ordinary variation is the fact that the descendants of hybrids are more apt to revert than the hybrids themselves. Darwin says (Variation, p. 65) that this is a general rule.

Now, whether reversion be due to the sudden excitement of a tendency which has long been transmitted in a dormant state by the ova, or whether it is due to the appearance of a new variation which resembles an old one, we can readily understand how, according to our theory of heredity, crossing should call this power into action. During the evolution of the species each hereditary peculiarity has been established in the egg by gemmules, and anything which prevents the egg from following its normal course and developing the recently acquired characteristics of the species, would allow older characteristics to appear in their place.

We know that animals which are very widely separated are infertile, and we can understand that even when the difference between two species is not great enough to prevent them from crossing, those cells of their bodies which have varied most may be so different from each other that gemmules from the one cannot fertilize the egg-particles which are to produce the other, or when they do fertilize them they may give rise to a variation which is so different from the normal cell that it cannot live. The cells which precede these in the order of growth being less different in the two parents, would be much more favorably situated, and would thus give to the embryo a characteristic of longer standing than the peculiarities of either parent. On the other hand, if reversion is simply variation, we can see that crossing might excite reversion just as it excites variability.

Summary of Chapter.

The study of hybrids gives us a means of comparing, within certain narrow limits, the parts which the two sexual elements play in heredity. The influence of each sex can, in a certain sense, be studied by itself when a given species is used in the one case as the father of a hybrid, and in another case as the mother. The value of crossing as an experiment in heredity is greatly limited, however, by the fact that, although we can study the influence of one sexual element unobscured by the other element from the same species, it is obscured and complicated by the influence of this element from an allied species, and in all organisms which can breed together the reproductive elements must be essentially alike.

Hybrids do, however, present a number of peculiarities which agree perfectly with what we should expect according to our hypothesis, and certain of these are inexplicable without it.

Hybrids and mongrels are highly variable, as we should expect to be the case, according to Darwin's pangenesis hypothesis. This hypothesis fails to account for the fact that hybrids from forms which have long been domesticated are more variable, than those from wild species or varieties, or for the very remarkable fact that the children of hybrids are much more variable than the hybrids themselves.

Our theory not only explains the variability of hybrids, but it also accounts for the two latter peculiarities, for crossing will not give rise to a marked or conspicuous variation unless the hybrid inherits numbers of gemmules, and as domesticated animals and plants live under unnatural conditions they are more favorably placed than wild forms for the production of gemmules.

The body of a hybrid is in itself a new thing, and therefore in a certain sense unnatural, and a male hybrid is, accordingly, more fitted for the production of gemmules than a male of a pure or unmixed race.

When a male hybrid is crossed with the female of either pure species or with a third species, the children are much more variable than those born from a hybrid mother by a male of a pure species. It would be difficult to devise an experiment better fitted than this to show that variation is caused by the influence of the male, and that the action of unnatural or changed conditions upon the male parent results in the variability of the child.

The remarkable history of reciprocal hybrids is directly opposed to Darwin's view that the functions of the two reproductive elements are essentially similar, for in some cases it is impossible to breed from a female of one species by the male of a second species, while the male of the first species readily fertilizes the ovum of the second and gives rise to fertile offspring. Even when both crosses are fertile the one is often much more so than the other.

The hybrids of one cross often differ remarkably from those of the other cross in general structure, and in many cases they show, in addition to the common characteristics of both parents, a tendency, more or less perfectly pronounced, to develop the recently acquired characteristics of that species which is used as the fa

ther.

This law is often obscured by the appearance of reversions, which are peculiarly apt to occur in hybrids, and by the presence, in certain cases, of a tendency for each parent to transmit its peculiarities to the hybrid, without fusion with those of the other parent. But when we

consider the great obscurity and complexity of the case, and the great difficulty in conducting rigid experiments, the balance of the evidence from hybrids seems to be greatly in favor of our view of the nature of heredity. It certainly presents features which are inexplicable in any other way, and perfectly simple and natural if our view is accepted.

CHAPTER VII.

THE EVIDENCE FROM VARIATION.

Causes of variation-Changed conditions of life induce variability-No particular kind of change is necessary-Variability is almost exclusively confined to organisms produced from fertilized ova-Bud variation very rare-History of the Italian orange-The frequency of variation in organisms produced from sexual union, as compared with its infre quency in those produced asexually, receives a direct explanation by our theory of heredity-Bud variation more frequent in cultivated than in wild plants-Our theory would lead us to expect this-Changed conditions do not act directly, but they cause subsequent generations to vary -Tendency to vary is hereditary-These facts perfectly explicable by our theory-Specific characters more variable than generic-Species of large genera more variable than those of small genera-A part developed in an unusual way highly variable-Law of equable variation-Secondary sexual characters variable-Natural selection cannot act to produce permanent modification unless many individuals vary together—Our theory is the only explanation of the simultaneous variation of many individuals-This theory also simplifies the evolution of complex structures-Saltatory evolution-This is explained by our theory of heredity -Correlated variation of homologous parts-Parts confined to males more variable than parts confined to females— Males more variable than females-Summary of last two chapters.

The Causes of Variation.

CERTAIN authors have held that variability is a necessary accompaniment of reproduction; that it is determined by something within rather than without the or