STUDEES ON THE MEMQRANE FiLTER TECHNIQUE AND ITS APPLECATION TO THE DETECTEON QF COLIFQRM QRGANESMS FROM WATER “nests For Hm Degree 05 pk. D. MICHEBAN STATE UNE‘JERSITY A. Marie McCarthy 1956 This is to certify that the thesis entitled Studies on the Membrane Filter Technique and Its Application to the D tection of Coliform Organisms from Water. presented by a. I-farie I-charthy has been accepted towards fulfillment of the requirements for Doctor of Philoszonh‘: degree in i-‘icrobiologjy and Public Health 1‘ - "\N v 0-169 STUDIES ON THE MEMBRANE FILTER TECHNIQUE AND ITS APPLICATION TO THE DETECTION OF COLIFORM ORGANISMS FROM WATER BY A. Marie McCarthy A THESIS Submitted to the School of Advanced Graduate Studies of Michigan State University of Agriculture and Applied Science in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Microbiology and Public Health 1956 VITA A. Marie McCarthy candidate for the degree of Doctor of PhilosOphy Dissertation: Studies on the MeMbrane Filter Technique and Its Application to the Detection of Coliform Organisms from Water Outline of Studies: Major subject: Microbiology and Public Health Minor Subject: Mycology Biographical Information: Born: April 2, 1924, Worcester, Massachusetts Undergraduate Studies: Clark University, Worcester, Massachusetts, 1942-1946; A.B., Biology, 1946. Graduate Studies: University of Massachusetts, 1947- 1949; M.S., Bacteriology and Public Health, 1949. Research Professor, Dr. J. E. Fuller. Thesis Title: Biological Studies of Coliform Bacteria from Lake Waters Michigan State University, 1951- 1956; Ph.D. 1956. Research Professor, Dr. W. L. Mallmann. Experience: Bacteriologist-Serologist, Worcester, Mass. City Health Dept. Laboratory, 1948 and 1949 (summers); Research Associate in Bacteriology, Amherst College, Amherst,Massachusetts, 1949- 1951; Research Associate, Marine Biological Laboratory, Woods Hole, Massachusetts, 1949- 1951 (summers); Graduate Teaching Assistant, Affiliations: University of Massachusetts, 1948-1949; Graduate Research Assistant, Michigan State University, 1951-1952; Graduate Teaching Assistant, Michigan State University, 1952- 1956. The Society of American Bacteriologists The Society of the Sigma Xi Sigma Delta Epsilon The Mycological Society of America The American Association for the Advancement of Science The American Public Health Association A. Marie McCarthy ABSTRACT A detailed study was made of the membrane filter technique for the purpose of evaluating its usefulness in determining the sanitary quality of water supplies. During this investigation certain technical problems arose, the consideration of which was essential to the application of this technique. Among those considered, the problem of moisture control was given special attent- ion. Studies, using a pure culture of Escherichia 221i, showed that a full recovery of these organisms could not be obtained with the membrane filter technique When EHC modified Endo medium was employed. The need for a more satisfactory selective medium was recognized and a search was undertaken in an attempt to formulate such a medium. Several different media formulations were compounded and tested. A medium was deve10ped which was found superior to EHC modified Endo medium for the examination of certain types of waters. This medium derived its selective action primarily from.bile salts and incorporated brom cresol purple which acted as an indicator to detect lactose fermentation. A. Marie McCarthy Difficulties in the use of the membrane filter were encountered as a result of the precipitation of iron from some samples. Although iron precipitation could be pre- vented by the addition of Versene, the presence of this com- pound further complicated the interpretation of the results obtained. It was found that the ability of an organism to fer- ment lactose with gas production could not be accurately predicted from its colonial appearance on the surface of the membrane filter. The results of the study raised questions as to the adequacy of the membrane filter technique for evaluating the sanitary bacteriological quality of waters. ACKNOWLEDGEMENT The author wishes to express her sincere thanks to Dr. W. L. Mallmann whose guidance, interest, and suggestions contributed greatly in this investigation. Grateful acknowledgement is also due to Mr. I. L. Dahljelm for his advise and valuable help in media formulations; and to Dr. H. J. Stefseth, Head of the Department of Microbiology and Public Health. The writer appreciates the aid of the Millipore Filter Corporation, Watertown, Massachusetts, in the compilation of reference and illustrative materials; and the Difco Company, Detroit, Michigan, for their media contributions. The author also wishes to thank Miss B. Tinklepaugh for her helpful encouragement and the preparation and typing of the manuscript. TABLE OF CONTENTS Page INTRODUCTION OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO REVIEW OF LITERATURE .00.....0...OOOOOOOOOOOOOOOOOI... MATERIALS ANDMETHODS 0.00.0000...OOOOOOOOOOOOOOOOOOOO Membrane Filters OOIOOCOOOOOOOOOOOOOOOOO00.0.0000... Filter Apparatus 000000000000.0000000000000000.00000 Sterilization Procedures ........................... Filtration Procedure 0.0.0.0000...OOOOOOOOOOOOOOOOOC Standard or comparison OOOOOOOOOOOOOOOOOOOOOOOOOOOOO COIlGCtion Of samples 0.0.000...OOOOOOOOOOOOOOOOO... EXPERIMENTAL STUDIES AND RESULTS ..................... Preliminary Experimental Problems .................. I. II. III. IV. Regulation of Humidity During Incubation .... Difficulties Encountered in the Preparation Of EEC MOdi-fied EIldO medim OOOOOOIOOOOOOOOOO Recovery of Escherichia coli with EHC Endo Medium and Comparison of the Membrane Filter Count with the Standard Plate Count ......... Recovery of Escherichia coli with M-Enrichment Broth and Comparison of the Membrane Filter Count with the Standard Plate count CO...OOOOOOOOOOOOOOOOOOOOOOOOO... Media Formulation Studies .......................... I. Formulation Number One (F-l) ................ 1 3 15 15 1e 16 19 25 24 25 25 25 27 30 34 54 Page II. Formulation Number Two (F-2) ............... 36 III. Formulation Number Three (F-3 EMB) ......... 57 1. Recovery of Escherichia coli on F-S EMB Medium and Comparison of the Membrane Filter Count with the Standard Plate CO‘mt OCOOOOOOOOOOOOOOOOOOO0.00.00.00.00 40 2. Recovery of Coliform Bacteria from Water on F-S EMB and Comparison of the Membrane Filter Count with the Multiple- tube MPN .0.0000......OOOOOOOOOOOOOOOOOO 40 IV. Formulation Number Four (F-4) .............. 45 V. Formulation Number Five (F-5) .............. 44 1. Recovery of Escherichia coli on F45 Medium and ComparISOn of the Membrane Filter Count with the Standard Plate Colm‘b .00....OOOOOOOOOOIOOOOOOOOOOO0.00.. 46 2. Recovery of Coliform Bacteria from Water on F-5 Medium and Comparison of the Membrane Filter Count with the Multiple-tube MPN ...................... 46 VI. Formulation Number Six (F‘G) coo-00000000000 49 1. Results Obtained with F—6 Medium in the Determination of the Number of Coliform Bacteria in Water Samples .............. 50 2. Gas Production in Lauryl Tryptose Broth of Isolates from F-6 Medium and from EEC Modified Endo Medium................ 55 DISCUSSION 0.00....OOOOOOOOOOOOOOOOOOOOOOO0.00.0.0... 57 SUMMARY AND CONCLUSIONS ............................. 65 BIBLIOGRAPHY C.0.0000000000000IOOOOOOOOOOOOOOOOOOOCOO 69 TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE II. III. IV. VI. VII. VIII IX. LIST OF TABLES Page Membrane Filter Results with EHC Endo Broth Compared to the Standard Plate COUIIt OOOOOOOOOOOOOOOOOOO0.0.0.0000 31 Membrane Filter Results with M- Enrichment Broth Compared to the Standard Plate comt OOOOOOOOOOOOOOOOOOOOOOOOOOOC. 35 Membrane Filter Results with F-l Medium Compared to the Standard Plate Count .... 55 Membrane Filter Results with F-S EMB Compared to the Standard Plate Count .... 41 Membrane Filter Results with F-4 Medium Compared to the Standard Plate Count .... 44 Membrane Filter Results on Water Samples with F-5 Medium Compared to the Standard Methods' Confirmed MPN .................. 48 Number of Coliforms Estimated Per 100 Ml of Sample-~Red Cedar River ....... 52 Number of Coliforms Estimated Per 100 Ml of Sample-~Lake Lansing .......... 55 Coliform Estimations on Contaminated wells 0..0......OOOOOOOOOOOOOCOOOOOO0.... 56 Figure Figure Figure Figure 1. 3. 4. LIST OF FIGURES Page Pyrex Filter HOIder .COOCOOOOCCOOCOOOOOOO 17 Assembled Filter Apparatus .............. 18 Bacterial Colonies on the Membrane Filter (Magnification 1.4K) ............ 21 Bacterial Colonies on the Membrane Filter (Magnification 10.7X) ........... 22 INTRODUCTION The membrane filter is a relatively new tool in the field of bacteriology. Since its release for general use by the U. 3. government in 1951, the application of the membrane filter to the bacteriological analysis of water samples has been very widespread. In fact, the membrane filter has found its most extensive use in the detection of coliform'bacteria from.water. The object of this investigation was to make a de- tailed study of the membrane filter technique and to at- tempt to evaluate its usefulness in the sanitary testing of water samples. Many articles have been written recently Which credit the membrane filter technique with.having many advantages over the Standard Methods' MPH technique for determining the sanitary quality of a water supply. Most of these re- ports depict the merits of this new technique and few men- tion its faults. In the experience of the author with the membrane filter technique, however, many problems were en- countered which had not been answered. This fact suggested the exigency for making a further study of the various as- pects of the membrane filter technique for testing the sani- tary quality of water. The major difficulty which is encountered in the appli- cation of the membrane filter technique to the bacteriological analysis of water lay in the lack of a satisfactory selec- tive medium which would permit the complete recovery of all coliform organisms present in water. Although the develop- ment of a successful medium formulation may be a highly per- plexing task, the need for such a formulation was great. For this reason, studies were undertaken in an attempt to de- velop a culture medium which would support the growth of all coliform bacteria on the membrane filter. REVIEW OF LITERATURE Collodion membranes have been used in biological in- vestigations for a number of years. Fick (1885) utilized such membranes in dialysis experiments. Sanarelli (1891) was the first to use collodion membranes for bacteriological work. His work included the ultrafiltration of blood plasma in zizg. He showed that these membranes were impervious to bacteria but not to their toxins. Ferry (1936) stated that Bechhold first produced collodion membranes of graded pore sizes in 1906. Disc membranes made of collodion were pro- duced by Bigelow and Gemberling (1907). This process was further deve10ped by Zsigmondy and Bachmann (1918). The latter were the first to develop a method whereby the mem- branes could be produced on a commercial scale and their method of preparation was patented. Elford (1951) used a similiar process to produce membranes on a smaller scale in England. According to Costs and Tsuneishi (1951), Grabar also produced sueh.membranes in France. Thus an interest in collodion membranes was again stimulated during the early 1950's. Schutz and Kruse (1947) credit the Russians as being the first to use membrane filters for estimating the coliform con— tent of water.1 Apparently they used an Endo medium for this 1 The terms membrane filter, molecular filter, milli- pore filter, and ultrafilter are synonymoun. The term millipore filter is a trade name used by the Millipore Filter Corporation, Watertown, Massachusetts. purpose in 1953-1954. During World War II, the Germans used the membrane filter technique as an emergency procedure for the sanitary testing of water supplies. The fact is, they were forced to use it due to serious shortages of both agar