
-------
ACKNOWLEDGEMENT
We wish to thank the attendees and speakers for their lively participation and interest. We also
acknowledge the wholehearted support given to the Symposium by the Subcommittee D-19 on Water,
ASTM.
VII
-------
LIST OF ATTENDEES
Dr. Robert K. Alico
Research Scientist
NASA Ames Research Center
Biological Adaptation Branch
Moffett Field, CA 94035
Mr. Thomas S. Arline, Jr.
Arline's Pollution Abatement
1627 N. 29th Ct.
Hollywood, FL 33020
Dr. Frank T. Baker
Water Microbiologist
Ecotest Laboratory
P. 0. Box 247
Elderton, PA 15736
Mr. D.G. Ballinger
Di rector-MDQARL
EPA
Cincinnati, OH 45268
Mr. Ken Barnhill
Director of Publications
Camp Dresser & McKee
One Center Plaza
Boston, MA 02108
Mr. Frank Bazogh
Biologist
Lee Co. Env. Protection
P. 0. Box 398
Ft. Myers, FL 33902
Mr. J.B. Bell
Supervisor, Microbiology
Laboratory
Dept. of Environment
5320 122 St.
Edmonton, Alberta,
Canada
Dr. Gerald Berg
Chief, Biological Methods Branch
MDQARL
NERC, EPA
Cincinnati, Ohio 45268
Dr. Herbert G. Berger
Regional Engineer
NCASI
P.O. Box 14483
Gainesville, FL 32604
Dr. Gary K. Bissonnette
Asst. Professor
West Virginia University
Rm. 401 Brooks Hall
Morgantown, WV 26506
Dr. Michael Bloomstein
Microbiologist
Pall Corp.
30 Sea Cliff Ave.
Glen Cove, NY 11542
Mr. Robert L. Booth
Technical Coordinator
MDQARL
NERC, USEPA
Cincinnati, Ohio 45268
Mr. Robert H. Bordner
Chief, Microbiological Methods
MDQARL, U.S. EPA
1014 Broadway
Cincinnati, OH 45202
Mr. James A. Brewtenstein
Environmental Specialist
City of Ft. Lauderdale
Utilities Department
4241 N.W. 11th Ave.
Ft. Lauderdale, FL 33309
Mr. Francis J. Brezenski
Edison Laboratory
U.S. EPA
Raritan Arsenal
Edison, NJ08817
Mr. Michael Brodsky
Scientist, Environmental
Bacteriology
Ont. Min. Health
Box 9000, Term "A"
Toronto, Ont., MSW 1R5, Canada
Dr. John D. Buck
Assoc. Prof, of Microbiology
Marine Research Lab.
University of Connecticut
Noank, CT 06340
Mr. Robert M. Carter
Product Administrator
Beckman Inst.
Box 6100
Anaheim, CA 92806
Mr. Richard Chen
Sr. Filtration Engr.
Amerace Corp.
Ace Road
Butler, NJ 07405
Mr. Harold P. Clark
Bio. Lab. Tech.
EPA
4676 Columbia Parkway
Cincinnati, OH 45268
Mr. Jas. L. Collins
The Standard Oil Co.
2109Glendale
Toledo, OH 43614
Mr. Richard A. Cotton
Technical Liaison
Millipore Corp.
Ashby Road
Bedford, MA 01730
Mr. D.W. Davis
Research Associate
Johns-Manville
Box 5018
Denver, CO 80217
VIII
-------
Mr. F.W. Dawson
Director of Research
Millipore Corp.
17 Cherokee St.
Bedford, MA 01730
Mr. Frank Dazzo
Microbiologist
Dept. of Soil Microbiology
University of Florida
2169McCarty Hall
Gainesville, FL 32611
Mr. Rodney S. Dehan
Microbiologist
Dept. of Pollution Control
Tallashassee, FL
Mr. R.M.Dille
Supervisor
Texaco
P. 0. Box 26747
Richmond, VA 23234
Dr. Alfred P. Dufour
Microbiologist
EPA
N.M.W.Q.L. Liberty Lane
W. Kingston, Rl 02892
Mr. Bernard J. Dutka
Head Microbiology
Canada Center of Inland Water
3450 Spruce Ave.
Burlington, Ont.,
Canada
Mr. Theodore A. Ehlke
Microbiologist
U.S. Geological Survey
P. 0. Box 1350
Albany, NY 12201
Mr. Karl M. Fox
Consultant
Research Service
300 Yale Ave.
Swarthmore, PA 19081
Mr. Thomas J. Furdek
Chief Chemist
USA Corps of Engineers
210 N. 12th St.
St. Louis, MO 64501
Mr. S.L. Furman
7035 Commerce
Pleasanton, CA 94566
Mr. Edwin E. Geldreich
Research Director
Microbiological Control in
Water Supplies
US EPA
Cincinnati, OH
Mr. Ralph E. Gentry
Microbiologist
EPA, Region IV, S & A Division
College Station Road
Athens, GA 30601
Mr. Walter Ginsberg
Chief, Water Bacteriology
Chicago Bureau of Water
1000 E. Ohio St.
Chicago, IL 60011
Mr. Norman H. Goddard
Export Area Manager
Sartorius Membranfilter
34 Gottingen
Weender, Landstr 96-102
W. Germany
Mr. Rosario J. Grasso
Sanitary Biologist
Comm. of Mass. D.P.H.
Lawrence Exp. Sta.
Lawrence, MA 01842
Ms. Barbara Green
Research Asst.
Univ. of Massachusetts
Dept. Envir. Sci.
Marshall Hall
Amherst, MA 01002
Mr. Phillip E. Greeson
Hydrologist
U.S. Geological Survey
National Center, MS 412
Reston, VA 22092
Dr. Edward F. Gritsavage
Microbiologist
U.S. EPA
240 Highland Ave.
Needham, MA02194
Dr. Leonard J. Guarraia
Microbiologist
U.S. EPA
401 M St., N.W.
Washington, DC 20460
Mr. Frederick L. Harris
Microbiologist
EPA
25 Funston Rd.
Kansas City, KS66115
Dr. Walter Harris
President
Med-Ox Chemicals Ltd.
145 Bentley Avenue
Ottawa, Ont., Canada
Ms. Patricia Haynes
Microbiologist
Carborundum Co.
R& D
Buffalo Ave.
Niagara Falls, NY 19302
Dr. Charles W. Hendricks
Microbiologist
EPA, WSD,WSME 1011
2005 Kenley Ct.
Alexandria, VA 20308
Dr. Alfred W. Hoadley
Assoc. Prof.
GA. Institute of Technology
School of Civil Engineering
Atlanta, GA 30332
Mr. Lyman Howe
Research Chemist
TVA Water Quality Branch
150-401 Bldg.
Chattanooga, TN 37402
Dr. James B. Hufham
Asst. Prof, of Life Science
University of Missouri
Rolla, MO 65401
Mr. Lothar Jeschke
Schleicher & Schuell
28 Greenbriar Road
Keene, NH 03431
Mr. Wayne Johnson
Chemist
Lee County
813 Dellena Ln.
Ft. Myers, FL 33905
IX
-------
Mr. Ernest Karvelis
Supervisory Aquatic Biologist
U.S. EPA-Cincinnati
5555 Ridge Ave.
Cincinnati, OH 45268
Ms. Harriet Kennedy
Laboratory Director
Evanston N.S. Health Dept.
1806 Maple Ave.
Evanston, IL 60204
Mr. Virgil Kessinger
Chief, Oper. Water Plant
Palm Springs Util.
226 Cypress Lane
Palm Springs, FL 33460
Mr. Robert T. Kirkland, Jr.
Chief, QW Service Unit
U.S. Geological Survey, WRD
244 Federal Bldg.
Ocala, FL 32670
Mr. William L. Klein
Manager, Surveillance Operations
ORSANCO
414 Walnut St.
Cincinnati, OH 45202
Mr. Edward C. Kosakoski
Microbiologist
Miami Reg. Lab.
1350 N.W. 14th St.
Miami, FL 33134
Mr. A. L. Lane
Director of Laboratories
Difco Laboratories
920 Henry St.
Detroit, Ml 48201
Mr. Edwin W. Lard
NSRDC Annapolis Lab.
U.S. Gov't Code 286
12703 Beaverdale La.
Annapolis, MD 21402
Dr. Morris Levin
Microbiologist
EPA
National Marine Water Qual. Lab.
Liberty Lane
W. Kingston, Rl 02892
Mr. Stan Liebaert
Technical Representative
Gelman Inst. Co.
600 S. Wagner
Ann Arbor, Ml 48106
Dr. Shundar Lin
Professional Scientist
III. State Water Survey
P.O. Box 717
Peoria, IL 61601
Dr. Warren Litsky
Professor
Institute of Agricultural and
Industrial Microbiology
Marshall Hall
Univ. of Mass.
Amherst, MA 01002
Mr. J.P. Lively
Chief, Laboratory Operations
Division
Environment Canada
Ottawa, Ont., Canada
Mr. John Loft
Products Manager
Amerace
10 Linda Lane
Summit, NJ 07901
Mr. Robert M. Lollar
Director of Environmental Affairs
Tanners Council
216 Eastern Ave.
Clarendon Hill, IL 60514
Dr. W.IM.Mack
Professor
Inst. Water Research
Michigan State University
Dearborn, Ml 48104
Mr. Jim Marshall
Manager, Membrane R & D
G,elman Inst. Co.
600 S. Wagner
Ann Arbor, Ml 48106
Dr. Maria T. Martins
Senior Biologist
CETESB
Av, Prof. Frederico Hermann Je 465
S.,Paulo, S.P.,
Brazil
Dr. Gordon McFeters
Assoc. Prof, of Microbiol.
Microbiology Dept.
Montana State University
Bozeman, MT59715
Mr. Gene Medley
Biologist
State of Florida, Dept. of Poll.
Control
1773 Pine Ave.
Winter Park, FL 32789
Mr. Stephen Megregian
Director, Water Quality Programs
Wapora Inc.
6900 Wisconsin Ave., N.W.
Washington, DC 20015
Mr. Amar S. Menan
Bacteriologist
Dept. of Environment
P.O. Box 2406
Halifax, Nova Scotia,
Canada
Mr. Gerald Moore
Med-Ox Chemicals Ltd.
145 Bentley Ave.
Ottawa, Ont., Canada
Dr. Bernard Newman
Grad. Dept. of Marine Science
C.W. Post College
Long Island University
Greenvale, NY 11548
Mr. William A. O'Connor
Laboratory Director
Serco Laboratories
2982 Cleveland Ave.
Roseville, MN 55113
Mr. Al Pendleton
Hydrologist
U.S. Geological Survey
USGS National Center
Reston, VA 22092
Ms. Priscilla Positano
Chemist
3588 N.W. 27th St.
Lauderdale Lakes, FL 33311
-------
Mr. Vincent Positano
Chief Chemist
Fort Lauderdale Utilities
3588 N.W. 27th St.
Lauderdale Lakes, FL33311
Dr. David A. Power
Manager, Technical Service
BioQuest
214 Burning Tree Rd.
Timonium, MD 21093
Mr. Robert Pratt
Director
Palm Springs Utilities
226 Cypress Lane
Palm Springs, FL 33460
Mr. Maynard W. Presnell
Research Microbiologist
USPHS-FDA
P.O.Box 158
Dauphin Island, AL 36528
Mr. William G. Presswood
Microbiologist
Tenn. Valley Authority
150-401 Chestnut St.
Chattanooga, TN 37402
Mr. Philip B. Reed
Manager, Ionics Lyo Products
Ionics, Inc.
65 Grove St.
Watertown, MA 02172
Mr. Louis A. Resi
Microbiologist
EPA
5555 Ridge Ave.
Cincinnati, OH 45268
Ms. Cynthia Root
Microbioiogist
Gelman Instrument Co.
600 S. Wagner
Ann Arbor, Ml 48106
Dr. Donald F. Rothwell
Professor Soil Services
University of Florida
2169McCarty Hall
Gainesville, FL 32601
Mr. Augustus Ruser
Microbiologist
Fla. State Div. of Health
P.O. Box 210
Jacksonville, FL 33033
Mr. Dave Rusnell
Environmental Specialist
Fla. Dept. Pollution Control
3201 Golf Course Blvd.
PuntaGorda, FL 33950
Dr. John E. Schillinger
Research Assoc.
Mont. State Univ.
816 N. 17th
Bozeman, MT59715
Mr. John W. Scottie
Assistant Director
Margate Utility Authority Inc.
6700 N.W. 6th Court
Margate, FL 33063
Mr. Irving Seidenberg
Chief, Microbiology, Region II
U.S. EPA
6 Sterling Ct.
E. Brunswick, NJ 08816
Mr. Karl V. Shallenberger
Biologist
City of Ft. Lauderdale
290 N.E. 40th St.
Ft. Lauderdale, FL33315
Mr. K. L.Shull
PEAC
320 N.Cleveland
Chagrin Falls, OH 44022
Mr. Joseph E. Sims
Dir. Membrane Development
Helena Laboratories
4180 Kenneth
Beaumont, TX 77705
Mr. William J. Stang
Senior Microbiologist
EPA, NF 1C-Denver
10626 W. 7th Ave., Apt. 102
Lakewood,C080215
Dr. David G. Stuart
Associate Professor
Montana State University
Dept. of Microbiology
Bozeman, MT 59715
Mr. Harry C. Torno
Staff Engineer
U.S. EPA
Office of R & D (RD678)
Washington, DC 20460
Mr. Ben Trasen, AW
Marketing Mgr.
JohnsManville
Greenwood Plaza
Denver, CO 80120
Mr. Albert D. Venosa
Research Microbiologist
U.S. EPA
National Environment Res. Center
Cincinnati, OH 45268
Mr. L.T. Vlassoff
Manager
Min. of Env. Canada
Box 213
Rexdale, Ont., Canada
Mr. Gene T. Waggy
Union Carbide
P.O. Box 8361
So. Charleston, WV 25303
Mr. Russell West
Operator
Margate Utility Auth. Inc.
924 Pine Ridge Dr.
Plantation, FL 33317
Dr. Ted Williams
Chief, San. Bact. & Chemistry Sect.
Michigan Dept. Public Health
3500 N. Logan
Lansing, Ml 48914
Mr. John Winter
Chief, Qale Br., MDQARL
U.S. EPA
NERC-CINTI
Cincinnati, OH 45202
Mr. Gilman Wommach
Director of Lab. Water Quality Program
Missouri Dept. of Natural Resources
1407 Rehagen
Jefferson City, MO 65101
Mr. Charles C. Wright
Senior Technical Advisor
ARAMCO
1539 W. 16th
Long Beach, CA90813
XI
-------
-------
A. Total coliforms
B. Fecal coliforms
C. Fecal streptococci
Plate 1. Recovery of Indicator Microorganisms
by Membrane Filter Methods.
-------
A. Fecal coliform and non-coliform colonies
B. Fecal coliform colony showing
crystalline structure
C. Total coliform sheen colony
D. Fecal streptococci colonies
Plate 2. Close-ups of Indicator Microorganisms on Membrane Filters.
2
-------
A. Poor distribution
B. Leaking filter assembly
C. Non-wetting area
D. Excessive turbidity and confluency
E. High background count
F. Wrinkled membrane
Plate 3. Problems in Recovery of Indicator Microorganisms as Shown in the Total Coliform Test.
-------
A. Problems in recognition
C. Turbidity effects
B. Poor colony definition
D. Channelling, poor distribution
and overcrowding
E & F. Normal recovery and effects of stress
Plate 4. Problems in Recovery of Indicator Microorganisms as Shown in the Fecal Coliform MF Test.
-------
WELCOME
Robert H. Bordner
On behalf of the American Society for Test-
ing and Materials and the U.S. Environmental
Protection Agency, we warmly welcome you to
ASTM Committee D-19 on Water, and to the
Symposium on the Recovery of Indicator Organ-
isms Employing Membrane Filters. We have esti-
mated the number of attendees to be 105. This
large attendance reflects the high interest in the
use of membrane filters to recover coliforms and
other bacteria from water.
ASTM Committee D-19 has been sojourning
to Florida every January for ten years and last
year the D-19 meeting attracted about 200 enthu-
siastic members to Fort Lauderdale. Committee
D-19 has grown rapidly from the original subcom-
mittees for organic substances, metals and inor-
ganic constituents in water and now includes
many other test areas, such as oil identification,
sediment chemistry, automated analyses, and
biological methods. This year the total attendance
of the D-19 meeting has doubled, reflecting the
increasing activity in biological and microbiologi-
cal monitoring.
The subject of this symposium falls under the
Subcommittee D-19.08 on Membrane and Ion Ex-
change Materials. A task group of this subcommit-
tee was formed to develop test procedures for eval-
uation of membrane filter materials. It is appro-
priate that this symposium be held under the
sponsorship of ASTM as well as EPA, because
of the Society's almost-unique structure that brings
together representatives of the manufacturers,
users, regulatory agencies and research workers
to attack problems of common concern.
-------
SUMMARY
The results of the studies of membranes
described in this symposium cannot be compared
directly because of non-uniform test conditions.
However, we can reach the general conclusion that
variable recoveries of coliforms, fecal coliforms
and fecal streptococci occur with testing of water
and wastewater samples by membrane filtration
(MF), further, that low counts result from injury
caused by natural stream conditions, chlorination,
the elevated incubation temperature of the fecal
coliform test and the structure of 0.45 iim mem-
brane filters.
The effects of stress on recovery are confused
by the variable and unpredictable low recoveries of
indicator organisms from different lots and brands
of membrane filters. Because most laboratories
are not conducting routine quality control checks
on materials, media, equipment and methodology
as part of a within-laboratory QC program, are
not analyzing enough split or replicate samples,
and are not verifying sufficient test results to in-
sure the validity of their data, further discrepancies
occur in comparative data.
It was also obvious from discussion that the
poor communications which exist between manu-
facturers and users contribute to the problem.
Proposed solutions to the problem of low
recovery of indicator bacteria on MFs were:
1. Incubation of filtered samples on a non-
selective medium prior to transfer and
incubation on the selective medium.
2. Short term incubation of filtered sam-
ples on a selective medium at 35 C prior
to incubation at 44.5 C.
3.
4.
5.
Short term incubation of filtered sample
on a non-selective medium at 35 C
prior to transfer and incubation on a
selective medium at 44.5 C.
Use of an overlay of a non-selective agar
on a base layer selective agar for short
term incubation at 35C prior to incuba-
tion at44.5C.
Use of
filter
standard filter.
a larger surface-pore membrane
in place of the 0.45 Aim pore
-------
RECOMMENDATIONS
To the ASTM Subcommittees
Develop MF collaborative testing procedures
for physical, chemical and microbiological char-
acteristics that include tightly written protocols,
which control test variables and include randomi-
zation and standardized statistical evaluation.
Develop a collaborative testing program for
comparison of methods for indicator organisms.
Develop a uniform test protocol for compar-
ing membranes, media and other test conditions.
Until this is done it will be impossible to evaluate
and select improved methodology for indicator
bacteria.
To Membrane Filter Users
Establish a quality assurance program within
the laboratory for supplies, equipment, and analy-
ses. Such control on membranes, media, and test
conditions will assure valid data.
Follow standard test protocols developed by
EPA/ASTM committees for future within-labora-
tory or interlaboratory studies.
Support ASTM, EPA and other testing groups
in evaluations of improved MF procedures using
the standard test protocols.
To Manufacturers
Expand and improve the quality control of
membrane filters, media, reagents, and equipment
used in MF tests.
Encourage the certification of specific pro-
ducts for water analysis.
Improve communications with users. Provide
them with necessary information on pore size,
configuration, additives (extractables), membrane
materials, manufacturing dates of media, changes
of formulations etc.
Establish a voluntary program to notify users
about unacceptable lots and recall such products
if necessary.
To EPA
Develop a test protocol for the comparison
of test methods.
To Researchers
Investigate the physiological basis for en-
vironmentally-stressed cells and apply the results
to improved MF methods.
Solve the problem of testing chlorinated
effluents by MFs.
Define MF specifications for a) a completely
inert, or b) highest count membranes.
-------
THE MEMBRANE FILTER DILEMMA
Robert H. Bordner
Chief, Microbiological Methods Section
Environmental Monitoring and Support Laboratory
Environmental Research Center
U.S. Environmental Protection Agency, Cincinnati, Ohio
ABSTRACT
Reported variances in recovery of indicator
organisms, symposium objectives and requirements
for improvements in membrane filter (MF) pro-
cedures are described. The applications of MF
methods to water quality control and recent
enforcement legislation are reviewed. The import-
ance of quality assurance procedures and the role
of The American Society for Testing and Materials
in the development of standardization test
methods for water are discussed.
INTRODUCTION
The dilemma that we face here today is that
for the past year and a half the membrane filter
has been the target of serious charges concerning
its inability to adequately recover indicator organ-
isms from water and wastewater. Conflicting re-
ports in the literature have documented the differ-
ences in recovery among various membrane filter
brands. Some laboratories have reported variations
in membranes and media from lot to lot. Other in-
vestigators have encountered low recoveries when
using the membranes to enumerate indicator
groups from marine waters and chlorinated ef-
fluents, or those containing toxic materials.
The requirements to enforce and monitor
water and wastewater standards recently estab-
lished by the U. S. Environmental Protection
Agency have greatly accentuated the need for
precise standard procedures and reliable, uniform
test materials.
SYMPOSIUM OBJECTIVES
This symposium was organized to focus on
the real ability of membrane filters to recover
indicator organisms. Much consideration will be
given to fecal coliforms, because they are the
indicator group of primary concern. Also, recovery
problems are intensified at the elevated tempera-
ture of the fecal coliform test.
The ultimate objectives of this symposium
are to:
1. Identify problem areas and future
needs for the use of membrane filters.
2. Review differences in the recovery of
microorganisms and investigate the cause
of these differences.
3. Determine the factors that affect re-
covery.
4. Define the type(s) of filters required for
water analysis.
5. Develop test procedures for the evalua-
tion of membrane filters that will
assure quality control.
THE MEMBRANE FILTER PROBLEM
The membrane filter, first introduced in this
country about 25 years ago, has developed over the
years into an estimated 5 million dollar a year
industry, and is widely accepted by water micro-
biologists for many different tests. The micro-
biologist has learned to appreciate the advantages
of membrane filter procedures: the rapidity, ex-
pediency of direct counts, ease of testing, minimal
space and labor requirements, ability to examine
large sample volumes and portability for field
testing. He must now consider which brand of
filter to use or, indeed, whether to use the mem-
brane filter at all to obtain good recovery.
-------
The manufacturers know the needs and con-
cerns of the users, and have taken a close look at
their product, conducted their own investigations,
and in some cases modified membrane formula-
tions in an effort to improve the filters for water
analysis. They are becoming more aware of the
importance of good quality control of membrane
filter materials.
There are several questions related to the
membrane filter dilemma that we should consider
during this symposium:
1. What degree of recovery is required? Is
the ultimate goal the recovery of all
organisms that will grow under one set
of test conditions for a given parameter
(e.g., specific membrane, medium, and
incubation temperature), or is it the
recovery of all viable organisms, in-
cluding attenuated organisms, to pro-
duce a result as close as possible to the
true count?
2. What type of membrane filter is neces-
sary? Do we need a filter that is com-
pletely inert and does not in itself af-
fect growth and recovery, or one that
does not contain nutrients, materials or
inhibitory materials? Should a filter be
considered that enhances recovery and
growth by releasing soluble materials
which stabilize pH or have a beneficial
buffering effect on the medium? Manu-
facturers point out that such materials
are available.
3. Will membranes provide good recovery
under prescribed test conditions for one
indicator but not provide equally good
recovery for other indicators under
different conditions? Is it desirable to
design a membrane material for the
optimal recovery of each specific indi-
cator group?
4. What is the relationship of the mem-
brane to the underlying media? For
example, is recovery affected by agar
or broth-saturated pad substrates?
5. Does the method of sterilization affect
the ability of the membrane to recover
organisms, as recently reported?
6. How can better recovery be provided
for problem samples such as marine and
estuarine waters, chlorinated effluents,
and wastewaters containing phenols,
metals or other toxic compounds?
7. What product specifications or limits
are critical for membrane filter mater-
ials?
8. What is the real shelf life of the mem-
brane filter?
9. What quality assurance does the manu-
facturer perform?
10. What quality control procedures must
the laboratory carry out on membrane
filters as well as other materials, such as
the media, reagents, distilled water and
other supplies?
11. What are the microbiological and chemi-
cal characteristics of membrane filter
materials for which ASTM subcommit-
tees should develop practical test pro-
cedures, for example, recovery, inhibi-
tory effects, retention and extractables?
12. What statistical measurements should be
used uniformly so that membrane
filter tests can be compared fairly among
laboratories?
USES OF MICROBIOLOGICAL ANALYSES
Microbiological analyses of water and waste-
water are conducted in order to:
1. Assure the quality of potable water at
the water treatment plant and in the
distribution system, raw water sources,
ground water, and bottled water.
2. Determine the quality of water for
recreational, argicultural, irrigation, in-
dustrial, shellfish-raising, and other uses.
3. Investigate the quality of municipal
and industrial wastewaters, and the
effectiveness of treatment.
4. Plan and develop water resources.
5. Perform in-plant studies.
6. Identify the source or trace the disposal
of bacterial pollutants.
-------
7. Carry out research investigations.
8. Monitor and enforce established stand-
ards for wastewater effluents and re-
ceiving streams.
WATER AND WASTEWATER
STANDARDS AND CRITERIA
The requirement for monitoring and enforc-
ing water quality standards is one of the most com-
pelling reasons for developing standard methods
and uniform materials.
The standards for potable water quality were
provided for in the Safe Drinking Water Act
(Public Law 93-523), dated December 16, 1974
(1). Within 90 days after enactment, the maximum
allowable levels of constituents should be pub-
lished in the Federal Register as EPA standards.
The MPN or MF procedure may be used to moni-
tor these limits. For the membrane filter tech-
nique, the quality limit is one total coliform per
100 ml and the action limit is more than 4 total
coliforms per 100 ml. The minimum action re-
quired is immediate repeat sampling. The volume
sampled by the MF technique must be 100 ml. The
minimum number of samples collected each month
is based on the population served by the supply.
A standard plate count incubated at 35 C for 48
hours is recommended.
The suggested criteria for natural waters
were spelled out in the Water Quality Criteria,
1972 (2). This updated volume of the original
Water Quality Criteria published in 1968 (3), was
developed for EPA by the National Science Foun-
dation. A supplemental document will be forth-
coming to support the enforcement of these
criteria. The recommended criterion for recrea-
tional water is 200 fecal coliforms per 100 ml. The
criteria for shellfish-raising waters are 70 total
coliforms and 14 fecal coliforms per 100 ml.
The Federal Water Pollution Control Act
Amendments of 1972 (Public Law 92-500) (4)
established guidelines for the levels of constituents
in municipal and industrial effluents. The monitor-
ing of these standards, under the new national per-
mit system, is entitled National Pollutant Dis-
charge Elimination System, (NPDES). The mini-
mum level of effluent quality attainable by secon-
dary treatment for fecal coliforms was described in
the Federal Register, August 17, 1973 (5). The Act
states that the geometric mean of the fecal coli-
form value for effluent samples collected over a
period of 30 consecutive days shall not exceed 200
per 100 ml for fecal coliforms. The geometric
mean for 7 consecutive days shall not exceed 400
per 100ml.
The microbiological guideline for industrial
effluents from the food processing, textile, feed-
lot, meat products, tanning and sugar processing
industries is a maximum fecal coliform value not
to exceed 400 counts per 100 ml at any time.
QUALITY ASSURANCE
One of the primary responsibilities of the
microbiology laboratory is the adoption of a for-
mal quality control program to assure the reli-
ability and validity of laboratory and field data.
The quality assurance program includes the syste-
matic practice of accepted sampling and analytical
procedures described in Standard Methods or in
the forthcoming EPA manual on microbiological
methods. Quality assurance by trained laboratory
and field personnel and application of good quality
control over materials, equipment, instrumenta-
tion analyses and the resultant data are required.
Assured validity of results is particularly
important if the data are used in court, in the
exchange or compilation of data from other
laboratories, or entered in a data storage bank for
other users.
MEMBRANE FILTER SPECIFICATIONS
In 1965 the Department of Defense, Office
of Medical Materiel, developed a detailed set of
interim specifications to control the quality of
membrane filter materials. Test procedures were
described for characteristics such as recovery,
toxicity, retention, flow rate, porosity, pore
size, and extractables. These specifications have
been superceded by Military Specifications dated
September, 1973 (6).
DEVELOPMENT OF ASTM TEST PROCEDURES
The American Society for Testing and Mater-
ials is directly concerned with the quality control
of materials such as membrane filters because of its
fundamental interest in the development of stand-
ards of characteristics and performance of ma-
terials and the promotion of this knowledge.
ASTM provides a unique system whereby
experts from several fields can get together and
develop, evaluate and approve test procedures by
group concensus. ASTM publishes the procedures
10
-------
using an established protocol and a standard for-
mat. If an evaluation process is required to charac-
terize materials available from various manufac-
turers, to determine that materials are uniform, or
to know that purchase from random sources will
give acceptable results, the mechanism is available
through ASTM. This is the real reason for this
symposium. The test procedures for membrane
filters are being developed through this system.
To develop a test procedure, a task group
chairman solicits proposed methods. These draft
methods are circulated for review until the chair-
man is satisfied that he has a consensus of agree-
ment on a proposed test procedure. This procedure
is then tested by volunteer laboratories in a round
robin procedure if it is amenable to this type of
collaborative testing.
Under ASTM rules, it is necessary to have
supporting data to show that the method is ac-
ceptable. Statistically significant results must
demonstrate the desired precision, and, if possible,
accuracy. The proposed test procedure is then
submitted to the main D-19 committee for appro-
val by ballot. If the procedure is approved and
negative votes, if any, have been resolved, the
method is published in the ASTM manual as an
official procedure with tentative status.
ASTM TASK GROUP ON MEMBRANE FILTERS
The present subcommittee task groups on
microbiological properties and physical-chemical
characteristics of membranes originated in June,
1971 with subcommittee D-19.08, which was
organized to investigate test procedures for mem-
branes used for separating processes such as ion
exchange, electrodialysis, reverse osmosis, and
ultrafiltration. At the initial meeting of this group
much confusion arose over the basic question,
"What is a membrane filter?" It soon became
apparent that the need for controlled pore size
filters varied with the use. Because membrane
characteristics are related to their application, sub-
sections were developed by use under specific
categories.
The present session of this particular sub-
committee that is developing test procedures for
membrane filter materials used in water analyses is
the third since the organizational meeting in June,
1973. The development of test methods has pro-
ceeded through the early discussion and review
stages, but not without growing pains. Test pro-
cedures for recovery and inhibitory effects were
submitted to the subcommittee. Preliminary
round robin tests were carried out on 6 brands of
filters by 10 volunteer water laboratories. The
statistical results indicate a need for a better test
procedure. A detailed report on this procedure
and the round robin test results will be presented
later at this symposium.
The present status of ASTM test procedures
for the microbiological properties of membrane
filter materials is: 1. A preliminary procedure for
recovery has been developed but requires modifica-
tion. 2. A modified test procedure for inhibitory
effects is ready for resubmission to the subcom-
mittee. 3. A proposed procedure for retention has
been prepared for task group discussion.
The conflicting and confusing reports on the
recovery of indicator organisms and the lack of a
test procedure for recovery precipitated the organi-
zation of this symposium to provide a forum for
the in-depth examination of these problems.
REFERENCES
1. Safe Drinking Water Act, Public Law 93-523,
December 16, 1974, 88 Stat. 1660. 42 United
States Code (USC) 300f.
2. Water Quality Criteria, 1972, EPA-R3-73-033,
December, 1973. Office of Research and De-
velopment, EPA, Washington, D.C.
3. Water Quality Criteria. Report of the National
Technical Advisory Committee to the Secre-
tary of the Interior, April 1, 1968. Federal
Water Pollution Control Administration.
Washington, D.C.
4. Federal Water Pollution Control Act Amend-
ments of 1972, Public Law 92-500, October
19, 1972, 86 Stat. 816, 33 United States
Code (USC) Sec. 1151.
5. Water Programs, Secondary Treatment Infor-
mation, EPA Federal Register CFR 40, Part
133, August 17, 1973.
6. Military Specification, Disk, Filtering, Micro-
porous, MIL-D-37005 (DSA-DM), Defense
Personnel Support Center, DPSC-ATT, Phila-
delphia, Pa., September 5, 1973.
11
-------
PERFORMANCE VARIABILITY OF MEMBRANE FILTER PROCEDURES
Edwin E. Geldreich, Microbiological Research Director
Water Supply Research Laboratory
National Environmental Research Center
U.S. Environmental Protection Agency
Cincinnati, Ohio
ABSTRACT
Performance variability in membrane filter
procedures can be traced to a variety of factors
including variations in membrane filter (MF)
manufacture, absorbent pad impurities, MF steri-
lization procedures, commercial media inconsis-
tencies, and technician knowledge, skill and judge-
ment in applying MF procedures to water analyses.
The acceptability of special MF devices in terms of
data reliability is also reviewed.
INTRODUCTION
Many of the problems to be identified in this
presentation on the membrane filter dilemma can
be traced to technological developments that have
failed to recognize the critical requirements asso-
ciated with microbiological applications. As a
consequence, a variety of factors contribute to the
overall problem whose magnitude threatens the
future of this valuable microbiological tool. Prior
to cataloguing these factors, a brief historical
resume should place these issues in better perspec-
tive and hopefully point to directions that both the
manufacturers and laboratories must take to re-
store confidence in membrane filter procedures.
Historical Development of the Membrane Filter
The initial attempt to develop an artificial
membrane as a substitute for those found in nature
has been credited to Fick in 1855 (1-3). However,
Fick experienced difficulties in using these fragile
collodion membranes and eventually abandoned
the idea. Once the conceptual design of a collo-
dion sac for dialysis became apparent, numerous
applications of this membrane form were reported
during the period of 1893 and 1905 (1). Soon it
was discovered that the mixture and concentration
of alcohol-ether and glacial acetic acid solvent
systems would change the porosity in the nitro-
cellulose during evaporation and that the resulting
permeability could be preserved by placing the
forming membranes in water before solvent evapor-
ation was completed (4-8). Glycerol, soluble in the
alcohol-ether mixture but insoluble in nitro-
cellulose, was found to increase membrane per-
meability (9). Improved flexibility of the finished
membrane was achieved by the addition of a small
concentration of castor oil. From these pioneering
investigations it became evident as early as 1915
that: a) the control of porosity was the key to the
successful development of nitro-cellulose mem-
branes; b) reproducibility of pore size was easier
in membrane sheets than in the production of a sac
configuration and; c) there was a need for careful
control of production methods and exact concen-
tration of ingredients used in the preparation of
membranes (10).
By 1916 the usefulness of a nitro-cellulose
membrane as a tool in bacteriology had been recog-
nized. Removal of bacteria from any reasonable
quantity of fluid and cultivation of the micro-
organisms in place on a membrane surface had long
been a goal for many investigators. However, the
usefulness of such membranes was extremely
restricted because their preparation was difficult
and the product was of uncertain quality and por-
osity. Zsigmondy and Beckman were the first to
develop a method for the preparation of a mem-
brane that could readily be adapted to commercial
production and they were issued a United States
patent in 1922 based on an application submitted
in 1919 (11-14). This procedure consisted of dis-
solving nitro-cellulose or nitro-cellulose acetate in
a mixture of acetone and glacial acetic acid, pour-
ing the solution on a glass plate in a thin layer;
allowing the volatile solvents to evaporate at 18 C
in a 60 percent relative humidity; and washing the
finished membrane in water. Pores of a specific
size range were produced by controlling the con-
centration of nitro-celhjlose in the basic solution,
12
-------
composition of solvent mixture and relative
humidity during solvent evaporation. Increasing
the percent of water in the solvent mixture or ele-
vating the percent relative humidity during the
drying process resulted in larger pore sizes.
The commercial manufacturing technology
for membrane filters remained basically unaltered
from the Zsigmondy process until the post World
War II period. From a review of German military
intelligence gained during a scientific reconnais-
sance after the war, Dr. Alexander Goetz prepared
a complete report on the manufacturing process,
characteristic properties and bacteriological appli-
cations of the Zsigmondy membrane filter (15).
Research into production techniques for the
U.S. Army Chemical Corps led Dr. Goetz to de-
velop a membrane filter that did not require stor-
age in water, to incorporate a grid imprint, and to
design an associated apparatus for use in filtering
water. This improved processing was then con-
tracted to the Lovell Chemical Corporation which
later organized the Millipore Filter Corporation for
American production of the improved filter made
from domestic materials. Numerous refinements
relative to automation of the process have since
been made to improve the uniformity and quality
of the product in a competitive market that in-
cludes several other American membrane filter
manufacturers: Gelman, Schleicher and Schuell,
Nuclepore, Helena, and Johns-Manville. Among the
foreign manufacturers, Oxoid (British) and Sar-
torius (German) are the most available mem-
branes in this country.
Application of the nitro-cellulose membrane
developed simultanously with the technological
advances in membrane filter manufacture. How-
ever, application to bacteriological procedures was
hindered in early years by the uncertain quality
and inadequate supply of the material. In 1919, a
nitro-cellulose filter was first reported to recover
organisms of tuberculosis from urine (16). Ap-
parently, the first attempts to culture micro-
organisms on membrane filters were done in
Russia and Germany during the 1930's (17-21).
Dr. Mueller (Hygienic Institute, Hamburg) adapted
these techniques to the urgent needs for an ade-
quate monitoring of public water supplies and for
emergency situations resulting from the war
devastations of Germany in the period of 1943 to
1945. Routine analysis of potable waters for coli-
forms was accomplished by placing the membrane,
after filtration, on a substrate of seven filter papers
saturated with Endo broth (20). Mueller was also
successful with this type of procedure in her inves-
tigation of an epidemic of typhoid fever occurring
in Hamburg (21). Samples were filtered through a
nitro-cellulose membrane filter which was then
placed on a modified bismuth sulfite medium for
cultivation of typhoid organisms. The interest in
membrane filter techniques for the bacteriological
examination of water became widespread through-
out the English speaking world following studies
in the nineteen-fifties by Clark et al. (22,23) and
Goetz et al (24, 25) in the United States and
Taylor and Burman (26, 27) in Great Britain.
Analysis of data available to the U.S. EPA
laboratory evaluation program indicates that state
health, state environmental, city-county health,
municipal water treatment and private laboratories
are examining approximately 3.5 million samples
annually from this nation's public and private
supplies and in gathering and monitoring data on
natural waters relative to state and federal stand-
ards for a variety of water quality uses (28). An
estimated one million additional samples are
analyzed by local laboratories in quality control
monitoring of industrial and municipal waste dis-
charges as required in the National Pollution Dis-
charge Elimination System. Of course, not all of
these analyses involve the membrane filter proce-
dure but national statistics suggest 52.1% of all
laboratories currently involved in the nationwide
laboratory evaluation program are using the
membrane filter procedure on a variety of water
samples estimated to be 1.8 million analyses per
year. Thus, there is a substantial amount of moni-
toring data being developed from membrane filter
procedures which should be reliably measuring
water quality.
Membrane Filter Quality for Microbiology
Commercial brands of membrane filters may
vary in performance as a result of manufacturing
technology, materials, and degree of quality con-
trol exercised. For microbiological applications,
there must be a complete retention of organisms
on or near the surface of a non toxic, inert matrix
which permits a continuous contact with nutrients
from a medium held in a substrate below the mem-
brane. These basic conditions place demanding
requirements on the quality of every membrane
used in the laboratory. Basic difficulties encount-
ered with membrane filters generally relate to pore
distribution, hydrophobic filter areas, grid line ink
restrictions, membrane materials, sterilization
practices, and poor storage characteristics that
cause increased filter brittleness and surface
warping (29-39).
13
-------
Membrane filter pores should be uniformly
distributed and have a diameter of 0.45 micron
(± 0.02 micron) for routine bacteriological tech-
niques. Pores of some commercial lots of mem-
brane filters have been found to be so small in
some areas of the filter that serious local reduction
in the flow rate occurs. The filter should be free of
visible non-porous areas which prevent the diffu-
sion of nutrients to the upper surface of the
membrane. Any bacterial cells entrapped on such
surfaces will not develop into visible colonies for
lack of nutrients. When M-Endo is used in a test of
diffusibility, non-wetting areas on the filter will
remain white and dry. Such observations should
not be confused with air bubbles, which can be
removed by reseating the membrane over the
medium-saturated pad or agar base. At the other
extreme, pores larger than 0.7 micron will not re-
tain organisms associated with indicator groups.
For complete bacterial separation from liquids,
membrane filter porosity of 0.22 micron is re-
quired to insure retention of the smallest bacteria
through physical impingement or electro-static
entrapment.
The ink used to imprint the grid system on
the membrane filter should be non-toxic to all bac-
teria cultivated on the filter' surface (30). Some
inks have been found to be bacteriostatic or
bactericidal. Such effects can be recognized
through restrictive colony development adjacent
to the imprinted lines. These growth restrictions
may not only be caused by inhibition from toxic
inks but also from thick ink imprints that "wall-
in" grid squares and by hydrophobic inks which
prevent nutrient diffusion to sites in the ink
imprint. As an additional characteristic, inks
selected for grid imprinting should not "bleed"
across the membrane surface after a 24 hour con-
tact with any medium normally used at 44.5 C
incubation. Heavy imprinting of the grid system
can also result in a network of "canal-like" inden-
tations that frequently become filled with con-
fluent growth.
The physical structure of the membrane
filter material should be such as to provide an
optimum retention of bacteria on the surface with
little migration to areas within the pore matrix.
When surface penetration occurs, growth may be
limited in development during the colony counting
procedure.
Chemical composition of membrane filters
has largely been limited to polymerized cellulose
esters since membrane filter technology initially
developed in this direction. Conventional media
designed for selective recovery of bacterial indica-
tor groups or pathogens using agar pour plates,
streak plates or broth cultures had to be redesigned
to compensate for the physical-chemical properties
characteristic of nitro-cellulose materials (29, 40,
41). For example, the selective adsorption of dyes
excluded the use of acid to neutral dyes as indica-
tor systems and necessitated the use of increased
amounts of brilliant green as a suppressive agent in
Kaufmann's Brillian Green agar to obtain the de-
sired suppression of some of the unwanted bac-
terial population. Similarly, various nutrients such
as tryptone, polypeptone and proteose peptone
No. 3 were found to be superior in membrane
filter media than in the same media used originally
with peptone in their formulation. The result has
been the creation of a family of media designed
specifically for use with nitro-cellulose membrane
filter products. With these experiences in mind,
manufacturers should be careful about revising the
Goetz membrane filter process. Changes involve
the risk that recommended media may suddenly
become less sensitive or less selective. Some com-
pounds introduced to the membrane filter may
improve flexibility, flow rate or stabilize porosity.
However, these substances should not become a
source of fermentable carbohydrates that cause
false colony differentiation, create pH shifts in the
indicator systems, are selectively toxic for specific
organisms, or adversely depress the selective
action of differential media by providing the
bacteria with a highly nutritive organic compound.
In essence, membrane filters should remain inert
to bacterial reaction, and unchanged in those
physical-chemical characteristics that effect media
selectivity and sensitivity.
Sterilization of the membrane filter is essen-
tial to all applications involving filtration of liquids
for bacterial removal or for use in bacterial cultiva-
tion. Prior to the development of the Goetz mem-
brane filter process, membrane filters were steri-
lized in the laboratory by gentle boiling in distilled
water for 20 minutes and repeating the procedure
a second time with fresh distilled water (15-29).
This procedure served the double purpose of
sterilizing the membrane and of extracting any
residual toxic substances. In retrospect, the con-
tinued use of this leaching and sterilization pro-
cedure would have avoided many of the variations
in membrane filter performance now evident.
However, the procedure does take more time to
execute and is a recognized inconvenience in busy
laboratories examining 50 to 200 samples per day.
Goetz proposed the use of ethylene oxide
(0.5 ml per liter volume) for 3 to 4 hours at room
temperature in a dessicator followed by air flushing
14
-------
for several hours to remove the sterilizing gas resi-
dual. This sterilization procedure is very effective
but requires a thorough flushing, preferably in a
vacuum system at an elevated temperature for
several hours, with the procedure being repeated
on two or three succeeding days for complete
removal of gas entrapped in pockets in the mem-
brane matrix. As a safety precaution relative to
the explosive nature of pure ethylene oxide in air
at certain concentrations, it is more desirable to
use a mixture of ethylene oxide and carbon
dioxide to decreased explosive and flammable
properties.
With improvements in membrane filter
technology, subsequent research demonstrated that
membrane filters (packed with absorbent pads at
top and bottom of a stack of filters wrapped in
Kraft paper) could be sterilized prior to use by
autoclaving at 121 C for 10 minutes. Immediately
following the time-temperature exposure, the
autoclave should be rapidly exhausted to atmos-
pheric pressure and the membranes promptly
removed to minimize total heat exposure. Exces-
sive exposure to sterilization temperatures can
cause membranes to become brittle and distorted.
This problem is also aggravated by sterilization
of membrane filter stocks held in storage for
periods beyond 18 months.
The introduction of prepackaged and pre-
sterilized membrane filters in resealable envelopes
by several manufacturers was considered a desir-
able convenience by the laboratory and immedi-
ately accepted. Now we have evidence from a
recent comparative study of these presterilized
membrane filters that there are significant in-
creases in bacterial recovery rates for steam steri-
lized membrane filters compared to ethylene
oxide presterilized membranes (37). As a result,
one manufacturer that previously used ethylene
oxide sterilization is now reported to be using
gamma radiation for sterilizing membranes pack-
aged in single service envelopes while another
manufacturer has switched to steam sterilizing
their packs of 10 menbrane filters. For laboratories
that currently have supplies of ethylene oxide
sterilized membranes it may be desirable to submit
them to steam sterilization (121 C for 10 minutes
with rapid steam exhaust) to further flush out
latent toxicities. These membranes should then be
compared with other membranes from the same lot
of ethylene oxide treated membranes in a pure
culture recovery experiment. Possibly some resi-
dual toxic effect may still persist either from en-
trapped ethylene oxide or its reaction products.
Despite manufacturing claims to the contrary,
nitro-cellulose membrane filters do undergo some
change in their physical characteristics upon stor-
age in the laboratory for periods beyond 18
months. Upon aging, membrane filters may lose
their flexibility and break apart at pressure points
created during manipulation. During filtration, sur-
face warping often occurs making a complete con-
tact with the medium substrate impossible. The
solution to this problem is not to stock pile mem-
brane filter supplies beyond what is estimated to
be needed for a 12 month period.
Other Variabilities in the Membrane Filter Test
The quality of the membrane filter is not the
sole source of unreliable performance. Bacteria
retained on the MF surface may receive nutrients
from a broth saturated absorbent pad or from an
agar based medium. When a liquid culture medium
is preferred, the absorbent pad substrate material
must be of high quality paper fibers, uniformly
absorbent and free of sulfites, acids, or other sub-
stances that could inhibit bacterial growth. Recent
quality control testing of absorbent pads supplied
with membrane filters of various manufacturers,
has demonstrated a significant reduction in colony
counts and colony size associated with use of the
absorbent pad substrate in comparison to the same
medium prepared in a 1.5 percent agar base (35).
Until the absorbent paper quality improves, it will
be necessary for the laboratory to remove residual
toxic materials, such as bleaching agents, by pre-
soaking pads in distilled water held at 121 C for
15 minutes in the autoclave, decanting the rinse
water, and repackaging pads in large petri dishes
for sterilization at 121 C for 15 minutes, using a
rapid exhaust to quick dry the pads (22).
The alternate approach is to prepare all MF
broths with the addition of 1.5 percent agar.
However, it should be noted that these agar prepar-
ations must be carefully added to culture dishes so
as to create a smooth, moist surface, free of pock
marks caused by foam and rapid mixing of air
bubbles in the liquid agar preparation.
Media manufacturers have also contributed
significantly to the membrane filter dilemma
through variations in media quality (42). Formu-
lations of media currently available and recom-
mended in various reference sources contain a
variety of peptones, bile salts and dyes which are
not chemically pure compounds and thus subject
to variations in composition and performance. As
a result, medium sensitivity and selectivity will
vary unless manufacturers maintain an adequate
15
-------
quality control program to insure that these prod-
ucts meet the requirements for their intended use.
Although a quality check is made of these com-
mercial products (43). it appears to be inadequate.
Poor quality total coliform sheen development and
significant reductions in coliform recovery on
M-Endo medium have been observed by several
laboratories in recent years. Apparently the use of
poor grades of basic fuchsin and inadequate dye-
sulfite balance in the medium are responsible. Basic
fuchsin may differ in dye content, both from lot to
lot and from manufacturer to manufacturer,
making it essential to standardize the fuchsin-
sulfite proportion used each time a new lot of dye
is employed (22). Variations in intensity of the
blue color of fecal coliform colonies on M-FC
medium may be caused by residual acidity in
absorbent pads or membrane filters and also from
unsatisfactory lots of aniline blue used in the com-
mercial preparation of this medium. The intensity
and structure of bile salt crystals that precipitate
on fecal coliform colonies relates to the type of
bile salts complex incorporated in the medium.
Formulations of commercial media containing
sodium azide (M-Enterococcus, KF and PSE agars)
have an approximate shelf life of two years after
production, because of the deleterious effects
created by the slow decomposition of the azide
compound. For these reasons it is desirable for the
laboratory to establish a quality control analysis
on each new lot of medium purchased, comparing
it with a lot of the same medium known to be
satisfactory in terms of differentia.1 qualities and
sensitivity.
No discussion on the variability of any
laboratory technique can be complete without a
recognition of the human element. Reliable labora-
tory performance by every technician in the deter-
mination of bacterial quality of water requires the
continued application of knowledge, skill and
judgment. Only through a uniform application of
careful technic and rigid adherences to details
will the procedures yield the maximum benefits
of reliability and accuracy (32). Deviations in
laboratory procedures occur as a result of many
factors including attempted shortcuts, ignorance
of technical procedures, inexperience in new
methods, equipment failures, inadequate facili-
ties, technical carelessness, shifts of competent
personnel to other laboratory assignments, and
lack of interest in the phase of public health bac-
teriology (28). These sources of variability in the
membrane filter procedure can be held to a mini-
mum through a vigorously pursued certification
program at both the Federal and State levels.
This program can best be achieved through train-
ing in proper methodology, supported by periodic
laboratory evaluations and a bacteriological refer-
ence sample protocol to test laboratory pro-
ficiency and to reaffirm the continued production
of reliable data (44).
Special Membrane Filter Application Considerations
The unique properties of the membrane
filter and the compactness of basic apparatus
stimulated the development of several devices that
appear to have special potential for field use ap-
plication. One of these devices, the field monitor,
serves initially as the filtration chamber and then
as the culture package upon injection of con-
centrated modified media into a pad below the fil-
ter.The unit is then ready for incubation and sub-
sequent colony counting. Several independent
evaluations of the field monitoring concept for
total coliform recovery from polluted water indi-
cate that only 70 percent recovery of the known
bacterial density is being obtained. The remaining
organism loss occurs from : a) some bacterial by-
pass around the filtration area to the pad below the
membrane or direct to discharge through the bot-
tom port; and b) failure of some debilitated cells
to grow on the membrane and medium. In an at-
tempt to seal off the by-pass loss, the manufacturer
has added a hydrophobic substance to the outer
periphery of the filter. Unfortunately, one lot
tested in our laboratory had a much reduced
effective filtering area due to the non-wetting
agent. Inclusion of a consistent amount of normal
strength medium is dependent upon displacement
of the water entrapped in the pad with 1.3 times
normal strength ampouled medium filtered
through the field monitor following water sample
filtration. Vapor blockage and uneven flow
through will result in uncertain medium concen-
trations in the pad substrate, ultimately affecting
bacterial growth. Ampouled media has a limited
shelf life that must be recognized by the labora-
tory; 6 months for M-FC and 18 months for
M-Endo when stored in the dark, preferably at
refrigerated temperatures.
The bacteriological "dip stick" appears to
offer the ultimate yet achieved in test simplicity
at some sacrifice in flexibility. This device consists
of a sterile rectangular shaped membrane filter
positioned above a medium impregnated pad, both
being secured to a plastic frame which is inserted
into a mating plastic case. The basic principle of
operation is the controlled absorption of one ml
16
-------
of sample through the membrane to the medium
impregnated pad of critical thickness when the
"dip stick" is held in a water sample for approxi-
mately 30 seconds. The small volume of sample
makes the "dip stick" self limiting for total coli-
form analysis in potable water because the test
base-line is established as "less than one coliform
per 100 ml." Preliminary evaluation of the "dip
stick" for use as a standard plate count measure-
ment in potable water compared to the Standard
Methods procedure shows the method to result in
significantly lower bacterial counts, possibly be-
cause of the toxicity inherent in the gray-black
membrane filter and to the inadequately enriched
medium. The fecal coliform "dip stick" appears to
offer the field investigator a convenient prelimi-
nary screening tool for water pollution surveys.
However, before this procedure can be accepted
as producing definitive data for stream standards
and effluent qualities, it must be evaluated on a
variety of waters including acid mine drainage,
highly nutritive paper mill wastes and chlorinated
effluents. The critical unknown factors involve the
effect of source water chemistry on indicator
bacterial survival and suppression of false positive
reactions from non-indicator organisms. In this
respect, a critical need exists to determine the
adequacy of the "dip stick" fecal coliform pro-
cedure for monitoring chlorinated sewage. Con-
sistent reliability of the bacteriological "dip stick"
has yet to be determined and will relate to basic
qualities of the membrane filter, absorbent pad and
media stability during storage.
Special qualities of the membrane filter must
be recognized by the researcher involved in radio-
active detection of bacterial indicators. An adverse
effect of membrane filters on bacterial release of
carbon labeled CC>2 from radioactive tagged so-
dium formate was observed by Levin et al. (45).
Filtered cultures of bacteria invariably released
much less radioactive C02 per cell than did cul-
tures that were not filtered. Either toxic material
from the membrane filter inhibits metabolic activ-
ity of the bacteria and thus, the release of radio-
active C02 or mechanical rupture of some of the
bacterial cells by vacuum filtration effectively
reduces ^C02 release. Studies on the problem
suggest that although there may be some mechani-
cal damage to cells by impaction, the mere pre-
sence of the membrane filter in the growth medi-
um along with unfiltered cells was enough to re-
duce the 14C02 re'ease- The problem needs more
research investigation so as to control this factor in
radioactive carbon measurement in a rapid test for
bacterial indicators in water.
Membrane filters can be used as one of the
adsorbents for the concentration of virus from
water (46, 47). The clogging effects of turbidity
can be partially circumvented by use of membrane
filters of 293 mm diameter. However, sterilization
of this size membrane by autoclaving may result
in increased brittleness, making UV sterilization
more desirable. The choice of membrane filter
material is critical, with nitro-cellulose membrane
filters having a high adsorption affinity for virus
particles at a low pH (in the absence of interferring
substances) even to pore sizes 285 times the virus
diameter (48, 49). By contrast, cellulose triacetate
filters absorb few virus particles, even at pore
sizes no larger than three times the virus diameter.
Some laboratories filter sterilize tissue culture
media for cell line maintenance. This practice may
introduce a toxic contaminant to the medium from
detergents incorporated in the filter to promote
filtration efficiency and sterilization by autoclaving
(50). Therefore, it may be advisable to flush all
membrane filters in hot distilled water followed by
an ice cold saline rinse before use.
The recent development of a dialysis chamber
with membrane filter side walls offers an excellent
opportunity to study bacterial survival in a variety
of natural and artificial water environments (51).
Membrane filters used for this purpose must be
sturdy enough to withstand the buffering effects
of water currents and for this reason tear resistant
micro-web membranes with a nylon backing are
recommended. Substitute membranes with a rein-
forced backing must also be free of biodegradable
materials that will encourage the development of
microbial films over their surface, restricting the
in-flow of water and solutes to interact with the
bacterial suspension. Finally, it is of critical im-
portance that these membrane filters be non-selec-
tive in their passage of solutes which might alter
the chemistry of the water under investigation
and thus effect the bacterial survival patterns pro-
duced.
SUMMARY
There can be no doubt that the membrane
filter dilemma is real and needs urgent resolution.
Membrane filter and media manufacturers must
heed the outcries from anguished microbiologists
and reevaluate their product and quality control
programs. If not, these corporations risk the loss
of their multimillion dollar market through a grow-
ing wave of no-confidence in any membrane filter
procedures with subsequent abandonment of this
17
-------
heretofore approved laboratory tool in water
microbiology. This indeed would be a tragic turn
of events that need not happen.
The laboratory staff should not be influenced
by advertising claims of product excellence which
we pay for but may not be getting. Possibly 15
percent of the laboratory activity should be appor-
tioned to a quality control program on membrane
filter and media lots and a variety of routine
quality control procedures. The researcher is also
cautioned to include adequate controls in all exper-
iments involving membrane filters so that variables
introduced from this source will not adversely
affect the data interpretation.
REFERENCES
1. Bigelow, S., and A. Gamberling, Collodion
Membranes. J. Amer. Chem. Soc. 29:1576,
1907.
2. Pierce, H.F. Nitro-cellulose Membranes of
Graded Porosities. J. Bio. Chem. 75:795,
1927.
3. Ferry, J.D. Ultrafilter Membranes and Ultra-
filtration. Chem. Rev. 18:373, 1936.
4. Bechhold, H. Kolloid Studien mit der Filtera-
tiones Methode. Z. Physik. Chem. 60:257,
1907.
5. Bechhold, H. Durchlassigkeit von UltrafiItem.
Z. Physik. Chem. 64:328, 1908.
6. Bechhold, D. Ultrafiltration. Biochem Ztschr.
6:379, 1907.
7. Brown, W. II. On the Preparation of Collo-
dion Membranes of Differential Permeability.
Biochem. Jour. 9:591, 1915.
8. Brown, W. VI. Further Contributions to the
Technique of Preparing Membranes for
Dialysis. Biochem. Jour. 11:40, 1917.
9. Schoep, A. Uber ein Neues Ultrafilter.
Kolloid Zeit. 8:80, 1911.
10. Walpole, G. XXVI. Notes on the Collodion
Membrane for Ultrafiltration and Pressure
Dialysis. Biochem. Jour. 9:284, 1915.
11. Zsigmondy, R. and W. Beckman, Uber Neue
Filter. Ztschr. F. Anorg. Chemie 103:119,
1918.
12. Zsigmondy, R. and G. Jander, Die chemische
analyse mit Membranfilter. Z. Anal. Chem.
58:241, 1919.
13. Zisgmondy, R. U.S. Patent 1,421,341; June
27, 1922.
14. Zsigmondy, R. and W. Beckman, Uber Fein-
porige Filter und Neue Ultrafilter. Biochem.
Ztschr. 171:198, 1926.
15. Goetz, A. Fiat Final Report 1313. Joint
Intelligence Objectives Agency, U. S. Dept.
of Commerce, Wash., D.C. 1947.
16. Rasumov, A. S. Priamoi metod uchets bak-
terii v vode. Srav nenie ego s metodom
kokha, Mikrobiologiya (USSR), 1, 131,
1932.
17. Rasumov, A. S. Novye metody i put! Kach-
estvennogo i Kolichestvennogo izucheniia
mikroflory vody, Mikrobiologiya (USSR), 2,
346, 1933.
18. Eschenbrenner, vbn H., Keimfreie Filtration
in Apothekenbetrieb. Pharm. Montash, 14:
169, 1933.
19. Thomann, von J. Uber die Herstellung
Kiemgreier Injektianss-Losungen in Apothe-
kenbetrieb, Pharmaceutical Acta Helvetias,
9:2, 1934.
20. Mueller, G. Lactose-Fuchsin Plate for Detec-
tion of Coli in Drinking Water by Means of
Membrane Filters. Ztschr. Hygiene u. Infek-
tionskrankh 127,3/4, 187, 1947.
21. Mueller, G. Eine trinkwassergebundene Ruh-
repidemie. Zentralbl. Bait. Ab. I. Orig. 152:
133, 1947.
22. Clark, H.F., E.E. Geldreich, H. L. Jeter, and
P. W. Kabler, The Membrane Filter in Sani-
tary Bacteriology. Pub. Health Rep. 66:951,
1951.
23. Clark, H.F., P. W. Kabler, and E. E. Geldreich,
Advantages and Limitations of the Membrane
Filter Procedure, Water & Sewage Wks.
104:385, 1957.
24. Goetz, A. and N. Tsuneishi, Application of
Molecular Filter Membranes to Bacterio-
logical Analysis of Water, J. Amer. Water
Works Assn. 43:943, 1951.
25. Goetz, A. R.H. Gilman, and A.M. Rawn,
Application of Molecular Filter Membranes to
Specific Problems in Water Analysis. J. Amer.
Water Works Assn. 44:471, 1952.
26. Windle Taylor, E., N.P. Burman, and C.W.
Oliver, Use of the Membrane Filter in the
Bacteriological Examination of Water. J.
Applied Chem. (London) 3:233, 1953.
27. Windle Taylor, E., and N. P. Burman. The
Application of Membrane Filtration Techni-
ques to the Bacteriological Examination of
Water, J. Appl. Bact. 27:294, 1964.
28. Geldreich, E.E. Application of Bacterio-
logical Data in Potable Water Surveillance.
J. Amer. Water Works Assn. 63:225, 1971.
29. Clark, H.F., H.L. Jeter, E.E. Geldreich, and
P.W. Kabler. Domestic and European Mole-
cular Filter Membranes. J. Amer. Water
Works Assn. 44:1052, 1952.
18
-------
30. Caspar, A.J., and J.M. Leise. Inhibitory
Effect of Grid Imprints on Growth of Pasteu-
rella tularensis on Membrane Filter. J. Bact.
71:728, 1956.
31. Loesche, W.J. and S.S. Socransky. Defect
in Small Millipore Filters Disclosed by New
Technique for Isolating Oral Treponemes.
Science 138:139, 1962.
32. Geldreich, E.E., H.L. Jeter, and J.A. Winter.
Technical Considerations in Applying the
Membrane Filter Procedure, Health Lab.
Sci.4:113, 1967.
33. Levin, G.V., V.L. Strauss, and W.C. Hess.
Rapid Coliform Organism Determination
with C14. J. Water Poll. Contr. Fed. 33:
1021, 1961.
34. Brown, O.R. Inhibition of Escherichia coll
on Cellulose Acetate Membrane Filters.
Microbios 7:235, 1973.
35. Presswood, W.C., and L.R. Brown. Com-
parison of Gelman and Millipore Membrane
Filters for Enumerating Fecal Coliform
Bacteria. Appl. Microbiol. 26:336, 1973.
36. Hufham, J.B. Evaluating the Membrane
Fecal Coliform Test by Using Escherichia coll
as the Indicator Organisms. Appl. Microbiol.
27:771, 1974.
37. Dutka, B.J., M.J. Jackson, and J.B. Bell.
Comparison of Autoclave and Ethylene
Oxide-Sterilized Membrane Filters Used in
Water Quality Studies. Appl. Microbiol.
28:474, 1974.
38. Schaeffer, D.J., M.C. Long, and K.G. Janar-
dan. Statistical Analysis of the Recovery of
Coliform Organisms on Gelman and Millipore
Membrane Filters. Appl. Microbiol. 28:
605, 1974.
39. Alico, R.K. and C.A. Palenchar. Staphylo-
coccus aureus Recoveries on Various Brands
of Membrane Filters. Health Lab. Sci. 12:
341-346, 1975.
40. Kabler, P.W., and H.F. Clark. The Use of
Differential Media with the Membrane Fil-
ter. Amer. J. Pub. Health 42:390, 1952.
41. Burman, N.P. Developments in Membrane
Filtration Techniques. 1. Coliform Counts
on MacConkey Broth. Proc. Soc. Water Treat.
Exam. 9:60, 1960.
42. Wolochow, H. The Membrane Filter Tech-
nique for Estimating Numbers. Appl. Micro-
biol. 6:201, 1958.
43. Vera, H.D. Quality Control of Diagnostic
Microbiology. Health Lab. Sci. 8:176, 1971.
44. Geldreich, E.E. A Total System Approach
to Monitoring the Bacteriological Quality of
Water. 2nd Annual Water Tech. Conf. AWWA.
Dec., 1974.
45. Levin, G.V., V.L. Strauss, and W.C. Hess.
Rapid Coliform Organism Determination
with C14. J. Water Poll. Contr. Fed. 33:
1021, 1961.
46. Oliver, D.O. Factors in Membrane Filtra-
tion of Enteroviruses. Appl. Microbiol.
13:417, 1965.
47. Wallis, C.M., M. Henderson, and J.L. Melnick.
Enterovirus Concentration on Cellulose Mem-
branes. Appl. Microbiol. 23:476, 1972.
48. Oliver, D.O. Factors in the Membrane Fil-
tration of Enteroviruses. Appl. Microbiol.
13:417, 1965.
49. Oliver, D.O. Virus Interactions with Mem-
brane Filters. Biotechnol. and Bioengr.
10:877, 1968.
50. Cahn, R.D. Detergents in Membrane Filters.
Science 152:195, 1967.
51. McFeters, G.A., and D.G. Stuart. Survival
of Coliform Bacteria in Natural Waters:
Field and Laboratory Studies with Membrane
Filter Chambers. Appl. Microbiol. 24:805,
1972.
19
-------
QUALITY CONTROL OF MEMBRANE FILTER MEDIA
David A. Power, Ph. D.
Manager of Technical Services
BioQuest, Division of Becton, Dickinson and Company
Cockeysville, Md. 21030
ABSTRACT
The manufacturers of dehydrated membrane
filter media provide quality control, lot-to-lot uni-
formity and relative stability of their products. The
quality control procedures test the media compon-
ents and the dehydrated products for suitable
physical and chemical characteristics, microbial
contamination and growth support. In addition,
the dehydrated products are subject to perfor-
mance tests based on the use of the specific media.
However, the users must maintain quality control
after receiving the media. Quality control proce-
dures in the laboratory include age of product,
storage conditions, accurate weighing of dehy-
drated media, good quality of distilled water,
clean utensils, complete mixing and solution,
controlled heating, accurate pH determinations,
approved supplements only, and appropriate
checking with stabilized test cultures. Tests that
the user should perform on the completed media
include observation of appearance, pH, sterility,
and membrane filter performance with selected
test organisms.
