144 EXCREMENTAL ORGANISMS 



In 72 samples the two methods agreed, that is the plate 

 count was in approximate agreement with the reciprocal of the 

 smallest quantity of the sample showing gas formation. In 25 

 samples the results differed by one dilution (the dilutions being 

 decimal fractions of a cubic centimetre), in two samples by two 

 dilutions, and in one sample by three dilutions. The agree- 

 ment in the averages is very reasonable when the chance errors 

 of distribution inherent to the tube method are considered, and 

 the differences between individual samples can be shown to be 

 well within the limits calculated by the theory of probabilities. 



The errors connected with rebipelagar are caused (1) by the 

 destruction of the characteristic colour of the B. coli colonies by 

 the diffusion of amines or other alkaline substances through the 

 medium and (2) by the development of red colonies by organ- 

 isms not of the B. coli group. When a dilution of the sample 

 is employed that prevents overcrowding of the colonies, the 

 first error is usually avoided unless there is a large excess of 

 alkali forming organisms present; this condition can be easily 

 recognised because either a yellow area is produced concen- 

 trically from a colony, or, as is usually the case, the whole- of 

 the medium is yellow. The error due to organisms other than 

 coliform bacteria is small and can be largely eliminated by 

 experience. The characteristic forms produced by coliform 

 organisms on the surface of 'the plate may either be a colony 

 of deep red colour producing a haze in the surrounding medium, 

 or one with a red centre surrounded by a yellowish or pinkish 

 aureole of slimy consistency. The subsurface colonies are of 

 the former variety but may not invariably produce the haze 

 which is due to the diffusion of acid into the surrounding 

 medium. The author, during the examination of several 

 hundreds of coliform colonies from milk plated on rebipelagar, 

 has only met with two organisms, one a coccus and the other a 

 bacillus, that produced colonies resembling those typical of 

 B. coli, but many organisms that ferment lactose with the pro- 

 duction of acid may, especially after prolonged incubation, 

 produce colonies that bear a superficial resemblance to those 



