sensibly with each other. The attempt to treat or study them as similar is leading to utter confusion in the study of evolution.[1]
If normality thus imagined can be shown to be a real phenomenon it is conceivable that we might then profitably attempt to determine in specified cases the average value of homotypic correlation for each case, but the average value for a miscellaneous collection of cases would still have no natural significance.
(Note, added November., 1901.)
On p. 287 Professor Pearson has added a note in which he seeks to meet a part of my objections. He says : "A diversity due to differentiation and a variability due to chance are quite distinct things. The one is the result of dominating factors which can be isolated and described ; the other of a great number of small factors, varying from organ to organ, and incapable of being defined or specified. Indeed, upon each dominating factor of differentiation is superposed such a chance variability. Of course all things which differ even by chance variation are in a certain sense differentiated." This welcome passage outlines the conception that must form the point of departure in any attempt to understand variation in its relation to Evolution. The same conception I have myself often laboured to express. On former occasions to these two kinds of diversity of which Professor Pearson speaks I have applied the terms "Discontinuous" and "Continuous." Though useful in practice, those terms are open to misconstruction and perversion. In the present paper I have suggested the nearly equivalent terms " Specific " and " Normal." Similarly, to variations occurring among repeated parts or homotypes we might apply the terms " Differentiant" and "Normal." Throughout nature the variations between the members of fraternities may be discontinuous and specific, and in like manner may the variations between repeated parts be specific and differentiant, though in both classes normal or continuous variations are always superposed on them.
In most cases the naturalist is seldom in much doubt with which he is dealing. But though these two great classes of variation can broadly be recognised and treated as distinct, the distinction may be evasive, and when the differentiation is irregular that distinction must often be obscured and not " statistically discoverable." Professor Pearson is mistaken in supposing that such differentiation must show itself in his seriations. It may appear only as a lowering
- ↑ For example, in his criticisms (p. 360 and elsewhere) of the view that sexually produced off spring are more variable than offspring not sexually produced, Professor Pearson is merely confusing different kinds of variations and applying to certain kinds conclusions derived from a study of another kind.