INTRODUCTION
The development of membrane filter techni-
ques for the isolation, enumeration and differen-
tiation of microorganisms in water, sewage, milk,
foods, air, solutions, specimens, etc., created a
need for culture media especially suited for use
with these techniques.
Kabler and Clark (5) reported that formulas
of most media for conventional use must be modi-
fied before the best results are obtained with the
membrane. This modification may be in the form
of changed quantities of ingredients or the sub-
stitution of nutrients. These authors noted that in
many instances there is no correlation between
results obtained with an agar medium using con-
ventional methods and the same formula without
agar when used on a membrane.
This need for specialized media has resulted
in the production by commercial dehydrated
media manufacturers of a broad line of media for
membrane filter techniques, which are denoted by
a "M" prefix to the product name. Dehydrated
membrane filter media offer the laboratory the
same advantages as standard dehydrated media
relative stability, lot-to-lot uniformity and the
quality control "built into" dehydrated products
by commercial media manufacturers.
QUALITY CONTROL BY THE
MANUFACTURERS
As with other dehydrated media, membrane
filter media are subjected to quality control pro-
cedures at least twice before reaching the user.
Initially, the raw materials incorporated into
dehydrated media are tested before inclusion in
various media formulas. Peptones are examined
for physical appearance of the powder, pH, clarity,
and color, and for microbial contamination. Pep-
tones are also subjected to growth support tests in
which a single peptone serves basically as the only
nutrient. Agar is tested for gel strength, clarity by
nephelometry, color by colorimetry, gelation and
melting temperature, pH, solution time, and visual
appearance. Only satisfactory ingredients are
employed in the manufacture of the various media.
The second check is performed on the dehy-
drated product. Dehydrated media are subjected
to the same types of tests described for ingredients,
such as appearance, pH, clarity, and color. Perfor-
mance tests with microbiological cultures are based
on the end use of the medium. For selective media,
it is necessary to demonstrate satisfactory growth
of desired species and inhibition of undesired
species, recognizing that a selective medium repre-
sents a compromise in that the selective agent may
20
-------
somewhat inhibit strains of desired species as well
as undesired species. The performance of differen-
tial media is assessed by the use of appropriate
species, including those producing positive and
negative reactions.
For example, the BBL laboratory evaluates
M-Coliform Broth, the BBL equivalent of M-Endo
Broth, by determining that the broth colorimetric
reading is satisfactory, and that the pH after heat-
ing and cooling to room temperature is 7.2 ± 0.2.
Growth support tests include the use of strains of
Enterobacter aerogenes, Escherichia coli, Pseudo-
monas aeruginosa. Salmonella typhimurium, and
Shigella sonnei. The medium is tested for the
ability to support growth following straight inocu-
lation of 10"' dilutions of test organisms, and for
the recovery and differentiation of E. coli in the
standard procedure by adding one ml of a dilution
containing 30 to 300 organisms to 50 ml of water,
followed by filtration through a membrane filter
subsequently incubated on a medium-soaked pad
for 24 hours at 35 C. Growth must be satisfactory
and reactions correct. Streptococcus fecalis is also
included in the performance evaluation, and
growth must be inhibited.
The use of stabilized, freeze-dried cultures
permit results to be predictable and reproducible
on standard or reference lots of media.
QUALITY CONTROL BY THE USER
The need for an internal quality control
program in the microbiology laboratory has been
documented in numerous papers and has been
legislated for laboratories involved in the interstate
practice of laboratory medicine through the
Clinical Laboratories Improvement Act of 1967
(3). Abuses resulting in poor performance have not
been uncommon, despite the quality control
procedures performed by commercial manufac-
turers on dehydrated media and media supple-
ments before their release and directions on labels
and on product or methodology manuals for re-
constitution and preparation of final media and
their handling (1, 2,6, 8).
Some of the factors that may contribute to
the preparation and use of unsatisfactory finished
media from dehydrated materials when little or no
attention is paid to the quality of the completed
media include the following (9):
1. Incorrect weighing of dry material,
through human error or use of a faulty
balance.
2. Use of dry material taken from previous-
ly opened bottles, which may have
deteriorated from exposure to heat,
moisture, oxidation, or other environ-
mental factors. The quantities in which
media are purchased should be regulated
by the rate of usage. Ideally, bottles
of dehydrated media, once opened,
should be used within a few weeks. The
purchase of media in 1/4 Ib bottles or
in sealed preweighed envelopes is en-
couraged.
3. Incorrect measurement of water and use
of tap water or water from a malfunc-
tioning still or deionizing resin column.
Water should generally meet the re-
quirements of the United States Pharma-
copeia (USP) XVIII (10) for purified
water or be of proven microbiological
quality.
4. Use of unclean containers or glassware,
especially those contaminated with de-
tergent or other chemicals.
5. Incomplete mixing or incomplete solu-
tion resulting in failure to prepare a
homogeneous medium. With agar media,
this may even product stiff medium in
some plates and soft medium in others.
6. Overheating occurring during prepara-
tion and sterilization, or resulting from
holding too long in the molten state be-
fore dispensing into plates, tubes or bot-
tles, can result in the loss of productivity
through hydrolysis of agar, carmelization
of carbohydrates, lowering of pH, in-
crease in inhibitory action, loss of dye
content in selective or differential
media, and formation of precipitates.
7. Improper determination of pH, resulting
in the addition of too much acid or
alkali. The pH of a medium should be
determined electrometrically; the elec-
trodes should be in contact with the
solidified agar medium, which may be
removed from a plate or tube and placed
in a beaker.
8. Improper addition or incorporation of
unsatisfactory supplements or enrich-
ments, or addition of supplements at
the wrong temperature, possibly causing
alteration of the supplements if the
21
-------
temperature is too high, or gelation of
media before proper mixing if too cold.
9. Failure of the laboratory to subject
samples of finished media to quality
control procedures with stabilized test
cultures before the media are used.
Regardless of the amount of quality con-
trol built into a product, users should be
aware of strain variations which influ-
ence recovery and growth and of the
purpose for which each medium was
designed.
STORAGE OF MEDIA
Once prepared, special attention must be paid
to proper storage. The selective and differential
membrane filter media are more susceptible to
deterioration than similar routine culture media.
For this reason, we recommend that broth media,
prepared in sealed tubes or bottles for later use, be
stored in a refrigerator maintained at 2 to 8 C and
used within a couple of weeks. It is generally re-
commended that M-Coliform Broth (M-Endo
Broth) and M-FC Broth be used within four to five
days (4, 7). Plates of agar media should be pre-
pared and used promptly, or be stored wrapped in
foil or plastic to limit water loss and used within
two weeks. Containers of prepared membrane
filter media must be protected from light.
RECOMMENDED QUALITY CONTROL
CHECKS BY THE USER
Checks that the user can perform on the
finished media include:
1.
2.
3.
Appearance if the medium is off-color,
or there is a change in appearance during
storage or signs of drying, contamina-
tion or deterioration, the medium
should be discarded.
pH measured electrometrically at
room temperature. The pH should be
within ± 0.2 of that stated on the label.
Sterility testing by incubating repre-
sentative samples, for two or more days.
Such sample tubes or plates should be
discarded and not used for later culture
work.
4. Performance as a minimal test, we
recommend the straight inoculation of
10~1 dilutions of test organisms. As a
test of the complete system, we recom-
mend the filtration through a membrane
filter of a water sample inoculated with
a diluted suspension of an appropriate
organism, followed by handling and
incubation according to the procedure
established for that medium. This test is
a quality control check for the entire
system.
If a problem is encountered, users are en-
couraged to contact the manufacturer. The lot
number of the product and dates on which the
medium was received and first opened should
accompany the observations.
REFERENCES
1. Blair, E.B., Media, Test Procedures and Chem-
ical Reagents. In H.L. Bodily, E.L. Updyke,
and J.O. Mason (eds.), Diagnostic Procedures
for Bacterial, Mycotic and Parasitic Infec-
tions, pp. 791-857, American Public Health
Association, New York, 1970.
2. Difco Manual, 9th ed. 1953. Supplementary
Literature. Difco Laboratories, Detroit, 1968.
3. Federal Register. Clinical Laboratories Im-
provement Act of 1967 Notice of Effective
Date. 33. No. 253. U.S. Government Printing
Office, Washington, D.C., 1968.
4. Standard Methods for the Examination of
Dairy Products, 13th ed., American Public
Health Association, Washington, D.C., 1972.
5. Kabler, P.W., and H.F. Clark. The Use of Dif-
ferential Media with the Membrane Filter.
Am. J. Public Health 42:390-392, 1952.
6. Rohde, P.A. (ed.). 1968. BBL Manual of
Products and Laboratory Procedures, 5th ed.,
BioQuest, Division of Becton, Dickinson and
Co., Cockeysville, Mr., 1968.
7. Standard Methods for the Examination of
Water and Wastewater, 13th ed., American
Public Health Association, Washington, D.C.
1971.
8. Vera, H.D., and M. Dumoff. Culture Media.
In E.H. Lennette, E.H. Spaulding, and J.P.
Truant (eds.), Manual of Clinical Microbio-
logy 2nd ed., pp. 881-929. American Society
of Microbiology, Washington, D.C., 1974.
9. Vera, H.D. Quality Control in Diagnostic
Microbiology. Health Lab. Sci. 8:176-189,
1971.
10. United States Pharmacopoeia. 18th ed. Mack
Publishing Co., Easton, Pa., 1970.
22
-------
DISCUSSION
Geldreich: Did you say that in the development
of M-Endo media BBL uses a color-
metric method to determine the
proper amount of basic fuchsin?
Power: There is some color determination. I
think it can vary with the lot of dye
involved, and there would be a range.
Geldreich: The problem that I am concerned
about is that many or all of us prob-
ably know that the actual dye con-
tents of these products vary consid-
erably. Dyes such as basic fuchsin,
aniline blue, brillant green, etc. are
not chemically defined materials.
Sometimes the microbiologist is work-
ing with a trade brand of material
which was originally intended for
dyeing clothes and not for preparing
media. Therefore, basic fuchsin, as an
example, may have a dye content
anywhere from 88% to 99%.
Power: That would be taken into considera-
tion in the formulation of the actual
dye.
Geldreich: This is our concern. Years ago we
suggested that the manufacturers per-
form a biological dye titration when
they made up M-Endo media. This
titration is done by holding the
amount of sodium sulfite constant
and running each new lot of basic
fuchsin against it. Each batch be-
comes an individual lot of media and
is checked for recovery of the organ-
isms with good sheen and no evidence
of toxicity. I don't think the manu-
facturers do this. We often find that
these products have poor sheen and
sometimes poor recovery. I think it
is because the sulfite-basic fuchsin is
not in the proper proportion.
Lane: I am quite sure that BioQuest titrates
the dye content in their MF media
just as Difco does. You are absolutely
right that there is variation in the
dyes, and one of the worst problems
has been with the aniline blue. Basic
fuchsin has not been too bad. The
only way that the dye content can be
determined satisfactorily is by per-
formance test. If the test organisms
(coliforms in this case) don't yield
typical colonial morphology or color,
the dye content is off. The manu-
facturers of dyes seem to have lost the
control that they had years ago. It
is very difficult to get a satisfactory
batch of basic fushion, aniline blue or
brilliant green. It is up to the manu-
facturer to standardize the media
so that the morphology and total
counts are satisfactory. We do titrate.
Power: We do have the standards. If you ex-
perience a problem or if you see a
variation, I wish you would let us
know.
Seidenberg: You made a comment that has me in
a dilemma; you said the M-FC and the
M-Endo agar plates can be kept for
4 to 5 days.
Power: Yes.
Seidenberg: We in EPA are very careful to discard
M-Endo plates after 48 hours, because
we have found color changes.
Power: I notice that our manual and APHA
manuals allow 4 to 5 days.
Seidenberg: I believe that Standard Methods for
the Examination of Water and Waste-
water states 48 hours.
Power: I believe that the holding times in the
Standard Methods for Water and
Dairy Products may differ by one
day or so.
Geldreich: Standard Methods allows the worker
to hold M-Endo and M-FC plates for
use during that work week. However,
I think some of the laboratories
have found that these media are light
sensitive and if they don't store them
in the dark there is a real problem.
One could cover half a plate, leave it
on a laboratory bench about two
hours, take that cover away and find
two shades of media, have resulted.
Power: I think it is best to use the media on
the day of preparation.
Geldreich: If possible, that is great. Many work-
ers ask how long they can keep the
media. The current Standard Methods
23
-------
recommends a work week. This re-
commendation may be changed in the
future.
Brodsky: I question the use of pure cultures as
the only quality control that you use
on your media. We are not accustom-
ed to receiving pure cultures in a
water sample. The relationship be-
tween the various organisms in their
natural environment is negated by the
use of pure cultures only. Perhaps
pure cultures are used just because
they are expedient.
Power: The use of the pure culture gives us a
base for the comparison from lot to
lot. Perhaps out of this meeting you
will come up with some improved
system as you indicated. Out of all
the possibilities one might select, I
frankly don't know which is most
practical. We feel that we have organ-
isms on which we have a history and
back over the years we have used
these cultures to test many lots. We
are trying to give the user a product
now that relates to previous lots, to
keep them as constant as possible.
One of the purposes of the meeting is
to talk about the membranes, pads
and media and to come up with final
specifications.
Lane: The manufacturers of media must
have a base line and the only base line
we can presently use is pure cultures
of Aerogenes, E. coli, or Salmonella.
However, the final assay, the final
approval, and this again is true for
BioQuest as well as Difco, is the use
of a natural specimen. One can titrate
a selective medium using a stock
culture, and can get certain results,
such as selectivity. We recognize that
when we use natural specimens such
as natural water, polluted water,
sewage and stool specimens, we get
entirely different results. The state-
ment that the manufacturer does not
take into consideration the different
results that could be obtained with
natural specimens may not be correct.
There is one point I would like to
raise; this is a problem that we en-
countered recently. We are taking part
in the revision of Standard Methods
for the Microbiological Examination
of Dairy Products. We were using a
buffered diluent as in water micro-
biology. We found that different lots
of phosphate gave different total
count results. Not only the different
lots of phosphates but the time that
organisms survived in these different
lots varied. The organism will survive
for an hour in one lot of phosphate
but in two hours the count goes
down. The count for another lot
might be satisfactory at zero time but
the count may go down after being
held for 15 minutes. We also en-
countered the problem of the varia-
tion in distilled water when preparing
a buffer from a good lot of phosphate.
This is a problem that nobody has
mentioned, and I think that you must
consider this.
Bordner: In water analyses we are aware of the
potential toxicity of phosphate buf-
fers and the importance of good
quality distilled water. Standard Meth-
ods allows an alternate dilution/rinse
water, 0.1% peptone. The next edition
will recommend the addition of mag-
nesium sulfate to the phosphate buf-
fer as an added protective against
toxicity. There is also a test for the
suitability of distilled water pro-
vided in Standard Methods.
I have a two-fold question. You have
stated, Aaron, that you periodically
use natural samples. Could you give us
any estimate on your sampling fre-
quency or upon what percentage of
the media sampling per lot is
based. Secondly, do you gentlemen
have any suggestions on how we as
water microbiologists could work
more closely with the media manu-
facturers as I know you do with other
groups. Could it be done with ASTM
or other organizations and could it be
a continuing relationship?
Power: That is why we are here, partly to tell
you our story and mostly to listen to
what you have to say, because ob-
viously you have problems. If we can
be of help, we certainly intend to
cooperate. Until this week, I haven't
24
-------
been totally aware of all the problems
involved and I certainly am willing to
take back to my company anything
that I hear. We are very willing to
work with you and try to come to
some solutions.
Lane: We check every lot of the M-Endo
broth MF; for example, every lot is
checked with river water. The other
membrane media are not checked as
frequently. We spot check them;
perhaps 1 out of 4 or 5 lots. The basic
media are checked every time we
make a new batch. Concerning other
media, for example, media for isola-
tion of Neisseria gonorrhea, all media
are assayed with clinical specimens.
All media for isolation of Salmonella
and Shi gel la are checked with stool
specimens which are seeded with
Salmonella and Shigella. Unfortun-
ately you can't obtain natural speci-
mens containing these pathogens so
we seed them with different concen-
trations. I am sure that we do the
same with the medium in which you
are primarily interested, the M-Endo
broth MF or the M coliform medium.
We do check them with river water
every time a batch is prepared.
Brezenski: I want to get back to the statement
concerning the variation in dye con-
tent, because I feel this is very im-
portant. Every time a laboratory gets
poor results there is a tendency to
say there was a problem with the
medium. The microbiologist talks to
the manufacturer and he says, "well
we have a problem with the raw in-
gredients which we don't have that
much control over." I am wondering
whether the manufacturer, for ex-
ample Difco or BioQuest, specifies
the amount of the active dye ingred-
ient that is supposed to be in aniline
blue and basic fuchsin and what the
percentage of inactive ingredients
should be. These specifications could
then be established to conform to
the quality control. For example, in
the FITC dye used in fluorescence
antibody CDC sets up a specification;
they must have 80 to 90 percent ac-
tive dye and only about 10 percent
inactive ingredients. If the dye doesn't
meet their specifications, they do not
buy it. I am wondering whether Difco
or BBL have specifications like this;
because if we don't, this is where we
should start.
Power: We do and on that particular product
our specification exceeds the CDC
specification.
Brezenski: The statement was made that varia-
tion is a problem; where does the
variation occur?
Lane: There is variation in natural dye
content. When you are comparing a
fluorescent dye with an aniline blue
and a basic fuchsin you are comparing
almost a pure chemical with a mix-
ture. The only way that you can test
and approve or reject a batch of basic
fuchsin or a batch of aniline blue is
by performance. You cannot set
specific criteria for actual dye con-
tent. You will have a variation, you
will have a range of actual dye con-
tent, but the remaining components
of a batch of basic fuchsin are really
not specific materials. They are not
even identified, so specifications can't
be set. The only specification that
can be set for any culture medium is
its performance. Does the medium do
what it was designed to do? Does this
coliform medium, the MF medium
that you are using to isolate the maxi-
mum number of coliforms present in
that sample, yield colonies which are
characteristic? This is what you are
after.
As you pointed out, there are many
variables in the dyes. Let me use pep-
tones as another example. The pep-
tone is not a specific substance. A
peptone is either an enzymatic digest
or acid hydrolysate of meat and
casein. The method of production
may be the same and the control of
production for dyes as well as pep-
tones may be the same; but the final
criterion of a good peptone or a good
dye is how does it perform in the
medium.
25
-------
STATISTICAL INTERPRETATION OF MEMBRANE FILTER BACTERIA COUNTS
K. J. Sladek, C. F. Frith, and R. A. Cotton
Millipore Corporation, Bedford, Massachusetts
ABSTRACT
Statistical design and interpretation of experi-
ments for enumerating bacteria by membrane
filters are discussed. To increase precision of mean
counts, large numbers of replicates can be used. In
fecal coliform tests, large numbers of replicates
can be achieved using a water sample stabilized by
dilution in phosphate buffered peptone. Another
way of increasing precision is to reduce the scatter
among replicates. However, the random fluctuations
of bacterial density, even in a well mixed sample,
place a lower limit on scatter. This limit is predict-
able as a theoretical minimum standard deviation,
which is useful as a yardstick in comparing with
experimentally determined standard deviations.
Experimental designs should provide for randomi-
zation of procedural variables so that unexpected
variables do not bias results.
INTRODUCTION
Comparing different methods for enumerating
bacteria is a familiar endeavor in microbiology
laboratories. These evaluations include comparing
MF techniques with other standards such as pour
and streak plates, comparing variations in proce-
dure, comparing membranes of different manu-
facture, and evaluating new media in comparison
to available ones. In addition, in production and
quality control operations, membrane manufactur-
ers are involved continuously in evaluating im-
proved membrane manufacturing and processing
methods in comparison to existing ones and in
qualifying production lots of membranes against
standards.
The present paper discusses experiments in
which the same water source or culture is used to
test different membranes, media or procedures. An
example is to evaluate effects of sterilization on
membrane filters. You may have had the sad exper-
ience, as we have on some occasions, of reviewing
pages of data only to find that each reviewer
reaches different conclusions from the study! We
will address here the formidable problem of carry-
ing out an experiment leading to conclusions upon
which all concerned can agree.
The factors contributing to a successful ex-
periment can be categorized as statistical factors,
experimental design principles, and bacteriological
considerations. Statistical factors include decisions
on the number of replicates to be used, the analy-
sis of the scatter in counts expressed by the stand-
ard deviation of replicate counts and calculations
of confidence limits. Experimental design must
provide for isolation of the differences under
investigation from extraneous effects; random
selection of samples to be tested is important
here. The bacteriological considerations include
choice of the water source or the culture to be
used, establishment of a detailed procedure, and
selection of a useful control method. This paper is
concerned mainly with statistical and experimental
design factors; some of the bacteriological consid-
erations in fecal coliform tests will be discussed
briefly.
Statistical Factors
The precision of a count determined by aver-
aging n replicate plates can be expressed by confi-
dence limits, which give a range in which the true
mean will lie. For example, the "95% confidence
limits" are given by ± 1.96 a/^n, where ^/ is the
standard deviation of the population of individual
measurements. The experimental mean has a prob-
ability of 95% of being within ± 1.96 a/Vn of the
true mean. To increase precision it is necessary to
reduce a or to increase n. Increasing n requires an
operating procedure which is identical for each
replicate and a water sample which is stable for the
life of the experiment. An example of sample
stabilization is given next. The scatter in replicate
counts, which is measured by a, is considered
afterwards.
26
-------
Fecal coliform counts for sewage samples
diluted three different ways are given in Figure 1.
Each plate was prepared by spreading a 0.1 ml
aliquot of diluted sample onto M-FC agar. The
lower graph, representing sterile water and phos-
phate buffer diluents, shows that the sample is un-
stable even for periods as short as 15 minutes. Use
of 0.1% buffered peptone (prepared by adding 1 g
peptone to 1 liter of phosphate buffer) stabilizes
the sample for a much longer period. "Example 1"
of the buffered peptone data exhibits a stable
count throughout the entire two hour test period.
In "Example 2", however, the counts begin to fall
outside the 95% confidence limits at a dilution age
of 60 minutes. Even with a 60 minute limit, how-
ever, peptone stabilized samples can be used in
running far more replicates than could be used
with water or phosphate buffer dilutents.
As illustrated earlier, the confidence limits
around an experimental mean depend not only on
the number of replicates, but also on the standard
deviation of individual measurements. An experi-
mental value, s, of the standard deviation can be
calculated from each set of replicates. However,
there is a particular source of scatter in membrane
filter counts which is predictable, and it is possible
to generate a theoretical value of a, against which
experimental values can be compared. The pre-
dictable source of scatter is the fluctuation in den-
sity of bacteria throughout a water sample.
At best, bacteria in a carefully mixed suspen-
sion will be distributed randomly through the sur-
rounding medium. At worst, they may be associ-
ated with particles, adhering in clumps, or con-
centrated at the walls of the container. If they are
O
o
o
o
o
60 1
1 "
50 9
40 1
30 p
60
40 1
30 1
O
20 1
A
30 1
20 A
:!
0
0.1% BUFFERED PEPTONE - EXAMPLE 1 I
0 0 c
e
n n R u *~
U 0 8 0 ° MEAN 1
-J
1
9.1% BUFFERED PEPTONE" EXAMPLE 2 °
C
o I
o ° o o I
9 o c
o © o c
/>,
° 1
A PHOSPHATE BUFFER
STERILE WATER
A A
A A ^
15 30 45 60 75 90 105 12
195%
CONFI-
DENCE
LIMITS
FOR
INDIVI-
DUAL
COUNTS
AGE OF DILUTION, minutes
Figure 1. Effect of Diluent on Stability of Water Sample.
27
-------
randomly distributed, samples of identical volume
will not contain exactly the same number of
bacteria.
The distribution of sample counts of parti-
cles taken from a random distribution in a sus-
pending medium is called a Poisson distribution
(2, 3). This distribution has a standard deviation
equal to the square root of its mean. For the range
of counts of usual interest in bacteriology, the
Poisson distribution can be closely approximated
by a normal distribution, which simplifies calcula-
tions considerably. Using this simplification, if
the mean count per aliquot of sample is 100
bacteria, the standard deviation ap due to random
distribution of bacteria through the liquid is>/100
or 10. The 95% confidence limits on individual
counts are ± 1.96 OR or ± 19.6. In other words, if
the average count is 100, we can expect 95% of the
measured counts to fall between 80 and 120, if the
random distribution in the water sample is the only
source of scatter in the data.
however, other sources of error arising from varia-
tions in experimental technique. These can appear
either as bias in the observed mean, X, or as scatter
in the data, an increase in s. In the following
section, we will discuss how an experiment can be
designed to prevent an unwanted bias from enter-
ing into the results.
Experimental Design
An example design is a study of the effect of
ethylene oxide sterilization on mixed cellulose
ester membranes. The purpose of the experiment
is to isolate the effect of sterilization and hence it
is necessary to eliminate effects of all other vari-
ables, known and unknown. The idea is to take a
group of identical membranes, to sterilize half, and
to compare these with the other half, which are
not sterilized. The isolated variable is sterilization,
but what about possible unknown variables: are
the membranes really identical, is the experi-
menter's technique identical for each plate, is each
Case
Number of
Replicates, n
Experimental
Mean, X
Experimental
Standard
Deviation, s
Predicted
Standard
Deviation, OR
Streak Plates
18
76.2
14.6
8.7
Membranes,
Unsterilized
Membranes, Ethylene
Oxide Sterilized
18
18
119.4
123.9
17.3
21.8
10.9
11.2
In summary, with the best case, random dis-
tribution of bacteria in the sample, we can expect
a fairly large but predictable standard deviation.
One example of using ap is given in the upper two
graphs of Figure I.JThe value of OR is the square
root of the mean X, found by averaging all the
counts in the experiment. Then the predicted 95%
confidence limits for individual counts are ± 1.96
\/X. These limits are represented on the figure as
dotted reference lines, showing when the experi-
mental points are within acceptable limits.
Another example is to compare an experi-
mental standard deviation, s, with the theoretical
minimum standard deviation, OR. This will be done
in a later section.
So far only one contribution to the scatter in
an experiment has been considered. There are,
water aliquot identical, is each plate incubated
identically, is each counted the same way . . .?
Although the bacteriologist exercises the strictest
control over all of these factors, it is still possible
for one of these or an entirely unknown variable
to intrude into the experiment. To prevent un-
wanted bias, it is safest to assume that unknown
variables are present, and to use randomization
techniques to eliminate their effects.
For the sterilization study 36 membranes
were chosen and coded for identification. Half
were selected by random number techniques and
were gas sterilized. Sterilization details are given in
our paper, "Optimum Membrane Structures for
Growth of Fecal Coliform Organisms" (1). Then
each of the 36 membranes was selected in random
order for testing. After incubation, plates were
again randomized before counting.
28
-------
The reason for all these randomizations was
to "mix up" the effect of any undesired variables.
Suppose, for example, that the technician who
counted the plates became fatigued and biased the
plates counted last towards lower values. If all the
unsterilized membranes had been counted last,
their lower counts would have been attributed
erroneously to the isolated variable under study.
Since, however, the plates were selected at random
for counting, the (hypothetical) counting bias ap-
peared scattered randomly throughout the results.
In summary, one can appreciate the value of
random sampling by assuming the worst: namely,
that in spite of your best efforts, undesired vari-
ables are present. By randomizing at each stage of
the experiment, the undesired effects are scattered
throughout the run so that they do not bias the
effect of the isolated variable. To put this another
way, the signal should be due to the variable under
study and all other factors should appear only as
noise.
Results of this sterilization study are given in
Table 1. It is useful to compare experimental
standard deviations with OR, as suggested earlier.
Values from Table 1 are summarized below.
Here, the three experimental standard devia-
tions exceed the values predicted for random dis-
TABLE 1. EFFECT OF ETHYLENE OXIDE STERILIZATION ON FECAL COLIFORM
COUNT
Replicate
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Mean,
X
Streak Plate
58
76
68
74
89
120
65
74
80
88
56
81
89
66
65
75
71
77
76.2
Membrane,
Unsterilized
127
128
129
85
115
117
143
142
135
114
129
102
135
125
112
126
102
84
119.4
Membrane, Ethylene
Oxide Sterilized
152
135
139
100
93
121
109
87
143
90
124
118
144
134
120
165
127
130
123.9
Experimental
Standard Deviations,*a
14.6
17.3
21.8
'Computed from s2 = 2 (X-X)2 / (n-1)
29
-------
tribution of bacteria in the sample. This indicates
that additional factors, such as the experimental
variables discussed above, have contributed to s,
and that randomizing at each stage of the experi-
ment was a necessary precaution.
Some contrasting data on standard deviations
are given in Table 2, which is a study of steriliza-
tion by three methods (1). Comparing s with OR,
here, we see that two s-values agree well with ap,
two are considerably lower than OR and one s is
considerably higher than /n
95% Confidence
Limits on X*
Unsterilized Membranes
Autoclaved Membranes
Ethylene Oxide Sterilized
Membranes
Irradiated Membranes
Streak Plates
5
5
5
5
10
98.8
107.5
103.0
94.3
82.0
4.3
10.5
6.0
9.4
14.2
9.9
10.4
10.2
9.7
9.1
±8.7
±9.1
±8.9
±8.5
±5.6
30
-------
tion is again useful. To randomize this variable it
would be necessary to list all the water sources of
importance for this type of test and to select at
random a group for testing. Then at the end of the
program results of all experiments would be
averaged together to eliminate any bias due to
water source effects. This kind of program is in-
deed a massive undertaking and requires the co-
operation of several laboratories over a long period
of time. We should thus regard the sterilization
studies presented here, using two sewage sources,
as only a beginning. A fine example of an extensive
study involving three laboratories and a great vari-
ety of sources is given in "An Improved Membrane
Filter Method for Fecal Coliform Analysis" later in
this Symposium (4).
A second important bacteriological considera-
tion is the selection of a control method. The
spread plate was the non-membrane control used in
the investigations of Table 1 and 2. The spread
plate itself, however, is subject to a whole set of
procedural details, one of which will be described
here. Table 3 presents fecal coliform counts for
0.1 ml samples of diluted sewage on two thick-
nesses of M-FC agar. The same experiment was
repeated on seven occasions using fresh sewage
diluted in phosphate buffered peptone. Attempts
to use agar thicker than 0.59 cm were not success-
full due to the presence of spreaders. Looking at
the overall means, there is clearly a significant
increase in count with agar thickness.
The agar thickness effect illustrates that the
spread plate control does not necessarily provide a
complete measure of the number of viable bacteria.
Although the spread plate does not provide an ulti-
mate standard, we believe that it is important to
include some non-membrane standard in mem-
brane evaluations.
Table 3 also shows a possible water source
effect. The first six runs show a statistically signifi-
cant agar thickness effect, while the last one does
not. If only the 12-27 run had been performed, we
would have missed the effect altogether!
CONCLUSION
In conclusion, some of the factors entering
into successful experiments evaluating membrane
filters have been discussed. To provide narrow con-
fidence limits on a mean count, it is useful to
stabilize the water sample so that large numbers of
replicates can be employed. The scatter in replicate
counts, which also affects the confidence limits,
cannot be reduced below a minimum which is
characteristic of the random distribution of
bacteria in the water sample. Procedural details
also affect the counts, and it is important to ran-
domize the order of procedures so that unwanted
variables appear as scatter rather than as bias in the
means. The sample source is also an important
variable which should be randomized. Finally, non-
membrane control methods are desirable in mem-
TABLE 3. EFFECT OF AGAR THICKNESS ON FECAL COLIFORM COUNTS ON
SPREAD PLATES
Test Date
Fecal Coliform Count*
On 15 ml Agar On 35 ml Agar
(0.25 cm thick) (0.59 cm thick)
* Each value is a mean of six replicates.
** At the 95% confidence level.
Size of Statistically
Significant
Difference**
11/12/74
11/14/74
11/15/74
11/20/74
11/21/74
11/30/74
12/27/74
Overall
40.8
30.2
68.7
24.8
48.8
60.8
27.3
43.0
49.7
46.8
94.3
46.7
69.8
82.6
27.7
59.7
7.6
7.0
10.2
9.3
8.7
9.6
5.9
3.1
31
-------
brane evaluation experiments, but these are subject
to procedural variables, too, and do not necessarily
provide a complete measure of the number of
viable bacteria.
REFERENCES
1. Sladek, K.J., R.V. Suslavich, B.I. Sohn, and
F.W. Dawson. Optimum Membrane Struc-
tures for Growth of Fecal Coliform Organ-
isms. This Symposium, 1975.
2. Moroney, M.J. Facts From Figures, Penguin
Books Ltd., pp. 96-107; pp. 220-1; pp. 261-3,
1971.
3. Bennett, C.A., and N.L. Franklin. Statistical
Analysis in Chemistry and the Chemical
Industry. John Wiley & Sons, pp. 115-8; pp.
172-5; pp. 601-11, 1954.
4. Rose, R.E., W. Litsky, and E.E. Geldreich.
An improved Membrane Filter Method for
Fecal Coliform Analysis. This Symposium
1975.
Geldreich: We see it this way and I just won-
dered if you recognized this problem
of lot to lot variation.
Sladek: I think, the best way that I could
reply is to say, if you want to study
sterilization, you should start with
things that are identical. I think too
often in the past people have found
differences between membrane filters
and may have attributed them to
sterilization whereas in fact the dif-
ferences were from other causes.
Geldreich: Part of this problem may be an inter-
reaction between ethylene oxide and
the membranes. Perhaps some of the
products on the membrane may give
a latent residual effect. Is there a
problem?
Dawson: Mr. Geldreich has mentioned the
possibility of residual ethylene oxide
or hydrolysis products. Karl would
you comment on residual ETO?
QUESTIONS AND ANSWERS
Geldreich: In these studies on randomization you
mentioned 18 membranes that you
picked at random. Are you also talk-
ing about 18 random lots of mem-
branes?
No, 18 membranes.
Sladek:
Geldreich:
Sladek:
Geldreich:
Sladek:
Geldreich:
How many lots are you talking about? Sims:
This particular study was for the pur-
pose of evaluating this one effect,
ethylene oxide sterilization. This
specific study was done with 36 mem-
branes which were as closely repli-
cated as we could choose.
Out of the same lot?
Yes, they were identical.
You recognize that the variation of
membrane filters from lot to lot is
going to be another great variable that
we are going to have to work with. Is
this correct?
Sladek: Quite possibly.
Saldek: I would be glad to. We have had
analyzed a number of membranes
that were sterilized by ethylene oxide.
We had the membranes analyzed by
an outside laboratory, the best one we
could find. They never found any
residual of ethylene oxide by chemi-
cal analysis. The limit of detection in
that test was 4 parts per billion.
One problem that will be coming up
over the next two days is: what is a
lot? Is a lot the amount of raw mater-
ials you mixed, that you manufactur-
ed, or that you sterilized? One lot,
number 500, may refer to each time
the sterilizer was used; it can be each
time a batch of raw materials were
mixed; it can be each time the
machine operated. This word 'lots'
can refer to at least three different
possibilities.
Sladek: I would like to comment on that. Its
true that sometimes there are diffi-
culties in defining a lot. However,
once you have decided what a lot is,
you get a lot of protection by doing
random sampling and using a large
number of replicates. For example,
32
-------
Litsky:
Sims:
we have often sampled lots of mem-
brane filters taking a sample of one
hundred membranes drawn at ran-
dom. This makes certain that if you
didn't recognize some difference be-
tween lots when you were defining
them, by random sampling you will
pick up these differences in your
quality control program.
Would you give us your definition of
a lot? What constitutes a lot? Would
you like to give an answer?
I was actually asking what constitutes
a lot for him. For me a lot is one day's
machine operation. In our plant one
solution is prepared per day. We
sterilize about 4 times a day and we
have to check each load after steriliza-
tion. I didn't know what they were
calling a lot but the definition of a lot
can be important when related to all
of the data and quality control and
experimental results.
Cotton: In our terminology we refer to a batch
as that material produced at any one
particular period of time during a day.
For the lot numbering purposes we
take a section of that batch and we
may treat it in one fashion or another.
We may autoclave pack it, or ethylene
oxide pack it. That section of the
day's batch which is produced and
packaged in a particular way is con-
sidered and given a lot number.
33
-------
EFFECT OF INJURY ON THE RECOVERY
OF INDICATOR BACTERIA ON MEMBRANE FILTERS
Review of Author's Research
Alfred W. Hoadley
School of Civil Engineering
Georgia Institute of Technology
Atlanta, Georgia 30332
ABSTRACT
Investigations have been undertaken which
demonstrate that indicator bacteria, which are
injured during aqueous suspension or by exposure
to chlorine, may fail to form colonies on mem-
brane filters incubated on selective media. Al-
though dividing cells of Escherichia coli were
recovered equally well on Trypticase soy agar and
membrane filters incubated on M-FC medium at
44.5 C, the efficiency of cell recovery decreased
with time of exposure to stress. Streptococcus
faecalis, in aqueous suspension, was recovered on
membrane filters incubated on KF agar and on
M-Enterococcus agar, with an efficiency of less
than 50%. Recovery of streptococci did not
decrease with time of suspension.
These findings helped to explain results ob-
tained by other workers and suggest the need to re-
duce the selectivity of membrane filter techniques
against injured cells. Comparative studies of the
recovery of decreasing, as well as growing, popula-
tions on rich and selective media ought to be in-
cluded in evaluations of selective media.
INTRODUCTION
Most microbiologists have been concerned
primarily with growth measurements of bacterial
populations. The food and sanitary microbiologists
are often required to estimate numbers of stressed
indicator bacteria and pathogens. This paper re-
views recent observations on the enumeration of
declining bacterial populations in water, with spe-
cial reference to the recovery of indicator bacteria
on membrane filters.
Background
In recent years, an extensive amount of
literature has appeared on the recovery of starved,
heated, frozen and thawed indicator bacteria,
leakage and degradation of cellular components,
and the repair of cell injury. A limited amount of
literature has appeared on the recovery of bacteria
exposed to disinfecting agents. However, little
effort has been made to evaluate the recovery of
stressed indicator bacteria in aquatic environments.
In 1961, McCarthy, Delaney and Grasso re-
ported that "weaker" coliforms failed to form
colonies on membrane filters incubated on M-
Endo broth. They found that pre-enrichment
on lauryl tryptose broth, prior to incubation on
LES .M-Endo agar, improved recovery from sur-
face water samples. Klein and Wu (3) recently
demonstrated significantly higher recoveries of
bacteria from stream waters on spread plates than
on pour plates. These results indicate that a signifi-
cant portion of the heterotrophic bacterial flora
in the streams consisted of injured cells that were
unable to tolerate the secondary stress of exposure
to melted agar at 42, 45, and 50 C. Hoadley and
Cheng (2) demonstrated injury of indicator bac-
teria suspended in a variety of aqueous suspending
media. Injury prevented recovery on selective
media.
In 1958, McKee, McLaughlin and Lesgourgues
demonstrated reduced recoveries of coliforms from
chlorinated settled sewage on membrane filters
that were incubated on dehydrated scheduled
nutrient pads containing an Endo-type medium.
On the other hand, recoveries of coliforms from
unchlorinated settled sewage on membrane filters,
34
-------
by most probable number tests, were in agreement.
By applying the pre-enrichment procedure of
McCarthy, Delaney and Grasso (7), Lin (5) was
able to obtain agreement between most prob-
able number values and MF counts of total and
fecal coliforms in chlorinated secondary sewage on
membrane filters incubated on M-Endo and M-FC
medium. Braswell and Hoadley (1) described
injury to Escherichia coli cells in chlorinated
secondary sewage.
It is the purpose of the present review paper
to describe briefly the recovery of injured and lag
phase indicator bacteria on membrane filters
employing standard selective media. Our results
suggested the need to reduce the selectivity of
membrane fitter techniques against injured cells,
and the need to understand cell injury better. It
is encouraging that the recovery of injured bacteria
is receiving so much attention at this symposium.
Hoadley and Cheng (2) examined the re-
covery, on membrane filters incubated on stand-
ard selective media, of E. coli and Streptococcus
faecalis after suspension for varying periods of time
in sterile stream water, double distilled water,
phosphate buffer (3.125 x 1Q-4M, pH 7.2), pep-
tone water (0.01%), and tap water. Escherichia
coli ATCC 11775 and Strep, faecalis ATCC 19433
were spread on plates of Trypticase soy agar (BBL)
and incubated for 18 to 24 hr. at 37 C and 35 C,
respectively. Suspensions of each organism were
prepared in each of the sterile experimental sus-
pending media and compared with McFarland
barium sulfate standards to obtain an inoculum
yielding a density of about 1000 organisms/ml
after addition to the test suspension. One liter
volumes of test suspensions were maintained at
20 C in water jacketed stirred flasks and samples
were removed at intervals of up to 24 hours for
counting.
Upon removal, samples were spread immedi-
ately in triplicate on Trypticase soy agar plates and
were filtered in triplicate through membrane
filters (type HA, Millipore Filter Corp.). Spread
plates inoculated with E. coli were incubated at
37 C and were counted at 24 and 48 hours. Mem-
brane filters carrying E. coli were incubated on
pads saturated with M-FC broth (Difco) at 44.5 C
for 22±2 hours. Spread plates inoculated with
Strep, faecalis were incubated at 35 C and were
counted at 48 hours. Membrane filters were
incubated on KF broth (BBL) and M-Enterococcus
agar (BBL) at 35 C and were also counted at 48
hours.
Counts of E. coli in double distilled water
are presented in Figure 1. As a rule, counts on
membrane filters initially resembled counts on the
rich, non-selective control medium. But as the
age of the suspension increased, so did the dis-
crepancy between counts on the two media.
Counts in double distilled water are presented
because the discrepancy between recovery on the
two media is greater than in phosphate buffer.
Many tap waters may be highly toxic, even in the
absence of residual chlorine. As a result, recovery
on M-FC medium was very poor (Figure 2). Re-
coveries from peptone water and stream water
(each of which contained between 545-550 mg/
liter total organic carbon) were nearly identical.
In each suspending medium, growth followed a
lag period (Figure 3). During the lag period, the
discrepancy between counts on selective and rich
media increased until about 10 hours, after which
cells recovered. Following recovery (18 hours),
counts were nearly identical on the two media.
Poor recovery of E. coli on a selective medium
during the lag phase was observed also by
Scheusner, Busta, and Speck (9).
The implications of these observations are
clear. Cells of E. coli were injured during suspen-
sion in water, either as a result of leakage or
degradation of cellular components. As a result, a
substantial portion of the population failed to
produce colonies on membrane filters incubated
on selective media.
In contrast, although recoveries of S. faecalis
were at all times low on both selective media, there
was no evidence of progressive cell injury in most
suspending media. Recoveries of S. faecalis sus-
pended in distilled water and phosphate buffer
were identical, and remained constant over a 24
hour period (Figure 4). However, recoveries on
both KF medium and M-Enterococcus agar were
about one half those on Trypticase soy agar. Re-
coveries from toxic tap waters on membrane
filters were very much lower (Figure 5). S. faecalis
grew both in stream water and peptone water fol-
lowing a short lag (Figure 6). Counts on each selec-
tive medium again were lower than those on the
rich medium, but they reflected the behavior of
the population as a whole. The consistent re-
coveries of S. faecalis from most aqueous sus-
pensions on selective media is a desirable attribute
in a recovery medium.
Recovery of E. coli from Chlorinated Sewage
Braswell and Hoadley (1) investigated the
recovery on membrane filters of E. coli ATCC
35
-------
200
100
O ^ 1
o < '
LU
cc >
0.1
I '
i I T
I i I i I i I i
8 12 16 20
AGE OF SUSPENSION (hr)
24
28
32
Figure 1. Recovery of E. Coli ATCC 11755 Suspended in Double Distilled
Water on Trypticase Soy Agar (Circles) and on m-FC Medium
(Triangles) (after Hoadley and Cheng, 1974).
100
35
AGE OF SUSPENSION (hr)
Figure 2. Recovery of E. Coli ATCC 11755 Suspended in Tap Water on
Trypticase Soy Agar (Circles) and on m-FC Medium (Triangles)
(after Hoadley and Cheng, 1974).
36
-------
1000
ig
SJ2
N? z
b=)
5- O
£ o
> %
O OQ
O <
UJ
DC >
o
o
»
o
^r
i
8 12 16 20
AGE OF SUSPENSION (hr)
24
28
32
Figure 3. Recovery of E. Coli ATCC 11755 Suspended in Stream Water
on Trypticase Soy Agar (Circles) and on m-FC Medium
(Triangles) (after Hoadley and Cheng, 1974).
200
100
CO
> o
£ 0
O 03
O <
UJ
cc >
0.1
8 12 16 20
AGE OF SUSPENSION (hr)
24
28
32
Figure 4. Recovery of Strep. Faecalis ATCC 19433 Suspended in Double
Distilled Water on Trypticase Soy Agar (Circles), KF Medium
(Squares), and m-Enterococcus Agar (Triangles) (after
Hoadley and Cheng, 1974).
37
-------
100
80
u_ co 60
LU
40
o
20
1
3 5
AGE OF SUSPENSION (hr)
Figure 5. Recovery of Strep. Faecalis ATCC 19433 Suspended in Tap
Water on Tryptlease Soy Agar (Circles), KF Medium (Squares),
and m-Enterococcus Agar (Triangles) (after Hoadley and
Cheng, 1974).
1000
~°100
o£
O m
0< 10
11 I v
8 12 16 20
AGE OF SUSPENSION (hr)
24
28
32
Figure 6. Recovery of Sterp. Faecalis ATCC 19433 Suspended in Stream
Water on Trypticase Soy Agar (Circles), KF Medium (Squares),
and m-Enterococcus Agar (Triangles) (after Hoadley and
Cheng, 1974).
38
-------
27622 suspended in secondary sewage and exposed
to chlorine. A suspension of cells was prepared, as
previously described, and was added to 500 ml of
sterile sewage in a 1000 ml beaker to yield approx-
imately 2 x 10^ organisms/ml. A sodium hypo-
chlorite solution was added to yield a dosage of 2
to 3 mg/liter and a total residual of 0.3 to 0.5 mg/
liter after 30 minutes. Ten ml samples were re-
moved after 1, 5, 10, 20, and 30 minute intervals,
and were placed immediately into 10 ml of sterile
sodium thiosulfate, after which appropriate dilu-
tions were made. Diluted samples were spread in
duplicate on Trypticase soy agar plates and filtered
in triplicate through membrane filters (type HA,
Millipore Filter Corp). Spread plates were incubat-
ed at 37 C for 24 hours. Membrane filters were
placed on pads saturated with M-FC broth (Difco)
and incubated at 44.5 C for 24 hours. In one ex-
periment, filters were placed on Trypticase soy
agar and on pads saturated with Trypticase soy
broth.
Counts of E. coli in chlorinated secondary
sewage are presented in Table 1. Counts on mem-
brane filters generally were close to those on
spread plates initially. With exposure of sewage to
Table 1. Recovery of E. coli ATCC 27622 from
chlorinated secondary sewage on Trypti-
case soy agar and on M-FC medium
(after Braswell and Hoadley (1)).
Chlorin
contad
time
(min)
1
5
10
20
30
e
[
Expt
TSAC
16,000
1,600
1,000
10
6
Counts/ml
1a
M-FC
15,400
17
0
0
0
Expt
TSA
40,000
26,200
9,600
1,000
100
2b
M-FC
14,000
1,400
460
0
0
aTemperature, 21 C; pH 7.0; chlorine dosage,
3 mg/liter; chlorine residual after 30 min, 0.75
mg/liter.
bTemperature, 21 C; pH 7.0; chlorine dosage,
2mg/liter; chlorine residual after 30 min, 0.35
mg/liter.
CTSA, Trypticase soy agar.
chlorine, however, recovery of E. coli on filters was
decreased more rapidly than recovery on spread
plates. In each of the experiments reported in
Table 1, counts on the rich medium were still 1000
bacteria/ml after colonies failed to form on the
selective medium. Scheusner et al (9) and Maxcy
(6) reported that E. coli exposed to chlorine and
other disinfectants did not form colonies on violet
red bile agar. Poor recoveries of coliforms from
chlorinated settled sewage on membrane filters
reported by McKee, Mclaughlin, and Lesgourgues
(8), and enhanced recoveries following pre-enrich-
ment reported by Lin (5), can be explained if it is
understood that injury occurs during exposure to
chlorine, thus preventing recovery on selective
media.
A single experiment was undertaken to deter-
mine whether injured cells form colonies as readily
on membrane filters as they do on agar surfaces.
Spread plates were prepared as usual and samples
were filtered through membrane filters, then
incubated in triplicate on Tyrpticase soy agar, and
pads saturated with Trypticase soy broth. All
plates were incubated at 37 C for 24 hours. Re-
sults are presented in Table 2.
As in previous experiments, initial counts on
membrane filters closely resembled those on spread
Table 2. Recovery of E. coli ATCC 27622 from
chlorinated secondary sewage on Trypti-
case soy agar (Braswell and Hoadley,
unpublished)^
Counts/ml
Chlorine contact
time (min)
Membrane filter
Spread Plate TSA TSBC
1
5
10
20
30
33,670
31,630
20,700
500
30
33,570
28,070
14,630
45
2
30,600
22,370
4,800
1
aTemperature, 21 C; pH 6.9; chlorine dosage, 2
mg/liter; chlorine residual after 30 min, 0.4 mg/
liter.
"TSA, Trypticase soy agar.
CTSB, Trypticase soy broth.
39
-------
plates. However, after the initial sample, recoveries
were lower on membrane filters, and again the
discrepancy increased as time of exposure to
chlorine increased. Furthermore, recoveries on
membrane filters were poorer when incubated on
pads saturated with Trypticase soy broth than
when incubated on Trypticase soy agar.
Conclusions
It is clear from the above observations that we
must pay more heed to the effects of injury on the
recovery of indicator bacteria from water. While
healthy, multiplying E. coli cells recover well on
membrane filters incubated on selective media,
cells exposed to stress may not form colonies on
filters or selective media after suspension in surface
waters, chlorination of wastes, or chlorination and
passage of potable waters through distribution
systems. Furthermore, it might be possible to
devise media, superior to those employed, on
which all viable cells are able to form colonies. The
significance of the discrepancy between counts on
rich and selective media is suggested by the obser-
vations of Speck and Cowman (11), that freeze-
injured salmonellae may be as pathogenic as un-
injured cells.
Evaluations of media and techniques should
include examination of the recovery of stressed
bacteria in pure culture. Cells of Pseudomonas
aeruginosa stressed in fresh waters and estuarine
waters were employed by Levin and Cabelli (4)
to evaluate their M-PA procedure. Such evaluations
might be applied profitably to existing, as well as
newly developed procedures.
REFERENCES
1. Braswell, J.R., and A.W. Hoadley. Recovery
of Escherichia coli from chlorinated second-
ary sewage. Appl. Microbiol. 28:328-329,
1974.
2. Hoadley, A.W., and C.M. Cheng. The recovery
of indicator bacteria on selective media. J.
Appl. Bacteriol. 37:34-57, 1974.
3. Klein, D.A., and S.Wu. Stress: a factor to be
considered in heterotrophic microorganism
enumeration from aquatic environments.
Appl. Microbiol. 27:429-431, 1974.
4. Levin, M.A., and V.J. Cabelli. Membrane
filter technique for enumeration of Pseudo-
monas aeruginosa. Appl. Microbiol. 24:864-
870, 1972.
5. Lin, S. Evaluation of coliform tests for chlor-
inated secondary effluents. J. Water Pollut.
Control. Fed. 45:498-506, 1973.
6. Maxcy, R.B. Non-lethal injury and limitations
of recovery of coliform organisms on selective
media. J. Milk Food Technol. 33:445-448,
1970.
7. McCarthy, J.A., J.E. Delaney, and R.J.
Measuring coliforms in water. Water Sew.
Works. 108:238-243, 1961.
8. McKee, J.E., R.T. McLaughlin, and P. Les-
gourgues. Application of molecular filter
techniques to the bacterial assay of sewage.
III. Effects of physical and chemical disinfec-
tion. Sew. Ind. Wastes. 30:245-252., 1958.
9. Scheusner, D.L., F.F. Busta, and M.L. Speck.
Injury of bacteria by sanitizers. Appl. Micro-
biol. 21:41-45, 1971.
10. Scheusner, D.L., F.F. Busta, and M.L. Speck.
Inhibition of injured Escherichia coli by sev-
eral agents. Appl. Microbiol. 21:46-49, 1971.
11. Speck, M.L., and R.A. Cowman. Injury and
recovery of frozen microorganisms. J. Milk
Food Technol. 34:548, 1971.
Discussion
Brezenski: I would like to ask you one question
with respect to the chlorination experi-
ments. Do you have some of the other
chemical characteristics of the second-
ary effluent that you use with respect
to solid and other organic removals?
Hoadley: We don't have parameters like TOC,
TSS, etc. These samples were just
autoclaved and we used the same
batches as much as possible.
Brezenski: We did some work several years ago in
our laboratory comparing the re-
coveries of membrane filter and MPN
systems from primary chlorinated ef-
fluents of sewage treatment plants. We
experienced atrocious recoveries. The
MF's , in every case, were many magni-
tudes lower. But, as we progressed
through various stages of treatment,
e.g., when we went up to secondary
treatment and to effluents from good
activated sludge plants, we found that
the results were coming closer together.
When we came to a tertiary plant, we
found that we almost experienced
comparable results. These results make
40
-------
Hoadley:
Alico:
Hoadley:
me wonder if we are dealing solely with Alico:
the recovery of injured cells. I agree
with you 100% that we do have a Hoadley:
problem of cells being stressed in the
environment, and also in chlorinated
waters. I also believe that we have to
take into consideration the physical
factors, for example, the level of Lane:
suspended solids. It is quite evident
from this data that if you do have a
high level of suspended solids, the
membrane filter is not going to give a
good recovery. There is no question
that McKee, in his early work in 1958,
showed that chloro-organic complexes
form on the membrane. And this does
have a growth-inhibition effect on these
colonies. I believe we are not only
talking about a stressed system, we are
also talking about a system of physical
characteristics coming into play. I think Hoadley:
the analyst, in this case, is going to have
to determine when he sees the sample
concerned, how he is going to treat it.
It becomes difficult because now we
have some subjectivity involved.
I agree. I think there are many factors
that are going to influence our re- Dawson:
coveries here. All I can do is demon-
strate a phenomenon. There is an ef-
fect. Initially our recoveries are much
the same. In all of this work we would
grow the bacteria on plates, suspend
the cells, adjust the turbidity and
inoculate our samples. This took 7
minutes usually. Then we would run
those things, initially the counts are
about the same.
Two questions, the first is, how many
replicates did you run and secondly
what were the means of sterilization
of the filters that you used?
We used the filters from the presteri-
lized packages as they came. Usually
tests were performed in triplicate. We Hoadley:
have run various experiments on dif-
ferent strains from time to time, and
you get variation from experiment to
experiment; but the phenomenon is
there.
What membranes were you using?
I don't think the brand is pertinent
here, because we are looking at the
phenomenon; but they were ethylene
oxide-sterilized.
I would just like to point out here the
importance of osmotic pressure and its
effects on recovery of injured cells. For
example, in clinical microbiology and
blood cultures, if you take one broth,
use it as is and add 10% sucrose to an-
other bottle of the same broth, you will
get a increased recovery of injured cells
in the broth with the 10% sucrose. This
has been demonstrated repeatedly, and
I think you may have a similar situation
here.
I think there are a lot of questions
about the environmental conditions.
I don't think we have good control of
that in our experiments, and I'm
ready to admit that. I imagine we were
not doing things much differently
than other people, so this is the way
our techniques are applied.
Just a comment. I think that osmotic
pressure may play an important factor
here but I think we also should look at
the quality of the water we use for pre-
paring dilutions in reagents. The gentle-
man from BBL indicated that he would
like to see DSP purified water used as
reagent grade water. Purified water
USP-18 is not all that good in the
light of modern-day technology. Single
distilled water can contain a high
quantity of amines and we know that
amines come over with a single distilla-
tion. There are numerous reports
published in the literature that amines
are toxic to both bacteria and to tissue
culture cells and I wonder if we are
seeing something here?
We would never use single distilled
water at all. It becomes very clear in
any of these comparisons that the
better the quality of distilled water the
better your recoveries are after sus-
pension.
41
-------
EFFECTS OF TEMPERATURE ON THE RECOVERY OF
FECAL COLI FORMS
J. B. Hufham, Ph. D.
Department of Life Science
University of Missouri at Rolla
ABSTRACT
The theoretical concept of cell and culture
death resulting from increased temperature is
important to an understanding of the inaccuracy
of the fecal coliform procedure. Results of our
studies on the differential effects of temperature
on various types of membrane filters show that the
type of filter is an important variable. Recoveries
of known densities of pure cultures of E. coli
varied from 10-70 percent from this variable
alone. The importance of selection of the proper
temperature for membrane evaluation cannot be
stressed too strongly.
Results show that the interaction of the
membrane with the cell, resulting in loss of cell
viability, is temperature dependent and varies
from brand to brand. A modified fecal coliform
procedure has been developed and field tested.
This new procedure lessens the detrimental effects
of the membrane on recovery of the organism.
While this symposium is entitled, "The Re-
covery of Indicator Organisms Employing Mem-
brane Filters", I sense that it is really an effort to
re-establish the M-FC method, or a modification of
that method, as recommended technique. I sin-
cerely hope that this symposium and the discus-
sions that follow will, in fact, provide investiga-
tors with a workable method.
It is no secret that there have been several
reports over the past year and a half which have
demonstrated poor results with the M-FC method
under some conditions. Most of these reports have
indicated that the fault lies in the quality of the
membrane filter. As the author of one of those
reports, I am certainly not going to deny that con-
clusion. I will modify it and say that the mem-
brane quality gives some of the error and most
of the variability.
I will limit my presentation to the M-FC
method. The subject of my presentation is the
effects of temperature on recovery, but in order to
fully develop my topic I would like to discuss
several aspects of our evaluations concerning the
procedures that were used. I hope that I will not
tread on the data of our other speakers.
In order to evaluate the problems with the
M-FC procedure, one must first develop a method,
based on sound reasoning, which restricts the vari-
ables with which one has to deal. I would like to
begin by discussing some of these variables and
describing our selection of a defined evaluation
procedure. (Tab. 1).
The organism
Any organism selected as a test organism in
evaluating this method must meet several criteria:
1) The organism must be a pure culture. This
is not to deny the use of field trials; nor of their
importance in the total evaluation of any analytical
procedure. It is necessary, however, to first estab-
TABLE 1. VARIABLES
A. Organism
1. Must be a pure culture.
2. Must be easy to obtain by the investigator.
3. Must be E. coli.
4. Must not be selected for by the M-FC
Procedure.
B. Medium
C. Filter membrane
D. Temperature
42
-------
lish the validity of the test procedure using a
known, pure culture of the organism which the test
is supposed to quantitate. I stress this because
some have proposed using mixed cultures. Mixed
cultures have a way of changing their makeup
depending upon how they are cultured.
2) The organism must be easy to obtain.
Specifically it should be on deposit with a type
culture collection and the strain number identified
so that each investigator is not introducing his own
variable. We have often found that we could not
compare our results with other investigators
because of the organism employed and the im-
possibility of isolating the same culture under the
same conditions.
3) The organism must be E. coli. This point
should be obvious. Notice, I did not say it should
be IMVIC positive. True, E. coli is, but that test is
not sufficient by itself to identify the organism.
Such studies have never been done. It does differ-
entiate the organism from Enterobacter aerogenes
and so far that is the only use of the test. Nor did I
mention other criteria such as lactose fermenta-
tion at 44.5 C, etc. There may be other organisms
that meet that test. If it is E. coli, it meets all of
the criteria. But if it meets all of the criteria, it
may not be E. coli.
4) The organism must not be selected for by
the M-FC Procedure. By using the M-FC Procedure
to obtain the test organism, it is possible that one
might obtain a mutant more resistant to the inhibi-
tory effect. This would lead to erroneous results
since the procedure itself is not designed to recover
only those resistant types. We isolated such a
colony and it gave 2.5 times the recovery as our
stock culture.
In our work, we selected ATCC strain 11775
because it is the neo-type strain, and we felt that
this organism best represented the organism for
which coliform tests were originally developed.
Cultures were grown in nutrient broth for 16
hours at 35 C.
The medium
The original investigators made the medium
from scratch. Most people probably employ
commercially available products. The question is,
are all products equal in quality and do they vary
from batch to batch. We had not had time to ex-
pand our studies into this area, but we feel that
someone should. I am also concerned about the
sources and quality of the rosolic acid employed.
We have used both Difco and BBL media with no
apparent difference. Our rosolic acid has always
been from Difco. In order to remove a variable
we are using Difco medium only, at the present
time.
The filter membrane
We did some work comparing the different
brands of membranes. The early work by Press-
wood and Brown, plus some of our own work
comparing Gelman and Millipore filters, showed
that while the brand of membrane has a decided
effect on the recovery in the M-FC Procedure, it
might not be the only effect. First we compared,
Then we chose Millipore because it was the worst
and we wanted to know why. Then we went to
Gelman because it gave the best results and we
would not have a second experimental variable..
And finally we decided to use none because we
couldn't trust any of them. One gets that way
when he opens a new package and finds nothing
between the blue paper.
The temperature
Temperature in this method is very import-
ant. 44.5 C is not hard to obtain, but most incu-
bators and waterbaths have a hard time maintain-
ing ± 0.2 C throughout the chamber. For this rea-
son we used a homemade circulating waterbath
that we had been using for our enzyme work. We
adjusted and checked it with a quartz thermometer
at first, but we now use a thermometer which has
been checked against an NBS standard thermom-
eter. I have seen many people who simply reach in
the drawer, pull out a thermometer and use it.
Perhaps we need to stress standardization more
than we do.
We have concentrated on this variable, tem-
perature, and it is these results which I would like
to discuss today.
What effect does temperature play in the
recovery of coliformsby the M-FC method? Let's
look at the results of several investigations compar-
ing brands of filters. Presswood and Brown com-
pared Millipore and Gelman filters. They found a
mean loss on Millipore of 53 percent when com-
pared with the Gelman filter. In their study, they
concluded that temperature itself was not detri-
mental, but that the Millipore filter was. We dis-
agree with those conclusions. Their organisms
were isolated with the M-FC Procedure and their
results were compared to cultures grown on M-FC
agar pour plates at 35 C. Since the difference
43
-------
showed up only at the elevated temperature it
would seem that temperature did have an effect.
The question is, is the effect on the membrane, on
the cells, or both. Recently Dutka, Jackson, and
Bell reported on a comparison of six different
filters. None of these filters gave better than 38
percent recovery at 44.5 C compared to a pour
plate control at the same temperature.
When we compared Millipore with Gelman we
found that the Millipore recovered only about 10
percent of the cells known to be in the sample.
(Tab. 2) The Gelman filter, however, recovered
about 55 percent. Our standard was the number of
cells which grew on plate-count broth at 35 C. We
feel this is the only reliable criteron for knowing
how many viable cells there are in the sample.
TABLE 2. RELATIVE ERROR IN THE FECAL-
COLIFORM METHOD ASA
FUNCTION OF THE BRAND OF
MEMBRANE FILTER EMPLOYED.
No. of E. coli cells
per 100ml
Experi-
ment
No.
1
2
3
a Rplat
Filter
Brand
Millipore
Gelman
Millipore
Gelman
Millipore
Gelman
i\/p orrnr =
Total
Coliform M-FC
Broth Broth
(35 C) (44.5 C)
50
48
46
48
79
73
count 35
6
26
5
21
6
50
C-count 44.5
Relative
Error3
(%)
88
46
89
56
92
32
C
count 35 C
One can improve the results with Millipore
filters by incubating the stock culture at 44.5 C
for an extended period of time. (Tab. 3) After 120
hours at 44.5 C the recovery on Millipore went
from 13 percent to 60 percent. Please notice that
the number of viable cells decreases with time but
the ratio of those countable at 35 C to those at
44.5 C changes. It was thought that by serially
diluting the culture, we could select for 100 per-
cent recovery.
TABLES. RECOVERY DATA FOR E. COLI
GROWN AT 44.5 C.
No. of E. coli cells
per 100 mla
Tube
no.b
1
2
3
4
Incubation
Time (1 hr)
(44.5 C)
24
72
96
120
48
72
96
24
48
24
Total
Coliform
1,180
300
20
30
920
240
110
1,780
510
1,440
M-FC
(44.5 C)
158
30
9
18
85
62
22
490
113
540
<%>
13
10
45
60
9
27
20
28
22
38
a. Counts are given as cells per 100 ml of a 10
dilution.
b. Each tube is a serial inoculation of the
previous tube.
If temperature alone affects the recovery,
then it should be noticable without any filter. We
ran standard plate counts at various temperatures
and found that the count dropped as the tempera-
ture was increased above 40 C. (Tab. 4) Please
note that the recovery is about 50-60 percent of
that obtained at 35 C.
We have tried to improve the technique by
lowering the incubation temperature. At 35 C
good differentiation was obtained between E. Coli
and E. aerogenes. (Tab. 5) The same would prob-
ably be true at 40 C with Gelman membranes
and we recommend that this be field tested by
those who are set up to perform extensive field
trials.
Let me briefly summarize our conclusions.
1. Incubation of E. coli at 44.5 C destroys
the viability of at least 40 percent of the
cells, even in the absence of a membrane
filter.
44
-------
TABLE 4. COMPARISON OF POUR PLATE
RECOVERIES OF E. COLI* ON
PLATE COUNT AGAR AT
VARIOUS TEMPERATURES.
Temp
30 C
35 C
41. 5 ± 0.5 C
44. 5 C
35 C
44.5 C
Average
Count
142
142
94
70
640
470
No. of
Plates
2
2
2
2
4
4
* E. coli#11775
TABLE 5. DIFFERENTIATION OF E. COLI
AND E. AEROGENES BY M-FC
METHODS AT 35C.
