in June.)
    In 1892, 35. 
    In 1893, 40. 
    In 1894, 38. 
    In 1895, 36. 
    In 1896, 36. 
    In 1897, 35. 
    Average, 37. (Omitting 1886, 1889, and 1891.)

Thus I have complete records for eight years, and incomplete records for three more; and the remarkable concord between the respective annual numbers of observations in these eight years not only affords us intrinsic evidence of the accuracy of my records, but, also, at once proves that there is an undeniable regularity in the occurrence of these sexual discharges, and, therefore, gives us reason for expecting to find this regularity rhythmical.  Moreover, since it seemed reasonable to expect that there might be more than one rhythm, I have examined my data with a view to discovering (1) an annual, (2) a lunar-monthly, and (3) a weekly rhythm, and I now proceed to show that all three such rhythms exist.

THE ANNUAL RHYTHM.

It is obvious that, in searching for an annual rhythm, we must ignore the records of the three incomplete years; but those of the remaining eight are graphically depicted upon Chart 8.  The curves speak so plainly for themselves that any comment were almost superfluous, and the concord between the various curves, although, of course, not perfect, is far greater than the scantiness of the data would have justified us in expecting.  The curves all agree in pointing to the existence of three well-defined maxima,—­viz., in March, June, and September,—­these being, therefore, the months in which the sexual instinct is most active; and the later curves show that there is also often a fourth maximum in January.  In the earlier years the March and June maxima are more strikingly marked than the September one; but the uppermost curve shows that on the average of all eight years the September maximum is the highest, the June and January maxima occupying the second place, and the March maximum being the least strongly marked of all.

Now, remembering that, in calculating the curves of the annual rhythm of the pulse, I had found it necessary to average two months’ records together, in order to bring out the full significance of the rhythm, I thought it well to try the effect upon these curves also of similarly averaging two months together.  At first my results were fairly satisfactory; but, as my data increased year by year, I found that these curves were contradicting one another, and therefore concluded that I had selected unnatural periods for my averaging.  My first attempted remedy was to arrange the months in the pairs December-January, February-March, etc., instead of in January-February, March-April, etc.; but with these pairs I fared no better than with the former.  I then arranged the months in the triplets, January-February-March, etc.; and the results are graphically recorded on Chart 7.  Here, again, comment would be quite futile, but I need only point out that, on the whole, the sexual activity rises steadily during the first nine months in the year to its maximum in September, and then sinks rapidly and abruptly during the next three to its minimum in December.