" precipitable substance." To this end lie added a drop of 

 antiserum and a drop of antigen to separate drops of each 

 superfluid. With the microscope he sought twenty minutes 

 later for a precipitate, find in each series he found a "middle 

 zone" in which the superlluid gave no precipitate with 

 antiserum nor with antigen. He concluded that in this 

 "middle zone" of the series the precipitate contained the 

 whole of the precipitin and antigen of the original mixtures. 

 He also made certain adsorption experiments to show the 

 variable composition of the precipitate. 



We have carried out many thousands of interactions of 

 the type employed by Yon Dungern, and we have tested 

 the superfluids with both antiserum and antigen (12, 16b 

 But we have found that in every case the subsequent 

 addition of antiserum lias led to the formation of a further 

 precipitate. The subsequent addition of antigen has, or 

 has not, yielded a precipitate, according as the original 

 interaction was partial or total (v. supm). We would draw 

 attention to the larger quantities of superfluid examined by 

 us, and the much greater time— frequently 48 hours- 

 allowed for interaction. 



P. T. Miiller (20) lias compared the elfect of lactoserum 

 on milk with the coagulation of caseinogen by rennin. He 

 has shown that lactoserum precipitates casein, and in many 

 other ways resembles rennin, e.<j. precipitation by lacto- 

 serum Is dependent on the presence of the salts of lime or 

 certain other divalent elements. The close similarity 

 between the actions of lactoserum and rennin indicates 

 that antisera prepared by injection of milk differ funda- 

 mentally in their action from antisera prepared with blood 



Phenomena of inhibition.-We have elsewhere discussed 



our experiments on lOhrlieh's hypothesis (29, 30). If 