and laboratory equipment. They seemingly had such success with the technique during the war that a widespread interest in it was deve10ped. Costs (1947) investigated the use of the meme brane filter technique in Germany and introduced this tech- nique into the United States. According to Goetz and Tsuneishi (1951), Goetz later deve10ped a new method for the production of membrane filters in the United States. These membranes were described as an improvement over those which had previously been produced in Germany. A procedure was described by Clark.g§'gl. (1951) for the estimation of coliform bacteria in water by the use of the mem- brane filter. They developed an Endo-type medium.for this pur- pose. These authors found that a significant increase was ob- tained in the numbers of coliform organisms recovered when in- cubation on the Endo medium was preceeded by a two hour pre- liminary enrichment period on a complete medium. This pro- cedure and the use of an Endo broth was also advocated by Costs and Tsuneishi (1951), Clark and Kabler (1952), and Kabler and Clark (1952) for the detection of coliform bacteria in water. These authors considered EHC Endo broth to be a satisfactory medium for use with the membrane filter in the enumeration of coliform organisms. Clark and Kabler (1952) stated that, "A few of the A, aerogenes and intermediates may be missed by this technique but the increased accuracy of the membrane procedure compensates for this loss when the data are compared with the MPN method". A complete description of the molecular filter technique for the bacterial examination of water is also given by McKee gt‘gl. (1955). Goetz, Gilman, and Rawn (1952) demonstrated the use of the molecular filter in determining the coliform density of ocean water. This technique was compared with the MPN procedure as given in Standard Methods. These authors contended that the molecular filter technique showed greater accuracy than stand- ard procedures and they pointed out the large deviations that can occur in coliform determinations which are made by the MPN procedure. 'Yet, in almost every case, their results showed the coliform estimation to be larger by the MPN procedure than.by the molecular filter technique. It might also be mentioned that the above authors tried to convey the impression that the molecular filter technique is a very simple Operation "as con- trasted with the more delicate and time-consuming technique of the lactose broth fermentation tests". Goetz (1955) has also proposed the use of the molecular filter technique for the bacterial assay of sewage and waste waters. Clark gt_al, (1952) compared domestic and EurOpean mo- lecular filter membranes. They described both the total re- coveries and the coliform densities obtained as being compar- able on the two types of membranes. However, they found that the domestic type membranes were somewhat easier and more con- venient to handle. A modification of an Endo medium to which was added 8- hydroxyquinoline, more commonly referred to as oxine, was emw ployed by Yee, Krabek, and Schaufus (1955). They reported that the addition of oxine reduced the number of non-coli- form colonies on the molecular filter membrane by approximately 25-50 per cent without causing any reduction in the number of coliform colonies present. These authors claimed that they generally obtained higher coliform counts with this mediwm and the molecular filter technique than were obtained by the MPN procedure. However, it should be pointed out that the above authors allowed the filters to dry before counting the colonies because this procedure resulted in higher coliform counts. This made it impossible to utilize confirmatory tests to determine whether or not the dried colonies which were counted were actually coliform colonies. Hajna and.Ihmon (1954) have deve10ped.a medium which con- tains sodium.desoxycholate and which does not require a pre- liminary enrichment period. This medium.is essentially EHC Endo broth to which sodium desoxycholate has been added. In 1954, Jeter, Geldreich, and Clark introduced a bril- liant green fuchsin medium (BGF medium) for use in examining waters which contained large numbers of non-coliform bacteria. With this medium, their coliform recovery was 8 per cent less than that which was obtained with EHC Endo broth. In this study, both EHC Endo and BGF medium gave lower coliform esti- mates when used with the membrane filter than were obtained by means of the Standard Methods MPN procedure. The results of Geldreich 22 21° (1955) likewise showed that coliform counts which were made by the membrane technique tended to be appreci- ably lower than those which were obtained by the MPN procedure. The findings of Kabler (1954) are in agreement with these ob- servations. Jeter, Geldreich, and Clark (1954) also pointed out that, in the case of the EHC Endo medium, 8 per cent of the colonies which were considered to be composed of coliform bacteria failed to produce gas When transfers were made to lactose broth fermentation tubes. On the other hand, 4 per cent of the colonies which were not considered to be composed of coliform bacteria were shown to produce gas from.lactose broth. The same was true with the BGF medium. Here 5 per cent of the colonies which were counted as coliform colonies failed to produce gas when transferred to lactose broth fer- mentation tubes, while 7 per cent of those which were not con- sidered to be coliform colonies on this medium produced gas when transferred to lactose broth. However, the authors Justified this situation by stating that, "These two sources of error tended, in this investigation, to cancel one another in coliform enumeration". Kabler (1954) compared the results which were obtained by the membrane filter technique with those which were obtained by the Standard Methods 5-tube MPN procedure in the examina- tion of water samples for coliform bacteria. He reported on the testing of 1,706 water samples by the two methods during which a comparison was made "on the basis of the 95 per cent confidence limits of the confirmed MPN”. He found that, when compared on this basis, 1,260 samples or 75.8 per cent were in agreement. However, after making the statement that, "Sources selected were to show a coliform MPN index between 10 and 200 per 100 ml.", Dr. Kabler included in his results 457 samples in which no coliform organisms were demonstrated by either procedure. In addition to this, he included 114 samples which gave an MPN of less than 5.4 coliforms per 100 ml. Of these 114 samples, 85 showed the presence of coliform organisms on the MPN procedure but not by the mem- brane filter technique, while 29 gave coliform organisms by the membrane technique but not by the MPN procedure. The fact that Dr. Kabler had no vindication for including the samples which showed no coliform bacteria by either procedure in the determination of his final percentages should be considered. It might be pointed out that, in the case of these negative samples, an agreement of 100 per cent could have been obtained even if the membrane filters were not incubated on any nutrient material whatsoever. If these 457 negative samples are not in- cluded, then it can be shown that only 64.5 per cent of the samples and not 75.8 per cent, as claimed by Kabler, are in agreement when compared on the basis of the 95 per cent con- fidence limits as indicated. It should also be mentioned that attempts to compare results on the basis of the 95 per cent confidence limits of the confirmed MPN do not really give any kind of a statistical comparison. Such an attempt is merely an endeavor to indicate an agreement in a case where evidently no good statistical comparison can be shown to exist. Accord- ing to Bush (1955) it is on the basis of these results Which have been reported.by Kabler (1954) that the membrane filter technique has been included as a tentative procedure in the tenth edition of "Standard Methods for the Examination of Water, Sewage and Industrial Wastes" (1955). Presnell, Arcisz, and Kelly (1954) compared the membrane filter and MPN methods in determining the coliform densities of samples of sea water. The 95 per cent confidence range of the MPN was used as a basis for comparison. Their results were 87.1 per cent in agreement when compared in this way. Levin and Laubausch (1954) reported on the testing of 207 samples of drinking water which.had.been delivered to "hotel" trains under emergency conditions. Of these 207 samples, only three showed the presence of any coliform bacteria on the confirmed MPN test, while two (of these same three) also produced coliformrlike colonies on the membrane filter. On the basis of these tests the authors concluded that, "It seems entirely feasible that, with the use of the membrane filter, taco 19.1.39 ed: to ale: vee'evig {Lise-x from 06 w $53926!!! a! :qmedds means;- viaaebtve ez‘edw'veaeo a at 47200011 .381” 0:! mode 2 Melch'aiwce'x east-S3 ‘10 ; «JIM candies: on: Jan. odd» HIE flabeoowq evids; ‘ r to momentum add an " .(aee.c) mammals became (£- ~8M¢fimb 30111.03 ed: : To cam-w sometime Jase - quarantined! .aeen - 4W 63-18, at -he'xsqmoo 3043329803 out no bed as! 'Wl'fobii‘md be 3 ms mm 10 .eaoi 8W .mfltlm' fills '10. 9::- ~ MI-«Cplfidbm‘twe’em its) ' fl .mflirtfll'zdflem a ”133’ dw‘bofinlorzoe Joann canines erfi "m - 10 small compact laboratories that could be carried on a man's back would be capable of providing bacteriological control over the quality of drinking water provided by makeshift means". Slanetz and Bartley (1955) reported a study of various selective media for the estimation of coliform‘bacteria in water. They found that their most satisfactory results were obtained by the use of M-Endo broth which was supplied by Difco and BBL Companies. They reported that these media gave higher coliform counts or more distinct coliform colonies than the other media tested. Taylor, Burman, and Oliver (1955) compared the membrane filter technique with the methods which are standard.procedure in Britain for the bacteriological analysis of water. They adapted a triple strength MacConkey broth which contained 12 times the usual amount of bromrcresol purple for use with the membrane filter. Their results were reported as promising and they appeared to have obtained a rather good agreement between the two methOds which were used. However, it is possible that some bacteria may have failed to develop due to the high in- cubation temperature which is required in their standard pro- cedures. These authors in a later report (1955) indicated further progress in this study. They used both Endo broth and a modified MacConkey‘broth with the membrane filter procedure. The results were compared with those obtained by the 5-tube method utilizing MacConkey broth and confirmation in brilliant- green bile broth which is the standard procedure that is used by 11 the British in testing the bacteriological quality of a water supply. They obtained fewer false positive results on the Endo medium, although a higher confirmed count was obtained with the MacConkey medium. Also, both media gave higher coli- form estimations when used with the membrane filter procedure than were obtained by the multiple tube technique. However, as the authors point out, the possibility must be considered that the high initial incubation temperature (42 C) which is used in the multiple tube procedure may have prevented the subsequent growth of some of the coliform bacteria. McCarthy (1955) used a two-hour incubation period on M-enrichment broth followed by M-Endo medium. He showed that when the results with the membrane filter were based on the appearance of a metallic sheen, the number of coliform bacteria obtained with this technique was considerably lower than that obtained by means of the MPN confirmed test. The author points out many of the difficulties and problems which are encountered in the use of the membrane filter technique. His work indicated the lack of a completely satisfactory medium which would fully permit the development and differentiation of coliform organisms from all waters. He emphasized that, "The most urgent need in connection with the membrane filter seems to be the formulation of a more satisfactory'mediump. McCarthy, however, did observe an excellent agreement between the membrane filter and the tube MPN test when sewage was examined, but he believed that this was due to the "freshness 12 of the coliforms in the sewage and the relatively low number of other organisms". The fact has been acknowledged by Thomas, Woodward, and Kabler (1955) that coliform densities which are obtained by the MPN method "are generally somewhat larger than those obtained by the membrane filter technic". These authors stated that, "While such differences may be statistically significant, it does not follow logically that the differences are of much practical import". They believed that "the MPN tends to overestimate the true density" and referred to the MPN as "a biased estimator of the true density". In addition, they made the statements that, "The arithmetic means of replicate MPN's tend to be too high by a factor of 25 per cent in a 5-5-5—tube test", and that, "If a large number of replicate 5-5-5-tube MPN tests are made from water containing 100 coli- form per 100 ml the arithmetic mean MPN will be 125 instead of 100". Nevertheless, it has been admitted by Thomas and Woodward (1955) that, in the case of certain waters, estimates in coli- form densities are greater with the MPN procedure than with the membrane filter technique and that the difference in densities is ”larger than may be accounted for by the bias of the MPN". However, the authors were of the Opinion that, "Some loss in recovery efficiency is acceptable and justifiable in view of the advantages gained by the new tool". 