1. Whooping cough is now in London the most important epidemic disease of childhood in causing physical suffering, loss of school attendance, impairment of physique and mortality.

2. The mean annual incidence of whooping cough in children under 10 in Battersea during the quinquennium 1925–9 is estimated to have been about 60 per 1000 living, and the fatality rate 1·26 per 100 cases. The fatality rates in the first two years of life were estimated to be 4·0 and 3·3 respectively, at ages 2 to 5, 0·7 and at ages 5 to 10, 0·2 per 100 cases.

3. Epidemics of whooping cough in Battersea and Greenwich tend to have a biennial periodicity; a summer epidemic is often immediately followed by or merges into a winter epidemic subsiding by midsummer, and then follows a year of low incidence, so that the total incidence of cases in successive years from midsummer to midsummer shows an alternation of epidemic and interepidemic years.

4. Epidemics do not always run concurrently in areas of London separated by several miles; the spread is checked to some extent by barriers to free communication and small districts thus occasionally escape an epidemic, or an interval of several months may separate the epidemic peaks in districts a mile or so apart.

5. The view that the epidemic cycle in whooping cough can be explained by a periodic reduction of the proportion of children not previously attacked below some critical limit, is shown to be untenable. Both the termination of epidemics and their periodicity can be explained by the occurrence of a latent immunisation of contacts as in measles, such immunity only lasting about 1 year instead of 2 or 3.

6. The occurrence of such a transient latent immunisation is indicated by (a) measurement of the rise and fall of apparent infectiousness of cases to other children in the same house during 6 years in Battersea, (b) measurement of the frequency of second cases in houses after different intervals of time during a decade in Greenwich, and (c) analysis of the attack rates amongst home contacts in Holborn and Battersea during several years following exposure to infection as compared with those among control groups of children. These phenomena would be difficult to explain as resulting from a “cycle of activity” of the organism.

7. In London as a whole a special enquiry into the previous histories of school entrants proves that 44 per cent. of children have an attack of whooping cough before their fifth birthday. According to Battersea records 27·7 per cent. are notified or reported to the Public Health authority as having whooping cough by this age, and 42·3 per cent. by the tenth birthday. It is concluded that in that borough about 60 per cent. of the true incidence is notified at ages under 4, about 72 per cent. at age 4, and 90 per cent. at ages 5 to 10 when school supervision is fairly complete. At all ages under 10 the recorded incidence in Battersea represents about 70 per cent. of the true incidence.

8. It follows that about 60 per cent. of London children have whooping cough by the tenth year, and about 61 per cent. at some time during life, this proportion being rather lower than estimates made from histories given by parents in a group of American towns. The bulk of the remaining 40 per cent. must escape attack by virtue of some kind of immunity rather than by avoidance of contact with infection through life, and a part of these no doubt escape by the repeated acquisition of transient latent immunity during epidemics, and the remainder by virtue of an inherent immunity to the disease.

9. During the year following an epidemic the proportion of children under 10 who are susceptible is approximately doubled and again halved during the next epidemic. This must be mainly due to temporary latent immunisation during an epidemic of x children to each one attacked, the value of x lying between the limits 1 and 4, this depending upon the unknown proportion who are inherently immune. If the bulk of the two-fifths of all children who escape attack have inherent immunity, as is suggested but not proved by the data, then the ratio of latent to recognisable infections lies between 1 and 2.

10. Superposed upon the gradual rise and fall of infectiousness believed to be due to these immunity changes, there occurs a sudden rise at the commencement of an epidemic which must be attributed to an enhanced real infectiousness. This probably arises from a sudden increase in activity of the organism itself under seasonal influences setting in at the moment when the herd immunity becomes insufficient to withstand the infection pressure, so that the balance is upset and the epidemic begins.

11. These findings are similar to those for measles, the chief differences being, (i) epidemics are usually spread over a year instead of 6 months; (ii) latent immunity apparently lasts about a year instead of 2 or 3; (iii) about 60 per cent. of children are attacked at some time during life instead of about 90 per cent.; (iv) there is probably inherent immunity in a certain proportion, but not in measles; (v) the ratio of latent to manifest infections is lower.

We are indebted to Miss M. Kirby for drawing the diagrams in this paper. Towards the expenses of this research a grant was made by the British Medical Association.