These reactions were obtained in four different strains of mice as well as in rats, guinea pigs, and hamsters. They were observed in mice after injection of mouse and egg-adapted virus, and all strains of virus tested gave these responses as long as sufficient quantities of the agents were injected. Control materials, such as normal allantoic fluids, particulate components of normal chorio-allantoic membranes, and suspensions of normal murine lungs, gave uniformly negative results. Activation of a latent virus as well as the inadvertent admixture of a neurotropic agent to the influenza cultures was excluded.

Propagation of the influenza viruses in the central nervous system could not be demonstrated and, in fact, the agents were no longer demonstrable in 4 days. It was concluded that the observed neurological reactions were the result of toxic activities rather than of virus propagation.

Comparison between the infectivity, hemagglutinating capacity, and the toxic activity showed that preparations with the higher titers of virus and hemagglutinin were also the more toxic ones.

The toxic agents could not be separated from the infectious particles by the use of such technics as differential high speed centrifugation, adsorption onto and elution from chicken red cells, and precipitation by protamine.

The toxic effect of influenza A virus preparations was specifically neutralized by anti-influenza A and not by anti-influenza B serum, and conversely. In addition, antigenic differences were noted between two strains of influenza A virus by this method of testing.