15 Thomas and Woodward (1955) have made it clear that, "The bias of the MPN derives only from.mathematical con- siderations and is independent of the growth efficiency of the media used". ‘Yet it is sometimes possible to isolate coli- form bacteria from previously inoculated lactose broth fer- mentation tubes in which no gas production was observed after an incubation period of 48 hours as advocated in Standard Methods. The work of Chambers (1950) has indicated that when a high ratio of non-coliforms to coliforms is present in a water sample, failure to produce visible gas in lactose broth may occur even though considerable numbers of coliforms are present in the sample. According to Chambers, "A negative test, to say nothing of one in which a small quantity of gas is produced, certainly does not preclude the possibility that coliforms were present". 0n the other hand, McCarthy (1955) took membrane filters which had no colonies with sheen, placed them.in lactose broth fermentation tubes, and confirmed all tubes showing gas. By this procedure, he was able to obtain an average MPN frum.mem- brane filters which was more than twice the MPN that was ob- tained when the results were based on the appearance of sheen colonies only. On the basis of results which have‘been reported thus far, the author was of the opinion that the membrane filter tech- nique had been pushed too fast and too far in the field of water bacteriology. Those favoring the acceptance of this 14 technique seem to have used every means at their disposal to make their findings appear better than they really were. In addition, this new tool seems to have held much appeal to technicians who are engaged in water analysis. Recently, this technique has been included as a tentative procedure in Standard Methods and although some of the pitfalls are pointed out, Standard Methods states that, "When the limitations of the test are fully recognized and the difficulties of interpreta- tion of the results are known, the technic may be used". However, Standard Methods does not intend that the membrane filter technique should be used as a substitute for the di- lution MPN procedure in the examination of drinking water. It was with these facts in mind that the experimental work was begun in this investigation. MATERIALS AND METHODS Membrane Filters The membrane filters which were used in this investi- gation were of the hydrosol assay, white, grid marked type. They were circular discs 47 mm in diameter and 150 microns in thickness and were composed of a cellulose derivative. The membranes resembled paper in appearance but in appearance only. (Anon.,1951 and 1955) These membranes were extremely porous and the pores were uniform in size. The pores comprised a volume of from 80 to 85 per cent of the total volume of the membrane. (Anon., 1955) The pore size was smaller on the upper surface of the mem- brane than on the lower surface. As a result of this prop- erty, filtration was achieved by means of a physical screen- ing action. (Goetz and Tsuneishi, 1951) The pores in the mem- brane filter are tubular and roughly parallel to the direct- ion of flow of a liquid during filtration. (Clark, 2£.El- 1952) The number of pores per square centimeter have been calculated at 50 million. (Bush, 1955) The pore size can be controlled during the manufacturing process. Goetz has reported that, "The pore size is measured by the determination of a factor 2 (zeit . time) which is necessary to pass 100 cc of distilled 'water at room.temperature through a filter area of 100 cm? at 15 16 a differential pressure of one atmosphere (15 lbs.)". (Goetz, 1947) The pore size has been calculated by the Millipore Filter Corporation to be 0.45 micron and the effective fil- tering area of these membranes is 9.6 cm2. (Anon., 1955) The upper surface of the membrane filter has been imprinted with a grid to facilitate counting. The membrane filters which were used in this study were manufactured by the Millipore Filter Gerporation, Watertown, Massachusetts. Filter Apparatus The filter apparatus used in this study was constructed of pyrex glass (Fig. 1). It consisted of a 250 ml glass funnel (part number 4) which was Joined to part number 2 by zneans of a clamp (part number 5). The membrane filter was supported on the fritted glass surface of part number 2. YEhe filter apparatus was mounted on a filter flask and a vacu- um was applied during the process of filtration (Fig. 2). The funnel was coated with Desicote1 so that no film of 'water would be left behind when the funnel was emptied. This allowed for a more complete rinsing of the funnel. Sterilization Procedures «The membrane filters were packed in groups of ten with ‘I’A product of Beckman Instruments, Inc., South Pasadena, California. 17 FUNNEL (PYREX) I ll PART=H= 4 %/ MP“ “'tupoae FILTER H0“ PARTO4 °.TMIAILI I lelm W”? ~ “We mu! “'|’| Md" WWW \ HOLDING CLAMP PART an 3 sssss wwwww ........... nnnnnnnnnnnnn oooooooooooooooooooooo .............. Figure l. MUMMOME :35: MM ram/a massacnusens warenowu 72 PYREX FILTER HOLDER CAT XX ID 047 00 FOR ANAYSIS OF HYDROSOLSrERs uses 47 MILIMETER TYPE HA FIL Figure 2. Assembled Filter Apparatus 19 absorbent pads between the filter discs, wrapped in Kraft paper, and sterilized in the autoclave at 15 lbs. pressure per square inch (121 C) for 10 minutes. At the end of this 10 minute period the steam pressure was rapidly released to prevent condensate from collecting on the membranes. Extra absorbent pads were wrapped in Kraft paper and sterilized as mentioned above. Both glass parts of the filter apparatus were wrapped separately in Kraft paper and sterilized in the autoclave at 121 C for 15 minutes. The filter apparatus was sterilized after each sample was examined. Forceps were sterilized prior to each use by dipping in 95 per cent ethyl alcohol and passing them.through the flame of a Bunsen burner. Filtration Procedure The following filtration procedure was employed: 1. A sterile membrane filter was placed with its grid side up on the fritted glass surface of the filter apparatus with the use of sterile forceps. The funnel was then attached by means of the holding clamp. 2. The desired volume of a water sample or of an apprOpriate dilution of a water sample was passed through the membrane filter. The funnel was rinsed three times using at least 50 ml of sterile distilled 5. 4. 5. 20 water each time. A vacuum of 50-40 cm.of mercury Which was provided by a central source in the building, was used to draw the water through the filter. The membrane filter was removed from.the filter appa- ratus with sterile forceps and placed on an absorbent pad which had previously been saturated with culture media. Care was taken so that no air bubbles would be trapped.between the membrane filter and the pad. The absorbent pads were contained in sterile 60 x 15 mm glass petri dishes. The petri dishes were incubated in an inverted position at a temperature of 55 C for whatever length of time was desired. The surface tension exerted kept the ab- sorbent pads in the bottoms of the petri dishes. In order to maintain an atmosphere of saturated.humidity during the incubation period, the petri dishes were placed in a plastic refrigerator tray the bottom of which had been lined with moist cheese cloth. This tray had a tight-fitting lid. During incubation the nutrient solution penetrated through the pores by means of capillary action to the bacteria which were deposited on the surface of the fil- ter. After a given period of incubation, colonies were observed on the membrane filter as shown in Fig. 3 and 4. A stereoscOpic microscOpe which.had a magnification of 10 diameters was used in counting the colonies. A light 21 Bacterial Colonies on the Membrane (Magnification 1.4X) Filter Figure 5. Figure 4. Bacterial Colonies on the Membrane Filter (Magnification 10.7X) 23 source which was nearly perpendicular to the mem- brane surface was used in making the counts. Standard of Comparison The results which were obtained in the estimation of coliform bacteria by the membrane filter technique were com- pared with estimations which were obtained simultaneously by means of the 5-tube, 3-decima1 dilution MPN confirmed test as given in the tenth edition of "Standard Methods for the Examination of Water, Sewage and Industrial Wastes" (1955). In the Standard Methods' procedures lauryl tryptose broth was used in the presumptive test and all tubes which showed any amount of gas within 48 hours were confirmed in brilliant- green lactose bile broth. Difco dehydrated media were used throughout in the Standard Methods' tests. The completed test was not carried out. The numbers of coliform organisms which were recovered by means of the two techniques will be given as numbers per 100 m1 of water sample. In some of the investigations, laboratory stock strains of Escherichia ggli were employed in testing various media ‘before these media were used in the cultivation of coliform “bacteria from raw waters. When this was done the standard plate count was used as the standard for comparison. The cultures were transferred daily in Brain Heart Infusion_broth (Difco) for at least three days prior to the date of testing. 