It follows that it is impossible to accept the hypothesis of the “deviation of the complement” suggested by Kyes and Sachs to explain the innocuousness of strong doses of venom. Besides, it would be difficult to reconcile this hypothesis with the fact, observed by Noguchi, that venom in a strong dose protects corpuscles, not only against the action of lecithin (complement), but also against distilled water, ether, &c. Noguchi, seeking more thoroughly to elucidate the mechanism of this protective action, finds that _Cobra_-venom forms a precipitate with blood-serum, when the latter is relatively poor in salts or when it is dilated with water. It likewise forms a precipitate with the aqueous extract of red corpuscles, and precipitates the globulins, hæmoglobin, or globin of the corpuscle, when treated separately. The precipitates are insoluble in water, but dissolve with the assistance of a small quantity of acid or alkali, and also in a great excess of saline solution. Noguchi supposes that red corpuscles, when treated with strong solutions of venom, are protected against destructive agents on account of the formation by the venom and certain constituents of the corpuscle (chiefly hæmoglobin) of a compound insoluble in water. When this compound is removed by repeated washings in physiological solution, the corpuscles can easily be hæmolysed afresh by the ordinary destructive agents. Venom, none the less, exerts a noxious influence upon the corpuscles in all cases; but when strong solutions are employed, this effect is masked by the protective action. All kinds of red blood corpuscles are not equally sensitive to the protective action of strong doses of venom. In this respect all degrees are observed in the action of venom. Thus the corpuscles of the dog are not protected at all by _Cobra_-venom. But it is interesting to observe that this venom in no way precipitates either the aqueous extract of dog’s corpuscles, or the hæmoglobin, or the globin of this animal. The venom of _Crotalus_ and that of _Ancistrodon_ likewise possess protective power, which is, however, less marked than in the case of _Cobra_-venom. Noguchi finally points out that corpuscles treated with venom are not hæmolysed by fluorescent substances such as eosin. They are also refractory to the hæmolysing action of tetanolysin. The resistance of the hæmolysins of venom to heat (which, according to Morgenroth, may extend to heating for thirty minutes at a temperature of 100° C.) explains how it is that the serum of horses immunised by means of venoms heated to 72° C. is distinctly antihæmolysing, and capable of perfectly protecting the red corpuscles _in vitro_ and _in vivo_.