PRODUCTION OF MILK OF LOW BACTERIAL CONTENT. 85 
Throughout the work it was found that, in spite of the variation 
in the number of bacteria in manure, a fairly constant relation 
existed between the quantity of sediment and the bacterial count. 
The relation can be established only when fresh, unstrained milk, 
handled in sterilized utensils, is examined. Some attempts have 
been made by investigators to establish a relation between the sedi- 
ment test and bacterial count in market milk, which, of course, has 
proved impossible, because the history of the milk was not known. 
Campbell (8) draws the following conclusions: “'The quantity of 
sediment or visible dirt present on the disk is no criterion as to the 
kind or number of bacteria contained in the milk.” He made tests with 
samples of milk which were collected on the railroad station platform 
from cans as they arrived from various farmers. The conclusion is 
of course correct, but it should be qualified by a statement as to where 
the samples were taken. In such cases there are three unknown 
factors, namely, whether the milk was strained, whether it was han- 
dled in sterilized or unsterilized utensils, and what proportion of the 
bacterial count is due to contamination and what proportion due to 
growth. If the milk has been strained on the farm it is certain that 
there is no relation between the sediment and the bacterial count; 
neither will there be any if the milk has been handled in unsterilized 
utensils, which may introduce large numbers of bacteria. Neither 
can any relation be expected unless the comparison is made on fresh 
milk, since bacteria multiply rapidly unless the milk is held at a low 
temperature. The only way in which the relation can be established 
between sediment and the bacterial count is by a study of fresh, un- 
strained milk at the farm, and where sterilized utensils are used. 
Under such conditions there is a general relation between the sedi- 
“ment and the bacterial count, as is shown in Plate V, which shows 
‘the sediment disks from a pint of milk together with the number of 
‘bacteria per cubic centimeter in it. At the beginning of the experi- 
ment the floor in the barn was clean, the cows were fairly clean, and 
for a period of nine days the manure was allowed to accumulate on 
the floor. During that time the cows were not cleaned; consequently 
they became dirtier each day. The experiment began on October 12 
and continued to October 23, during which time samples of milk were — 
examined from each morning’s and night’s milking. The first two 
rows in Plate V show the sediment test and bacterial count of samples 
of milk at each morning’s milking, the lower two rows the results 
from the night’s milking. At the beginning, when conditions were 
fairly clean, it will be noted that there was a little sediment and that 
the bacterial count was low. On each successive day, as the cows and 
barn floor became dirtier, the quantity of sediment gradually in- 
creased together with the bacterial count. There were slight fiuctu- 
