384 BACTERIOLOGY. 
lus which killed by means of diphtheria toxin—or, in 
other words, not simply a virulent bacillus, but a viru- 
lent diphtheria bacillus. When the bacilli to be tested 
grow poorly in the simple nutrient bouillon they should 
be grown in bouillon to which one-third its quantity of 
ascitic fluid has been added. Quite a number of bacilli 
have been met with which killed 250 gramme guinea- 
pigs in doses of 2 to 15 c.c., and yet were unaffected 
by antitoxin. These bacilli, though slightly virulent 
to guinea-pigs, produce no diphtheria toxin, and so 
cannot, to the best of our belief, produce diphtheria in 
man. 
