4 BULLETIN 940, U. S. DEPARTMENT OF AGRICULTURE. 
varied slightly in that sterile paraffin was poured into each tube after 
heating. This formed a plug over the milk, and it was not necessary 
to place the tubes under anaerobic conditions. 
It was considered advisable to try the test on milk produced under 
conditions that would represent the worst grade of milk which might 
be encountered under commercial conditions. In order to do this, 
four cows were placed in a small barn which had been used for simi- 
lar experimental purposes. The loft above the cows was composed 
of narrow boards laid from 1 to 2 inches apart. Hay and cobwebs 
hung down from these openings. The walls were soiled with manure 
and dirt. All the cows were allowed to become dirty and their 
udders and flanks were more or less covered with partly dried ma- 
nure. The manure was removed from the floor only twice a week. 
Open pails, not sterilized, were used for milking. 
To show the relation between the Savage sporogenes test and the 
milk, the sediment from 1 pint of milk, the total count, and the re- 
sult of the Savage test are shown in Plates I, II, and III. In the 
upper right-hand corner of each square is a number designating the 
number of tubes showing a positive sporogenes test out of the 10 
tubes used for each sample. Keeping in mind the arbitrary standards 
set by Savage (that is, or 1 -j- = good milk, 2, 3, or 4 + = unsatis- 
factory milk, 5 or more -f- = bad milk) the results are interest- 
ing. It will be noted that according to this test the milk from 
Samples 1 to 35, inclusive, would be called good milk. It is believed 
that the sediment disks and counts make further discussion unneces- 
sary. Particular attention is called to the difference in sediment be- 
tween Sample 1 and Samples 19 and 20. None of the three showed a 
positive test by the Savage method. On Plate III samples from 36 
to 52, inclusive, would be classed as unsatisfactory by the Savage 
sporogenes test. No one would dispute this statement, although 
many were no worse than those called good on Plates I and II. 
Samples 53, 54, and 55 are classed as bad by the test, yet they are no 
worse than some called good. 
It is further evident from the results shown on the plates that there 
is no relation between the sporogenes test as used by Savage and the 
total count. In this connection it may be noted that the milk ex- 
amined was fresh milk. Savage also found very little relationship 
between the test and the total count. 
The question naturally arises in connection with the sporogenes 
test as to the accuracy of the test itself. Will a number of tests with 
a given sample of milk show the same results? To answer this ques- 
tion, 5 sets of 10 tubes each, with 2 c. c. of milk in each tube, were 
prepared from a sample of milk. In other words, the Savage method 
was applied 5 times to, the same sample of milk. From the results of 