No. of cells per 100 ml
Experiment Culture
1
2
3
4.
5.
E. coli
E. coli
E. aerogenes
E. coli
E. aerogenes
Mixture (l:l)
E. coli
E. aerogenes
Mixture (l:l)
TGY
Broth
(35 C)
58
51
59
M-FC Broth
(35 C)
59
48
58a
73
154a
230a
156
253a
389a
a. E. aerogenes colonies were cream to light
green in color.
2. Use of the Millipore membrane inhibits
another 50 percent of the cells. This is
probably a heat soluble factor.
3. Previous incubation on Millipore mem-
branes at 44.5 C would select for
mutants capable of growing under
these conditions.
I believe that a test strain of E. coli should be
established. I also believe that counts of this cul-
ture on plate count medium at 44.5 C should
equal the counts on the same medium at 35 C.
Until they do, I do not feel that the M-FC Pro-
cedure should even be considered.
Question and Answer Session
Grasso: You made mention of your comparison
studies of the fecal coliformson Gelman
and Millipore filters, and mention of
numbers that were recovered. You didn't
make any mention of colony size or
characteristic. Did you detect any dif-
ferences on the two filters at the higher
temperatures?
Hufham: Yes. There is a tremendous difference.
At 35 C, colonies were about the same
size. When you get to 44.5 C, you find
a lot of variation. You find pinpoint
colonies and, if you let the culture
incubate a little more, you sometimes
can see additional colonies to count
that weren't there previously. I don't
know what causes that.
Grasso: The reason that I mention it is because,
as I will state in my paper, we found
marked differences at 44.5 C in the
characteristics of fecal coliform colonies
on the Gelman and Millipore filters, as
far as size and color are concerned. I
think that in addition to all the physical
characteristics and media problems,
there is also another problem; the actual
definition of a fecal coliform.
Hufham: Let me answer that by saying we came
across so many variable we didn't feel
we could make a good solid study.
We tried to eliminate as many variables
as possible. We tried to get rid of media
that seems to have a problem. We
tried to get rid of the membrane and we
tried various cultures to see if they grew
at one temperature or another. We
ended up in frustration.
Ginsburg: I want to make a comment about the
rosolic acid. We found that early in our
use of M-FC tests we also had a problem.
But now we do not incorporate rosolic
acid and get comparative results.
45
-------
OPTIMUM MEMBRANE STRUCTURES FOR GROWTH OF
FECAL COLIFORM ORGANISMS
K.J. Sladek, R.V. Suslavich, B.I. Sohn, and F.W. Dawson
Millipore Corporation, Bedford, Massachusetts
ABSTRACT
The purpose of this study was to determine
the optimum membrane filter structure and charac-
teristics for recovery of coliform organisms.
Additionally, other factors such as sterilization
method and membrane composition were ex-
amined. Fecal coliform growth tests with varied
samples indicated that the most critical factor in
recovery was surface pore morophology and not
other factors previously suspected. Fecal coliform
counts showed a dramatic increase with increasing
surface opening sizes. Membrane structures with
surface openings large enough to surround the en-
trapped bacteria are required for optimum growth
of fecal coliform organisms. Maximum fecal coli-
form recoveries are obtained using membranes
composed of mixed esters of cellulose exhibiting a
surface opening diameter of 2.4 ^m and a retention
pore size of 0.7 /im.
INTRODUCTION
Since its introduction as a tentative method
for coliform enumeration in the 10th Edition of
Standard Methods in 1955, the membrane filter
has gained wide acceptance not only for total
coliform, but also for fecal coliform, total bacteria,
and a wide variety of other bacterial tests. The
unique advantage of the membrane over other test
methods is its ability to concentrate and localize
bacteria from large samples. Hence, the membrane
increases the sensitivity of quantitative bacteri-
ology into the range well below one organism per
ml. Once the bacteria are localized, the membrane
provides a structure for counterdiffusion of nu-
trients and metabolic products as well as "hospit-
able" growth environment. In these functions, the
membrane differs little from the earlier pour and
streak plate methods.
The earliest technique for bacteriological
analysis with membrane filters involved direct
microscopic examination of bacteria trapped
on the membrane surface. Here, the optimum
structure required pores smaller than the organisms
being trapped for examination so that they would
lie in a single microscopic plane. This surface
planar retention facilitated finding the organisms
under high power microscopy. The above require-
ments evolved naturally to the practice of retaining
organisms on the membrane surface for various
culture techniques. At that time, not much
thought was given to developing an optimal mem-
brane structure for colony growth.
The ideal characteristics of a membrane for
quantitative bacteriology would appear to be pores
small enough to retain bacteria but open enough to
provide paths for liquid transports, and a "hospit-
able" surface growth. However, upon examin-
ing the variety of bacterial methods utilizing
membranes, one finds a considerable range of
bacteria sizes, types, and metabolic requirements.
These considerations led us to wonder if it was pos-
sible to develop membranes which would be
especially favorable for the growth of particular
types of organisms, such as the coliform group.
The critical step in development of a colony
from a single bacterium is the onset of cellular
division, and it is not unreasonable to expect that
this delicate process could be affected by the ex-
tent and nature of the contact of the organism
with the solid, and the extent and thickness of the
nutrient film surrounding the organism. Further,
nutrient supply by diffusion of medium and re-
moval of subsequent metabolic waste products
must be a function of membrane structure and
pore morphology.
With these factors in mind, we began this
study with the objective of defining the optimum
membrane structure for growth of coliform bac-
teria.
46
-------
MATERIALS AND METHODS
Membranes were obtained from a variety of
commercial sources and from our experimental
membrane development activities. The surface
structures of these were characterized using a
Coates & Welter CWICSCAN 100-4 scanning elec-
tron microscope. Before observation, the mem-
branes were coated with a 100 - 200 A layer of
gold.
Fecal coliform and total coliform determina-
tions were performed in accordance with Standard
Methods (1), Sections 408 A and B with the fol-
lowing modifications: To achieve the closest pos-
sible similarity between membrane tests and
streak plate control, the membranes were plated
on a 0.34 cm thickness of agar medium in 47 mm
petri dishes; each streak plate was prepared by
spreading a 0.1 ml aliquot of sample onto a 0.34
cm thickness of agar in a 90 mm dish. The reason
for using a controlled thickness of agar is that we
had found, in earlier experiments, that fecal coli-
form recovery is a function of agar thickness (2).
M-FC Agar and M-Endo Agar were obtained
from the BioQuest Division of Becton, Dickinson,
and Company. Plates were stored at 5 C and were
used within 48 hours of preparation. Fecal coli-
form plates were incubated at 44.5 C ± 0.2 C in
Blue M waterbaths equipped with calibrated re-
cording thermistors. Total coliform plates were
incubated at 35 ± 0.5 C in circulating air incuba-
tors.
Most of the water samples were untreated
sewage, obtained from the masher section of the
Billerica, Massachusetts, Sewage Treatment Plant.
River samples were also used. Samples were stored
at 5 C and were used within 30 hours of collection.
Some refinements of technique were needed
to allow us to run experiments involving large
numbers of samples. Initially, it was found that
noticeable die-off occurred in 15 minutes when the
source water was diluted with phosphate buffer.
The use of 0.1% buffered peptone, however,
stabilized the count for a period of one hour (2).
We also found that it was important to restrict
the time between plating and incubation to 15
minutes or less. The complete procedure was then
as follows: A preliminary count was obtained when
the sample was taken. The following day, a dilu-
tion was prepared to give a count of 200 1,000
bacteria/ml, using buffered peptone diluent. The
diluted sample was mixed for 30 minutes on a
mechanical shaker. Then groups of about 18 mem-
branes and 9 streak plates were prepared from 0.1
ml aliquots, plated, and incubated. This was re-
peated throughout the experiment. Using this
method, up to 100 membranes plus associated
streak plate controls could be run within the one
hour limit. To confirm fecal coliforms, typical
blue colonies were transferred into Lauryl Tryp-
tose broth and then into EC broth.
Surface Pore Morphology
Membrane filter structure can be character-
ized by several parameters. The retention pore size
is a measure of the smallest particle which is re-
tained by the structure, and is best measured by
direct determination of passage of particles (or
microbes) of known size. This technique is des-
cribed by Rogers and Rossmoore (3).
In the present investigation, we were inter-
ested not only in bacterial retention but also in
how the bacteria are situated on the membrane.
It is reasonable to expect that the environment of
retained bacteria depends on the retention pore
size as well as the structure of the surface layer
in which they are retained.
Figure 1 gives scanning electronphotomicro-
graphs of a series of eight membranes made from
mixed esters of cellulose. The photomicrographs
show similar structures which differ only in the
size of the openings. In each photomicrograph
relatively large surface openings were characterized
by the surface opening diameters reported on the
Figure. These were determined by direct measure-
ments on each photomicrograph, or in the case of
the smaller size openings by measuring enlarge-
ments of the photomicrographs. The retention
characteristics of these membranes for coliform
organisms were determined by passage tests, as
described in the following section.
In summary, the way in which bacteria are
situated on a membrane is determined by a new
parameter, the surface opening diameter, which is
observable from scanning electron photomicro-
graphs. The retention of bacteria is determined by
the more familiar retention pore size, which is
found from passage tests.
Results
Figure 2 shows fecal coliform counts on the
series of membranes described above. There is a
47
-------
j^sK.
«m«-
*^te$
&ESti,
M
s£2?«&
0.7,um
2.0 ftm
3.0//m
0.8,um
2.4yum
4.0yum
Figure 1. Scanning Electron Micrographs of a Series of Mixed Ester of Cellulose
Membranes. Numbers shown are Surface Opening Diameters.
48
-------
I^U
1
o
o
I 80
o
LJ_
5 60
o
0
< 40
o
LLt
LL.
or>
A)
n
T
/^\
/ \
f
/
tL
I
I
MEAN OF 5
REPLICATES AND
95% CONFIDENCE
LIMITS. SEWAGE
SAMPLE.
SB \G^X'*'
' . . ^bS-'-"'' ,
0 1.0 2.0 3.0 4.0
SURFACE OPENING DIAMETER, /u m
STREAK
PLATE
Figure 2. Fecal Col i form Count versus Surface
Opening Diameter.
remarkable increase in counts at surface opening
diameters between 1.0 and 2.0 /im. The decrease
in counts at the largest opening size is evidently
due to passage of organisms through this very
coarse structure. The dotted line labeled "passage"
was obtained by re-filtering the effluent through a
bacterial retentive membrane and plating this
membrane on M-FC Agar in the usual way. On the
basis of both growth and passage tests, the opti-
mum membrane structure was determined to have
a 2.4 jum surface opening diameter with smaller
(fecal-coliform retentive) voids of approximately
0.7 urn internally. Results of this plus three other
fecal coliform runs are given in Figure 3. In all four
runs, the abrupt increase in recovery at a surface
opening diameter of 1.0 to 2.0 Aim is evident,
with the optimum structure - i.e., zero passage
and optimum growth occuring with 2.4 /im surface
openings.
Typical blue colonies were picked for confir-
mation from the 0.7, the 1.4, and the 2.4 /im sur-
face opening membranes. The ratios of confirmed/
picked were 18/20 for the 0.7, 19/20 for the 1.4,
and 17/20 for the 2.4 Aim surface opening mem-
brane.
SEWAGE SAMPLE
T SEWAGE SAMPLE
A MILL RIVER
D CHARLES RIVER
EACH POINT IS AN
AVERAGE OF 5
REPLICATES
FFpAl 1 1
ru^rtL ' * ~
COLI-
FORM 1<2
COUNT 1 Q
FECAL
COLI- 0.8
FORM
COUNT °-6
ON 2.4 Q4
jum
OPEN- 0.2
ING
MEM 0.0
o
n
X""?\
/* \y
T *.
/
/
; /
n/D
- A6 PASSAGE ^ TX*
- 3* >
* ' ' ->^*f '
T
1.0 2.0 3.0 4.0 STREAK
bKAINt RLATE
SURFACE OPENING DIAMETER, urn
Figure 3. Normalized Fecal Coliform Counts versus
Surface Opening Diameter.
To compare the total and fecal coliform
effects more directly, a culture was prepared from
a typical fecal coliform colony from the Mill River
source. This was run on M-FC agar and M-Endo
agar using the same series of samples, all from the
same dilution. Results are presented in Figure 5.
Here, the total coliform test shows only a very
slight surface opening size effect while the effect
is considerably more evident in the fecal coliform
test. Evidently, while the fecal coliform test re-
quires a surface opening diameter of 2.4 ^m for
optimum growth, the total coliform test is less
demanding and is performed well on membranes of
surface opening diameter in the range, 1 to 3
Figure 4 presents the results of two total
coliform experiments on the same series of filters.
Here, a light effect may be observed occurring
only at the smallest and largest surface opening
sizes.
At this pointjt appeared that surface opening
diameter was definitely a primary determinant of
fecal coliform recovery. However, other factors,
such as chemical composition and methods of
sterilization, remained to be investigated.
49
-------
TWO SEWAGE
SAMPLES. EACH
POINT IS AN
AVERAGE OF 5
REPLICATES.
50
o
°40
O
IJ 30
o
o
^ 20
10
0
0 1 0 2.0 3.0 4.0 STREAK
PLATE
SURFACE OPENING DIAMETER,// m
Figure 4. Total Coliform Count versus Surface
Opening Diameter.
INCUBATION ON
MF-ENDO AGAR, 35°
INCUBATION ON
M-FC AGAR, 44.50
40
30
ID
o
o
20
o
o
10
0 1.0 2.0 3.0 4.0 STREAK
SURFACE OPENING DIAMETER,yw m PLATE
Figure 5. E. Coli Counts versus Surface Opening
Diameter. A Comparison of Total and
Fecal Coliform Tests.
Effect of Chemical Composition
In the foregoing set of tests, membranes
employed were composed of mixed esters of cellu-
lose. A second series of experiments were designed
employing cellulose acetate membranes. Cellulose
acetate has a much smaller affinity for proteins,
and presumably bacteria, than does the mixed
esters material used in the previous tests. Thus, if
surface adhesion affects growth, a difference be-
tween the acetate and mixed esters results should
be evident.
In the next experiment, recovery on the 2.4
Aim (surface opening) mixed cellulose esters mem-
brane was compared with that of a 3.8 fj.m (surface
opening) cellulose acetate membrane. In addition,
an experimental non-cellulosic membrane composi-
tion having a 3.0 ;urn diameter surface openings was
included. Results are given in Figure 6. Here, we
have plotted the actual counts on each of five repli-
cates, and the passage count obtained by re-filter-
ing the effluents from each. The results show very
little difference in count between the three mem-
brane compositions. These results suggest that
COUNT ON EACH
MEMBRANE
x PASSAGE ON EACH.
MEMBRANE
o
o
50
30
20
< 10
o
LLJ
"- o
REPLICA
NO.
MIXED
ESTER
OF
CELLULOSE
xx x x
X
%
12345 12345 12345 12345
/ / / /
CELLULOSE NON- STREAK
ACETATE CELLULOSIC PLATE
Figure 6. Fecal Coliform Count on Membranes of
Three Different Compositions.
50
-------
membrane composition is not an important factor
in fecal coliform recovery.
Effect of Sterilization Method
Several authors (4, 5) have suggested that
bacterial recoveries may be affected by the method
of sterilization. They did not, however, present
data derived from comparing identical membranes,
where the only variable was the method of sterili-
zation. To test for possible sterilization effects,
membranes were selected from the group exhibit-
ing optimum growth characteristics (2.4 /im sur-
face openings). These membranes were then
divided into four groups using random sampling
techniques. One group was left unsterilized, one
was autoclaved at 121 C for 15 minutes, one was
exposed to ethylene oxide using a standard sterili-
zation cycle * and was aerated three days, and the
fourth group was sterilized by irradiation at a dose
of 1.0 megarads using gamma rays from a cobalt
60 source. Mean counts and 95% confidence limits
on the means are given in Table 1, and are discus-
sed in more detail in our other paper (2). There are
no significant differences between counts on the
unsterilized membranes and counts on the mem-
branes sterilized by the three methods used.
TABLET EFFECT OF STERILIZATION ON
MIXED CELLULOSE ESTER
MEMBRANES HAVING 2.4;um
SURFACE OPENING DIAMETER
Total Coliform
Count**
Unsterilized
Ethylene Oxide
Sterilized
Autoclaved
Irradiated
Streak Plate
42 ±6
44 ±6
38 ±6
40 ±6
50 ±6
Fecal Coliform
Count**
99
103
108
94
82
±9
±9
±9
±8
±6
Two different sewage samples were used. Each
mean is an average of five replicates.
The cycle used a two hour exposure of 12%
ethylene oxide at 130 F and 60% relative
humidity (6).
Discussion
The data collected to this point strongly
suggest that neither chemical composition nor
method of sterilization has any significant effect,
but that the primary determinant of fecal coliform
growth on a membrane filter is that of the surface
pore morphology (specifically with respect to the
size of upper surface openings).
We speculated that since surface effects are
strongest at surface void sizes which are close to
coliform dimensions, some sort of fit of the organ-
ism into the pore might be required for optimum
growth. In particular, the mechanism of the effect
could be that organisms which are deposited on
very fine surface structures are incompletely sur-
rounded by nutrient, while ones that fit into sur-
face openings can be cradled below the level of
nutrient that is drawn up by capillary forces.
Because of evaporation, an incompletely surround-
ed bacterium might be subjected to a locally
hypertonic solution, with resulting plasmolysis
and death. This effect would be particularly evi-
dent at the elevated temperature (44.5 C) of the
fecal coliform test.
To test this hypothesis, three methods of sup-
plying nutrient were compared. The 0.7 ;um and
the optimum 2.4 p.m surface opening cellulose
ester membranes were used. One set was plated
in the standard manner, one set was plated face
down on the M-FC Agar, and the third set was
plated right side up with 2.0 ml of M-FC Agar
overlayed onto each membrane. Results are sum-
marized in Table 2.
Due to the confluence of colonies, accurate
counts could not be obtained from the membranes
placed face down. However, it was clear that the
number of colonies on membranes having the
smaller surface openings (0.7 /jm) was substantially
increased by placing face down. Overlaying these
membranes gave a dramatic increase in counts. The
increase in growth thus seen from inverting the
filter, plus the close agreement in counts of the
two membrane groups when the lower yield filters
were overlayed with nutrient, gives strong evidence
that complete nutrient coverage of the organisms
is required and that this is achieved only with
larger surface opening sizes.
During the comparison testing of the mem-
branes for the 0.7 /zm and 2.4 jum surface opening
groups, some additional benefits were noted rela-
51
-------
Figure 7. A Fecal Coliform Organism Cradled within the Surface Opening
Diameter of a Type HC Filter (15,500X).
TABLE 2 EFFECT OF PLATING METHOD AND PORE SIZE FECAL COLIFORM TEST,
SEWAGE SAMPLE
Membrane Plated and
Overlayed
Streak Plate
Mean Counts and 95% Confidence Limits
Method of Plating
Membrane Plated in
Standard Manner
Membrane Inverted on Agar
Membranes with
0.7 MTI Surface Openings
14±3
Approx. 30
Membranes with
2.4 nm Surface Openings
44+ 10
Approx. 45
46 ±7
53 ±8
35 ±5
52
-------
tive to the latter. These predictable, but nonethe-
less important, phenomena were an increase in the
flow rate through the membrane, an increased dif-
fision rate of media to the membrane surface and,
significantly, increased capacity to filter large
volumes of water particularly those where algae
or other collodial turbidity would otherwise limit
the sample size.
In summary, the factors expected to have an
effect on fecal coliform recovery were investi-
gated. The only one showing a significant effect
was that of surface pore morphology. The evidence
suggests that unlike other organisms, fecal coli-
forms specifically must be cradled slightly below
the membrane surface for optimum recovery. This
suggests an optimum membrane structure with
surface pores slightly larger than the fecal coliform
organisms, but with internal bacterial retentive
pores. See Figure 7.
Until now, membranes recommended for
bacterial testing have been specified by a retention
pore size of 0.45 Aim. Typical 0.45 Aim retention
membranes have surface opening diameters of
1 to 2 jum. As can be seen in Figures 2 and 3, a
slight shift of position on the curve in the range of
1 to 2 /zm surface openings can have a large and
significant effect on recovery.
Since membranes of different manufacturers,
all having 0.45 jum retention size, may exhibit
differences in surface morphology (i.e., in relative
surface opening diameters) they may also exhibit
considerable differences in fecal coliform recovery.
A change to the optimum 2.4 /im surface
opening size will not only provide higher fecal
coliform counts, but will also lead to a smaller
sensitivity to small differences in surface morpho-
logy.
For the total coliform test (Figure 4 and 5),
however, membrane performance is not sensitive
to surface morphology (except in the range below
1 /urn surface opening size). The new 2.4 Aim sur-
face opening/0.7 ^m retention pore size membrane
developed in this work should be regarded as an
improvement for fecal coliform tests, and may also
be used for total coliform with results equivalent
to 0.45 Aim retention membranes.
REFERENCES
1. Standard Methods for the Examination of
Water and Wastewater. 13th Ed., APHA,
1971.
2. Sladek, K.J., C.F. Frith, and R.A. Cotton.
Statistical Interpretation of Membrane Filter
Bacteria Counts. This Symposium, 1975.
3. Rogers, B.G., and H.W. Rossmoore. Deter-
mination of Membrane Filter Porosity by
Microbiological Methods. Dev. Ind. Micro-
biol. 11,453, 1970.
4. Presswood, W.C., and L.R. Brown. Com-
parison of Gelman and Millipore Membrane
Filters for Enumerating Fecal Coliform
Bacteria. Appl. Microbiol., 26, 332, 1973.
5. Dutka, B.J., MJ. Jackson, and J.B. Bell.
Comparison of Autoclave and Ethylene
Oxide-Sterilized Membrane Filters Used in
Water Quality Studies. Appl. Microbiol.,
28,474, 1974.
6. Kereluk, K., and R.S. Lloyd. Ethylene Oxide
Sterilization. J. Hospital Research, 7, 7,
1969; Cycle is shown in Figure 32, p.67.
QUESTIONS AND ANSWERS
Hufham: We did a study which we are not quite
through with yet, in which we washed
membranes in weak sodium bicar-
bonate at 50 C, then in distilled water,
and then reautoclaved. We tried to see
if we could get an inhibitor out, and
got an increase of 50% on our counts
with your particular filter. We were a
little afraid however, that we were
distorting the filter. I wonder if you
would comment on a procedure like
this; might we be making the pore
sizes bigger as you describe in your
paper?
Sladek: It's possible that something like boil-
ing or strong autoclaving can have a
slight effect on- the surface morphol-
ogy of the membrane.
Ginsberg: I, and perhaps others, don't quite
understand what the difference is be-
tween the pore size and the surface
opening size?
Sladek: I am glad that you asked that, because
I would like to make it absolutely
clear. With filtering to remove
particles or bacteria we are interested
in retention of particles and bacteria.
That is the way we characterize our
material, by retention pore size.
53
-------
Bordner:
Sladek:
Bordner:
Sladek:
Grasso:
Sladek:
The way you measure retention Grasso:
pore size is by retaining something.
That is the only certain way to char-
acterize a membrane for retention, by
performing a passage test. Generally
these are done with microorganisms
since the size distribution is quite
narrow. I can give you some refer-
ences to the procedure. This is the re-
tention pore size; it's something that
is experimentally determined. You
can only speculate how this relates
to the structure. You can't really
make a microscopic study using a
scanning electron microscope and
say what the retention pore size is.
I would like to view the retention
pore size and the surface morphology
characterized by surface opening
diameter as two separate and inde-
pendent parameters characterizing a
membrane. In fact, they aren't en-
tirely independent. As you see, one is
constrained as to what you can
manufacture, so that as you increase
the surface opening diameter you
also increase the retention pore
size, generally speaking, for any given
type of membrane.
Your remarks are certainly interesting
and I think you have given us a lot of
new information. I don't know
whether to go home and pour selec-
tive agar over all my membranes or
throw them all out and buy this new
proposed material that you describe.
Do I understand that these are experi-
mental materials with the larger sur-
face porosity that you are talking
about, not the ones that I have been
buying recently?
G rasso:
Sladek:
If you give the right answer I might
not come back tomorrow and give
my paper. Do you feel that with this
increased pore size the temperature
effect and all the other variables men-
tioned today are overcome? In other
words, you could use this new filter
with the larger surface pore size and
with the M-FC broth and overcome
these difficulties and problems?
Yes, I think you have asked me if this
is the cause of all the controversy?
Right.
I think to a large extent, yes. In the
figure that I gave on the fecal coliform
count versus the surface opening
diameter, this slope was very steep on
the left. This means that different
membranes that are manufactured by
various companies striving for the
same retention pore size were not con-
trolling the surface morphology dir-
ectly. The surface morphology was
down in a range where it was a very
sensitive parameter with regard to the
fecal coliform test. I think we can
look forward to a period when all
those differences will go away, be-
cause now we move up further on the
curve towards the peak.
Are these experimental membranes
going to be available?
As far as I know they will be very
shortly.
Sladek:
These materials that I showed you
were made specifically for this test.
Now they do have a lot of similarities Sladek-
to present 0.45 /inn membranes.
Then are we talking about the possi- Dawson:
bility of a membrane filter formulated
specifically for fecal coliforms?
Yes, indeed we are.
Seidenberg: How much vacuum did they use to
pull those samples through, and what
is your recommended vacuum?
Let me refer this question to our
bacteriologist, Mr. Dawson.
I don't think vacuum has any tremen-
dous effect on whether or not an or-
ganism is impinged deeper into a
smaller pore or remains near the sur-
face. It's been known for a long time
that the gram negative microorgan-
54
-------
isms are extremely sensitive to hyper-
tonic solutions. Indeed this is the
recommended method for preparing
protoplasts from gram negative organ-
isms. I believe in our laboratory we
normally use something like 14"
vacuum for running samples. I've been
working with membranes since about
1955 and I haven't seen any effects
that were due to vacuum.
Brezenski: I think I recall one paper that dis-
cusses cavitation and the effect of
vacuum on coliforms, so I don't quite
believe what you said.
Levin: Two questions really one is on the
steep part of the curve, where you
were talking about the surface effects.
I didn't really catch it. When you
boil or autoclave you go up in terms
of improving recovery or do you go
back down in terms or decreasing
recovery after treating them in your
own laboratory?
Sladek: We have found very little effect of
either procedure. I showed you some
data on autoclaving.
Levin: It really doesn't make any difference?
Sladek: No.
Levin: The second question is only indirectly
related but since you're a statistician
I thought that I would put you
on a spot, if I could. You've given
data in two papers and your counts
have ranged from as low as 9 or 10 up
to about 120 for the average count
and yet we think of 20 to 80 as being
the optimum numbers. I'm wondering
where you would draw the line? For
instance, if I'm doing a one hun-
dredth, a tenth, one ml and 10 ml,
and my 10 ml averages 12, and my
1 ml averages 100, which one do I
believe and have the most faith in?
Sladek: Yes, let me comment on that. I'm sure
that you realize in terms of the statis-
tics that you want to avoid low
numbers, so it is to your advantage
to get up into the higher numbered
range. We have not established for
certain whether there is an upper
cutoff beyond which you should go.
We have had good results going up to
about 140. I mean "good" in the
sense that the scatter didn't increase
extraordinarily. I've seen data up
beyond 200 where the scatter was
really terrible and I am not sure how
safe it is to conclude that above 200
something goes wrong . . . perhaps
someone else has some information
on that.
Litsky: Did you examine brand X for cavita-
tion and if so, what did you find?
Sladek: You are asking if we examined other
brands of membrane filters so far as
their surfaces go? Let me answer that
in this way. I would like to refer to
my earlier paper. I stressed there the
idea of random sampling. Random
sampling is a very important concept
in not producing bias in your data. We
have tested what I will call a few
boxes of competitor's membrane fil-
ters. I will not tell you the results be-
cause I know that this does not repre-
sent a random sample of their produc-
tion. I can tell you the results that we
have obtained on Millipore filters be-
cause we are very careful to take a
representative random sample of our
production and we base results that
we report on that. I can't really com-
ment in public about other manufac-
turers' filters because I don't have a
proper random sample of their pro-
duction.
Litsky: May I take my prerogative as the
chairman and ask any other represen-
tatives, Gelman, Johns-Manville or
anyone else, if they examined their
filters and observed the cavitation
effect?
Sladek: Please don't call it a cavitation effect.
Litsky: Litsky stands corrected.
Brezenski: I was going to mention that cavitation
isn't what we are talking about here. I
want to get back to surface morpho-
55
-------
Sladek:
Brezenski :
Sladek:
logy and the reasons that you gave
why there was no effect on the re-
covery of the total coliforms, yet
there was a decrease in recovery. You
assumed this because of the surface
morphology and that fecal coliforms
stay on the surface and don't get
deep down into the layers where they
can get more medium. I haven't seen
any data that you've presented which
shows specifically that surface mor-
phology effects the fecal coliform
and not the total coliform. Do you
have any data which shows the total
coliforms in the upper layers and the
fecal coli, or E. coli in the bottom
layers? This seems to be the crux of
your explanation.
No. This is not the crux of my explan-
ation.
Oh, I
that?
am sorry. Would you clarify
The organisms are of course the same
size. They are from the same group.
We are speaking of what happens to
an organism that is filtered on a very
fine surface structure, and ends up
"on a moutain top". The explanation
was that because of evaporation, hav-
ing to do with dehydration rates, va-
por pressures, etc. a fecal coliform or-
ganisms may be in contact with a pool
of nutrients, which will shrink in size
and consequently become concen-
trated, or locally hypertonic, causing
plasmolysis of the organism. If this is
indeed the mechanism, you would
expect it to be much less pronounced
at a lower temperature where the
evaporation rate is lower. In trying to
tie all of these things together, we
have shown rather conclusively that it
is the way in which nutrient is sup-
plied that is the origin of the effect.
We did this by turning the membrane
over and pouring agar on top of it,
etc. We have also shown that the
effect is very strong in the fecal coli-
form tests, but rather weak in the
total coliform test. The way we
bridged the gap and put all this back
together has to do with the evapora-
tion rate.
Winter: I just have one quick question. I can't
really quite understand evaporation as
the reason, or should I say the hypo-
thesis, because we are dealing with a
saturated humidity chamber whether
you are on M-FC in a bag or in a super
tight fitting plate. I wonder whether
this really has an effect. Aren't
we really reaching a bit to explain
why something happens that we don't
know anything about?
Secondly, although you may not have
tested many Gelman or other filters,
it would seem ironic that some of
the other manufacturers are ahead of
you, because they, by some process,
were able to prepare a membrane
with larger surface openings, hence
large recoveries. If we just look at the
summary of results they seem to con-
clude that under present manufactur-
ing techniques, Gelman seems at least
to have given higher recoveries by the
majority of experimenters and no one
has explained this fact. We may have
some random error occuring, and we
know we have systematic error in
everyone's work, but this is random-
ized, by the number of manufacturers
putting out membranes and the
number of people doing the work,
when we have had perhaps as many as
50 people coming up with results
which all tend to point in the same
direction. I am wondering aren't we
reaching a bit at this time?
Sladek: Let me respond to your first question.
Aren't we reaching a little in the ex-
planations? Yes, indeed we are reach-
ing in the explanation. Concerning
your comment that it is a closed
system, with respect to water vapor,
there are always some small gradients
in these systems and water vapor does
not move around inside the plate, so
that is about the only support that I
can give to the idea that small changes
in concentration can occur.
Again, with regards to your other
question, I would have to defer any
56
-------
comments on what the real status of
different membrane manufacturers'
membranes are, pending a study. I
believe the D-19 round robin study
has been completed and I certainly
would not accept your statement that
everyone agrees that this kind of filter
is better than that kind, because
everyone doesn't.
57
-------
A COMPARISON OF MEMBRANE FILTERS AND MEDIA
USED TO RECOVER COLIFORMS FROM WATER
M.H. Brodsky and D.A. Schiemann
Ontario Ministry of Health,
Laboratory Services Branch
Environmental Bacteriology Laboratory
Toronto, Ontario. Canada M5W 1R5
ABSTRACT
Many laboratories involved in water analysis
are using membrane filtration methods for the
enumeration of pollution indicator organisms in
water. The Ontario Ministry of Health, Laboratory
Services Branch, analyzes approximately 350,000
water specimens annually, almost exclusively by
membrane filtration. We have observed that there
are pronounced differences in the abilities of the
filters produced by various companies to recover
coliforms. This paper reports the results of an in-
vestigation which evaluated three brands of mem-
brane filters. Seven of our laboratories participated
in this study. Parallel analyses for total coliforms
from routine water samples were performed using
filters supplied by the Johns-Manville Company of
Canada (045 MO 47SG), the Millipore Corporation
(HAWG 47SO) and the Sartorius Company
(11456). Statistical evaluation of the results
indicated that the Johns-Manville and Millipore
filters were equivalent and much superior to the
Sartorius filters for the enumeration of coliforms
from water.
LES Endo agar is the only solid medium re-
cognized by Standard Methods for the Examina-
tion of Water and Wastewater (13th ed. 1971) for
the direct recovery of coliforms from water. M-
Endo broth media are also recognized by Standard
Methods for use in membrane filtration; however,
it is recommended that these broth media prepara-
tions be used with sterile pads. This latter proce-
dure as outlined in Standard Methods, adds an
additional time factor to the processing of each
water specimen which would present difficulties to
a high volume laboratory such as ours.
The relative cost of these two types of Endo
preparations as well as the problems created by the
recently experienced shortages of Endo based
products prompted us to compare LES Endo agar
with various M-Endo broths with agar added for
coliform analyses.
INTRODUCTION
Specifications for membrane filters used for
the bacteriological analysis of water are presented
in the 13th edition of "Standard Methods for the
Examination of Water and Wastewater".
All manufacturers of membrane filters claim
or intimate in product advertising that their
filters meet the criteria specified in Standard
Methods for bacterial recovery. Recent investiga-
tions indicate that, despite manufacturers' claims,
there exists considerable variation among com-
mercial brands of membrane filters in their ability
to recover coliform and faecal coliform organisms
from water. There is, however, some disagreement
concerning experimental designs and statistical
evaluations used in these studies. Consequently,
any general conclusions to be drawn from such in-
vestigations must be guarded.
The Laboratory Services Branch of the
Ontario Ministry of Health analyzes more than
350,000 water specimens annually, almost ex-
clusively by membrane filtration. Our laboratory
personnel had also observed inconsistencies in
coliform and faecal coliform enumeration on mem-
brane filters produced by various companies. As a
result of these observations we carried out a series
of investigations to quantitatively compare coli-
form and faecal coliform recoveries on three
brands of membrane filters Johns-Manville, Milli-
pore and Sartorius.
58
-------
MATERIALS AND METHODS
The comparative evaluations were done in
three phases. Phase 1 was a preliminary field study
involving five regional laboratories. Phases 2 and 3
were more rigidly controlled investigations con-
ducted solely in our central laboratory.
Membrane Filters:
The Millipore filters (Catalogue No. HAWG-
047SO) and the Sartorius filters (Catalogue No.
11456) used throughout the investigation were
obtained from the Ministry of Health Laboratory's
stock supplies. The Johns-Manville filters (Cata-
logue Nos. 045M047SG and 045M047LG) were
supplied by the company.
Three different lot numbers of the three filter
brands, pre-sterilized by ethylene oxide, were in-
cluded for comparison in the first and second
phases of this investigation. For the third phase,
unsterile Johns-Manville filters were obtained and
autoclaved in our laboratory.
Source of Cultures:
In the preliminary phase 1 of the study we
used routine water samples. Appropriate dilutions
of the samples were filtered in duplicate for total
coliforms only. All the participating laboratories
did not stock the same brands of filters. Three
laboratories compared Johns-Manville with Sar-
torius filters and two others compared Johns-
Manville with Millipore filters.
Similar routine water samples were used for
the second phase of this study. After being
analyzed by our routine procedure for total and
faecal coliforms, these water samples were refrig-
erated overnight. The following day, those samples
having at least 20 faecal coliforms per 100 ml
were selected for further processing. Ten ml of
each of these samples were added to 10 ml of
double-strength MacConkey broth in screwcapped
fermentation tubes. Following incubation for 24
to 48 hours at 35C, two loopfuls of each positive
broth culture were subcultured into EC broth. The
EC broths were incubated at 44.5C for 22 to 24
hours. Five replicate filtrations of a dilution of
each EC broth culture, standardized by optical
density, were performed for both total and faecal
coliforms for each of the three brands of filters.
In phase 3 of the study, we collected eight,
one-litre samples of water from a known polluted
surface source, the Number River, over a one
week period. On the day of collection, a pre-
screening membrane filtration was done on each
sample, to determine coliform and faecal coliform
densities. Based on the screening densities, appro-
priate test dilutions of the refrigerated samples
were prepared to provide 20 to 80 total coliform
colonies and 20 to 60 faecal coliform colonies per
filter. Ten replicate filtrations per sample per
brand of filter were completed on each water
sample for total and faecal coliforms.
Cultural Techniques:
Throughout the study, M-Endo MF broth
(Difco) with 1.5% agar added was used for total
coliform recovery, and M-FC broth base (Difco)
with 1.5% agar added was used to culture faecal
coliforms. These solid media were prepared
in 15 x 150 mm plastic petri plates, which accom-
modate 5 filters. Incubation times and tempera-
tures were as specified in Standard Methods. The
M-FC plates were heat sealed in waterproof plastic
bags before being immersed in a constant tem-
perature water bath at44.5C.
Statistical Analysis:
The Students t test for comparison of means
was used to evaluate the results of the preliminary
field study. Eighty-two comparisons between the
Johns-Manville and Sartorius filters were tabulated.
Sixty eight comparisons between Johns-Manville
and Millipore filters were analysed.
Analysis of variance (ANOVA) was applied to
the results of phases 2 and 3 of this investigation.
When the F ratio indicated a significant difference
in the means at the 5% significance level, a multi-
mean comparison test (the Tukey Test) was used
to determine where the difference occurred.
RESULTS AND DISCUSSION
Our results clearly demonstrate that varia-
tions in experimental design can lead to very dif-
ferent conclusions regarding the superiority of one
MF brand over others. Tables 1 and 2, summarizing
the statistical analysis of the preliminary field
study, show that the Johns-Manville filters were
superior to Sartorius filters but equivalent to Milli-
pore filters for total coliform recovery. But the re-
sults of phase 2 conflicted with this conclusion.
Recovery of faecal coliform isolates on m-Endo
medium (Table 3) concluded that the three brands
59
-------
Table 1: Total coliform recovery from routine
water samples in three laboratories with
Johns-Manville and Sartorius membrane
filters.
No. Comparisons
Means (x)
Standard
Deviation
t-Test
Analysis
Johns-Manville Sartorius
82 82
14.9 8.1
15.50 11.94
t = 3.17
(t .05, 162= 1.92}
of filters were equivalent; but, recovery on M-FC
medium concluded that Johns-Manville filters
were superior to both the Millipore and the Sar-
torius filters (Table 4). There was no significant
difference in faecal coliform recovery between
Millipore and Sartorius filters.
We attempted to resolve this conflict of data
by the experimental design of phase 3 using natural
water samples. Statistical analysis of total coliform
recovery by the three filters (table 5) supports the
findings of the preliminary study, i.e. the Johns-
Manville filters were superior to Sartorius filters
but equivalent to Millipore filters for total coliform
recovery. As in the preliminary study, these results
conflict with the results of phase 2. Similarly,
statistical analysis of faecal coliform recovery
rates from natural water samples (Table 6) indicate
that there was no difference among the three
Table 2: Total coliform recovery from routine
water samples in two laboratories with
Johns-Manville and Millipore membrane
filters.
Johns-Manville
No. Comparisons
Mean (x)
68
192.3
Millipore
68
184.5
Standard
Deviation
t-Test
Analysis
499.60
t= .05
(t.05, 134= 1.98)
335.80
Table 3: Recovery of faecal coliform isolates on
M-Endo medium with Johns-Manville,
Sartorius and Millipore membrane filters.
Johns-Manville Sartorius Millipore
No. Comparisons 100
Mean (x) 52.0
Standard
Deviation
30.70
100
44.5
30.87
100
43.7
32,28
Analysis of Variance (ANOVA)
Source SS df MS F ratio3
Within 290885.70 297 979.31
2.14
Between 4194.06 2 2097.03
a F .05°° = 3.00
brands of filters at a significance level of .05. This
finding also disagrees with phase 2 of the study,
which used laboratory cultures of faecal coliform
isolates. Similar disagreements have been described
by other investigators. Presswood and Brown (4),
and Harris (3) concluded that Gelman membrane
filters were superior to Millipore membrane filters
for recovery of E. coli. Schaeffer et al (5) dis-
agreed with their statistical conclusions. Schaeffer's
group found when using natural water samples,
that Gelman and Millipore filters were equivalent
for faecal coliform recovery, but, that Gelman
filters were superior to Millipore filters for total
coliform recovery.
In a recent paper, Dutka et al (2) reported
conflicting results in two studies employing the
same experimental design. Field samples and broth
cultures of E. coli ATCC 25922 were used for
comparative recoveries with autoclaved and ethy-
lene oxide sterilized filters. The results of their first
study (March 1973) concurred with findings of
Schaeffer et al (5). At a significance level of .01,
Gelman and Millipore filters were equivalent and
superior to Sartorius filters for faecal coliform
recovery; however, Dutka et al (2), also reported
that both Millipore and Sartorius filters were
60
-------
Table 4. Recovery of faecal coliform isolates on
M-FC medium with Johns-Manville, Sar-
torius, and Millipore membrane filters.
Johns-Manville Sartorius Millipore
Table 5. Total coliform recovery from Humber
River water samples with Johns-Man-
ville, Sartorius, and Millipore membrane
filters.
No. Compariso
Mean (x)
Standard
Deviation
JUIIIIb-IVIdllVII
ms 100 100 100
No. Comparisor 80
40.5 30.3 33.1
Mean (x) 49.1
32.47 26.75 28.19 Standard
Deviation 25.81
lie oariurius iv _uuie
80 80
34.7 42.7
22.01 27.12
Analysis of Variance (ANOVA)
Source SS df MS F ratio3
Within 25387.02 297 854.78
3.22
Between 5496.17 2 2748.08
Multimean Comparison Test (Tukey Test)
Tukey Johns-Manville Millipore vs
Calculation13 vs Millipore vs Sartorius Sartorius
(Xl-x2)±Tc +17.Q3 to +19.83 to +13.48 to
- 2.33 + 0.47 - 6.88
Conclusion Not Not
(
-------
TableB. Recovery of faecal coliforms from
Humber River water samples using
Johns-Manville, Sartorius and Millipore
membrane filters.
Johns-Manville Sartorius Millipore
No. Comparisons 70 70 70
Mean(x) 30.1 26.7 26.0
Standard
Deviation 12.39 10.15 10.40
Analysis of Variance (ANOVA)
Source SS df MS F ratio3
Within 25150.87 207 121.50
2.75
Between 668.41 2 334.20
a F .05°° =3.00
the mean coliform counts on the Johns-Manville
filters were statistically greater than on Millipore
and Sartorius filters. The autoclaved Johns-Man-
ville filters however, were noticably more brittle
and less flexible than the ethylene oxide ster-
ilized filters. We also observed distortion of the
faecal coliform colonies on autoclaved filters.
An additional undesirable features of the
Johns-Manville and Millipore filters was the inhibi-
tion of growth by grid markings. Colonies growing
near the grid lines developed flat edges, conforming
to the restrictions imposed by the lines. Colonies
which straddled the lines were split. Interestingly,
grid line interference was more pronounced with
total coliform than with the faecal coliform
colonies.
We also noted, as did Dutka et al (2), that the
Sartorius filters had irregular hydrophobic areas
which became evident when the filters were wet-
ted. These areas of reduced permeability likely
contributed to the decreased bacterial recovery we
observed with Sartorius filters. This was brought
to the attention of the Sartorius representative
about six months ago and we have yet to hear a
satisfactory explanation of why this has occur-
red.
We realize that our culture media, M-Endo
MF broth with agar added and M-FC broth base
with agar added, are not recognized as standard
solid media for membrane filtration by Standard
Methods. LES Endo agar is the only solid medium
accepted for the direct recovery of coliforms
from water. However, we have conducted a com-
parison of coliform recovery on M-Endo MF broth
with agar and LES Endo agar by membrane filtra-
tion of 101 natural water samples (unpublished).
Statistical analysis indicated that there was no
significant difference between coliform recovery
on these two media.
The conflicting conclusions of our investi-
gation and of other similar studies comparing
membrane filters need to be resolved. Experi-
mental designs including source of the test organ-
ism and statistical evaluations, must be standard-
ized so that some logical conclusion regarding
membrane filter performance can be made.
ACKNOWLEDGEMENTS
We wish to acknowledge the technical assist-
ance of Mr. B. Ciebin.
REFERENCES
1. American Public Health Association. Stand-
ard Methods for the Examination of Water
and Wastewater. 13th. ed. American Public
Health Association Inc., New York, 1971.
2. Dutka, B.J., M.J. Jackson and J.B. Bell.
Comparison of Autoclave and Ethylene
Oxide-Sterilized Membrane Filters Used in
Water Quality Studies. Applied Microbiology,
28:474-480, 1971.
3. Harris, F.J. Coliform Recoveries on Mem-
brane Filters. E.P.A. Newsletter No. 22: 4,
1974.
4. Presswood, W.G. and L.R. Brown. Com-
parison of Gelman and Millipore Membrane
Filters for Enumerating Fecal Coliform Bac-
teria. Applied Microbiology, 26:332-336,
1973.
5. Schaeffer, D.J., M.C. Long, and K.G. Janar-
dan. Statistical Analysis of the Recovery of
Coliform Organisms on Gelman and Millipore
Membrane Filters. Applied Microbiology,
28:605-607, 1974.
62
-------
QUESTION AND ANSWER SESSION
Geldreich: You mentioned that the use of the
agar preparation of M-Endo and M-FC
is a non-standard method. I would like
to say that in the next edition of
Standard Methods we have become
so concerned about this problem, that
I have written a paragraph saying that
that you can certainly use it with
1.5% agar, and it will be a standard
method. In fact, when I go out and
do laboratory evaluations I certainly
recommend it in my report.
Brodsky: We ran into problems, getting Endo
base media from various companies.
We were trying to find alternate
procedures and this is how we got
into it.
Brezenski: Could you please summarize. For
example you said for EC positive
organisms that Johns-Manville was
equivalent to Millipore. Would you go
over this? I think this is the crux of
the issue, and I am a little confused
because someplace where the line goes
equivalent from here to here, it's not
equivalent from here back again.
Brodsky: I must admit that I was confused too.
As you realize the study was divided
into 3 phases. Perhaps I should leave
phase 1 out completely, because I
really don't want to base a judgement
on such a loosely controlled study.
Let me do the 2nd and 3rd phases
which are more complete and more
rigidly controlled. In phase 2 we selec-
ted EC positive cultures, but for the
sake of argument we will call them
fecal coliform cultures, and then we
compared their recovery, that is EC
positive cultures on M-Endo media
and on M-FC media in parallel, using
the three same brands of filters. We
did 5 replicates for each filtration.
We determined that Johns-Manville
and Millipore and Sartorius were
equivalent when used with M-Endo
MF medium for these EC cultures.
When we used M-FC medium, obvi-
ously selected for fecal coliforms,
we found that Johns-Manville was
superior to Sartorius; however, Johns-
Manville was equivalent to Mi-llipore,
and Millipore was equivalent to Sar-
torius. Now, there's some logic if you
try and think if A=B, and B=C, then
A must = C. Think of it this way, if
A is 40, and B is 35, and C is 30. The
difference between 40 and 35 is not
significant. The difference between 35
and 40 is not significant, but the dif-
ference between 30 and 40 may be
significant. It is confusing but that is
the best explanation I can give you as
to why this natural logic doesn't
apply.
Then in phase 3 we said "fine." Let's
see what happens now if we use
natural cultures or a natural source of
water, rather than using a laboratory
culture in which we have given it the
best possible condition to grow and
allow them to overcome any possible
inhibition. We were hoping to find
some sort of parallel. Perhaps I
shouldn't say that, because that
isn't fair scientific judgement. What
happened was for total coliforms
from a natural polluted sample such
as the Number River, (I think anyone
from the Toronto area can vouch for
the fact that the Humber River is a
polluted water source) we found that
Johns-Manville was superior to Sar-
torius for total coliforms. If you
recall in the first part we said that
they were equivalent for total coli-
forms using the laboratory cultures.
Johns-Manville and Millipore were
equivalent and Millipore and Sartorius
were equivalent for total coliforms.
When we looked at fecal coliforms we
didn't find any difference at all be-
tween the three brands of filters for
recovering fecal coliforms from a
natural polluted source. We performed
two different studies; one used
laboratory cultures and one used
natural samples. We got two different
results. We have to resolve this prob-
lem. The question that I'm asking
is where do you get your source of
culture? Obviously the source of
culture is going to have a tremendous
influence on conclusi6ns.
63
-------
COMPARISON OF MEMBRANE FILTERS IN RECOVERY
OF NATURALLY INJURED COLIFORMS
by
David G. Stuart, John E. Schillinger and
Gordon A. McFeters
Department of Microbiology
Montana State University
Bozeman, Montana 59715
ABSTRACT
Raw sewage and cultures of E. coli were
exposed to natural stream conditions in test
chambers for 24 hours then recovered and enum-
erated with the M-FC test with and without enrich-
ment and with different brand filters.
ANOVA, F and t statistics showed that the
Gelman filters had only a slight advantage in
recovery over Millipore filters. Both brands re-
covered significantly better than the Nuclepore.
Enrichment gave significantly better recovery with
variation by days. Pure cultures were more sensi-
tive to test and filter variations than were the sew-
age cultures. The major conclusion was that Nucle-
pore membranes should not be used for coliform
analyses in water.
INTRODUCTION
Beginning with Dr. William G. Walter's initial
investigation (1964) into the relative bacteriologi-
cal quality of water produced by adjacent open
and closed watersheds, numerous studies of water
quality in high mountain watersheds have been
carried out in our laboratory using Millipore mem-
brane filters exclusively. Presswood and Brown's
article in 1973 (6) along with papers and discus-
sions at the 1974 American Society for Micro-
biology meetings in Chicago, indicating that Milli-
pore filters might be yielding erroneously low
counts, caused some concern about the data col-
lected over the last 10 years. With recent publica-
tions (2, 4, 7) adding conflicting data and interpre-
tations to the issue, it seemed wise to follow the
advice of Geldreich et al (3) and compare the
performance of different brands of membranes in
our laboratory.
MATERIALS AND METHODS
The preliminary experiments reported here
were performed with suspensions of Escherichia
coli C320 MP 25, isolated from water in our labor-
atory and with raw sewage. Procedures followed
Standard Methods (1).
Aliquots of 24 hour cultures of E. coli were
washed twice with gelatin phosphate buffer and
dilutions yielding 1()5 to 106 cells per ml were
placed in chilled river water, taken immediately to
the stream site, and submersed in the flowing river.
A 1 ml sample was taken from the chamber, placed
in 9 ml gelatin phosphate buffer, iced and trans-
ported back to the lab where dilutions were
plated on TSY agar (35 C) to yield a 0 time count.
After 24 hours of exposure to the natural aquatic
environment to allow injury to occur, a 1 ml
sample of the contents of the chamber was trans-
ported to the lab in 9 ml of iced gelatin phosphate
buffer. One ml of this cell suspension was placed
in 9 ml of Trypticase soy broth + 0.5% glucose +
0.3% yeast extract and incubated at room tempera-
ture for 2 hours before filtering (enriched). An-
other 1 ml was taken from the buffer suspension,
diluted, and filtered immediately (non-enriched).
M-Endo MF medium (35 C) was used with the
E. coli experiments. (mEndoMF-E. coli).
The sewage experiments were performed in an
identical manner except that undiluted raw sewage
was placed in the membrane filter chambers and
64
-------
M-FC fecal coliform medium was used and incu-
bated at 44.5 C (M-FC-sewage).
Colonies were counted with the aid of a bin-
ocular microscope (7X) and reflected light. Counts
were statistically analyzed with regression analy-
sis and analysis of variance utilizing the classifica-
tions: filter, day and enrichment.
RESULTS AND DISCUSSION
Results from three M-FC-sewage runs with 10
replicates with each brand of filter (Millipore,
Gelman and Nuclepore) on non-enriched and
enriched samples along with four identical
mEndoMF-E. coli runs are reported in this paper.
Counts from the 10 replicate plates for each
filter brand generally followed the Poisson distri-
bution. The mean counts on Gelman filters were
slightly higher than those on Millipore filters. Both
Gelman and Millipore filter counts always ex-
ceeded counts on Nuclepore filters.
All counts were transformed using square
roots and analyzed with a computer program for
ANOVA. Both F and t statistics were computed
for various interactions of means by using the
Day-enriched-filter interaction mean square as the
error term. Use of an interaction mean square as
the error term is common in randomized block
designs. Here one would consider a day as a block.
Day and enrichment differences were found
to be significant at the P = 0.005 level for both the
M-FC-sewage and the,mEndoMF-E. coli situations.
Enrichment effect varied significantly with differ-
ent days (i.e. interaction) for both mEndoMF-
E. coli (0.005) and M-FC sewage (0.05). Day by
filter interactions and enrichment by filter inter-
actions were not significant for M-FC-sewage but
were significant (0.005 and 0.05 respectively) for
mEndoMF-E. coli. This would seem to indicate a
greater injury effect for the washed E. coli cells
than for the raw sewage coliforms. The in-stream
conditions varied during the course of the experi-
ments resulting in changing degrees of injury which
are reflected in the above statistics.
Differences among the 3 filters (all treatments
grouped) were significant at the 0.01 level for
M-FC-sewage and at the 0.005 level for mEndoMF-
E. coli. Analysis of all Millipore counts versus all
Gelman counts showed no statistical differences
in either M-FC-sewage (0.2) or mEndoMF-E. coli
(0.5) trials while differences between Millipore
versus Nuclepore counts and Gelman versus Nucle-
pore counts were significant. These differences
were much larger for the mEndoMF-E. coli situa-
tion (P = 0.001) than for the M-FC-sewage differ-
ences (P = 0.01 to 0.005) again suggesting a greater
degree of injury for the E. coli compared to sewage
coliforms, and subsequently, some kind of injury -
Nuclepore filter interaction. If enrichment had
overcome this inhibitory effect with Nuclepore
filters, one would expect Nuclepore versus en-
richment t values to be higher than those of
Millipore and Gelman versus enrichment. This was
not the case as shown by mEndoMF-E. coli t values
of 49 and 54 for Millipore and Gelman versus en-
richment effect and a t value of 41 for Nuclepore
versus enrichment effect. Although the t values
were smaller (3.9, 4.7, 3.2) this interpretation is
corroborated by the results of the M-FC-sewage
experiments.
A breakdown of filter comparisons into non-
enriched and enriched trials showed Millipore
versus Gelman differences to be insignificant for
M-FC-sewage (0.5, 0.4) but were barely significant
(0.05, 0.10) for mEndoMF-E. coli results. Millipore
and Gelman were significantly different from
Nuclepore at the 0.005 level in the mEndoMF-
E. coli situation and from 0.05 to 0.005 in the case
of the M-FC-sewage.
The practical significance of the differences
between filter counts should be examined with an
understanding of day to day changes in the natural
environment and of errors and variation inherent
in membrane filter techniques. For example,
assuming that a set of replicate counts follows the
Poisson distribution, the square root of the mean
will give a reasonable estimate of the standard
deviation one can expect between individual
plate counts. Thus, for the mean of 52.6 observed
for the M-FC-sewage Millipore counts, a standard
deviation of ±6.7 would be expected. The actual
difference between Millipore and Gelman means
was observed to be 7.7 which is about equal to
the expected variation among Millipore counts.
For all practical purposes, these data indicate
that the bias towards higher counts obtained with
Gelman as compared to Millipore filters is small.
The difference between Gelman and Nuclepore
means was 33.9, much larger than the expected
standard deviation of 6.7. Thus, the error when
using Nuclepore filters would be considerable.
65
-------
5.
LITERATURE CITED
American Public Health Association. Standard 7.
Methods for the Examination of Water and
Wastewater, 13th ed. American Public Health
Association, Inc., New York, 1971.
Dutka, B.J., M.J. Jackson, and J.B. Bell.
Comparison of autoclave and ethylene oxide-
sterilized membrane filters used in water
quality studies. Appl. Microbiology. 28:474-
480,1974.
filters for enumerating fecal coliform bacteria.
Appl. Microbiology. 26:332-336, 1973.
Schaeffer, DJ.,M.C. Long and K.G. Janardan.
Statistical analysis of the recovery of coliform
organisms on Gelman and Millipore mem-
brane filters. Appl. Microbiology. 28:605-
607,1974.
QUESTION AND ANSWER SESSION
Geldreich, E.E., H.L. Jeter, and J.A. Winter. Geldreich:
Technical considerations in applying the
membrane filter procedure. Health Lab.
Science, 4:113-125, 1967.
Hufham, J.B. Evaluating the membrane fecal
coliform test by using Escherichia coli as the
indicator organism. Appl. Microbiology. 27:
771-776, 1974.
McFeters, G.A., and D.G. Stuart. Survival of
coliform bacteria in natural waters: Field and Stuart:
laboratory studies with membrane filter
chambers. Appl. Microbiology. 24:805-811,
1972.
Presswood, W.G., and L.R. Brown. Compari-
son of Gelman and Millipore membrane
Dave I think the reason you see so
much difference between the Nucle-
pore and the other two is that the
Nuclepore is not really the same
material. As we tried to say this morn-
ing, if that membrane were to be used,
you would have to redesign a whole
family of media for it.
I remember you saying that, and the
reason that we used it was we don't
use it for bacterial counts but when we
are using algae in our chambers, we use
a Nuclepore sidewall and it ended up
being a pretty nice control.
66
-------
EFFICIENCY OF COLIFORM RECOVERY
USING TWO BRANDS OF MEMBRANE
FILTERS
Frederick L. Harris
and
Carl A. Bailey
U.S. Environmental Protection Agency
Surveillance & Analysis Division
25 Funston Road
Kansas City, Kansas 66115
ABSTRACT
The comparative study of Gelman and Milli-
pore membrane filters by Presswood and Brown
(5) prompted the evaluation of the two brands of
membrane filters employing routine samples. The
study included a total of 100 samples from a
variety of non-chlorinated aquatic sources. Gelman
filters averaged 2.5 times greater recovery of fecal
coliforms than did Millipore filters. A comparative
study, with fewer samples, was also made utilizing
total coliform analyses. Data indicate that total
coliform recovery is similar with the two brands.
For verification as fecal coliforms, some typical
blue colonies were subcultured from both filter
brands to confirmatory media. Different lots of
each brand of filters were used.
INTRODUCTION
Various brands of membrane filters have been
under close scrutiny by several investigators. A
disparity in the ability of different brands of
membrane filters to support the growth of coli-
form bacteria from both natural and stock sources
was found. Using a typical strain of fecal coliform.
Levin et al. (4) observed that Gelman filters
exhibited a much less adverse effect on the micro-
organism than did Millipore and Oxoid filters.
Presswood and Brown (5), utilizing pure strains,
found that Gelman filters recovered 2.3 times
more fecal coliforms than did Millipore filters.
Comparative analyses of river water for fecal coli-
form bacteria gave results comparable to those for
pure cultures. In the study, total coliform recovery
was statistically higher with Gelman filters than
with Millipore filters. In a field study employing
Gelman, Millipore, and Sartorius membrane
filters, Dutka et al. (2) observed that Gelman
filters generally produced the highest counts. Huf-
ham (3) found that Gelman filters demonstrated
higher counts of a strain of typical fecal coliform
at 44.5 C than did Millipore filters; however, both
filter brands showed similar results at 35 C. Using
natural samples with Gelman and Millipore filters,
Shaeffer et al. (6) obtained higher total coliform
counts with Gelman filters. The fecal coliform
counts were similar with the two membrane
filter brands.
Over the past few years, the membrane
filter test has become an official method (1) and a
valuable laboratory tool. However, with the
mounting data of various investigators indicating
membrane filter brand disparity in microorganism
enumeration, doubts have been raised concerning
the present accuracy of the test.
As an in-house quality control measure,
prompted by the results of Presswood and Brown
(5), a study was initiated to evaluate Gelman and
Millipore membranes using routine samples from
non-chlorinated aquatic sources.
MATERIALS AND METHODS
Sample Sources and Sampling.
Water samples were obtained from three types
of sources: (i) aerobic lagoon (influent and ef-
67
-------
fluent); (ii) river; and (iii) sewage treatment plant
effluent. All samples from these sources were
collected in autoclaved sterilized bottles iced en-
route to the laboratory, and processed within 8
hours of collection time.
Procedure, Culture Media, and Reagents.
All procedures, media, and reagents used were
in accordance with those described in Standard
Methods for the Examination of Water and Waste-
water (13th ed.) part 400.
Samples were taken from the same dilution
bottle and filtered simultaneously through each
brand of filter using a Millipore membrane filtering
apparatus with a 3 place-Hydrosol manifold.
Membrane Filters.
Three lots each of two commercial brands of
0.45 Aim porosity membrane filters were used in
the study: Millipore HAWG 047SO (Millipore
Corp., Bedford, Mass.) sterilized with ethylene
oxide by the manufacturer; and Gelman GN-6
(Gelman Instrument Co., Ann Arbor, Mich.)
sterilized in an autoclave by the manufacturer.
Confirmation of Colonies.
Fecal coliform colonies from membranes of
both brands were tested to establish the validity of
counts. Ten blue colonies were picked at random
from each of 10 randomly selected membranes.
The confirmation study was carried out in two
phases: (I) subculture of 10 colonies per mem-
brane to EC broth incubated at 44.5 C for 24
hours and (II) subculture of 10 more colonies per
membrane to tryptophane broth, MR-VP broth,
and citrate agar.
RESULTS AND DISCUSSION
Fecal coliform colonies grown on Millipore
filters appeared larger, smoother, and more mucoid
than on Gelman filters. The fecal coliform colonies
on Gelman filters, although generally higher in
number than Millipore filters, appeared small and
often dull. This observation was also noted by
Presswood and Brown (5). When utilizing the M-FC
test, it was found that Gelman filters appeared
blue while Millipore filters appeared beige-yellow.
These have also been the findings of other investi-
gators (2, 5). There has been speculation that
this phenomenon is due to a difference in.pH and
that this could possibly be responsible for the
disparity in counts on the two filters (5).
Table 1 shows that during 100 test trials,
colony counts on Gelman filters were almost con-
sistently higher than Millipore filters. On 3 trials,
Gelman filters were lower or equal to Millipore
filters in fecal coliform count. In considering the
overall data, Gelman filters recovered 2.5 times
more fecal coliform bacteria when the same
samples and identical processing methods were
used.
TABLE 1. COMPARATIVE STUDY OF GELMAN AND MILLIPORE FILTERS FOR THE
RECOVERY OF FECAL COLIFORMS.
Colonies Per
Millipore
23
10
10
14
9
17
33
18
12
18
12
7
18
19
26
Membrane
Gelman
58
44
20
44
28
36
59
42
20
42
2-
21
43
38
36
Ratio
G/M
2.52
4.40
2.00
3.14
3.11
2.12
1.79
2.33
1.67
2.33
1.67
3.00
2.39
2.00
1.38
Colonies Per
Millipore
29
23
6
4
7
31
13
32
26
11
10
10
7
44
11
Membrane
Gelman
49
59
35
20
32
36
28
44
36
36
36
30
46
43
22
Ratio
G/M
1.69
2.56
5.83
5.00
4.57
1.16
2.15
1.38
1.38
3.27
3.60
3.00
6.57
0.98
2.00
68
-------
Table 1 cont'd.
Colonies Per Membrane
Millipore Gelman
9
19
11
14
49
27
42
16
23
19
7
19
12
23
29
30
35
39
32
18
26
15
16
8
13
11
9
41
15
16
20
25
23
25
34
Total count
Total count
Mean count
Mean count
No. of times
No. of times
2 of G/M
35
49
26
38
39
44
42
25
42
59
22
30
30
39
46
40
44
49
51
31
46
31
26
45
56
44
22
53
24
39
32
30
31
46
55
Millipore =
Gelman
Millipore
Gelman
Millipore recovered
Higher counts
Lower counts
Gelman recovered:
Higher counts
Lower counts
=
Mean ratio of G/M
Ratio
G/M
3.89
2.58
2.36
2.71
0.80
1.63
1.00
1.56
1.83
3.10
3.14
1.58
2.50
1.70
1.57
1.33
1.26
1.26
1.59
1.72
1.77
2.07
1.62
5.62
4.31
4.00
2.44
1.29
1.60
2.44
1.60
1.20
1.35
1.84
1.62
1896
3821
19
38
:
2
97
97
2
251.51
2.52
Colonies Per
Millipore
22
36
16
4
38
9
12
12
13
31
13
23
20
13
32
16
42
23
16
38
11
10
8
6
8
7
16
7
11
18
18
19
20
5
23
Membrane
Gelman
39
52
39
23
49
33
36
25
44
44
28
59
37
36
58
56
61
44
49
49
55
49
37
35
27
29
42
24
20
39
30
23
36
21
29
Ratio
G/M
1.77
1.44
2.44
5.75
1.29
3.67
3.00
2.08
3.38
1.42
2.15
2.56
1.85
2.77
1.81
3.50
1.45
1.91
3.06
1.29
5.00
4.90
4.62
5.83
3.38
4.14
2.62
3.43
1.82
2.17
1.67
1.21
1.80
4.20
1.26
69
-------
Table 2 indicates that total coliform recovery
is similar with the two brands. The contrasting data
of Tables 1 and 2 would indicate a possible
elevated temperature-membrane inhibitory effect.
This difference in recovery has been observed by
other investigators (2, 3, 5). However, Presswood
and Brown (5) found a statistically significant
difference at both 35 C and 44.5 C. The compara-
tive total coliform study is not as comprehensive
as the fecal coliform study due to: (1) initial tests
indicated the two membrane filter brands gave
similar counts (2) the majority of samples pro-
cessed in our laboratory are for fecal coliform de-
termination. It is hoped that a more in-depth
membrane comparative study with total coliform
procedures will be a part of our quality control
program in the near future.