24 ApprOpriate dilutions were made using 90 ml sterile saline dilution blanks. Five replicate plates were poured for each determination and an average of the five plates was taken as the standard plate count. Brain Heart Infusion Agar was used as the plating medium. An incubation period of 24 hours at 55 C was also used. Collection of Samples The samples were collected in sterile, wide-mouth bottles with ground glass stOppers, according to the pro- cedure outlined in Standard Methods. Prior to sterilization the tops of the sampling bottles were covered with Kraft paper and tied securely. All samples which failed to show the presence of coli- form.bacteria by either the membrane filter or the MPN pro- cedure were eliminated and these samples were not considered in the tabulation of results. ********** Where deviations frmm the above procedures occurred, znention will be made in the prOper places. EXPERIMENTAL STUDIES AND RESULTS Preliminary Experimental Prdblems Certain technical problems were encountered during early studies with the membrane filter technique. The con- sideration of these problems was essential to the successful application of this technique. Therefore, it was necessary to deal with each of these problems, separately, as they arose and to reach an immediate solution or conclusion with regard to each before further progress could be made in the subsequent studies to be undertaken. The nature of each of these problems is described.below. I. Regulation of Humidity During Incubation In the early studies which were made by the author, an attempt to maintain an atmosphere of saturated.humidity dur- ing incubation by inverting the bottoms of petri dishes (which.sontained the membrane filters) on a moist towel placed in a glass tray was found to be most unsatisfactory. This :method was suggested.by McKee gtmgl. (1950) and by Goetz and Tsuneishi (1951). When the membrane filters were incubated in.this manner, the towel was found to be completely dried after a period of 18 hours in the incubator at 550 and the 25 26 absorbent pads were also found to be in a partially dried state. This resulted in extremely low counts on the meme brane filter when a pure culture of Escherichia ggli was used. Such a result was to be expected since the nutrient was not supplied to the bacteria during the full incubation period. During this study, 10 replicate plate counts were made uti- lizing Brain Heart Infusion Agar (Difco) as a plating medium. At the same time, 10 replicate membrane filters were also run. EHC modified Endo medium was employed for use with the mem- brane filters. This was preceded by a 2-hour preliminary en- richment period on Albimi M broth (Clark 32.31., 1951). The average count per'membrane filter was found to be 51, whereas the average plate count was 170. Thus, the percentage of organisms recovered on the membrane filter, as compared to the plate count, was 50 per cent. The above experiment was repeated with the following changes. The petri dish bottoms, containing the membrane filters, were inverted on a moist towel as previously men- tioned. However, this time the petri dishes were covered with a second moist towel. A metal lid was then placed on top of the tray which contained the petri dishes and incubation was allowed to proceed in the ubual manner. In this case, the hue lnidity produced was apparently too great and, as a result, such extreme spreading occurred that it was impossible to make colony counts. 27 It was then discovered that a constant amount of mois- ture could be insured by the following procedure. The mem- brane filter was placed on an absorbent pad in the bottom of a petri dish. The petri dish was inverted and a sheet of moist filter paper was placed inside the cover. The dish was sealed with Scotch brand masking tape and placed in the incubator at 350. This method gave satisfactory results,‘but it became too laborious when one had many filters to run. Another procedure which gave reliable results and proved to be more practical is described below. In this procedure, incubation was carried out in a plastic refrigerator tray lined with moist cheese cloth and with a tight-fitting cover. Petri dishes, which contained membrane filters, were placed inside these trays in an inverted position. The trays were set in an incubator at 550. In the experience of the author, the most dependable results were obtained by this method. IHenceforth, in this study all membrane filters, when used 'with absorbent pads, were incubated in this manner. II. Difficulties Encountered in the Preparation of BBC Modified Endo Medium The EEG modified Endo medium.was prepared as directed in the tenth edition of "Standard Methods for the Examination of Water, Sewage, and Industrial Wastes"(1955). In the preparation of the basal medium, it was observed ‘that.Neopeptone (Difco) formed a precipitate in distilled water. 28 The amount of precipitate increased on boiling. It was possible to dissolve this peptone completely by the addition of acid, but, as soon as the pH of the solution approached neutrality, a precipitate immediately formed. Preliminary studies were made in which Bacto-Peptone, Bacto-Tryptose, Bacto-Tryptone, Proteose Peptone (Difco), Proteose Peptone No. 5 (Difco), and Peptone M (Albimi) were individually substituted for NeOpeptone. When a pure cul- ture of‘g.‘ggli was used, Bacto-Peptone, Bacto-Tryptose, and Baeto-Tryptone gave results which were similiar to those ob- tained with Neapeptone both in total recovery and in produc- tion of metallic sheen. However, the total number of tests run was not sufficient and no significant conclusions could 'be drawn. In preparing the complete medium.from liquid indicator solutions, some of the basic fuehsin crystallized when this dye was added to the sodium sulfite solution. The amount of crystallization varied depending upon the manner in which these ‘two solutions were combined. Rapid.mixing produced less crys- tallization, but some crystals were always formed. Thus the (puestion arose in the mind of the author as to whether or not ‘flhis condition could result in a uniform dye solution in all batches of finished medium. Powdered indicators also presented difficulties in pre- paration. Basic fuchsin, sodium sulfite, and lactose were ground to as fine a powder as was possible with a meter and 29 pestle. The homogeniety of this mixture, however, was highly questionable. When the complete medium was prepared from powdered indicators, the amount of dye which failed to go in- to solution was greater than could be accounted for by the amount of inert material initially present in the dye powder. III. Recovery of Escherichia coli with BBC Endo Medium and Comparison of the Membrane Filter Count with the Standard Plate Count Early investigations with a pure culture of g. £311 in- dicated that all of these organisms were not recovered when EHC modified Endo medium.was employed with the membrane filter technique. Therefore, the following study was undertaken to determine the recovery of E. ggli in the use of this technique when the above mentioned medium was utilized. In this study a series of 29 suspensions of a pure cul- ture of E. 231i were prepared in sterile saline dilution blanks. The number of organisms present in these suspensions was determined by both the membrane filter technique and the standard plate count. EHC modified Endo medium, prepared as directed in Standard Methods, was used with the membrane .filter technique. A 20-hour incubation period on this medium ‘was preceded by a 2-hour preliminary enrichment period on IMéenrichment broth (Difco). (Note: The criterion for the re- cognition of coliform colonies on EHO Endo medium is the de- 'veIOpment of a metallic sheen.) Brain Heart Infusion agar 30 (Difco) was used as the plating medium in determining the plate counts. Five replicate membrane filters and five re- plicate plate counts were prepared for each series. The results which are given in Table I are reported as averages of the five determinations made in each series. The average membrane filter count was 174 colonies, whereas the average plate count was 229 colonies. This gave a recovery of 76.0 per cent on the membrane filter as compared with the recovery on the standard plate count. The question arose as to Whether this loss in recovery was due to inhibitory effects of the medium used or Whether some of the organisms had been lost during the filtration process. Another question was also in- troduced here as to the possibility of obtaining a total re- covery of E.‘ggli with the membrane filter technique. In an attempt to answer these questions the following experimental study was conducted. IV. Recovery of Escherichia coli with M-Enrichment Broth and Comparison of;the Membrane Filter Count with the Standard Plate Count In this investigation M-enrichment broth, a non-selective Inedium, was used with the membrane filter technique to elimi- Jmate the introduction of any inhibitory substances. Otherwise 'the procedure was the same as that which was employed in the pre- ceeding section. A series of 23 suspensions of a pure culture of TABLE I MEMBRANE FILTER RESULTS WITH EHC ENDO BROTH COMPARED TO THE STANDARD PLATE COUNT Series Colony Count Per Plate Count MF* Count as Filter Percentage of Plate Count 1 161 210 _ 76.7 2 96 155 ’ 62.7 5 115 165 69.7 4 209 296 70.6 5 175 217 80.6 6 195 265 72.8 7 177 255 76.0 8 256 295 87.4 9 140 175 . 80.9 10 165 255 70.2 11 201 265 75.8 12 254 547 75.2 15 108 149 72.5 14 267 525 82.2 15 175 255 68.4 16 252 515 74.1 17 154 211 75.0 18 219 509 70.9 19 158 195 81.9 20 186 295 65.5 21 165 217 76.0 22 157 205 76.6 25 105 124 84.7 24 152 169 78.1 25 152 227 67.0 26 176 205 86.7 27 146 189 77.2 28 198 229 86.5 29 170 167 101.8 Total 5040 6628 Average 174 229 76.0 * Membrane Filter 51 52 ‘§.‘ggli was made and the recovery, Which was obtained with the use of M-enrichment broth and the membrane filter tech- nique, was compared with that obtained by the standard plate count utilizing Brain Heart Infusion agar. The results are summarized in Table II and they are re- ported as averages of the five determinations which were made in each series. The average membrane filter count was 197. The average plate count was 205. The result was a recovery of 97.4 per cent on the membrane filter as compared to the re- covery with the standard plate count. This indicated that a good agreement between these two procedures was possible. It also indicated that a satisfactory recovery can be obtained with the membrane filter technique if the medium which is used will allow the full development of the organisms desired. This verified the fact that the loss of 24 per cent in the recovery of E. ggli Obtained with the membrane filter in section III was due to the inhibitory effect which the EHO modified Endo ‘broth had upon these organisms. During this study, several attempts were made to recover ;§, 221i from the filtrates which were passed through the filters. ‘At no time were these attempts ever successful. This sub- stantiated the finding that the loss in recovery was not due to the fact that E. 921i was passing through the filters. The conclusion reached by the author was that the main difficulty encountered in the recovery of E..ggli with the meme ‘brane filter technique was due to the lack of a satisfactory I ‘4‘..\ .- J .- u 'l"t-—. s..;\ . . i ‘. . t 4-“ :J TABLE II MEMBRANE FILTER RESULTS WITH M-ENRICHMENT BROTH COMPARED TO THE STANDARD PLATE COUNT Series Colony Count Per Plate Count MFw Count as Filter Percentage of Plate Count 1 95 120 79.1 2 274 540 80.6 5 204 202 101.0 4 164 154 106.5 5 245 254 105.8 6 207 186 111.5 7 198 217 91.2 8 185 192 95.5 9 257 275 94.1 10 214 211 101.4 11 155 154 99.5 12 114 106 107.5 15 120 151 91.6 14 255 212 109.9 15 125 142 88.0 16 254 240 105.8 17 256 242 97.5 18 252 249 101.2 19 195 207 94.2 20 510 545 90.5 21 215 195 _ 111.4 22 155 165 95.9 25 142 151 94.0 Total 4525 4644 Average 196.7 201.9 97.4 * Membrane Filter 55 54 selective medium. For this reason, the search for the formur lation of such a medium was undertaken. Media Formulation Studies I. Formulation Number One (F-l) As a result of exploratory studies which were carried out, the F-1 medium formulation was developed. F-l BaCto-Tryptose O...O0....00.0.0.0..00000004000g BaCto-I'aCtose OOOOOOOOOOOOOO0.0.00000000002000 Yeast Autolysate (Albimi) ................ 6.0 Dipotassium.Phosphate .................... 2.75 MonOpotassium.Phosphate .................. 2.75 SOdiIm Chloride O...OOOOOOOOOOOOOOOOOOOOOO 5.0 SOdim Laury1 sulfate OOOOOOOOOCOOOOOOOOOO 0.2 Brom Thy-5101 Blue OOOOOOOOOOOOOO0.0.0.0.... 0.1 Distilled water 000......OOOOOOOOOOOOOOOOOOOOOOO 1000 ml KOH to give a pH of 7.0 Suspensions of‘g. 321$ were prepared and.the recovery which was obtained by the use of this medium.and the membrane filter technique was compared with that obtained by the stand- ard plate count. The results are given in Table 111. Each figure recorded represents the average of five determinations. In the study of the F-1 medium, the average colony count per membrane filter was 197 and the average plate count was 204. Thus, a recovery of 96.7 per cent was obtained on the membrane 55 filter as compared with the plate count. A satisfactory re- covery was observed with the F—1 medium, but the colonies of .