A total of 454 sampling experiments were performed on the groupings of 100 samples each from separate Poisson populations in order to establish estimates of frequency of rejection of the hypotheses of homogeneity by means of the z, or index of dispersion, test. Results of these artificial realizations are presented in Figs. 1–6, and compared with the corresponding non-central ξ2-distributions.

It is a pleasure to mention the technical assistance of Dennis Hamilton of the Research Computer Laboratory in programming the detailed computational phases of this investigation, which was partially supported by the State of Washington Initiative 171 Funds for Research in Biology and Medicine.

Highly purified crystalline diphtheria toxin-protein, and toxoid prepared from it, were labelled with 131I, without change of potency, and injected intradermally into immune and allergic guinea-pigs. It is probable that actively immunized guinea-pigs localize these antigens and that localization is due to hypersensitivity of a type that cannot be transferred with serum; it is possibly of the delayed (tuberculin-type) of allergy.

It is a pleasure to acknowledge the gift of highly purified (crystalline) diphtheria toxin, and toxoid prepared from it, from Dr C. G. Pope of the Wellcome Research Laboratories, England. Dr S. P. Masouredis of the Central Blood Bank of Pittsburgh kindly labelled these antigens, using the technique he devised. Dr W. J. Kuhns generously provided the human precipitating and non-precipitating diphtheria antitoxin*.

The collimator head was made by Mr J. Nechaj of the University of Pittsburgh.

All these collaborators gave me advice on matters of which I was ignorant; for this I am grateful.

On the basis of starch synthesis and degradation by the strains three subdivisions of Group A streptococci were made. In defined cultural conditions about a third of the strains accumulated starch, the majority belonging to one subgroup. Of patients from whom starch-accumulating strains were isolated at the onset of infection, the observed numbers who developed either acute nephritis or acute rheumatic fever were more than three times the expected numbers. Of eighty-six strains, including forty-six members of Type 12, and forty belonging to several other serotypes, which were associated with either nephritis or rheumatic fever, more than 90% were starch-accumulating strains. Though 47% of all Type 12 strains accumulated starch, there was a primary association of starch-accumulating strains and the sequelae, irrespective of serotype. On the basis of type-frequency, and the starch synthesizing and degrading activities of the three most common serotypes, Types 1, 4 and 12, together with the host factor (for only 25% of patients infected with starch-accumulating strains developed the sequelae) the chances of Type 12 incidents are four to one against Types 1 and 4. The probability of the findings occurring by chance was considerably less than 0·001.

Technical assistance was paid for by grants from the Royal Free Hospital Endowment Fund, and I thank the Board of Governors of the Hospital. The drawings were made by Mrs Audrey Besterman, medical artist. For all their help and patience I wish to thank Dr R. E. O. Williams, Dr M. Patricia Jevons, Mr W. R. Maxted, Dr P. J. L. Sequeira, Dr W. J. Whelan and Dr Rebecca Lancefield.