Ten generations of mice (with one exception) were inoculated with a partially purified toxic fraction isolated from Bact. typhi-murium, the next generation being bred each time from the survivors to the test. By this selective process a stock was produced with a substantially increased power of resistance to the toxin. For instance, in the fourth and fifth generations (combined), at which point of time the dose inoculated was stabilized at 4 mg., 36% of the mice with resistant ancestors survived inoculation compared with 12% of normal mice derived from the same original stock; in the ninth and tenth generations (combined) 64% of the resistant stock survived and only 14% of the controls. Also, those mice which failed to survive inoculation took slightly longer to die, on the average, in the resistant stock than in the normal stock.

The mice of each generation were not inoculated until about the age of 12 weeks, but to test the possibility of the transference of a passive immunity from mother to young the seventh generation was bred from without previous inoculation. The eighth generation was composed therefore of descendants of animals selected for resistance by inoculation but its own parents had not been tested. Its survival rate, 59%, was substantially greater than that of the normal stock, 17%, proving that the differences observed in previous generations were not the results of a transference of passive immunity.

Some mice of the ninth and tenth generations were inoculated with the living organism Bact. typhi-murium in place of the toxic fraction derived from it. Although these generations were relatively highly resistant to inoculation with the toxin itself they were less resistant than normal mice to the living organism.