18 BULLETIN 642, U. S. DEPARTMENT OF AGRICULTURE. 
almost always a lower count than that from the unwashed udders. - 
The results again lead to the conclusion, as in Experiment No. 3, that 
three simple factors are necessary for the production of milk practi- 
cally free from visible dirt and with a low bacterial content when 
drawn, namely, sterilized utensils, clean cows (especially the udder 
and teats), and the small-top pail. 
During Experiment No. 4 the udders of two cows were washed and 
two left unwashed, and frequently the practice was reversed. The 
change was made in order to eliminate the bacterial variation due 
to a difference in number of bacteria in the milk from the udder 
of each cow. 
Occasionally the bacterial content of the middle milk drawn di- 
rectly from the cow was determined, as in Experiment No. 3. The 
average bacterial content of such milk drawn directly from the un- 
washed udders was 757 per cubic centimeter, and from the washed 
udders 739. Subtracting these numbers from the averages shown 
in Table 4, there remain differences of 3,767 bacteria per cubic 
centimeter in the case of the unwashed udders and 1,415 in the case 
of the washed udders, which represent the external contamination. 
However, as already pointed out in connection with the previous 
experiment, there is probably less external contamination than the 
figures show. 
The small additions of bacteria in milk from both washed and 
unwashed udders must be kept in mind, as they represent the total 
external contamination caused by factors formerly regarded as 
important but which have not been considered in these experiments 
in the production of low-count milk, which also was relatively free 
from visible sediment. 
In order to show further that the fresh milk produced under such 
conditions was of high quality from a bacteriological standpoint 
and compared closely with the middle milk as drawn from the 
udder, the bacterial groups in a number of samples were determined 
by what is designated as the milk-tube method. This consisted of 
picking off each colony on a plate and inoculating into litmus-milk 
tubes. Tubes were incubated for 14 days at 30° C. (86° F.), and the 
bacteria from the plate were then divided into groups according to 
the reactions produced. By that method it was possible to divide 
the bacteria developing on a plate into 6 groups, namely, the acid- 
forming, coagulating-acid forming, the inert (which produce no 
change in milk), the alkali-forming, the peptonizing, and the acid- 
coagulating peptonizing groups. A number of samples of milk were 
studied by that method, and as often as possible samples were taken 
directly from the udder at the same milking. The results of the 
alg dle (a cli iil 
