This detailed protocol has been recited in considerable length because it contains what appears to be unequivocal evidence that a strain (Rawlin) of B. typhosus has been filtered not once, but actually thrice serially through Berkefeld filters, the first two filtrations through Berkefelds of N porosity, the third filtration through a Berkefeld of the W porosity, making three filtrations in all in a period of four days. These filtrations were performed on November 3, 5 and 6, respectively. Between filtrations, and prior to each of them, growth was elicited in K medium. The organisms developing after the first, the second and the third filtrations (November 3, 5 and 6) were recovered successively in the non-filterable state in nutrient broth and tested for their identity and purity by sugar reactions, by agglutination with specific typhoid serum and by plating upon agar. These tests were performed not only with the cultures of filtrates in plain nutrient broth, but also with subcultures made from K medium, in plain nutrient broth, with the exception of the K culture from the third filtration. This failed to grow in nutrient broth, although it grew well in successive K medium transfers. Concerning the proportion of filterable to non-filterable forms, no definite information is available. However, inasmuch as the cultures were diluted with salt solution to 1/5 their original concentration prior to filtration, and inasmuch as 1 drop and 1/10 cc, respectively, of the diluted filtrates gave growths that were subsequently identified as B. typhosus, by fermentation reactions, by agglutination and by plating, it may be assumed that at least 1/50 cc of the unfiltered cultures in K medium contained viable typhoid microbes in the filterable state.