144 Elements of Water Bacteriology. 



On the other hand, it is equally true that in a large pro- 

 portion of cases colon bacilli are isolated from positive 

 dextrose-broth tubes. Longley and Baton (1907) in the 

 examination of 3553 samples of Potomac water obtained 

 positive tests 794 times, while B. coli was actually present 

 529 times; 67 per cent of the "presumptive tests were 

 therefore correct. Gage (1902), in the Massachusetts 

 work cited above, found that 70 per cent of his fermented 

 dextrose tubes contained B. coli. 



The work of recent years has made it clear, however, 

 that both the coincidence of negative presumptive tests 

 with the; absence of B. coli and the general coinci- 

 dence of positive presumptive tests with the presence of 

 B. coli, are open to disastrous exceptions. In the study 

 of the samples, tabulated on page 142, the presumptive 

 test closely coincided with sanitary conditions. In the 

 Kennebec River, too, Whipple found a close corre- 

 spondence between the results of the presumptive 

 test and the complete isolation of B. coli (Whipple, 

 1907). With other waters, however, discordant results 

 have been reported. Stoughton (1905), in a study of the 

 New York Supply, showed that B. coli could frequently 

 be isolated when the presumptive test was negative. 

 Fuller and Ferguson's (1905) results at Indianapolis 

 and those of many other observers have led to the same 

 conclusion. With heavily polluted waters the presump- ' 

 tive test breaks down entirely. Gas production may be 

 absent or atypical in a large proportion of tubes inocu- 



