s for a number of years to be greater among boys than among
girls, until in the period of adolescence the death-rates of the two
sexes are equal. In adult life the death-rate among men is nearly always
higher than that among women, but this is due largely to the fact that
men pursue occupations where they are more exposed to death. In such
cases, and particularly where deaths are due to accident, the mortality
may not only be non-selective, but is sometimes contra-selective, for
the strongest and most active men will often be those who expose
themselves most to some danger. Such a reversal of the action of natural
selection is seen on a large scale in the case of war, where the
strongest go to the fray and are killed, while the weaklings stay at
home to perpetuate _their_ type of the race.

A curious aspect of the kind of natural selection under
consideration,--that which operates by death without reference to the
food-supply,--is seen in the evolution of a wide pelvis in women. Before
the days of modern obstetrics, the woman born with an unusually narrow
pelvis was likely to die during parturition, and the inheritance of a
narrower type of pelvis was thus stopped. With the introduction and
improvement of instrumental and induced deliveries, many of these women
are enabled to survive, with the necessary consequence that their
daughters will in many cases have a similarly narrow pelvis, and
experience similar difficulty in childbirth. The percentage of
deliveries in which instrumental aid is necessary is thus increasing
from generation to generation, and is likely to continue to increase
for some time. In other words, natural selection, because of man's
interference, can no longer maintain the width of woman's pelvis, as it
formerly did, and a certain amount of reversion in this respect is
probably taking place--a reversion which, if unchecked, would
necessarily lead after a long time to a reduction in the average size of
skull of that part of the human race which frequently uses forceps at
childbirth. The time would be long because the forceps permit the
survival of some large-headed infants who otherwise would die.

But it must not be supposed that lethal, non-sustentative selection
works only through forms of infant mortality. That aspect was first
discussed because it is most obvious, but the relation of natural
selection to microbic disease is equally widespread and far more
striking.

As to the inheritance of di