§°.22l$ were colorless and did not exhibit any differentiating characteristics which would make it possible to differentiate between these colonies and colonies of non-coliform bacteria. For this reason, the F-1 medium was not used in any future study. TABLE III MEMBRANE FILTER RESULTS WITH F-l MEDIUM COMPARED TO THE STANDARD PLATE COUNT Series Colony Count Per Plate Count MF* Count as Filter Percentage of Plate Count l 197 211 .95°4 2 179 184 97.5 5 251 265 95.4 4 162 158 102.5 Total 789 816 Average 197 204 96.7 * Membrane Filter 56 II. Formulation Number Two (F-2) The possibility of using neutral red as an indicator for the detection of lactose fermenters from non-lactose fermenters growing on the membrane filter was considered. The F-2 medium was developed in an attempt to determine whether or not this indicator would give a satisfactory performance for use with the membrane filter. The F-2 medium.differed in the following way from.the F—l medium.which was used in the preceding section. The sodium.laury1 sulfate was reduced to 0.1 g and the brom thymol blue was replaced with 0.08 g of neutral red and 0.001 g of crystal violet. A suspension of‘g.‘ggli was made in sterile saline solu- tion. A total of ten replicate counts was made from this sus- pension utilizing the membrane filter technique and the F-2 :medium. An average of 179 colonies was obtained for the ten filters. A total of ten replicate standard plate counts, made simultaneously, yielded an average of 201 colonies. This gave a recovery of 89.1 per cent on the membrane filter as compared *with the standard.plate count. The F-2 medium.produced a pink coloration on the entire membrane filter as was expected. The colonies, however, be- came only very slightly colored and no differentiating charac- 'terdstics were observed. When acid was produced by the colo- zxies, whole areas of the filter surface changed color. In other words, any color changes produced were not confined to 57 the colonies. Because of this, neutral red was not an ade- quate indicator for use with the membrane filter. Other studies showed that Andrade's indicator was also un- satisfactory. When this indicator was incorporated into the F-1 medium, no hint of differentitation could be detected. The membrane filter and all colonies were white or colorless. III. Formulation Number Three (F-5 EMB) The following investigation was undertaken in an effort to deve10p an eosin methylene-blue medium which could be used successfully in the enumeration of coliform'bacteria with the membrane filter technique. It has been stated that, "Eosin- methylene-blue broth to date has not been satisfactory for membrane filter work because the eosinate of methylene blue apparently does not diffuse through the membrane." (Kabler and Clark, 1952) These authors had found it feasible to use this combination of dyes in broth "only after the membrane has been saturated with methylene blue and washed free of the excess dye before filtration". (Clark, gt 31., 1952) Several different media formulations were developed and investigated as this study was pursued. Numerous trials were made with various combinations of ingredients. Changes were also made in the composition of each medium under development by changing the concentration of each individual ingredient. The recovery of E. coli Which was obtained by the use of each 58 formulation deve10ped was compared with the recovery obtained by means of the standard plate count. These studies resulted in the deve10pment of F—5 EMB medium. F-5 EMB Base Medium BaCtO'Peptone eeeeeeeeeeeoeeee 20.0 g BECtO-LaCtose eeeeeooeeeeeeeoe 20.0 Bacto-Agar ................... 6.0 Eosin Y'(95% dye content) ............ 24 m1 (an 8% aqueous solution) Distilled Water ..................... 900 ml The base medium was dispensed in 90 ml quantities and sterilized in the autoclave at 121 C (15 lbs. pressure) for 15 minutes. Dipotassium Phosphate Solution Dipotassium Phosphate ........ 2.0 g DistilledVJater OIOOOOOOOOOOOOOOOOOOOO 153ml This solution was dispensed in convenient quantities and sterilized in the autoclave at 121 C for 15 minutes. Methylene Blue Solution Methylene Blue (88% dye content) ... 2.0 g Make up to 100 ml with distilled water Preparation 2; Finished Medium The finished medium was prepared by adding the following substances, in the amounts indicated, to 90 ml quantities of the base medium: Dipotassium phosphate solution ......6.6 m1 Methylene blue solution .............5.6 ml The concentration of dipotassium phosphate in the finished medium was found to be very critical with regard to sheen deve10pment in the colonies. The concentration which will give Optimum sheen production must be determined for each lot of dipotassium phosphate used. 59 The F-5 EMB medium.was poured into standard size petri dishes in approximately 25 ml quantities. The petri dishes were allowed to stand undisturbed at room temperature for at least 50 minutes. A sheet of sterile Whatman No. 1 filter paper was then placed on the surface of the medium in each petri dish. After filtration the membrane filters were laid on the surface of the filter papers. It was necessary to re- move the membranes from the filter papers and re-lay them in order to liberate the air bubbles Which became trapped beneath the membranes. Because the F-5 EMB was a semi-solid medium, the petri dishes were not inverted during incubation. For this reason, the use of clay top petri dishes was preferable. The petri dishes were incubated at 55 C for 22 hours. No at- tempt was made to control the humidity during the incubation period. When F-5 EMB was used in this way, no difficulty was experienced and both dyes appeared to diffuse evenly through the membrane filters. However, when the agar was not included and attempts were made to use the medium as a broth, many difficulties occurred and the dyes failed to diffuse through the filters evenly. The results which are reported.below ‘were obtained with the agar medium. Coliform colonies on IF-5 EMB medium were characterized by the presence of a me- tallic sheen. No preliminary enrichment period was used with this medimm. 4O 1. Recovery of Escherichia coli on F-5 EMB Medium.and Comparison of the Membrane Filter Count with the Standard Plate Count Thirty-four separate suspensions of a pure culture of pg..ggli were prepared using sterile saline solution. The re- covery of E. 321i from these suspensions by the use of the membrane filter technique utilizing F-5 EMB was compared with the recovery which was obtained by means of the standard plate count. The results are summarized in Table IV. Each series represents the average of five replicate membrane filter counts and of five replicate plate counts. The average filter count was found to be 148, while the average plate count was 165. This represents a recovery of 90.7 per cent on the membrane filter as compared with the standard plate count. These re- sults with a pure culture of E. coli indicated a loss in re- covery of approximately 10 per cent with F-5 EMB medium. 2. Recovery of Coliform Bacteria from.Water on F-5 EMB and Comparison of the Membrane Filter Count with the Multiple-tube MPN The Red Cedar River was selected as the source of water which was to be used in this investigation. The stream, which flows through the campus of Michigan State University, is heavily polluted. A series of dilutions was made of this water using 90 ml sterile saline dilution blanks. The MPN's were determined on the basis of the 5-tube, 5-decimal dilution con- firmed test by the procedure as given in Standard Methods. TABLE IV MEMBRANE FILTER RESULTS WITH F-5 EMB COMPARED TO THE STANDARD PLATE COUNT Series Colony Count Per Plate Count MF* Count as Filter Percentage of Plate Count 1 104 165 65.8 2 179 204 87.7 5 146 178 82.0 4 192 198 97.0 5 126 185 68.9 6 208 255 A 88.5 7 154 148 90.5 8 148 145 102.1 9 155 156 99.5 10 155 152 102.5 11 200 219 91.5 12 197 206 95.6 15 121 144 84.0 14 150 162 92.6 15 177 175 102.5 16 108 94 114.9 17 145 165 86.7 18 159 106 112.1 19 111 ' 126 88.0 20 166 197 84.5 21 94 105 91.5 22 202 188 107.4 25 154 152 101.5 24 165 174 95.7 25 158 140 98.6 26 151 146 89.7 27 142 150 94.7 28 126 167 74.6 29 154 174 77.0 50 129 192 67.2 51 159 171 95.0 52 157 157 100.0 55 119 128 95.0 54 175 194 89.2 Total 5020 5552 Average 148 165 90.7 * Membrane Filter 41 42 In an attempt to evaluate the F-5 EMB medium, a comparison was made between the recovery of coliform organisms obtained with this medium and the membrane filter technique and the re- covery obtained by the MPN confirmed test. In this study, 20 membrane filters and 20 MPN's were run and estimations of the number of coliform bacteria present in the water were ob- tained by each procedure. The average MPN per 100 m1 of same ple computed from.the membrane filter counts was 9,580. The average 5-tube MPN was 18,500. Therefore, the MPN calculated from.the Standard Methods' confirmed test was found to be approximately twice that calculated from the membrane filter test utilizing F-5 EMB medium. Random isolations were made from.496 of the colonies, grown on the membrane filters, which exhibited a metallic sheen. Of these, only 85 or 17.1 per cent produced gas in lauryl tryptose broth. The high dye concentrations which were necessary for the production of sheen apparently had an ad- verse effect upon the coliform organisms. For this reason, no later studies were carried out in an attempt to further develop an EMB mediums It was discovered, however, that coliform bacteria could be grown on the membrane filter in the absence of dyes and that a metallic sheen could then be produced.by placing the filters on F—5 EMB medium for four hours at room temperature. The F-4 :medium.was deve10ped for use with the membrane filter in the study of this phenomenon. 45 IV. Formulation Number Four (F-4) F-4 Bacto-Tryptose ......OOOOOOOOOOOOOO 40.0 g Bacto-Lactose ......OOOOOOOOOOOO... 20.0 Dipotassium.Phosphate ............. 2.75 Monopotassium.Phosphate ........... 2.75 Sodium Chloride ................... 5.0 Sodium Lauryl Sulfate ............. 0.15 BECtO-Agar eeeeeeeeeeeeoeeeeeeeeeee 600 Distilledwater 0.000000000000000000000000. 1000 ml The medium was dispensed in 100 m1 quantities and steri- lized in the autoclave at 121 C for 15 minutes. Because F-4 was a semi-solid medium, the method of preparing the petri dishes for incubation with the membrane filters was the same as that which has been described in the preceding section for the F-3 EMB medilm. A series of four suspensions of a pure culture of‘g. coli was prepared. For each series, five replicate membrane filter counts using F-4 medium, without a preliminary enrichment period, were obtained. Five replicate standard plate counts were also simultaneously obtained for each series. The meme brane filters were incubated at 55 C on the Fb4 medium for 18 hours. They were then removed and placed on F—5 EMB at room temperature for four hours. A metallic sheen developed on the colonies. The recovery of E. coli, which was obtained by the two procedures, is given in Table V. An average of 209 colonies 44 was counted on the membrane filters, while the average plate count was 205. This represented a recovery of 101.8 per cent on the membrane filter as compared with the plate count. TABLE V MEMBRANE FILTER RESULTS WITH F-4 MEDIUM COMPARED TO THE STANDARD PLATE COUNT Series Colony Count Per Plate Count MF* Count as Filter Percentage of Plate Count 1 254 220 106.4 2 181 187 96.8 5 158 144 109.7 4 261 268 97.4 Total 854 819 Average 209 205 101.8 * Membrane Filter The results which were obtained with a pure culture of ‘E. coli appeared to be satisfactory. However, when water samples were tested, very excessive overgrowth occurred on the membrane filters with F-4 medium. This overgrowth was, for the most part, caused by non-coliform organisms. Because of this, F-4 medium was not used in further studies. V. Formulation Number Five (F-5) Studies were carried out in search of a selective agent which could be incorporated into a medium designed for the 45 enumeration of coliform bacteria in water. Several different media formulations were investigated in the study of the following selective agents: Dowicide K-76451 Sodium Lauryl Sulfate Oxgall Sodium Azide Brilliant Green Di Sodium Versenate Acid Fuchsin Ethyl Violet Sodium.Desoxycholate Bile Salts None of the substances tested gave results which were completely satisfactory. The most promising results, however, were obtained with the use of Di Sodium Versenate (Versene). The F-5 medium was developed as a result of this investigation. F-5 Base Medium Bacto-Tryptose ................. 