ves of the same type; the resemblance would then
be perfect and the correlation, as we call it, would be expressed by a
coefficient of 1."

If all mating were at random, evolution would be a very slow process.
But actual measurement of various traits in conjugal pairs shows that
mating is very rarely random. There is a conscious or unconscious
selection for certain traits, and this selection involves other traits
because of the general correlation of traits in an individual. Random
mating, therefore, need not be taken into account by eugenists, who must
rather give their attention to one of the two forms of non-random
mating, namely, assortative and preferential.

2. If men who were above the average height always selected as brides
women who were equally above the average height and short men selected
similarly, the coefficient of correlation between height in husbands and
wives would be 1, and there would thus be perfect assortative mating. If
only one half of the men who differed from the average height always
married women who similarly differed and the other half married at
random, there would be assortative mating for height, but it would not
be perfect: the coefficient would only be half as great as in the first
case, or .5. If on the other hand (as is indeed the popular idea) a tall
man tended to marry a woman who was shorter than the average, the
coefficient of correlation would be less than 0; it would have some
negative value.

Actual measurement shows that a man who exceeds the average height by a
given amount will most frequently marry a woman who exceeds the average
by a little more than one-fourth as much as her husband does. There is
thus assortative mating for height, but it is far from perfect. The
actual coefficient given by Karl Pearson is .28. In this case, then, the
idea that "unlikes attract" is found to be the reverse of the truth.

If other traits are measured, assortative mating will again be found.
Whether it be eye color, hair color, general health, intelligence,
longevity, insanity, or congenital deafness, exact measurements show
that a man and his wife, though not related by blood, actually resemble
each other as much as do uncle and niece, or first cousins.

In some cases assortative mating is conscious, as when two congenitally
deaf persons are drawn together by their common affliction and mutual
possession of the sign language. But in the greater number of cases it
is wholly unconsc