BACTERICIDAL TEST-TUBE EXPERIMENTS. 137 



of the postulates of Ehrlich's theory, and it is this quantitative 

 difference which constitutes the essential point in the deflection of 

 complements described. 



The following table includes ten different goat sera, among them 

 three bactericidal immune sera (columns 2, 3, 4), one antitoxic serum 

 (column 5), four hsemolytic immune sera (columns 6, 7, 8, 9), and 

 one anticomplement serum directed against the complements of 

 horse serum (column 10). All of these have been tested as to their 

 complement-deflecting power against vibrio Metchnikoff. 



The experiment has been slightly modified from the former, 

 for in this I made use of a mixture of 0.1 cc. active guinea-pig serum 

 (in itself not bactericidal, Control II) plus 0.01 cc. inactive goat 

 immune serum (against vibrio Metchnikoff). This mixture completely 

 killed the amount of bacterial culture used, namely, 1Tr V(7 cc. of a 

 one-day bouillon culture of vibrio Metchnikoff (see Control II). De- 

 creasing amounts of various inactive goat sera were added to this, 

 as is shown in columns 1 to 10. 



According to this, the Metchnikoff immune serum exerts a specific 

 action, and it is certainly too hazardous to assume that the Metchni- 

 koff goat used by us happened to possess an unusually large amount 

 of normal anticomplement. Furthermore, it is possible, as I shall 

 show later, to furnish positive proof that the deflection of complement 

 is caused, not by the anti complements, but by the amboceptors; 

 for by removing the amboceptors it is possible to prevent the deflec- 

 tion. The following experiment furnishes still further evidence 

 against Metchnikoff 's view: I examined the serum of a rabbit before 

 and after immunization with vibrio Metchnikoff and found that the 

 normal serum was entirely inactive, whereas after eight days the 

 immune serum of this animal caused strong deflection. In these 

 cases, therefore, it will not do to ascribe the deflection of complements 

 *to a normal anticomplement. That normal anticomplements do 

 occur and that they may at times simulate the phenomenon above 

 described is, of course, possible and has already been emphasized 

 by Neisser and Wechsberg. In such cases suitable control tests, 

 above all the absorption method described in the next section, will 

 guard against errors. From all this it follows that the phenomenon 

 of complement deflection which can be observed in suitable cases is 

 not to be ascribed to the presence of a normal constituent but to one 

 produced by immunization. 