40.0 g BaCto-LaCtose eoeeeeeeeeeeeeeeeo 2000 Sodium Chloride ................ 5.0 BaCtO'Agar eeeeeeeeeeoeeeeeeeeee 600 Distilled Water ........................ 950 ml The base medium.was tubed in 28 m1 quantities in 50 m1 test tubes and sterilized in the autoclave at 121 C for 15 minutes. Dipotassium.Phogphate Solution Dipotassium.Phosphate .......... 2.0 g Distilled “'ater OCOOOOOOOOOOOOOOOOOOOOO 133 ml Versene Solution Di Sodium Versenate ............ 5.0 g Distilledwater ......OOOOOOOOOOOOOOOOO 60 m1 l A productof Dow ChemicEI’Company,’Mid1and, MiEhIgan. 46 Both the dipotassium phosphate and the Versene solutions were tubed in convenient quantities and sterilized in the autoclave at 121 C for 15 minutes. Preparation.gf Finished Medium The finished medium was prepared by adding 2.0 ml of dipotassium phosphate solution and 0.4 ml of Versene solution to 28 m1 quantities of the base medium. The F-5 medium, being semi-solid, was prepared for incubation in the same manner as described in section III for the F-5 EMB medium. 1. Recovery of Escherichia coli on F-5 Medium and Comparison of the Membrane Filter Count with the Standard Plate Count A suspension was made of a pure culture of g. 221i. Estimations of the number 0f.§-.221$ present in the suspension were made by the use of the membrane filter technique ems ploying F-5 medium. Simultaneous plate counts were made from the same suspension. A total of 20 counts was made on this suspension with each method. The average membrane filter count was 145. The average plate count was 145. This in- dicated a recovery of 98.6 per cent on the filter as compared with the plate count. Thus, a highly satisfactory recovery was obtained When a pure culture of E. coli was used. 2. Recovery of Coliform Bacteria from.Water on F-5 Medium and Comparison of the Membrane Filter Count with the Multiple-tube MPN When F-5 medium was employed with the membrane filter, coliform colonies growing on the filter did not exhibit any 47 characteristics by which they could be differentiated from other organisms. However, this medium did inhibit the growth of almost all non-coliform bacteria. Since a need existed for a study to determine whether or not it was possible to obtain a comparable recovery between the membrane filter tech- nique and the 5-tube decimal dilution confirmed MPN with a raw water, the following investigation was undertaken. A total of 51 samples was collected from the Red Cedar River. The recovery of coliform.bacteria was determined.by the membrane filter technique and.by the MPN confirmed test. Since coliforms could not be differentiated from.non—coli- forms on the F-5 medium, transfers to lauryl tryptose broth were made from all colonies which developed on the membrane filters. During this study, isolations were made from 2,102 colonies. Of these, 1,724 or 82.0 per cent produced gas with- in 48 hours at 55 C in lauryl tryptose broth. In the case of the membrane filters, the MPN's per 100 m1 of sample were com- puted on the basis of gas production of the isolates. The results are given in Table VI. The average MPN with the meme brane filter technique was 15,600 coliforms per 100 m1 of sample, and an average of 14,200 was obtained by means of the 5-tube, 5-decimal dilution MPN. The results of this investigation showed that it was possible to obtain a satisfactory agreement between these two procedures with the river samples which were examined. The F-5 medium, however, was not practical for the routine TABLE VI MEMBRANE FILTER RESULTS ON WATER SAMPLES WITH F-5 MEDIUM COMPARED TO THE STANDARD METHODS' CONFIRMED MPN Sample Membrane Filter MPN Multiple Tube Confirmed Per 100 ml of Sample MPN per 100 ml of Sample 1 10,000 15,000 2 51,000 49,000 5 28,000 55,000 4 56,000 25,000 5 25,000 44,000 6 51,000 ' 17,000 7 29,000 12,000 8 22,000 51,000 9 26,000 15,000 10 16,000 11,000 11 19,000 17,000 12 14,000 17,000 15 6,000 24,000 14 8,000 15,000 15 6,000 15,000 16 8,750 7,900 17 5,500 5,500 18 7,250 ‘ 4,000 19 4,750 7,000 20 4,500 7,900 21 8,500 6,200 22 5,750 4,600 25 6,500 7,000 24 9,500 7,900 25 8,500 11,000 26 5,750 4,900 27 7,750 8,400 28 10,400 7,900 29 9,000 5,400 50 5,600 7,000 51 8,800 7,900 Total 421,800 440,500 Average 15,600 14,200 48 49 examination of water samples because it was impossible to differentiate between coliform.bacteria and non-coliform bacteria by colony appearance on the membranes. Therefore, the search for a satisfactory medium was continued. VI. Formulation Number Six (F-6) Since the results which were obtained with the F-5 medium showed.a satisfactory recovery of coliform bacteria from water samples by the use of the membrane filter tech- nique, the search for an adequate medium was pursued further. In this investigation, 145 different medium formulations or variations in medium formulations were compounded and tested. The F-6 medium was the result of this investigation. F-6 BaCtO-Tryptose 00......00.000..00.00 1 Sodium Chloride .................... BaCtO‘LaCtose 0....0..000....00.00.. 1 BaCto-Bile salts 000.000.000.0.000.0 0 Broerresol Purple ................. . 5 Methylene Blue (88% dye content) ... 0.01 Ethy1v1016t .....OOOIOOOOOOOOOOOO. 0.0175 Distilled Water ............................ 1000 ml OU‘IOUIO'I HOOOO The medium.was tubed in 25 m1 quantities and sterilized in the autoclave at 121 C for 15 minutes. The pH was 1 The amount indicated is on the basis of 100 per cent dye content. 50 adjusted with KOH so that a final pH of 6.8 resulted after autoclaving. Before use, 0.25 ml of a sterile one per cent solution of sodium lauryl sulfate was added to each 25 m1 quantity of medium. The medium was added in 2.2 m1 amounts to sterile ab- sorbent pads contained in 60 x 15 mm glass petri dishes. Plastic refrigerator trays, lined with moist cheese cloth, were used to maintain an atmosphere of saturated.humidity during incubation. An incubation period of 14 hours at 55 C was found to yield the best results. If incubation is al- lowed to continue for more than 15 hours, the results obtained cannot be considered reliable. No preliminary enrichment period was necessary with the F—6 medium. After an incubation period of 14 hours, coliform bacteria produced yellow or pinkish-yellow colonies while colonies of non-coliform bacteria were blue or colorless. The F-6 medium, like the MacConkey broth developed by Taylor, Burman, and Oliver (1955 and 1955), incorporated bile salts as a selective agent and.high concentrations of brom.cresol purple as an indicator to detect lactose fermentation. 1. Results Obtained with F-6 Medium in the Determination of the Number of Coliform Bacteria in Water Samples Three types of water samples were used in testing the F-6 medium. The first group of samples was taken from the Red 51 Cedar River which represents a very heavily polluted stream. The second group was collected from Lake Lansing. This lake was selected because it contained a very high non-coliform population in relation to the coliform pepulation. The third group of samples was taken from various contaminated wells. Coliform determinations were made on each Sample by means of the membrane filter technique. Both F-6 and EHC modified Endo medium were used in.making these determinations. The results were compared with coliform determinations which were simultaneously obtained by means of the Standard Methods' 5-tube, 5-decima1 dilution confirmed test. The results, which were obtained with the Red Cedar River and Lake Lansing, are given in Tables VII and VIII. Each membrane filter count is based on an average of two filters. The average of the membrane filter counts for a total of 26 Red Cedar River samples was 12,225 with the F-6 medium and 7,845 with EHC modified Endo medium. The average of the 5-tube confirmed MPN's was 15,754 for these same samples. Thus, F-6 medium.gave a better agreement with the MPN determination than that obtained with the Endo medium. In the case of the Lake Lansing samples, the average membrane filter count for 20 samples was 21 with F—6 and 17 with Endo medium. The average for the tube MPN's was 29. Because, all three coliform determinations were fairly close, no conclusion could be drawn from.these results. However, TABLE VII NUMBERS OF COLIFORMS ESTIMATED PER 100 ML OF SAMPLE RED CEDAR RIVER Sample F-6 EHO Modified Multiple Number Broth Endo Broth Tube MPN 1 10,600 8,800 25,000 2 22,400 16,900 40,000 5 18,400 14,800 21,000 4 20,400 10,200 17,000 5 11,700 7,700 15,000 6 14,200 5,200 2,100 7 4,700 5,400 1,700 8 12,200 6,100 11,000 9 9,100 7,200 7,900 10 4,500 5,500 5,500 11 8,500 6,900 7,900 12 7,900 4,500 2,500 15 4,500 5,200 5,200 14 14,400 14,400 15,000 15 28,500 19,500 51,000 16 17,500 14,500 55,000 17 18,400 15,100 19,000 18 12,600 5,700 17,000 19 17,500 5,160 17,000 20 5,700 1,700 2,500 21 15,500 6,100 17,000 22 11,500 9,200 25,000 25 6,900 5,800 4,900 24 11,400 6,400 11,000 25 8,500 6,100 7,900 26 5,400 4,700 4,100 Total 517,800 205,960 557,600 Average 12,225 7,845 15,754 52 TABLE VIII NUMBERS OF COLIFORMS ESTIMATED PER 100 ML OF SAMPLE LAKE LANSING Sample F-6 EHC Modified Multiple Number Broth Endo Broth Tube MPN 1 20 14 22 2 12 18 55 5 10 4 25 4 4 4 l5 5 8 0 7.8 6 0 8 25 7 4 2 17 8 61 20 49 9 55 45 45 10 57 41 64 11 16 19 25 12 18 12 l7 15 15 8 51 14 0 4 17 15 4 12 14 16 8 8 55 17 8 8 15 18 55 15 55 19 45 52 46 20 51 58 54 Total 429 552 576 Average 21 17 29 55 54 excessive overgrowth of non-coliform bacteria was observed with both F-6 and EHC modified Endo medium. Because of this, neither medium was completely satisfactory with the Lake Lansing samples. The well samples were collected on the day prior to the test date. They were then stored in the refrigerator at 8 C for 24 hours. The samples were clear at the time of collection, but, on standing, iron precipitated from many of them and during filtration this iron was deposited on the surface of the membrane filters. These iron deposits caused extreme.spreading among the colonies on the filters which often resulted in confluent colonies. This was especially true when EHO modified Endo medium was used. It also re- sulted in the failure of colonies to produce a metallic sheen on this medium. With F-6 medium, however, discrete, isolated colonies were obtained regardless of whether or not iron deposits were present on the filters. Because of the difficulty with iron precipitates, all well samples were collected in duplicate 16-oz sampling bottles. One sampling bottle contained approximately 0.04 g of sodium.thiosulfate while the second bottle contained 0.05 g of di sodium Versenate (Versene). Otherwise the procedure was the same as was used with the two water sources previously described.. 55 The results which were obtained with the well samples are summarized in Table IX. When Versene was present in the sampling bottles, a five-fold increase was observed in the coliform counts on the membrane filters and a three to four- fold increase was observed in the tube MPN determinations. 