still quite obscure. A satisfactory explanation, based on experimental evidence, has not yet been offered as to why the valves are involved so much more frequently than the mural endocardium. While it is generally assumed that the bacteria localize by implantation, the greater frequency of mitral and tricuspid than semilunar endocarditis remains to be explained. Kbster' demonstrated that extension in an established human endocarditis may be due to embolism and suggested that endocarditis may begin as an embolic process. Experimental proof of this mode of origin is still lacking, however. Lissauer," after an extensive review of the literature, concludes that while endocarditis may occur exceptionally as an embolic process, no one has been able to show experimentally that this mode of origin really plays a r61e in the localization of bacteria on the heart valves. The great difficulty of producing endocarditis experimentally without previous injury to the valves is well known. Ribbert3 produced endocarditis in rabbits by intravenous injections of emulsions of staphylococci and potato particles, and showed that the endocarditis was due, at least in some cases, to implantation. Orth and Wyssowitsch,4 and Fulcis had similar results from injections of suspensions of staphylococci and also of streptococci with finely pulverized charcoal. These authors were not able, however, to produce endocarditis by intravenous injection without the foreign particles. Recently Lissauer obtained endocarditis in two of 20 rabbits after repeated intravenous injections of a non-virulent, white staphylococcus. One rabbit died in two months after six injections, the other in six months after io injections, both showing