2 ------------ / \ 5 6 ------- ------- / \ / \ 11 12 13 14 --- --- --- --- / \ / \ / \ / \ 23 24 25 26 27 28 29 30 In the three diagrams the nuclei with the same numbers have the same rank in descent, and therefore, according to the theory of Roux and Weismann, have the same qualities, while the nuclei with unlike numbers differ in qualities. Let us now notice how the nuclei in the three processes of division, of which two are abnormal, are placed in the mass of the egg. After the first division, the nuclei are alike in all three cases; after the second difference appears. In A1 and B1 nuclei 3 and 5 lie to the left; 4 and 6 to the right of the second cleavage-plane, which, according to Roux's hypothesis, corresponds to the median-plane of the future embryo; while in C they are forced into two layers, one above the other, nuclei 4 and 6 being dorsal, 3 and 5 ventral. In the third cycle of division there is no agreement between the three cases. In the diagrams A2 and B2 the nuclei still lie similarly to the right and left of the middle line; but in A2 they are arranged in two layers, in B2 in a single layer. The nuclei 8, 10, 12, and 14, which compose the upper layer in A2, form the middle of the disc in B2; and 7 and 9, 11 and 13, the ventral nuclei of A2, occupy the ends of the single-layered disc of B2, being closely pressed against each other. In the diagram C2 there is actually no median-plane after the third cycle of division. The nuclei 9, 10, 14, 13, which in A and B form the right side of the mass, here form a dorsal layer with nuclei 7, 8, 12, 11, forming a ventral layer. In the fourth cycle of division the nuclear matter is still more variously distributed through the mass, as may be seen from comparison of diagrams A3, B3, C3. Although, under normal conditions, the multiplication and division of the nuclear material occurs in an almost invariable and definite fashion, the mere altering of the spherical form to a cylinder or to a disc produces a method of division completely different, so far as the nuclei are related to each other in a genealogical tree. In the one and the other method of division the nuclei are brought into relation with different regions of the protoplasmic mass, and are united with these regions to form cellular individuals. I had quite enough reason for what I said in my essay: 'If the doctrine of Roux and Weismann be true, and the successive divisions by which nuclei arise really place different qualities in the nuclei--qualities according to which the masses of protoplasm surrounding them become different and definite parts of the embryo--what a pretty set of malformations must result from eggs in which the nuclear matter has been shuffled about so wantonly! As such malformations do not occur, it is plain that the doctrine is untenable.'