204 THE PRINCIPLES OF IMMUNOLOGY 



and careful study of the cases clinically, as well as a failure to observe 

 minute details of the test. 



Antigens are of the utmost importance, and numerous forms have 

 been suggested. There appears to be well-founded evidence for using 

 several strains of the human type bacillus associated with one or more 

 strains of bovine type. The methods of making antigen vary and 

 include the use of saline suspensions or extracts of tubercle bacilli, 

 living or dead, intact or pulverized ; filtrates of broth cultures ; ether 

 alcohol extracts of whole or autolyzed bacilli, and extracts of tubercu- 

 lous organs. Apparently those extracts which contain both lipoids and 

 proteins are most satisfactory. The antigenic substance is thermostable. 



The human serum is inactivated, and in Petroff's hands appears to 

 be most satisfactory if collected one or two days before the test. 

 Accurate titration of complement, to be used in doses of two units, 

 and of hemolytic amboceptor is essential. Guinea-pig complement and 

 a sheep-rabbit hemolytic system are satisfactory. It is absolutely essen- 

 tial that glassware be perfectly clean and that measurements be accur- 

 ate. The incubation of the mixture of antigen, tuberculous amboceptor 

 and complement should be from one and one-half to two hours at the 

 optimal temperature of 35-4O. 



Wilson, using a lipoid-free bacillary antigen, attaches great im- 

 portance to the complement and finds that there is not a universal 

 adaptability of guinea-pig complement. That from some guinea-pigs 

 appears to be fixed more readily than that from others. Therefore, in 

 general, pooled complements are likely to give the best results. If a 

 single complement is used tests should be made to determine the extent 

 of fixation. Von Wedel states that preservation of the patient's serum 

 in the ice-box for five to seven days favors the reaction, but Petroff 

 found that fresher sera are preferable. It is desirable to make several 

 tests at intervals upon the same patient. As the result of 1555 tests on 

 713 cases Petroff obtained the following results: 



Cases Positive Negative Percent. 



Clinically active tuberculous 212 199 13 93.9 



Quiescent tuberculous 158 89 69 56.3 



Apparently cured more than two years 58 5 53 8.5 



Normals 78 3 75 3.8 



Suspected 166 65 101 39.1 



Other diseases 41 6 35 14.6 



An -analysis of these figures shows that under proper conditions 

 complement fixation is of distinct value in the diagnosis and prognosis 

 of tuberculosis. Basing his conclusions on experimental data Petroff 

 considers " the complement-fixation test in tuberculosis more specific 

 than the Wassermann test " in syphilis, an opinion in which we concur. 

 Nevertheless, its most ardent advocates do not regard the test as pathog- 

 nomonic and Petroff regards it as " only one of the many links in the 

 tuberculosis diagnostic chain." It is unfortunate that the complement- 

 fixation test gives the highest percentage of positive results in cases in 

 which the need for such diagnostic aid is least evident, namely in those 



