## A Contribution to The Aetiology of Manic-Depressive Insanity (1940)

*Yearly Rhythm of Incidence*

The problem of seasonal incidence may be investigated in yet another way. According to Lange^{1} certain number of manicdepressives fall ill again and again at the same time of the year. He is not, on the other hand, of the opinion that any such rule holds good for manic‑depressives in general. Nevertheless I have attempted to discover by statistical means whether the liability to illness for the single manic‑depressive is not concentrated on a particular time of the year. That is to say, whether some do not fall ill mainly in the summer, others in spring, autunui, or winter. I proceeded in the following way.

Let us assume that a manic‑depressive has two illnesses. The onset of both illnesses can occur in the same month, at the same time of the year, or at a quite different time of the year. The two months of onset can show a difference of 0, 1, 2 ‑ 6 months. January is separated from July by six months, by five months from June and August, by one month from December and February. I will take as an example Case 159, who had five attacks, to wit: one in April and two each in August and October. If one calculates the difference in time of each illness from every other, twenty such differences are obtained ( = 5 x 4, i.e. n(n‑1), where n = the number of illnesses). These differences are:

4 of 0 months

8 of 2 "

4 of 4 "

4 of 6 "

Proceeding in this way with all the cases, the following table is obtained.

The probability that two events occur in the same month or in two months separated from each other by a distance of six months is obviously only half as great as the probability that they occur in two months separated from each other by a period of any value from one to five months. January, for instance, has only one month, January, separated from it by a period of 0 months; and only one, July, separated from it by a period of six months. On the other hand, it has two months, December aud February, separated from it by a period of one month; two months, Novenber and March, separated from it by a period of two months, and so on. Accordingly, to obtain corrected values, one has to divide the figures in Table 4 for the one‑ to five‑month periods by two. In this way the following table (Table 5) is obtained. These figures are represented graphically in Figure 4.

The figures and the curve show a clear fall in frequency from the 0‑ to the 6‑months values, from which one can conclude that the individual manic‑depressive does show a tendency to have recurrences of illness at the same time of year. The statistical significance of the figures can be tested by calculation of the X^{2} value, which works out in this case to be 27,810. For the calculation it is necessary to use the empirical figures of Table 4, all divided by two. For a 1% probability and six degrees of freedom, the corresponding value of X^{2} is 16,812. As the calculated value of X^{2} is very considerably larger than this, the statistical significance of the finding may be taken as assured.

(1) Lange. Bumkes Handbuch der Geisteskrankheiten, Vol. 6, p. 181. Berlin, 1928.