INFLUENCE OF DISEASE UPON MILK 89 
each according to the stage of lactation and once each 
week the milk from each group was put into a separate 
can, the cows being groomed and milked in the usual 
manner by the regular attendants. A sample of milk 
was taken from each can for examination. Two guinea 
pigs were inoculated from each sample—one with the 
cream and one with the sediment. This was repeated each 
week for six weeks. Then the cows in the stable were 
again examined in the same way and, no symptoms indi- 
-cating tuberculosis being found, samples of milk were 
collected and examined as before once a week for another 
period of six weeks. Altogether, 96 guinea pigs were 
inoculated. Thirty died of intercurrent disease and the 
other 66 were chloroformed two months after inoculation. 
A post-mortem examination was made of every animal, 
but in no instance were any lesions of tuberculosis found. 
One of the cows in the experiment had reacted to tuber- 
culin over 8 years before, one 7 years, two 6 years, two 4 
years, two 8 years, two 2 years, one 1 year and one 4 
months before. 
These observations show that non-clinical reactors 
play a minor réle in the infection of market milk with 
tubercle bacilli, even when the virulence of the milk is 
tested by the delicate inoculation test. That there is a 
vast difference between the number of tubercle bacilli 
necessary to produce infection by the mouth and by in- 
oculation has been demonstrated by a number of investi- 
gators. Ostertag and others have shown that two and a 
half million times more material is required to infect an 
animal by feeding than by inoculation. Schroeder and 
Cotton found that milk which would produce tubercu- 
losis in guinea pigs when 5 c.c. was injected into the 
peritoneal cavity could be fed 30 days without producing 
