THE DETERMINATION OF BACTERIA IN ICE CREAM, 7 
The maximum and minimum bacterial count of the samples from 
each gallon of ice cream, together with the per cent of variation, is 
shown in Table II. 
In the samples from Plant No. 1 there was a variation of 300 per 
cent among the samples from lot D, which showed a minimum count 
of 20,000 and a maximum of 80,000 per cubic centimeter. This case, 
as may be noted, is extreme and may be accounted for by the fact 
that the 80,000 count was obtained from one plate only, the duplicate 
being covered with spreaders. Reference to the complete results in 
Table I for this lot shows that among the other samples the colony 
counts were very nearly alike. A similar explanation holds for lot E 
from Plant No. 1, in which there was a variation of 81.81 per cent, 
with a maximum and minimum count of 13,000 and 7,150. 
Throughout the rest of the samples the only high percentage of 
variation was among the samples taken from lot A, Plant No. 6, a 
variation of 105.32 per cent. In this case the melted cream was a 
thick, viscous mass, which made it difficult to measure accurately in 
a pipette. Special care was taken in the remaining determinations 
of bacteria in the ice cream from this manufacturer, with the result 
that the percentage of variation in bacterial counts was very low. 
The general variation among the samples from each gallon of ice 
cream was from 20 to 30 per cent, which is decidedly low, although 
at first thought it may seem high. 
In Reprint 295 of the Public Health Reports 1 it is stated that in 
analyzing duplicate samples of milk the general average variation in 
each of four laboratories ranged from about 110 to 380 per cent. 
Just what percentage of variation in duplicate counts is normal to 
the method of bacterial analysis we shall not attempt to say, but a 
variation of 20 per cent means only the difference between 100 and 
120 colonies on a Petri plate. 
The small variation in our results indicates that in the ice cream 
examined the bacteria were rather evenly distributed and that an 
analysis of one sample taken in the manner described would show 
for all practical purposes the bacterial content of any other sample 
in the 1 -gallon lot. 
i Conn, H, W. Standards for determining the purity of miik. 
