254 LOCAL ILSSl'E REACIIVLiY 



niortalit) occimcd with 200 reading units mixed with 200 anti- 

 toxic units ol serum, tlian with '^00 reactino units mixed witli 

 200 antitoxic luiits ol serum. The difference, although not very 

 striking, may be possibly explained as a prozoiie reaction. 



It becomes also obviotis from Diagram 4 that hemorrhage, necro- 

 sis and subsequent complete regression may be obtained in a large 

 proportion ot tumors with partially neutralized toxic filtrates of 

 low lethal potency. Thus, one combination (Group 2) showed 

 no early mortality (i.e., ^vithin the first twenty-foin^ hotns) , a low 

 late mortality and a strikingly high proportion of severe reac- 

 tions in the tumor tissue and of complete regression. The inci- 

 dence of hemorrhagic necrosis continued to be high as the nimi- 

 ber of toxic luiits in the mixtme Avith the serum ^vas raised. 

 There Avas a somewhat louver incidence of lienKjrrhage and 

 necrosis in mice treated with toxin alone. This was possiljly due 

 to the fact that some mice died too early. An asstimption of this 

 sort may be safely made since some of the reactions recjuire fidly 

 twenty-four hours for their aj)j)earance. It is also seen from Dia- 

 gram 4 that the hemorrhage and necrosis ^vere follo^ved by com- 

 plete regression in a high propc^rtion of animals. Many of the 

 survi\'ing mice ^vith complete regression were kept for periods as 

 Icjng as foiu" to five months. No resiunjjtion of growth or metastasis 

 ^vas obserxed in these mice. From the analysis of the cohunn indi- 

 cating the proportion of tumor regression, it may appear that 

 the incidence of regression decreased as the number of toxic 

 luiits increased. This conclusion ^vould be erroneous, ho^vever, 

 since many of the mice died ^vithin the first t^venty-foiu' horns. 



Thus, the experiments recorded in Diagram 4 proxe definitely 

 that the ability of a given bacterial filtiate to elicit hemorrhage, 

 necrosis and regression of sarcoma 180, is a function independent 

 of its lethal effect, and that the high death rate obtained in 

 tiunor-bearing mice with non-neutralized filtrates was not due to 

 the processes elicited in the tumor. It was thought that high lethal 

 effect of bacterial filtrates upon tumor-bearing mice might be in 

 the nature of an anapyhlactoid or hypersensitiveness reaction 

 made possible by sensitization ^vith secondary bacterial invaders 

 commonly observed in these mice. The following experiment ^vas 

 done in order to study this point; 



Four normal mice (Group 1) and 3 mice bearing twelve day old uunors 

 (Group 2) were each injected intravenously with 1 c.c. of eighteen hour old 

 broth culture of a "rough" strain of B. entcriiiiUs (;^H) diluted 1:100 in 0.85 