2. Gas Production in Lauryl Tryptose Broth of Isolates from.F-6 Medium.and from.EHC Modified Endo Medium During the previous study on water samples from.the Red Cedar River, Lake Lansing, and various contaminated wells, isolations were made from many colonies growing on the meme brane filters. These isolations were made from both F-6 and from.EHC modified Endo medium. The isolates were tested for I their ability to produce gas in lauryl tryptose broth within 48 hours at 55 C. In the study of the F-6 medium, isolations were made from 1,028 yellow colonies. Of these, 941 or 91.5 per cent pro- duced gas from.lactose. Of 155 isolations which.were made from.pinkisheyellow colonies, 114 or 75.5 per cent produced gas. There were also 255 isolations made from.other colony types and of these only 5 or 0.01 per cent produced gas. From 515 isolations which were made from.sheen colonies on EHC modified Endo medium, 298 or 95.2 per cent produced gas. Of 500 isolations which were made of non-sheen colonies, 42 or 14 per cent produced gas. 56 «we 83 «on new «mm 8N owes»... omRH . ~83 83H 83 83 8mm its 03 on ow an «a «A 3 om.» owe cue owe 03 R 3 owe om... one 03 R... om 3” 8mm omen 83 3. Sn owe ma mm mm mm mm 2 3 S m... 4 S w... e o 2 8mm 83 8% 03 a 03 S cm. 3 mm mm o e o 0mm 3. in one 5.3 3 m comm 00mm come com com 02. e 08. .5 oo 02 Ma mm o o: 3 NS omm 8 «3 m 8m 3m 03 0mm we ca 4 8mm ooam 8mm 0mm ova com m ... N 3 mg. 0 m m R o «H S m «H H E: 33 £95 35 59a 2% BS flown 03m 52m won—Ba .3322 goddess 8m one #3sz Eu cum sesame £30m 335m 3 388.3» 888 S 0.35m wafiflnsdm a.“ enema—55.39 gadom .Raem no as 03 use essence no use; a Eggazoo zo EOHEEHBQM 2.5.3900 NH 593. 57 DISCUSSION The membrane filter, a relatively new tool in the field of bacteriology, has found its widest application in the bac— teriological analysis of water. Satisfactory results in de- tecting coliform organisms in water by the use of the mem- brane filter technique have been reported by many authors. The technique has received rather wide acceptance among technicians who are engaged in water analysis. Kenyon (1955) intimates that the membrane filter has "displaced methods for the examination of contaminants in potable waters which have remained standard for nearly fifty years." He also refers to the Standard Methods' dilution tube technique as “the old M.P.N. procedures". Such assumptions are premature, of course, for Standard Methods states, "It must be understood that this is in no way a standard technique, and it cannot be considered an acceptable substitute for the dilution tube method." In the experience of the author with the membrane filter technique, several difficulties were encountered which cannot be ignored. When membrane filters were incubated on absorbent pads which contained nutrient media, problems arose concerning the regulation of humidity during incubation. Taylor, Burman, and Oliver (1953 and 1955) have experienced difficulties simi- lar to those of the author with regard to moisture control. J, \J 58 0n.the other hand, Slanetz and Bartley (1955) have indicated that incubation in an atmosphere of saturated humidity did not seem to be necessary and they reported that comparable counts were obtained when incubation was carried out under normal incubator conditions. However, it was the experience of the writer that humidity control could not be disregarded in the application of the membrane filter technique if valid results were to be obtained. Certain difficulties were also encountered in the prep- aration of EHC modified Endo medium. As a result, the writer considered it highly doubtful that all batches of complete medium, prepared from either liquid or powdered indicators, contained uniform amounts of dye in solution. With a pure culture Of.§;.£2££: a recovery of only 76.0 per cent was obtained when EHC modified Endo medium was em- ployed with the membrane filter technique as compared with the recovery obtained by the standard plate count. Thus, if it is impossible to obtain complete recovery on EHC modified Endo medium with a strong laboratory culture 0f.§;.§9li’ then it is certainly unwarranted to expect a complete recovery when dealing with attenuated coliforms which are commonly encountered in the examination of a water supply. It was later shown that a full recovery could be obtained on the membrane filter with a pure culture of §fi_ggli_when a non-selective medium was employed. This observation, together with the finding that all attempts to recover §;_coli from the I . I ‘4 ~ VJ ‘l/ 1 , 1' . C “l . \l . - U . | ' ' v 1‘ . V . ' fl ‘ ~ \ I . x . , F ’44 x . . I _ _ ,V, l« , ‘ \. h l A ‘. v‘ . " _.,‘ g ‘_, x ‘ \w w - _, A . V _ I , . I t ‘ : 59 filtrates were unsuccessful, indicated that the loss in re- covery was not due to the fact that §;.2911.Was passing through the filters. It also showed that the loss in re— covery was not due either to a toxicity of the filters for the organisms or to any effect resulting from the mechanical process of filtration. The only other factor, then, which could account for the loss in recovery was that a toxicity for §;_ggli_was exhibited by the EHC modified Endo medium. Thus, it was concluded that this medium was toxic to E; coli and that the main difficulty in the application of the mem- brane filter technique to the bacteriological analysis of water lay in the lack of a satisfactory selective medium. Experimental studies were carried out in an attempt to develop a more satisfactory selective medium for the enumer- ation of coliform bacteria from water by the use of the mem— brane filter technique. During these studies six main media formulations were developed. Five of these formulations were eventually discarded and eliminated from further studies. This was due to either a lack in development of differenti- ating characteristics of the coliform bacteria growing on the media or as in the case of the F-3 EMB, to an adverse effect which the medium had upon these bacteria. The most satisfactory results in this investigation were obtained with the F-6 medium. Although small amounts of methylene blue and ethyl violet were present, this medium de- rived its selective action primarily from bile salts and it .llf]..lxlu\ OI’DIIJI 6O incorporated brom cresol purple which acted as an indicator to detect lactose fermentation. The MacConkey broth of Taylor, Burman, and Oliver (1953 and 1955) and the F-6 medium are the only media which have been developed thus far for the de- tection of coliform bacteria from water which do not employ high concentrations of a toxic dye as a selective agent and which do not rely upon the production of a metallic sheen as the criterion for the indication of coliform colonies. To date, all media which have received wide acceptance for use with the membrane filter in the detection and enumer- ation of coliform bacteria have been modifications of an Endo— type broth and coliform production on these media is deter- mined by the presence or absence of a metallic sheen. The amount of basic fuchsin which is required for the production of this metallic sheen is large and such a quantity apparently exhibits a toxic effect upon some of the coliform bacteria as well as upon non-coliform bacteria. Three types of water samples were selected for the testing of the F-6 medium. The first group of samples were taken from a heavily polluted river, the second group from a lake which contained a very high non-coliform population in relation to the coliform population and the third group from various contaminated wells. Coliform determinations were made on each sample by means of the membrane filter technique. Both F—6 and EHC modified Endo medium were em- ployed in the use of this technique. Simultaneous coliform 61 determinations were also made by means of the Standard Methods' 5-tube, 3-decimal dilution confirmed test. F-6 medium was found to be superior to EHC modified Endo medium in the coliform determinations which were made on the river samples. In the testing of 26 samples, an average coli- form estimation of 12,223 per 100 ml of sample was obtained with F-6 medium. An average estimation of 7,845 was obtained by the use of EEO modified Endo medium, whereas the 5-tube confirmed MPN gave an average of 13,754 for the same 26 sam- ples. In the case of the lake samples, neither F—6 medium or EHC modified Endo medium was completely satisfactory. Diffi- culties were experienced with both media due to excessive overgrowth of the surface of the membrane filters by non— coliform bacteria. With the well samples, difficulties were experienced due to the precipitation of iron. Filtration resulted in the COD? centration of iron deposits on the surface of the membrane filters. This condition caused both extreme spreading which resulted in confluent colonies and failure to produce a me- tallic sheen when EHC modified Endo medium was employed. When F—6 medium was used, however, discrete, isolated colo- nies were obtained regardless of whether or not iron deposits were present. Since the results obtained with F-6 medium are not dependent upon the presence or absence of a metallic sheen, v 62 any difficulties which were encountered in the use of this medium were greatly minimized. It was learned that the precipitation of iron in the well samples could be prevented by the addition of di sodium Versenate (Versene) to the sampling bottles. However, the presence of Versene in the sampling bottles resulted in a five-fold increase in the average coliform counts obtained with the membrane filters and a three to four-fold increase in the 5-tube MPN determinations. Therefore, if Versene is used to prevent the precipitation of iron, this would neces- sitate a revision in the standards to be met for such waters. This would be a definite disadvantage of the membrane filter technique since it would be far from desirable to have two sets of standards, one for well waters or those which con— tained iron and one for waters of other types. As far as the F-6 and the EHC modified Endo media were concerned, comparable results were obtained with both media on the well samples and the results were in agreement with those obtained by means of the Standard Methods' multiple- tube MPN confirmed test. During this study, numerous isolates from both F~6 medium and EHC modified Endo medium were tested for their ability to ferment lactose with gas production. It was found that it was impossible by means of colony appearance on the surface of the membrane filter to differentiate ac- curately between an organism which was capable of fermenting 0 J ‘ H x,- .7) s L_‘ ) l ’ x) 1‘ ‘1 _ lactose with the production of both acid and gas and one which was capable of fermenting lactose with the production of acid only. This finding was in agreement with observa- tions which have also been made by Taylor, Burman, and Oliver (1953 and 1955). The author is of the opinion that many of the advan- tages which are claimed to make the membrane filter tech- nique superior to standard procedures are more apparent than real. It is true that the short incubation period used with the resulting saving of time and the fact that two or more different media can be applied in succession are factors which favor the adoption of this technique. The increased size of the sample which may be examined has often been stated as an advantage of this technique since this made 'possible the examination of waters which have a low bacterial density. However, it is unnecessary to examine larger volumes of water than is permitted by the use of present standard techniques since these techniques have proven to be more than adequate for the detection of a safe water supply. In fact, these techniques might be criticized on the basis that their use may condemn a safe water supply, but certainly not for their inability to detect a supply which is unsafe. Claims have been made that the membrane filter technique is simpler in operation than present standard techniques. (Goetz, 1947) (Goetz and Tsuneishi, 1951) The opinion of the writer is in disagreement with this viewpoint. ‘J g Q \J KI 64 It is also to be questioned as to whether the membrane filter technique is more economical than standard procedures especially when the high cost of the membranes is considered. And although small volumes of media are employed in this tech- nique, the fact that the media has to be used in high concen- trations must be taken into account. Field applications of the membrane filter technique have been suggested by Geldreich et_al;_(l955), Levin and Laubausch (1954), and others. However, in the bacteriological testing of water it is necessary that the quantity of water which is passed through the membrane filter be adjusted according to its coliform density. Small quantities of heavily polluted waters must be diluted with sterile water so that errors will not occur due to the use of too small a sample. For this rea: son, all proposed field tests can be used only with low dens- ity waters since it would be unfeasible to carry sterile di- lution bottles out into the field to prepare dilutions. In addition, the membrane filter technique cannot be used satisfactorily in the examination of waters which contain sus— pended solids as this would result in the clogging of the fil- ters. In order to examine such waters by means of the mem- brane filter technique, it would be necessary to utilize the "Concentrometer" as devised by Goetz (1953a) which would fur- ther add to the cost of the examination. g1 65 SUMMARY AND CONCLUSIONS 1. A detailed investigation of the membrane filter technique was made in order to evaluate its usefulness for determining the sanitary quality of water supplies. Studies were conducted on various aspects of this technique and its application to the bacteriological analysis of water. During early studies that were carried out, certain technical prob— lems arose. The solution of these problems was necessary before subsequent studies could be undertaken. The nature of each of these problems has been described. 