It seems possible that rats and mice may become infected with certain bacteria of this group without human intervention, and that these infected animals may be the means of contaminating foodstuffs and so causing outbreaks of food poisoning. Proof of the frequency with which this actually occurs is naturally difficult to obtain. There is no escape from the conclusion that in any given case of food poisoning the exact source of infection is often largely conjectural. Even when suspicion falls strongly on a particular article of food, it may not be possible to establish beyond a reasonable doubt whether the material (meat or milk) came from a diseased animal or whether it was infected from other sources (man or other animals) at some stage during the process of preparation and serving. The most definitely attested cases yet put on record are those in which it is possible to trace the infection to food derived from an ailing animal. _Means of prevention._--The most obvious and probably the most important method of preventing infection with paratyphoid bacilli is the adoption of a system of inspection which will exclude from the market as far as possible material from infected animals. To be most effective such inspection must be directed to examination of the living animal. The milk or the meat from diseased animals may give no visible sign of abnormality. In the Ghent outbreak of 1895 the slaughter-house inspector, a veterinary surgeon, was so firmly convinced that the meat which he had passed could have had no connection with the outbreak, that he ate several pieces to demonstrate its wholesomeness. The experiment had a tragic ending, as the inspector was shortly attacked with severe choleraic symptoms and died five days later, paratyphoid bacilli being found at the autopsy. Müller[89] also has described a case in which paratyphoid bacilli were found in meat that had given rise to a meat poisoning outbreak although the meat was normal in appearance and the organs of the animal showed no evidence of disease to the naked eye. It is evident that inspection of the live animal will often reveal evidence of disease which might be missed in the ordinary examination of slaughter-house products. Although inspection of cows used for milking and of food animals before slaughter is highly important, it does not constitute an absolute protection. Emphasis must be repeatedly laid on the fact that meat, and especially milk that is derived from seemingly healthy animals, may nevertheless contain paratyphoid bacilli. To meet this difficulty in part the direct bacterial examination of the carcasses of slaughtered food animals has been proposed, but this seems hardly practicable as a general measure. In spite of all precautions taken at the time of slaughtering it seems probable that occasionally paratyphoid-infected meat will pass the first line of defense and be placed on the market.