Before discussing what may follow from Mendel’s discovery for the theory of heredity, we must lay stress on the fact that there are many exceptions to his rule. In quite a number of cases the hybrids are of one or more types, which remain constant: there is no splitting at all in the second generation. But that does not affect the rule of Mendel in those cases where it is true. Where there is a “splitting” in the second generation, there also are the numerical proportions stated by Mendel; there never are other relations among the numbers of individuals of the mixed and of the two pure types than those given by his rule. I regard it as very important that this real meaning of Mendel’s principle should be most clearly understood. From the fact of the splitting of hybrids in the second generation most important consequences may be drawn for the theory of inheritance; the split individuals, if crossed with each other, always give an offspring which remains pure; there is no further splitting and no other change whatever. The germ-cells produced by the split individuals of the second generation may therefore be said to be “pure,” as pure as were those of the grandparents. But that is as much as to say that the pureness of the germ-cells has been preserved in spite of their passing through the “impure” generation of the hybrids, and from this fact it follows again that the union of characters in the hybrids must have been such as to permit pure separation: in fact, the germ-cells produced by Mendelian hybrids may hypothetically be regarded as being pure themselves.[131] [131] This hypothesis was first suggested by Sutton and is at present held by orthodox Mendelians; but probably things are a little more complicated in reality, as seems to be shown by some facts in the behaviour of so-called “extracted recessives.” In Morgan’s *Experimental Zoology*, New York, 1907, a full account of the whole matter is given.