Roux[13] has based his mosaic theory upon experiments upon frogs' eggs. According to the theory, the first two segmentation spheres contain not only all the formative material for the right and left halves of the embryo respectively, but also the differentiating and elaborating forces for these, so that on the destruction of one cell, the other can give rise only to one lateral half of the embryo (hemiembryo lateralis). Roux, therefore, considers that by the first cleavage the nuclear material is broken up into unlike halves, by which the development of the corresponding cells is directed diversely, _i.e._, is determined in a specific fashion. [Illustration: FIG. 3.--DIAGRAMS OF THE EGGS OF FROGS, which show how alteration of the cleavage process changes the mode in which the nuclear material is distributed. The nuclei indicated by the same numbers have the same descent in all the diagrams. All the eggs are viewed from the animal pole. A. Normally developing eggs. B. Eggs developing under compression by horizontal plates. C. Eggs developing under compression by vertical plates.] The error in these representations of Weismann and of Roux has been shown by varied experiments of my own. The eggs of frogs on the point of cleaving were flattened to a disc between vertically or horizontally placed glass-plates. In the first case they were flattened in the dorsoventral direction, _i.e._, the axis passing through the animal and vegetative pole was shortened; in the second case an axis at right angles to this was shortened. In both cases the course of cleavage, and the resulting distribution of the nuclei in the yolk, was artificially modified. The diagrams A, B, C (Fig. 3) will make the results plain to the reader. A, represents the distribution of the nuclei after normal cleavage; B, the same, when the egg was pressed between horizontally-arranged parallel glass-plates; C, the same, where the flattening was produced by vertically-placed parallel glass-plates.[14] The diagrams show the positions of the segmentation spheres and of the contained nuclei as seen from the animal pole. In stages where two layers of cells as a result of division lay one above the other, the cells of the lower layer are distinguished in the figure by shading. In the three diagrams the nuclei are numbered so that the reader may know how far they are removed from the nuclei of the first two segmentation spheres. The numbers are further exhibited in the following two genealogical trees: 1 ------------ / \ 3 4 ------ ------ / \ / \ 7 8 9 10 --- --- --- --- / \ / \ / \ / \ 15 16 17 18 19 20 21 22