SUMMARY : An examination of freshly isolated Haemophilus pertussis strains for agglutinability, growth requirements, haemagglutination, virulence for mice, alum precipitability and protective properties shows that each of these characters varies quantitatively, and furthermore that they vary independently. The presence of one strongly marked character is no indication that another character will be strongly marked or even present. Not all strains necessarily have the characters to be expected in freshly isolated strains ; one or possibly more may be missing. On subculture of the strains on Bordet-Gengou media the order in which these characters degenerate or are lost and the rate at which they are lost differs with the strain. Agglutinability and agglutinogen content vary considerably in freshly isolated strains, and therefore the expression 'agglutinates to titre' cannot be considered as a criterion of a Phase I organism. The obvious complexity of the serology of organisms within ' Phase I ' suggests that this term has served its useful purpose and should now be dropped as a description of a certain type of strain. The term 'Smooth', which indicates the possession of the antigenic components commonly associated with the pathogenic state of the organism, is preferable. Leslie & Gardner (1931) analysed thirty-two strains of Haernophilus pertussis by agglutination and agglutinin absorption and found that they fell into four well-marked groups which they called Phases I, 11, I11 and IV. They suggested that Phases I and I1 probably correspond to the smooth (S) and Phases I11 and IV to the rough (R) phase of other bacteria. This work was confirmed by Lawson (1939) with respect to Phases I, I11 and IV but he did not find any strains in Phase 11. He stated, however, that ' there are undoubtedly other intermediate stages in the dissociation process, their number dependent on the strain of the organism and its environmental condition ', and suggested that the degradation proceeds gradually, often making it difficult to determine the border-line between the phases. The investigation of H. pertussis has tended to proceed along one of two lines ; either the investigation of the agglutinogenic relationship of the phases, as made by Flosdorf, Dozois & Kimball (1941), who presented evidence that variation is even more complex in minor relationships than former workers had found, or the characterization of Phase I organisms in the hope of finding a single character which would clearly distinguish organisms in this phase. What then is Phase I, and are all organisms in Phase I identical? Historically Phase I was defined serologically, but recent work (Smolens, McLaren, McAleer & Flavell, 1947; Kendrick, Updyke & Eldering, 1949; Cohen, 1949) has shown that freshly isolated strains may vary considerably in their agglutinogen content, and that a small number of typical strains are of con- sistently low agglutinability. Leslie & Gardner implied, though they did not