but directly unattainable thing." In this sense alone do we maintain that there are three ideal racial types in Europe to be distinguished from one another. They have often unfortunately dissolved in the common population; each trait has gone its own way; so that at the present time rarely, if indeed ever, do we discover a single individual corresponding to our racial type in every detail. It exists for us nevertheless.
Thus convinced that the facts do not warrant us in expecting too much of our anthropological means of isolating racial types, we have recourse to a second or inferential mode of study. In this we work by geographical areas rather than by personalities. We discover, for example, that the north of Europe constitutes a veritable center of dispersion of long-headedness. Quite independently we discover that the same region contains more blond traits than any other part of Europe; and that a high average stature there prevails. The inference is at once natural that these three characteristics combine to mark the prevalent type of the population. If one journeyed through it, one might at first expect to find the majority of the people to be long-headed and tall blondes; that the tallest individuals would be the most blond, the longest-headed most tall, and so on. This is, as we have already shown, too good and simple to be true, or even to be expected. Racial combinations of traits indeed disappear in a given population, as sugar dissolves, or rather as certain chemical salts are resolved into their constituent elements when immersed in water. From the proportions of each element discovered in the fluid, quite free from association, we are often able to show that they once were united in the same compound. In the same manner, we, finding these traits floating about loose, so to speak, in the same population, proceed to reconstitute types from them. We know that the people approach this type more and more as we near the specific center of its culmination. The traits may refuse to go otherwise than two by two, like the animals in the ark, although they may change partners quite frequently; and they may still manifest distinct affinities one for another nevertheless.
The apparent inference is not always the just one, although it tends to be. Suppose, for example, that one observer should prove that sixty per cent of ten thousand natives of Holland were blondes: and another, studying the same ten thousand individuals, should prove that a like proportion were very tall—would this of necessity mean that the Hollanders were mainly tall blondes? Not at all! It might still be that the two groups of traits merely overlapped at their edges. In other words, the great majority of the blondes might still be constituted from the shorter half of the population. Only twenty per cent need necessarily be tall and blond at once, even in this simple case where both observers stud-
