A variety of suggestions have been made, several of which can be shortly dismissed. It was thought, at one time, that infection might be conveyed by the animal eating the infected fleas. But it is very improbable that this means of infection is of any real importance, even if it may sometimes occur. Experiments in feeding have shown that an animal is unlikely to become infected by swallowing material containing plague bacilli, unless the amount is considerable. Moreover we know that infected fleas confined in test-tubes readily convey the disease when allowed to bite an animal. In such cases the situation of the primary bubo corresponds with the area of skin upon which the fleas are placed. That the transmission of plague is due to the _bite_ of the flea seems abundantly clear. It has also been suggested that the proboscis of the flea acts as a mechanical instrument for the transference of the bacilli. No doubt the outside surface of the flea’s proboscis must become contaminated, when it sucks the blood of a plague-stricken rat; but it is difficult to suppose that contamination of the proboscis can explain cases of continued infectivity during which the flea has been feeding regularly upon healthy animals. Next, there is a hypothesis that the salivary glands of the flea become infected and that the bacilli are inoculated along with the saliva. The reader will remember that when a flea sucks, a stream of saliva is pumped down the mandibles into the puncture. But this hypothesis is shattered by the fact that plague bacilli are apparently confined to the alimentary canal of the flea, and that they have never been found in the salivary glands. An apparently more probable explanation, that the contents of the stomach (in which as we know the bacilli may multiply) are regurgitated and transferred to the wound by the mouth-parts, is rendered less credible when we remember that there is a valvular arrangement at the opening of the flea’s stomach which seems to make such a thing impossible.