hose years
only which are characterized by a Sunday maximum; and, accordingly, in
Chart 12 I have depicted the curves in a more suitable form.

Chart 12 _A_ is obtained by combining the data of 1888, 1892, and 1896:
the years of a Sunday maximum. Curve 12 _B_ represents the results of
1894, the year of a Tuesday maximum--multiplied throughout by three in
order to render the curve strictly comparable with the former. Curve 12
_C_ represents 1886 and 1897--the years of a Thursday maximum--similarly
multiplied by 1.5. In Curve 12 _D_ we have the results of 1887--the year
of a Friday maximum--again multiplied by three; and in Curve 12 _E_ those
of 1893 and 1895--the years of a Saturday maximum--multiplied by 1.5.
Finally, Curve 12 _F_ represents the combined results of all nine years
plus (the latter half of) 1891; and this curve shows that, on the whole
period, there is a very strongly marked Sunday maximum.

I hardly think that these curves call for much comment. In their general
character they display a notable concord among themselves; and it is
significant that the most regular of the five curves are _A_ and _E_,
representing the combinations of three years and of two years,
respectively, while the least regular is _B_, which is based upon the
records of one year only. In every case we find that the maximum which
opens the week is rapidly succeeded by a minimum, which is itself
succeeded by a secondary maximum,--usually very secondary, although in
1894 it nearly equals the primary maximum,--followed again by a second
minimum--usually nearly identical with the first minimum,--after which
there is a rapid rise to the original maximum. The study of these curves
fortunately amplifies the conclusion drawn from our study of the annual
rhythm, and suggests that, in at least part of the year, the physiological
condition of man requires sexual union at least twice a week.

As to Curve 12_F_, its remarkable symmetry speaks for itself. The
existence of two secondary maxima, however, has not the same significance
as had that of our secondary maximum in the preceding curves; for one of
these secondary maxima is due to the influence of the 1894 curve with its
primary Tuesday maximum, and the other to the similar influence of Curve
_C_ with its primary Thursday maximum. Similarly, the veiled third
secondary maximum is due to the influence of Curve _E_. Probably, any
student of curves will concede that, on a still larger average, the