224 G. C. SuppLEE, W. A. WHITING, AND P. A. Downs 
Sherman (1916) points out the higher counts obtainable by the use of 
lactose agar in place of plain agar, and also the increase in the size of the 
colonies and the better differentiation of the types. 
Breed and Stocking (1917) published a preliminary report on a series of 
comparative determinations, in which they conclude that the plate method, 
when used by careful workers, will give more reliable results than those 
reported by Conn, which had been obtained under routine conditions 
and possibly, in some instances, by inexperienced operators. Obviously, 
inexperience and carelessness are factors to be avoided in any method of 
enumerating bacteria, especially when the results are for the determination 
of municipal regulations. The same authors (1920), reporting a similar 
but more extensive investigation, found the plate method and the micro- 
scopic method (Breed method) productive of reasonably uniform and 
accurate results for the total number of bacteria present, all factors known - 
to introduce inaccuracies having been first reduced to a minimum. For 
the plate method they report an average coefficient of variability of 8.3; 
for the microscopic determination of groups of bacteria, consisting of one 
or more individuals, 11.7; and for the microscopic determination of individ- 
aal bacteria, 13.4. Altho these results are remarkably uniform, it must 
be remembered that they are obtained from samples which were artificially 
inoculated in order to reduce the clumping tendency to a minimum, and 
that the time and labor necessary to obtain this degree of accuracy by the- 
microscopic method could not be expected in regular, routine examinations. 
. PRESENT EXPERIMENTS 
Comparison of media and incubation temperatures 
The experimental work reported herein was for the purpose of demon- 
strating the variations in counts obtained by plain and carbohydrate 
media at different incubation temperatures. It may indicate a further 
reason why the count at 37° C. for forty-eight hours, as used in routine 
work, may be more subject to discrepancies than counts obtained from 
longer incubation periods at lower temperatures. 
The samples. used for this work were selected at random from the 
ordinary market milk of the Ithaca city supply, at intervals extending 
over a period of one and one-half years.. Twenty-seven plates were made 
from the same dilution of each sample. Nine of the plates were 