Due to the higher fecal coliform counts on
Gelman membranes, it was necessary to confirm
typical colonies in order to eliminate the possibil-
ity of a high number of false positives on Gelman
membranes. Table 3 indicates that the two mem-
brane filter brands gave similar confirmatory re-
sults in both phase I and II.
TABLE 2. COMPARATIVE STUDY OF GELMAN
AND MILLIPORE FILTERS FOR THE
RECOVERY OF TOTAL COLIFORMS.
TABLE 3. VERIFICATION OF TYPICAL BLUE
COLONIES ON MILLIPORE AND
GELMAN MEMBRANES
No. Colonies* No. EC
Filter Brand Picked Positive
^uiunies rei ivi
Colonies
Millipore
50
25
25
25
26
44
33
21
29
19
19
42
17
32
19
Total count
Total count
Mean count
Mean count
emuidiie
Membrane
Gelman
54
24
24
27
28
38
24
17
35
24
26
30
24
25
29
Millipore
Gelman
Millipore
Gelman
Ratio
G/M
1.08
0.96
0.96
1.08
1.08
0.86
0.73
0.81
1.21
1.26
1.37
0.71
1.41
0.78
1.53
= 426
= 429
= 28
= 29
I Millipore 100 92
Gelman 100 90
No. Colonies* IMViC No. of
Filter Brand Picked Pattern Cultures
II Millipore 100 + + - - 97
1
- + - - 2
Gelman 100 + ~ 95
1
- - + + 4
* 10 typical blue colonies were picked at random
from 10 randomly selected membranes.
In order for a microbiologist to report accur-
ate data, he must have efficient tools. Hopefully,
there will be a standardization and a centralized
routine evaluation of membrane filter brands.
The varying efficiencies of membrane filter brands
cannot be allowed to continue.
No. of times Millipore recovered:
Higher counts
Lower counts
No. of times Gelman recovered:
2 of G/M
Mean Ratio
Higher counts
Lower counts
of G/M
7
= 8
= 8
= 7
= 15.83
1.06
SUMMARY
(1) A comparative study has been conducted with
two brands of membrane filters.
(2) The data indicated that Gelman filters re-
covered 2.5 times more fecal coliforms than
Millipore filters.
70
-------
(3) Recovery of total coliform bacteria was
similar with Gelman and Millipore filters.
(4) There is a need to improve the quality control
of membrane filters for use in water micro-
biology.
ACKNOWLEDGEMENTS
The authors wish to express their apprecia-
tion to H. Bunczewski of the Bio-Analysis Group,
U.S. Environmental Protection Agency, Region
VII Laboratory, for media-reagent preparation and
filtration of various samples. A special note of
thanks to S. Finley of the same laboratory for
her clerical assistance.
REFERENCES
1. American Public Health Association. Standard
Methods for the Examination of Water and
Wastewater, 13th ed. American Public Health
Association Inc., New York, 1971.
2. Dutka, B.J., M.J. Jackson, and J.B. Bell.
Comparison of Autoclave and Ethylene
Oxide-Sterilized Membrane Filters Used in
Water Quality Studies. Appl. Microbiology.
28:474-480, 1974.
3. Hufham, J.B. Evaluating the Membrane Fecal
Coliform Test by using Escherichia coli as the
Indicator Organism. Appl. Microbiology. 27:
771-776, 1974.
4. Levin, G.W., V.L. Strauss, and W.C. Hess.
Rapid Coliform Organism Determination with
C14. J. Water Pollution Control. Fed. 33:
1021-1037, 1961.
5. Presswood, W.C., and L.R. Brown. Compari-
son of Gelman and Millipore Membrane
Filters for Enumerating Fecal Coliform Bac-
teria. Appl. Microbiology. 26:332-336, 1973.
6. Schaeffer, D.J., M.C. Long, and K.G. Janardan.
Statistical Analysis of the Recovery of Coli-
form Organisms on Gelman and Millipore
Membrane Filters. Appl. Microbiology. 28:
605-607, 1974.
QUESTION AND ANSWER SESSION
Sladek: The only comment I have is with re-
spect to your statement on good lots
and bad lots. I don't believe we came
to exactly that conclusion. One of the
lots that Mr. Harris studied was about
5 years old and we had manufactured
it at a time before we had begun to
quality control those filters with
respect to the fecal coliform test.
Harris: Well how do you explain that the two
good lots gave lower recovery com-
pared to the Gelman lots?
Sladek: I think without being able to inspect
the numbers and the experimental
conditions, it is not really worth
discussing.
Dazio: I would like to ask you or anybody
else in the audience, if you could tell
me what basic difference there is be-
tween different brands, let's say
Gelman and Millipore, which could
account for the differences for the
recovery of fecal coliforms. Does
anyone know? I think we should
being asking some basic questions,
and analyse and perhaps chemically
determine what differences exist in
the composition of membranes which
could account for differences in
recovery.
Harris: Well, there is one possibility. The
Gelman membrane has phosphate in
it whereas Millipore does not. Milli-
pore filters are supposedly almost
completely inert. So it could be a
possibility that the phosphate in Gel-
man membranes acts as a nutrient
source for the fecal coliform organism
and therefore tends to enhance the
growth of the fecal coliform organism
on the membrane.
Question Somebody, this afternoon, said that
from the the phosphate buffer was very dela-
Floor: terious to fecal coliform growth.
Grasso: I want to ask the gentleman from the
University of Florida, "Will you be
here tomorrow?" I think that Dr.
Litsky put me in my place and said to
hold my information until I present
my paper. So, without letting it out
of the bag today, I think I could
probably give you some encouraging
results on what is the effect of differ-
ent brands of filters tomorrow in our
paper. Thank you.
71
-------
Harris: We have some comments that ethy-
lene-oxide leaves a toxic residue but
according to the presentation by
Millipore Corporation this morning
they can't find it. Harris:
Brodsky: Not to put down the individual
companies producing membrane fil-
ters, but I am curious to know where
you got your membrane filters? Were
they, pardon me, a donation from the
company or were they purchased
from stock?
Lane:
Harris:
Presswood:
Harris: They were purchased from stock.
Brodsky: From a stock, both brands?
Harris: Right, both brands.
Brodsky: O.K. Because I think that is one
aspect of the investigation that
really should be settled.
Harris: We did not go directly to Gelman to
get the filters or directly to Millipore,
they were from the clearing house.
Rusnell: I just wanted to know if the samples
were chlorinated or unchlorinated.
Harris: We took this into consideration be-
fore starting our study, and we de-
cided that it wouldn't be wise to use
chlorinated sources.
Mack: I was wondering if anybody had taken
a look at the coliform organism under Question
the electron microscope, because we from the
are talking about filter sizes and all Floor:
the scientific work that goes with this
and nobody has come up with the fact Furman:
that many of these organisms are
terribly large and some are quite
small. Some of them have capsule-like
material and others don't, and this
would have a great deal to do with
whether or not they are retained.
Lane: You said that the Gelman membranes
were autoclave-sterilized and Millipore
were ethylene-oxide sterilized. How
can you compare one membrane that
is sterilized by one method with a
membrane that is sterilized by an
entirely different method? Shouldn't
they both be sterilized the same way?
According to the manufacturers they
are supposed to give equal results.
There are not supposed to be inhibi-
tory effects from ethylene oxide or
autoclave sterilization. Millipore says
ethylene oxide is nontoxic so we used
their filters and compared to auto-
el aved filters.
I can't visualize that heating a mem-
brane to the temperature of 121° C
for 20 minutes or a half an hour is
not going to do something to the pore
size of the membrane whereas ethy-
lene-oxide might be nontoxic.
But the fact remains one gets higher
counts than the other.
Ours was a comparison of filters
similar to Mr. Harris' and we also
used autoclave and ethylene oxide
membranes. The reason we did this
was that this is the way the manufac-
turer sells them. I think Millipore
does sell autoclaved filters but as Mr.
Harris said the ethylene-oxide is
suppose to give comparable results.
Also some of the Millipore filters be-
come distorted if you do autoclave
them. You can see by the grid lines,
that the filters are distorted.
What about Nuclepore?
The Nuclepore membranes do not
grow colonies with media that you
use, so I would like to publicly tell
you not to use Nuclepore for that pur-
pose and with these media. We have
never made the claim that the poly-
carbonate would work for water
analyses. If you have some research
needs for polycarbonate and you can
use some wrinkles, add surfactants to
the media to improve growth. But for
the routine use we do not recommend
Nuclepore for microbiological pur-
poses. Thank you.
72
-------
COMPARISON OF MEMBRANE FILTER BRANDS FOR
RECOVERY OF THE COLIFORM GROUP
A. P. Dufourand V.J. Cabelli
Environmental Protection Agency
National Marine Water Quality Laboratory
Narragansett, R.I.
ABSTRACT
Recoveries of pure cultures of E. coli, K.
pneumoniae and E. cloacae and fecal coliforms
from natural waters were compared using the M-FC
test and different lots and brands of membrane
filters. MPN recoveries were used as reference
values for measuring accuracy.
Except for Nuclepore, the brand of mem-
brane filters did not significantly affect recoveries
of pure cultures or natural source fecal coliforms.
Variability increased with natural water samples.
The variability (precision) of results from lot to
lot within a brand and from culture to culture
within an MF brand precluded any generalization
about acceptance of one brand over another.
INTRODUCTION
Our interest in the recovery efficiencies of
membrane filters originated a few years ago when
we began an epidemiological/microbiological study
of the relationship between pollution levels and the
incidence of disease at marine bathing beaches. It
resulted from the necessity to evaluate standard
membrane filter (MF) methods in marine waters
prior to their possible use in the study. The evalua-
tion of the M-FC test revealed that it measured
much lower fecal coliform densities than the EC
most probable number (MPN) test. This in turn led
to a further investigation of various membrane
filter brands in order to determine if the filters
themselves affected the accuracy and precision of
the assay method. For the purpose of this pre-
sentation, I want to define those two terms as we
use them.
The accuracy of a method for an organism or
a group of organisms, which may be defined as the
ability to detect the "true density" of bacteria in a
given volume of a water sample, ideally should be
determined using natural samples. However, since
"true density" is, in fact, unknown in natural
samples, the accuracy can be determined only
relative to some "standard" or "reference"
method. Secondly, when a group of organisms is
being enumerated i.e., coliforms or fecal coliforms,
the relative accuracy may vary from location to
location depending on the particular distribution
of the component biotypes in the particular
sample.
The precision of a method, which may be
defined as the degree of dispersion of obtained
values around a mean estimate, is equally as
important as its accuracy or recovery efficiency.
This is so because of the practice of determining
bacterial density values with a single filter which
is a common practice. Thus, a lack of precision
caused by the filter would affect the ability of a
method to detect differences between bacterial
densities of two water samples.
With these two parameters in mind, the
following experimental methodology was used to
determine which type of filter would be most
efficient for our purposes. I would like to state
that all of these were done with surface waters in
mind, particularly bathing beach surface waters,
and I would not like to expand this to other types
of water samples.
MATERIALS AND METHODS
Media Dehydrated M-FC broth, M-Endo
broth, nutrient agar, Trypticase soy broth, EC
broth, lactose broth and brilliant green lactose
bile broth were obtained from Difco and prepared
73
-------
for use according to the directions of the manu-
facturer.
M-Endo broth, M-FC broth and nutrient agar
media.
Test Organisms Strains of E. coli, Klebsiella
pneumoniae and Enterobacter cloacae were main-
tained on nutrient agar slants. The Klebsiella
strains were able to produce gas in EC broth in-
cubated at 44.5 C.
Filters Membrane filters from the following
manufacturers were obtained from commercial
sources: Millipore (M), Gelman (G), Sartorius (S),
Schleicher and Schuell (s/s) and Nuclepore (N).
All of the membranes were packaged sterile with
the exception of the Nuclepore and Schleicher and
Schuell membranes, which were sterilized by
autoclaving for 15 minutes at 121 C. All mem-
branes, with the exception of Nuclepore, were
gridded and had an average pore size of 0.45
microns. Nuclepore membranes were ungridded
and had a 0.4 micron pore size.
Control Procedure Spread plates used to
determine the density of the test cell suspensions,
were prepared by pipetting 0.2 ml of the test cell
suspension onto each of five plates. The suspen-
sions were spread over the surface of the nutrient
agar medium with a sterile glass rod and allowed to
dry, after which the plates were incubated in an
inverted position at 35 C.
The MPN procedures were carried out as des-
cribed in Standard Methods for the Examination
of Water and Wastewater (1).
Statistical Analysis Tukey's Studentized
Range Procedure for comparing several means
was used for the statistical analysis of the data (2).
The dispersion about the mean was described as
the coefficient of variation.
Natural Samples Natural samples were obtain-
ed from three locations in the Rhode Island area.
They were: (1) Wickford Harbor, a salt water cove
which receives mainly septic tank overflows, and
this would be human waste of which the major
component was E. coli. (2) the Saugatucket River,
which receives mainly industrial effluent from a
textile finishing plant and these effluents con-
tained Klebsiella species. (3) the Pawcatuck Es-
tuary whose bacterial pollution comes from
domestic sources, as well as industrial effluents.
The samples usually arrived at the laboratory
within three hours of collection and they were
immediately assayed.
Test Cell Suspensions Eighteen to twenty
hours before each experiment, a loopful of the test
strain was transferred to Trypticase soy broth.
After 18 hours incubation at 35 C, the culture
was diluted in sterile, phosphate-buffered saline
(pH 7.2) to a density of between 20 and 60 organ-
isms per ml.
Membrane Filtration One ml of the test cell
suspension, or an appropriate volume of a natural
sample, was passed through each filter after being
mixed with 20 ml of buffer. The filter was then
rolled onto a broth saturated pad or the agar
surface of a randomly chosen plate. The M-FC
plates were incubated in "whirl-pak" bags in a
44.5 C water bath. The M-Endo plates and mem-
brane nutrient agar plates were incubated in an
inverted position in a 35 C incubator. All mem-
brane filter brands were tested in triplicate on
RESULTS AND DISCUSSION
The accuracy of the M-FC procedure for fecal
coliforms as it is affected by the brand of mem-
brane filter used, was determined by comparing the
densities in Klebsiella and E. coli test suspensions
obtained by the MF procedure with those from
nutrient agar spread plates. It was assumed that the
density, as determined from nutrient agar spread
plates, would provide the best estimate of the
number of bacteria actually present in a given test
cell suspension. The effect of five membrane
filter brands on the accuracy of the M-FC proce-
dure with four fecal coliform strains is shown in
Table 1. The percent recoveries for the E. coli
strains ranged from 16 to 75%, and for the Kleb-
siella strains they ranged from 14 to 94%. Al-
though the filter brands are ranked according to
their mean recovery values, most of the observed
differences were not statistically significant as
shown by the underscoring of mean recovery
values of each organism. The underscoring indi-
cates that any two means, not underlined by the
same line, are significantly different at the 95%
confidence level. Statistically significant differ-
ences appeared more frequently with the Kleb-
siella strains than with the E. coli strains. Of the
brands tested, the Gelman product appeared to
provide the most, and Nuclepore the least accurate
density estimates with the four fecal coliform
strains tested. However, none of the filter brands
was consistently accurate. This appeared to depend
more on the basic method relative to the strain
being tested rather than the filter brand.
74
-------
TABLE 1. RECOVERY OF E. COLI AND K. PIMEUMONIAE ON THE M-FC MEDIUM WITH
VARIOUS MEMBRANE FILTER BRANDS
E. coli no. 3
Filter type
Mean (2)
% Recovery
ST. (3)
E. coli no. 5
Filter type
Mean
% Recovery
S.T. ST.
K. pneumoniae no. 450
Filter type
Mean
% Recovery
S.T.
K. pneumoniae no. 444
Filter type
Mean
% Recovery
ST.
(1)
N
14
16
N
22
18
M
34
39
s/s
38
43
S
41
47
G
46
52
N
50
38
s/s
57
44
S
71
55
M
77
59
G
98
75
M
21
14
S
29
19
s/s
31
20
s/s
66
53
S
86
68
N
47
30
M
99
80
G
75
48
G
118
94
1. Gelman (G), Millipore (M), Nuclepore (N), Sartorius (S) and Schleicher and Schuell (s/s)
2. Relative to bacterial density on nutrient agar spread plate incubated at 35C.
3. S.T. = Statistical Test: Tukey's Studentized Range Procedure; all means not underscored by the
same line differ significantly at the P = 0.05 level.
The effect of membrane filter brand on the
precision of the M-FC test is shown in Table 2. The
filters are ranked according to the magnitude of
dispersion around the mean. The precision of the
Sartorius filters was consistently better (a lower
coefficient of variation) than the other brands,
and the Nuclepore and S & S filters were con-
sistently poor. Unlike the accuracy, the precision
appears to be filter brand dependent rather than
strain dependent.
The recovery of fecal coliforms from natural
samples is shown in Table 3. The trend, with regard
to filter efficiency, appears to be similar to that
found with pure culture suspensions. However,
there were fewer statistically significant differences
between filter brands with natural sample suspen-
sions. These differences were probably due to the
greater filter variability of all brands (see Table 4).
This variability would, in fact, cause the statistical
test used to be less sensitive for detecting recovery
differences between filter brands.
Table 4 indicates that the precision of the
M-FC test was much lower with all brands of filter
when compared with the pure culture data. It can
also be noted that the rank position of the filter
brands was not similar to those obtained with
pure cultures. This is probably indicative of the
heterogeneous nature of the fecal coliform group.
75
-------
TABLE 2. PRECISION OF MEMBRANE FILTER BRANDS ON M-FC MEDIUM
E.
E.
K.
K.
coli no. 3
Filter type (D
C.V. (2)
coli no. 5
Filter type
C.V.
pneumoniae no. 450
Filter type
C.V.
pneumoniae no. 444
Filter type
C.V.
N
43
s/s
18
N
19
N
45
s/s
24
N
12
M
14
s/s
18
G
17
G
12
s/s
10
M
14
M
12
M
6
G
7
G
3
S
5
S
3
S
4
S
2
1. See footnote 1, table 1.
2. C.V. = Coefficient of Variation; (standard deviation/arithmetic mean) x 100
TABLES. FECAL COLIFORM RECOVERIES FROM NATURAL SAMPLES WITH VARIOUS
MEMBRANE FILTER BRANDS USING THE M-FC TEST.
Wickford
Harbor
Saugatucket
Pawcatuck
Estuary
Filter type (D
Mean (2)
% Recovery
ST. (3)
R.
Filter type
Mean
% Recovery
9 T
O. 1 ,
Filter type
Mean
% Recovery
N
8.3
12
N
4
33
N
1
7
s/s
12
17
s/s
5
38
s/s
4.7
32
S
17.7
25
M
6
46
M
5.2
36
M
20.3
29
G
8
62
S
12
82
G
21.3
30
S
12
92
G
16
109
1. See footnotes 1 and 3, table 1.
2. Relative to E.G. MPN estimate
76
-------
TABLE 4. PRECISION OF MEMBRANE FILTER BRANDS ON M-FC MEDIUM
Wickford
Harbor
Filter type M)
C.V. (2)
Saugatucket R.
Filter type
C.V.
s/s
35.8
N
78
G
35.7
M
45
S
25
G
44
N
18.7
s/s
20
M
7.5
S
17
Pawcatuck
Estuary
Filter type
C.V.
N
100
M
40
S
25
s/s
22
G
19
1. See footnote 1, table 1.
2. See footnote 2, table 2.
The effect the filter brand has on the re-
covery of total coliforms by the M-Endo proce-
dure was also examined. Table 5 shows that, with
pure cultures of E. coli, Klebsiella, and Entero-
bacter cloacae, recovery between brands was
neither appreciable nor statistically significant.
The one exception was the Nuclepore brand, which
was consistently poor in its ability to recovery
coliforms. Relative to spread plates on nutrient
agar (NA), all of the filter brands, with the excep-
tion of Nuclepore, were reasonably accurate when
using Klebsiella or Enterobacter as the test strain.
The recoveries with E. coli no. 3 were rather poor.
In order to determine if this might be due to inhib-
itors present in M-Endo medium, the mean re-
coveries on M-Endo medium were compared to
those of membranes also placed on nutrient
agar. As can be seen in Table 6, the relatively poor
recoveries of E. coli no. 3 on M-Endo were also
obtained on NA, suggesting a general filter effect
or a nutrient deficiency for this particular organism.
The relative accuracy of the various filter
brands, when examined using natural samples was
essentially the same as noted above. The recovery
relative to the MPN procedure was excellent in two
of the three samples examined. However, the fact
that the MPN procedure was used to estimate the
"true" coliform density does not allow substantive
conclusions to be drawn about these data in this
regard. See Table 7.
The variability, from lot to lot, of some
filter brands would preclude broad generalizations
about the acceptability of one brand over another
except for those brands actually tested under the
same conditions. This point is illustrated in Table
8. In two out of the three brands tested, there
were statistically different mean recovery values
between lots. Thus, it would appear that with some
membrane filter brands, the recovery efficiencies
can not be predicted from lot to lot. And you will
notice that this is a different E. coli strain than we
used in the other three experiments. What happen-
ed is that we had lost this strain, but last week we
found a lyophilized culture of it. So, we ran it
through the same test procedure and, if you will
put on the last slide, you will see the strain to
strain variabilities causing differences with the
Gelman filters.
SUMMARY
The following conclusions can be made:
1. The use of pure cultures to detect dif-
ferences in accuracy in the M-FC pro-
cedure between membrane filter brands
is more sensitive than using natural
samples. However, there is a strain to
strain variation in the way in which pure
cultures react to each filter brand and
77
-------
TABLE 5. RECOVERY OF E. COLI, K. PNEUMONIAE AND E. CLOACAEON M-ENDO MEDIUM
WITH VARIOUS MEMBRANE FILTER BRANDS
E. coli no. 3
Filter type (D
Mean (D
% Recovery
ST. (D
K. pneumoniae no. 444
Filter type
Mean
% Recovery
ST.
K. pneumoniae no. 450
Filter type
Mean
% Recovery
ST.
E. cloacae no. 491
N
32
36
N
68
51
G
58
66
s/s
59
67
M
61
69
S
68
77
G
107
86
M
109
88
s/s
112
90
S
120
105
N
8
44
s/s
14
78
S
15
83
G
15
83
M
20
111
Filter type
Mean
% Recovery
N s/s S
11 23 24
58 121 126
G
243
128
M
26
137
1. See footnotes 1, 2 and 3, table 1.
TABLE 6. COMPARISON OF E.COLI RECOVERIES ON M-ENDO AND NUTRIENT AGAR WITH
VARIOUS MEMBRANE FILTER BRANDS
E. coli
% Mean Recovery Value
(1)
no. 3
Filter type (1'
mEndo
Nutrient Agar
N
36
52
G
66
70
s/s
67
68
M
69
69
S
77
75
1. See footnotes 1 and 2, table 1.
78
-------
TABLE?. TOTAL COLIFORM RECOVERIES FROM NATURAL SAMPLES WITH VARIOUS
MEMBRANE FILTER BRANDS ON M-ENDO MEDIUM
Wickford Harbor
Filter type
Mean
% Recovery
ST. (D
Saugatucket R.
Filter type
Mean
% Recovery
S.T.
Pawcatuck Estuary
N
g
26
s/s
33
94
G
37
106
S
38
109
M
39
111
N
27
15
G
56
32
S
67
38
s/s
71
41
M
83
47
Filter type
Mean
% Recovery
N
8
22
s/s
31
86
G
33
92
S
36
100
M
37
103
1. See footnotes 1 and 3, table 1.
2. Relative to completed total coliform MPN estimate.
this may influence the choice of a "best"
membrane filter. Furthermore, pure cul-
ture may not truly mimic the physio-
logical state of organisms in natural
samples.
2. The precision of the M-FC test is in-
fluenced by the filter brand, and this is a
rather consistent characteristic from
strain to strain with pure cultures.
3. Differences in precision due to mem-
brane filter brand have a tendency to be
lost when natural samples are used as the
test inoculum in the M-FC test. This is
assumed to be caused by the hetero-
geneity of the fecal coliform group and
the physiological state of the organisms.
4. The relative accuracy of the various
brands of membrane filters was essen-
tially the same whether pure cultures or
natural samples were used. However, in
the latter case the differences were mask-
ed due to the decreased precision when
natural samples were used.
5. In general, acceptable total coliform
recoveries were obtained by the M-Endo
procedure with all the filter brands
(except Nuclepore).
6. If pure cultures are used to compare
various filter brands or lots within a
filter brand, the choice of test organism
is exceedingly important, since, with
some strains, an overall membrane
filter effect may mask true differences.
If natural samples are used, a large
enough number of filters must be
examined to compensate for the de-
creased precision observed with this
type of test suspension.
7. Three factors that must be considered
in evaluating membrane filters are
accuracy, precision, and lot to lot
variability.
REFERENCES
Standard Methods for the Examination of
Water and Wastewater, 13th Edition, Ameri-
can Public Health Assoc., New York, 1971.
79
-------
TABLES. RECOVERY OF E. COLI ON M-FC MEDIUM WITH DIFFERENT LOTS OF MEM-
BRANE FILTER BRANDS
Millipore
(E. coli no. 104)
Filter lot #
Mean
% Recovery (D
S.T. (D
Millipore
(E. coli no. 104)
Filter lot #
Mean
% Recovery
S.T.
Gelman
(E. coli no. 5)
Filter lot*
Mean
% Recovery
S.T.
Sartorius
(E. coli no. 104)
1295
12.3
10.4
NN
25.7
21.7
9396
26.3
22.3
6112
38.3
33
8752
47.7
40
3283
6.3
6.3
80714
70
73
9733
89.7
76
5000
16.7
16.7
9768
17.7
17.7
9733
62
62
80706
75
78
80730
76
79
Filter lot #
Mean
% Recovery
S.T.
10
65
34
30
27
14
1. See footnotes 2 and 3, table 1.
2. Lecture Notes in Applied Statistics, Joseph L.
Ciminera, Villanova Univ. Press, 1956.
QUESTION AND ANSWER SESSION
Geldreich. What do you think of the possibility
of using pure cultures in some of the
evaluations of these products? We
can't say, as you are saying, to use a
natural sample, although I always like
to use them in preference to pure
cultures. What about the possibility
of using a mixture of pure cultures in
a lyophilized condition in which you
can standardize and then use them as
a way to come up with a test group
which would allow us to look for
sensitivity, as well as selectivity in
suppressing some organisms. You
think this is a possible way out of the
problem?
Dufour. I don't really know. Dr. Hufham out-
lined this rather well in his presenta-
tion this morning. Any one culture
has a certain part of the distribution,
even if we called it pure culture, that
is killed by the elevated temperature. I
don't know whether putting in dif-
ferent types of organisms and using
this as a test suspension will overcome
that shortcoming.
Geldreich. One of the problems that we have
here is that if you use an E. coli of a
certain type, you are only checking it
80
-------
for one thing, and that is sensitivity
to let's say E. coli. But, if you are
trying to check it for suppression of
other organisms, you would be miss-
ing that point.
Dufour. Yes. This is why we used samples
from what we thought were three
different environmental situations.
But obviously, we couldn't show
anything because of the great dis-
persion we found between filters.
Geldreich. We noted, and you may have too, for
instance that Gelman membranes may
at times have more background count
of other organisms growing on them.
This may occur because of some
nutrient material present there.
Dufour. Again, this was not one of the prob-
lems that we-ran into because of the
way we chose our natural sample.
Presswood. I noticed that if you are using pure
cultures, the source where you isolate
the culture makes a difference. I
have isolated some bacteria from raw
sewage before it was chlorinated and
those bacteria did not respond as well
to the membrane filter technique as
bacteria isolated from river water.
Even after passing them through 2 or
3 passes of EC medium, or some
other nutrient broth, they still did not
respond as well to the membrane
filter. Especially at 44.5C as E. coli,
or what we call fecal coliform bacteria
isolated from river water. I don't
know why this is, but it happens.
Dufour. It was mentioned today, that all of
our strains were not isolated using the
M-FC procedure. They were isolated
using a non-inhibitory medium and
found to be fecal coliforms using EC
broth. So, at least we didn't have that
shortcoming.
Bordner. It just occurs to me that one of the
plans we have in mind at MDQARL,
Methods Development and Quality
Assurance in Cincinnati, is to accept
bids for a contract that will develop
pure cultures from environmental
samples and fecal samples. These can
then be used as reference samples. It
occurs to me that it might be appro-
priate to look at this type of culture
which we hope to have lyophilized
and made available to our laboratories
and others later on for membrane
filter and media evaluation.
81
-------
A COMPARISON OF MEMBRANE FILTERS, CULTURE MEDIA, INCUBATION TEMPERATURES,
POLLUTED WATER AND ESCHERICHIA COLI STRAINS IN THE FECAL COLIFORM TEST
Paul J. Glantz
Department of Veterinary Science
The Pennsylvania State University
University Park, Pa. 16802
ABSTRACT
Twenty lots of membrane filters from 3
manufacturers produced variable results when the
fecal coliform test was used with polluted water
samples and Escherichia coli (E. coli). The growth
of nineteen E. coli cultures varied with incubation
temperatures of 35, 43, and 44.5 C, with mem-
brane filter lots, and with culture media (Trypti-
case soy agar, M-FC, and violet red bile agar).
Growth was better on violet red bile agar than
on M-FC medium, which was not always due to
the 44.5C incubation temperature. Some of the
E. coli strains tested did not grow as well in poured
agar plates as they did on membrane filters. Evalua-
tion of the results indicated that the efficacy of
the membrane filter for the fecal coliform test
could be affected by, (1) the strain of E. coli, (2)
the method utilized for growth (pour plate, pad
and broth, pad and agar), and (3) the tempera-
ture inside the plate. Therefore, it is apparent
that test methods must be standardized for these
factors.
INTRODUCTION
The membrane filter method used for mea-
suring fecal coliform bacteria in polluted water
has been critically evaluated by several investiga-
tors. Poor recovery of fecal coliform or E. coli
strains has been attributed to the variation of the
quality of the components in the M-FC agar (10),
to the differences in the brands and lots of the
membrane filters (1,5,8,11), to the variation in
the incubation temperatures (3), and to the type
of water tested (1,7,9).
This study was undertaken to determine the
efficiency of three brands of membrane filters to
recover fecal coliforms and E. coli isolated from
polluted water. Three incubation temperatures
(35, 43, and 44.5 C) were studied to determine
the effect of temperature on the growth of the
bacteria. A non-inhibitory medium, Trypticase
soy agar (TSA, BBL), was used to determine
the actual counts of the E. coli cultures being
compared for growth on M-FC and violet red
bile agar (VRB, BBL). Additional studies were
initiated to determine the cause of the erratic
behavior of the membrane filters and the E. coli
bacteria.
MATERIALS AND METHODS
The manufacturer's name and lot number
of the membrane filters (with the prefix used as a
test number in this study) were as follows:
Test Millipore
No. Lot No.
Test Gelman Test Sartorius
No. Lot No. No. Lot No.
M-1
M-3
M-5
M-6
M-7
M-8
M-11
M-12
M-13
95434-3
93448-10
11013-4
95434-2
12732-6
12667-3
06602-8
06618-3
34798-4
G-3
G-6
G-7
80730
80822
80901
S-1
S-2
S-3
S-4
S-5
S-6
S-7
S-8
713649
753 649
773 284
993 447
993711
993 186
013870
773 568
All of the above except M-11 and M-12 were
pre-sterilized by manufacturer. These two lots
were used as supplied (not sterile).
M-FC broth BBL lot 204625) with 0.01%
rosolic acid (Difco lot 488535) added was used for
82
-------
tests with pads, and 1.5% agar added for pour
plates. These two media were dissolved by heating
to a boil (as was the VRB). The TSA was auto-
claved at 121 C for 15 minutes. The agar media
/vere cooled to 45 C before pouring plates.
Sterile demineralized water (pH 6.5) was used
to prepare dilutions of the E. coli culture and as
a dispersal solution (20 to 30 ml) during filtration
of all samples. Thirty to 40 ml of this water was
filtered after every 5 replicates as controls. A 1 ,000
Ai1 Eppendorf, or a 5 to 10 ml Oxford pipet with
sterile tips, was used to transfer 1 to 10 ml of
diluted E. coli culture, or creek water, for filtra-
tion and agar pour plates. Membranes were placed
on pads in sterile 48 x 8.5 mm Millipore or 12 x 50
mm Falcon plates. Sterile 15 x 100 mm plastic
petri dishes were used for pour plates. The upper
half of the Gelman magnetic filter funnel was
placed under a germicidal lamp (G.E. G8T5)
between filtrations. Plates were sealed inside Whirl-
pak (Nasco) or Ziploc (Dow Chemical Co.) plastic
bags prior to immersion in the water baths. After
24 hours incubation, typical blue M-FC or red
(VRB) colonies were counted at 7X magnification
with daylight fluorescent illumination. The 15 x
100 mm agar pour plates were counted using a
Quebec colony counter. Plates with 20 to 100
colonies for the 48 x 8.5 mm plates and 30 to 300
for the 100 x 15 mm plates, were considered
countable.
The nineteen E. coli cultures isolated from
creek water were incubated at 44.5 C on M-FC
medium. Biochemical reactions of the E. coli
cultures tested produced acid-slope and acid-gas
butt on triple sugar iron agar (TSI, BBL). The
Minitek (Bioquest) test method gave the following
results: positive for indole, methyl red, lactose,
arabinose, lysine decarboxylase; negative for
Voges-Proskauer, citrate, phenlalanine, inositol,
malonate,
Variations occured with ornithine decarboxy-
lase and rhamnose respectively as follows: +, +,
Nos. 2A, 3A, 9, 16, 16A, 16X, 16Z, 24; -, +, Nos.
1A, 5; -, -, Nos. 1, 2, 2Z, 3, 3X, 3Z; 4, +, -,
Nos. 2B, 2X. Culture #4A gave results typical of
Klebsiella pneumoniae.
Stock cultures were maintained at room tem-
perature on TSA slants sealed with waxed corks.
Growth cultures were transferred from stock plants
to VRB agar plates, incubated at 35 C overnight,
and a single colony suspended in sterile deminer-
alized water (pH 6.5). One ml of the suspension
was transferred to membrane filters, or to pour
plates for subsequent tests.
The temperatures of the Hotpack Incubator
#5528 (35 C), the GCA-Precision Scientific Co.
Coliform Incubator Bath (44.5 C), and the Blue-M
non-circulating waterbath (43 C) were continually
monitored during the study. The waterbath tem-
peratures and the temperatures of the corres-
ponding immersed plates did not vary.
The different lots of Gelman, Millipore and
Sartorius membrane filters were tested with 11
E. coli cultures. The membrane filters were
placed either on corresponding pads saturated with
M-FC medium or on TSA plates and incubated at
44.5 C (GCA-Coliform Bath). The same test
cultures were also plated with TSA agar and in-
cubated at 35 C. One ml of a suitable dilution of
the culture was used for filtration and for poured
plate counts. All tests were done in five repli-
cates.
The results based on the average counts, are
expressed as percent (%) recovery of E. coli. The
pour plate counts obtained with TSA at 35 C were
considered representative of the actual number of
E. coli in the 1 ml volume tested. Membrane filters
were placed on solid TSA agar for comparison with
membrane filters placed on pads. A dilution of one
strain of E. coli was tested with all membrane
filters listed for that strain at one time.
RESULTS
Comparison of E. coli Recovery by M-FC Brand
and Lot
In Table 1 the results are presented in 3 cate-
gories, A, B, and C. Category A is the percent re-
covery of the E. coli counts on membrane filters
83
-------
d
in
5
Q
Z
<
O
in
CO
<
<
CO
1
T.
H
3»
>
Q
LU
DC
<
a.
S
O
o
0
LL
s
z
o
Ij
o
U
ai
1 1
0
>
DC
LU
>
O
o
LU
DC
h-
2
O
DC
LU
a.
LU
i
.j
CO
<
I
1
in
*t U3
Tf CO
°< <
CO CO
1- h-
LO
5 u>
m co
LL <
£ £
" 3
»rMCM« ^I^CM^ CMr^^^^fr^T incocM^'Oi^i'OJin^'
cocoincocoo5'-ooooo5»-'-ooo30oooo)o>ooinoor»oop^
« « « T « r T
O)OOOO)^-'-r>oocNinr>.oO'-CMcocMcDin«*oo^co^-o
inr^-coco^-rv-O)^'CDtN-inooooooocDcooocDtocDr^-cor^r>-i>>
incM.cor«srvr^r^r^cor>a>ooo5
«<<colnCOrxoocoCDtt>oo'7{Or^c?'-
S(3coC5coIEC3cn2ci)coSC3coC3coSC3coC3coSc3coCDco
co'-o^f' cococococooocM^tinr^^i-O'-LnincM^t
ococMOTt T-or"»O)cooor^coinrxcor-.oocoo>O)CO
T t T r r
o 05 » cMOOOcoOor^ococoincNr^rMinoocMCOco
ooooooom CM «- inr^incDinincDinincoin
*~
co^ri^«r>.CMOCooocMO'*Tj-O5»-oo'tr~oincM^r
oooouscoin CM CM r*r-o>cocor»o>O)inr-.cQ
««« «<«
*
0) co CO CO
topooooo 5'7coT_coco^-T7Cor-L9toroCOioT7{?'7
i^r /« » ,. L ~y "^ /n > «>. ^^ /H **» ^5 ^n * *\ ^S ^^ */> ^^ f }\ ^> (^ r f\
^tDcOCOZSUCO^UCO^VJCOSC-'COwCO^tJCO
^
CD
OJ
4-»
c
i
5
E
3
c
1-1
^
52
*
i^
CD
CO
JO
E
o
X)
L
CO
05
CO
>
o
C/)
CD
(A
CO
CJ
'(-I
Q.
>
H
C
o
~n
w
0>
CJ
CO
Q.
L_
d)
^r
^-
0
"8
L_
CO
a
E
o
CJ
E
3
T3
CU
E
o
LL
^
^
-C
i
a
CO
Q.
C
0
x_
O)
r^:
H
a>
c
£
n
E
CD
E
c
o
1
0}
g
y
0)
^
(1)
UJ
L.
3
(-»
3
CJ
"5
U
LU
"o d
c ^q
OJ ^1-
s 5
CD 4.!
a. co
<
<
CO
1
c
o
1 J
c
"2
o
o
5!
CO
a.
CO
O5
CO
D
£
3
O
a.
(->
2
"D
£
(D
a
E
o
CJ
O
in
5
U
CO
E
_3
T3
CD
E
o
LL
^
2:
jz
I
"D
CO
a
c
0
x_
CD
^^
CD
C
CO
L~
jO
E
CD
E
c
0
D
2?
CD
O
0
CD
l_
n\
£
3J
+-<
D
O
1
LU
'o
C "
05 in
fc? co
05 +-
a. co
00
d
in
CO
*j
CO
CO
O
4-1
c
23
O
U
CD
+J
JO
a
L_
CO
05
CO
a
£
^
o
a
o
+j
~a
CD
1 _
CO
a.
0
CJ
O
in
*t'
^~
+-<
CO
<
&
c
o
53
*4
0)
CD
L-
-Q
E
CD
E
c
0
T5
2?
CD
O
0
CD
^.
£
D
M
3
CJ
1
LU
*+
O
<-<
05
£f
05
a.
O
84
-------
(M-FC pad) compared with the average counts of
filters on TSA at 44.5 C. The percent recoveries
ranged from 44 to 133%, with cultures 2A, 3A,
and 24 providing good recoveries while cultures
1A and 3 were poor.
When the 1A culture was filtered and the
membrane filters were placed on pads with M-FC
medium at 44.5 C, growth occurred on Gelman
lot numbers 3 and 6, but not on Millipore lot
numbers 3 and 13, and Sartorius lot numbers 1
and 4. Growth was produced on all 6 lot numbers
when the filters were placed on TSA medium and
incubated at 44.5 C. Since this 1A culture was
unusual, it will be omitted in the remainder of
results for Table 1. However, further tests were
conducted later on this culture ( Refer to Special
test-A).
The second category-B shows the percent re-
recovery of E. coli counts on membrane filters
M-FC at 44.5 C compared with poured TSA
plate counts at 35 C. The percent recovery ranged
from 47 to 108%, with culture numbers 1 and 9
providing best results.
The third category C presents the percent re-
covery of E. coli on membrane filters (TSA) at
44.5 C compared with the poured agar plate count
on TSA incubated at 35 C. The percent recovery
ranged from 55 to 131%, with culture numbers 1,
9 and 3Z providing best results. These results are
applicable only for the lot numbers of membrane
filters tested.
With respect to efficiency of each filter brand
and lot number, the variations apparently were due
to the culture (1A and 3), or to the filter used
(S-3 with culture 2A). Since one specific culture
dilution was tested at one time with the different
filters, the efficiency of the latter can be compared
for that particular culture. Due to lack of supply,
not all filter lot numbers could be tested with all
cultures and some of the results may require
further tests.
Gelman lot number 3, tested with 11 cultures,
had a percent recovery of 71 to 114% (Category
A), 65 to 109% (Category B), and 64 to 131%
(Category C). Compared with the other lot num-
bers, G-3 provided the best overall percent
recovery for the 11 cultures tested. Lot number
G-6, tested with 6 cultures, did not provide results
that were as good as G-3.
Lot M-13, tested with 5 cultures had a per-
cent recovery of 55 to 83% (Category A), 51 to
74% (Category B) and 72 to 95% (Category C).
Lot M-6 had a percent recovery of 83 to 95%
(Category A), 59 to 80% (Category B), and 62 to
96% (Category C). Lot numbers M-3, M-5, M-7,
M-8, and M-12, each were used with only one test
culture, with M-5 and M-8 producing comparable
results for all categories. Lot M-12 had an unusual
result with culture 24, a high (133%) recovery for
Category A compared to 59% for Category B.
Lot numbers S-1 and S-3 were used with 4
cultures. Lot S-1 varied in percent recovery, having
64 to 99% (Category A), 53 to 70% (Category B),
and 55 to 84% (Category C). For the same cate-
gories respectively, lot S-3 had recoveries of 44 to
87%, 47 to 82%, and 70 to 107%.
Lot numbers S-4, S-5, S-6, S-7, and S-8 each
were tested with 2 cultures. Lot S-4 provided best
results with culture #3A, and S-7 was good with
culture #24, but not with culture #9. Lot S-5 was
good with culture 2A but not quite as good with
culture #9.
There were some filter lot numbers that gave
nearly identical results with the same culture,
such as S-5 and G-3 with culture #9, S-6 and G-6
with culture #16, and M-6 and S-3 with culture
#1.
In the test with culture #1 where no filter
was used, the recovery on poured agar plate counts
for Categories A and B was lower than that ob-
tained with the filters. Further tests (Special test
A) indicated a much lower count for poured agar
plates with this culture.
The results in Table I are an indication of the
erratic behavior of membrane filters with respect
to recovery of E. coli test strains. Filtering the
culture dilution and placing the membrane filter
on TSA medium at 44.5 C sometimes gave a lower
percent recovery (57 to 69%, G-3 and S-4, culture
3A, Category C) than when the filter was placed on
M-FC at 44.5 C (102 to 104%, Category A). How-
ever, the percent recovery (Category B) of culture
3A was similar when filters were placed on M-FC
(pad) at 44.5 C (60 to 70%) or on TSA at 44.5 C
(57 to 69%) and compared with poured TSA
plate count at 35 C.
85
-------
The opposite effect was obtained when cul-
ture #9 was used with filter lot number S-5, and
filter G-3 was used with cultures #9 and 4.
*
Comparison of E. coli Recovery on Pour Plates at
35, 43, and 44.5 C
The comparison of the percent recovery of
nine E. coli cultures (average count of 5 replicates),
obtained by the pour plate method on TSA and
M-FC media incubated at 35 (air), 43 (Blue-M),
and 44.5 C (GCA) is listed in Table 2. Results ob-
tained with culture 1A are not included as it
did not grow on M-FC or TSA at 44.5 C, and was
markedly inhibited on M-FC at 43 and 35 C (30
colonies), but had good growth on TSA at 43 and
35 C (126 colonies).
The percent recovery on M-FC agar as com-
pared with TSA was best for culture numbers 9,
16, and 24 at 35 to 44.5 C. The lowest percent
recovery occurred when the counts of the other
6 cultures (1,2, 2A, 3A, 3Z, and 4) on M-FC were
TABLE 2 PERCENT RECOVERY OF E. COLI CULTURES ON TSA AND M-FC MEDIA (POURED
PLATE COUNTS9) INCUBATED AT 35, 43, AND 44.5 C.
% Recovery of E. coli on media at temperature listed.
Media
Temp
TSA
44.5
TSA
43
TSA
35
E.G.
No.
1
2
2A
3Z
3A
4
9
16
24
1
2
2A
3Z
3A
4
9
16
24
1
2
2A
3Z
3A
4
9
16
24
TSA
44.5
100
100
100
100
100
100
100
100
100
100
103
108
116
90
99
107
117
118
90
109
100
116
88
113
114
109
129
TSA
43
100b
97
93
86
111
101
93
85
85
100
100
100
100
100
100
100
100
100
90
105
93
100
98
114
106
93
110
TSA
35
111
92
100
86
113
89
88
92
77
111
95
108
100
102
88
94
108
91
100
100
100
100
100
100
100
100
100
Media E.G.
Temp No.
1
2
2A
M-FC 3Z
44.5 3A
4
9
16
24
1
2
2A
M-FC 3Z
43 3A
4
9
16
24
1
2
2A
M-FC 3Z
35 3A
4
9
16
24
TSA
44.5
76C
75
76
70
62
51
98
104
112
69
65
74
66
49
49
93
107
100
63
59
90
52
51
53
96
107
112
TSA
43
76
72
70
61
69
51
91
89
95
69
63
68
57
54
50
86
91
85
63
57
83
45
57
54
89
91
95
TSA
35
84
69
76
61
70
45
86
96
86
77
60
74
57
55
44
82
98
77
70
55
90
45
58
47
84
98
86
a Average of 5 replicates for each medium at each temperature.
b Average count on TSA at 44.5 C ^ average count on TSA at 43 C.
c Average count on M-FC at 44.5 C + average count on TSA at 44.5 C.
86
-------
#1 -
#2 -
#2A-
#3A-
#3Z -
83 to 121%
7 9 to 126%
82 to 122%
79 to 127%
74 to 135%
#4
#9
#16
#24
compared with TSA at all 3 temperatures of in-
cubation. The percent recovery, when counts on
TSA were compared with TSA at the 3 tempera-
tures, dropped below 85% in only one instance
(77%, culture #24). Although not listed in Table 2,
the counts obtained for each culture on M-FC
agar were cross-compared for the 3 incubation
temperatures, and correlation in percent recovery
was obtained as follows:
93 to 108%
95 to 105%
98 to 102%
89 to 112%
From the results of the poured plate counts, it
appears that E. coli cultures #9, #16 and #24
would be a good choice for a standardized test.
Comparison of E. coli Recovery on TSA, M-FC,
and VRB at 35,43, and 44.5 C
The percent recovery of 9 E. coli cultures
as poured plate counts on TSA, M-FC, and VRB
media incubated at 35 (air), 43 (Blue-M), and
44.5 C (GCA) is listed in Table 3. Cultures 1A and
4A (Klebsiella) which did not grow at 44.5 C on
TSA and M-FC in previous tests, produced similar
results on VRB media and will be excluded. As
noted in prior tests (Table 2), the percent recovery
of all nine cultures on TSA at the three tempera-
tures showed close correlation (83 to 121%).
The percent recovery of M-FC compared with
that of TSA, was lower at all three temperatures.
Culture numbers 2, 2A, 3A, and 9 gave best
results (61 to 98%), culture numbers 1 and 4 next
best (54 to 80%), while cultures # 24 (21 to 40%),
#3 and #16X (4 to 10%) were very low. The re-
sults obtained with VRB agar were better than
with M-FC for all nine cultures. Recovery of cul-
tures #2 and #3A on VRB were nearly identical
with TSA at all three temperatures. Culture num-
bers 1, 2A, and 9 also had a good recovery (72 -
107%), with culture numbers 4 and 24 somewhat
lower (24 to 73%) and numbers 3 and 16X the
poorest (0-29%).
Comparision of E. coli Recovery on TSA, M-FC,
and VRB at 35 and 44.5 C
Twelve E. coli cultures were tested using
poured plate counts on TSA, M-FC and VRB
media incubated at 44.5 (GCA) and 35 C (air)
(Table 4). The percent recovery, based on the
counts obtained on TSA at 44.5 C, was good
(± 10%) for 10 of the 12 cultures. Recovery of
cultures 3X and 24B were low (73 and 77%) on
TSA at 35 C. When the counts on M-FC were
compared with those on TSA, the percent recovery
varied from 31 to 88%, with cultures #16X and
#2Z lowest at 31 to 47%. Seven cultures (num-
bers 2B, 2Z, 16X, 16Z, 24, 24B, 24X) had a
similar (± 5%) recovery on M-FC at 35 and 44.5 C
when compared with TSA at 44.5 C, and at 35 C.
Culture 3A was better on M-FC at 44.5 C than at
35 C (82% compared with 64%). Culture #24 had
74% recovery on M-FC at 44.5 and 35 C vs. TSA
at 35 C, compared to 69% on TSA at 44.5 C.
The percent recovery, varying from 50 to
105%, on VRB was higher than on M-FC at both
temperatures. Culture #16X was again low (32 to
40%) in recovery. Six cultures, numbers 2B, 3A,
3X, 3Z, 16Z, and 24 had recoveries between 71
to 105%.
Variations noted with M-FC media were ap-
parent with VRB. Culture numbers 2B and 24 had
higher percent recovery (85 to 96%) when VRB at
44.5 C was compared with TSA at 44.5 and 35 C,
than when VRB at 35 C (71 -80%) was compared
in similar manner. The reverse (VRB at 35 C vs.
TSA at 44.5 and 35 C) occurred with cultures 2X,
3A, 3Z, and 16Z where recovery was 82 to 105%
on VRB at 35C as compared to 69 - 86% on 44.5C.
Culture numbers 2Z, 3X, 16A and 24B
varied in another manner. On VRB incubated at
44.5 and 35 C and compared with TSA at 44.5 C
the recovery was 75 - 100%, while the TSA at
35 C, the recovery was 63 - 77%.
Comparision of Fecal Coliform Recovery on M-FC
and VRB Media at 43 and 44.5 C
Variations in the average colony counts (5
replicates) and percent recovery of fecal coli-
form bacteria were apparent for different brands
and lot numbers of membrane filters in two dif-
ferent tests (Table 5). Sartorius lot 8 and Gelman
lot 7 provided the best results when 20 ml of
filtered creek water were incubated on 2 lots of
M-FC and VRB media at 44.5 C (GCA). Millipore
lots 6 and 11 gave the poorest results. While the
colony counts and percent recovery of the other
six lots were comparable in most cases after incu-
bation at 44.5 C, they were one-fourth to one-half
that obtained on VRB agar at 43 C (Blue-M).
87
-------
tn
Q
z
CO
in
CO
h-
Q
LU
H
OQ
O
Z
CO
to
z
o
o
LU
|-
_J
a.
Q
LU
tr
D
O
Q.
<
Q
LU
2
m
QC
Q
Z
^
co"
LL
^^
*»
-
12
Z
o
CO
LU
DC
D
t
O
O
O
LU
LL
O
DC
LU
0
CO
LU
DC
h-
Z
LU
CO
DC
LU
CL
CO
LU
1
m
1-
-o
CD
+-I
0)
3
to
CD
a
£
+-*
CO
CO
"§
c
o
1
LU
"o
a>
cd
DC
°
CO CO
1
to «*
1-
< in
tO Tfr'
i ^~
<-> 0
LU Z
co cL
TJ P
.£
o o o>
^" rL
<
CO 00 00
CO 00 00
00 CO O)
<
00 tO CN
oo en en
o o o
o o o
<
0
°
o
CO
m
CO
CO
CO
CO
co
LL
s
CO
en
o
o
CO
h-
o in oo CN in
o in i-> CN CD
o co en CM in
*~'~
< X
«- CN
s
en co co CN
C3> LO CD CN
LO LO o oo ^r
O) co r~ CN
< X
«- CN
LO
5
O CN CN CN O
O CO 00 O O
o o o o o
o o o o o
< X
«- CN
5
^~
CO CO 00 O
r**» 05 r*1"" c^
PI g °° J;
00 O 00 «-
*
<
m
DC
r» r~- rv
CD r*»- r*>
O «t CO CD <»
«- CN
"
o o o en oo
88888
CN CO LO LO CN
o r- r- en en
< X
«- CN
CO
^
SScnSS^
O 00 LO O5 CN ^~
,
O> O) O CN O IT)
^ r
-CNCNCOCO^
m
DC in
SLO^COCO
in r^ r~ oo in
t CO CO C5 ^- CO
co co r^ co CD
< <
«- CN CN CO CO t
CO
LL in
888888
^c in o in oo co
o o o oo en o>
CO CO 00 O5 O 00
«- «- O 00 O O
< <
3*>
L_ CO
5
CO
(35
m
CO
CO
en
^
en
o
o
in
en
00
o
SS
CO CN
CN in
X
«- CN
"S
^ °
r^ r»
CO
X
CD t
T- CN
§O
0
CO «-
O CN
to <-
O) «-
X
<- CN
8
X
o
CO
q-
4-J
CO
to
1-
c
o
c
o
o
- £
~D ci)
CD >
C CO
JI '
(J LO
S 5
O -^
4 (D
S3 <
" to
CO l_
c_) ^^
c~
Q. 0
i_ *-*
O
"o o
CD CD
O) O)
CO CO
CD CD
CO CO
co -Q
88
-------
TABLE 4 PERCENT RECOVERY OF E. COLI CULTURES ON TSA, M-FC, AND VRB MEDIA
(POURED PLATE COUNTS)9 INCUBATED AT 35 AND 44.5 C.
% Recovery of E. coli on media at temperature listed
Media
Temp.
TSA
44.5
TSA
35
E.G.
No.
2B
2X
22
3A
3X
3Z
16A
16X
162
24
24B
24X
2B
2X
22
3A
3X
32
16A
16X
162
24
24B
24X
TSA
44.5
100
100
100
100
100
100
100
100
100
100
100
100
93
90
125
100
130
103
106
91
111
93
133
114
TSA
35
108b
100
90
100
77
97
94
110
90
108
73
88
100
100
100
100
100
100
100
100
100
100
100
100
Media TSA
Temp. 44.5
62
47
39
82
M-FC 79
44.5 60
61
36
76
69
72
56
66
55
47
64
88
M-FC 48
35 36
32
70
69
76
59
TSA
35
68
53
31
82
60
58
58
40
68
74
53
49
72
61
38
64
67
47
34
35
63
74
55
52
Media TSA
Temp. 44.5
88
69
86
80
VRB 100
44.5 85
75
59
86
85
86
77
74
82
81
88
97
VRB 95
35 75
50
105
71
93
83
TSA
35
96
76
69
80
77
83
70
65
78
92
63
68
80
91
64
88
74
92
70
55
95
77
68
73
a average of 5 replicates for each medium at each temperature.
° average count on TSA at 44.5 C -=- average count on TSA at 35 C x 100.
Transparent colonies were frequently ob-
served on the S-8 and M-6 filters placed on M-FC
medium lot A (H6DBXP), but not on lot B
(910666). The transparent colonies were lactose
negative, dextrose acid on triple sugar iron agar
(BBL) and IMViC reactions were+.
The results for Test 1 in Table 5 were ob-
tained from 20 ml of creek water which normally
produced 25 or more colonies. While the low
counts obtained with M-FC and VRB at 44.5 C
may appear insignificant, they do indicate a
decided drop from the count obtained on VRB
agar at 43 C.
Comparison of Fecal Coliform Recovery with
Different Membrane Filter Lots on M-FC and
VRB Media
In Table 6 two different volumes (10 and
20 ml) of one sample of creek water were tested
with 5 lots of membrane filters on M-FC and VRB
media incubated at 44.5 C (GCA). The fecal coli-
form counts per 100 ml water were higher (80 -
155) on VRB agar than on M-FC medium (60 -
125). The most consistent results on M-FC medium
were produced by Gelman lot 7, Sartorius lot 5,
and Millipore lot 7 for the two volumes tested.
However, the difference in count per 100 ml,
between manufacturer and lot number, are ap-
parent (M-8, 60-125 vs. G-7, 110 - 120).
The highest count on VRB agar was obtained
with G-7 filters, with S-5 next highest and the
other three filters about equal. The M-8 filter was
the only one that recovered 100% more fecal
coliform on M-FC medium than on VRB agar. The
89
-------
TABLES PERCENT RECOVERY OF FECAL
COLIFORM FROM 20 ML. CREEK
WATER USING MEMBRANE
FILTERS ON TWO LOTS OF M-FC
MEDIA AT 44.5 C AND ON VRB AT
43a AND 44.5 C.
Media, incubation, lot no., count °
Test
No.
1
2
Filter,
Lot No.
M-6
G-7
S-7
S-8
M-11
M-12
G-7
S-8
M-FC
Lot A
1b
8
4
6
15
27
30
19
44.5
LotB
3
9
5
6
24
32
30
20
VRB
43
20
32
23
25
50
58
56
44
Agar
44.5
5
8
7
5
23
30
28
23
a 43 - actual temperature 41.8 to 43.6 C in plate.
" Average colony count, 5 replicates.
percent recovery of the other four lots varied from
60 to 94%.
Comparison of Fecal Coliform Recovery with
Different Membrane Filter Lots at 43 and 44.5C
Results of six daily tests using 20 ml volumes
of creek water with different lots of filters on
M-FC and VRB media incubated at 43 and 44.5 C
are summarized in Table 7. For the percent re-
covery of fecal coliform, the average count (5
replicates) M-FC at 44.5 C, (GCA) was compared
with the averages obtained on M-FC at 43 C
(Blue-M) and VRB at 43 and 44.5 C for each
individual filter lot number.
Overall, the lowest percent recoveries were
obtained when M-FC at 44.5 C was compared
with VRB at 43 C. At times the percent recovery
on M-FC at 43 C was identical with that obtained
with VRB at 43 C (M-8, G-6, S-8). In the majority
of tests, the percent recovery was as good or better
on VRB as on M-FC at 44.5 C.
Gelman lot 7 was most consistent, and the
best tested in four out of five tests.
Comparison of Fecal Coliform Recovery on
Rinsed and Unrinsed Membrane Filters
In an effort to determine whether residues on
the membrane filters were affecting fecal coliform
counts, four different lots were rinsed in sterile
phosphate buffer, pH 7.2, just prior to use. A
separate beaker, containing 100 ml of buffer, was
used for rinsing no more than five filters. The only
difficulty encountered was the reduction in flow
when the M-13 filters were transferred from the
buffer rinse to the filtration unit.
As indicated in Table 8, filter rinsing im-
proved the percent recovery of fecal coliform on
M-FC media with all except the M-6 filters. The
TABLE 6 FECAL COLIFORM COUNTS8 PER 100 ML. FROM 10 AND 20 ML. CREEK WATER,
FILTERED AND INCUBATED ON M-FC AND VRB MEDIA AT 44.5 C.
Average of 5 replicates.
M-FC
VRB
Filter,
Lot
M-7
M-8
S-5
S-6
G-7
count/100 ml.
count/100 ml.
count/100 ml.
count/100 ml.
count/100 ml.
20ml.
44.5
75
125
75
75
120
10ml.
44.5
70
60
80
60
110
20ml.
44.5
90
100
110
125
155
10ml.
44.5
120
110
130
80
150
90
-------
TABLE 7 PERCENT RECOVERY OF FECAL COLIFORM FROM 20 ML. CREEK WATER USING
MEMBRANE FILTERS ON M-FC AND VRB MEDIA INCUBATED AT 43a AND 44.5 C.
% Recovery
Recovery
Test
No.
1
2
3
Filter
Lot No.
M-8
G-6
S-6
S-7
S-8
M-1
M-3
G-7
S-2
M-5
M-6
G-7
S-3
M-FC
43
20
50
25
150
67
100
29
42
58
77
72
90
82
VRB
44.5
33
100
50
150
100
150
83
160
117
93
98
96
97
VRB
43
20
50
20
100
67
60
45
40
47
82
76
84
78
Test Filter
No. Lot No.
M-6
S-8
4
G-7
S-7
M-11
M-12
5
G-7
S-8
M-7
M-8
6
G-7
S-5
M-FC
43
NDC
ND
ND
ND
63
84
100
95
78
89
98
104
VRB
44.5
20
120
100
57
65
90
107
83
58
64
92
85
VRB
43
5
24
25
17
30
47
54
43
41
49
68
71
3 43 - Actual temperature 41.8 to 43.6 C in plate.
° Average count 5 replicates (M-FC at 44.5 C) ^ average count 5 replicates, (media and temp, listed)
x 100. For example, in Test 1, M-8, M-FC at 43 C, the 20% means that M-FC at 44.5 C recovered
20% of M-FC at 43 C.
c Not done.
non-rinsed filters recovered 73 - 81% of the buffer
rinsed filters.
On VRB agar, the non-rinsed filters of M-6
and M-13 recovered 80% of the counts obtained on
the rinsed filter, while with the G-7 and S-6 filters
the percent recovery was 104% and 110% respec-
tively. The M-13 filters gave the only consistent
results (75 - 80%) on M-FC and VRB media.
Special Test - A
E. coli Grown on Surface or Embedded in Medium
Culture #1A, originally isolated as a deep
blue colony on M-FC medium incubated at 43 C,
produced few colonies on TSA medium and none
on VRB and M-FC agars at 44.5 (GCA) as poured
plate counts. This #1A culture grew best on TSA
plates with marked inhibition noticeable on M-FC
and VRB when incubated at 43 and 35 C.
In a repeat test, culture #1A was plated using
M-FC and TSA agars (5 replicates each). One ml of
diluted culture was used for 15 x 100 mm plates
and 0.1 ml of the same dilution and a more con-
centrated suspension was used with two sets of
12 x 50 mm plates. No growth occurred on any of
the poured plates after overnight incubation at
44.5 C in the (GCA) water bath.
In contrast, the #1A culture on membrane
filters and pads with M-FC medium that were
incubated at 44.5 C had growth on Gelman lot
numbers 3 and 6, but not on Millipore lot num-
bers 3 and 13 and Sartorius lot numbers 1 and 4
(Table 1). Growth was produced on all six lot
91
-------
TABLE 8 FECAL COLIFORM RECOVERY FROM 20 ML. CREEK WATER USING BUFFER RINSED
OR NOT RINSED MEMBRANE FILTERS ON M-FC AND VRB MEDIA INCUBATED AT
44.5 C.
M-6
M-FC-Pad
VRB - Agar
Filter,
Lot No.
No
Rinse
Rinse
No
Rinse
Rinse
Mean count a
%b
14 12
117
16 20
80
M-13 Mean count 12 16
% 75
20 25
80
G-7 Mean count 26 32
O/ O *I
/O O I
27 26
104
S-6 Mean count 16 22
% 73
23 21
110
a Average of 5 replicates
" % = Mean counts, no rinse -=- rinse, x 100
numbers when the filters were placed on TSA
medium and incubated at 44.5 C.
The average colony count of 0.1 ml of con-
centrated suspension of E. coli #2, filtered, and in-
cubated on TSA agar at 35 C was 115, while that
on an M-FC (pad) was 101. The average colony
count of the 1 ml filtered and placed on a pad
containing M-FC broth and incubated at 35 C was
179. Due to excess moisture and the volume
(0.1 ml) streaked on the 12 x 50 mm plates, the
bacterial growth ran together and counts were
too erratic to be used.
Culture #2 was tested by filtering (M-13)
0.1 ml and 1 ml of two different concentrations
and placing the filters on 12 x 50 and 15 x 100
mm plates of M-FC and TSA. Then the same
volumes (0.1 and 1.0 ml) were respectively spread
on M-FC and TSA agar plates (12 x 50 and 15 x
100 mm). Five replicates of each category were in-
cubated in a 44.5 C (GCA) waterbath.
The agar plates that were streaked or received
the filters (1 ml) had average colony counts
(44.5 C) as follows (E. coli #2):
E. coli #2
1 ml (filter)
1 ml (filter)
1 ml (streaked)
1 ml (streaked)
12x50
119
169
omitted
omitted
15 x 100
134
204
73
71
Media
M-FC
TSA
M-FC
TSA
The average counts (71 - 73) of the surface
streaked (1 ml) plates therefore were much lower
than when the filters were placed on agar or pads.
E. coli #2, in two concentrations, was then
tested by the pour plate method using M-FC and
TSA agars in 12 x"50 (0.1 ml) and 15 x 100
(1 ml) plates incubated at 44.5 C (GCA bath).
The colony count of #2 culture in the M-FC agar
with both concentrations and plate sizes was about
92
-------
one-half that which occurred with the TSA agar.
Repeating the experiment with the #2 culture
gave the same results with poured plate counts.
One ml of E. coli culture #1 was then tested,
as mentioned above for #2, using filters on agar
(12 x 50 mm plates), and agar poured plates (15 x
100 mm). The results (average colony counts) were
as follows:
Filter M-FC TSA Temperature
M-6 20 24 44.5 C (GCA)
G-5 27 32 44.5 C (GCA)
S-8 20 30 44.5 C (GCA)
S-3 20 23 44.5 C (GCA)
Poured 14 27 44.5 C (GCA)
Poured 10 25 35 C Incub.
When 0.1 ml of concentrated #1 suspension
was poured in 12 x 50 plates, the average count on
M-FC was 12 and on TSA, 24. Using 15 x 100
plates, the poured plate counts were 5 on M-FC
and 22 on TSA. Incubation was 44.5 C (GCA)
in both tests.
It should be noted that where two concen-
trations of a culture were used, the counts are
comparable only with one or the other concen-
tration.
There was, therefore, a difference between
placing the culture on the surface of the agar
medium via a filter or by spreading the liquid and
incorporating the culture in the medium as in the
poured plate method with E. coli #1, #1A, and
#2.
Special Test - B
Tests were conducted to determine: 1)
whether the bacteria are trapped on the filter sur-
face but not all will grow, 2) some are trapped
deep in the membrane and would grow through
the other side (bottom of filter), and 3) some of
the bacteria actually escape through the membrane
into the filtrate. It was assumed that all bacteria
trapped on the surface of a membrane filter
would be transferred by contact to the surface of
an agar plate . The contact time used was one, two,
and four hours, or the filter was left in place.
