246 G. C. SuppLEE, W. A. WuiTING, anv P. A. Downs 
bacteria counts from miscellaneous samples of milk. Plain agar at 37° 
for forty-eight hours is unquestionably the least favorable combination] 
for this purpose; the use of lactose agar at this temperature appears tof 
have few, if any, advantages over plain agar; and altho dextrose agar at 
37° has distinct advantages over those media, nevertheless the majority 
of the results obtained from it are lower than those produced at 20° or 
30° for five days. 
For developing the maximum counts, dextrose agar at 30° for five days 
seems to be superior to any of the other combinations considered in this 
paper. This medium at 20° for five days is also preferable to plain or 
lactose agar at either 30° or 20°. The same lack of a distinct superiority 
of one of these latter two media over the other exists at the lower tem- 
peratures, as well as at 37°. 
Counts obtained at 37° after forty-eight hours are probably subject to 
greater discrepancies than those obtained at the lower temperatures for 
longer periods of time. It has been shown that the normal variations in 
temperature thruout a mass of tightly packed plates is sufficient to cause 
as high as a fiftyfold variation from the same sample of milk; whereas 
only a threefold variation was found when the plates were arranged to 
allow a free circulation of air around each one. To avoid gross discrep- 
ancies, it 1s necessary, on the basis of the previous results, to provide 
such ventilation in 37° incubators as will heat each individual plate in a 
block at an approximately equal rate. 
Possible variations in bacteria counts resulting from the present plate 
method of enumeration should not be considered as a condition eliminating 
their usefulness. Bacteria counts, together with the discrepancies to which 
they are subject, should be considered only with full knowledge of their 
limitations and of the fact that they constitute but one item of the 
evidence necessary to grade milk into distinct classes according to its 
wholesomeness and keeping quality. 
In order to harmonize these variations with existing numerical bacterial 
standards, it is essential that all factors tending to cause variations and 
discrepancies be reduced to a minimum. Furthermore, results obtained 
from such manipulations, even tho necessitating the statement of fixed 
numbers, should be interpreted in a manner which recognizes the lack 
of intrinsic value of these numbers for denoting such exact degrees of 
bacterial quality as the statement of fixed numerical expressions implies. 
