In the third chapter I stated that serums were known which give precipitates only with the blood of man and of his near relatives the anthropoid apes. Serum of this kind has a definite specific action. Serums may be prepared that dissolve only the red corpuscles of particular species of animals, and that are without action on the other organic elements. It has been found possible, even, to prepare a serum that arrests instantaneously the movements of human spermatozoa, and that is neutral to the similar cells of other animals. These serums are all prepared in the same way. The cellular elements in question, spermatozoa or red corpuscles, cells of the liver or of the kidney, taken from one animal, are injected into an animal of another species. After several injections have been made, the serum of the animal operated on becomes active with respect to the cells introduced into its body. These serums were discovered by J. Bordet of the Pasteur Institute, but the results have been confirmed by investigators in other countries. The serums are specifically _cytotoxic_, that is to say, they poison particular kinds of cells. Now it has been shown that such serums, employed in small doses, do not kill or dissolve the specific tissue elements, but actually strengthen them.[299] Here the case is analogous with the action of poisons, such as digitalis, which kill in strong doses, but which in weak doses improve or strengthen the action of certain tissues. In accordance with this indication, experiment has shown that small doses of a serum which is capable of dissolving the red corpuscles of human blood, actually increase the number of those in the body of a patient treated by injections. In the same way, in the case of a serum large doses of which destroy the red corpuscles of a rabbit, small doses increase the number of these elements in the blood.