Two cultures, #2 and 3A, were diluted and
1 ml (5 replicates for each) filtered through Gel-
man, Millipore, and Sartorius membrane filters.
For comparison, 1 ml was also plated with TSA for
pour plate counts and incubated at 44.5C and 35C.
One group of filters was placed upside down
and a second group right side up on corresponding
pads saturated with M-FC medium. In a similar
manner the same brand and lot numbers of filters
were placed on M-FC agar (Test A) or VRB agar
plates (Test B) and incubated at 35 C and 44.5 C.
One hour later, some of the upside down filters
were removed from the plates, with others left in
place. This procedure was again repeated after
four hours incubation, and all plates returned to
their respective incubation overnight (Table 9).
No bacterial colony growth was evident on
either side of the filters placed upside down on the
pads or from the filtrates. However, the colonies
present on filters upside down on the agar plates
could be counted through the agar. The colony
counts on plates (and percent recovery) with filter.
removed after one hour incubation were lower
than those removed four hours later. The latter
colony counts were higher than where filters were
left on the plates, (Test A, Table 9), with the ex-
ception of culture 3A on Millipore lot #13. How-
ever, the recovery of E. coli by the filters on the
pads and agar plates did not equal the poured plate
counts on TSA medium for culture 2 and 3A on
M-FC at 35 C and 44.5 C.
In test B (Table 9), when the same filters and
cultures were used, there was a marked increase in
present recovery of culture #2 with filters placed
on VRB agar at 35 and 44.5 C. A slight increase
occurred when filters were removed after two
hours. However, results of E. coli #3A on VRB
agar were about the same as with M-FC medium.
DISCUSSION
To determine efficiency of membrane filters
for the fecal coliform test one must test polluted
water samples, and confirm that the colonies are
fecal coliform. To detect deficiences in the test
procedure, E. coli strains are used, but these vary
in different reports (1, 5, 6, 8). However, all test
methods and equipment must be accurate and
rigidly adhered to. The water baths and incubators
used must maintain a temperature of 44.5 C inside
the plate, but actual reports are scarce. The GCA -
Precision Scientific Co. water bath was the only
93
-------
TABLE 9
COLONY COUNTS AND PERCENT RECOVERY ON MEMBRANE FILTERS PLACED
ON PADS AND AGAR MEDIA AND REMOVED OR LEFT IN PLACE.
M-FC Medium or agar plate and incubation
Test A
Culture
2
2
2
3A
3A
3A
Filter3
Lot No.
G-6 count
%d
S-4
%
M-13
%
G-6
%
S-4
%
M-13
%
Pad
44 35
43
60
35
49
49
68
74
71
57
55
61
59
55
79
32
46
51
73
77
74
67
60
73
65
1 hr.
44 35
21
29
17
24
27
38
23
22
54
52
65
63
31
44
33
47
40
57
24
21
44
39.
38
34
4hr.
44 35
47 57
65 81
36 35
50 50
51 47
71 67
75 86
72 77
43 63
41 56
63 76
61 68
Left on
44 35
18 23
25 33
30 22
42 31
31 32
43 46
50 72
48 64
54 68
52 61
51 65
49 58
TSAC
44 35
72 70
100 100
104 112
100 100
Filter Removed
Test B
Culture
a
b
2
2
2
3A
3A
3A
Filter8
Lot No.
G-6 count
%
S-4
%
M-13
%
G-6
%
S-4
%
M-13
%
Gelman, Sartorius, Millipore
Pili-orc romr\\/orl aftor 1 hr A.
VRB
44 35
40
80
44
88
47
94
24
77
18
58
18
58
(average
hrc nr If
48
92
54
104
49
94
23
82
20
71
14
50
of 5
aft nr
2hr
44 35
56
112
58
116
46
92
18
58
16
52
15
48
replicates)
i nlato
58
112
55
106
60
115
18
64
18
64
15
54
4hr
44 35
ND
ND
ND
ND
ND
ND
Left on
44 35
ND
ND
ND
ND
ND
ND
TSAC
44 35
50 52
100 100
31 28
100 100
c Poured Trypticase soy agar plate count.
" Percent recovery, count at 44 =- TSA count at 44.5 C.
ND = not done.
35 -j- TSA count at 35 C.
94
-------
one used in our tests that kept a steady 44.5 C
temperature inside the plate and the surrounding
water.
Three incubation temperatures were utilized
to determine differences in percent recovery. The
number of bacteria that grew on TSA medium at
35 C should be an indication of the number of
E. coli present in the dilution being tested. From
this base, comparisons with counts obtained at 43
and 44.5 C were made, not only with TSA medium
but also with M-FC and VRB. The 43 C incubation
temperature (41.8 to 43.6 C inside plate) was used,
as we had noted these temperatures in improperly
working water baths and air incubators set at
44.5 C.
In prior reports, the exact number of bacteria
cells being tested was based on poured plate counts
(8), or on counts obtained on membrane filters
with pads, using a non-inhibitory medium (5). That
either or both of these methods could effect the
results obtained with E. coli cultures was apparent
in our study.
The results we obtained indicated that eleven
of the E. coli test strains varied not only in their
ability to grow at 35, 43, and 44.5 C, but also on
or in the culture media used. Culture 1A would
grow on TSA agar, but not on M-FC agar at 44.5 C.
Theoretically, a temperature tolerant culture
should grow as well at 35 C as at 44.5 C on a non-
inhibitory (TSA) medium, but this did not always
occur (Table 1, cultures 2 and 9). At times, the
percent recovery of membrane filters on M-FC vs.
TSA at 44.5 C was better than that obtained on
TSA at 44.5 (C) and 35 C (B), (Table 1, cultures
2A and 3A). This is similar to the results obtained
by Presswood and Brown (8) with three E. coli
cultures. However, while they used M-FC medium
for both the membrane filter and poured plate
counts, there is agreement that the membrane
filter was responsible for low percent recovery
more frequently when M-FC medium was used.
When the E. coli strains were tested by the
pour plate method, the percent recovery of nine
cultures on TSA medium was very close for the
three temperatures of incubation (Table 2). In con-
trast, only three cultures (9, 16, 24) grew as well
on M-FC as on TSA at the three temperatures
of incubation. The addition of VRB medium for
comparison with TSA and M-FC media by the pour
plate method indicated that VRB had a better
percent recovery than M-FC (Tables 3, 4). Culture
#9 again produced good results, but in contrast
culture #24 showed inhibition on M-FC and VRB
agar. This would indicate variations of temperature
tolerance in some of the test cultures.
A comparison of VRB agar with M-FC
medium in respect to recovery of fecal coliforms
from polluted water indicated that VRB was better
than M-FC when the incubation temperature was
44.5 °C (Tables 5, 6, 7), and quite superior at
43 C (Table 5).
The possibility that the membrane filter-
culture media interaction was responsible for low
recovery was obvious when the membrane filter
counts were compared with poured plate counts.
An attempt to transfer bacteria on the membrane
filter to M-FC medium did not improve percent
recovery as well as on VRB medium (Table 9).
Placing three lot numbers of filters containing
E. coli #2 on VRB agar at 35 and 44.5 C pro-
duced much better results than on M-FC with
pads (Table 9, Test B). An improvement in percent
recovery was also noted when the same culture
and membrane filters were placed face down on
the VRB agar, and then removed two hours later.
However, when E. coli #3A was tested in the same
manner, there was little difference in percent
recovery on VRB and M-FC media. The results
obtained on M-FC with E. coli #2 and #3A (Table
9, Test A) are similar to those obtained in Table 1
(Category B). Therefore, we must assume that
VRB was better than M-FC, at least for culture #2
in this test. Whether placing the membrane filter
face down on the VRB agar and stimulating better
growth by direct contact can be compared with the
two layer technique (7, 9) requires further study.
Temperature alone did not affect the growth
of the nineteen E. coli strains tested in an equal
manner since all grew well on TSA medium. The
E. coli strains were affected more by their ability
to form less colonies on M-FC agar than on TSA
and VRB media. Differences in strains or isolates,
with some showing no growth inhibition while
others were markedly inhibited, were apparent. It
therefore appears that the M-FC medium did
support good growth of some, but not all, of the
E. coli strains. This applied to pour plates as well
as membrane filter tests. Since one lot medium
was used for most all tests, different lot numbers
were not responsible.
A comparison of different brands and lot
numbers of membrane filters indicated variations
in percent recovery of E. coli cultures were due to
95
-------
the filter, the culture, or the test method. Culture
#1 as poured plate count recovered 52 to 58% of
the E. coli on M-FC when compared to TSA at
44.5 and 35 C (Table 1), and similar results were
obtained with poured plate counts (Special Test
A). In each of these two tests, the membrane
filter counts were higher (67 - 109%) and mem-
brane filters M-6, S-8, and S-3 produced identical
results in respect to percent recovery (67 - 87%)
at 44.5 C. Cultures #1A and 2 also produced
lower counts when tested by the poured plate
method and compared with jnembrane filter
counts. The M-FC agar was more inhibitory (low-
ered counts) than the TSA for these cultures via
the membrane filter or poured plate method. With
culture #2, this inhibitory activity of M-FC was
less marked when streaked on the solidified M-FC
and TSA plates, even though the surface streaked
count was much lower than the poured plate count
(Special Test A).
Therefore, in evaluating the efficiency of the
membrane filters in respect to percent recovery of
E. coli bacteria, the ability of the test strain to
grow in or on the medium should be considered.
In Table 1, when the membrane filters were placed
on M-FC medium (pads) or on TSA agar, Category
A probably provides a fair comparison of the dif-
ferent lot numbers. That the percent recovery
might be due to the method, i.e. Category C,
where the membrane filter was placed on TSA
(44.5 C) and compared with poured plate count
on TSA (35 C) is evident with cultures #1 and
#9 vs. #2 and #3A. All four cultures (in fact all
cultures tested), were almost equal in percent
recovery when compared at 44.5 C and 35 C on a
poured plate basis (Tables 2, 3, 4). In this respect,
there was little difference reported in bacteria
counts obtained when membrane filters were
tested with plate count broth and total coliform
broth at 35 C, but a decided drop occurred in the
counts when plate count broth and M-FC broth
were used at 44.5 C (5).
Why the membrane filters are so erratic is
still not clear. Rinsing the filters prior to use
helped somewhat \jn limited tests (Table 8). The
bacteria trapped on the membrane filter surface
required more than one hour contact time with a
solid culture medium to grow. It is possible the
bacteria on the filters may be in clusters, or some-
how injured and do not grow as readily as in
poured agar plates. However, we did find two
strains of E. coli that grew better on the surface
of the medium than in the poured agar. If this was
due to melted agar being too hot, all samples
done at that time would have been affected, but
this did not occur.
While it appears that VRB medium might be
as effective or better than M-FC medium, this
should be evaluated by others. Klein and Fung
(6) reported that the VRB poured plate method
was as good as the MPN and membrane filter
method for the fecal coliform tests. However, an
air incubator at 44.5 C and only two replicates
per sample were used in their tests.
SUMMARY
The conclusions to be drawn from this study
emphasize that a test method should be standard-
ized with respect to the E. coli strain, the culture
medium used to determine the actual number of
bacteria in the dilution being tested, the best
method to be utilized (poured plate, pad and
broth, or solidified agar plate) and control of tem-
perature of incubation inside the medium. Studies
to further standardize the culture medium to be
used at 44.5 C have been initiated (9) and hope-
fully this will provide more uniform blue colony
types.
ACKNOWLEDGEMENT
The technical assistance of Greg Bossart,
Mike Davis, Loretto Yanaitis and Charlotte Smith
is greatly acknowledged.
REFERENCES
1. Dutka, B.J., M.L. Jackson, and J.B. Bell.
Comparison of autoclave and ethylene oxide
sterilized membrane filters used in water
quality studies. Appl. Microbiology. 28:
474-480, 1974.
2. Evaluation of Water Laboratories, U.S. Dept.
H.E.W. Public Health Service Public No. 999-
EE-1,p44, 1966.
3. Fishbein, M. The aerogenic response of E. coli
and strains of Aerobacter in EC broth and
selected sugar -broths at elevated tempera-
tures. Appl. Microbiology. 10:79-85, 1962.
4. Hartman, P.A., P.S. Hartman, and W.W.
Lanz. Violet red bile 2 agar for stressed coli-
forms. Appl. Microbiology. 29:537-539, 1975
5. Hufham, J.B. Evaluating the membrane fecal
coliform test by using E. coli as the indicator
organism. Appl. Microbiology. 27:771-776,
1974.
96
-------
6. Klein, H. and D.Y.C. Fung. Identification and
qualification of fecal coliform using violet red
bile agar at elevated temperature. ASM Alle-
gheny Branch Meeting, Pittsburgh, Pa., 1974.
7. Nash, H.D. and E.E. Geldreich. Applicability
of a modified membrane fecal coliform med-
ium and technique. ASM abstracts, 75th
Meeting, N.Y., N.Y., 1975.
8. Presswood, W.C. and L.R. Brown. Compari-
son of Gelman and Millipore membrane filters
for enumerating fecal coliform bacteria. Appl.
Microbiology. 26:332-336, 1973.
9. Rose, R.E., E.E. Geldreich, and W. Litsky.
Improved membrane filter method for fecal
coliform analysis. Appl. Microbiology. 29:
532-536, 1975.
10. Shahidi, S.A. and M.H. Backer. Effect of fecal
coliform organism medium. ASM Abstracts,
73rd Meeting, 9., 1973.
11. Sladek, K.J. Optimum membrane structures
for growth of coliform organisms. Ft. Lauder-
dale, Fla., This Symposium, 1975.
12. Standard Methods for the Examination of
Water and Wastewater. APHA, AWWA,
WPCF, 13 ed. 1971.
97
-------
RECOVERY CHARACTERISTICS OF BACTERIA INJURED
IN THE NATURAL AQUATIC ENVIRONMENT
by
Gary K. Bissonnette, James J. Jezeski,
Gordon A. McFeters and David G. Stuart
Department of Microbiology
Montana State University
Bozeman, Montana
ABSTRACT
The recovery and enumeration of indicator
organisms from natural waters were found to be
adversely influenced by several phases of the
system. These dictated the degree of sub-lethal
injury observed in water-borne bacteria that ren-
dered them incapable of growth under conditions
that are routinely used for their identification.
Therefore, the recovery of indicator organisms
following water-induced injury was examined
using various isolation procedures. A high degree of
variation was found and the membrane filtration
technique was the least efficient. The injury
inflicted upon E. coli and S. faecalis by the
aqueous environment could be reversed by a short
enrichment treatment. This procedure was es-
pecially applicable to membrane filtration pro-
cedures.
INTRODUCTION
In evaluating the problem of detecting par-
ticular microorganisms from various sources,
proper consideration must be given to the in-
fluence of environmental factors upon detection
methods. Data are available indicating that after
exposure to freezing, heating, or freeze-drying,
some microorganisms are either physiologically
debilitated or injured to such an extent that
significant problems arise upon attempts at de-
tection and enumeration. Such stress-injured
microorganisms become more sensitive to in-
hibitory agents in specific selective media and are
unable to grow and produce colonies.
Most sanitary indicator organisms and enteric
water-borne pathogens are bacteria whose natural
habitat is the intestine of man and warm-
blooded animals. Once these microorganisms are
deposited into water they are in an environment
that is not favorable to the maintenance of viabil-
ity for most heterotrophic bacteria. Therefore,
proper interpretation of sanitary water quality
data relies partly on a basic understanding of
survival characteristics of bacteria in water. In the
majority of reported survival studies, only two
sub-populations of the total have been considered:
those cells which can withstand the aquatic en-
vironment as reflected by their detection and
enumeration when using standard laboratory
procedures and conversely, those cells which
cannot persist in the unfavorable environment,
resulting in death and non-detection. There is a
paucity of available literature concerning the
possibility that a substantial fraction of the total
population of cells in water may be injured to the
extent that they fail to grown on selective media.
This report presents research directed toward
determining whether aquatic environments pro-
voke stress upon indicator bacteria such that these
cells become physiologically debilitated and
cannot be detected by direct selective procedures.
The report also describes methods to recover
these injured cells.
In the first study, membrane chambers were
filled with washed suspensions of a typical EC+
strain of Escherichia coli, type I, and immersed
at different stream sites, and sampled daily. The
organisms were tested for their ability to form
colonies on Trypticase soy agar supplemented with
yeast extract (TSY agar), to indicate the maximum
number of recoverable organisms, and on desoxy-
cholate agar (DLA), to yield the number of bac-
98
-------
teria that were able to form colonies in the pres-
ence of this selective medium. From these data,
the percentage survival and the percentage death
were obtained in addition to the fraction of the
total viable population that was injured to the
extent that they were unable to grow on the
selective medium. This portion of the total viable
population that became debilitated in water repre-
sents the cells that would not be enumerated using
the current standard procedures.
The results of the first study indicated that
the number of organisms in these segments of the
total bacterial population varied considerably
among the stream sites that had different water
quality characteristics. Also, at the sites where
the greatest percentages of death was observed,
there was a greater proportion of the bacteria
that were injured. The extent of this non-lethal in-
jury varied from 10% to 96% of the total viable
population after four days, depending on the
physical and chemical characteristics of the water.
In addition, it was observed that the proportion of
the survivors that reflected non-lethal injury
increased as the length of exposure to the aquatic
environments increased. This observation was also
found when comparable experiments were con-
ducted using Streptococcus faecalis.
Because the isolation of various indicator
bacteria is the primary objective in assessing the
microbiological quality of water, the use of selec-
tive media is required to suppress the growth of
other organisms that could interfere with the
detection and enumeration of the desired micro-
organisms. Inhibitory agents in these media may
exert unexpected inhibition on cells subject to
stress. Thus, the combination of environmental
stress with the subsequent utilization of selective
media may result in the diminished recovery of
injured cells. This problem is further compli-
cated by the varying degrees of injury observed as
a function of time and water characteristics.
Therefore, improved enumeration methods should
be developed to more efficiently recover injured
bacteria.
In the second study, experiments were done
to establish the relative efficiency of various
selective media in recovering E. coli that were
progressively debilitated in natural water. Samples
were enumerated daily for four days by different
methods using several media. These studies re-
vealed the superior recovery efficiency of liquid
media (MPN) over broth plating and membrane
filtration methods: MPN with TSY = MPN with
lactose broth > MPN with brilliant green lactose
bile broth > plating with desoxycholate lactose
agar > membrane filtration with M-Endo MF
medium > membrane filtration with M-FC med-
ium. It should be emphasized that the membrane
filter procedures were less efficient in recovering
injured bacteria found in natural waters than the
other selective procedures by a statistically sig-
nificant margin. Similar results were found when
comparable experiments were done using S. fae-
calis and the appropriate media. These findings in-
dicate that, while conventional methods might be
adequate to enumerate bacteria from water exert-
ing minimal environmental stress or cells deposited
into the water shortly before it was sampled, care-
ful consideration should be given to the develop-
ment of new methods to recover debilitated
bacteria from stressful water environments. This
is particularly true, since injured cells were found
to be a large fraction of the total population when
exposed to certain aquatic environments for as
little as two days. Therefore, it is important that
this damaged population be recovered to more
correctly evaluate the bacterial quality of many
waters.
In the third study, experiments were con-
ducted to determine if cells injured by environ-
mental stress in natural waters had the capacity to
repair themselves when exposed to a suitable en-
vironment. An EC+ strain of E. coli was exposed
to water in membrane chambers for two days, the
organisms removed, inoculated into liquid TSY
medium and enumerated every hour by plating on
TSY and DLA agar for six hours. This procedure
was used to compare the growth kinetics of bac-
teria exposed to water with a control suspension
of the same organism that was not subject to
environmental stress. As in the previous experi-
ments, the difference between the counts on the
TSY and DLA media reflected the debilitated
population. Control cultures that were not exposed
to the water contain virtually no cells in this
weakened physiological state and the bacteria
exhibited normal growth kinetics. However, the
cell suspension that had been exposed to the
stresses of the aquatic environment for a period
of two days contained a substantial proportion
(95%) of injured cells, and the lag period was three
times longer than in the control suspensions. As
the injured cells were exposed to the TSY broth,
the injured population progressively repaired itself,
so that after three hours, they were capable of pro-
ducing colonies on both TSY and DLA agar.
Further experiments, where bacteria were in the
water for longer times, demonstrated increases in
99
-------
the lag period and recovery time. The same kinds
of observations and relationships were found
when similar experiments were conducted using
S. faecalis. These studies indicate that appreciable
repair from environmental injury may be attained
by exposing the cells to a rich, non-selective
medium for a short period prior to the use of a
selective medium. This procedure affords the
more complete enumeration of indicator bacteria
from the aquatic environment on selective media.
Enrichment techniques appear to be especially
applicable to membrane filtration methods that
have a low efficiency of recovery for injured cells
and are easily adapted to enrichment procedures.
Additional studies were conducted to deter-
mine if a two hour enrichment step, before ex-
posure to the selective medium, would enhance the
recovery of total and fecal coliform bacteria from
a mixed natural population of bacteria, as they
were in contact with the aquatic environment for
various times. In these experiments, suspensions
of raw sewage were placed in membrane chambers
and then in a stream, followed by daily sampling
using the membrane filter procedure. Then dupli-
cate filters were incubated on M-Endo agar at 37 C
and M-FC agar at 44.5 C, after a two hour enrich-
ment step on TSY agar at 37 C. These experiments
indicated that the enrichment procedure improved
the recovery, in comparison to control samples,
where the enrichment was omitted. Throughout
the entire three day period that the bacteria from
the sewage were in contact with the aquatic
environment, the most efficient method for re-
covering total and fecal coliform bacteria was a
two hour enrichment on TSY agar followed by
transfer to the selective medium. The two hour
enrichment period provides a non-toxic, nutrient
rich environment for the gradual adjustment and
repair that is needed by these organisms to success-
fully grow on the selective medium. The adoption
of enrichment techniques would help overcome
some of the limitations of the membrane filtra-
tion procedure, arising from the exclusive use of
selective media, by providing the necessary bridge
for bacterial adaptation between the environment
encountered in natural waters and the selective
media in the laboratory.
ACKNOWLEDGEMENTS
This project was supported by funds from the
U.S. Department of the Interior authorized under
the Water Resources Research Act of 1964, Public
Law 88-379, and administered through the Mon-
tana University Joint Water Resources Research
Center (grants OWRR B-035 Mont, and B-040
Mont.).
100
-------
A LAYERED MEMBRANE FILTER MEDIUM FOR IMPROVED RECOVERY
OF STRESSED FECAL COLI FORMS
Robert E. Rose, Edwin E. Geldreich and Warren Litsky
ABSTRACT
A two layered agar method employing tem-
perature acclimation and lactose enrichment with
diffusion transfer into M-FC agar is proposed for
improved recovery of stressed fecal coliforms on
the membrane filter. The procedure was field
tested in three laboratories using samples of raw
and chlorinated wastewater, reservoir, river and
marine waters. Verification of 1013 fecal coli-
form colonies isolated from 61 water samples
averaged 92 percent using this proposed procedure.
Comparisons with the Standard Methods M-FC
procedure revealed the two-layered agar method,
provided an overall increased sensitivity for fecal
coliform detection in chlorinated secondary
effluents, marine waters and any natural waters
that contained pollutants with heavy metal ions.
INTRODUCTION
Recent reports of reduced recovery of fecal
coliforms from chlorinated sewage effluents, when
the membrane filter procedure is used (1-3), have
caused much concern both to regulatory agencies
and thos laboratories involved in monitoring.
Until this problem is resolved, Federal require-
ments for the bacterial quality assessment of ef-
fluents under the National Pollution Discharge
Elimination System has specified that fecal coli-
form densities must be determined by the multiple
tube procedure. As a result of this decision, many
small laboratories are unable to meet this analysis
protocol because of a limited bacteriological
testing capability based solely on the membrane
filter concept.
Review of the attenuated fecal coliform
recovery problem suggested that chlorine inactiva-
tion of some coliform cells might be reversed
provided enrichment (4, 5) and temperature
acclimation (5, 6) were possible without com-
promising specificity of the test. All enrichment
procedures, previously developed for the mem-
brane filter technique, required a manual transfer
of the membrane filter cultures from one medium
to another (4-9). Recognizing media manipulations
is time consuming in the laboratory so a new ap-
proach incorporating a two layer enrichment-
differential growth medium was explored. The
method was evaluated on a variety of waters which
might contain attenuated fecal coliforms.
MATERIALS & METHODS
Preparation of a two layer medium (Table 1)
was accomplished by dispensing approximately
5 ml of M-FC agar into each culture dish (50 x
12mm), permitting the agar to solidify, then add-
ing 2 ml of normal strength lactose broth in 1.5
percent agar over the M-FC agar. Since the ingred-
ients of the two agar layers will eventually diffuse
TABLE 1. FORMULATION OF THE TWO-
LAYER MEDIUM
Differential Medium (Botton Layer) 3.7 gm
M-FC Medium 1.5 gm
Agar 100 ml
Distilled Water
Resuscitation Medium (Top Layer) * 0.3 gm
Beef Extract 0.5 gm
Peptone 0.5 gm
Lactose 1.5gm
Agar 100 ml
Distilled Water
* Resuscitation medium equals 1x lactose broth
plus 1.5% agar.
101
-------
into each other, it is suggested that the base
M-FC agar be prepared in advance and the lactose
agar overlay added within one hour prior to use.
After the membrane filter was placed on the
two-layer medium, the plates were incubated at
35 C for 2 hours after which the temperature was
increased to 44.5 C for 22-24 hours to attain the
necessary selectivity. All blue colonies were count-
ed with the aid of a binocular scope employing
10-15x magnification and a flourescent light
source. Verification of fecal coliforms isolated on
the test medium was performed by subculturing
each blue colony into either phenol red lactose
broth or lauryl tryptose broth for 24 to 48 hours
at 35 C. Tubes showing gas production within this
period were subcultured to E.G. broth and incuba-
ted in a water bath for 24 hours at 44.5 C ± 0.2 C.
Samples were collected from diverse waters
that included estuarine waters of Massachusetts,
raw sewage and chlorinated sewage effluents from
the Billerica Massachusetts sewage treatment plant,
polluted stretches of the Merrimack, Fort and Mill
rivers, and sampling at varying depths of a raw
water impoundment near Walton, Ky. The reser-
voir water samples were collected during a period
of prolonged dry weather and following a signifi-
cant stormwater runoff into the impoundment.
All bacteriological examinations of these waters
were performed at one of three different labora-
tories located near the sampling sites: Millipore
Corporation Laboratory at Bedford, Mass.; Uni-
versity of Massachusetts Department of Environ-
mental Sciences Research Laboratory at Amherst,
Mass.; and the US EPA Water Supply Research
Laboratory, Cincinnati, Ohio. Three or five repli-
cate portions were prepared for cultivation on both
the two-layer experimental medium and the
M-FC agar direct method as recommended in
Standard Methods (10).
RESULTS AND DISCUSSION
A total of sixty-one water samples were
analyzed in the evaluation of the two-layer
medium procedure. The choice of samples used in
this evaluation were oriented to those waters that
might have attenuated fecal coliform populations
resulting from exposure to chlorination of sewage
effluents, contact with the marine environment,
antagonistic action of metal ions in chemically
polluted fresh waters and to natural forces of
self-purification induced during storage of im-
pounded natural waters. Data presented in Tables
2 to 5 are based on average colony counts per
membrane filter test rather than as counts per 100
ml so as to avoid distortion from factoring dilu-
tions to the base 100 ml level.
The fecal coliform colony counts on the two-
layer method were greater than those detected by
the companion direct M-FC procedure when
chlorinated sewage effluents were examined
(Table 2). The average colony count ratio obtained
by the experimental procedure and the standard
M-FC method was 18.2. Inspection of the com-
parative data for raw sewage revealed only three of
18 samples had a ratio higher than the lowest ratio
calculated in the chlorinated sewage effluents.
These preliminary findings suggest attenuated
fecal coliforms are more numerous in chlorinated
TABLE 2. COMPARISON OF THE 2-LAYER
AGAR VS DIRECT M-FC
PROCEDURES COLIFORM
DENSITIES FROM RAW AND
CHLORINATED SEWAGE
Ratio
M-FC 2-LayerAgar 2-LayerAgar
Source Count Count Direct M-FC
Raw Sewage 57
23
32
16
16
10
4
1
2
46
5
15
24
15
25
52
12
103
26
72
20
26
31
11
10
14
91
12
23
33
22
42
98
32
1.8
1.1
2.3
1.3
1.6
3.1
2.8
10.0
7.0
2.0
2.4
1.5
1.4
1.5
1.7
1.9
2.7
5.7
Chlorinated Sewage6
1
26
5
40
228
26
127
19
Avg. 2.9
38.0
26.0
4.9
3.8
Avg. 18.2
102
-------
sewage than might be expected in raw sewage and
thus support the observations of Lin (1).
Attenuated fecal coliforms also appear to be
present in the estuarine samples collected from a
coastal site in Massachusetts (Table 3). Stevens
et al. (9) were of the opinion that the initial shock
at 44.5 C adversely affected reproduction of meta-
bolically injured cells present in the marine water
environment. Here again, the two layered medium
recovered from 3.8 to 7 times more fecal coliform
colonies than the direct M-FC procedure.
TABLES. COMPARISON OF THE 2-LAYER
AGAR VS DIRECT M-FC
PROCEDURES FOR FECAL
COLIFORM DENSITIES FROM
MARINE WATERS
Ratio
M-FC
Source Count
Marine Waters 3
21
3
21
3
30
20
2- Layer Agar
Count
12
79
12
79
16
210
92
Avg.
2-Layer Agar
Direct M-FC
4.0
3.8
4.0
3.8
5.3
7.0
4.6
, 4.6
Attenuated fecal coliform occurrences in
fresh waters are more varied as related to the in-
tensity of heavy metal ions found in a particular
stretch of river (Table 4). The toxic effect of heavy
metal ions in river water has been reported to be a
factor in coliform recovery from transported
samples (11-13). Adsorption of metal ions from a
water sample may also occur on the membrane fil-
ter, producing a concentrated toxic effect (11).
Thus, the more frequent occurrence of attenuated
coliforms observed in the Merrimack River, as
contrasted to data obtained on Fort River, is
assumed to be a reflection of the more numerous
industrial waste discharges to the Merrimack
River.
Results obtained from a study of a water
reservoir supply (Table 5) indicate the least
amount of difference between the proposed two
TABLE 4. COMPARISON OF THE 2-LAYER
AGAR VS DIRECT M-FC
PROCEDURES FOR FECAL
COLIFORM DENSITIES FROM
RIVER WATERS
Ratio
M-FC 2-Layer Agar 2-Layer Agar
Source Count Count Direct M-FC
Merrimack
River
Fort River
Mill River
4
76
4
76
23
15
8
11
4
7
9
24
6
22
25
183
25
183
53
38
56
21
11
10
16
29
12
60
6.3
2.4
6.3
2.4
2.3
2.5
7.0
1.9
2.8
1.4
1.8
1.2
2.0
2.7
Avg. 3.1
layered medium and the direct M-FC medium.
This water contained few fecal coliforms per
100 ml during the day weather and no industrial
waste discharge in the drainage basin. Fecal pollu-
tion that enters from stormwater runoff is from
cows grazing on the hills surrounding the reser-
voir. For these reasons, those few attentuated fecal
coliforms present, represent debilitated cells com-
mon to natural die-off.
Verification of 1013 typical blue colonies
(Table 6) from the two-layer agar procedure con-
firmed our contention that the reported increased
fecal coliform recoveries attributed to this pro-
posed procedure were valid. Of 1013 colonies
picked from all samples tested, 930 produced gas
at the elevated temperature, for an average verifi-
cation rate of 92%.
CONCLUSIONS
The results indicate that the proposed two-
layer agar membrane filter procedure allows for
103
-------
TABLES. COMPARISON OF THE 2-LAYER
AGARVS DIRECT M-FC
PROCEDURES FOR FECAL
COLIFORM DENSITIES FROM
RESERVOIR WATER
Ratio
M-FC
Source Count
2-LayerAgar 2-LayerAgar
Count Direct M-FC
Water Supply
Reservoir
(Dry Period)
Water Supply
Reservoir
(Storm Water
Runoff)
11
17
44
10
21
94
6
8
36
91
32
86
100
140
80
95
53
130
10
30
44
13
31
89
10
10
43
96
44
95
106
170
100
125
55
160
0.9
1.8
1.0
1.3
1.5
0.9
1.7
1.3
1.2
Avg. 1.3
1.1
1.4
1.1
1.1
1.2
1.3
1.3
1.1
1.2
Avg. 1.2
TABLE 6. VERIFICATION OF BLUE COLONIES
FROM TWO-LAYER CULTURES
Colonies as
Number of Verified
Source Colonies Fecal Percent
Subcultured Coliforms Verification
Raw Sewage
Chlorinated
Effluent
River Water
Marine Water
Reservoir
538
70
145
80
180
477
69
132
79
173
88.7
98.6
91.0
98.8
96.1
repair and subsequent reproduction of those fecal
coliform which have been debilitated by exposure
to chlorine, industrial waste or marine waters. The
decision to use the slightly more involved two-lay-
ered medium procedure in preference to the direct
M-FC method should be based on a demonstration
of increased verified recovery of fecal coliforms
from samples routinely examined. With a major
interest in the fecal coliform test being related to
the bacterial quality assessment of effluents, the
proposed technique should be considered as an
alternative Standard Methods fecal coliform
membrane filter test, specifically intended for
those waters known to have significant levels of
attenuated fecal coliforms.
ACKNOWLEDGEMENTS
The authors wish to thank Mr. David Lentine,
Mrs. Barbara Green, Dr. H.D. Nash, Mr. D.F.
Spino and Mrs. M. Rutland for their technical
assistance.
REFERENCES
1. Lin, S. Evaluation of Coliform Tests for
Chlorinated Secondary Effluents, Jour. Water
Poll. Contr. Fed. 45, 498, 1973.
2. Green, R.A., R.H. Bordner, and P.V. Scar-
pino. Applicability of the Membrane Filter
and Most Probable Number Coliform Pro-
cedures to Chlorinated Wastewaters. Abs.
Ann. Meeting, Amer. Soc. Microbiol. p. 34
1974.
3. McKee, J.E., R.T. McLaughlin, and P. Les-
gourgues. Application of Molecular Filter
Techniques to the Bacterial Assay of Sewage.
III. Effects of Physical and Chemical Disin-
fection. Sew. and Ind. Wastes. 30, 245, 1958.
4. McCarthy, J.A., J.E. Delaney, and R.J.
Grasso. Measuring Coliforms in Water, Water
and Sew. Works 108, 238, 1961.
5. Taylor, E.W., N.P. Burman, and C.W. Oliver.
Membrane Filtration Technique Applied to
the Routine Bacteriological Examination of
Water. Jour. Inst. Waters Engrs. 9, 248
1955.
6. Burman, N.P., E.W. Oliver, and J.K. Stevens.
Membrane Filtration Techniques for the Iso-
lation from Water, of Coli-aerogenes,
Escherichia coli, Faecal Streptococci, Clos-
tridium perfringens, Actinomycetes and
Microfungi. 127-135 in Isolation Methods for
Microbiologists, Technical Series No. 3 ed.
Shapton, D.A. and G.W. Gould, Academic
Press, p. 178, 1969.
104
-------
7. Clark, H.F., E.E. Geldreich, H.L. Jeter, and
P.W. Kabler. The Membrane Filter in Sani-
tary Bacteriology. Pub. Health Repts. 66,
951, 1951.
8. Goetz, A. Application of Molecular Filter
Techniques to the Bacterial Assay of Sewage.
I. Proposed Technique. Sew. and Ind. Wastes
25,1136, 1953.
9. Stevens, A.P., R.J. Grasso, and J.E. Delaney.
Measurement of Fecal Coliform in Estua-
rine Water. Presented at the 7th National
Shellfish Sanitation Workshop. New Orleans,
La., June, 1974.
10. American Public Health Association. Standard
Methods for the Examination of Water and
Wastewater. 13th ed. American Public Health
Association, Inc., New York, 1971.
11. Shipe, E.L., and A. Fields. Chelation as a
Method for Maintaining the Coliform Index
in Water Samples. Pub. Health Rpts. 71,
974, 1956.
12. Coles, H.G. Ethylenediamine Tetra-acetic
Acid and Sodium Thiosulphate as Protective
Agents for Coliform Organisms in Water
Samples Stored for One Day at Atmospheric
Temperature. Proc. Soc. Water Treat. Exam.
13,350, 1964.
13. Hoather, R.C. The Effects of Thiosulphate
and of Phosphate on the Bactericidal Action
of Copper and Zinc in Samples of Water.
Appl Bact. 20, 180, 1957.
QUESTIONS AND ANSWER SESSION
Lin. Are there any shortcomings when
you combine two media in one?
Geldreich: Yes. By putting a top layer as an en-
richment medium in that position,
within two hours if diffuses into the
M-FC. They diffuse together and be-
come the same medium. At the begin-
ning, the membrane is separated from
the more specific medium, the M-FC
medium, by this layer of an enriched
material. We used the procedure
many years ago when we were trying
to work out an overlay concept for
pour plates. We were at one time
hoping that we could come up with a
coliform pour plate approach which
would be cheap and wouldn't use
membranes and other materials. That
is the concept. Now you cannot lay
the top layer on more than an hour
in advance, because it will diffuse into
the other medium and you will have
the fused media and you won't have
the advantage of an automatic transfer
concept.
Dufour: I would like to know what percentage
of debilitated organisms you are
recovering. In other words, what were
you using as the best estimate, or why
didn't you use the MPN for instance,
as the best estimate?
Geldreich: We are using MPN. The material is so
preliminary in nature that I didn't
have time to add the MPN results.
That is being done for the ASM meet-
ing.
Dufour: What bothers me is the fact that you
may be getting seven times more re-
covery but if your regular M-FC count
is down by two logs it really doesn't
make much difference.
Geldreich: Well, we will be able to give you more
information related to a base of an
MPN in May in the ASM.
McFeters: Is your verification data based strictly
upon bacteria that were isolated that
appeared as fecal coliforms from the
medium?
Geldreich: I would like to have done what some
of the rest of you have done but I
didn't have time. That is to take some
of these organisms out of the environ-
ment, which you know you are having
trouble with, or pure cultures which
you know don't work too well,
put them through the system and then
look at it. We didn't have this time as
of now.
McFeters: I just wondered about the background
of non-fecal coliforms.
Geldreich: We had very little background on all
samples except those which were in
the reservoir. The reservoir samples
occasionally gave us an increased
background of other bacteria which
were white colonies. That is one of
the difficulties when you begin play-
ing around with changing temperature
105
-------
and media enrichment. You are going
to increase the risk of getting some
bacteria that you don't want as back-
ground.
Ginsberg: The instructions for performing the
M-FC test state, rather adequately,
that the plates must be incubated
within 10 to 20 minutes after filtra-
tion. Now, you are suggesting that
you incubate at the lower temperature
for two hours. How do you explain
this contradiction?
Geldreich: I explain this for the very reason
that we are trying to get an enrich-
ment started at 35 C, not at room
temperature. If you lower the temper-
ature further to room temperature, we
are going to have more and more
problems with background organisms.
I would like to avoid completely, any
use of a lower temperature, but we
need a temperature acclimation, and I
think what you are going to see may
be in the next paper. I've talked to
Jack Delaney and Grasso and they
have come to the conclusion, in their
preliminary work, that one of the
problem is a shock of these organisms
when coming out of a cold environ-
ment and suddenly put at a jarring
44.5 C temperature. Maybe they
need a period of acclimation.
Brezenski: I am a little disturbed. If I read you
correctly, you said this is a proposed
method to substitute the routine
membrane filter procedure for chlor-
inated effluents. In Figure 2, you
showed four chlorinated effluent
samples and I think that three out of
the four were below the valid statisti-
cal range. I'm happy to see something
like this but I am afraid that I don't
see enough data to make a proposal
at this time, based on this amount of
data. Are the sewage effluents chlori-
nated?
Geldreich: As I told you when I was reading the
paper, we recognize this as prelimi-
nary data and I myself would not
begin to vote for it to be put in the
Standards Methods or EPA Methods
until I see not only more data from
our laboratories but from other
laboratories that have checked it out
in a field test. This is only the begin-
ning, and if these preliminary results
still prove to be promising as others
check it out, then I would entertain
that idea; but if it doesn't, let's forget
it.
Brezenski: During these sessions, we have been
using such terms as, attenuated cells,
injured cells, damaged cells and a few
others that I don't remember at this
point. I don't know if anybody has
done anything to describe what these
terms really mean. I get confused
between what is injured and what is
dead. Would somebody define these
terms for me so that I can get clear in
my mind what we are talking about.
So far, I have seen no data to show
any physiological problems with the
cells. There is no enzyme work. No
one has shown me a normal physio-
logical reaction taking place within
the cell. I would like to see what has
happened to the cell as a result of
the aquatic environment. If it was in
salt water, is some enzymatic system
blocked and therefore when it comes
into contact with a certain type of
medium does it need a specific sub-
strate as a booster? I just fail to see
the damage assessment because I
haven't seen the proof.
Geldreich: Well, I agree with you. We are assum-
ing many things but we have never
proven any of it.
Hufham: We have seen the damage; it results in
different types of effects on the cell.
You can get lack of separations with
chromosomes duplication, in which
case you get filaments up to 500
microns long. Unfortunately, you
cannot see these as colonies because
the width of the cell does not change.
This is a stress factor. There have been
several studies on this in which they
are trying to form filaments in cells
and they use heat shock to do it. This
is one of the ways of getting cultures
synchronized by using heat shock and
then bringing them back to the appro-
priate incubation temperature.
106
-------
Brezenski: You can see these filamentous forms Bordner:
which Zobel also showed with great
hydrostatic pressures. But, as it re-
lates to these studies in terms of your
recovery data, and relating back to
the reasons why the recoveries were
low, we just say they were stressed
cells. Zobel showed that when cells
are subjected to a certain depth at a
certain hydrostatic pressure, the cells
formed filamentous branches. We have
shown a difference in count assumed
on the differences of a stressed cell.
But nobody has shown the differences
in the stressed cell.
Geldreich: There are various forms of doing this.
Zobel used pressure, various people
used temperature or antibiotics and
we used oxygen. By increasing the Geldreich:
percentage of oxygen in the culture
you can get the same phenomenon. If
you return those to the normal condi- Bordner:
tions, the cell immediately starts to
divide and you get a colony formed.
But we have not done it on solid agar. Geldreich:
We can see it in the liquid media.
Theoretically, if these injured cells
are not kept too long and you return
the plate to 35 C, you should get a
colony formed. I think in the pre- Bordner:
enrichment medium you are getting
the micro-colony. Some of these are
capable of growing at higher tempera-
tures.
Williams: Getting back to basics, one of the
reasons we chlorinate is to attenuate
the cells. We like to kill them and I
guess my point is that evidently we
are making an effort to recover more
and more of these attenuated cells
and at what point do these attenuated
cells no longer have a sanitary signifi-
cance.
Geldreich: Well, there is some evidence. I think if
you remember this morning, Ted, one
of the speakers commented about
Salmonella cultures which had been Geldreich:
severely attenuated through a freeze-
drying condition and they were
proven to be very pathogenic. What
we really want is a kill.
Ed, you know that we have been
investigating chlorinated effluents for
sometime. Originally, in some work
that a graduate student did with us,
we were shocked at the very low
levels that were recovered by the MF
as compared to the MPN. Most of the
results did not fall within, or approach
the lower range of the very broad 95%
confidence level of the MPN. The MF
was that much lower than the confi-
dence levels of the MPN. Also, we
have been looking at the M-FC agar
for two or more hours at 35 C before
placing that same membrane on the
same agar at 44.5 C. It looked promis-
ing since we knew the interest in the
overlay technique.
Are you talking about an overlay or a
two layer?
A single layer of M-FC is what we
were reviewing.
These membranes are sitting on top
of the lactose agar which is overlayer-
ed on M-FC agar. I thought you
meant agar over the membrane.
I am talking about the comparison of
two-step M-FC and the overlay of the
lactose agar over the M-FC agar. In
the twenty-two samples of chlorinated
effluents and twenty with stream
samples, we see very little difference.
The recoveries were comparable. I
would like to call it temperature
acclimation with a ratio of one to
one in both cases. There were three
samples of chlorinated effluents where
there was a background of tiny pin-
point blue colonies. We found them in
both the lactose overlay and the
non-overlay. It seemed to me that
there is some possibility of avoiding
the overlay. I am not sure at this point
but I think that it would be worth
getting more data.
I am intrigued as much as many of
you here with this concept of a Milli-
pore membrane that we have heard
about today and wonder if that will
help too. But I think one of the im-
107
-------
portant things that we must try and
get across here is that these are
methods that we are only discussing
today. They are preliminary and we
must have many laboratories field test
these things before they go into any
published procedures. I don't want
any of you to have the illusion that
because these three labs did it this is
the final work. This is just the begin-
ning and I hope that we evaluate these
ideas in many different geographical
areas. Many times in the past we
personally developed some media that
worked great in the Ohio area, but out
in the West or in New England it did
not work as well, because bacterial
flora are different. So whenever we
evaluate these procedures it is of great
importance that we have as many
people involved in it and as many
different kinds of samples. We should
make our minds up from all of these
numbers whether that method goes
into the Standard Methods or an EPA
manual. Thank you very much.
108
-------
MEASUREMENT OF FECAL COLIFORMS IN ESTUARINE WATER
by
Alanson P. Stevens, Rosario J. Grasso and John E. Delaney
ABSTRACT
Recoveries of fecal coliform bacteria from
estuarine waters were compared using the standard
MF procedure and the standard MPN EC Count.
The experimental two-step procedure recovered
85% of the MPN count as compared to 24% by the
standard MF procedure. Ninety-three percent of
the colonies in the experimental procedure did
verify.
In later work, significant variation in recover-
ies using different batches and brands of membrane
filters suggested possible influences on the earlier
work. Further tests are being conducted.
INTRODUCTION
The fecal coliform concentration in sea water
can be measured, at the present time, by the E C
MPN procedure, and by a membrane filter method
using M-FC broth as the culture medium. However,
the inherent disadvantages associated with these
procedures have prevented either one from achiev-
ing absolute acceptance by marine microbiologists.
The multitube method requires a period of three
days to complete, and yields a concentration esti-
mate that embodies neither precision nor accuracy.
In addition, the MPN procedure is time consuming,
requires considerable incubator space, and large
amounts of sterilized media and glassware. These
requirements restrict the number of samples that
can be processed. The direct-count, membrane
filter (MF) procedure, on the other hand, although
more rapid, possessing greater built-in precision
and requiring a minimum of preparatory labor and
laboratory glassware, recovers only a fraction of
the fecal coliform population in a sea water
sample. A study in which over 200 sea water
samples were analyzed by both the E C MPN and
standard MF procedures indicated that, on the
average, only 10% of the fecal coliforms enumer-
ated by the multitube technique were measured
by the current MF procedure.
Our studies have indicated that the low
recovery exhibited by the membrane filter pro-
cedure is intrinsically associated with the im-
mediate exposure of the fecal coliforms to the
elevated selective temperature 44.5 C. The stress
imposed by this temperature on the individual
fecal coliform cells, during the initial 15 minutes of
exposure, has been shown to be lethal to the
majority of the fecal strains filtered from sea
water. Since the poor MF recovery of fecal coli-
form in sea water is unquestionably related to the
immediate exposure of these organisms to the
elevated temperature, our initial attempts at
developing a more accurate MF test procedure
focused on methods of acclimatizing these species
before exposing them to the selective temperature.
Enrichment on Minimal Medium at 25 C Methods
It was recognized at the outset, that the med-
ium and the temperature employed during the
acclimatization period would have to control the
number of bacterial generations, so that subse-
quent exposure to the selective stage would not
foster colony overgrowth. A number of media
formulations were tested for their ability to pro-
vide optimum resuscitation conditions, while
simultaneously controlling growth for the fecal
coliforms in sea water samples. Initial studies
employed a one per cent (1%) solution of M-FC
broth as the minimal medium and full-strength
M-FC agar as the growth-indicator medium. This
media combination in the two-step, two-day pro-
cedure markedly increased the fecal coliform
recoveries from sea water, when compared to the
results from the standard method one-day, direct
count procedure. However, extensive testing of
this media combination failed to yield the desired
recovery efficiency to warrant its acceptance as
a standard test system.
Recent work has focused on formulating and
testing a minimal medium composed mainly of
109
-------
simple carbohydrates. It was theorized that these
substrates would be easily attacked and metabol-
ized by the fecal coliforms and would foster
bacterial cell repair and resuscitation for those that
had been attenuated by exposure to sea water.
This minimal medium, coupled with full-strength
M-FC broth as the selective-growth medium, has
yielded very satisfactory recovery levels for fecal
coliform in sea water.
The formulation of the minimal medium,
labeled L.E.S. Minimal Holding Agar, is as
follows:
L.E.S. MINIMAL HOLDING AGAR
Tryptose
Dextrose
Lactose
Oxgall
Sodium Chloride
Agar
0.5 grams/I
0.5 grams/I
0.5 grams/I
0.25 grams/I
0.4 grams/I
15.0 grams/I
L.E.S. Two-step, Two-day Procedure
The minimal holding agar is allowed to warm
to room temperature before "prepared" mem-
brane filters are placed on the agar surface. All
standard precautions are exercised at this stage to
guard against improper placement of the filter on
the agar surface. The dishes are tightly sealed and
placed into a watertight, plastic bag, making sure
that all plates are facing upright. The sealed bags
are inverted and incubated in a water bath or air-
jacketed incubator regulated at 25 C for 18 ± 2
hours. Subsequent to the enrichment period, the
membranes are transferred to absorbent pads in
tight sealing petri dishes that have been saturated
with Standard Methods M-FC broth. Approxi-
mately 1.7 to 1.8 ml of broth are required to
saturate an absorbent pad. Excess media should be
discarded to prevent excessive and "running"
bacterial growth. The dishes are tightly sealed,
placed into water-tight plastic bags and incubated
in an inverted position in a water bath at 44.5 ±
0.2 C for 24 hours.
Counting of Fecal Coliforms
The fecal coliform colonies are counted with
the aid of a stereomicroscope and a light source
above, that is approximately perpendicular to the
plane of the membrane being counted. The fecal
coliform colonies are recognized by their blue
coloration and the crystallized deposits on the
surface. Both of these identifying characteristics
must be employed in counting the fecal coliform
colonies. The "quartz" surface appearance has
been found to be a complimentary and distinc-
tive characteristic of fecal coliform colonies.
Certain non-fecal coliforms capable of growing
under the test conditions will form blue-colored
colonies but will lack the distinguishing crystalline
characteristic on their surface.
RESULTS
Fecal Coliform Recovery MF vs MPN Procedures
Twenty-five samples of sea water containing
varying concentrations of fecal coliform and salt
content were examined to determine the recovery
efficiency of the L.E.S. two-step, two-day pro-
cedure. From each sample, five 10 tube E C MPN
analyses were made and ten membrane filters
(Millipore HAWG 047SO) were prepared and pro-
cessed by both the Standard Methods MF pro-
cedure and the experimental MF technique. The
results, obtained from each of the MF procedures,
were logarithmically averaged and proportioned
against the average logarithmic MPN result of each
sample. Table I presents the percent of recoveries
obtained by the two MF procedures for each
sample analyzed, and employ the logarithmic
average E C MPN result as the true estimate. These
data indicate that the two-step, two-day procedure
yields a significant recovery increase over that
attainable by the Standard Methods MF proce-
dure. The Standard Methods MF procedure pro-
duced an average recovery of only 24% while the
L.E.S. two-day technique yielded an 86% recovery
of the fecal coliform concentration in the sea
waters examined. It is worthwhile noting that the
lowest recovery obtained by the experimental
method (64%) is higher than the highest recovery
obtained by the Standard Methods MF procedure
(46%). Table I also presents data on the selectivity
incorporated into the two-step, two-day MF pro-
cedure. Over 93% of the colonies exhibiting the
two identifying characteristics of fecal coliform
verified as bona fide fecal coliform strains. Overall,
the percent verification in the 22 samples ranged
from 87 to 100.
Figure I presents a graphical analysis of the
recovery data obtained by the E C MPN procedure
110
-------
TABLE 1. FECAL COLIFORM RECOVERIES
BY MEMBRANE FILTER
PROCEDURE
Percent Recoveries (*)
2.0
1.0
Standard Methods
Procedure (1)
17
39
34
16
11
12
6
9
34
5
46
40
27
22
31
37
22
30
16
15
21
16
32
29
24
L.E.S.
Procedure (1)
126
114
96
64
89
97
75
76
77
80
71
71
89
72
79
92
97
78
87
66
82
84
106
117
90
/o
Verification of
L.E.S. Colonies
92
95
88
93
96
95
100
90
95
91
87
100
95
91
92
92
96
94
96
91
87
100
90
100
(*) E C MPN ten tube procedure used as Control.
(1) % Recoveries based on 5 MPNs and 10 MFs
per sample.
and the two membrane filter methods. The ratios
(Table I) were arrayed in order of magnitude and
every other value plotted against the appropriate
percentile on log-probability paper. The resulting
lines of best fit clearly indicate that a significant
difference exists between the recovery attainable
by the Standard Methods MF procedure and by the
experimental MF method. Based on the 50 per-
centile (median) ratios, the average recovery of the
Standard Methods MF procedure was 22%, while
the experimental MF procedure recovered an aver-
age of 84% of the fecal coliform concentration in
the sea water samples. This analysis indicates that
the experimental method, on the average, recovers
2.9 times more fecal coliforms from sea water than
0.1
CURRENT L.E.S.
PROCEDURE
STANDARD
METHODS
PROCEDURE
1 10 50 80 95 99
PERCENT EQUAL TO OR LESS THAN
Figure 1.
the Standard Methods MF procedure. A further
analysis of the plots in Figure I clearly demon-
strates the improvement in fecal coliform recovery
as a result of the two-step, two-day procedure. The
plots indicate that the Standard Methods MF pro-
cedure recovered less than 19% of the fecal coli-
form in 40% of the samples and less than 40% in
90% of the samples, while the experimental pro-
cedure recovered at least 60% in all samples and
less than 72% in only 20% of the samples analyzed.
SUMMARY
The direct count procedure for fecal coliform
in estuarine waters, developed under this research
project 1, appears to possess the necessary accur-
acy, precision and selectivity to warrant its ac-
ceptance by marine microbiologists as a more than
adequate replacement for the E C MPN pro-
cedure. The recovery efficiencies of the L.E.S.
experimental MF procedure were calculated from
MPN values that had not been corrected for their
inherent positive bias. If this bias factor had been
used in the computation of the recovery ratios, the
median recovery percentage (84%) of the experi-
mental procedure would consequently have been
adjusted to a more favorable level, namely 92%.
This truer recovery capability of the L.E.S. two-
step, two-day MF procedure, when coupled with
the MF advantages of precision, time of analysis,
less preparatory labor and incubator space, are
strong arguments for accepting and employing the
experimental MF technique as a standard proce-
dure for measuring fecal coliform concentrations
in sea water.
This research project was supported, in part,
by a Research Grant from the Massachusetts Water
Resources Commission. We acknowledge this
support and their active interest in this project.
111
-------
Recent Investigations
Gelman
Millipore
Recent studies have indicated that different
brands of membrane filters have an effect on the
recovery levels of fecal coliform organisms (1, 2).
Since all the data generated in developing the
L.E.S. two-step, two-day procedure for fecal coli-
forms was compiled using the Millipore Type HA
filter, we deemed it necessary to compare the fecal
coliform recoveries attainable by using two brands
of membrane filters (Gelman GN-6, lot No. 80706
and Millipore HAWG 047SO, lot No. 934487) in
our newly developed method.
This work was not motivated by the desire to
prove one membrane filter brand superior over
another, but rather by a strong curiosity as to why
different reputable brands of filters should produce
statistically significant recoveries of fecal coliform.
The results of our comparative studies showed
that Millipore filters consistently gave lower re-
coveries than the Gelman filters in the Standard
Methods MF procedure for fecal coliform. The
Gelman filters, conversely, gave poor recovery
compared to the Millipore filters when used in the
two-step, two-day L.E.S. procedure. In addition,
the Gelman filters produced non-typical fecal coli-
form colonies subsequent to incubation. These
colonies were small with irregular shapes, and
varied in color from light brown to the typical
fecal coliform blue. It should also be noted, that
even with the higher recoveries of fecal coliform
by the Gelman filters in the Standard Methods
fecal coliform procedure, only 30% of the actual
concentrations of fecal coliform in the samples,
as judged by the E C MPN method,were recovered.
From these results, it seemed plausible, that
any substances present in the membrane filters
capable of affecting development of fecal coliform
colonies during the incubation period would be
soluble in the media, producing pH and other
physico-chemical changes. In order to determine
the water soluble chemicals in these two filter
brands, six filters of each were immersed in 100 ml
of distilled water for 18 hours at room tempera-
ture, the filters were removed and the following
results were obtained upon analysis of the water.
It was thus possible to project that the dif-
ferences in quantities of soluble components
GN-6
Lot #80706
pH 3.23
Ammonia-Nitrogen 13.0 mg/l
Ortho-Phosphate (as P) 43 mg/l
Total dissolved solids 185 mg/l
HAWG
Lot #93448-7
6.20
0.24 mg/l
0.06 mg/l
6 mg/l
between brands of filters could somehow be
responsible for the differences that were obtained
in fecal coliform recoveries. Neither the Standard
Methods M-FC broth, nor the two media used in
the L.E.S. two-step, two-day procedure have an
inorganic buffer system in their formulation.
Low recoveries exhibited by the Gelman
filters with the new L.E.S. procedure could be
attributed to the increased hydrogen-ion concen-
tration that these filters imposed on the holding-
minimal agar, while the higher recoveries by
Gelman filters on standard M-FC broth at 44.5 C
could be attributed to the leaching of beneficial
inorganic nitrogen and phosphate into the medium,
as well as the suppressing of the pH change by
buffering action of the organics in the formulation.
Based on this hypotheses, we added a monopotas-
sium phosphate and dipotassium phosphate buffer
system to the L.E.S. minimal holding agar and also
to the M-FC broth. Results obtained from experi-
ments employing this buffered media, produced
encouraging results with both brands of filters.
At present, additional tests are being conducted on
estuarine waters using the new buffered media with
the two-step, two-day L.E.S. procedure in order to
substantiate that either brand of filters can be
employed in this technique.
REFERENCES
1. Presswood, W.B., and L.R. Brown. Compari-
son of Gelman and Millipore membrane
filters for enumerating fecal coliform bacteria.
Appl. Microbiol, 26:336, 1973.
2. Hufham, J.H., Evaluating the membrane
fecal coliform test by using Escherichia
coli as the indicator organism. Appl. Micro-
biol, 27:771, 1974.
112
-------
EVALUATION OF METHODS FOR DETECTING COLIFORMS
AND
FECAL STREPTOCOCCI IN CHLORINATED SECONDARY
SEWAGE EFFLUENTS
by
Shundar Lin
Water Quality Section
Illinois State Water Survey
P.O. Box 717
Peoria, Illinois61601
ABSTRACT
Total coliforms (TC), fecal coliforms (FC)
and fecal streptococci (FS) recoveries in chlori-
nated secondary sewage effluents were investigated
using the membrane filter (MF) and multiple-tube
(Most probable number, MPN) methods. The LES
two-step MF method was found to be comparable
to the MPN procedure for determining TC. The TC
detection was 1.5 times greater when using the
LES two-step technique than that obtained by the
M-Endo one-step MF procedure. Fecal coliform
recovery by the M-FC MF procedure was lower
than the recovery obtained using the MPN method.
The use of each of azide-detrose broth, brain-
heart infusion broth, and peptone yeast-extract
Casitone with the M-Enterococcus agar MF2
(2-day incubation) procedure was not satisfactory
for the recovery of FS. The M-Enterococcus agar
procedure with bile broth enrichment (MF2> or
prolonged incubation for 3 days (MFs) signifi-
cantly increased FS recovery and were comparable
to the MPN method. The results cited should be
useful in assessing the efficiency of disinfection
practices for waste treatment plants employing
effluent chlorination.
INTRODUCTION
The year-round disinfection of wastewater
treatment plant effluents has become mandatory
in Illinois and in several other states. The most
common method of disinfection at treatment
plants is chlorination. Its effectiveness has gener-
ally been measured by residual chlorine. The
Illinois Pollution Control Board (1) requires a
limitation on fecal coliforms (FC) densities inde-
pendent of residual chlorine thus requiring deter-
minations for FC densities in chlorinated effluents.
The Board's rules stipulate that fecal coliforms
densities in a waste effluent shall not exceed
400/100 ml.
Total coliforms (TC) have been used for mea-
suring the disinfective efficiencies of water and
wastewater treatment units. The TC index is still
valid and reliable for the water industry. In Euro-
pean countries fecal streptococci (FS) are com-
monly looked for in the sanitary analysis of water
supplies (2). In the United States, they are used
currently in conjunction with FC for determining
the sanitary quality of water. Although FS deter-
minations are not required by most regulatory
agencies the usefulness of the procedure should
not be overlooked.
The requirement for bacteria enumeration in
treated effluents necessitates the development of
adequate and economical procedures for determin-
ing bacteria densities in chlorinated effluents. The
series of investigations described in this report were
undertaken with these objections in mind.
Indicator Organisms. The purpose of the rou-
tine bacteriological examination of water samples
is usually to estimate the hazard due to fecal pollu-
tion and the probability of the presence of patho-
genic organisms. The isolation of pathogens from
water and sewage is expensive and laborious. It
is not a routine practice. Normally occurring
bacteria in the intestines of warm-blooded animals
have been used as indicators of fecal pollution.
Total coliforms, fecal coliforms, and fecal strep-
113
-------
tococci have all been used as pollution indicators
at various times (3, 4). Other bacterial indicators
have been proposed. These include Clostridium,
Pseudomonas and Aerobacter. Presently their
value has been considered questionable or irrele-
vant (5).
Correlations between coliforms and patho-
genic bacteria have been cited frequently i.e.,
coliforms vs Salmonella (6, 7, 8, 9). Less known is
the relationship, if any exist, between coliforms
and viruses. A coliform index is not a reliable
index for viruses (10, 11). In spite of the lack of
documented relationships there is little evidence
that enteroviral or other microbial diseases are
transmitted frequently by the drinking water
route in the absence of coliforms (5).
Until more definitive studies are completed
on the relationship of pathogens and indicator
organisms, the use of TC for water supplies and FC
and FS for sewage and stream quality, as indicators
or enteric pollution, are valid.
Bacteria Enumeration. The basic methods for
the assay of pollution indicators (TC, FC, and FS)
in waters are outlined in Standard Methods (4).
These include the multiple-tube, or most probable
number (MPN) technique and the membrane
filter (MF) procedure. Standard Methods (4),
however, states that "Experience indicates that the
MF procedure is applicable to the examination
of saline waters but not chlorinated wastewaters".
Because the MF technique is not comparable to the
MPN procedure and is less time consuming, it
seems unfortunate that the MF technique cannot
be used as a control procedure by the waste plant
operator who uses chlorination.
McKee et al. (12) reported on the lack of
correlation between MPN and MF techniques
while assaying chlorinated settled wastewater for
total coliforms. Because monochloramine is the
predominant bactericidal agent in chlorinated
wastes, they advanced the hypotheses that partial
reversibility is responsible for the discrepancy
between MPN and MF results; that is, the MF
technique produces considerably fewer colonies
than the numbers that develop by the MPN method.
Presumably, when inactivated cells are deposited
on a membrane with limited nutrient availability,
the cells cannot rid themselves of monochloramine
and therefore cannot grow. However, when inacti-
vated cells are put in an aqueous medium rich in
organic matter, such as lactose broth, the mono-
chloramine may diffuse outwardly from the cells,
permitting them to recover, grow, and produce gas.
In the McKee et al. (12) investigations, the
culture media used for the MF technique was the
same as that previously described (13). Dehydrated
scheduled nutrient (DSN) pads were used. They
contained two elements with an upper leaf impreg-
nated with an Endo-type inhibitory nutrient. The
results obtained using DSN pads with the MF tech-
nique were comparable to those obtained from the
confirmed MPN procedures on raw settled waste-
water. McCarthy et al. (14), though working
initially with water, were not satisfied with the
one-step, M-Endo broth MF techniques. Their
work suggested that enrichment plus an agar
substrate (E&A) was superior to the one-step
technique based on a higher degree of coliform
recovery. Examinations of natural waters and
wastewater demonstrated that the E&A results
were comparable to standard MPN data. From
their work an agar-based medium (LES M-Endo
agar) was developed. Its use with the MF technique
is basically a two-step enrichment procedure.
The need has developed not only for deter-
mining total coliform but also for enumerating
fecal coliform densities. Geldreich et al. (15)
recommended the use of an M-FC medium at
incubation temperature of 44.5 ± 0.5 C as part of
the MF technique for the direct county of fecal
coliform. It has been reported (16, 17, 18) that the
determinations for fecal coliform rather than total
coliform are a more realistic measurement of the
public health significance of microbial discharges
in wastewater plants. Illinois requirements specify
maximum permissible limits for fecal coliform
concentrations in treated effluents. This will
require fecal coliform enumeration in chlorinated
effluents.
Several investigators (19, 20, 21) reported
that Escherichia coli injured during a physical or
chemical treatment failed to form colonies on
membrane filters incubated on M-FC medium, to
grow and produce gas in lactose broth or to grow
on selected media. Braswell and Hoadley (21) sug-
gested that standard methods for enumeration of
total and fecal coliforms in water and wastewater
should not be applied to chlorinated effluents.
Even for unchlorinated samples Hufham (22)
claimed that a large relative error in the results of
MF method was found to be dependent on the
brand of MF used, the medium, and the tempera-
ture of incubation. A study by Presswood and
114
-------
Brown (23) showed FC counts incubated on
Gelman filters at 44.5 C averaged 2.3 times greater
than those on Millipore filters. Hufham (22) sug-
gested that the MF method for FC recovery should
not be accepted.
