traits) are still uncertain in their results. The characters involved are too complex to offer any simple sequences. If two parents have brown eyes, it can not be said that all their children will have brown eyes; still less can it be said that all the children of two musically gifted parents are certain to be endowed with musical talent in any given degree.
Prediction is possible only when uniform sequences are found. How are such sequences to be found in heredity, if they do not appear when a parent and his offspring are examined? Obviously it is necessary to examine a large number of parents and their offspring,—to treat the problem by statistical methods.
But, it may be objected, a uniformity gained by such methods is spurious. It is merely shutting the eyes to the mass of contradictions which are concealed by an apparent statistical uniformity.
This objection would be valid, if the statistical results were used for prediction in individual cases. The statistician, however, expressly warns that his conclusions must not be used for such prediction. They are intended to predict only general trends, only average results; and for this purpose they are wholly legitimate. Moreover, evolution itself is a problem of statistics, and therefore the statistical method of studying heredity may offer results of great value to eugenics, even though it can not furnish in individual cases the prediction which would be desirable.
From this standpoint, we return to attack the problem of the relation between parent and offspring. We noted that there is no uniform sequence in a single family, and illustrated this by the case of brown eyes. But if a thousand parents and their offspring be selected and some trait, such as eye-color, or stature, or general intelligence, be measured, a uniformity at once appears in the fact of regression. Its discoverer, Sir Francis Galton, gives this account of it: