87 
blood serum to cause a visible precipitation when added to horse 
serum. 
The following studies show still more clearly that the specific 
precipitins have nothing to do with this toxic reaction. 
We prepared guinea pigs by injecting them many times at short 
intervals with horse serum, so as to develop a high degree of pre- 
cipitin in their blood serum against horse serum. The blood serum 
of guinea pigs so prepared was then added to horse serum in the 
proportions of about one of the former to two of the latter and 
allowed to stand until the precipitating action was complete. This 
guinea-pig serum was so active that very large quantities of the pre- 
cipitum settled to the bottom of the test tubes. The mixture was 
now filtered, so as to separate the precipitum entirely, and the 
filtrate injected into sensitized guinea pigs and found to be quite 
as toxic as untreated horse serum. 
G. P. No. 107. Seventeen subcutaneous injections of 1 c. c. each normal horse serum (roan) 
and one peritoneal injection of 6 c. c. antitoxic horse serum (Xatl. IX, 19) were given 
between February 8 and March 11. 
Six c. c. normal horse serum (roan) were mixed with 3 c. c. of the serum of this guinea 
pig and allowed to stand about 3 hours at room temperature. The large amount of pre- 
cipitum which developed was filtered off and the filtrate injected into the following two 
guinea pigs: 
G. P. No. 4996. Two c. c. of the above filtrate injected into peritoneal cavity. Dead in 20 
minutes. 
[Previous treatment: 27 days prior, inoculated subcutaneously with 0.22 c. c. toxine 
No. 7+^-0 c. c. antitoxic horse serum (Wellcome 474A).] 
G. P. No. 5000. Five c. c. of the above filtrate injected into the peritoneal cavity. Dead 
in 10 minutes. 
[Previous treatment: 28 days prior, inoculated subcutaneously with 0.22 c. c. toxine 
No. 7+f^o c - c - antitoxic horse serum (Wellcome 477B).] 
These experiments were repeated under varying conditions, espe- 
cially allowing the two serums to remain in contact a longer time 
before filtering, and similar results obtained. 
The results of this part of our work are strengthened by the con- 
clusions of Pirquet and Schick, who found that the antibodies, what- 
ever they may be, producing symptoms of the serum disease, are not 
identical with the precipitins. They found, for instance, that the 
symptoms of the serum disease appear within eight to thirteen days 
following the first injection of horse serum into man, whereas it 
requires about three weeks for precipitins to appear in the blood in 
children after the injection of horse serum. 
The formation of precipitins does not take place as readily in 
man following the injection of horse serum as it does in rabbits. In 
fact, Pirquet and Schick found that sometimes even after the injec- 
tion of 200 c. c. there was no production of precipitins, and, further, 
that the length of time that the precipitin reaction remains in the 