Lattanzi and Mood (24) used the Winter and
Sandholzer method for the detection of entero-
cocci. Later Litsky et al. (25) suggested the use of
glucose azide broth as a presumptive medium and
ethyl violet azide (EVA) broth as a confirmatory
medium for enterococci detection with MPN
procedures.
Slanetz and Bartley (26) proposed the use of
M-Enterococcus agar for the isolation of FS by the
MF method. Kenner et al. (27) introduced the KF
streptococcus agar. Rose and Litsky (28) found
they could increase the recovery of FS from river
water by more than 2-fold when using peptone
yeast-extract Casitone (PYC) compared to M-
Enterococcus agar. Recently Pavlova et al. (29)
suggested that fluorescent antibody techniques
may be useful, for FS detection, in determining the
presence and source of fecal pollution in water.
Objectives
During the course of the study two separate
investigations were performed. One dealt princi-
pally with TC and FC; the other with FS. The pur-
poses of the study were:
1. To determine whether or not the MF
technique for TC, FC and FS detections
in chlorinated secondary effluents is
comparable to the MPN method.
2. To determine whether or not the LES
two-step enrichment MF technique for
TC detection, in chlorinated secondary
effluents, is comparable to recommend-
ed MPN methodology.
3. To develop improvements in the MF
method for the detection of FS in chlor-
inated secondary sewage effluents.
MATERIALS AND METHODS
Grab samples of final settling tank effluents
from three wastewater treatment plants serving the
cities of Peoria, Morton, and Washington in Illinois
were used in the study. A minimum of five effluent
samples from each plant were examined. The
Peoria plant employs the high-rate activated sludge
process treating a combination of domestic and
industrial wastewaters. Contact stabilization com-
parable to the standard-rate activated sludge pro-
cess is used at Morton. This plant treats principally
domestic wastewater. Washington is served by a
standard-rate trickling filter plant, treating domes-
tic wastewater also.
One liter portions of each effluent were dosed
with calcium hypochlorite (HTH, 70% available
chlorine) up through 6 mg/1 of chlorine. The
samples were stirred gently but intermittently, and
after varying periods of contact (up to 30 min.)
they were dechlorinated with an excess of sodium
thiosulfate. The dechlorinated samples were
assayed immediately for bacterial densities using
parallel MPN and MF methods.
The MPN procedures were performed by
inoculating a series of four decimal dilutions per
sample, using five tubes for each dilution. Lauryl
tryptose (LT) broth was used for the presump-
tive tests in TC and FC determinations. The TC
test was confirmed using brilliant green bile (BGB)
medium, and was completed with gram-strain. For
FC confirmation EC medium at 44.5 ± 0.5 C
(water bath) was used. In the MPN procedure for
FS tests, axide-dextrose (AD) broth was used for
the presumptive test; while ethyl violet azide
broth was used for confirmation.
In the MF procedures for TC, FC, FS, three
duplications for each sample were filtered through
an 0.45 //m membrane filter (Millipore) for each
bacterial test. For TC tests, the two-step enrich-
ment for LES M-Endo agar (14) was followed.
Occasionally, parallel tests with the standard one-
step M-Endo procedure were performed. For TC
verification purposes, representative colonies (3 to
6 sheen colonies per filter) were subcultured
through LT broth into BGB broth (30). The pro-
duction of gas on BGB broth was deemed verifi-
cation.
When using MF procedures for FC detections,
the recommendations of Geldreich et al. (16)
were followed. Several colonies (3 to 5 blue
colonies per filter) grown on the M-FC medium
were verified by inoculating in phenol red lactose
broth for a 24 to 48 hour period at 35 C and
noting gas production. All positive tubes were con-
firmed at 44.5 C (water bath) in EC broth.
In the determination of FS densities by the
MF technique, the standard one-step M-Entero-
coccus agar (4) was used. According to Sies (31)
115
-------
M-Enterococcus agar is superior to the KF strep-
tococcus agar for sewage effluents because
some of the non-streptococci species in sewage
samples grow red and pink colonies on KF Strep-
tococcus agar. The FS counts on the membrane
filters were generally made after 2, 3, 4, and 7
days incubation. Parallel tests with the two-step
enrichment were also performed. The enrichment
media used include AD broth, brain-heart infusion
(BHI) broth, bile broth medium (prepared by
adding 40 ml sterile 10% oxgall solution to 60 ml
sterile BHI broth), and PYC broth. The period of
the pre-enrichment was 2-3 hours. For the purpose
of FS verification, red and pink colonies (3 to 6
colonies per filter) were fished at random from the
membrane filter and inoculated onto a brain-heart
infusion agar (BHIA) slant, followed by a catalase
test. If the catalase test was negative, then the
growth on the BHI A slant was sub-cultured into
both a BHI broth and into a bile broth medium
for confirmation.
With slight variation all bacteria assay pro-
cedures followed Standard Methods (4). Generally
all the media used were freshly prepared; none of
the media used was more than four days old.
RESULTS AND DISCUSSION
Total Coliforms
Multiple-tube versus membrane filter. Consis-
tent with Standard Methods (4) recommendations
that a comparison be made between MPN and MF
techniques before using the MF procedure, a series
of bacterial assays on unchlorinated samples from
a variety of sources was performed. This evaluation
included enumeration for total coliforms as well as
fecal coliforms. Table 1 summarizes the results.
TABLE 1. COMPARISON OF THE MPN AND THE MF COLIFORM DENSITIES OF
UNCHLORINATED WATERS FROM SEVERAL SOURCES
Source*
Spoon River
Spoon River
Spoon River
Spoon River
A. S. effluent
A. S. effluent
A. S. effluent
A. S. effluent
A. S. effluent
A. S. effluent
A. S. effluent
A. S. effluent
A. S. effluent
A. S. effluent
T. F. effluent
T. F. effluent
T. F. effluent
Tert. pond effluent
Total coliforms/100 ml
Completed MPN
LES-MF
2,400
3,300
460
1,300
92,000,000
13,000,000
7,900,000
5,400,000
790,000
350,000
240,000
3,500,000
2,700
3,100
1,200
2,600
80,000,000
12,000,000
7,400,000
6,700,000
1,300,000
670,000
360,000
4,000,000
Fecal coliforms/100 ml
MPN
MF
Illinois River
Illinois River
Illinois River
Illinois River
1,700
1,300
790
1,400
2,000
1,200
230
490
170
79
640
330
270
160
490
170
140
790
4,900,000
2,400,000
460,000
33,000
79,000
79,000
49,000
33,000
490,000
490,000
790,000
330
270
250
790
35,000,000 19,000,000
1,000,000
2,100,000
400,000
30,000
52,000
80,000
59,000
60,000
600,000
410,000
600,000
*A. S. = Activated sludge process
*T. F. = Trickling filter process
116
-------
Using the paired data t-test technique in testing
the hypothesis (H0) that the mean of the first
population is equal to the mean of the second; the
results (Tests 1 and 3 of Table 2) do not indicate
significant differences in TC and FC recoveries
determined by the MPN and MP methods. The
comparison for the purposes of this study, there-
fore, were considered acceptable.
M-Endo (one-step) versus LES M-Endo (two-
step). Samples of chlorinated effluents from an
activated sludge process were evaluated for TC
densities using M-Endo one-step and LES M-Endo
agar two-step MF procedures. McCarthy et al. (14)
performed a similar assessment on unchlorinated
water samples from rivers, lakes, and ponds leading
to the development of the LES agar-based medium.
The results obtained on chlorinated secondary
effluents were comparable to those observed by
McCarthy et al. (14). As shown in Figure 1, the
plotted data lie above the equality line, indicating
that total coliform recovery by the LES two-step
procedure was superior to the M-Endo one-step
method. A better development of sheen colonies
was also observed on the LES medium.
5 240
o
o
£ 200
160
o
Q
120
CO
CO
80
40
n=34
Y = 0.968
CHLORINATED ACTIVATED
SLUDGE EFFLUENTS
40 80 120 160 200 240
ONE-STEP M-ENDO MF COUNT
Figure 1. Comparison of Total Coliform Counts
made on M-Endo Broth and on LES
M-Endo Agar.
The experiences of McCarthy et al. (14) and
McKee et al. (12) were similar with regard to total
coliform recovery from unchlorinated wastewater
samples. The McCarthy group found no advantage
in using an enrichment phase when compared with
a one-step agar method on unchlorinated wastes
and polluted waters. They suggested that the
recovery efficiency for total coliform&was a func-
tion of the number of coliforms in the sample and,
therefore, in natural waters where smaller numbers
of coliforms are likely to exist, an enrichment
phase in the MF technique is required, whereas
with polluted waters and unchlorinated wastewater
the enrichment two-step procedure can be omitted
without significant effect on coliform recovery.
McKee and his colleagues, however, experienced
the lessening of coliform recovery on chlorinated
settled wastewater similar to that described for
water with a smaller number of coliforms. This
suggests that equivalent conditions are encountered
when temporarily inactivated colonies exist or a
smaller number of colonies are present. In both
cases, an enrichment phase would more than likely
be required to attain satisfactory coliform recovery
using the MF technique.
Although the number of colonies per filter
as depicted in Figure 1 exceeded the desirable
range of 20 to 80/filter, they were considered
satisfactory for comparison purposes. The results
correlated well (r = 0.968), and the relationship
between the two procedures can be expressed as
TC2 = 0.64+ 1.56TC.,
where TC-j and TC2 are, respectively, the total
coliform colonies determined by the one-step and
two-steo MF techniques. The total coliform re-
covery on chlorinated effluents by the LES two-
step procedure is about 1.5 times greater than that
attained by the M-Endo one-step method.
LES (two-step) versus multiple-tube. The
multiple-tube method is considered acceptable for
assaying the total coliform densities in chlorinated
wastewater effluents. A comparison of the total
coliform data resulting from the LES two-step
method, which was used in this study, with bacter-
ial densities obtained from parallel multiple-tube
observation was therefore pertinent. Using meth-
ods described by Thomas (32) the total coliform
data for all chlorinated secondary effluents ex-
amined are shown in Figures 2 and 3. Figure 2
represents observations of the LES two-step MF
technique, and Figure 3 represents observations of
the multiple-tube procedure. The figures reflect
simply the geometric distribution of the bacterial
densities for all effluents using two techniques.
Similar curves could have been presented for each
type of effluent.
117
-------
CO
O
I-
cc
LU
CO
00
o
Q
LU
CC
LU
LU
LU
00
LU
LU
DC
LU
LL
LU
O
O
CO
DC
O
CO
LU
LU
IE
I-
LL
O
CO
CO
LU
QC
CN
LU
_J
OQ
I-
CNI
II
a.
"o
IE
*
H-
T3
LO"
CD
CJ
i_ iL C
QJ QJ ».
-Q y, ^
|°-B
z*S 5
"8
« cS
00 +j
in
T3
O
+-*
0)
E
"D
2?
CO
a
E
o
O
CD
tt E
CP 3
I- C
CD
CO
oq
o
ro
o
_a>
CD
O
O
a
(D
CN
O
LL.
O5
CN «-
r- LO
O) CD
CO
in*-
00
LO
CO
cxD
cqr--
o ^
r^
CM"
cxj
o
CO
>
COCO
COCOCOCO
o
o
3
LLc/)
o
C
C\
o
c
CN
CLL.(nLLv)LL.
CM
CMC CNC CNc CO
a>
o
D
L_
o
cu
a>
o
0}
T3
0)
~ 'o
"S &
3 CD
3 «
118
-------
106
o
o
105
CO
uu
Q
EXPLANATION
HIGH RATE ACTIVATED
I- SLUDGE
^CONTACT
STABILIZATION
_°TRICKLING FILTER / Mg x dg,
n=71 f 350,000
P104
O
o
103
Mg-Og
2400
Mg= 29,000
.12.14
102
0.5 10 50 90 99.5
PERCENT PROBABILITY OF NOT EXCEEDING
o
o
106
105
o
0.
10*
O
o
103
102
EXPLANATION
HIGH-RATE ACTIVATED
"^SLUDGE
CONTACT
STABILIZATION
oTRICKLING FILTER
Mg X dg
370,000
Mg-28,000
d = 13.55
Mg-dg
2100
0.5 10 50 90 99.5
PERCENT PROBABILITY OF NOT EXCEEDING
Figure 2. Total Coliform Analysis on LES Two-Step Figure 3. Total Coliform most probable number
Membrane Filter Samples from Chlorinated
Effluents.
Ana|ysis
ch|orinated Eff|uents
More important for comparative purposes is
the summary included in Table 3. For the MF
technique, including all data, the geometric mean
was 29,000 total coliforms/100 ml; the geometric
standard deviation was 12.14; and the arithmetic
mean computed from geometric parameters (32)
was 650,000/100 ml. Similarly, the MPN data
reflected a geometric mean of 28,000 conforms/
100 ml, a geometric standard deviation of 13.55;
and an arithmetic mean of 700,000/100 ml. All of
the data, including that for each type of effluent
summarized in Table 2 suggest that the LES two-
step MF method for chlorinated effluents is com-
parable in coliform recovery efficiency to the
multiple-tube procedure.
Figure 4 is a graphical presentation for com-
parative purposes also. For the 71 examples ex-
amined, 32 of the MF results are higher and 34
of the MF results are lower than concurrent MPN
results. Five observations were found to be identi-
cal. The ratios of MF:MPN varied from 0.44 to
5.03 with a median of 1.00.
o
o
CO
LU
Q
;104
o
o
2102I2
EXPLANATION
HIGH-RATE ACTIVATED SLUDGE!
CONTACT STABILIZATION
TRICKLING FILTER
n = 71
102
10*
106
Figure 4. LES Two-Step Membrane Filter and
complete most probable number
results on Chlorinated Effluents.
119
-------
TABLE 3. COMPARATIVE RESULTS FOR TOTAL COLIFORM DATA
Chlorinated
Effluent
High rate
Act. si.
Contact
Stabilization
Trickling Filter
Total
No. of
Observations
21
24
26
71
Test*
MF
MPN
MF
MPN
MF
MPN
MF
MPN
Geometric
Mean, Per
100ml
40,000
36,000
5,100
4,600
110,000
120,000
29,000
28,000
Geometric
Standard
Deviation
18.00
20.43
5.16
5.20
6.31
7.15
12.14
13.55
Arithmetic
mean t, Per
100ml
2,600,000
2,900,000
190,000
150,000
6,000,000
7,000,000
650,000
700,000
*LES (two-step) MF and completed MPN
tM - C Mg ag 1.15 log ffg. C = 1.0 for MF, C = 0.851 for 5-tube MPN
As shown in test 2 of Table 2 the TC re-
covered by the MPN and LES M-Endo methods
are not significantly different. It is concluded that
the LES two-step MF technique was as good as the
multiple-tube method for assaying total coliforms
densities in chlorinated secondary effluents. From
the standpoint of time, convenience, freedom from
bias, and equipment needs, the LES two-step
technique would seem preferable to the multiple-
tube technique for chlorinated effluents.
LES (two-step) and multiple-tube verifica-
tions. Occasionally, coliform bacteria may fail to
reproduce colonies on membrane filters and non-
coliform organisms may develop sheen colonies.
Verification procedures were undertaken for
coliform organisms on all membranes and multiple-
tube samples in accordance with procedures des-
cribed by Geldreich et al. (30). Calculations for
verification include:
Percent verified (MF) =
BGB verified sheen colonies
total sheen colonies tested
Percent verified (MPN) =
Coliform count by the completed test
Coliform count by the confirmed test
X 100
X 100
The results of verification for the LES (two-step)
MF and confirmed multiple-tube methods are
summarized in Table 4. From 263 membrane
filters, 1110 sheen colonies were selected for
verification; 89.6 percent were verified as coliform
organisms. Trickling filter effluent displayed the
highest verification (97.5 percent) from the MF
technique. It was also the highest (93.3 percent)
in using the MPN procedure. From 97 MPN
samples a wide range of verifications (22-100
percent) were observed; however, 80 percent of
the MPN samples reflected 100 percent verifica-
tion. The average verifications for both the MF and
MPN methods were higher than reported by Geld-
reich et al. (30): 78.1 percent for MF and 70.3
percent for MPN on samples of natural waters and
sewages.
Time effect after dechlorination. During the
course of the investigation, the question arose as to
whether or not, after the dechlorination of
samples, the observed bacterial densities signifi-
cantly fluctuated with time. This seemed an
important consideration because of the time ele-
ment involved in performing comparative tech-
niques. To investigate this, samples of three types
of secondary effluent were chlorinated at varying
dosages for a contact time of 15 minutes after
which they were dechlorinated as previously
described and kept at room temperature (20 to
22 C). Bacterial density assays were undertaken
using the LES two-step and the multiple-tube
120
-------
TABLE 4. VALIDITY OF TWO-STEP MF AND CONFIRMED MPN TESTS
MPN Confirmed Test
MF Test
Chlorinated No. of No. of Avg. % Number 100% % Verified
effluent Membranes Colonies Verified Tested Verified Avg. Min.
High rate
act. si. 66
Contact
Stabilization 101
Trickling
341
393
86.2
88.8
27
35
22
26
92.4
91.0
46.8
50.0
Filter
Total
96
263
376
1,110
97.5
89.6
35
97
30
78
93.3
92.0
22.0
22.0
methods at 15 minute intervals for a 2-hour
period. The procedure not only permitted an
assessment of the time element but also provided
an opportunity for more comparative analyses of
the MF versus MPN techniques. The results are
summarized in Table 5.
There was no significant change in coliform
densities during the more than 2-hour period. A
comparison of the paired MF and MPN results
indicates the inherent precision of the MF method
over that of the MPN.
Fecal Coliforms
Comparison for assaying fecal coliforms was
made using the M-FC MF technique recommended
by Geldreich et at. (16) and the confirmed MPN
procedures (4). These procedures have been accept-
ed for fecal coliform enumerations on unchlori-
nated wastewater. Four chlorinated effluents were
examined. One effluent, representative of the
Bloomington-Normal, III., Sanitary District's acti-
vated sludge process, was collected from a chlorine
contact tank effluent stream and immediately
dechlorinated; the other three were treated with
various dosages of chlorine as previously described.
The results of the two assay methods on the
four effluents are shown in Figure 5. It is apparent
that most of the plotted points lie below the line
of equality. In fact, 78 are below, 12 are above,
and 6 are on the line. Adjusting the equality line
for MPN bias as described by Thomas (32) does
little to change the pattern; 74 points are below
and 22 are above the line. In several cases the dis-
-4
u_
s
u_
O
Q o
0
EXPLANATION
HIGH-RATE ACTIVATED SLUDGE
^CONTACT STABILIZATION
oTRICKLING FILTER S-*
iCI2 CONTACT TANK
EFFLUENT SI*
ALLOWANCE FOR
MPN BIAS
0246
LOG OF CONFIRMED MPN/100 ml
Figure 5. Fecal Coliform densities as determined
by the Membrane Filter and most
probable number techniques on
Chlorinated Effluents.
crepancy is by a factor of 10 or more which are
not apparent in Figure 5.
It can be concluded that the M-FC MF tech-
nique for fecal coliform detection, when applied
to chlorinated wastewater effluents, is less efficient
in recovery than the confirmed MPN procedure.
Other media or an enrichment step similar to
that used in the LES two-step procedure for total
121
-------
TABLE 5. COMPARISON OF TOTAL COLIFORM DENSITY (PER 100 ml) IN EFFLUENTS
AFTER DECHLORINATION, DETECTED BY MF AND MPN TECHNIQUES
Time (min.) after High rate act. si.
dechlorination MF MPN
Contact stabilization
MF
MPN
Trickling filter
MF
MPN
0
15
30
45
60
75
90
105
120
135
Arithmetic mean
Geometric mean
Median
Mode
Coeff. of
variation, %
% verified avg.
Range of %
verified
1,500
1,500
1,500
1,600
1,800
1,500
1,600
1,500
1,500
1,500
1,600
1,500
1,500
1,500
6.9
92.0
77.0-100
1,700
1,700
1,700
1,700
1,700
2,200
2,400
2,400
2,200
2,200
2,000
2,000
1,900
1,700
15.7
96.9
69.0-100
6,400
4,100
4,000
4,000
5,200
3,200
3,800
4,500
3,900
4,400
4,300
4,000
4,000
21.4
93.5
83.3-100
7,900
6,300
4,900
3,300
3,500
11,000
7,900
3,500
3,500
5,800
5,200
4,900
3,500
46.8
97.9
78.8-100
23,000
21,000
25,000
25,000
23,000
24,000
22,000
23,000
23,000
24,000
23,000
23,000
23,000
23,000
5.6
91.4
73.4-100
23,000
35,000
17,000
28,000
35,000
35,000
49,000
22,000
22,000
22,000
29,000
27,000
25,000
22,000
33.2
95.3
53.2-100
Chlorine dosage
mg/1
Contact period,
mm.
4.0
15
2.0
15
3.0
15
*MF = LES (two-step)
MPN = Completed tests
coliform might improve the recovery efficiency.
Further investigations along these lines would seem
justified. Braswell and Hoadley (21) found that the
use of Trypticase soy agar was superior to the MPN
and MF techniques for E. coli recoveries in chlori-
nated secondary sewage.
The minimum and maximum fecal coliform
ratios of MF/MPN for all tests were 0.17 and 1.46
respectively. The median ratio was 0.70. Based on
observations from 96 comparative runs, the rela-
tionship of fecal coliform densities in chlorinated
effluents for the two procedures can be expressed
as
log MF = 0.012 + 0.942 log MPN
The correlation coefficient is 0.987. Until a more
precise procedure is developed for using MF
techniques in recovering fecal coliforms from
chlorinated wastewater, a mathematical expression
of this nature may be useful for estimating MPN
densities. The results should be multiplied by the
factor 0.851 as described by Thomas (32) for an
estimate without bias.
Verifications of membrane developed colonies
were made using a phenol red lactose broth and EC
broth. A total of 616 blue colonies were fished for
verification; 87.7 percent were verified. This was
lower than 93.2 percent verification reported by
Geldreich et al. (16) on pure cultures.
122
-------
Fecal Streptococci
Multiple-tube versus membrane filter. To
compare the MPN and MF techniques a series of
FS tests on unchlorinated samples from a variety
of sources were performed. The results are sum-
marized in Table 6. A statistical test of the ob-
served data was made using the t-test of pairing
observations to determine whether there is a
significant difference in FS recoveries by the
MPN and MF2 methods. The results indicate there
is no statistical difference in the mean value of the
bacterial counts determined by the two methods
(test 4 of Table 2). The FS densities obtained
from both procedures are comparable and prob-
ably have the same sanitary significance. Therefore
the laboratory techniques of this study were con-
sidered acceptable.
One hundred and thirty-one chlorinated
samples taken from three secondary sewage ef-
fluents were concurrently assayed for FS densi-
ties by the MPN and MF procedures. The colonies
developed on the membrane filter were counted
after 2,3,4, and 7 days incubation and were desig-
nated MF2, MF3, MF4, and MFy, respectively.
The MF2 and MPN method is recommended by
Standard Methods (4).
The comparative results of MF2 and MPN on
chlorinated samples are presented graphically in
Figure 6. It is apparent that most of the plotted
points lie below the line of equality. In fact, 107
plotted points are below, 19 are above, and 5 are
on the line. From adjusting the equality line for
the MPN bias, as described by Thomas (32), most
of the plotted points (97 points) are below the
MPN bias reference line, 32 points are above and 2
are on the line. Statistically significant differences
were found in FS recoveries, when the MPN
procedure with the MF2 method (Test 5 of Table
2). It can be concluded that the MF2 procedure
gives lower FS recovery on chlorinated effluents
than does the MPN technique. It seemed reason-
able that enrichment and prolonged incubation
might improve FS recovery using the MF method.
TABLE 6.
MOST PROBABLE NUMBER AND
MEMBRANE FILTER COUNTS,
FECALISTREPTOCOCCI PER 100
ML IN UNCHLORINATED WATERS
Source
MPN
Illinois River
Spoon River
Spring Lake
Twin Lake
Havana Farm Pond
Fiatt Farm Pond
High-rate Activated
Sludge Process effluent
Contact Stabilization
Process Effluent
140
140
130
4,600
2,200
790
700
540
350
280
230
220
170
1,300
340
110
1,400
35,000
33,000
27,000
22,000
22,000
3,000
9,400
4,900
4,900
4,600
64
60
140
3,100
1,600
900
830
900
300
300
240
200
230
1,400
420
75
1,200
34,000
30,000
30,000
26,000
24,000
2,000
7,700
5,400
4,700
4,300
Enrichment. Azide dextrose broth is the me-
dium used for the presumptive test of the MPN
method for fecal streptococci in waters. Brain-
heart infusion broth and bile broth medium are
the confirmation media of FS for the MF method.
These three media were used in this study for en-
richment purposes in efforts to enhance FS re-
covery in chlorinated effluents. The results of FS
recovery on M-Enterococcus agars (MF method)
Trickling Filter
Process Effluent
24,000
11,000
9,400
4,600
24,000
13,000
7,000
5,400
*0ne-step M-Enterococcus agar MF count with
two-day incubation
123
-------
10.5
o
o
103
CM
101
EXPLANATION
HIGH-RATE ACTIVATED SLUDGE,
^CONTACT STABILIZATION
oTRICKLING FILTER
n=131
ALLOWANCE FOR
MPN BIAS
10'
103
MPN/100 ml
105
Figure 6. Fecal Streptococci densities as
determined by the One-Step
Membrane Filter and most probable
number technique on Chorinated
Effluents.
240
0
o
o 80
o
o
o
Q- O
LU or
co £
CHLORINATED HIGH'RATE ,
ACTIVATED SLUDGE EFFLUENTS
n=24 /
/
0 80 160 240
ONE-STEP M-ENTEROCOCCUS AGAR MF2 COUNT
Figure?. Comparison of Fecal Streptococci Counts
on M-Enterococcus Agars with and
without Azide Deztrose Broth Enrichment.
with and without enrichment for chlorinated
samples are shown in Figure 7, 8, and 9. All FS
counts in these figures were made after a two-day
incubation. Although the number of colonies per
filter as depicted in these figures exceeded the
desirable range of 20 to 100 per filter, they were
considered satisfactory for comparison purposes.
With AD broth enrichment, all plotted points
lie below the equality line (Figure 7). In other
words, the FS recovery from chlorinated effluent
on M-Enterococcus agar with AD broth enrich-
ment falls far short of that without enrichment.
This is substantiated by the t-test (Test 6 of Table
2) and it is concluded therefore that enrichment
with AD broth inhibits the FS recovery of chlori-
nated samples on membranes.
Figure 8 shows no appreciable difference in
FS counts with or without BHI broth enrichment.
Eleven plotted points lie above, 9 lie below, and 5
points are on equality line. A statistical test (Test 7
of Table 2) suggests no significant difference in
FS recoveries from chlorinated effluents deter-
mined by the MF method with or without enrich-
ment. Using the least square regression technique,
the plotted points in Figure 8 can be fitted as
follows:
x 9.
LU «=
LU CC
cc. <
Q. O
03 <
T CO
^ O
^ O
V)
CHLORINATED TRICKLING
FILTER EFFLUENTS
n = 25
i /
/
/
OLL
0 40 80
ONE-STEP M-ENTEROCOCCUS AGAR MF2 COUNT
Figure 8. Comparison of Fecal Streptococci Counts
on M-Enterococcus Agars with and
without Brain.Heart Infusion Broth
Enrichment.
124
-------
o
^160
LLJ <£
CO
80
&
0
/
»/_
/
A X
-;x
A*£/
^A/X EXPLANATION
. 'A n = 53
o TRICKLING FILTER
ACONTACT STABILIZATION
HIGH-RATE ACTIVATED SLUDGE
0 80 160
ONE-STEP M-ENTEROCOCCUS AGAR MF2 COUNT
Figure 9. Comparison of Fecal Streptococci Counts
of M-Enterococcus Agars with and
without Bile Broth Enrichment.
Y = 0.097 + 0.98 X
in which Y = FS counts by M-Enterococcus agar
MF2 with BHI medium enrichment, in organisms
per 100 ml; X = FS counts by one-step M-Enter-
ococcus agar MF2 method, in organisms per 100
ml. The correlation coefficient is 0.97. Equation 5
shows the slope to be 0.97 with an intercept of
0.097. Thus the regression line expressed by
Equation 1 is almost identical with a 45 degree
line. From these tests, it can be reasonably con-
cluded there is no advantage to BHI broth enrich-
ment for the MF method on chlorinated effluent
samples.
It is quite evident from the data depicted in
Figure 9 that the FS recovery using bile broth
enrichment is higher than FS recovery by non-
enrichment techniques. Fifty-three comparisons
were made on three effluents and only two ef-
fluent samples showed the enrichment FS counts
slightly less than the nonenrichment. This is con-
firmed by statistical analyses (Test 8 of Table 2).
It is concluded that bile broth enrichment did
improve FS recovery on chlorinated effluent
samples.
A peptone yeast-extract casitone enrichment
broth was suggested by Rose and Litsky (28) for
use with the MF method for the enrichment of
FS recovery in unchlorinated waters. To deter-
mine the efficiency of PYC broth on chlorinated
effluents samples, parallel tests were made with
PYC broth, bile broth medium, and without
enrichment on portions of the same samples.
About one-half of the experimental results
were discarded due to extremely high or low
counts. The results (68 samples), where filter
counts were in the desirable range of 20 to 100,
are summarized in Table 7. The values of Table 7
represents a two-day incubation period. For all
tested effluents, with few exceptions, the recovery
of FS increased with enrichment and especially
with bile broth enrichment.
The recovery ratios of enrichment to non-
enrichment for each effluent are presented in Table
8. The highest ratios were 2.45:1 and 1.77:1 for
bile broth and PYC enrichment, respectively.
Similarly, the overall average ratios for the 68
samples were 2.14:1 and 1.60:1. In comparison
with the work done by Rose and Litsky (28), the
recovery ratio of PYC enrichment to M-Enter-
ococcus agar was 2.44:1 in waters.
Nineteen chlorinated samples were examined
for FS densities by both MPN and PYC enrichment
MF methods. The results from these assays are
depicted in Figure 10. The equality line was
adjusted for MPN bias as described by Thomas (32)
and used for reference. Fourteen plotted points
lie below the equality line, 4 are above and 1 is on
the line. A t-test analysis confirmed the differences
105
O o
^ 103
CM
10
EXPLANATION
HIGH-RATE ACTIVATED SLUDGE /'
ACONTACT STABILIZATION /
^TRICKLING FILTER / '
n=19 A/X& '
/
/°
101
103
MPN/100 ml
Figure 10. Comparison of Fecal Streptococci
densities determined by the MPN and
the PYC Enrichment MF methods.
125
-------
TABLE 7. COMPARISON OF RECOVERY OF FECAL STREPTOCOCCI ON
M-ENTEROCOCCUS AGARS WITH AND WITHOUT ENRICHMENT
Sample
Number
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
Average
Activated Sludge
NE
12
75
61
17
32
57
36
19
25
12
29
65
12
17
45
19
31
28
42
52
46
35
PYC
22
74
95
30
57
73
47
30
37
20
46
71
22
28
52
30
48
45
66
83
86
51
Bile
50
180
160
54
111
97
52
54
68
26
59
90
26
35
74
36
58
48
70
93
96
73
Contact Stabilization
NE
54
28
48
72
61
23
69
44
20
53
59
44
34
88
39
24
53
31
23
40
72
47
PYC
75
32
80
95
86
29
91
62
30
104
90
65
36
91
72
34
71
46
38
65
100
66
Bile
83
38
90
112
91
29
107
63
30
90
104
80
52
95
62
38
82
70
58
98
122
76
Trickling Filter
NE
20
16
20
18
26
32
40
60
24
18
60
16
30
20
24
18
11
12
21
50
16
30
18
24
30
16
26
PYC
35
22
52
34
60
66
80
100
40
29
96
30
38
36
42
32
25
21
22
89
38
36
32
28
56
30
45
Bile
40
46
50
38
58
90
100
110
54
45
126
38
54
50
48
45
40
39
34
132
36
62
52
56
90
50
61
Incubation time was 48 hours for all cases; NE means M-Enterococcus agar without enrichment;
PYC means with PYC broth enrichment; and Bile means with bile broth medium enrichment.
TABLE 8. FS RECOVERY RATIOS OF ENRICHMENT OF NON-ENRICHMENT
Ratio of*
PYC/NE
One-step M-Enterococcus agar MF count with two-day incubation
126
Bile/NE
Chlorinated effluent
High-rate Activated Sludge
Contact Stabilization
Trickling Filter
Overall
Range
0.99-1.87
1.06-1.96
1.05-2.60
0.99-2.60
Average
1.54
1.44
1.77
1.60
Range
1.38-4.16
1.08-2.45
1.62-3.25
1.08-4.16
Average
2.24
1.67
2.45
2.14
-------
(Test 9 of Table 2). From this test it is concluded
that the recovery of FS from chlorinated effluents
on PYC enriched membrane filters is less than for
the MPN procedure. Although prolonged incuba-
tion through seven days on PYC enriched filters
showed increasing counts with time. No attempt
was made to compare prolonged PYC enriched
MF counts with MPN values.
Bile broth enrichment, as mentioned earlier,
gave the highest recovery of FS from chlorinated
effluent samples. To compare the bile broth en-
riched MF2 results with the MPN data, 45 chlori-
nated samples collected from three sewage efflu-
ents were subjected to FS assays, in parallel, by
both methods. The results of the analyses are pre-
sented in Figure 11. The ratios of the bile enriched
MF2 to the MPN FS densities were calculated,
arrayed in order of magnitude, and plotted on log-
probability paper. The line of the best fit was
drawn. The median, or 50 percentile of the 45
ratios is 1.00. In fact, 4 ratios are equal to, 21 are
greater than, and 20 ratios are less than unity. This
indicates that the bile enriched MF2 data is in very
close agreement to that data obtained by MPN
techniques. It was also observed that there was no
significant increase in FS count on the bile en-
riched filters for prolonged incubation up through
seven days. It is concluded that the bile enrichment
MF2 method is superior to the PYC enrichment
MF2 method and comparable to the MPN proce-
dure for the recovery of FS in chlorinated
secondary sewage effluents.
3.0
Q_
1.0
00
0.3
EXPLANATION
HIGH-RATE ACTIVATED SLUDGE
^CONTACT STABILIZATION
o TRICKLING FILTER ^
n=45
0.5 10 50 90
PERCENT PROBABILITY OF NOT EXCEEDING
Figure 11. Analysis of Fecal Streptococci Recovery
made on Chlorinated Effluents by use of
Multipletube (MPN) Test and of
M-Enterococcus Agar MF with Bile Broth
Enrichment.
Prolonged Incubation. As stated earlier, the
M-Enterococcus agar MF2 (non-enriched) techni-
que tends to produce lower FS recovery than the
MPN procedure on chlorinated effluents (see
Figure 6). Colonies developed for two-day incuba-
tion were generally small. To check the effects of
prolonged incubation on FS recovery for the
M-Enterococcus agar MF technique all filters
were counted at the end of 2, 3, 4, and 7 days
incubation periods. Figure 12, a typical example,
shows the general trend of the FS counts with
incubation time. The FS recovery increased signifi-
cantly up through the three-day period. After
three days, the FS counts leveled off for chlori-
nated effluents. For the unchlorinated effluent
sample, no significant increase was found in FS
counts after a two-day incubation. The ratios of
MF3 to MF2 for unchlorinated and chlorinated
effluents are summarized in Table 9. For chlori-
nated samples the average MF3/MF2 values ranged
from a low of 1.27 for contact stabilization efflu-
10,000r;
5000
o
o
o
o
o
£
cc
in
CJ
LU
1000
500
100
i I I I i i i _
UNCHLORINATED CONTACT -
STABILIZATION
-v v
HIGH-RATE ACTIVATED-
SLUDGE
TRICKLING FILTER
CONTACT STABILIZATION
-A
CHLORINE DOSAGE: 2 mg/l
CONTACT TIME: 15 min
I I I I I I i
2468
INCUBATION TIME, days
Figure 12. Recovery of Fecal Streptococci on
M-Enterococcus Agar from one
Unchlorinated and three Chlorinated
Secondary Effluents.
127
-------
TABLE 9. FECAL STREPTOCOCCI COUNT MF3/MF2 RATIO
Unchlorinated
Chlorinated
Type of Effluent
Number
of
Sample
Range
Average
Number
of
Sample
Range
Average
High-rate Activated
Sludge
Contact
1.00-1.33
1.14
41
1.17-4.68
2.12
Stabilization
Trickling Filter
Overall
4
4
12
1.00-1.09
1.00-1.14
1.00-1.33
1.07
1.06
1.09
39
44
124
1.07-1.67
1.07-4.28
1.07-4.68
1.27
2.07
1.84
ent, to a high of 2.12 for high-rate activated sludge
with an overall average of 1.84.
To compare the non-enriched MF3 data with
the MPN results, 124 comparisons, made on three
chlorinated effluents, are depicted in Figure 13.
Seventy-six plotted points are above the line of
equality, and 38 are below. Using the corrected
MPN bias as a reference line, 101 points are above,
21 are below, and 2 are on the line. The MF3 re-
sults were found to be slightly higher than the
MPN data, especially when the FS counts were less
than 500/100 ml (Figure 13). For the 124 in-
stances, the geometric mean values were 1,300
MF3/100 ml and 1,000 MPN/100 ml. The geo-
metric standard deviations were 3.98 and 4.93 for
the MF3 and the MPN methods, respectively. How-
ever, a statistical test (Test 10 of Table 2) did not
indicate a significant difference between the MPN
and MF3 methods.
When comparing MF3 and MPN results for
124 chlorinated effluent samples in a manner
similar to that depicted in Figure 11 the median, or
50 percentile, for the MF3/MPN is 1.11. Although
the MF3 values are slightly higher than the MPN
data, the MF procedure for 3-day incubation on
M-Enterococcus agar, without enrichment, appears
applicable for the FS assay of chlorinated efflu-
ents.
Verification. A total number of 967 colonies
were fished from 306 membrane filters and sub-
jected to the verification procedure outlined in
Standard Methods (4). The results of the verifica-
tion are summarized in Table 10. These include
105
\ 103
CO
EXPLANATION
HIGH-RATE ACTIVATED SLUDGE
^CONTACT STABILIZATION
^TRICKLING FILTER
n 124
ALLOWANCE FOR MPN BIAS
101
103
MPN/100ml
105
Figure 13. Comparison of Fecal Streptococci
densities determined by the MPN and
M-Enterococcus Agar MF3 methods.
colonies grown on filters placed on M-Enterococ-
cus agars with and without enrichment. After two-
day incubation, all of 688 colonies isolated from
unchlorinated and chlorinated effluents were, veri-
fied as fecal streptococci. Although Kenner et al.
(34) reported similar 100 percent recovery of FS
from the membranes for the fecal samples, Rose
and Litsky (28) experienced a 94.6 percent FS
verification from filters placed on M-Enterococcus
agars with and without PYC enrichment for
128
-------
TABLE 10. VERIFICATION OF FS GROWN ON FILTERS PLACED ON M-ENTEROCOCCUS AGARS
WITH AND WITHOUT ENRICHMENT
Sample
Unchlorinated Effluents
Chlorinated Effluents
Overall
Growth after
days of
Incubation
2
2
3
4
No. of
Filter
27
189
74
16
306
No. of
Colonies
Examined
92
596
234
45
967
Positive Verified
No. of
Colonies Percent
92
596
223
42
954
100
100
95.3
93.3
98.6
natural waters. From markings placed on the back
of petri dishes during this study it was possible to
distinguish two, three, and four days growth colo-
nies. About 5-7 percent of the three and four day
growth colonies were not verified as FS (Table 10).
SUMMARY AND CONCLUSIONS
Two series of laboratory assays were per-
formed to determine whether or not the standard
membrane filter (MF) procedure for total coli-
forms, fecal coliforms, and fecal streptococci
detections on chlorinated secondary sewage efflu-
ents was comparable to that obtained by the mul-
tiple-tube (MPN) method. If not found to be the
case, efforts were made to improve bacteria re-
coveries using various modifications of the MF
method.
Grab samples of secondary effluents were
collected with up through 6 mg/1 of chlorine,
stirred, and dechlorinated by sodium thiosulfate.
After varying periods of contact the samples were
assayed for bacteria. Based upon the results derived
from this work, the following conclusions were
developed.
For chlorinated secondary sewage effluents,
the recoveries of TC, FC, and FS by the standard
MF (one-step nonenrichment) method is signifi-
cantly less than that obtained by the standard
MPN procedure.
The use of the LES two-step MF method is
comparable to the completed MPN procedures for
total coliform detection. Total coliform recovery
by the LES two-step MF technique is approxi-
mately 1.5 times that obtained using the M-Endo,
one-step MF procedure. From 273 filters using the
LES two-step MF procedure and 1,110 sheen
colonies, 89.6 percent were verified as colifnrm
organisms.
Estimates of FC MPN densities may be de-
rived from the MF procedure by using a mathe-
matical relationship similar to log MPN = 1.062
log MF 0.014. For FC verification, 87.7 percent
of 616 blue colonies were verified.
The use of azide-dextrose broth, brain-
heart infusion broth and peptone yeast-extract
Casitone for enrichment purposes, with the M-
Enterococcus agar MF2 procedure, did not satis-
factorily increase the sensitivity of the procedure
for FS assays. Enrichment with bile broth medium
of the M-Enterococcus agar MF2 procedure signifi-
cantly increases the FS recovery to the extent that
the procedure is comparable to the multiple-tube
method.
The recovery of FS using the membrane filter
technique with M-Enterococcus agar increased
significantly after three days incubation (MF3>
compared to two days incubation (MF2>; and the
MF3 procedure is comparable to the multiple-tube
method for FS detection. The membrane filter
technique preferred for FS assays is the MF2
procedure using M-Enterococcus agar with bile
broth enrichment. All of 688 colonies for two-day
incubation on filters were verified as fecal strep-
tococci.
129
-------
ACKNOWLEDGEMENTS
The writer wishes to express appreciation to
Ralph L. Evans, Head of the Water Quality Sec-
tion, Illinois State Water Survey, Peoria, for his
encouragement and review of this paper. Special
acknowledgements are also extended to Davis B.
Beuscher, Jack W. Williams, Pamella A. Martin and
Donald H. Schnepper of the Water Quality Section
for their assistance during the study; and to John
W. Brother, Jr. of the Survey who prepared the
illustrations.
REFERENCES
1. Illinois Environmental Protection Agency.
Water Pollution Regulations in Illinois. 36 pp.
March 7, 1972.
2. Stanier, R.Y., M. Doudoroff, and E.A. Adel-
berg. The Microbial World. Prentice Hall,
Inc., Englewood Cliffs, N.J., 43 pp., 1963.
3. Kabler, P.W. Microbial Considerations in
Drinking Water. J. Amer. Water Works Assn.,
60(10): 1173-1180, 1968.
4. Standard Methods for the Examination of
Water and Wastewater, 13th ed. Amer. Pub.
Health Assn., Inc., New York, 1971.
5. The Working Group of Water Quality of the
Subcommittee on Water Quality, Interde-
partmental Committee on Water. Guidelines
for Water Quality Objectives and Standards, A
Preliminary Report. Technical Bulletin No.67,
Inland Water Branch, Department of the En-
vironment, Ottawa, Canada, pp. 14-25., 1972.
6. Gallagher, T.P., and D.F. Spino. The Signifi-
cance of Numbers of Coliform Bacteria as an
Indicator of Enteric Pathogens. Water Re-
search, Pergamon Press, Great Britain v.
2(1):169-175, 1968.
7. Geldreich, E.E. Applying Bacteriological Para-
meters to Recreational Water Quality. J.
Amer. Water Works Assn. (2):113-120, 1970.
8. Smith, R.J., and R.M. Twedt. Natural Rela-
tionships of Indicator and Pathogenic Bacteria
in Stream Waters. J. Water Poll. Contr. Fed.
43(1):2000-2209, 1971.
9. Smith, R.J., R.M. Twedt, and L.K. Flanigan.
Relationships of Indicator and Pathogenic
Bacteria in Stream Waters. J. Water Poll.
Contr. Fed. 45(8): 1736-1745, 1973.
10. The Committee on Environmental Quality
Management of the Sanitary Engineering Divi-
sion. Engineering Evaluation of Virus Hazard
in Water. Amer. Soc. & Civil Eng., Journal of
Sanitary Engineering Division, 96(SAI):111-
161, 1970.
11. Geldreich, E.E.,and N.A. Clarke. The Coli-
form Test: A Criterion for the Viral Safety of
Water. In: Virus and Water Quality: Occur-
rence and Control. Proceedings 13th Water
Quality Conference, Department of Civil
Engineering, University of Illinois at Urbana-
Champaign, ML, pp. 103-113, 1971.
12. McKee, J.E., R.T. Mclaughlin, and P. Les-
geurgues. Application of Molecular Filter
Techniques to the Bacteria Assay of Sewage.
III. Effects of Physical and Chemical Disin-
fection. Sewage Ind. Wastes, 30(3):245-252,
1958.
13. McKee, J.E., and R.T. Mclaughlin. Applica-
tion of Molecular Filter Techniques to the
Bacteria Assay of Sewage. II Experimental
Results for Settled Sewage. Sewage & Ind.
Wastes 30(2):129-137, 1958.
14. McCarthy, J.A., J.E. Delaney, and R.J.
Grasso. Measuring Coliforms in Water. Water
& Sewage Works 108(6):238-243, 1961.
15. Geldreich, E.E., H.F. Clark, C.B. Huff, and
L.C. Best. Fecal-Coliform-Organisms Medium
for the Membrane Filter Technique. J. Amer.
Waterworks Assn. 57(2):208-214, 1965.
16. Geldreich, E.E. Sanitary Significance of Fecal
Coliforms in the Environments. U.S. Depart-
ment of Interior, FWPCA publication WP-20-
3, 122pp., 1966.
17. Evans, F.L. Ill, E.E. Geldreich, S.R. Weibel,
and G.G. Robeck. Treatment of Urban Storm-
water Runoff. J. Water Poll. Contr. Fed.
48(5)R162-R170, 1968.
18. ORSANCO Water Users Committee. Total
Coliform: Fecal Coliform Ratio for Evalua-
tion of Raw Water Bacterial Quality. J.
Water Poll. Contr. Fed. 43(4)630-640, 1971.
19. Maxey, R.B. Non-lethal Injury and Limita-
tions of Recovery of Coliform Organisms on
Selective Media. J. of Milk and Food Technol.
33(9):445-448, 1970.
20. Scheusner D.L., F.F. Busta, and M.L. Speck.
Injury of Bacteria by Sanitizers. Appl. Micro-
biol. 21(1):41-45, 1971.
21. Braswell, J.R., and A.W. Hoadley. Recovery
of Escherichia coli from Chlorinated Second-
ary Sewage. Appl. Microbiol. 28(2):328-329,
1974.
22. Hufham, J.B. Evaluating the Membrane Fecal
Coliform Test by Using Escherichia coli as the
Indicator Organism. Appl. Microbiol. 27(4):
771-776, 1974.
23. Presswood, W.G., and L.R. Brown. Compari-
son of Gelman and Millipore Membrane Fil-
ters for Enumerating Fecal Coliform Bacteria.
Appl. Microbiol. 26(3):332-336, 1973.
130
-------
24. Lattanzi, W.E., and E.W. Mood. A Compari-
son of Enterococci and Escherichia coli as In-
dices of Water Pollution. Sewage & Ind.
Wastes. 23(9) :1154-1160, 1951.
25. Litsky, W., W.L. Mailman, and C.W. Fifield.
A New Medium for the Detection of Enter-
ococci in Water. Amer. J. of Pub. Health.
43(7):873-879, 1953.
26. Slanetz, L.W., and C.H. Bartley. Number of
Enterococci in Water, Sewage and Feces
Determined by the Membrane Filter Techni-
que with an Improved Medium. J. of Bact.
74(5):591-595, 1957.
27. Kenner, B.A., H.F. Clark, and P.W. Kabler.
Fecal Streptococci I. Cultivation and Enumer-
ation of Streptococci in Surface Waters. Appl.
Microbiol.9(1):15-20, 1961.
28. Rose, R.E., and W. Litsky. Enrichment
Procedure for Use with the Membrane Filter
for the Isolation and Enumeration of Fecal
Streptococci in Water. Appl. Microbiol.
13(1):106-108, 1965.
29. Pavola, M.T., E. Beauvais, FT. Bresenski,
and W. Litsky. Rapid Assessment of Water
Quality by Flourescent Antibody Identifica-
tion of Fecal Streptococci. Proceeding of the
6th International Water Pollution Research,
Pergamon Press Ltd. London, Great Britain,
A/3/5/1 -A/3/5/8, 1972.
30. Geldreich, E.E., W.L. Jeter, and J.A. Winter.
Technical Considerations in Applying the
Membrane Filter Procedure. Health Lab. Sci.
(4)2:113-125, 1967.
31. Seiz, D. Personal Communication. Illinois
Department of Public Health, Springfield.
32. Thomas, H.A. Jr. Statistical Analysis of
Coliform Data. Sewage & Ind. Wastes 27(2):
212-222, 1955.
33. Dixon, W.J., and F.J. Massey, Jr. Introduc-
tion to Statistical Analysis. McGraw-Hill,
New York, pp. 384, 1957.
34. Bartley, C.H., and L.W. Slanetz. Types and
Sanitary Significance of Fecal Streptococci
Isolated from Feces, Sewage and Water.
Amer. J. Pub. Health. 50(10): 1545-1552,
1960.
QUESTION AND ANSWER SESSION
Geldreich: Dr. Lin, with enterococci, we find
when we try to go through enrich-
ment with extended incubation time,
we run into pediococci and other
organisms which are not fecal streps.
I don't recall whether you had a table
containing this data. Did you check
verification when you extended either
the time of incubation or the use of
these other enrichment devices before
you went to M-Enterococcus to
show that this increase was still from
fecal strep rather than possibly some
of these other species that we know
will grow on M-Enterococcus or
some other strep media? What was
that increase, 80% or 90%? That was
good.
Lin: Eighty to 90% of the increase was
from fecal strep.
Geldreich: So even though you extended the
time and you added enrichments you
didn't get any more false positives.
Good.
Bordner: Dr. Lin, Did you evaluate media other
than M-Enterococcus agar for fecal
streps, for example PSE agar or the
KF medium?
Lin: We restricted our studies to the M-
Enterococcus.
Bordner: Shifting back to the M-FC medium, I
understood you to say that you
haven't looked into enrichment, for
fecal coliforms. Is this your next
plan?
Lin: You mean this one? Yes, but I didn't
succeed.
Bordner: Do you plan to do this in the future?
Lin: Yes.
Bordner: May I ask what brands of filters you
usually use?
Lin: Millipore.
Brodsky: Perhaps I should direct this short
question to Geldreich. I have read in
the literature that M-Enterococcus
agar is selective for certain groups
or species of streptococci. In compar-
ison with PSE agar and KF agar the
terms fecal streptococci and enter-
ococci tend to be used interchang-
ably. Could you clarify these terms?
131
-------
Geldreich: We find that M-Enterococcus agar is
very selective for enterococci but
there are other fecal streptococci that
we are concerned about. These are
from other warm-blooded animals.
Feed lots have a tremendous number
of streptococci which do not recover
too well on M-Enterococcus agar. KF
agar and PSE agar recover these
species much better. We know we get
equivalent results with M-Enterococ-
cus and KF when we use it on domes-
tic sewage because we are looking at
enterococci, the sub-group of fecal
strep. Since we are working on ratio
development of fecal coliform and
fecal strep in a stream, there are times
when we do have animal feedlot dis-
charges and slaughter house waste. It
would be easier to stay with one
medium that will recover all of the
members of the fecal strep group be-
cause otherwise your ratio won't
mean a darn thing. We have always
recommended the KF agar. Recently
PSE agar which is being introduced,
looks like an excellent medium. I am
sure Warren and Fran have used it
and with excellent results. It may be
far superior to M-Enterococcus agar,
particularly for relationships with
fecal coliforms and pollution from
feedlots or from domestic wastes.
132
-------
THE ASTM PROPOSED MEMBRANE FILTER TEST PROCEDURE
FOR THE RECOVERY OF FECAL COLIFORMS
Margareta J. Jackson
Microbe One
Ann Arbor, Michigan
Donald W. Davis
George R. Kinzer
Johns-Manville R&D Center
Denver, Colorado
ABSTRACT
Results of a collaborative study of the pro-
posed ASTM test procedure for the recovery of
fecal coliforms on membrane filters (MF) were
presented. The test procedure is one of a series
planned by the ASTM subcommittee D 19.08
.04.02 to evaluate MF materials. Recoveries of
fecal coliforms on MF's were compared to those
on pour or spread plates using M-FC agar. Analysis
of variance indicated differences in filter brands
and in the results obtained in different labora-
tories. The laboratory effect could not be sep-
arated from sample effect.
The test procedure did not satisfy all of test
objectives and must be rewritten to eliminate
variables and options.
INTRODUCTION
Since the acceptance of the membrane filter
technique for the isolation and enumeration of
total and fecal coliform and fecal streptococcid),
conflicting papers (2,3,4,5,6) have appeared in the
literature about the use of membrane filters as a
method of evaluating the quality of water. In June
of 1973, ASTM brought together microbiologists,
membrane filter manufacturers, media manufac-
turers, regulatory agencies, consumers, and users of
membrane filters to discuss the problems that con-
fronted them. At this meeting, it was decided to
examine the following parameters of membrane
filters:
1. Inhibitory effects
2. Recovery
3. Retention
4. pH
5. Shelf life
6. Sterility
Since existing test procedures for inhibitory
effects and recovery were based on 35C (7,8), it
was agreed to proceed with a draft for inhibitory
effects and recovery using the elevated tempera-
ture of 44.5±0.2C for fecal coliforms.
The standard test method for recovery was
designed to determine the ability of a membrane
to recover fecal coliform organisms from un-
treated water samples on a selective differential
medium. The test method was based on recovery
by the M-FC method currently being used in water
testing laboratories (1) compared to pour and
spread plate procedures.
METHODS AND MATERIALS
Four polluted waters and one raw sewage
were collected by each participating laboratory and
serially diluted to obtain fecal coliform densities
of 20-60 organisms per ml.
Five replicate dilutions of each sample were
filtered through test membranes, transferred to
M-FC agar and incubated at 44.5 C ± 0.2 C for 22-
24 hours. The same dilutions of each sample were
also tested on M-FC agar by the pour plate or
133
-------
spread plate technique. The blue colonies were
counted with a stereomicroscope for the MF
technique and with a Quebec colony counter for
the pour and spread plates. Twenty colonies per
sample from one representative pour and/or spread
plate were verified using lactose broth and EC
broth according to the test procedure.
By comparing relicate membrane filter
counts with replicate pour and/or spread plate
counts, the recovery rate of fecal coliforms on
membrane filters was determined. If the arithmetic
mean counts on five membrane filters was 85 per-
cent or greather than the arithmetic mean of the
five pour and/or spread plate counts, the mem-
brane filter had met the criteria for recovery of
fecal coliforms.
A preliminary study of round robin was
initiated to determine whether a valid test proce-
dure had been drafted. A common source of
media and membrane filters was essential for the
study. Six manufacturers supplied membrane
filters (0.45 micron pore, white, gridded, sterile
47 mm). Each participating laboratory received the
same lot number of membrane filters from the
manufacturers. Difco supplied media with the
same control numbers to each laboratory.
Media
Code
1.
2.
3.
4.
5.
Bacto-Peptone
M-FC Agar
Rosolic Acid
Lactose Broth
EC Medium
0118-01
0677-01
3229-09
0004-02
0314-02
Control No.
602509
586112
596061
597469
598803
Several laboratories were able to test all six
membranes, whereas, the other labs tested two to
four membrane filters. A common data sheet was
used to record each laboratory's evaluation. These
data sheets were then submitted for statistical
analysis to determine whether a valid test proce-
dure had been drafted.
LABORATORIES PARTICIPATING IN
PRELIMINARY ROUND ROBIN - COMMITTEE
D 19.0804.02
1. Methods Development and Quality Assurance
Research Laboratory
U. S. Environmental Protection Agency
Cincinnati, Ohio
2. Canada Center Inland Waters
Microbiology Laboratory
Burlington, Ontario
3. Millipore Corporation
Bedford, Massachusetts
4. Department of Environmental Sciences
University of Massachusetts
Amherst, Massachusetts
5. Division of Laboratory Services
Illinois Environmental Protection Agency
Chicago, Illinois
6. Gelman Instrument Company
Ann Arbor, Michigan
7. Sartorius Membrane Filter GMBH
West Germany
8. Ministry of the Environment
Division of Laboratories
Bacteriology Branch
Rexdale, Ontario
9. Johns-Manville R&D Center
Denver, Colorado
MEMBRANE MANUFACTURERS OR
DISTRIBUTORS PARTICIPATING IN
PRELIMINARY ROUND ROBIN TESTS
1. Sartorius
2. Gelman
Beckman
Anaheim, California
(Manufactured in Germany)
Gelman Instrument Company
Ann Arbor, Michigan
3. John-Manville John-Manville Corporation
Denver, Colorado
4. Oxoid
5. Millipore
6. S&S
Med-Ox Chemicals Limited
Ottawa, Canada
Millipore Corporation
Bedford, Massachusetts
Schleicher and Schuell, Inc.
Keene, New Hampshire
134
-------
RESULTS
(DATA ANALYSIS)
The raw count data from the supplied forms
were converted to a machine-readable form and the
mean and standard deviations were computed for
each set of replicates. These calculated values were
then used to determine the recovery of fecal coli-
form as a percentage of the pour plate (PP) and
spread plate (SP) results. In addition, the recovery
calculations were repeated, rejecting all counts
outside of the ranges of 20 to 60 counts on mem-
brane filters and 30 to 300 colonies on pour or
spread plates. These calculations are listed on Table
1. The "ERR" listed is determined from the
standard deviations of both measurements by cal-
culating the standard deviations as a percentage of
the mean. These percentages were summed and the
recovery was multiplied by this percentage to
obtain the "ERR." The recovery data was used in
the remaining statistical analyses.
The data from all participating laboratories
were plotted. In general, the data appeared scat-
tered for all laboratories except one. This labora-
tory showed no differences between filters.
Data from three laboratories were selected
for the analysis of variance. The laboratories
were selected on the basis that:
1. All six filters were tested.
2. At least 5 water samples were run.
3. Both spread and pour plate standards
were run on all five samples tested.
These laboratories included one filter manu-
facturer, one university, and one government
agency.
Two analyses of variance were run, one using
the spread plate standard and one using the pour
plate standard. The three variables used in the
analyses were:
Laboratories (L)
Filter Membrane (F)
Water Samples (S)
Since the water samples selected by each
laboratory were different, the variability due to
the samples, the laboratory-sample interaction, and
the filter-sample interaction are meaningless for
these analyses.
The results of the analysis of variance are
summarized in Tables 2 and 3.
The following conclusions can be drawn from
the analyses:
1. In both cases the F x L interaction was
not statistically significant, i.e., the fil-
ters behaved the same in all laboratories.
2. There are differences between filters
with manufacturer 3 supplying the best
filter and manufacturer 2 supplying the
poorest filter. All other filters supplied
were about the same. Statistical signifi-
cance represented the 95 to 97.5 percent
level.
3. The difference between laboratories was
highly significant (over 99 percent). It
is not clear from this analysis if these dif-
ferences are due to technique in the
laboratories or due to differences in
water samples used, since these sample
differences are included in this effect
and in the residual and cannot be sepa-
rated.
The contribution of the various sources to
the total test variance were determined using com-
ponents of variance analysis for both sets of data.
Since the F x L interaction was not significant, a
better estimate of residual error can be made by
pooling the sum of squares for this term and for
the "residual". The calculation of these variances
is listed in Tables 4 and 5. It should be noted
that the variance contribution of the filters is only
5.5 percent in the case of spread plates or 7.3 per-
cent in the case of the pour plates of the total
variance. The large sources of variance are the
"residual" (unexplained or random) variance and
the inter-laboratory differences.
The net effect of these large variances is to
render the results of the test suspect. The expected
value of the test of any filter by any laboratory
would fall in a range of ± 2 standard deviations 95
percent of the time. This factor is ± 78 for spread
plates and ± 85 for pour plates. This variability
makes the proposed test useless.
The proposed test procedure permitted the
use of either spread or pour plates as the standard
at the discretion of the testing laboratory. A
scatter plot of the results obtained in this test from
spread and pour plates on the same samples from
135
-------
all laboratories is shown in Figure 1, although
higher counts were obtained using spread plates.
Statistical analysis of the data indicated that the
variability of the data from the two methods
was not statistically significant.
200
O
o
100
LLJ
cc
Q_
CO
0 100 200
POUR PLATE COUNT
Figure 1. Scatter Plot of Spread and
Pour Plate Results.
CONCLUSIONS
We can reach the following conclusions based
on the data analyses:
1. There are differences between filters.
2. Different results are reached in different
laboratories although filters behave simi-
larly in all laboratories.
3. Unknown causes contributed markedly
to the variability of test results. This
could be due to differences in bacterial
population and pollutants in the test
samples.
4. There is no evidence that pour or spread
plates are more variated. Spread plates
have a higher recovery than pour plates.
5. We do not have a satisfactory test.
RECOMMENDATIONS
The test as presently set up is not satisfactory.
Since we cannot separate the effects of samples
from the effect of laboratories with the present
data, we cannot be sure if the variability is due to
differences between technique or to sample dif-
ferences. If the same samples could be run at a
limited number of laboratories with a limited
number of filters, these effects could be clarified.
This would require either a mixed pure culture
approach or the shipping of samples by air from a
common point with all its inherent problems.
Neither proposal is ideal.
The procedure should be rewritten to leave
nothing to the discretion of the person running
the test before another round robin is proposed.
A decision should be made limiting the standard
to either spread or pour plates to reduce the
amount of work required.
REFERENCES
1. Standard Methods for the Examination of
Water and Wastewater, 13th Ed., Am. Pub.
Health Asso., Nqw York, N.Y., 1971.
2. Levin, G.W., V.L. Strauss, W.C. Hess. Rapid
Coliform Organism Determination With C14,
J. Water Pollut. Cont. Fed. 33:1021-1937,
1961.
3. Hufham, James B. Evaluating the Membrane
Fecal Coliform Test by Using Escherichia coli
as the Indicator Organism. Appl. Microbiol.
27:771-776, 1974.
4. Presswood, W.C., and L.R. Brown. Compari-
son of Gelman and Millipore Membrane
Filters for Enumerating Fecal Coliform Bac-
teria. Appl. Microbiol. 26:332-336, 1973.
5. Dutka, B.J., M.J. Jackson, and J.B. Bell.
Comparison of Autoclave and Ethylene
Oxide-Sterilized Membrane Filters Used in
Water Quality Studies. Appl. Microbiol. 28:
474-480, 1974.
6. Schaeffer, D.J., M.C. Long, and K.G. Jan-
ardan. Statistical Analysis of the Recovery
of Coliform Organisms on Gelman and
Millipore Filters. Appl. Microbiol. 28:605-
607, 1974.
7. Interim Federal Specification, NNN-d00370
(DSA-DM) dated 13, April, 1965.
8. Specification NIH-01-119, dated March 25,
1071.
136
-------
Table I. Fecal Coliform Recovery Data
L
A
B
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
I
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
H20
SPL
TYP
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-2
S-2
S-2
S-2
S-2
S-2
S-?
S-2
S-2
S-2
S-2
S-2
S-2
S-2
S-2
S-2
S-2
S-2
M
F
G
1
1
1
1
2
2
2
2
3
3
3
3
4
4
5
5
5
5
£
6
6
6
1
i
1
1
2
2
2
2
3
3
3
3
4
4
5
5
5
5
S
T
D
GA
GA
PA
PA
GA
GA
PA
FA
GA
GA
PA
PA
GA
GA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
GA
GA
PA
PA
T
S
T
PP
SP
PP
SP
PP
SP
P?
SP
PP
SP
PP
SP
PP
S?
PP
S?
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PO
SP
OP
SP
PP
SP
P?
SP
PP
SP
*
*
«
ft
*
ft
ft
*
*
ft
ft
ft
ft
ft
ft-
ft
ft
ft
ft
ft
ft
*
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
RECOVERY - PERCENT OF STO
ALL DATA * SELECTED DATA
AVG ERR * AVG ERR
92.92 57*32 * 83.33 28.46
80.15 43.27 * 88.49 27.87
46.01 4S.66 *
39.69 38.92 *
12.38 9.97 *
10.68 7.77 *
77.87 68.70 *
67.17 54.08 *
114.15 54.80 * 86.00 32.38
93.47 39.68 * 91.32 31.96
139.82 113.48 * 105.33 74.58
120.61 88.59 * 111.85 76.23
69.91 65. 86 *
60.30 52.16 *
111.50 49.28 * 84.00 28.42
96.18 35.10 * 89.20 27.82
107.07 6!f>.P>6 * 87.30 40.13
92.36 5C.30 * 92.92 40.15
69. C2 54. f^ ft 71,66 19*24
59.54 42.54 * 76.10 18.42
109.73 98.65 * 104.44 72.84
94.65 77.80 * 110.91 74.41
42.94 20.14 * 35.79 10.13
29.13 15.84 *
33.76 19.64 * 39.52 16.14
22.89 15.03 *
3.63 1.18 *
2.46 0.99 *
32.47 10.95 * 35.52 8.83
22.02 9.08 *
66.66 10.36 * 61.43 6.22
45.21 10.42 ft
60.68 15.86 * 53.41 11.14
41.15 13.84 *
36.53 18.73 * 41.39 16.82
24.78 14.56 *
48.07 23.60 * 39.79 12.74
32.60 18.45 *
1)8.76 21.02 # 44.33 24.76
39.85 17.25 *
*
#
ft
#
*
*
ft
ft
ft
»
ft
*
#
#
ft
*
«
*
*
«
#
Jt
*
«
ft
ft
«
#
#
ft
*
«
*
«
»
ft
«
ft
*
*
ft
*
*
CODES -
H20
SPL TYP COLUMN
TST/STD COLUMNS:
: P = POLLUTED WATER S = SEWAGE
GA = GRIDDED MEMBRANE ON AGAR
PA = PLAIN MEMBRANE ON AGAR
SP = SPREAD PLATE, PP = POUR PLATE
137
-------
Table I. Fecal Coliform Recovery Data (Cont'd.)
