Nelson, in his study of dreams and their relation to seasonal ecbolic manifestations, does not present any yearly ecbolic curve, as the two years and a half over which his observations extend scarcely supply a sufficient basis. On examining his figures, however, I find there is a certain amount of evidence of a yearly rhythm. There are spring and autumn climaxes throughout (in February and in November); there is no December rise. During one year there is a marked minimum from May to September, though it is but slightly traceable in the succeeding year. These figures are too uncertain to prove anything, but, as far as they go, they are in fair agreement with the much more extensive record, that of W.K. (_ante_ p. 113), which I have already made use of in discussing the question of a monthly rhythm. This record, covering nearly twelve years, shows a general tendency, when the year is divided into four periods (November-January, February-April, May-July, August-October) and the results summated, to rise steadily throughout, from the minimum in the winter period to the maximum in the autumn period. This steady upward progress is not seen in each year taken separately. In three years there is a fall in passing from the November-January to the February-April quarter (always followed by a rise in the subsequent quarter); in three cases there is a fall in passing from the second to the third quarter (again always followed by a rise in the following quarter), and in two successive years there is a fall in passing from the third to the fourth quarter. If, however, beginning at the second year, we summate the results for each year with those for all previous years, a steady rise from season to season is seen throughout. If we analyze the data according to the months of the year, still more precise and interesting results (as shown in the curve, Chart 3) are obtained; two maximum points are seen, one in spring (March), one in autumn (October, or, rather, August-October), and each of these maximum points is followed by; a steep and sudden descent to the minimum points in April and in December. If we compare this result with Perry-Coste's also extending over a long series of years, we find a marked similarity. In both alike there are spring and autumn maxima, in both the autumn maximum is the highest, and in both also there is an intervening fall. In both cases, again, the maxima are followed by steep descents, but while in both the spring maximum occurs in March, in Perry-Coste's case the second maximum, though of precisely similar shape, occurs earlier, in June-September instead of August-October. In Perry-Coste's case, also, there is an apparently abnormal tendency, only shown in the more recent years of the record, to an additional maximum in January. The records certainly show far more points of agreement than of discrepancy, and by their harmony, as well with each other as with themselves, when the years