It has been shown (Barkworth & Irwin, 1938) that the distribution of coliform organisms in milk follows a Poisson series, and on this basis it should be possible to estimate the population from the results of replicate tests at several dilution levels by reference to tables. In later experiments with water (Barkworth & Irwin, 1941), it was found that the number of positive tubes did not always reach expectation, having regard to the known numbers of organisms inoculated. The peculiar physico-chemical constitution of milk naturally raises the question whether coliform organisms, though present, might fail to give a positive reaction in the presumptive test. It is well known that in favourable circumstances lactic acid bacteria will overgrow all other organisms in milk. In the presumptive test the greatest concentration of milk in routine technique is 1 ml. of milk in 5 ml. of medium. This dilution would not affect the ratio of coliform organisms to lactic acid bacteria. Even when the coliform contamination at the time of testing is ‘present in 0·001 ml.’ the agreement with the fermentation test is only about 50% (Barkworth, 1929). It will be argued that the gas tube would disclose the presence of gas in more cases than the fermentation test, nevertheless it is not perfect, for Chalmers (1928) recovered Bact. coli from plates of 1: 10 dilution of samples of milk which had given a negative presumptive test. There is therefore some reason to expect a small proportion of false negative coliform results to occur with milk samples, and the object of the present experiment was to see if inoculation with a known number of organisms would produce similar numbers of positive tubes in milk and in water.