L
A
B
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
H20
SPL
TYP
S-2
S-2
S-2
S-2
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P- 1
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
M
F
G
6
6
6
6
1
1
1
1
2
2
2
2
3
3
3
3
4
/;
5
5
5
5
6
6
6
6
1
1
1
1
2
2
2
2
3
3
3
3
4
S
T
D
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
GA
GA
PA
DA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
T
S
T
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
£P
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
pp
SP
PP
SP
PP
SP
PP
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
*
ft
ft
ft
*
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
*
ft
ft
ft
*
RECOVERY -
ALL
AVG
9.40
6.37
33.76
22.89
68.55
68.55
44.92
44.92
44.63
44.63
bl.59
bl . 59
74.20
74.20
76.66
76.66
bO.72
bOo 72
62.17
62.17
65.21
65.21
bO.OO
bO.CO
62.60
62.60
62.61
62.33
72.52
72.19
b6.75
b6.50
68.46
68.16
119.81
119. ?8
90.99
90.58
72.97
DATA
ERR
4.70
3.66
11.25
9.35
45.89
24.87
47.50
33.72
27.25
13,56
34.81
18.99
44.84
22.09
47.81
24,30
37,12
21.57
37.32
18.25
46.16
26.16
40.26
24.93
37.76
18.56
22.14
21.82
30.30
29.91
17.57
17.29
22.76
22.42
61.68
60.97
29.73
29.28
26.86
PERCENT OF STD *
ft
*
#
#
ft
ft
*
ft
ft
ft
ft
#
*
ft
ft
»
#
ft
*
a
#
*
#
ft
ft
ft
ft
*
ft
ft
ft
ft
#
«
*
ft
*
#
»
ft
*
SELEC
AVG
33.76
39.64
58.57
53.57
62.61
62.33
72.52
72.19
56.75
56.50
68.46
63.16
103.60
103.13
90.99
90.58
72.97
TED DATA
ERR
11.25
14.72
5.84
12.35
22.14
21.82
30.30
29.91
17.57
17.29
22.76
22.42
37.01
36.48
29.73
29.28
26.86
*
#
*
#
«
*
*
«
*
ft
*
*
*
«
#
«
«
*
#
«
*
*
*
*
#
*
ft
*
*
«
*
#
ft
*
#
*
*
ft
*
ft
ft
CODES -
H20 SPL TYP COLUMN:
TST/STD COLUMNS:
GA
PA
SP
= POLLUTED WATER S = SEWAGE
GRIDDED MEMBRANE ON AGAR
PLAIN MEMBRANE ON AGAR
SPREAD PLATE, PP = POUR PLATE
138
-------
Table I. Fecal Coliform Recovery Data (Cont'd.)
L
A
B
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
H20
SPL
TYP
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-4
P-4
P-4
P-4
P-4
P-4
P-4
P-4
M
F
G
4
5
5
5
5
6
6
6
6
1
1
1
1
2
2
2
2
3
3
3
3
4
4
5
5
5
5
6
6
6
6
1
1
1
1
2
2
2
2
S
T
D
GA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
T
S
T
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
FP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
*
*
*
*
*
*
*
*
*
ft
*
*
*
*-
*
*
*
*
*
*
*r
*
*
*
*
*
*
*
*
*
*
*
*
*
#
*
*
*
*
*
*
*
RECOVERY - PERCENT OF STD
ALL
AVG
72.64
79.72
79.37
90.99
90.58
44.14
43.94
71.17
70.85
110.06
84.69
84.75
65.21
79.26
60.99
100.00
76.95
i 1 0 i 3 6
84.92
114.32
87.97
80.18
61.70
92.68
71.31
90.24
69.44
65.54
bO.43
96.95
74.60
37.32
25.27
38.52
26.08
6.13
4,18
42.71
28.91
DATA
ERR
26.48
25.71
25.31
26.02
25.58
15.44
15.21
25.84
25.47
51.02
43.29
26.66
23.62
23.96
21.34
35.79
31.20
26.46
25.94
27.63
25.45
35.63
30.35
26.11
23.49
22.37
20.52
27.39
23.48
23.87
21.92
14.88
8.01
16.74
9.20
9.26
5.93
21.57
12.24
#
*
#
*
#
#
*
#
*
*
*
*
*
*
#
*
*
*
#
*
#
#
#
«
*
#
#
*
*
*
*
»
#
*
*
*
*
*
jt,
*
#
SELECTED DATA
AVG
72.64
79.72
79.37
90.99
90.58
48.79
48.57
71.17
70.85
89.17
75.97
75.71
64.50
74.69
63.63
80.03
68.18
72.02
61.36
81.55
69.48
87.27
74.35
65.54
55.84
44.70
46.13
51.15
ERR
26.48
25.71
25.31
26.02
25.58
13.77
13.53
25.84
25.47
17.41
10.20
18.38
11.73
17.58
11.09
15.95
9.44
27.41
19.61
14.04
7.73
19.06
11.71
27.39
19.93
10.90
12.90
17.91
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
#
#
*
#
#
*
*
*
*
*
*
*
*
*
*
*
#
*
*
*
*
#
*
*
#
*
*
CODES -
H20
SPL TYP COLUMN
TST/STD COLUMNS:
: P =
GA = GR
POLLUTED
WATER S
IDDED MEMBRANE ON
PA = PLAIN MEMBRANE
SP = SPREAD PLATE,
= SEWAGE
AGAR
ON AGAR
pp =
POUR PLATE
139
-------
Table I. Fecal Coliform Recovery Data (Cont'd.)
L
A
B
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
2
2
2
H20
SPL
TYP
P-4
P-4
P-4
P-4
P-4
P-4
P-4
P-4
P-4
P-4
P-4
P-4
P-5
P-5
P-5
P-5
P-5
P-j>
P-5
P-5
P-5
P-5
P-5
P-5
P-5
P-5
P-5
P-5
P-5
P-5
P-5
P-5
P-5
P-5
S-2
S-2
S-2
V
F
G
3
3
3
3
4
4
5
5
6
6
6
6
1
1
1
1
2
2
2
2
3
3
3
3
4
4
5
5
5
5
6
6
6
6
1
1
1
S
T
D
GA
GA
PA
PA
GA
GA
GA
GA
GA
GA
PA
PA
GA
GA
PA
PA
GA
CA
PA
PA
GA
GA
PA
PA
GA
GA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
T
S
T
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
S?
Dp
SP
PP
i,p
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
pp
SP
PP
*
*
*
*
*
*
#
*
*
*
*
*
*
*
*
#
*
*
*
*
*
*
*
*
#
*
*
*
*
*
#
*
*
#
*
*
#
*
*
*
RECOVERY -
ALL DATA
AVG
t>2.49
35.54
50.89
34.45
43.51
29.45
48.90
33.10
33.33
22.56
35.92
25.00
83.50
31.21
34.09
12.0?
24.52
5.64
115.70
40.81
125.28
44.18
111.11
39.18
87.73
30.94
102.29
36.08
109.19
38.51
76.24
26.89
62.45
22.02
lt>2.00
145.01
117.06
ERR
19.18
10.08
19.56
10.43
15.66
8.19
23.84
13.43
17.32
9.83
22. RO
13.39
25.21
11.47
25.33
9.93
1C*5<.
4.43
17.79
9.64
34.37
15.77
38.37
16.77
22.77
10.58
23.55
11.28
28.59
13.26
28.38
12.23
222.23
80.20
30.00
29.33
44.56
PERCENT OF STD #
* SELECTED DATA *
* AVG
* 62.86
*
# 60.95
*
* 52.11
*
* 54.68
*
* 39.92
#
* 50.79
*
*
*
*
*
#
1C
#
#
*
*
*
*
*
#
*
*
*
*
*
*
*
*
*
*
*
ERR *
13.23 *
*
13.98 »
*
10.68 *
*
17.52 *
*
14.56 *
*
12.36 *
*
*
«
*
*
*
if
#
*
*
#
*
*
#
*
*
*
*
*
*
*
*
*
*
*
*
CODES -
H20 SPL TYP COLUMN: P = POLLUTED WATER S = SEWAGE
TST/STD COLUMNS: GA = GRIDDED MEMBRANE ON AGAR
PA = PLAIN MEMBRANE ON AGAR
SP = SPREAD PLATE, PP = POUR PLATE
140
-------
Table I. Fecal Coliform Recovery Data (Cont'd.)
L
A
B
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
7.
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
H20
SPL
TYP
S-2
S-2
S-2
S-2
S-2
S-2
S-2
S-2
S-2
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-3
P-3
P-3
P-3
P-3
M
F
G
1
3
3
3
3
6
6
6
6
1
1
1
1
3
3
3
3
6
6
6
6
1
i
1
1
3
3
3
3
6
6
6
6
1
1
1
1
3
S
T
D
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
T
S
T
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
ft
RECOVERY -
ALL DATA
* AVG
#
ft
*
*
«
*
ft
ft
*
*
ft.
*
*
*
ft
«
ft
*
*
*
*
ft
«
»
*
ft
ft
ft
ft
ft
*
#
*
#
*
*
*
*
111
135
125
120
110
100
79
120
95
179
113
1*4
93
182
133
167
127
301
293
2*8
2*6
164
147
1*5
140
271
2*1
101
94
190
165
205
178
764
126
749
123
1275
.68
.59
.13
.05
.79
.69
.65
.25
.12
.60
.88
.97
.?&
.47
.57
.43
.15
,30
.96
.11
.10
.48
.22
.55
.12
.84
.74
.55
.04
.08
oc
.50
.39
.70
.21
.01
.62
.86
1
1
1
1
1
1
2
1
1
ERR
43.
21.
27.
40.
43.
41.
40.
28.
30.
88.
29.
22.
54.
11.
44.
90.
32.
52.
69.
07.
37.
84.
87.
02.
03.
02.
09.
80.
45.
95.
37.
31.
64.
433.
3
69.
69.
59.
922.
06
69
16
30
52
54
01
26
89
66
43
02
26
64
75
64
10
02
88
87
68
18
21
77
28
68
13
43
13
81
27
25
31
57
66
43
12
44
DE
*
*
ft
#
ft
ft
ft
#
*
*
*
ft
»
*
ft
ft
#
*
*
*
*
#
*
ft
#
#
*
ft
ft
ft
#
*
«
#
*
#
*
*
#
«
:E/NT OF STD *
SELECTED DATA *
AVG ERR *
*
«
*
#
#
*
*
«
*
*
*
*
*
*
*
*
#
*
*
*
ft
«
ft
*
*
ft
CODES -
H20 SPL TYP COLUMN: P = POLLUTED WATER S = SEWAGE
TST/STD COLUMNS: GA = GRIDDED MEMBRANE ON AGAR
PA = PLAIN MEMBRANE ON AGAR
SP = SPREAD PLATE, PP = POUR PLATE
141
-------
Table I. Fecal Coliform Recovery Data (Cont'd.)
L
A
B
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
H20
SPL
TYP
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-4
P-4
P-4
P-4
P-4
P-4
P-4
P-4
P-4
P-4
P-4
P-4
S-l
5-1
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-2
S-2
S-2
S-2
S-2
V
F
G
3
3
3
6
6
6
6
1
i_
i
i
3
3
3
3
6
6
6
6
1
1
2
2
3
3
4
4
5
5
6
6
1
1
2
2
3
S
T
D
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
PA
PA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
T
S
T
SP
PP
SP
PP
SP
PP
SP
pp
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
*
*
*
*
*
*
#
«
*
#
#
*
*
#
*
*
*
,t
H-
*
*
tt
*
*
w
*
»
R-
#
*
#
*
#
#
*
*
*
«
*
RECOVERY -
ALL
AVG
187
1117
164
1516
237
1530
239
391
lb8
117
47
447
169
840
318
375
248
349
231
83
92
95
105
96
107
DO
56
68
76
227
2t>2
92
S8
107
103
118
.81
.24
.46
.27
.09
.23
.27
86
.96
.44
.64
56
.90
.24
9fl
.18
.30
.63
i f°*
. 4w
.60
.72
.08
.45
.7?
.27
.81
.36
.85
.36
.86
.72
.07
.15
92
.31
.81
DATA
ERR
55
856
56
1196
115
1221
119
136
88
127
61
426
137
702
221
439
93
507
157
41
55
2S
43
29
^4
20
28
23
34
132
174
38
35
30
27
36
.91
.65
.15
.57
.82
86
.12
.18
.53
.21
.58
.09
.35
.96
.43
.85
.97
.80
.03
.21
.67
.70
.16
.91
.69
.17
.42
.28
.02
**6
.04
.19
.26
.57
.75
.56
PERCENT CF STD
* SELECTED DATA
* AVG ERR
*
#
*
#
*
*
*
*
* 113.26 30.49
#
*
* 196.55 52.80
* 113.43 55.21
*
*
*
*
#
*
*
«
*
*
#
#
*
*
«
#
*
*
* 92.07 38.19
* 88.15 35.26
* 107,92 30.57
* 103.31 27.75
* 118.51 36.56
*
*
#
*
*
*
*
*
#
*
*
*
#
#
*
#
*
*
*
#
*
*
*
#
*
*
*
*
*
*
«
*
*
#
#
#
*
*
#
CODES -
H20 SPL TYP COLUMN: P = POLLUTED WATER S = SEWAGE
TST/STD COLUMNS: GA = GRIDDED MEMBRANE ON AGAR
PA = PLAIN MEMBRANE ON AGAR
SP = SPREAD PLATE, PP = POUR PLATE
142
-------
Table I. Fecal Coliform Recovery Data (Cont'd.)
L
A
8
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
H20
SPL
TYP
S-2
S-2
S-2
S-2
S-2
S-2
S-2
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-i
P-l
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-3
P-3
P-3
P-3
P-3
P-3
P-3
M
F
G
3
4
4
5
5
6
6
1
1
2
2
3
3
4
4
5
5
6
6
1
1
2
2
3
3
4
4
5
5
6
6
1
1
2
2
3
3
4
S
T
D
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
T
S
T
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
pp
SP
pp
SP
pp
SP
pp
SP
pp
SP
pp
SP
pp
SP
pp
SP
pp
SP
pp
SP
pp
SP
p?
SP
?p
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
*
ft
ft
RECOVERY -
ALL
AVG
113.74
103.46
99.05
103.46
99.05
82.17
73.67
114.59
122.47
87.12
93.11
102.14
109.17
83.69
89.44
88.84
9^.95
109e 01
116.51
76.74
72.52
103.43
97.80
111.62
105.49
73.25
69.23
89.53
84.6,1
P.O. 23
75.82
107.65
161.06
106.63
Ib9. 5^*
119.89
179.33
79.08
DATA
ERR
33.33
50.23
46.63
50.22
46.63
31.84
29.32
39.13
45.54
32.22
37,27
31,33
36,80
32.92
37.90
37.90
43.40
2 6 i 9 2
43.06
17.20
19.20
26.05
28.59
22.03
25.11
19.52
21.26
13*19
15.91
22.23
24.09
45.04
24.36
52.29
26.12
64.6**
49.35
41.79
PERCENT OF STD *
*
*
ft
ft
#
*
#
*
ft
ft
ft
ft
#
ft
ft
a-
ft
ft
#
«.
ft
ft
ft
#
ft
ft
ft
ft
ft
ft
#
»
ft
»
#
*
*
#
#
*
SELEC
AVG
113.74
103.46
99.05
103.46
99.05
82.17
78.67
96.56
103.21
87.12
93.11
102.14
109,17
83.69
89.44
88.84
94.95
1 ri -a A o
110.09
99.18
107.65
161,06
106.63
159.54
110.96
166.03
79.08
TED DATA
ERR
33.33
50.23
46.63
50.23
46.63
31,84
29.32
14.28
18.40
32.22
37.27
31.33
36,80
32.92
37. 9C
37.90
43,40
2 4 ? 9
40.64
6.23
45.04
24.86
52.29
36.12
53.56
36.7^
41.79
ft
*
#
ft
#
ft
*
»
ft
«
«
«
#
ft
ft
ft
ft
*
#
JC
ft
ft
ft
*
*
ft
ft
ft
ft
ft
*
ft
*
ft
ft
ft
ft
ft
ft
ft
CODES -
H20 SPL TYP COLUMN:
TST/STD COLUMNS: GA
PA
SP
P = POLLUTED WATER S = SEWAGE
= GRIDDED MEMBRANE ON AGAR
= PLAIN MEMBRANE ON AGAR
= SPREAD PLATE, PP = POUR PLATE
143
-------
Table I. Fecal Coliform Recovery Data (Cont'd.)
L
A
B
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
H2C
SPL
TYP
P-3
P-3
P-3
P-3
P-3
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-l
5-1
S-l
S-2
S-2
S-2
S-2
S-2
S-2
S-2
S-2
S-2
S-2
S-2
S-2
P-l
p-1
P-l
P-l
P-l
P-l
P-l
M
F
G
4
5
5
6
6
1
1
2
2
3
3
4
4
5
C,
^
6
6
1
1
2
2
3
3
4
4
5
5
6
6
1
1
2
2
3
3
4
5
T
D
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
T
S
T
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
pp
SP
PP
SP
PP
SP
PP
*
*
*
*
ft
*
ft
*
*
ft
*
*
ft
ft
fr
«
ft
*
*
*
ft
ft
ft
ft
*
*
ft
*
ft
*
*
ft
*
*
*
*
tt
*
*
RECOVERY -
ALL
AVG
118.32
79.08
118.32
61.73
92.36
74.48
99.31
57.14
76.19
75.00
100.00
62.75
83.67
68.37
91.93
55.61
74.14
58.67
77.96
45.40
60.33
68.36
90. 34
i>0.00
66.44
51.53
68.^7
52.04
69.15
94.09
85.35
12.42
11.26
102.48
92.95
105.59
DATA
ERR
31.29
41.79
31.29
34.87
27.79
37.93
36.47
29.60
28.64
33.30
30.23
3?. 76
31. 50
30.95
28.23
32.50
32.80
34.26
30.30
27.24
24.41
32.16
24.99
25.91
21.46
25.46
20.46
23.43
17.63
28.82
34.98
13.77
13.66
35. ?5
42.15
27.72
PERCENT OF STO
*
*
*
*
*
#
#
»
#
#
ft
*
*
*
*
ft
ft
*
it
#
#
#
*
»
#
»
ft
ft
#
»
#
ft
#
#
#
»
*
#
SELECTED
AVG
118.32
79.08
118.32
66.32
99.23
74.48
99.31
60.58
80.78
75.00
100.00
66.96
69.28
68,87
91.33
60. 5R
80. 7S
68.02
90.29
58.67
77.96
68.36
90.84
56.12
74.57
59.94
79.66
58.67
77.96
81.52
83.33
85.40
87.30
DATA
ERR
31.29
41.79
31.29
33.63
24.12
37.93
36.47
28.56
26.61
33.30
30.20
30.68
28.22
30.95
28.23
29.98
28.50
32.26
25.21
19.49
10.68
32.16
24.99
21.20
13.60
25.33
18.10
19.49
10.68
27.41
23.43
18.16
13.75
ft
*
#
»
ft
*
ft
#
ft
»
#
ft
#
ft
*
»
ft
#
»
«
*
*
#
ft
»
»
ft
*
#
«
ft
»
*
»
»
#
#
*
«
CODES -
H20 SPL TYP COLUMN:
TST/STD COLUMNS: GA
PA
SP
P = POLLUTED WATER S = SEWAGE
= GRIDDED MEMBRANE ON AGAR
= PLAIN MEMBRANE ON AGAR
= SPREAD PLATE, PP = POUR PLATE
144
-------
Table I. Fecal Coliform Recovery Data (Cont'd.)
L
A
B
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
H20
SPL
TYP
P-l
P-l
P-l
P-l
P-l
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-4
P-4
P-4
P-4
P-4
P-4
0.4
P-4
P-4
V
F
G
4
5
5
6
6
1
1
2
2
3
3
4
4
R
5
6
6
1
i
2
2
3
3
4
4
5
5
6
6
1
1
2
2
3
3
4
4
5
5
^
D
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
T
S
T
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SD
PP
SP
PP
SP
= p
SP
PP
SP
pp
SP
PP
*
*
*
*
*
*
*
*
*
*
*
#
*
&
*
*
*
*
a
a-
#
*
*
*
#
#
*
*
#
*
*
#
*
*
«
*
*
#
*
*
*
RECOVERY -
ALL
AVG
95.77
65.52
i>9.43
66.14
60.00
126.5*
40.28
139.82
44. 5C
114.15
36.33
100.00
31.83
96.46
30.70
112.38
35.77
144.73
46.47
111.40
35.77
142.10
45.63
86.84
27.88
121.05
38.87
122.80
39.43
214.03
lt»4.43
142.10
102.53
317.54
229.11
223.07
164.55
238.59
DATA
ERR
35.06
16.91
21.49
22.05
26.21
68.27
25.65
50.95
20.55
31.77
13.65
29.48
12.48
35.27
14.22
39.22
15.97
44.96
20.39
60.65
24.06
42.11
19.37
25.52
11.77
30.26
14.70
58.72
23.91
58.72
110.42
55.25
85.05
100.36
173.38
56.72
113.45
72.50
PERCENT OF STO
*
*
*
*
*
#
*
*
*
#
*
*
*
*
*
*
*
*
*
#
*
*
*
*
*
*
#
*
*
#
*
*
*
*
*
*
*
*
*
*
SELEC
AVG
65.52
66.98
66.14
67.61
134.04
50.00
134.46
50.15
109.78
40.95
100.00
37,30
103.54
33.62
114.89
42.85
137.50
52.38
129,16
49.20
135.00
51,42
91,66
34,92
115,00
43.80
131.25
50.00
203.33
112.61
135.00
74.76
216.66
120.00
216.66
TED DATA
ERR
16.91
13.59
22.05
13.81
62.56
20.77
47.65
15.20
29.45
8.83
25.86
7.73
31.23
9.67
32.39
9.89
34.39
14.67
50.00
20.52
31.84
13.67
11.13
5.29
21.79
9.61
36*44
15.38
43.48
47.27
44.32
39.94
40.77
47.29
51.83
*
*
«
*
*
*
*
*
*
#
*
*
*
«
»
*
#
#
*
w-
*
*
*
#
#
*
#
*
*
#
*
#
*
*
#
#
*
*
*
*
*
CODES -
H20 SPL TYP COLUMN:
TST/STD COLUMNS: GA
PA
SP
P = POLLUTED WATER S = SEWAGE
= GRIDDED MEMBRANE ON AGAR
= PLAIN MEMBRANE ON AGAR
= SPREAD PLATE, PP = POUR PLATE
145
-------
Table I. Fecal Coliform Recovery Data (Cont'd.)
L
A
B
4
4
4
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
H20
SPL
TYP
P-4
P-4
P-4
S-l
S-l
S-l
S-l
S-l
S-l
S-l
S-?
S-2
S-2
S-2
S ?
S-2
P-l
P-l
r>- 1
P-l
P-l
P-l
P-2
P-2
P-2
P-2
P-2
P-2
P-3
P-3
P-3
P-3
P-3
P-3
M
F
G
5
6
6
1
1
2
3
4
5
6
1
2
3
A
tj
6
1
2
3
A
5
6
1
2
3
A
5
6
1
2
3
A
5
6
S
T
D
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
T
S
T
SP
PP
SP
PP
SP
SP
SP
SP
SP
SP
PP
PP
PP
PP
PP
p?
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
SP
s°
SP
SP
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
*
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
RECOVERY -
ALL
AVG
172.15
289.47
208.86
763.15
100.11
76.00
131.46
66.51
9A.76
122.90
71.31
5A.28
SB, 16
35.15
1 n * e r, c,
81./>5
156. Al
159.85
211.12
191.78
259.50
297.80
129.23
158. A7
196.61
133.24
186.77
157.89
98.13
10A.62
136.71
107.32
103.02
152.77
DATA
ERR
128.17
75.17
1A6.28
205.21
A0.85
23. 1A
A9.65
A6.00
18.08
40,18
24.42
20.42
23*01
27.27
32s 00
41 .48
37. 3A
113.77
115.59
84.91
155.43
100.47
70.02
29.71
78.66
28.25
69.31
56.53
33.16
53.62
47.19
22.63
32.60
6A.61
PERCENT OF S
#
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
#
*
ft
*
«.
it
ft
ft
ft
*
«
*
*
ft
»
*
ft
ft
#
*
ft
*
*
ft
SELECTED
AVG
120.00
73.68
116.77
158. A7
161. ?6
133.24
172.19
147.36
98.13
92.59
119.16
107.32
108.02
136.38
TD
DATA
ERR
53.42
1'A.IC
53.49
29.71
33.^0
28.25
53.43
^3.50
33.16
40.43
29.45
22.63
32.60
30.07
ft
ft
*
*
*
*
*
ft
ft
ft
ft
«
ft
«
«
«
ft
«.
*
»
ft
*
*
#
#
*
«
*
«
ft
*
ft
ft
«
#
«
*
CODES -
H20 SPL TYP COLUMN:
TST/STD COLUMNS: GA
PA
SP
P = POLLUTED WATER S = SEWAGE
= GRIDDED MEMBRANE ON AGAR
= PLAIN MEMBRANE ON AGAR
= SPREAD PLATE, PP = POUR PLATE
146
-------
Table I. Fecal Coliform Recovery Data (Cont'd)
L
A
B
5
5
5
5
5
5
6
6
6
6
6
6
6
6
&
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
7
7
7
H20
SPL
TYP
P-4
P-4
P-4
P-4
P-4
p-4
S-2
S-2
S-2
S-2
S-2
S-2
Sr2
S-2
S-2
S-2
S-2
S-2
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
S-2
S-2
S-2
M
F
G
1
2
3
4
5
6
1
1
2
2
3
3
4
4
5
5
6
6
1
1
2
2
^
3
4
4
5
5
6
6
1
2
3
5
T
D
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
G A
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
T
S
T
PP
PP
PP
PP
PP
PP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
PP
SP
DP
SP
PP
SP
PP
SP
pp
SP
pp
PP
pp
*
*
#
*
*
*
#
*
*
*
#
*
*
*
*
*
*
1C
*
*
*
#
*
#
*
#
*
*
*
#
*
*
*
*
*
*
RECOVERY -
ALL
AVG
78.49
80.89
108.89
87.09
102.51
100.74
97.68
106.28
80.30
53.80
128. P7
95.00
78.12
75.75
9 s * 0 3
82.67
126.31
124.85
81.84
92.11
4.89
4.37
121.44
111.71
b8. 22
66.02
65.33
89.13
131.74
2bl.29
86.07
5.52
92.65
DATA
ERR
11.91
22.11
29.63
23.68
24.85
17.06
38.12
43.17
40.20
39.63
59.05
46.72
46.85
52.36
30.92
26.92
53.47
94.07
83.86
133.94
11.65
12.00
91 .54
126.30
22. *7
34.48
39.53
93.34
64.40
197.60
10.30
1.86
10.30
PERCE.N
T OF
STD
* SELECTED DATA
*
* 78
* SO
* 108
* 87
* 102
* 100
* 97
* 106
* 97
* 69
* 120
* 95
* 96
* 93
* 9S
* 82
* 126
* 111
*
*
*
*
*
*
* 53
* 66
* 66
* 76
#
#
*
*
AVG
.49
.89
.39
.09
.51
.74
.68
.28
.34
.79
.90
.00
.35
.43
*03
.87
.31
.48
.22
.02
.93
.71
ERR
11.91
22.11
29.63
23.68
24.85
17.06
33.12
43.17
24.57
39.00
49.96
46.72
20.21
23.15
3 0 9 2
26.92
53.47
74.62
22.87
34.46
11.86
30.56
#
*
#
*
*
#
*
*
*
#
#
*
*
*
*
*
*
A
*
*
#
*
#
#
#
*
*
*
*
*
*
*
*
*
#
*
CODES -
H20 SPL TYP COLUMN:
TST/STD COLUMNS:
P = POLLUTED WATER S = SEWAGE
GA = GRIDDED MEMBRANE ON AGAR
PA = PLAIN MEMBRANE ON AGAR
SP = SPREAD PLATE, PP = POUR PLATE
147
-------
Table I. Fecal Coliform Recovery Data (Cont'd.)
L
A
R
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
9
9
9
9
9
9
9
9
9
H20
SPL
TYP
S-2
S-2
S-2
P-l
P-l
P-l
P-l
P-l
P-l
P-2
P-2
P-2
P-2
P-2
P-3
P-3
P-3
P-3
P-3
P-3
P-4
P-4
P-4
P-4
P-4
S-2
S-2
S-2
S-2
S-2
S-2
S-2
S-2
S-2
M
F
G
4
5
6
1
2
3
4
5
6
1
2
4
5
6
1
2
2
4
5
6
1
3
4
5
6
1
1
2
2
4
4
5
5
6
S
T
D
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
T
S
T
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
SP
PP
SP
op
SP
PP
SP
PP
*
*
*
ft
*
*
*
*
*
ft
*
ft
*
*
*
*
*
ft
ft
#
ft
»
*
ft
»
ft
*
*
*
*
ft
*
ft
ft
*
*
*
RECOVERY -
ALL
AVG
71.13
75.44
91.13
94.63
0.56
90.11
77.96
88.13
85.87
94.51
85.65
72.15
75.94
77.21
89.14
1.31
90.78
62.50
77.96
79.27
89.04
90.41
73.97
85.61
86.30
238.35
84.26
245.20
86.68
272.60
96.36
2t>4.10
89. S3
236.30
DATA
ERR
12.16
8.15
12.63
8.49
0.80
12.00
9.22
11.97
13.69
12.10
10.21
10.70
11.92
17.13
19.08
2.30
22.65
19. 5C
14.92
13.48
18.05
18.99
18.98
21.23
19.65
29.27
27.45
46*^4
34.72
44.06
35.14
57.27
38.45
50.17
PERCENT OF STD
«
#
ft
*
ft
ft
*
ft
»
#
ft
#
ft
ft
ft
ft
*
ft
ii-
fc
ft
#
ft
ft
ft
#
li-
ft
ft
ft
#
#
*
#
#
ft
SELECTED
AVG
63.67
73.83
84.74
77.96
81.92
76.97
94.51
85.65
72.15
75.94
77.21
86.34
86.75
62.50
77.96
79.27
89.0^
90.41
77.91
8 5-. 61
86.30
DATA
ERR
9.92
7.34
5.24
9.22
7.06
5.76
12.10
10.21
10.70
11.9?
17.13
16.77
19.09
19.50
14.92
13.48
18.05
18.99
16.43
21.23
19.65
#
ft
ft
ft
*
*
ft
*
#
*
*
*
ft
ft
»
*
#
#
«
*
#
»
ft
ft
«
ft
»
ft
ft
*
#
»
«
*
»
*
ft
CODES -
H20 SPL TYP COLUMN: P = POLLUTED WATER S = SEWAGE
TST/STD COLUMNS: GA = GRIDDED MEMBRANE ON AGAR
PA = PLAIN MEMBRANE ON AGAR
SP = SPREAD PLATE, PP = POUR PLATE
148
-------
Table I. Fecal Coliform Recovery Data (Cont'd.)
L
A
B
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
H20
SPL
TYP
S-2
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-l
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-2
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-3
P-4
P-4
P-4
P-4
P-4
P-4
M
F
G
6
1
1
2
2
4
4
5
5
6
6
1
1
2
2
4
4
5
5
6
6
1
1
2
2
4
4
5
5
6
6
1
1
2
2
4
S
T
0
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
4 GA
T
S
T
SP
PP
SP
PP
SP
pp
SP
pp
SP
pp
SP
pp
SP
pp
SP
pp
SP
pp
SP
pp
SP
pp
SP
pp
SP
pp
SP
pp
SP
pp
SP
pp
SP
po
SP
pp
SP
*
RECOVERY - PERCENT OF STD
* ALL DATA
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
#
*
#
*
*
»
*
*
*
*
*
*
*
*
*
*
#
*
*
*
*
AVG
83.53
189.72
78.02
188.35
77,46
193.83
79.71
202.05
83.09
195. 89
80.56
200.00
73.73
20s. 21
76.76
197.26
72.72
219.17
80.80
215.75
79.54
186.98
80.29
191.09
82.05
205.47
88.23
206.16
88.52
210.27
90.29
204.79
77.86
199.31
75. 7B
195.89
74.47
ERR
34.69
47.87
22.29
34.90
16.93
43.50
20.54
47.15
22.16
32.91
16.22
38.74
32.00
45.53
35.23
49.44
35.70
42.25
34.99
41.61
34.45
35.59
25.83
31.01
24.14
37.18
27.61
41.71
29.59
52.59
34.49
53.12
19.93
42.51
15.90
33.35
12,42
* SELECTED DATA
* AVG
*
* 171.23
* 70.42
* 183.21
* 75.35
* 178.08
* 73.23
* 172.94
* 71.12
* 187.21
* 76.99
* 188.35
* 79.42
* 184.93
* 77.97
* 187.50
* 79.06
* 202.05
* 85.19
* 195.20
* 82.31
* 180.65
* 77.57
* 191.09
* 82.05
* 188.35
* 80.88
* 184.93
* 79.41
* 189.49
* 81,37
* 132.64
* 69.44
* 1P.6.07
* 7C.7<*
* 191.73
* 72.91
36
17
31
15
29
14
19
10
26
13
29
24
38
28
41
29
25
23
29
25
28
22
31
24
19
18
28
22
29
23
32
12
36
13
30
11
ERR
.36
.30
.16
.33
.85
.71
.89
.55
.31
.59
.30
.66
.05
.13
59
.91
.25
.85
.40
.15
.47
.46
.01
.14
.02
.84
.36
.66
.60
.45
.29
.04
.03
.45
.90
.50
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
#
*
*
«
*
*
*
*
*
*
*
#
*
*
*
*
#
*
*
*
*
*
*
#
*
CODES -
H2
0
SPL TYP COLUMN: P = POLLUTED
TST/STD COLUMNS:
GA = GRI
PA = PLA
WATER S
DDED MEMBRANE ON
= SEWAGE
AGAR
IN MEMBRANE ON AGAR
SP = SPREAD PLATE, PP =
POUR
PLATE
149
-------
Table I. Fecal Co I i form Recovery Data (Cont'd.)
L
A
B
9
9
9
9
10
10
10
10
10
10
10
10
10
10
10
10
10
H20
SPL
TYP
P-4
P-4
P-4
P-4
S-2
S-2
S-?
S-2
S-2
S-2
S-2
P-l
P-l
P-l
P-l
P-l
P-l
M
F
G
5
5
6
6
1
1
2
3
4
5
6
1
2
2
4
5
6
S
T
D
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
*
S
T
PP
SP
PP
SP
PP
SP
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
PP
*
#
*
#
»
*
*
*
*
*
#
*
#
*
«
#
*
*
ir
#
RECOVERY -
ALL
AVG
191.78
72.91
180.13
68.43
121.63
54.16
131.49
130.40
111.69
103.50
143.09
li>2.55
107.82
128.94
135.00
i:>2.72
ib6.57
DATA
ERR
45.45
17.03
37.04
13.85
44.90
15.63
53.12
41.07
34.16
37.12
36.19
43.56
68.74
42. 12
55.00
50.62
39.49
PERCENT OF STD
#
*
*
*
*
*
*
«
*
#
*
*
#
#
#
#
#
*
*
SELEC
AVG
183.21
69.66
172.94
65.75
121.63
54.16
118.09
130.40
111.69
103.50
143.09
152.55
133.63
TED DATA
ERR
38.96
14.57
28.82
10.73
44.90
15. 6S
25.27
41.07
34.16
37.12
36.19
43.56
26.07
*
*
*
*
#
*
*
*
*
#
*
*
#
#
*
*
*
*
#
*
CODES -
H20 SPL TYP COLUMN: P = POLLUTED WATER S = SEWAGE
TST/STD COLUMNS: GA = GRIDDED MEMBRANE ON AGAR
PA = PLAIN MEMBRANE ON AGAR
SP = SPREAD PLATE, PP = POUR PLATE
150
-------
Table 2 ANOVA TABLE - SPREAD PLATE STANDARD
Source
Sum of Squares
DF
Mean Squares
Filters (F)
Samples (S)
FS
Labs (L)
FL
SL
Residual
Total
10257.6269
22436.6289
9214.7500
39769.3829
7386.6250
74822.8908
32624.6289
196512.5316
5
4
20
2
10
8
40
89
2051-5253
5609.1572
460.7375
19884.6914
738.6625
9352.8613
815.6157
2.51
6.87
0.56
24.37
0.90
11.46
Significant - 97
#
*
.5%
Significant - 99+%
Not Significant
#
Meaningless - since they refer to differences attributed to samples which we know are different.
Table 3 ANOVA TABLE - POUR PLATE STANDARD
Source
Sum of Squares
DF
Mean Squares
Filters (F)
Samples (S)
FS
Labs (L)
FL
SL
Residual
Total
14464.0000
21314.5000
10630.3769
47449.2579
8669.1269
104435.5158
37086.0000
244048.7816
5
4
20
2
10
8
40
89
2892.8002
5328.6250
531.5188
23724.6289
866.9125
13054.4394
927.1500
3.12
5.74
0.57
25.58
0.93
14.08
Significant - 97
*
*
.5%
Significant - 99+%
Not Significant
*
Meaningless - since they refer to differences attributed to samples which we know are different.
151
-------
Table 4 COMPONENTS OF VARIANCE Table 5 COMPONENTS OF VARIANCE
ANALYSIS - SPREAD PLATE ANALYSIS - POUR PLATE
STANDARD STANDARD
02 Residual = 7386.625 + 40011.245 = 800 a2 Residual = 8669.+ 37086 = 915
50 50
02 I OK = 19885 - a2 Residual = RQC o
L8b b b a2 Lab - 23725 - a2 Residual = 760
30
a2 Filters = 2052 - a2 Residual = 83
15 a2 Fj|ters = 2893 - a2 Residual = 132
15
a2 For a single determination on any filter
in any laboratory is:
a2 for a single determination on any filter
in any laboratory is:
a2 = a2 Residual + a2 Lab + a2 Filter =
800 + 636 + 83 1519 a2 = a2 Residual + a2 Lab + a2 Filter =
a = ± 39 2a = ± 78 915 + 76° + 132 = 18°7
a = ± 42.5 2a = ± 85
GRAND MEAN = 81.5
GRAND MEAN = 95.5
152
-------
CRITIQUE ON ASTM TEST FOR RECOVERY OF FECAL COLIFORMS AND
PROPOSAL FOR MODIFIED METHOD
Norman H. Goddard
Sartorius Membrafilter GmbH
ABSTRACT
As both a manufacturer of membrane filters
and as a participating laboratory, in the ASTM
Collaborative Study on MF's, Sartorius (Goddard),
made the following suggestions for change in the
study plan:
1. Conduct the study with a flora more favor-
able than polluted waters in sewage. For
example, use a potable water spiked with a
standard pure culture.
2. Standardize sample mixing conditions.
3. Require separate counts of coliforms and
non-coliforms.
4. Randomize sample positions in the incuba-
tion.
5. Verify accuracy of thermometer and incuba-
tor settings.
The requirements of a routine method for the
testing for the recovery of fecal coliforms include
the following:
Agreement must be obtainable with a compar-
ison method such as the pour plate method.
The results must be reproducible irrespective
of the location of the testing laboratory or of the
person performing the test.
Evaluation must be clear, without the possi-
bility of subjective errors, so that the culture
medium must not only be selective but also allow
simple recognition of the resulting colonies.
The method itself must be simple and practi-
cal, capable of use in any testing laboratory.
As the leading European manufacturer of
membrane filters, we welcome this attempt by the
American Society for Testing and Materials and the
United States Environmental Protection Agency to
standardize a method for testing the suitability of
membrane filters for fecal coliform determinations,
and for comparing different brands of membrane
filters. We were therefore very pleased to be invited
to participate actively in this work and to carry out
the round robin comparison tests on various
filters. Such a filter comparison requires that all
influences which can affect the result, excepting
the filters, be removed. Unfortunately, we believe
that this was not the case with the procedure
used, and that modifications are required for
meaningful results to be obtained.
One point of interest which arose from our
tests was, however, that there appeared to be some
correlation between the flow rate through a
filter under standard conditions and the efficiency
of recovery of that filter. The faster the flow rate,
the higher the efficiency.
We heard yesterday that higher recoveries can
be obtained by increasing the average surface open-
ing diameter of the filter to an optimum size. Now
the surface opening diameter is directly related to
the pore size. The former is measured by electron
microscopy, the latter is characterized by the
mercury intrusion method, by retention charac-
teristics or by flow rate measurements. In general,
the faster the flow rate, the larger the pore size,
and the larger the average surface opening diame-
ter.
It was also stated yesterday that a reason for
the higher recovery with a larger pore size filter
may be the better provision of nutrient to a micro-
organism sitting in a surface pocket compared to
one perched on the top of the filter and exposed
to different evaporation effects.
153
-------
The structure of a membrane filter is uniform,
the pore size being the same throughout the width
of the filter. Another possibility is therefore that
the use of a larger pore size filter simply results in
an easier passage of the nutrient from the bottom
to the top surface of the filter, thus favoring
growth.
Our experiments in this direction are not yet
completed, but it could be that an optimum sur-
face opening diameter, or pore size, for the re-
covery of fecal coliforms can be decided upon, and
that comparisons between competitive filters need
then be made only on the basis of the measure-
ment of physical parameters: the sizes, and the
reproducibility of the sizes, of the surface openings
or pores on each side of the filter, and the flow
rate through the filter under defined conditions.
This is purely conjecture at present, and I will
therefore return to the test under discussion.
We are not happy with the samples suggested,
polluted water and raw sewage, as a number of E.
coli and coliform types can occur which affect the
reported recovery because of subjective evaluation.
Different investigators may not agree as to which
colonies are fecal coliforms.
Additionally, such samples can contain high
concentrations of sub-lethally injured micro-
organisms, and there is a large variation in the
samples taken at the different locations of the
investigating laboratories.
The culture medium, M-FC agar, is apparently
not sufficiently selective. Not only coliform
colonies but also numerous other colonies are
obtained, which may or may not work as antago-
nists, either by affecting the growth of the bacteria
we are searching for, or by changing the pH of the
surrounding nutrient by alkali formation and so
affecting blue color formation.
We filtered lake water through a filter and
cultivated as usual. We then selected four of the
blue colored colonies and three different non-blue
colonies and subcultured them.
We streaked the blue colony culture on each
of three different M-FC agar plates and streaked
the culture of the non-blue colonies vertically to
the blue colonies, one on each of the three plates.
On incubating, differing effects of the non-
blue on the blue colonies could be seen. This shows
that according to the type and the amount of
alkali producing bacteria which are present, the
typical blue color of the coliforms may be weak-
ened or even completely prevented from occuring.
Fewer non-col if orms appear to grow on Endo
media than on M-FC and strong alkali producers
appear to have less effect.
Returning to the present procedure, the
length of incubation using M-FC agar appears to be
insufficient, as we have often noted blue color
developing after completion of the standard incu-
bation time. The incubation temperature of
44.5 C. is difficult to maintain in routine use.
With regard to the reference method, using
samples containing high concentrations of antago-
nists, their effect depends on the contact area with
the nutrient. The relationship of sample volume to
contact area should be the same with the filter
and the reference method, i.e. smear plates should
be of the same diameter as the filter diameter used.
Pour plates of the same diameter have a far larger
contact area, and conditions within the culture
medium may affect different antagonists in differ-
ent ways.
The sample suggested, polluted waters, con-
tain bacterial agglomerates. Shaking will dispense
them to an extent, depending on the amount and
method of shaking. Some aggregate break-up
occurs in the pour plate method on mixing the
sample with the warm viscous agar.
We would therefore suggest that a sample be
used which contains favorable bacteria flora for
this test. We recommend one in which the numer-
ous types of protein decomposing, alkali forming
bacteria occurring in raw sewage do not occur, so
that the difference in contact area between test
and reference method is not important.
Such a sample could be potable water with
added pure culture.
The conditions of shaking the sample are less
important with potable water than with polluted
water but should nevertheless be standardized.
M-FC agar would be suitable as culture med-
ium for this sample, whereas with a sample con-
taining a bacteria spectrum approaching that of
raw sewage a more specific culture medium is re-
quired. Regardless of which culture medium is
used, the total number of non-coliform colonies
154
-------
occuring should be stated in addition to the
coliform count.
Again, using such a sample, the incubation
time of 22-24 hours would be sufficient and dif-
ficulty in evaluation would not occur.
Streaked spread plates of the same diameter
as the filter would be optimal, but not practical be-
cause of the small sample volume which can be
spread. Pour plates in 60 mm. Petri dishes should
therefore be used with a good relationship between
sample volume and contact area.
Randomization of sample positions during
incubation is necessary, as is a check on the accur-
acy of the thermometer used.
QUESTIONS AND ANSWERS
Geldreich: Was rosolic acid added to the medium
which you used?
Goddard: (I don't really know.) I must apolo-
gize, I am not a microbiologist and if
I were timid, I would not have read
this technical paper, but we did want
to comment on comparisons of mem-
brane filters.
Geldreich: I would like to know if rosolic acid
was added, to cut down the back-
ground of other organisms?
Bordner: The method carried out is exactly as
in the ASTM draft procedure given us
and the draft procedure did include
the addition of rosolic acid. Any other
questions related to this paper?
Brezenski: One of the criteria established in the
comparison of the replicate membrane
filters was an 85% recovery. Why was
this number chosen - what is the basis
for it? I just want to know what the
reference is, that's all.
Bordner: This was just a goal that was set
arbitrarily by the subcommittee. I
think the percentage was from the De-
partment of Defense specifications for
membrane filters. That percentage
was also quoted for previous recovery
tests that were identified in the DOD
specs that we discussed yesterday.
Any other questions?
Litsky: I cannot leave this room without
asking the question I asked yesterday
morning, and that is the problem of
the extractables. We have perfectly
good procedures on extractables. If
the specifications require changing,
let's try to get a uniform product. If
the definition needs changing, let's
change it. But its rather ironic that
three months ago I sat in an EPA
committee concerning disinfectants
and detergents, and they stuck to the
letter of the law! Any deviation from
the specifications caused a riot. If
the specs written ten years ago do not
apply now, let's change them.
Sims: During the last two days we have seen
a dilemma develop for the manufac-
turer. We are responsive to you, the
customer, and we can provide you
with whatever you want. The repre-
sentative from Millipore pointed out
this morning that if you create this
surface - what they are calling greater
surface pore size - you create a differ-
ent atmosphere. It can be provided.
Also, cells in the surface cavity will
have a tendency to run and colonies
won't look the same. In purchasing
membranes you are going to have to
be more particular about what results
you require and not as particular
about cell appearance.
The specs were written ten years ago
and our membrane was developed to
meet these specs, whether correct or
not, it is the law. If you are selling to
the US Army, they buy by those
specs. You live with extractables.
Everybody does the same thing. If
somebody adds more extractables and
they can justify it and give you a
product, fine. If you want zero
extractables, the membrane can be
washed. It will be brittle and it might
have usage effects that would really
turn you off. You, the customer, are
deciding what you want to buy. The
manufacturer will meet your de-
mands. What we want from this
symposium is to find out what your
demands are.
Right now, from the attitudes I have
seen, you want specific microbiologi-
155
-------
cal properties. You will accept, per-
haps, not as even a gridline, not as
dark a print. You will take perhaps a
little more brittle product or a prod-
uct packaged in a different way. May-
be you want to autoclave the mem-
brane and it wrinkles, maybe it
doesn't. I think what we have been
discussing is a two-way street. I am
looking forward to getting the re-
sults of the committee meeting over
the next few days and as manufactur-
ers we will provide what you want.
The thing that did interest me about
this surface phenomenon is that
membranes can be manufactured to
create a very tight system, a uniform
pore throughout where bacteria sit
right on the top, fluid has to nuture
the cell from both sides and the cell
grows into a little round colony. If
you have a sponge-like network the
cell is down in it, the fluid surrounds
Bordner:
it almost on four sides, but always on
three, the colony growth might spread
out differently, depending on the
structure. If you are affected drasti-
cally by colony appearance and how
easy it is for your technician to count,
you might get lower counts. It's
something, I think, that is very vital.
The manufacturer can provide you
with either structure - its no problem.
The only other question I have for the
representative from Millipore concerns
the 0.7 Aim filter. Will Serratia mar-
cescens still be the species used to
standardize filters? This is what the
specs now read and as a manufacturer
I live by these specs. Are we going to
vary the way we control retention?
We hope to provide some of these
answers in the ASTM committee
meeting.
156
-------
SUMMARY OF SYMPOSIUM
John A. Winter, Chief, Quality Assurance Branch
Environmental Monitoring and Support Laboratory, Cincinnati, EPA
and
Francis T. Brezenski, Chief, Technical Support Branch
Edison Water Quality Laboratory, Edison, EPA
SUMMARY AND DISCUSSION
Bordner (Moderator) This morning the
agenda is shorter than yesterday. We plan to end
the Symposium by noon. It will be followed this
afternoon by a meeting of ASTM subcommittee
D19.08.04.02. You are welcome to stay this after-
noon and participate. You are also encouraged to
join the subcommittee as permanent members.
The last item on the program this morning is
a summarization of the papers in the Symposium.
It will be given by Fran Brezenski and John Winter,
both microbiologists in EPA.
The papers are placed into four categories,
however, the audience should be aware that there
is a considerable overlap of subject matter be-
tween categories.
I. General papers.
II. Stressed/injured cells and observed ef-
fects on MF recoveries.
Comparisons between
brands.
MF lots and
IV. Solutions to apparent low recoveries.
General Papers
1) Bordner - The Membrane Filter Dilemma
Bordner described the general preference
in water analyses for MF techniques based
on simplicity and speed. Increasingly, how-
ever, the literature has reported low or vari-
able MF results, particularly for fecal coliform
bacteria and has attributed these results to
brand or lot differences in MFs, stressing of
bacteria in natural waters, inhibitive incuba-
tion at 44.5 C, chlorination, and other test
factors.
ASTM Committee D-19 and the U.S.
EPA sponsored this seminar to: disseminate
known information on the problem, identify
the better possible solutions, establish the
standard test procedures needed for validly
comparing media, membrane filters, and
other test factors, and select the best solution
using these standard tests.
2) Geldreich - Performance Variability of
Membrane Filter Procedures
The variable results reported from mem-
brane filter procedures are ascribed not only
to differences in membrane filter materials
and methods of MF sterilization but also to
inconsistencies in absorbent pads, commer-
cially-prepared media, and the knowledge and
experience of the technician. The need for
cooperation between manufacturers and users
and the need for improved quality control
programs by both were emphasized.
3) Powers - Quality Control of Media
Powers reviewed the manufacturing
techniques and the quality control techniques
practiced by the manufacturer. He suggested
that the manufacturer loses control over
product quality once it leaves his plant. He
then described the abuses in laboratory test
conditions which influence microbial recover-
ies, growth and colonial characteristics and
stated that most media problems result from
mishandling of media in the laboratory. This
157
-------
attitude moves final QC responsibility from
the manufacturer to the user.
4) Sladek - Statistical Interpretation of MF
Counts
Sladek cited the importance of statistical
design in any comparison of microbial re-
coveries and reviewed the main factors to be
measured in such studies.
He used examples of actual studies to
describe measures of precision and to demon-
strate the importance of using large numbers
of replicates. He cited the problems of stabi-
lizing samples for true replication over time.
Sladek emphasized that the theoretical
minimum deviation inherent in the method
must be established to compare with subse-
quent experimental deviations. He stressed
that valid studies can only be made by re-
moval or control through randomization of
unwanted variables which include sampling,
types of samples, procedures, materials and
media.
Finally, Sladek recommended the use
of a control which is independent of the test
methods, i.e., non-MF control for comparing
MF tests.
II. Injury Papers/Effect Papers
1) Hoadley - Effects of Injury on Recovery
of Indicators on Membrane Filters
Hoadley studied the recovery of E. coli
and S. faecalis as control cultures and as cul-
tures stressed by exposure to stream condi-
tions and to chlorination. He used spread
plate and membrane filter tests on selective
and non-selective media. Control cultures of
S. faecalis showed low recovery on selective
media while E. coli controls recovered well.
After stressing, E. coli yielded low re-
covery on MFs, while S. faecalis showed good
recovery. E. coli counts on MFs were reduced
with increased stress.
Hoadley concludes S. faecalis may be a
better indicator than E. coli because it does
not show low recovery after stress. Further,
he recommends that any studies for improved
methodology should include evaluation of
stressed cell populations.
2) Hufham - Effects of Temperature on
Recovery of Fecal Coliform
Hufham cites extreme effects of tem-
perature on recovery of fecal coliforms, sug-
gesting that these high temperature effects
(44.5 C) can overlap effects of membrane
differences. Temperature may have been a
major factor in earlier papers citing severe
differences in membranes.
For future recovery studies, Hufham
proposes the use of a standard E. coli culture
which has not been selected from MF cultures
at 44.5 C. He proposes that a standard plate
count of cells grown at 35 C is necessary to
avoid temperature-related losses and to estab-
lish the reference cell numbers for MFs or
other comparisons.
III. Comparison Papers
1) Brodsky - A Comparison of Membrane
Filters and Media Used to Recover
Coliforms from Water
Two to seven laboratories analyzed
water samples and cultures for coliform bac-
teria using membrane filters from Johns-
Manville of Canada, Millipore Corporation
and the Sartorius Company.
Cultures were both routine water
samples and total and fecal coliform mixed
cultures obtained by passage through Mac-
Conkey broth and EC broth. Analyses were
performed in parallel using LES Endo Agar
and M-Endo agar.
Although the Johns-Manville and Milli-
pore filters showed higher recoveries than
Sartorius in Phase I of the study, Phase II
results showed the three filters to be equal.
Recoveries on LES Endo agar and M-Endo
agar were similar. The authors reported that
their results varied with the test conditions
used. They concluded that test design and
quality control must be carefully selected and
standardized before the results are meaning-
ful.
2) Stuart - Comparison of MFs in Recovery
of Naturally-Injured Coliforms
In a series of studies, Schillinger et al. ex-
posed pure cultures of E. coli using MF
158
-------
chambers, in situ in a stream. Cells were re-
covered, enriched and tested by the M-FC
procedure, using Gelman, Millipore and
Muclepore filters. Results to date show no
significant difference in recovery of E. coli on
Gelman or Millipore filters. Nuclepore filters
gave lower recoveries than the other brands.
It was concluded the Nuclepore brand should
not be used for culture work.
3) Harris - Efficiency of Coliform Recovery
Using Two Brands of MFs
Harris and Bailey compared three lots
each of Gelman and Millipore filters in the
testing of aerobic lagoon and sewage samples
for total and fecal coliform bacteria. Gelman
filters were steam-autoclaved and Millipore
filters were sterilized with ethylene oxide.
Total coliform counts showed no significant
difference with lot or brand. However, twice
as many fecal coliforms were recovered on
Gelman filters as on Millipore filters. Colonies
on Gelman filters were smaller than on
Millipore. Millipore filters were blue and Gel-
man filters beige, suggesting a pH effect on
the dyes.
This work confirms Presswood and
Brown's work. In later work by Harris and
Bailey, experimental membrane filters from
Millipore showed recoveries comparable to
Gelman. The authors urge more quality con-
trol for uniformity between batches and
brands.
4) Dufour - Comparison of MF Brands for
the Recovery of the Coliform Group
Dufour and Cabelli studied the recov-
eries of pure cultures of E. coli, K. Pneu-
moniae and fecal and total coliforms from
natural samples, using membrane filters pro-
duced by Gelman, Millipore, S&S, Sartorius
and Nuclepore.
They found that strain differences in the
organisms and differences between lots of
membranes confused the recovery compari-
sons between brands. They noted that the
precisions were consistent with brands when
using pure cultures.
The accuracy of recoveries was the same
whether pure cultures or natural samples were
used. Because precision decreased with natu-
ral samples, studies done with these samples
should use a larger number of replicates than
for pure cultures. Acceptable total coliform
counts were obtained with all filters except
Nuclepore.
5) Glantz - Comparison of Millipore and
Gelman Filters, Culture Media, Incuba-
tors and Escherichia coli strains
Pure culture isolates of E. coli were test-
ed for recovery on Trypticase soy agar (TSA)
pour plates, violet red bile agar (VRB) spread
plates and M-FC broth membrane filtrations,
using different brands and lots of membrane
filters. Recoveries were compared at incuba-
tion temperatures of 35 C, 43 C and 44.5 C.
Although some cell counts were reduced
at 44.5 C, counts varied most significantly
between E. coli strains and between different
MFs. M-FC broth gave lower recoveries than
VRB. Standard strains of E. coli are recom-
mended for future evaluations.
6) Davis - The ASTM Proposed Membrane
Filter Test Procedure for the Recovery
of Fecal Coliforms
Jackson and Davis described the pre-
liminary test procedure developed by the
ASTM subcommittee 019.08.04.02 for re-
covery of fecal coliforms by membrane filters
and detailed the round robin test performed
by ten laboratories using the procedure. They
summarized the statistical analysis of the
results and concluded that the preliminary
procedure did detect differences in filters but
was unsatisfactory because it did not separate
the effects of individual laboratories and
techniques from the natural sample differ-
ences and other procedure variables.
7) Goddard - Critique on ASTM Test for
Recovery of Fecal Coliforms and Propo-
sal for Modified Method
Goddard feels that the necessary stand-
ardization and control of variables was not
attained in this first ASTM test effort.
He discussed briefly membrane charac-
teristics and suggested that the better re-
covery reported with large surface pores
could be due to the cradling effect or simply
be due to easier passage of nutrients. He sug-
159
-------
gested that an optimal pore size for fecal
coliforms be established as a standard and
comparisions of filters then made based on
other physical characteristics of the filter.
Goddard critiqued the ASTM procedure
with the following suggestions for improve-
ment: Use a standardized potable water
sample to obtain a more uniform and favor-
able flora for the test, report noncoliform and
coliform counts, use spread plates of same
size as filters for fairer establishment of ref-
erence count, place samples randomly in in-
cubator and verify accuracy of thermometers.
IV. Solutions
1) McFeters - Recovery Characteristics of
Bacteria Injured in Natural Aquatic
Environment
Bissonnette, et al. continued the MF
chamber studies described by Schillinger,
et al.. They placed raw sewage, E. coli and S.
faecal is cultures in the chambers in situ in
stream, and sampled the chambers daily over
time. They plotted recoveries in Trypticase
soy yeast extract agar (TYSA) and desoxy-
cholate lactose agar (DLA) and calculated the
number of bacterial cells injured (unable to
grow).
Further studies added MPN lactose
broth, MPN BGB broth, M-Endo MF and
M-FC membrane filter media. Comparative
recoveries were:
MPN - TSY = MPN - LB > MPN - BGB > DLA
>M-Endo-MF>M-FC-MF
The numbers of survivors increased with
increased time of exposure. Membrane filter
media were less efficient than other proce-
dures in recovering injured bacteria. Exposure
of uninjured cells to TSY broth repaired the
cells so that they could reproduce. A two-
hour enrichment on TSY agar was suggested
before enumeration of indicator bacteria on
a selective medium.
2) Geldreich - An Improved MF Method
for FC Analyses
Rose et al. described the development
and testing by three laboratories of a two-
layer agar method. The procedure used an
overlay of lactose broth agar on M-FC agar
and two-hour incubation at 35 C prior to the
44.5 C incubation for 22-24 hours. Limited
work with raw and chlorinated wastewaters,
and results from reservoir, stream and marine
samples showed improved recoveries in 59 of
61 samples and a median improved recovery
ratio of 1.9. More work must be done to
verify these results.
3) Grasso - Measurement of Fecal Coliform
in Estuarine Water
Stevens, Grasso and Delaney cited at-
tempts to improve recoveries of fecal coli-
form from seawater using Millipore filters in
two-step procedures. The first series using
minimal media at 25 C failed to yield con-
sistent recoveries.
The second approach utilized a two-step,
two-day procedure of incubation at 25 C for
18 hours on a minimal LES medium then
transfer to M-FC for incubation at 44.5 C for
24 hours. Grasso reported an average increase
of 2.9 in the recovery of fecal coliforms as
compared with EC counts. Ninety-three per-
cent of the picks of fecal coliform colonies
did verify.
Later work suggested that batch and
brand differences in membrane filters may
have influenced the study data which was
generated using Millipore filters only. Com-
parison of Millipore and Gelman filters
showed a high level of dissolved solids in the
Gelman filters. Examination of water ex-
tracts from the filters showed a low pH and
significant mg/liter levels of ammonia-nitro-
gen and orthophosphate in the Gelman
filters. Gelman filters showed higher recov-
eries than Millipore filters on regular M-FC
medium but lower recoveries with the new
LES procedure. Further tests are being con-
ducted.
4) Lin - Evaluation of Method for Detect-
ing Coliforms and Fecal Streptococci in
Chlorinated Secondary Sewage Effluents
In a massive series of tests, Lin studied
total coliform, fecal coliform, and fecal
streptococci recoveries from unchlorinated
waters and wastewaters and chlorinated efflu-
ents, using membrane filters and MPN tech-
niques.
160
-------
MF recoveries with chlorinated second-
ary effluents were significantly lower for total
coliform, fecal coliform and fecal streptococci
than the counts by the standard MPN. M-FC
counts were lower than EC recoveries. No
fecal coliform enrichments were reported.
Low recoveries of total coliforms using a
single-step procedure were improved 1.5
times by use of the LES two-step procedure
and were comparable to the MPN values for
these effluents. Equal recoveries of total
coliforms were made with the LES two-step
MF procedure and the standard MPN con-
firmed test when testing chlorinated second-
ary effluents.
Use of a bile broth enrichment for two
and three days increased fecal streptococci
counts to a level comparable to MPN values.
5) Sladek - Optimim Membrane Structures
Sladek studied test factors affecting re-
covery of fecal coliforms in MF tests and con-
cluded that the most critical factor in re-
covery was the structure of the membrane
filter. Studies on maximum recovery of fecal
colifrom bacteria showed that there was an
apparent optimum top surface opening of
2.4 jum accompanied by a 0.7 /zm internal
pore size. This surface opening was reported
as critical for maximum recovery of fecal coli-
form organisms because the cells are exposed
to a high evaporation effect at 44.5 C. Evap-
oration produces hypertonic solution areas
unless the cells are cradled in surface open-
ings. Millipore feels this theory explains the
low M-FC recoveries on 0.45 /im membranes.
The better recoveries on the new 2.4 ^m
surface pore, 0.7 ;um internal pore membranes
confirm this. Data are limited and must be
substantiated.
GENERAL COMMENTS
Winter: Let's forget about any more compari-
sons of MF's. There have been enough
such studies. If we do any more work
we had better look at the solutions
suggested in this Symposium.
Almost all papers given here were fo-
cussing on some variable and ignoring
another variable of lot, batch, medi-
um, incubation temperature, time and
Editor's
Note:
Winter:
so forth. A number of solutions were
proposed and these proposals need
more supportive data, but only after
factors which affect the MF test are
controlled or randomized as described
by Karl Sladek, so that we can get a
valid comparison of numbers.
We would urge everyone to partici-
pate in collaborative studies per-
formed with ASTM, EPA and other
testing groups. No proposed method
should be accepted until it has had a
satisfactory collaborative study. Fail-
ure to do so was how mistakes were
made in the past. If methods get into
books, there is a great deal of diffi-
culty getting them out.
At this stage we shouldn't try for an
instant solution. We have a serious
problem of test recoveries made more
serious because of compliance moni-
toring regulations promulgated by
EPA and implemented by the States.
Specific industrial and municipal dis-
chargers are required to get NPDES
permits and prove that they are not
exceeding the limits for fecal coli-
forms set in their permit. The fecal
coliform test that must be used at this
time for chlorinated wastes is the
MPN. The MPN is the only technique
we know of which seems to withstand
the effects of chlorination and give us
the maximum recoveries from chlori-
nated effluents.
Since the time of the Symposium in
January 20-21, 1975 EPA has review-
ed public response to the proposal
that only the MPN be used to analyze
effluents for fecal coliforms in the
presence of chlorine. The current
proposal is that both the MPN and MF
may be used but the method used
must be identified. The MPN is the
method of choice if controversy is
anticipated because the MF has been
reported to yield low and erratic
results from chlorinated effluents.
The users need the cooperation of the
MF manufacturers in solving the prob-
lem of recoveries on MF's. If they
come out with a new formulation of
161
-------
membranes such as the one spoken
of yesterday, the HC membrane
filter from Millipore, I hope that we
don't find out that it has phosphate
or some other stimulating factor in
it. We don't want more surprises
about membrane filters. We were
told that membrane filters were inert.
The specs have always said they shall
not enhance or inhibit growth. What
should we do now? If user and sup-
plier exchange information and work
together they can find hopefully a
solution to the MF recovery problems.
Brezenski: Bob Bordner started the Symposium
with a question: why are we here? I
think the answer was to cope with the
problems of the lack of recovery by
the membrane filter systems, and the
number of conflicting reports in the
literature. After reviewing the papers
and tapes and listening to the discus-
sion yesteraday we concluded those
two objectives were met. So from that
point of view, the meeting was suc-
cessful, but the ASTM Committee on
MF's should continue its work. I
would like to make a suggestion to the
manufacturers. One company contri-
buted two papers to the Symposium.
There are other companies manufac-
turing MF's. I would like to see a
greater research contribution by the
companies themselves. I am sure that
most of them develop data in the pro-
cess of controlling the membrane
manufacturing process. This data can
contribute to the knowledge about
the physical and chemical character-
istics of the membranes.
A number of papers described the
phenomenon of the stressing of cells.
There were a number of terms used
yesterday: stressed cells, attentuated
cells, resuscitated cells, damaged cells,
debilitated cells, unresuscitated cells.
All of these terms have a somewhat
different connotation. I would urge
some group, government, academic or
private, to investigate what these
debilitated cells are physiologically.
Once we have this information, the
medium and the whole testing pro-
cedure can be modified to recover
these cells. Right now we see a lag
time in growth and say that the cells
were debilitated or they were stressed.
