The mind of primitive man

by Franz Boas

A number of years ago I had an opportunity to investigate a considerable number of Indian half-bloods; that is to say, of descendants of Indian mothers and white fathers. The most characteristic difference between the American Indian race and the European race, so far as these differences can be expressed in metrical form, is found in the width of the face. An extensive series of measurements of width of face made among half-bloods showed conclusively that the width of face does not tend to range around a certain intermediate value located between the width of face of the white race and that of the Indian race, but there was a decided tendency in the children to resemble either the Indian race or the white race; in other words, that feature of Mendelian inheritance which brings about the occurrence of mixed characteristics in the first hybrid generation was not found, but instead of this a decided tendency of reversion to either type, and to comparative rarity of intermediate forms. The results seem also to indicate that the Indian form in this mixture seems to dominate over the white form, but not in the Mendelian sense, which would require the presence of dominant features in a certain definite number of individuals, but only in the sense that the Indian type was a little more frequent than the European type, with the effect that the average width of face of the whole series was a little nearer to the Indian group than to the white group.

While this single observation is not by any means sufficient to determine fully the characteristic traits of heredity which govern the phenomenon in question, they indicate decidedly and beyond cavil that, in this case at least, we find what has been called by Karl Pearson “alternating inheritance.” It is worth remarking that not all the features of the body of the half-blood Indian exhibit the same tendency; that, for instance, in the case of stature, a general increase in the stature of the mixed people over that of the pure races may be observed.

Attention has been called by Felix von Luschan to a similar phenomenon which occurs in the mixed population of southern Asia Minor, where he believes to have found an alternating inheritance of the head-form, particularly of the proportions between width and length of head; some of the people retaining the short, high head-forms of the Armenoid type of the interior of Asia Minor, while the others have the long, low head of the Semites of Syria.

For a clear understanding of the laws of heredity, it seems important to know whether a similar alternating inheritance occurs in marriages of members of the same type. I have been enabled to investigate this question by a study of the East European Hebrews living in New York. A simple consideration shows, that, if the children tend to follow a type intermediate between the type of their parents, then the children of one family will show the same degree of resemblance among themselves, no matter how great the difference between the parents; for, if they simply tend to reproduce a middle type, it would not make any difference whether the mother is excessively short and the father excessively tall, or whether both parents are of middle stature. In both of these cases the intermediate value would be the same, and we should therefore expect that the effect upon the children would be the same. If, on the other hand, there is any kind of alternation in inheritance, the effect upon the family would be quite different. We should expect, in a family of which both parents are near the typical average, to find the children also to be near this average. On the other hand, if the mother is excessively short and the father excessively tall, we should expect some of the children to follow the mother in regard to shortness of stature, others to follow the father in regard to tallness. It will therefore be seen that in the case of alternating inheritance, we must expect an increased variability among the children. The compilation of material obtained from several thousand families shows very definitely that the variability among children both of whose parents belong to the same racial type, even to the same local type, increases quite considerably with the increased difference of the parents; so that we may assume a decided tendency to alternating inheritance in these cases. There is, however, no evidence whatever of the dominance of one type over another.

Quite a number of investigations have been made in relation to the intensity of heredity of parents and of grandparents; and, notwithstanding the uncertainty of the quantitative result, it seems reasonably certain that the intensity of heredity for each parent may be expressed by the value of about one-third (Pearson, Boas). It is somewhat difficult to explain clearly the significance of this value. I may, however, briefly indicate it in the following manner. Provided the mother differs in her stature by an amount of 9 cm. from the racial norm,—for instance, if she is 9 cm. taller than the average individual,—then we may expect the child to be one-third of 9 cm., or 3 cm., above the average. It will thus be seen that if both parents differ in the same direction from the average, the effect of both will be cumulative; and if both differ from the average of their people by the same amount, the joint effect of the two parents may be expressed by the coefficient of about two-thirds. In case, for instance, both father and mother should be 9 cm. above the type average, we should expect the child to be about two-thirds of 9 cm., or 6 cm., above the average.

Although definite information on the amount of heredity of previous generations is not yet available, the probability seems to be that the grandparents have jointly an influence of about two-ninths, the great-grandparents jointly of about two twenty-sevenths, etc., upon the offspring.

When we study these problems according to statistical theories, and take into consideration the observations on the resemblance of brothers and sisters, it can be shown that the theory of alternating inheritance cannot be taken too literally; for, if there were an absolute reversion of any one trait to pure ancestral types, we might say that the probability would be very small that two brothers should happen to repeat the bodily form of the same ancestor, because the number of ancestors in remote generations is very large. In other words, there must be an additional cause of resemblance between brothers and sisters. It is possible to show, that in case the inheritance has the strength denoted before, and if bodily form of a certain generation were due only to alternating inheritance acting from parents, grandparents, great-grandparents, and so on, and directly upon the generation in question, and without an occurrence of the same individuals in various places in the line of ancestors, then the resemblance between brothers and sisters, or, as we say, between the members of a fraternity, would result in a degree of resemblance which is much lower than the one actually observed. When the total number of ancestors is small, the recurrence of the same forms would become more probable, and the similarity of the series would increase. On the whole, the data seem to be best explained if we assume that there is not only alternating inheritance, but also a direct dependence upon the combination of the two parental types.

I should like to repeat here that these results have not been obtained with absolute certainty, and that it seems improbable that the laws of heredity in regard to various ancestral traits are the same. I do not enter into a discussion of the question of in how far these traits follow the laws of Mendelian inheritance,—a question that cannot be answered definitely at the present time (Davenport).

These problems have a fundamental importance for a clearer interpretation of the conditions which prevail in the form of local types of man.

In a large population which is as little stable in its habits as that of modern Europe and modern America, the number of ancestors of a single person increases very rapidly, the number of parents being two; of grandparents, four; of great-grandparents, eight; the theoretical number of ancestors twenty generations back would be over a million, or, more accurately, 1,048,576. Twenty generations represent, according to the rate of increase of modern times, about seven hundred years; according to the rate of increase of older times, about four hundred years as a minimum. These figures would apply to the series of generations represented by first-born males; for first-born females the respective numbers would be about five hundred years and three hundred and fifty years. If we consider, however, the actual descent of families, including individuals later born, we might perhaps assume that twenty generations in Europe would represent from eight hundred to nine hundred years, and among primitive peoples perhaps only little less, since in former times the differences between the rapidity of successive generations in Europe and among primitive peoples was not very great. This makes it obvious that it is entirely impossible that as great a number of ancestors as the theory requires can have contributed to the development of the individuals of the present generation. The reason for this is plain. Owing to intermarriages between the same families, large numbers of ancestors will be duplicated in different paternal and maternal lines; and in this way the real ancestry of each individual appears to be much more complex than the purely arithmetical treatment would suggest. The calculation for the ancestor table of the German Emperor, for instance, is instructive. According to O. Lorenz, the numbers of his ancestors in successive generations were as follows:—