2. Studies, using a pure culture of §;_cgli, showed that a recovery of only 76.0 per cent of the organisms was obtained by the use of the membrane filter technique when EHC modified Endo medium was employed. However, when a non-selective medium was used, a recovery of 97.4 per cent was obtained with this technique. These results showed that EHC modified Endo medium had an inhibitory effect upon E; 9911, Since there was a loss of 24 per cent in the recovery of 24.22222 it was realized that this Endo medium would not permit the complete recovery of all coliform bacteria from water samples. Therefore, the need for a more satisfactory selective medium was recognized. 3. A search was undertaken in an attempt to formulate a culture medium for use with the membrane filter which would permit the complete recovery of all coliform organisms from water samples. During this study six main media formulations were developed and tested. Five of these formulations were eventually discarded. However, it was shown that a satis- factory agreement in results could be obtained between the membrane filter technique and Standard Methods' 5—tube, 3- decimal dilution confirmed test. 4. The F—6 medium resulted from the compounding and testing of 145 different formulations or variations in form- ulations. This medium was the most satisfactory of those developed. In the examination of the river samples, it was found superior to EHC modified Endo medium in that it gave a better agreement with the Standard Methods' MPN confirmed test. With lake samples, which contained a very high non- coliform population in relation to the coliform population, neither F-6 or the Endo medium was completely satisfactory. In the case of well samples, comparable results were obtained with both F-6 and the Endo medium as far as coliform deter— minations were concerned. However, when iron deposits were concentrated on the filters extreme spreading took place which resulted in confluent colonies. In addition these colonies failed to produce a metallic sheen on the Endo medium. With the F-6 medium spreading among the colonies was greatly reduced and, since the criterion for coliform detection did not depend upon the production of a metallic sheen, less difficulty was experienced in the use of this medium. 67 5. It was discovered that the precipitation of iron from the well samples could be prevented by the addition of Versene to the sampling bottles. This, however, resulted in a substantial increase in the coliform estimations which were obtained with both the membrane filter technique and the Standard Methods' MPN confirmed test. The use of Versene in the sampling bottles would therefore necessitate a re- vision in the standards which are to be met. 6. Many isolates from both F—6 medium and EHC modified Endo medium were tested for their ability to ferment lactose with gas production. It was found that gas production could not be accurately predicted from the colonial appearance of organisms growing on the surface of the membrane filter. 7. Several advantages have been reported in favor of the membrane filter technique which seemingly make this technique superior to the Standard Methods' MPN procedure. The writer is in disagreement with many of these so—called advantages. 8. The results of this study indicate the inadequacy of the membrane filter technique for evaluating the sanitary bacteriological quality of waters. The main difficulty en- countered in the use of the membrane filter technique results from the lack of a satisfactory selective medium. Although the F-6 medium was found to be more satisfactory than EHC modified Endo medium in the examination of certain waters, it was not 68 found to be completely satisfactory with all types of waters. Before the membrane filter technique can be accepted as a standard procedure for the examination of waters, a selec- tive medium must be deve10ped which will permit the complete recovery of all coliform bacteria from all types of waters. Otherwise, the use of this technique can produce a distorted view of the condition of the water by giving misleading re- sults. 2. 5. 4. 8. 9. 10. 69 BIBLIOGRAPHY Anon. 1951 Characteristics and uses of the HF Millipore Filters. Lovell Chemical 00., Watertown 72, Massachusetts. Anon. 1955 The role of the molecular filter membrane in the field of public health. (Editorial) Am. J. Public Health, Vol. 45, No. 6: 766-768. Anon. 1955 MF Millipore Filters Supplemental techni- cal information regarding bacteriological and other analytical procedures. Millipore Filter Corp., Watertown 72, Massachusetts. Bigelow, S. L., and Gemberling, A. 1907 Collodion membranes. J. Am. Chem. Soc., 29: 1576-1589. Bush, J. H. 1955 Status of the Membrane Filter in the bacteriological examination of water. J. New England Water Works Assoc., vol. 69, no. 1: 1-8. Chambers, 0. W. 1950 Relationship of coliform bacteria to gas production in media containing lactose. Pub. Health Rep. 65, no. 19: 619-627. Clark, Harold F., Geldreich, E. E., Jeter, H. L. and Kabler, Paul L. 1951 The membrane filter in sanitary bacteriology. Pub. Health Rep. vol. 66, no. 50, reprint no. 5101: 951-977. Clark, H. F., Jeter, H. L., Geldreich, E. F., and Kabler, P. W., 1952 Domestic and European molecular filter membranes. J. A. W. W. A. 44, no. 11: 1052-1056 Clark, H. F., and Kabler, P. W. 1952 The membrane filter in water quality tests. A. J. P. H. 42, no. 4: 385- 588 0 Elford, W. J., 1951 A new series of graded collodion membranes suitable for general bacteriological use, especially in filterable virus studies. J. Path. Bact. 54: 505-521. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 70 Ferry, J. D. 1956 Ultrafilter membranes and ultra- filtration. Chem. Rev. 18: 575-455. Fick 1885 Ueber Diffusion. Pogg. Annal. 94:) 59-86. Cited from Ferry, J. D. (11) Geldreich, E. E., Kabler, P. W., Jeter, H. L., and H. F. Clark 1955 A delayed incubation membrane filter test for coliform bacteria in water. Am. J. Pub. Health, vol. 45, no. 11: 1462-1474. Goetz, A. 1947 Materials, techniques and testing methods for the Sanitation (Bacterial Decontamination) of small-scale water supplies in the field used in Germany during and after the war. (Abstract) U. S. Lept. of Commerce, Joint Intelligence Objective Agency of the Armed Services, Fiat Final Rep. 1512: 8. Washington, D. C. Goetz, A. 1955 The concentrometer method of apply- ing molecular filter membranes. J. Am. Water Works Assoc., vol. 45, no. 9, Sept. 1955: 955-944. Goetz, A. 1955 Application of molecular filter tech- niques to the bacterial assay of sewage. Sewage and Industrial Wastes, vol. 25, no. 10, Oct. 1955: 1156- 1144. Goetz, A., Gilman, R. E., and Rawn, A. M. 1952 Application of molecular filter membranes to specific problems in water analysis. J. Am. Water Works Assoc., 44, no. 6: 471-485. Goetz, A. and Tsuneishi, N. 1951 Application of molecular filter membranes to the bacteriological analysis of water. J. Am. Water Works Assoc., 45: 94 3" 984 0 Hajna, A. A. and Damon, S. R. 1954 Coliform detection in water by a single-step technique using the membrane filter. Pub. Health Rep., vol. 69, no. 1: 58-60. Jeter, H. L., Geldreich, E. E., and Clark, H. F. 1954 Type differentiation of coliform organisms with mem- brane filter technique. Jour. Am. Water Works Assoc., vol. 46, no. 4: 586-594. Kabler, Paul 1954 Water examinations by membrane fil- ter and most probable number procedures. Am. J. Pub. Health, vol. 44, no. 5: 579-586. 22. 23. 24. 25. 26. 27. 28. 29. 50. 51. 71 Kabler, P. W. and Clark, H. F. 1952 The use of differential media with the membrane filter. Am. J. Pub. Health, vol. 42, no. 4: 590-592. Kenyon, Walter W. 1955 A remarkable development in the control of water quality. Pub. Works Mag., vol. 86: 129-130 0 Levin, G. V. and Laubausch, E. L. 1954 Membrane filter and emergency water supply control. Am. J. Pub. Health, vol. 44, no. 1: 55-65. McCarthy, J. A. 1955 Comparative coliform densities in water by membrane filter test and by multiple tube technic. Am. J. of Pub. Health, vol. 45, no. 12: 1569- 1577. McKee, J. E., Derby, R. L., Goetz, A., Noble,-R. E., and Streicher, L. 1955 Technique of bacterial examination of water with molecular filter membranes. J. Am. Water Works Assoc., vol. 45, no. 11: 1196- 1210. Presnell, M. W., Arcisz, W., and Kelly, C. B. 1954 Comparison of the MF and MPN techniques in examining sea water. Pub. Health Rep., vol. 69, no. 5: 500- 504. ‘ Sanarelli 1891 Abt. I, 9 Cent. Bakt. Parasitenk: 467-472. Cited from.Ferry, J. D. (11) and Goetz, A. (18). Schutz, F. and Kruse, H. 1947 Die Anwendung der Membranfilter in der bakteriologischen Wasserunter- suchung, insonderheit zur Bestimmung des Cali-Titers (The use of membrane filters in bacteriological exam- ination of water, more especially for the quanitative examination of coliform organisms. Zent. f. Bakt., 1947, I. Abt. Orig., 152: 155-147. Summary in English Bull. Hyg., Lond., 1950, 25: 289. Slanetz, L. W. and Bartley,.C. H. 1955 Evaluation of membrane filters for the determination of coliform bacteria in water. Applied Microbiology, vol. 5, no. 1: 46-51. Standard Methods for the Examination of Water, Sewage and Industrial Wastes, 10th ed., 1955. Am. Pub. Health Assoc., New York. ' 52.. 35. 55. 56. 57. 58. 72 Taylor, E. W., Burman, N. P., Oliver, 0. W. 1955 Use of the Membrane Filter in the bacteriological examination of water. J. Applied Chem- 5: 255-240. Taylor, E. W., Burman, N. P., and Oliver, 0. W. 1955 Membrane filtration technique applied to the routine bacteriological examination of water. J. of the Inst. of Water Engineers, vol. 9: 248-265. Thomas, H. A., Jr. 1955 Statistical analysis of coliform data. Sew. Indust. Wastes, 27: 212. Thomas, H. A., Jr. and Woodward, R. L. 1955 Water potability control with membrane filters. Presented at the 44th Meeting of the Committee on Sanitary Engineering and Environment, National Research Council, Division of Medical Sciences, Washington, D. C., Oct. 7, 1955. Thomas, H. A., Jr., Woodward, R. L., and Kabler, Paul. 1955 Estimation of coliform density by the membrane filter and the fermentation tube methods. Am. J. Pub. Health, vol. 45: 11. 'Yee, George 8., Krabek, W. B., and Schaufus, C. P. 1955 New medium for bacteriological analysis with molecular filter membranes. J. Am. Water Works Assoc., vol. 45, no. 9: 945-951. Zsigmondy, R., and Backman, W. 1918 Uber neue Filter. Ztschr. f. anorg. u. allgem. Chem. 105:. 119-128. 0' A! H I IIHIHIHIWW RH mlllll Blllll H! Y” ”I" “WWW E" Vll N” U" 3 03145 3222 12 air 3