In what way? This is a major problem
that was not discussed in the Sympo-
sium. Yesterday, Ed Geldreich men-
tioned the tremendously large number
of MF tests run in the US each year. It
brings to mind that if we have defec-
tive MFs or problems of recovery, we
have had many defective numbers
produced and used over these past
years.
I would like to close by saying that it
was an enjoyable experience. I have
been in a lot of other meetings and
this is the first time I have seen so
many worthwhile contributions and
such active participation in a one-day
session. I think we are well on our
way to solving this problem. Thank
you.
FINAL DISCUSSION
Cotton: I would like to commend the review-
ers for putting the program of yester-
day in such great perspective. I would
like to speak briefly on behalf of
membrane manufacturers. I was with
Miliipore in the beginning when the
company first took the initial con-
tract and developed the membrane. In
fact, I was responsible for part of the
development and for the initial work
on the coliform test done by Milli-
pore. One point which was left out
was that although MF's are a big busi-
ness for other purposes, the commer-
cial development of the membrane
filter stemmed from the coliform
test. It was felt that the coliform test
provided a significant commercial use
for the membrane filter. This is where
it all started. Although the use of the
membrane filter in sanitary bacteriol-
ogy is not the major commercial appli-
cation, in spite of the millions of tests
that are being run, it is a significant
one and should be of great concern to
the membrane manufacturers.
Mr. Geldreich made a point on filter
specifications and quality control that
162
-------
I think needs clarification. I know
that our Company and other mem-
brane manufacturers are aware of the
need for extreme quality control tech-
niques in their own laboratories. The
problem is that there are many as-
pects of the filter that need to be con-
trolled. Sometimes when you find a
little piece of paper, as we saw yester-
day, or when a box of filters comes
out of production without the filters
and only the separator pads, you can
get a good laugh, but this represents a
very small percentage of the total
products produced. It is only one of
about 14 different aspects which need
careful control. All the manufacturers
have brought some fine control condi-
tions to the filter manufacturing pro-
cess.
One point brought out was extract-
ables in filters. I don't think this was
completely clarified yesterday. One
manufacturer's representative pointed
out that glycerol was put into the
filter to plasticize it. This is not cor-
rect. Glycerol is used in pore forma-
tion. It is a means to control the pore
size of the filter, and after manufac-
ture it is removed. The degree of re-
moval will have an effect on the elas-
ticity of the filter and for certain
applications it is not important to
remove all of the glycerol or even
most of it. For some applications
however, it is very important, espe-
cially in gravimetric analysis. In the
case of the coliform test, if too much
glycerol remains in the filter, false
positives will result because glycerol
is broken down to aldehydes which
react with the fuchsin sulphite system
and produce sheen. So, it is very im-
portant to limit the amount of glycer-
ol that remains in the filter. This is
one of the reasons for the tight
government specifications. Originally
the specifications on total extractables
in the membrane were set at 2.5%
Recently, the extractable condition
on specifications in the government
specs was dropped, and we are look-
ing into this. We think it should be
re-instituted primarily because of the
false positives in the coliform test.
Also, in the gravmetric analysis, which
is done in water and in fuels, false
readings can result because of extract-
ing some of the weight of the filter
which will counteract the-dry weight.
Another membrane manufacturer
mentioned the inclusion of the wet-
ting agent, which is an alkylaryl-
polyether alcohol. This is well re-
searched. It was selected because of
low toxicity and is in the formulation
primarily for pore formation and con-
trolled pore size. The amount which
is removed in the manufacturing
process must be controlled so that
the filter is wettable. Otherwise,
filtration won't take place or non-
wetting spots will result. The state-
ment was correct, but the primary
reason for it is control of pore for-
mation in membrane production.
Subsequently, the procedure calls for
removing both the wetting agent and
the plasticizer in the processing.
I want to point out that during the
development of the total coliform MF
test, thousands of tests were run.
There was a competition between the
membrane filter and the MPN pro-
cedures. The problem was to get
acceptance from the Standard Meth-
ods Committee of the MF method in
comparison to the MPN. Hundreds of
papers, many contributed by some of
the researchers here today, were
written before this test was accepted,
first as a tentative standard and then
as an alternate to the standard MPN
test. Subsequently, however, when the
fecal coliform test was developed as a
better sanitary indicator, something
slipped in. It was assumed that be-
cause this was also a coliform test, the
same filter should be the right product
for the test. Other aspects of this test
were examined. Recently* reports
were published that observed a dif-
ference in recoveries in the fecal
coliform test. Dr. Presswood's paper
appeared first and other followed,
but there was a lot of confusion.
When you look at these reports to-
gether you find that different research
found diametrically opposed results.
163
-------
The ASTM committee that will be
meeting here this afternoon to estab-
lish methods for evaluating mem-
brane filters has decided that the fecal
coliform test would be the optimum
test for evaluating all microbiological
membranes because of its sensitivity.
They worked on the test procedure
and modified it and as is the final step
in all ASTM methods, planned a
round robin study. A number of labo-
ratories, including our own, agreed to
run the test and compare the results
to make sure that the test works. The
results of the round-robin were rather
confusing and you heard about that
in the report by Mr. Davis. It was dif-
ficult to determine what had hap-
pened.
Our company began a research pro-
gram to find the problem. We were
looking for what we call the big red
X, this confusing factor that led to
what seemed to be total confusion in
the fecal coliform test. Dr. Sladek's
group spent a great deal of time re-
searching the point. They found out
that there can be an effect of the dif-
ferent factors. The question of differ-
ent methods of sterilizing filters,
autoclaving versus ethylene oxide
versus high voltage were discussed. I
think this point was just a little con-
fused yesterday. Proper sterilization,
regardless of the method, should not
be a significant factor. If you leave
ethylene oxide in a filter and test it,
it's going to be toxic to the organisms,
there is no question about it. That's
what the ethylene was there for, to
kill organisms. But if you properly
remove the ethylene oxide, it will
have no subsequent effect.
Let me re-draw the curve that Dr.
Sladek showed in the presentation, be-
cause I feel that surface effects are not
the only factor, but the key factor.
After we found out about the surface
pore phenomenon, we ran tests on a
tremendous number of different pore
size filters, comparing the surface
pore openings and pore sizes against
recoveries of fecal coliform in the test.
With the fecal coliform test, there was
a very clear pattern. If the ordinate is
recovery shown as percent recovery or
cell numbers and the absissa is the sur-
face pore size measured with the
electron microscope, then there is a
very pronounced curve of this nature.
We repeated the tests many, many
times. Because of time limitations, Dr.
Sladek showed only a few of the
results, but what was shown clearly
was that the .45 urn filter which we
manufacture has a pore size opening
which lies close to the top of this very
steep curve. I will not speak about the
other membrane manufacturers' prod-
ucts. However, a similar curve should
develop for their products. If there is
a slight difference in the surface pore
openings from lot to lot, and batch to
batch, it is a factor that has never be-
fore been quality controlled. We did
not realize and no one knew that it
was an important factor, so we were
looking at retention pore size and not
surface pore size. In the future this
will change.
Slight changes in the surface pore
size or pore openings could create a
tremendous difference in the recov-
eries and this, we are convinced, is the
reason why attempts to compare
autoclaved packed filters with ethy-
lene oxide packed filters showed one
thing one time, and something oppo-
site the next time. We were not care-
ful to use filters which were the same,
from the same batch or with the same
surface pore size. What we are doing
as a company is recommending that
for the fecal coliform test we go up
into this region where slight changes
in the surface pore size will have very
little effect in the recovery of fecal
coliform. In the limited tests we have
run comparing this for optimum sur-
face pore size filter, we found a very
good uniformity of results and we
have found recoveries usually higher
than what we are using as the control,
the spread plate. That's where our
technology stands at this particular
time.
In this research we came up with a
few other points which are worth-
164
-------
while reiterating. I think Dr. Sladek
mentioned these yesterday but in
terms of an appropriate control for
this type of work the thickness of agar
in the spread plate control is an im-
portant factor. We were just casually
pouring our spread plates without
carefully measuring the amount of
thickness and found that this was an
important factor. We use a 90 mm
diameter petri dish. We ran tests using
different quantities from 15 ml up to
35 ml of M-FC agar and we found a
recovery which looked essentially like
this with a peak at about 30 ml in the
dish. But if we had too little, we had
low recovery so that is a factor that
we are now carefully measuring for
our controls.
A second factor mentioned by several
people is the dilution water. Initially,
the fecal coliform test was set up with
phosphate buffer dilution water and
phosphate buffer is toxic to fecal coli-
forms. It is now written as peptone
dilution water and it should be cor-
rected, incidentally, to phosphate
buffer peptone water (we will talk
about that this afternoon). If you
have good pH 7 distilled water, pep-
tone water is fine, but if your distilled
water is off in pH, you need a buffer.
This is another factor keeping your
control sample count low.
To finish off, I have another point for
clarification. The question was raised
about the effect of vacuum on bacter-
ial cells. Let me say clearly that the
degree of vacuum has no effect on
the bacterial cells. When the cells are
in the water they don't know whether
your vacuum pump is pulling one inch
or three atmospheres of pressure.
They are not affected by the vacuum
unless they are on the surface of the
water where the vapor pressure has
an effect. So during filtration they do
not know what the vacuum is and
don't care. The rate of impingement
against the filter is not significantly
different. It is very slow regardless of
whether it is high vacuum or low
vacuum. At the end of the filtration,
when there is no more water going
through the filter, there is a blockage
effect because of Poiseville's law of
capillarity. The vacuum is below the
filter but they again don't know to
what degree the vacuum is below the
filter. So unless you let the filter sit
there and dry for a long time where
the drying effect could kill the cells,
there is no effect of degree of vacuum
on the cells.
Question Is there anything in the literature that
from we could read to find out all of this?
Audience: I've heard you say this often, and I
have heard others talk that way but
have never read anything about it.
Cotton: Well, that particular point isn't worth-
while doing a research study on, be-
cause there is no possible effect. It
can't happen. The cells don't know
what the vacuum is. However, I don't
know of literature on that particular
study.
I think I have just about covered all
the points that I had here, except
that as a result of our information we
are going to make available a filter
which has a surface opening pore size
which will have a retention pore size
of .7 urn. I thank you for allowing me
this time to speak on behalf of the
membrane manufacturers. I would be
happy to answer any questions.
Bordner: Mr. Cotton, you have just given us
another paper which was not on the
agenda but provided additional in-
sight into filter manufacturing pro-
cesses.
Winter: We are looking at a graph showing the
recoveries of fecal coliform relating
to the surface pore opening. One of
the points in the paper"was passed
over yesterday. It shows that passage
of microorganisms begins, interesting-
ly enough, at 2.4 /zm. This means that
cells pass through your membrane. If
we will take Sladek's description of
the standard deviation as being the
square root of the mean, and the
mean pore opening size is 2.4 vm,
then one standard deviation is about
1.2 //m and that means that half the
165
-------
pore opening sizes are greater than
2.4 /im and hence are passing cells?
Sladek: I was speaking of a certain contribu-
tion to scatter in measuring counted
data, such as you have in bacteriology
where you count a number of colo-
nies. Now this is not counted data,
this is a measurement which is more
like a measurement of temperature,
and on something like a pore distri-
bution there is no necessity for the
standard deviation to have any partic-
ular size. That is, there isn't a Poisson
distribution involved. This is not a
counted sort of phenomenon. In fact,
the width of the pore distribution in
membrane filters is controlled entirely
by the detail of the manufacturing
process. It is considerably narrower
than what you were just saying.
Winter: Well, how narrow is it? This is a point
that has always bothered us. In any of
the literature, manufacturers talk
about A5nm openings. They show
neat pore size distribution in their
literature which has no tails. I have
always been curious as to how they
do something that nobody can do,
which is to cut off the tails of the
distribution. It shows that as you
go up in pore size opening on the sur-
face, you are also enlarging the size
of the internal pore so that at 2.4 ;um
the internal pore size is not .45 but
0.7 /urn. So pores are coming very
close. Are we retaining all of the cells?
Cotton: May I speak about that question
since I'm the person that orginally
cut off the tail of that curve. That
curve comes from work in measuring
pore size with the mercury intrusion
technique. I won't go into detail be-
that takes quite a while, but there is a
tail on both ends of that curve and
that was cut off for the purpose of
explaining our case. The key factor
as Karl pointed out yesterday, is re-
tention of particles which are filtered
through the filter and the criteria for
the .45 p.m filter as established by the
government was 100% retention of
Serratia marcescens cells which aver-
age in size of .6 to .7 micrometers in
Bordner:
diameter. The coliform cells that we
know about, and as described in
Bergey's manual are all larger than
that. I think Bergey states 1.2/zm to
2 ^m. They may expand on the range,
but our experience is with filters
which average this size and in which
there is no passage. That is, the filters
made at the optimum size in the mid-
dle of this curve showed no passage
whatsoever. We began to see passage
at the 3 jum size, so if you look care-
fully at the graph we showed you will
see some passage at the 3 urn surface
pore opening and this would answer
the spread question also. The passage
is probably from the larger pores in
that particular filter. Its not that the
precise size shows no passage, but it's
the filter at that pore size that has
pores larger or smaller which show no
passage.
Are there any questions related to
Dick Cotton's comments?
Seidenberg: Do I understand that you are develop-
ing another membrane with a surface
pore size which will be suitable for
fecal coliform?
Cotton: Yes. Developing may not be the
proper term, because we can now pro-
duce membranes with any graded pore
sizes. Let's say we have another mem-
brane which we are going to make
available.
Seidenberg: Does this mean that we'll have to have
one membrane for total coliform and
another for fecal streptococci?
Cotton: Not necessarily. It is our current
feeling that the filter which we are
going to recommend for fecal coli-
form will be completely satisfactory
for total coliform, although not neces-
sary for total bacteria count. As for
fecal streps, we can't answer your
question just yet because we will have
to evaluate that parameter.
Seidenberg: Then it's possible that when we do
tests we may have to use one of more
different types of membranes, is that
right?
166
-------
Cotton: That's possible.
Seidenberg: That's more confusion. Thank you.
Cotton: This is true, of course, with a great
many of the procedures even now.
There are a variety of membranes
available; you may just be concentra-
ting on the one .45 jum membrane,
but there are not 14 different pore
size membranes.
Seidenberg: I'm only thinking of bacteriology.
Cotton: For the optimum test, it may be
necessary to use different filters.
Bordner: Dick, can we review? Is this new
membrane with the wider surface
pores, the HC experimental mem-
brane?
Cotton: At the present time, our company is
calling this the HC experimental
membrane.
Bordner: Is this a membrane that has an ap-
proximate mean pore size, if we can
use those terms, of 2.4 jum on the top
surface and .7 Aim in the bottom
which we can call the retention pore
size?
Cotton: Surface pore size of 2.4 micrometers,
plus or minus, we don't know ex-
actly yet, and a retention pore size of
.7 micrometer.
Brodsky: I wanted to ask whether Millipore or
other membrane filter manufacturers
in their quality control work on mem-
branes have looked at the filtrate to
see the rate of passage of the organ-
sims through the filter. I discussed
this, this morning with somebody, and
I think it is a valid point. When you
show on this point of the graph that
any slight variation in retention pore
size can have a tremendous effect on
surface pore size, and we know that
some organisms are always going to
pass through, they don't always line
up in any particular direction; they
don't polarize themselves, so that
you get an increased loss of organisms
through the filter due to slight varia-
tions in surface pore retention size.
Cotton: Let me clarify that point. Passage of
organisms through the filter is a very
important factor with us. We do check
very carefully to determine whether
or not the organisms have passed
through, both in this test and more
significantly in sterile filtration pro-
cedures. Passage test data were pre-
sented on the graph that Dr. Sladek
showed, and all filters with the larger
pore size-surface pore openings were
checked for passage. This explained to
us why we got a tail-off on this curve,
it was due to passage. That passage
data was presented here and if you got
a copy of Dr. Sladek's paper I think it
is explained more clearly. What we
found was that at the optimum sur-
face pore size, there was no passage.
Actually, we got just as high a re-
covery with larger surface-pore size,
but because of passage, the recovery
dropped. If you added the passage
data to the recovery data you came
out about the same until you reached
the point where the pores were so
large that media transports were ex-
cessive and you got sloppy colonies,
spreading and difficult-to-read colo-
nies. So that was really the upper
level, except for the passage factor
which was carefully studied.
Bordner: Thank you, Dick. Because Mr. Cotton
is from Millipore Corp., I feel a need
to invite comments from other MF
manufacturers. If none, we welcome
any additions or comments on this
summary.
Litsky: You know, there is an old trick: If
you want to keep a guy quiet you
make him a moderator. As long as I
am not a moderator, I'm going to talk.
I can't help thinking that this is the
first time I have seen so many big-wigs
from EPA in one room and I am going
to take the opportunity to remind
them what they want to forget. Years
ago we woke up one morning with a
book published by the APHA and
sanctioned by EPA which required
everyone to use a fecal coliform MF
test. We could not fight this because
if you ever went to college you
learned that you follow what the
167
-------
Bordner:
coach believes and the coach in this
case accepted only fecal coliforms. It
was pushed down our throats without
proper evaluation because, if we had
had time to evaluate we would have
found the test did not work for all
types of samples. Most of the fault lies
with EPA; however, we in the uni-
versities and in the other non-federal
laboratories are at fault because
microbiologists are very lazy people.
When someone publishes a method
which gives two steps, we accept it
without question because 1) the
coach accepted it and 2) no one had
the guts to question it. I was a little
teed off with Presswood because we
were content. We were very comfor-
table although we knew it wasn't such
a good method, but EPA wanted it
and it saved us some time and money,
so we accepted it. I hope to God we
don't repeat this mistake.
We (Litsky, Rose, and Geldreich)
proposed a method yesterday. We
proposed the method only to ask all
of you people to try it and tell us
whether we are on the right track. I
think we are trying to make a point.
I think we all look to EPA for guid-
ance and this is the way it should be;
however, I hope EPA takes the warn-
ing that we are not going to accept
any other methods without the proper
round-robin testing, the proper field
testing and without the proper data to
prove that the method proposed is
better than the method we have now.
I hope EPA takes the lead because I
speak for all the private, city, and
state laboratories. We are anxious to
cooperate. Use us, but don't abuse us.
Thank you Dr. Litsky. I feel that not
only have our minds been stimulated
but also, at least psychologically, that
area of anatomy that young children
usually have stimulated after getting
into trouble. It occurs to me that EPA
has always tried to work through
the concensus of opinion and through
Standard Methods of which you and
several others within and outside of
EPA are members. So we try to pro-
tect the Agency from single interest
pressure. We will accept the guidance
and, I hope, the help of interested
microbiologists, academic and non-
government. Are there other additions
or critiques on the summary?
Hendricks: I really don't know what to say after
Dr. Litsky's comments. Certainly I'm
not an EPA big-wig. I'm big, but the
stature and the agency position do not
correlate. I would like to say to Dr.
Litsky that his comments are true in
many respects. However, I would say
this about the fecal coliform tests.
They have served us well. We are for-
tunate that we have reduced the
intestinal disease rates in this country
and other parts of the world where
this problem has existed.
There is no doubt that we can do bet-
ter and I would like to make three
comments. One is about the organisms
which are used in tests to evaluate
media, filters and culture procedures.
We all know that when we grow a
pure culture in a broth and recover it
under stress conditions, whether the
stress is temperature, nutrient concen-
tration, or inhibitors, the results will
be low. Organisms that you introduce
in the environment become stressed
by environmental parameters as Drs.
McFeters and Stuart have shown. To a
great extent we are going to have to
control the way we use our organisms,
whether they are pure or natural
cultures, if we are ever going to
achieve uniformity in the procedures
that we have been talking about this
week. There is plenty of data to dem-
onstrate this. We need a standard way
of treating our cultures.
Secondly, the papers of Presswood
and Brown and others, I think show
quite clearly that there is a problem
with membrane procedures. Let's look
at the reasons why these results occur.
I think we have had enough of the
observations.
Thirdly, Ed Geldreich mentioned yes-
terday that there can be one coliform
per 100 ml in drinking water. So if we
168
-------
recover additional coliforms by im-
proved procedures, what is this going
to do to our standards? We should
keep in mind the general significance
of an improved technique. Something
has to be done with the data and we
had better know what we are going to
do with it once we get it. The concept
that recovery may be a function of
survival is a valid one.
Fourthly, I am aware of a paper by
Kerr, University of Georgia, that
showed visual differences by electron
microscopy in Millipore, Nuclepore,
and I think, Sartorious membranes.
This technique might be of value to
those of you who want to observe
bacteria after culturing by membrane
procedures. Thank you.
Bordner: I might add that Dr. Kerr was in-
vited to give a paper at this sympo-
sium and would have accepted had he
had the travel funds.
Power: I would like to just make a comment
or two. We had this discussion yester-
day on media. The point of the discus-
sion was to tell you that we have spe-
cifications for all our media and that
we use the best ingredients we can
obtain. Speaking for my own com-
pany, we are not a chemical manu-
facturer. We buy many of the ingred-
ients from reputable companies. On
receipt, we do quality control tests
to insure that the product is what we
ordered and what it says on the label.
Just as the user can take a good
product and ruin it in the preparation,
we are capable of taking good ingred-
ients and ending up with a product
that doesn't perform. That's the pur-
pose of the quality control laboratory.
We do not release any product that
we feel won't perform for the pur-
pose for which it was intended.
From the discussion, I understand
that some of you have had problems. I
personally have handled in my depart-
ment all product reports for four of
the last six years. I am not aware of
any great number of problem prod-
ucts so I would ask that you would
let us know if you feel that you are
having a problem. This can be done in
several ways. If you ever see a Bio-
Quest representative you can inform
him. Most of you probably never do,
so please call or write and tell us of
the problem. I will send you an
authorization form to return the
product. A tremendous amount of
material comes to our receiving
department and to insure that it gets
into the right channels I would like
to provide an authorization form. We
would like to have return goods for
testing. If that's not possible, at least
provide the lot number. Then I can
check the production record and see
if there was anything in there that
would indicate that the product might
be any different or any suspicion that
it has deteriorated. I can't do any-
thing without the lot number.
If I can get returned goods we will set
up an evaluation with our reference
shelf sample of the same lot and the
current lot and see the results. We
may be able to find the problem. If
we don't know about it, we can't do
anything.
You are paying good money for the
product and we are putting a lot of
effort into them. There is no sense for
anyone not to use the best available.
We can make changes. We are making
media according to established
formulations of course. We are here to
serve you. We hope to give you a
degree of stability and a lot-to-lot
uniformity and to take out some of
the variables. I would encourage you
to contact us. Of course, one of the
problems is turn-around time, but I
will inform my people that if any
reports come in on these types of
products to please let me know. I
don't handle every one of them indi-
vidually any more. I will try to get
you an answer as quickly as possible.
There are some comments that I have
heard about the storage problem. If
the medium starts to cake, you have
moisture and can have deterioration.
So I encourage the purchase of small
169
-------
quantities, tight covers and the use of
a dessicator particularly for small
bottles. I'll go back and see if there is
anything that we can do as far as
stability.
Someone said that if they reorder
there is a long lag time before they get
the product. This is a problem. It
could be our fault through poor
scheduling. It could be a problem due
to lack of availability of some ingred-
ients. We have made an effort in the
past year to improve our scheduling
and cut down on the backorders, but
I know it is a problem at times. I
would encourage you to call my of-
fice or write if you prefer. If you will
keep in touch with us we will do all
we can to help you. All of the phone
calls, letters, product reports, etc. of a
technical nature come to my atten-
tion. We are there to help you and
you can take advantage of it. Thank
you.
Winter: I have a question for you, Dave, and
for Aaron Lane, if he is here, EPA is
now coming out with its own manual
of methodology which includes a sec-
tion on quality control. In it we are
urging a limited holding time for
media and one of the problems is
that we can't get the manufacturers
to put a date on the bottles.
Power: By law, we do have a date on them.
There is an expiration date and I
don't have the figures with me but I
would imagine it is two years on these
dehydrated materials.
Winter: Do you have a list of recommended
times for holding different media?
The recommended holding times dif-
fer with each media. Some, of course,
like lactose broth are pretty stable. Is
this list available, I think it would
guide the labs. One problem with
some laboratories, particularly state
or federal laboratories, is the need
for unique type media which they
buy in quarter pounds and use only
once in a year.
Power: The stability and expiration date that
we put on will be for the unopened
Winter:
Power:
Winter:
Power:
bottle, because after it is opened, I
don't know. The more it is opened the
faster it is going to deteriorate. I
don't think that we can say, we have
to have data to show that a bottle of
unopened medium is good for a given
time period and expiration date. Once
it is opened, each bottle is unique and
stability depends on whether you are
in a humid climate or a dry climate.
Of course the product has been stored
at a distributor; also it may be stored
in your facility and then stored in
your lab before its opened. The user
opens it, some put the cap on tight,
some put it on loose and some put it
on crooked; so there is no way that
we can control it. I think in some
cases the criticism is somewhat un-
justified because we feel the product
that we send out is okay. So what
happens after that? We have not had
expiration dates until this year. As of
September 15, under the new FDA
regulations, every newly manufac-
tured product should have an expira-
tion date and every product must have
a lot number. Perhaps you can come
up with some guidelines or some help
from that standpoint. I don't think
that we can.
You are saying that your responsi-
bility essentially ends when the
product leaves your plant, but that we
could supply guidelines for storage.
In your manual, you could certainly
use guidelines. We have enough to do
to justify the dating. Perhaps we are
talking abour relatively few products
of interest to the people here. Our
company and our competitors are
talking about thousands of products.
Once the bottle is opened, everybody
is going to handle it a little differen-
tly. I have no objections to anything
you might want to say about storage
or handling media in the laboratory.
We are only going to take it up to that
point and verify how long it should be
good unopened.
Could you provide the data on your
recommended guidelines?
Yes.
170
-------
Bordner: Could we have the comments from
the other media manufacturers?
Would you gentlemen mind remaining
up front for a minute because I think
we might have questions related to
media.
Lane: Concerning Dave Power's paper, the
comment was made this morning that
there is a shift in responsibility of
quality assurance from the manufac-
turer to the user. I don't believe that
was the intent of the paper. Certainly,
we don't ask the user to do all the
quality assurance testing of a culture
medium. Both the manufacturer and
the user must have a quality assurance
program. Our quality assurance, as
Dave pointed out for his company,
stops when the bottle leaves our
plant. Every component of the
culture medium - protein source,
carbohydrate source, buffer, indica-
tor, selective agents - is tested alone
and in combination. The complete
medium is tested for productivity,
pH, and appearance when it is bottled.
Then the bottle is sealed, capped and
shipped to your laboratory.
The user has responsibilities for the
media. Before you use it, you must
make sure that the medium is placed
into a refrigerator if the label so
states. When the medium is delivered
into the laboratory the date of receipt
and the date that the bottle was
opened are both on the labels. Don't
open the bottle until the previous lot
has been used up. That will give you a
longer period of use. Store the medi-
um at temperatures below 25 C in a
low humidity environment and out of
direct sunlight. Do not store the medi-
um near an autoclave or drying ovens.
When you weigh the medium, use a
balance and weights which are fre-
quently checked for accuracy. Do not
weigh in a draft or high humidity
area.
Do not leave the dehydrated medium
exposed to the air. Get it into solution
quickly or it can harden, then prompt-
ly return the cap to the bottle and
tighten securely. Return unused por-
tions to their proper storage. Most
dehydrated media are very hygro-
scopic and the ingredients may be
sensitive to excessive moisture and
light and heat. Exposure to such
conditions, especially when the cap
on the bottle is not securely tight-
ened, may result in moisture uptake
which alters the physical, chemical
and bacteriological properties of the
medium. The result could be harden-
ing of the freeflowing powder, darken-
ing of the powder, oxidation of some
of the components, change in pH,
change in solubility, change in the ap-
pearance of the color of the dissolved
medium and reduction or loss of pro-
ductivity, selectivity or differential
characteristics.
Dissolving of the medium is another
point. If you use glassware that has
not been thoroughly washed, residual
detergents may cause low counts.
Dirty or improperly washed glassware
can change the color of a medium, in-
crease or decrease in pH, cause a pre-
cipitate reaction between the residue
on the glassware and the medium
component, or produce toxicity from
residual detergent.
Impure water or water which has been
stored in a soft glass bottle or exposed
to the laboratory air, can easily alter
the quality of the medium. A good
distilled water which has been in a
bottle on the shelf for two or three
weeks with a lose stopper will take up
C02 and that, too, can effect the pH
of the medium. It can yield a precipi-
tate in the medium or impart toxicity.
Test the water that you are using to
dissolve the medium for conductivity,
metals and other inhibitory sub-
stances. A pH of a good distilled
water is usually between 6 and 6.5.
Do not use water suspected of con-
taining chlorine, copper, lead or de-
tergents. We have run into this prob-
lem when we had complaints of cul-
ture medium. We would send a man
to check how the distilled water was
made. In one instance, we found dis-
171
-------
tilled water being made in a copper
still.
Avoid scorching the medium on the
bottom of a flask or other container
by stirring during heating; and if you
are heating an agar medium be very
careful. Distribute the medium uni-
formly and dissolve completely. To
avoid overflow of the medium, do not
prepare media, especially agar, in a
flask less than 2% times the volume of
the medium. Prepare only sufficient
quantities suitable for use in a week
or less.
Do not attempt to readjust the pH of
a medium unless you have proper
equipment and know the procedure.
Adjusting the pH of a broth is not
really difficult, but adjusting the pH
of an agar is quite difficult. The final
pH of a medium after autoclaving and
cooling to 25 C is on the label. A
medium prepared according to direc-
tions in distilled or deionized water
and which has not overheated during
dissolving and has not been over
autoclaved should have that pH.
Be sure that the temperature and
pressure gauges on the autoclave are
accurate. Careful timing during auto-
claving is essential. Remember not to
start timing when the steam starts
entering the autoclave. Avoid under
or over-autoclaving, especially over-
autoclaving. Frequently check the
efficiency of the autoclave with a
biological indicator such as Bacillus
stereothermophilus. Do not keep a
sterilized agar medium in a water bath
at 50 C for more than 45 to 60 min-
utes. The agar can settle out and
phosphates can precipitate. Do not
autoclave a medium containing heat
labile enrichments or additives which
precipitate by heating. Since heat
penetration is slow in culture media,
especially media containing agar, it is
important that the recommended ster-
ilization period be strictly adhered to.
The time required to autoclave a
medium depends not only on the
efficiency of the autoclave but the
volume of medium in the bottle and
its size and shape. Over-autoclaving a
medium, especially an agar medium,
can cause: development of precipi-
tate, change in pH, carmelization or
darkening, depolymerization of the
agar and reduction in gelation, reduc-
tion in productivity, selectivity and in-
crease in inhibitory substances.
After the preparation, some media
can be left at room temperature in
screw-capped tubes. Agar plates must
be in sealed plastic bags, preferably in
the refrigerator. As Dave pointed out,
the best prepared dehydrated medium
can be destroyed if it is not handled
properly, so I do take exception. I
don't believe that the manufacturer
of dehydrated media wants to shift
the responsibility of quality assurance
to you. You are part of it.
Bordner: Aaron, would you be willing to share
that material that you read with us?
Lane: This is a quality assurance manual
that we recently completed. It is
available. We can send it to you at
any time.
Power: In the transcript of the paper I gave
yesterday is a reference to a paper
published a couple of years ago by Dr.
Vera who at the time was head of
quality control laboratory. It says
very much the same thing that Mr.
Lane read. If you write to us we will
send you a copy of our manual as
well.
Geldreich: The points you are making are not
new. We have evaluated state labora-
tories for over 15 years. We have train-
ed state people who in turn evaluate
the laboratories examining public
water supplies throughout the state.
The laboratory certainly has an im-
portant responsibility in this area.
Once a medium leaves the manufac-
turer there are lots of things that can
go wrong with it. I recently had one
laboratory throw out over $200 worth
of media which was caked. We write
strong reports on these subject mat-
ters. If we don't have standardization
172
-------
Bordner:
Lane:
or control over media, the results will
be meaningless. For instance, one
laboratory in Hawaii reported that
they couldn't get coliforms to ferment
brilliant green when they used it as
normally prepared, but of they
doubled the concentration of the
brilliant green they could get gas
production. These things do happen.
Through laboratory evaluation, EPA
will encourage the people in the
laboratory to maintain a good quality
control program. That's their respon-
sibility.
However, there is another form of
quality control where manufacturers
can help. We know there are bad
batches of media. This generally is
what happens: the laboratory calls the
representative of the scientific supply
house; the representative comes in and
replaced the old batch with a new lot
of medium. However, others may have
been using that bad lot of medium.
We don't know its bad unless we have
quality control check it. I would
urge the manufacturers who find a
bad lot or batch of media to recall it
so that the rest of us don't have to
find this out the hard way. In New
England where two states next to each
other had the same lot of medium
they had the same problems with it.
The company recalled the medium,
but I know of another state close by
that used that medium. They didn't
quality control it so they consumed
the batch of medium. I am asking you
and others who manufacture media
that if once you find that you have a
bad batch tell us you will replace it
or refund our money. Look up in the
records and find where in the market
the product is being used and recall it.
Thank you.
Aaron, do you want to reply to this
comment?
I'm quite sure that our quality assur-
ance laboratories do not approve a
batch of medium unless it performs in
the manner for which it was designed.
With the coliform MF media as I men-
tioned yesterday we use river water
and ATCC cultures. I am not saying
that those state laboratories mishand-
led the medium. Something may have
happened. The problem is that the
manufacturers are not being informed.
If we were given the opportunity to
test the sample that the state is using
against our official sample, there may
not have been a problem. Our quality
control laboratory must report all
problems and complaints they are
made aware of. Did these state labor-
atories talk with our representatives?
Geldreich: Yes, I am told they did. I was not
there. The man gave them a new lot
and took the old lot away.
Lane: I would just like to say that many of
the products are purchased through
distributors and in the case of per-
formance complaints I would ask that
you come back to the manufacturer.
You are probably visited by distribu-
tor representatives far more often
than you are by our own sales force.
In such instances, I don't know
whether the reports get back to us. I
do get letters once in a while from one
supply house in particular.
They have a form letter which is very
helpful. I think you should talk with
the manufacturer. If you have other
problems, deal with the distributor -
he's the one you brought it from.
When you have performance prob-
lems which are serious, deal with the
manufacturer. If you want to call,
that.s fine. We will take it from there.
Without the lot number, one doesn't
know how many years that product
has been sitting around. I know some
of the distribution houses have not
rotated their stock. I think that is
another advantage of the expiration
date. I know that once in Dallas, I
started to get reports of people re-
ceiving material that was far too old.
It was traced back to one distributor
who had found a box that he didn't
realize he had and shipped it all out.
In the future this should improve.
If on a re-test we find that the refer-
ence shelf sample has deteriorated
173
-------
prior to the expiration date we would
be obligated to do a withdrawal. We
know to whom we shipped it. It's up
to the distributors also to have records
to whom they shipped the material.
I think we have a fairly good chance
of getting back through the distribu-
tion system. Anything that can hap-
pen, will happen, sometime, some-
where. We must realize that we are
dealing with variables in biological
materials and transportation system,
laboratories, samples and technicians.
We will do the best that we can, but
I would encourage you to come back
and talk with us. At least there is a
chance that it will be recalled.
Bordner: The key word is communication,
right?
Alico: Its been a fine symposium and I want
to offer my gratitude to the two
people who summarized the presen-
tations. I have suggestions on the title
of the symposium. It reads: "The
Symposium on the Recovery of Indi-
cator Organisms in Applying Mem-
brane Filters." I would like to add
"For Coliform and related Organisms"
since the general papers, the stress
papers, and comparative papers dealt
with coliform-related organisms. I sug-
gest this so that when we get the pro-
ceedings of the symposium it will be
clear that it deals with this one aspect
of membrane filters and not with
others.
One other comment I have is about
the Nuclepore membrane filters. I
believe about a year and half ago at
the ASM meeting in Miami, I received
information from the people at
Nuclepore on using these filters for
the enumeration of microorganisms.
It was stated yesterday that they
should not be used in enumeration
of microorganisms. I think Nuclepore
should notify people that their prod-
uct should not be used for this pur-
pose.
Harris: I think what I have to say is germane
to the issue. The last speaker from
EPA slipped in a little statement
which I think is the whole reason for
our being here. He made the rather
astounding statement, in my view,
that gastrointestinal diseases were on
the wane. This is just not so! The
incidence of salmonellosis in North
America, I'm talking of the whole
continent now, is rapidly on the in-
crease. We are talking of notified
cases. Bear in mind that the average
physician does not notify. Probably
only one in ten is notified. So which
ever way you slice the cake the reason
for our being here is not simply to
design a system. Our basic reason for
being here is the eventual reduction
in G. I. diseases. If we take this as our
measure, we can fail, ladies and gentle-
men, fail dismally. I think that it is
important that we do design a system,
whether it is the fecal coliform indica-
tor system or not. I'm speaking as a
microbiologist and as a physician. I'm
seeing both sides of the issue. I know
what happens from the physician's
side of the fence. He is presented with
a lot of data. He is told thus and such
an incidence of such and such an or-
ganism. He has not been trained. Even
many of our physicians in the Public
Health field have not been adequately
trained to interpret the laboratory
data.
We have been very smug as scientists
in designing indicator systems, coming
up with good systems for protecting
public health, but leaving it there. We
have not taken the trouble to bridge
the gap between the scientist and the
physician. I think that we should not
just be satisfied in designing a system
but also make sure that there is some
carry-over from this type of meeting
to the people who are going to go into
the field and implement our findings.
Unless we do that we can sit here for
the next 20 years designing better and
better indicators but not getting to
the root of the problem which is the
reduction of G. I. diseases. I would
like to correct the statement made by
a gentleman from EPA, that G. I.
diseases are on the decrease.
I wanted to speak before the gentle-
man from BBL and Difco got their
174
-------
plug in. Maddox Chemicals supplies
the Oxoid media in Canada exclusive-
ly. I go along with large measure with
what Dr. Lane said. Many of faults are
at the user level. I am not saying that
the media manufacturers are faultless.
But many of you who use Oxoid
media are aware that if you run into
problems we send people down. I
would underline that we don't do any-
thing about a bad product or bad use
of a product unless we know some-
thing about it. Thank you.
Bordner: We are glad to have the opportunity
to exchange ideas - and problems -
with the media manufacturers at this
symposium.
Hendricks: First of all, let me be the first to
applaud the physician from Canada
who took exception to what I had
said about the incidence of intestinal
disease. I think it is true that there is
an increase in terms of "running
rampant" is a relative term. I consider
venereal disease in this country to be
rampant, but whether this is due to
reporting, or increased actual number
of cases, I don't know. I rather sus-
pect that it is a combination of the
two. I will stand by what I said earlier
that where the coliform tests are used
to monitor water quality, I think the
rate of serious intestinal disease is
much lower than those areas that do
not employ such procedures. Of
course, these microbiological proce-
dures have to be followed with ade-
quate treatment of some sort. I do
believe that they have served us well.
The mechanisms by which intestinal
disease seems to be increasing un-
doubtedly is due to a variety of
things. I would hate to see us abandon
our coliform procedures, to say that
they are no good and throw them out,
because I think the results of not
monitoring water quality would be
disastrous. This is really what I was
saying about the significance of any
technique that's going to have a
tendency to increase numbers. We
know that our measurements of coli-
form procedures may at times be low.
This may very well be one of the rea-
sons why we can recover pathogens,
bacterial pathogens including viruses
in water that appear to be of excellent
quality. Again, I applaud the gentle-
man, but I think there are two sides to
the picture and we should not elimi-
nate procedures that have served us
well. I think we should improve upon
them and be well aware of what we
are counting. Thank you.
FINAL REMARKS
Frith: I think everyone agrees that this was
a very timely conference and we had
an opportunity to share some of our
general knowledge as well as to open
up some additional avenues for in-
depth study. I don't think that any-
one will leave thinking that they got
short-changed from this meeting. It is
important to understand the function,
and I don't want to give you a long
detailed and boring explanation of
ASTM but as your co-chairman for
the sub-committee I would like to
explain the real function of ASTM in
this whole operation and ask you or
invite you to become a participant.
It is obvious that the only way to
eliminate confusion is to develop a
test method that will help everyone to
know that they are buying a stand-
ardized product, that will represent
the state of the art. For about three
years, ASTM has been trying to do
this with just the membrane filter,
not the media - nor the various con-
trols. We have written three draft
procedures: one on recovery that you
have heard about today, one on in-
hibitory effects and one other being
proposed today on retention. This
afternoon we will be working on two
or three of these procedures that have
been round robin tested, from which
you have seen some of the data. We
are trying to eliminate some of the
confusion.
As an ASTM member, you become a
voter, a person who has a chance to
see the draft copies and to submit
negative ballots if you feel that there
is something wrong with the way the
175
-------
technique is being run or if you have
data that does not support what is
being presented. These negative bal-
lots or the additional data supplied
have to be overridden to either prove
or disprove your point. Ginsberg:
Active membership and voting mem-
bership is available by joining the Frith:
ASTM Society. You are not restricted
from attending any of our meetings
as a non-ASTM member. You can
come, you can participate, you can
offer data and comments. ASTM
would like to see you become an
active member and be involved with
us. We hope not only to develop Ginsburg:
specifications for recovery, inhibitory
effects and retention, but also to con-
sider extractables, surface charac- Frith:
teristics and other general topics such
as membrane composition and impuri-
ties. You can see the charter that we
have ahead of us is quite broad.
We will be doing the same thing, I
am sure, with various types of media.
We will be determining how you
actually monitor a temperature of
44.5 + 0.2 C. All of these issues
will be coming out of ASTM in the
next few years. Your active partici-
pation can help that move faster. Dr.
Litsky has said, "Let's either write the
specs by the state of the art and make
them meaningful or forget it." I'm
thankful that each of you has taken
time for a day and a half to attend. I Ginsberg:
think it has been a fantastic sympo-
sium and because of your participa-
tion, we will gain a lot more and
move a lot quicker than we have done Frith:
in the last ten years. Again I say, you
are invited this afternoon to come and
listen and participate. Are there any
questions about ASTM?
test methods on how do you really
monitor the variations in membranes
or monitor the difference in water
samples, etc?
This same idea, is it carried through
the week?
No. The test method will only carry
through this afternoon. Tomorrow,
there is going to be an organizational
meeting for indicator organisms. We
are concerned not only with coliforms
and fecal coliforms but also with
different indicator organisms.
What is the procedure for participat-
ing in any future round robins?
Well, you'll hear about that at the
meeting today. If you as a laboratory
representative would like to partici-
pate you are welcome. We want to get
enough different labs around the
country to get a representative sam-
ple. If you feel that your lab can
spend the time (it is a very exicting
program but it takes an awful lot of
money and time to do what is re-
quired) you will be getting notes as a
member of the committee, and you
will be notified that a round robin will
be taking place. Margareta Jackson,
who unfortunately could not be here,
is the one actually in charge of the
afternoon session.
Do I understand that ASTM supplies
the materials or do we have to buy
them?
In the first round robin we asked the
membrane manufacturers and one
media manufacturer to supply mater-
ials but you will make the biggest
contribution which is man hours.
Ginsburg: This meeting that is coming up this
afternnon, will it deal with the same
problems that the symposium cover-
ed?
Frith: That is correct. We hoped to learn a
lot from the symposium and to use
the knowledge we have gained the last
day and a half to make meaningful
Vlassoff: When the people fill out these forms,
won't they have some idea what's
happening?
Frith: Yes. That is an ASTM form, if you
have not filled out one of these, you
should. We will be submitting to you
everything that has happened at this
symposium. How fast several of us can
176
-------
Question:
Frith:
get together and either get these
papers in some form to send out or
get into a publication with which
both ASTM and EPA are concerned
will depend on some man hours. We
will keep you informed if you leave
your name and address, whether you
are an ASTM member or not, by this
route.
If this bulletin is published, will it
go automatically to ASTM members
or will this be a personal-type mailing?
We are not sure right now. I think the
question has yet to be resolved - how
and in what form this symposium be
Bordner:
provided to you. It will not be pub-
lished until some degree of satisfac-
tion has been reached between ASTM
and EPA, as they were the sponsors.
Again, many thanks for your time and
your talents and we will look forward
to those of you who would like to
join us at 1:30 back in this room.
I want to thank all of you for par-
ticipating in this symposium - those of
you who made special efforts in pre-
paring papers and presenting them,
who participated by offering thought-
provoking comments, and who sum-
marized what was said. To all of
you we owe particular appreciation.
177
-------
APPENDIX
The following paper intended for the Seminar was
not received in time for presentation. However, it
is relevant and is presented here for the considera-
tion of the readers.
COMPARISON OF MEMBRANE FILTER COUNTS AND PLATE COUNTS
ON HETEROTROPHIC AND OIL AGAR USED TO ESTIMATE
POPULATIONS OF YEAST, FUNGI AND BACTERIA
J. D. Walker, B. F. Conrad, P. A. Seesman and R. R. Colwell
Department of Microbiology
University of Maryland
College Park, Maryland 20742
ABSTRACT
Comparison of filter and plate counts of
yeasts and fungi on heterotropic and oil agar re-
vealed higher counts were nearly always obtained
with filters. Comparison of filter and plate counts
of heterotrophic estuarine and marine bacteria re-
vealed that, on the average, filter counts were 80%
lower than the plate counts. These results should
be considered when evaluating methods for enumer-
ation of microorganisms in the marine environment.
INTRODUCTION
Microorganisms present in seawater are usually
enumerated using membrane filters because of
their low numbers. However, in the estuarine en-
vironment, generally characterized by larger micro-
bial populations, microorganisms are frequently
enumerated using plate counts. In the study report-
ed here, Chesapeake Bay water was tested, permit-
ting comparison of plate and filter counts for yeast,
fungi and bacteria.
MATERIALS AND METHODS
Heterotrophic yeasts and fungi were enumer-
ated using yeast medium, details of which have
been published elsewhere (1). Yeasts and fungi
capable of growth on petroleum were enumerated
using oil agar #2, pH adjusted to 5.5 and supple-
mented with streptomycin and tetracycline (50
ug/ml of each), as described by Walker and Colwell
(2). Heterotrophic bacteria were enumerated using
the basal medium of Walker and Colwell (D.
Bacteria capable of growth on petroleum were
counted using a silica gel-oil medium (3). All plates
were incubated at 15 C for two weeks.
RESULTS AND DISCUSSION
Heterotrophic yeasts and fungi count plates
were incubated for a minimum of one week to in-
sure appearance of the maximum number of
colonies (Tables 1 and 2). The results of the hetero-
trophic counts compared with counts for yeasts
and fungi on oil agar plates incubated for at least
two weeks are shown in Tables 3 and 4. It was
necessary to examine the plates periodically, after
the initial three days of incubation, to avoid over-
growth with fungi.
Before plate counts were compared with the
filter counts, a percent comparison was calculated
between the plate and filter replica plate counts
(Tables 5 and 6). These calculations indicated at
least 60% comparison between replicate counts was
obtained. Similar comparability was observed for
results of the yeast and fungi replicate counts.
Comparison of the plate and filter counts of
heterotrophic yeast showed that the highest counts
were always obtained when membrane filters were
used. The percent comparison was always indica-
178
-------
tive of a significantly lower count on the plates
compared with the filters (Table 7). In most in-
stances, higher counts of heterotrophic fungi were
obtained using membrane filters, compared with
the spread plate technique (Table 8). Half of the
percent similarities were significant. The number of
yeasts growing on oil agar, using the filter method
of enumeration, was higher than on the oil agar
plates. In only one case did the two methods give
a reasonably close count (Table 9). As in the case
of the heterotrophic fungi, higher counts of fungi
on oil agar were obtained using membrane filters,
than with plate counts, although the counts were
minimally comparable (Table 10). Results obtained
when plate and filter counts of yeasts and fungi on
heterotrophic and oil agar were compared suggested
that quite different counts were obtained by these
methods and that filter counts yielded higher num-
bers of yeasts and fungi when estuarine and marine
water and sediment samples were examined.
Comparison of plate and filter counts of
heterotrophic bacteria indicated that plate counts
yielded higher counts in most samples, and only a
few samples gave similar results by plate and filter
counts (Table 11). Plate and filter counts of bacteria
on silica gel medium were not high enough to com-
pare results (Table 12). The filter procedure gave
higher counts on silica gel oil for bacteria from
sediment and from Eastern Bay water, whereas
plate counts provided the best estimation of the
bacterial populations in Colgate Creek water enu-
merated on silica gel-oil medium.
The two areas in Chesapeake Bay included in
this study, Colgate Creek in Baltimore Harbor and
Eastern Bay, differ ecologically. Colgate Creek is
continuously contaminated with oil, whereas
Eastern Bay is an oil-free, commercially productive
shellfish area. The higher counts of yeast, fungi
and bacteria on oil media for Colgate Creek samples
can be explained. By virtue of larger populations,
more comparisons of plate counts and filter counts
were possible using inocula from Colgate Creek
than from Eastern Bay.
Plate and filter counts of seawater collected at
stations along the U.S. east coast were compared
and, generally speaking, plate counts of micro-
organisms in seawater, concentrated using an
Aminco concentrator (American Instrument Co.,
Silver Spring, Md.) yielded higher counts than the
filter procedure (Table 13). By dividing the average
filter count by the average plate count and multi-
plying by 100,. a comparison was possible for estua-
rine and concentrated seawater samples. The
average filter count for estuarine samples was 19.7%
of the average plate count and 19.2% for the con-
centrated seawater. Thus, for the average of the 14
estuarine samples and 10 marine samples studied,
filter counts were about 80% lower than the plate
counts.
Nuclepore (General Electric, Pleasanton, Calif.)
filters were compared with Millipore (Millipore
Filter Corp., Bedford, Mass.) filters by scanning
electron microscopy (Todd and Kerr (4)), but
not for efficiency in enumeration of bacteria.
Nuclepore filters are thin (10 urn) films of polycar-
bonate plastic, with an average pore diameter ap-
proximating the individual pore diameter. Millipore
filters are thick (150 um) films of cellulose, the
average pore diameter differing significantly from
the individual pore diameters. Millipore counts
were compared with Nuclepore counts. Data
for the estuarine samples showed no trend toward
higher counts for Millipore filters in comparison
with Nuclepore filters (Table 14). This was unlike
the marine samples which always gave higher
counts with Millipore filters (Table 15). Two dis-
tinct disadvantages of using Nuclepore filters in
laboratories aboard ocean research vessels are that
Nuclepore membranes are markedly thin, making
it difficult to place them on sintered glass filter
holders and on agar surfaces. Air bubbles trapped
between the agar and the filter are often indis-
tinguishable from translucent bacterial colonies.
As a general statement, membrane filters are
more efficient for counting yeasts and fungi but
plate counting for estuarine and marine water
bacteria is recommended over the membrane filter
method. The ease of use and their sturdier nature
make Millipore filters preferable to Nuclepore
filters for field work.
ACKNOWLEDGMENT
This work was supported by Contract No.
N00014-67-0239-0027 between the Off ice of Naval
Research and the University of Maryland.
REFERENCES
1. Walker, J.D. and R.R. Colwell. Microbial
Degradation of Model Petroleum at Low Tem-
peratures. Microbiol. Ecol. 1:63-95, 1974.
2. Walker, J.D. and R.R. Colwell. Factors
Affecting Enumeration and Isolation of Acti-
nomycetes from Chesapeake Bay and South-
179
-------
eastern Atlantic Ocean sediments. Mar. Biol. Petroleum-Degrading Microorganisms. Micro-
In Press, 1975a. bial Ecol. Submitted, 1975b.
3. Walker, J.D. and R.R. Colwell. Microbial De- 4. Todd, R.L. and T.J. Kerr. Scanning Electron
gradation of Petroleum: Enumeration of Microscopy and Microbial Cells on Membrane
Filters. Appl. Microbiol. 23:1160-1162, 1972.
Table 1. EFFECT OF INCUBATION TIME ON FILTER COUNTS OF HETEROTROPHIC YEASTS
Incubation
time
(days)
3
14
Dilution/Volume
filtered
(ml)
0/100
0/250
-2/10
0/100
0/250
-2/10
0/100
0/250
-2/10
Counts
Colgate
Creek
sediment
2
2
2
for
Colgate Eastern
Creek Bay
water sediment
1
3
<1
2
OQ3
<1
2
OG
<1
Eastern
Bay
water
1
3
2
3
2
3
aOvergrown with fungi
Table 2. EFFECT OF INCUBATION TIME ON FILTER COUNTS OF HETEROTROPHIC FUNGI
Incubation
time
(days)
3
7
14
Dilution/volume
filtered
(ml)
0/100
0/250
-2/10
0/100
0/250
-2/10
0/100
0/250
-2/10
Counts
Colgate
Creek
sediment
2
11
for
Colgate Eastern
Creek Bay
water sediment
5
9
<1
10
OGa
3
10
OG
Eastern
Bay
water
2
12
5
19
5
19
aOvergrown with fungi.
180
-------
Table3. EFFECTOR INCUBATION TIME ON FILTER COUNTS OF YEASTS ON OIL AGAR
Incubation
time
(days)
3
7
14
Dilution/volume
filtered
(ml)
0/100
0/250
-2/10
0/100
0/250
-2/10
0/100
0/250
-2/10
Counts for
Colgate Colgate Eastern
Creek Creek Bay
sediment water sediment
<1
<1
<1 <1
<1
OGa
<1 <1
5
OG
1 <1
Eastern
Bay
water
<1
<1
1
2
1
2
aOvergrown with fungi.
Table 4. EFFECT OF INCUBATION TIME ON FILTER COUNTS OF FUNGI ON OIL AGAR
Incubation
time
(days)
3
7
14
Dilution/volume
filtered
(ml)
0/100
0/250
-2/10
0/100
0/250
-2/10
0/100
0/250
-2/10
Counts
Colgate
Creek
sediment
<1
3
3
for
Colgate
Creek
water
0
2
2
OGa
OG
OG
Eastern Eastern
Bay Bay
sediment water
0
0
<1
3
6
<1
3
6
<1
aOvergrown with fungi.
181
-------
Table 5. COMPARISON OF REPLICATES FOR DUPLICATE PLATE COUNTS OF CHESAPEAKE
BAY BACTERIA
Bacteria/ml
Replicate 1
5.2 x 105
1.7 x 104
6.4 x 102
1.4 x 102
2.3 x 103
7.3 x 104
3.0 x 101
2.8 x 102
4.3 x 104
1.7 x 103
1.0 x 105
Replicate 2
4.7 x 105
2.7 x 104
1.0 x 103
1.2 x 102
1.9 x 103
7.1 x 104
4.0 x 101
2.0 x 102
4.2 x 104
1.5x 103
1.6 x 105
Percent comparison
90.4
62.9
64.0
85.7
82.6
97.3
75.0
71.4
97.7
88.2
62.5
Table 6. COMPARISON OF REPLICATES FOR DUPLICATE FILTER COUNTS OF CHESAPEAKE
BAY BACTERIA
Bacteria/ml
Replicate 1
30
15
33
44
51
45
50
Replicate 2
20
10
25
37
39
42
41
Percent comparison
66.6
66.6
75.6
84.1
76.5
93.3
82.0
182
-------
Table 7. COMPARISON OF PLATE AND FILTER COUNTS OF HETEROTROPHIC YEASTS
Inoculum
Sediment
Sediment
Sediment
Water
Water
Water
Sediment
Sediment
Sediment
Water
Water
Water
Source Plate
Colgate Creek 3.0 x 101
5.0 x 10°
5.0 x 101
Colgate Creek 1.0x10°
<10°
5.0 x 10°
Eastern Bay <101
<101
<101
Eastern Bay <10°
<10°
<10°
Yeasts/ml
Filter
1.7x 102a
1.0x 101
2.0 x 102
1.7x 101b
4.0 x 10'1b
1.0 x 10°b
5.0 x 10°
<^101
<^1Q1
1.0x 10'2C
5.0 x 10-3C
2.0 x 10-2C
Percent
comparison
17.6
50.0
25.0
5.8
N.D.d
20.0
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
aResults obtained by filtering 10 ml of a 10~2 dilution of sediment samples in this and subsequent tables
for yeasts and fungi.
bResults obtained by filtering 100 ml of Colgate Creek water in this and subsequent tables for yeasts and
fungi.
cResults obtained by filtering 1000 ml of Eastern Bay water in this and subsequent tables for yeasts and
fungi.
determined.
TableS. COMPARISON OF PLATE AND FILTER COUNTS OF HETEROTROPHIC FUNGI
Inoculum
Sediment
Sediment
Sediment
Water
Water
Water
Sediment
Sediment
Sediment
Water
Water
Water
Fungi/ml
Source Plate
Colgate Creek 1.0 x 103
7.0 x 102
1.0 x 103
Colgate Creek 1.5 x 10°
<1QO
5.0 x 101
Eastern Bay <101
6.0 x 101
<101
Eastern Bay <10°
<1QO
1.0 x 10°
Filter
2.0 x 103
9.0 x 101
1.0 x 103
1.7 x 10°
2.5 x 10°
1.0 x 10-1
<101
5.0 x 101
3.0 x 102
5.0 x 10-2
7.0 x 10~2
5.0 x 10'2
Percent
comparison
50.0
12.9
100.0
88.2
N.D.
0.2
N.D.
83.3
N.D.
N.D.
N.D.
5.0
183
-------
Table 9. COMPARISON OF PLATE AND FILTER COUNTS OF YEASTS ON OIL AGAR
Inoculum
Sediment
Sediment
Sediment
Water
Water
Water
Sediment
Sediment
Sediment
Water
Water
Water
Table 10.
Yeasts/ml
Source Plate
Colgate Creek 3.0 x 101
1.0x101
Colgate Creek 1.0x 10°
<1QO
<10^
Eastern Bay 5.0 x 10°
<10^
-------
Table 11. COMPARISON OF PLATE AND FILTER COUNTS OF HETEROTROPHIC
ESTUARINE BACTERIA
Inoculum
Sediment
Sediment
Sediment
Sediment
Sediment
Sediment
Water
Water
Water
Water
Sediment
Sediment
Sediment
Water
Source Plate
Colgate Creek 3.8 x 105
4.5 x 106
6.4 x 105
2.8 x 104
3.2 x 106
1.9 x 105
Colgate Creek 2.9 x 104
9.1 x 104
2.5 x 103
8.1 x 104
Eastern Bay 9.3 x 104
4.7 x 105
1.6 x 104
Eastern Bay 2.2 x 102
Bacteria/ml
Filter
2.6 x 105
4.0 x 105
1.0 x 105
3.2 x 104
7.0 x 105
2.5 x 104
1.2 x 105
6.7 x 104
4.8 x 103
6.8 x 104
5.6 x 104
2.0 x 104
1.0x 104
3.0 x 101
Percent
comparison
68.4
8.8
15.6
87.5
21.9
13.2
4.1
73.6
52.0
84.0
60.2
4.2
62.5
13.6
185
-------
Table 12. COMPARISON OF PLATE AND FILTER COUNTS OF BACTERIA ON SILICA GEL-OIL
MEDIUM
Bacteria/ml
Inoculum
Sediment
Sediment
Sediment
Sediment
Sediment
Water
Water
Water
Water
Water
Sediment
Sediment
Sediment
Sediment
Sediment
Water
Water
Water
Water
Water
Source Plate
Colgate Creek <102
<102
<102
<102
<102
Colgate Creek 5.0 x 101
3.5 x 102
1.5x 102
1.5 x 102
5.0 x 101
Eastern Bay <102
<102
<102
<102
<102
Eastern Bay <102
1.0 x 10°
<10°
<1QO
<1QO
Filter
3.0 x 101
3.0 x 101
1.7 x 102
2.0 x 101
1.0x 101
<1.0x 10-1
<1.0x 10-1
<1.0x 10-1
<1.0x 10-1
6.1 x 10°
1.8 x 102
5.0 x 10°
<1.0x 10-1
<1.0x 10'1
<1.0x 10-1
9.0 x 10-2
<1.0x 10"2
<1.0x 10'2
7.5 x 10-2
5.0 x 10"2
Table 13. COMPARISON OF FILTER AND PLATE COUNTS OF HETEROTROPHIC MARINE
BACTERIA
Bacteria/ml
Inoculum
Sea water concentrate
Sea water concentrate
Sea water concentrate
Sea water concentrate
Sea water concentrate
Sea water concentrate
Sea water concentrate
Sea water concentrate
Sea water concentrate
Sea water concentrate
Plate
3.4 x 105
1.5x 105
1.8x 106
5.0 x 105
9.0 x 104
2.0 x 106
3.0 x 105
2.6 x 103
6.2 x 102
1.1 x 104
Filter
4.8 x 105
5.0 x 104
1.8x 105
1.4 x 105
7.4 x 104
5.6 x 104
6.0 x 103
7.1 x 101
7.2 x 102
3.5 x104
Percent comparison
70.8
33.3
10.0
28.0
82.2
2.8
2.0
2.7
86.1
31.4
186
-------
Table 14. COMPARISON OF FILTER COUNTS OF ESTUARINE BACTERIA OBTAINED ON
MILLIPORE AND NUCLEPORE FILTERS
Inoculum
Sediment
Sediment
Sediment
Sediment
Sediment
Sediment
Sediment
Water
Water
Water
Water
Sediment
Sediment
Sediment
Sediment
Sediment
Sediment
Sediment
Water
Table 15.
Source Millipore
Colgate Creek 2.6 x105
4.0 x 105
1.0x 105
3.2 x 104
7.0 x 105
2.5 x 104
20
Colgate Creek 1.2x 105
6.7 x 104
4.8 x 103
6.8 x 104
Eastern Bay 5.6 x 104
30
2.0 x 104
16
1.0 x 104
6
3
Eastern Bay 3.0 x 101
Bacteria/ml
Nuclepore
2.8 x 105
6.0 x 105
1.0x 105
3.2 x 104
1.9 x 106
1.0 x 104
8
7.9 x 104
8.3 x 104
3.4 x 103
5.3 x 104
7.4 x 104
30
1.0 x 104
2
1.2x 104
12
3
1.0x 101
COMPARISON OF FILTER COUNTS OF HETEROTROPHIC MARINE
OBTAINED ON MILLIPORE AND NUCLEPORE FILTERS
Bacteria/ 100 ml on
Inoculum Millipore
Sea water 360
Sea water 162
Sea water 260
Sea water 268
Nuclepore
340
147
106
100
Percent
comparison
92.8
66.6
100.0
100.0
36.8
40.0
40.0
65.8
80.7
70.8
77.9
75.7
100.0
100.0
12.5
83.3
50.0
100.0
33.3
BACTERIA
Percent
comparison
94.4
90.7
40.8
37.3
187
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/9-77-024
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
Proceedings of the Symposium on the Recovery of
Indicator Organisms Employing Membrane Filters
5. REPORT DATE
September 1977 issuing date
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Robert H. Bordner, Clifford F. Frith and John A. Winter
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Environmental Monitoring & Support Lab.
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
- Cin., OH
10. PROGRAM ELEMENT NO.
.1BD612A
11. CONTRACT/GRANT NO.
In-House
12. SPONSORING AGENCY NAME AND ADDRESS
SAME AS ABOVE
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
EPA/600/06
15. SUPPLEMENTARY NOTES
Held at the meeting of the American Society for Testing and Materials, Committee D-19
on Water, Ft. Lauderdale, Florida. January 20-21, 1975.
16. ABSTRACT
The Symposium on the Recovery of Indicator Organisms Employing Membrane Filters
brought together users, manufacturers, research scientists and representatives of
government agencies to exchange technical information and review the performance of
membrane filters for water and wastewater analyses. Problems with the recovery of
bacterial indicators had been reported;they were most pronounced in the fecal coliform
test. A key question was whether the cause was differences in sample types, membrane
filters or the test method employed.
Professionals experienced in water analysis presented relevant field experiences,
laboratory data and research findings and discussed problems concerning recovery of
organisms stressed or injured by environmental factors. Media, transport phenomena,
physical and chemical characteristics of membranes, membrane sterilization methods,
incubation temperatures, techniques for comparison of methods, data analysis, and the
status of the proposed ASTM methods for evaluating membrane filters were discussed.
Solutions suggested at the Symposium included use of two-step incubation, overlay
and/or enrichment techniques and modification of membrane filter structures. Recom-
mendations were made to manufacturers and to users to develop and improve intralabora-
tory quality control programs, to standardize interlaboratory testing procedures, to
participate in these collaborative studies and to generally improve communications
among users, manufacturers and standard-setting organizations.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Aquatic microbiology
Coliform bacteria
Indicator organisms
Membranes
Fluid Filters
Water
Water analysis
Water
Water
Tests
microbiology
pollution
Bacterial tests
Environmental factors
Fecal coliforms
Fecal streptococci
Membrane characteristics
Recovery of bacteria
Stressed microorganisms
06L
06M
13B
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (ThisReport)
UNCLASSIFIED
21. NO. OF PAGES
200
20. SECURITY CLASS (Thispage)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
188
U. S. GOVERNMENT PRINTING OFFICE: 1977-759-092
-------
-------
0) to
^ CB
o 5
? CD'
Go
9- 3
si
o- CD g.
5. ? 3
ft) 0]
C "*
""s ^D
3 o
o
CD
CD CD
"S3
o
"
§
8-
o
r
\
>j
z
rn
D
C
|
0
PPORTU
z
H
m
D
n
O
m
m
Z
p
H
"S
-n
O
"0
j>
H
m
C
CO
m
CO
o
0
0
Tl
Tl
ICIAL BU
0)
Z
m
0)
en
o
^
5'
a>
O
^"
o'
en
O)
00
i
CD
O
IT
O
0)
o'
B
0>
o'
ZJ
CO
(-*
QJ
g
CD
O
*»
33
esearch
0)
Q.
Develoi
T3
3
CD
i-^-
C
m
55
O
z
<^
m
z
H
T)
3)
O
1
ECTIOh
£
O
m
CD
o
o
CO
-Q
CD
0
"
_
T]
0
C
-1
C
w
m
Z
^
o
z
m
Z
J >
i r
CJ Tl
Tl
O
(f)
-\
>
C)
m
>
O
o
n
(A
01
01
O m
H m
rn CO
0 13
H
O
Z
>
D
Q
m
Z
O
------- |