INTRODUCTION 167 



same. The bacteria always remain alive in the antibody-com- 

 plement mixture for a much longer time than in pure physiological 

 saline. 



These experiments show that cholera vibrios, or dysentery 

 bacilli, are rapidly destroyed in saline; that they remain alive 

 longer in the presence of a normal fresh serum containing com- 

 plement or in an antiserum; and that they remain alive still 

 longer in the presence of both the antiserum and complement. 

 It is therefore evident that the bacteria, sensitized and having 

 fixed complement, far from being subjected to lysis, are more 

 resistant than normal bacilli. 



The sera termed " antibacterial' ' do not, in vitro at least, play 

 any bacteriolytic role. Everything indicates that it is the same 

 in vivo. We know that at times the antisera, derived from horses 

 thoroughly immunized by the injection of living bacilli, are con- 

 taminated by bacilli of the same species as those which have been 

 injected, and that it is possible to demonstrate them by culture. 

 Anti-rouget serum provides a remarkable example. The serum 

 from a horse hyperimmunized by a series of injections of living 

 culture possesses extremely marked curative and preventive 

 properties, but it is, nevertheless, rather frequently contaminated 

 by the bacillus of rouget, even if the serum is withdrawn some teD 

 to twelve days after an injection. How can we reconcile this 

 fact, which indeed is not an isolated observation, with the hypothec 

 sis that the immunity which it confers is, by some mechanism 

 at present unexplained, associated with the presence of an " anti- 

 bacterial antibody?" If it should produce bacteriolysis, it cer- 

 tainly ought to do so in the hyperimmunized animals themselves, 

 where the bacterium finds itself in contact with an abundance of 

 antibody and of complement. 



On the other hand, sera very rich in antibody may entirely 

 lack preventive power, and the opposite is also true. Metchni- 

 koff and his collaborators have furnished many examples of this. 



If we pass to a consideration of natural immunity we readily 

 discern that there is absolutely no parallelism between the state 

 of the patient and the antibody content of the blood. In human 

 typhoid fever, for example, the fatal relapses may occur when the 

 antibody is at its maximum. Finally, in diseases which result 



