48 
Journal of Agricultural Research 
Vol. VIII, No. ? 
milligram. The second weight of the paper and bottle was used, regard¬ 
less of whether it was higher or lower than the first. 
After being used for filtration, the papers containing a precipitate were 
replaced in their respective bottles, dried for 14 hours in the air oven at 
ioo° C., and weighed. They were then dried a second time for two hours 
and weighed. It was found that 14 hours* drying almost always dried 
the papers and precipitates completely; the second weighing differed 
from the first only by 2 or 3 mgm. Occasionally, when the second 
weighing was less than the first by 6 mgm. or more, a third drying and 
weighing was made of the particular bottle. The lesser of the two 
weights was used. As is well known, the increase in weight after pro¬ 
longed drying is due to oxidation. 
A large number of blank filtrations, 23 in all, were made as follows: 
Through a weighed paper a filtrate obtained from a globulin or similar 
determination was passed. The paper was washed free from sulphate, 
then with alcohol and ether, dried 14 hours, and weighed as if it had 
contained a precipitate. The object was to ascertain the extent of the 
change in weight due to the mechanical ^handling, drying, etc. Out of 
the 23 blanks, the differences were 3 mgm., or less, in 18 blanks; generally 
there was a slight increase in weight. In the 5 other blanks the differ¬ 
ences were 5, 5, 6, 6, and 7 mgm.; 3 of these were gains and 2 were losses 
in weight. These blanks were not used in correcting the weights of 
precipitates, as they were small enough to be negligible. 
In so far as the weight of a precipitate is obtained by difference, it is 
obvious that care must be used to be certain that the conditions of dry¬ 
ing, etc., are such as to lead to a minimum of error. It is probable that 
some investigators have not realized that errors of a few milligrams are 
almost unavoidable; one investigator using this method published results 
to the tenth of a milligram. It is believed that a minimum of error will 
result under the following conditions: If the first and second weights of 
the empty dry papers generally differ by 3 mgm. or less, either weight 
should be used consistently—that is, either the first weight should be 
used throughout or the second weight throughout, regardless of whether 
the differences are plus or minus. After drying the paper and precipi¬ 
tate, the minimal weight should be used, regardless of whether this is 
the first, second, or even third. Should differences of more than 3 mgm. 
appear often between consecutive weighings, it is probable that the 
drying was incomplete. 
The analytic data obtained on the serums of horses 48 and 96 and 
mules 148 and 149 are contained in Table V, together with the data 
obtained on the globulin preparations. The data in Tables I to IV are 
obtained from those of Table V. It will be apparent from the last 
table that when a different technic is used in the determination of globulin 
in the same serum the results are different. Whenever several results 
were obtained on one serum that result obtained by single precipitation 
and centrifuging was regarded as correct. The variations from this 
