There seems no doubt that certain cases of paratyphoid food poisoning are caused by contamination of the food during preparation and are, sometimes at least, due to infection by human carriers. The bacilli in such cases are usually (according to many German investigators) or always (according to most English bacteriologists) of the _B. suipestifer_ type. Other cases are due to pathogenic bacteria derived from diseased animals, and these bacteria are often, possibly always, of a slightly different character (_B. enteritidis_ Gärtner). It is still unsettled whether both types of food poisoning bacteria are always associated with disease processes of man or animals, or whether they are organisms of wide distribution which may at times acquire pathogenic properties. In certain regions, as in North Germany and England, such bacteria are rarely, if ever, found except in connection with definite cases of disease. In parts of Southwest Germany, on the other hand, they are said to occur with extraordinary frequency in the intestines of healthy men and animals. Savage[74] believes that there is some confusion on this subject owing to the existence of saprophytic bacteria which he calls "Paragaertner" forms and which bear a close resemblance to the "true" Gärtner bacilli. They can be distinguished from the latter only by an extended series of tests. The bacilli of this group show remarkable variability, and in the opinion of some investigators "mutations" sometimes occur which lead to the transformation of one type into another.[75] In spite of the present uncertainty regarding the relationship and significance of the varieties observed, a few facts emerge plainly from the confusion: (1) The majority of meat poisoning outbreaks that have been bacterially studied in recent years have been traceable to one or another member of this group and not to "ptomain poisoning." (2) Bacteria of the _paratyphoid enteritidis_ group that are culturally alike but agglutinatively dissimilar can, when taken in with the food, give rise to identical clinical symptoms in man. (3) Food poisoning bacteria of this group, when derived directly from diseased animals, seem more likely to be of the Gärtner type (_B. enteritidis_) than of the _B. suipestifer_ type.