The following conclusions may be drawn from the foregoing series of experiments. The complementary activity varies within a definite limit in different specimens of guinea pig serum. With sera which stood in contact with the clot for twenty hours, the strongest and weakest were in the ratio of 0.015 cubic centimeter to 0.04 cubic centimeter. The former was 2.66 times stronger than the latter. The variation observed with the same series of sera after forty-six hours was still more striking. The strongest was 0.013 cubic centimeter, and the weakest, 0.06 cubic centimeter, that is, the former was 4.6 times stronger than the latter. These findings agree with those made by Massol and Grysez. The variations were not so marked with the majority of sera. It is noteworthy that a large number of the sera gained in the complementary activity when remaining in contact with the clot for forty-six hours, while some sera became weakened during the same length of time.
The amount of serum fixed by given constant quantities of syphilitic serum and antigen varies much more markedly than the variations in their complementary activity. One serum failed altogether to be fixed. On the other hand, one sample of serum was so easily fixable that 0.24 cubic centimeter (corresponding to 9.6 complement units of this specimen) disappeared, while the average quantity fixed was only 0.098 cubic centimeter (corresponding to 4.64 complement units). The normal standard of fixability was shown in about 50 per cent. of the specimens examined. If the zone of normal fixability is enlarged in both directions by one unit, the percentage of normal fixability would become 65.8. There is no definite relationship between the complementary activity and the fixability of a given specimen of guinea pig serum.
The facts derived from our present experiments, especially in regard to the exceptions in the fixative quality of this serum, demand the utmost precaution from those intending to employ it for diagnostic purposes, as, for example, in the Wassermann reaction. No quantitative work is possible with the complement fixation reaction unless the experimenter is capable of determining the fixability of the serum in use. One of us (Noguchi) has long realized this source of error, and in order to reduce it he has advised the employment of a mixture of sera from more than two guinea pigs.