1. A bacteriological examination was made of 1108 samples of water submitted to the laboratory for routine analysis. Of these, 550 contained coliform bacilli and were examined by two different methods. The first method, referred to as method I and officially recommended by the Ministry of Health, involves enrichment in MacConkey broth at 37°C., followed by plating and subsequent identification of selected colonies. The second method, described by G. S. Wilson and his colleagues in their report on milk and referred to as method IV, consists in subculturing every positive 37°C. MacConkey tube into MacConkey broth at 44°C. for Bact. coli I and into Koser's citrate at 37°C. for organisms of the iritermediate-aerogrenes-cloacae group (I.A.C.). In both methods the absolute numbers of coliform bacilli and of the different types are calculated from probability tables.

2. From every fermented tube of MacConkey broth at 37°C. that was plated coliform organisms were isolated. In not a single instance was a false positive, due to non-coliform bacilli, encountered. This merely confirms the general experience in this country of the high degree of specificity of MacConkey broth for members of the conform group. Altogether 2840 strains were isolated by method I and classified according to the scheme given in Table I (p. 310).

3. In the 550 samples Bact. coli I was demonstrated 417 times by method I and 446 times by method IV. The corresponding numbers for I.A.C. were 213 and 449. Method IV therefore proved rather more delicate for the detection of Bact. coli I, and very much more delicate for the detection of I.A.C.

4. Evidence is brought to show that the frequent failure of method I to detect I.A.C. is due partly to a dilution effect and partly to the overgrowth of I.A.C. by Bact. coli I. The chances of demonstrating I.A.C. by method I are remarkably small unless these organisms are present alone, or in numbers greater than or approximately equal to those of coli I.

5. The specificity of the 44°C. MacConkey test for Bact. coli I, and for the infrequent Irregular II which closely resembles it, is very high. For instance, among 2840 strains tested, only fifteen produced gas in MacConkey at 44°C. which could not be classified as Bact. coli I or Irregular II. Only six out of a total of 1086 I.A.C. strains reacted positively to this test. Finally not a single strain, other than Bact. coli I, isolated from 100 samples of human faeces was found capable of producing gas in MacConkey broth at 44°C.

6. The specificity of the citrate test for I.A.C. is not so high as that of the 44° C. MacConkey test for Bact. coli I. In eight out of eighty-one samples of water a positive citrate test was given by organisms other than I.A.C. Examination showed that these consisted of Bact. coli I, non-lactose-fermenters, or of irregular types. Since neither Bact. coli I nor the non-lactose-fermenters grew in citrate in pure culture, and since they were found in citrate tubes that had been subcultured from fermented MacConkey broths inoculated with relatively large quantities of polluted water, it seems possible that their growth was due to the carrying over of small amounts of organic matter into the citrate medium.

7. Observations on the association of Bact. coli II, Irregular I and Irregular II with other coliform types render it probable that Bact. coli II is not mainly of faecal origin, that Irregular I is almost certainly of faecal origin, and that Irregular II may be of faecal origin, but that further evidence will be required to establish this definitively.

8. Owing to the fact that Irregular I does not produce gas in MacConkey broth at 44°C., the faecal coli count by method IV is underestimated, and owing to the fact that Irregular II does produce gas at 44°C., the faecal coli count is overestimated. In this series of examinations the faecal coli count was falsely estimated owing to the presence of Irregular I in 0·4% and owing to the presence of Irregular II in 0·5 % of samples. The combined error of 0·9 % is insignificant in relation to the experimental error of the technique as a whole.

9. By the use of method IV it is possible to demonstrate that (a) I.A.C. is nearly as frequently present in polluted water as Bact. coli I, and when the pollution is only slight may exceed it; (b) I.A.C. tends to be the dominant organism in polluted water that has been subjected to chlorination; and (c) I.A.C. is often present in human faeces (61% of 100 samples), and may sometimes constitute the dominant coliform type. In view of these facts it would seem unwise to neglect organisms of the I.A.C. type in the interpretation of water analyses in this country. Their presence is likely to be of most significance in water of a fairly high degree of purity in which, owing to chlorination or other causes, Bact. coli I can no longer be detected.

10. It is therefore concluded that it would be advisable in future to adopt a method of analysis, such as method IV, which affords a more delicate index of the presence of Bact. coli I, and particularly of I.A.C., in water than the method officially recommended at the present time. The fact that method IV is both simpler, quicker, and cheaper than Method I provides an additional recommendation for this change.