124 THE ENUMERATION OF BACTERIA IN MILK 



method of enumeration was also compared with the direct 

 microscopical method of Breed but this will be dealt with later. 



From a consideration of this work Conn pointed out that 

 variations in technique are much more important than the com- 

 position of the medium, and that variations in results may 

 reasonably be expected, even under the best conditions due 

 (1) to clumping of the bacteria, and (2) to the bacteria being 

 in non-uniform suspension and not in solution. These two 

 factors render it improbable that two small samples will contain 

 equal numbers of organisms, and the lower the total number of 

 bacteria the greater will this divergence become. Conn ex- 

 pressed the opinion that " individual counts under the best 

 conditions are subject to considerable variation and that no 

 single individual count can be relied upon." ..." It is not 

 possible to rely upon a greater accuracy than 100 per cent even 

 when the average of more than one sample is obtained, although 

 most of the results fall considerably below this limit." 



During 1915 the author made a series of duplicate examina- 

 tions of milk by plating one of the routine samples in duplicate 

 daily; in this series plates containing ytnr c.cm. and TWO c.cm. 

 were inoculated and counted with a low-power glass after forty- 

 eight hours incubation at 37 C. Porous covers were used to 

 prevent loss of plates by spreaders. In 142 samples the differ- 

 ence between duplicate determinations varied from zero to 

 464 per cent with an average variation of 24.7 per cent. Ex- 

 pressed as 'a variation factor the average was 1.25 (1.247) with a 

 maximum of 4.64. The bacterial count varied from 1600 per 

 c.cm. to 1,200,000 per c.cm. and it was with the best grade milks, 

 i.e., those containing less than 10,000 per c.cm., that the vari- 

 ations were the largest. This was anticipated from a consider- 

 ation of the frequency distribution in the largest amount of 

 sample plated and could have been reduced by inoculating 

 larger quantities. This was not done because the labour in- 

 volved in so treating all samples, when but very few were of 

 this grade, was not justified by the increased precision so ob- 

 tainable, for whether a sample contains 1600 or 5000 organisms 



