BACTERIAL COUNT OF MILK AND DIET TEST. 3 
method was also about two minutes, and we found it to be the most 
satisfactory for field work. 
Fifty samples of milk were collected on the railroad station plat- 
form from milk cans as they arrived from various farmers throughout 
the section. Upon arrival at the laboratory the temperature was 
taken and a bacterial count made. After preparing plates each 
sample was passed through one of Gerber's sediment tubes. The 
sediment disks were kept and compared with the bacterial count. 
A similar comparison was also made with the Wizzard and Lorenz 
apparatuses, using 50 samples in each case. 
After 50 samples had been tested with each apparatus, 20 samples 
were filtered through 4 pieces of cheesecloth, 20 through one thick- 
ness of absorbent cotton, and 20 through one of Canton flannel. 
Each of these samples was then subjected to the sediment test and 
a bacterial count made in each case; this was done to determine 
the effect that straining the milk would have upon the test. We 
also made a comparison of the filtered samples with the bacterial 
count after passing them through the cotton disks used in the Lorenz 
apparatus. 
The writer wishes to thank Dr. John R. Mohler, assistant chief of 
the Bureau of Animal Industry; Dr. Louis A. Klein, dean of the 
veterinary school, University of Pennsylvania; and Dr. C. J. Marshall, 
State veterinarian of Pennsylvania, for many valuable suggestions 
in the work. 
METHOD OF COLLECTING SAMPLES. 
The milk in the can was thoroughly shaken and 1 pint taken as a 
sample. The sediment in this kind of sample would, in our opinion, 
represent the amount of dirt contained in an ordinary bottle of milk. 
A few inspectors believe that the sample should be collected from 
the bottom of the cans before shaking, but it seems to us that this 
may at times be unfair to the producer. 
DETAILS OF THE EXPERIMENTS. 
In our experiments the character and quantity of sediment upon 
the cotton disks is represented by the words "good," "fair,'" "me- 
dium," and "bad." (PI. I, fig. 1.) This gives four classifications, 
which we considered sufficient for all practical purposes. These 
classifications are illustrated in Plate I. 
COMPARISONS WITH UNFILTERED MARKET MILK. 
Table 1 shows the laboratory results obtained by comparing the 
bacterial count with the Gerber sediment test on 10 average samples 
out of 50. 
