1. After injection into the renal medulla of rats Escherichia coli 06 variants reverted rapidly in vivo in the absence of penicillin. These variants had previously been shown to be stable in vitro. 2. Variants failed to survive following intramedullary injection when animals were receiving penicillin. 3. Late reversion of variants also failed to occur in animals treated with penicillin for only 1 or 2 days. 4. Variants survived and reverted more readily when injected in the renal medulla, compared with liver and spleen. Classical bacteria injected into the kidney, liver, and spleen were recovered in approximately equal numbers. 5. The histologic response to nonreverting variants, medium not containing variants, and killed variants was similar and was characterized by a fibrotic reaction with moderate round cell infiltration. 6. In contrast, the histologic response to reverting variants and to classical E. coli was characterized by an intense, acute, polymorphonuclear leukocytosis typical of acute pyelonephritis.

The "endogenous serum pyrogen" that appears in the circulating blood after a single intravenous injection of endotoxin does not produce leukopenia in normal animals, fails to provoke the local Shwartzman reaction, and elicits no "tolerance" when injected daily. Suppression of the febrile response to endotoxin by prednisone does not prevent the appearance of pyrogen in the blood. Animals given large amounts of endotoxin daily continue to respond with high fevers despite failure of endogenous serum pyrogen to appear in detectable amounts after the first two or three injections. Analysis of the response to daily injections shows clearly that the fever during the first 2 hours after administration of endotoxin is unrelated to levels of endogenous serum pyrogen; in contrast, the magnitude of the fever after the 2nd hour correlates well with endogenous pyrogen in some instances. The leukopenic response to endotoxin could not be correlated with the appearance of endogenous serum pyrogen. The differences between endotoxin and endogenous pyrogen and the similarities between leukocyte extracts (sterile exudates) and endogenous pyrogen are summarized and discussed. Dissociation of the febrile response to bacterial endotoxin and levels of endogenous serum pyrogen are discussed and it is concluded that a mechanism involving both direct and indirect action of endotoxins offers the best explanation for the pyrogenic action of these bacterial products.

Intravenous administration of bacterial endotoxins in dogs is followed within 2 hours by the appearance of a fever-producing substance in the blood. This endogenous pyrogen differs from the endotoxins originally administered by its ability to produce fever in tolerant recipients and failure to promote tolerance after repeated daily injections. Endogenous serum pyrogen is destroyed by heating at 90°C. for 30 minutes, and is also inactivated to some degree by incubation at 37°C. for 24 hours. Suppression of fever by aminopyrine does not affect appearance of the endogenous factor. Animals made febrile with dinitrophenol, kaolin, or lysergic acid do not elaborate a fever-promoting substance in the blood. Sterile abscesses, accompanied by elevations in body temperature of the host, are unassociated with detectable amounts of secondary pyrogen in the serum. The absence of endogenous pyrogen in the blood of febrile dogs made leukopenic with nitrogen mustard favors the idea that polymorphonuclear leukocytes injured by endotoxins release the endogenous factor. On the other hand, the finding that the granulocytopenic animals are febrile when no circulating endogenous pyrogen is present, casts doubt upon the essential role of this substance in endotoxin fever.