The cells which compose the ganglion (fig. I I, _sp.g_) are easily distinguished from those of the root. Each cell is elongated with an oval nucleus, large in proportion to the cell; and its protoplasm appears to be continued into an angular, not to say fibrous process, sometimes at one and more rarely at both ends. The processes of the cells are at this stage very difficult to observe: figs. I a, I b, I c represent three cells provided with them and placed in the positions they occupied in the ganglion. The relatively very small amount of protoplasm in comparison to the nucleus is fairly represented in these figures, though not in the drawing of the ganglion as a whole. In the centre of each nucleus is a nucleolus. Fig. I b, in which the process points towards the root of the nerve, I regard as a commencing nerve-fibre: its more elongated shape seems to imply this. In the next stage special bundles of nerve-fibres become very conspicuous in the ganglion. The long diameter of an average ganglion-cell is about 1/1600 of an inch. The whole ganglion forms an oval mass, well separated both from the nerve-root and the nerve, and is not markedly continuous with either. On its outer side lies the downward process of the nerve-root before mentioned. The nerve itself is still, as in the last case, composed of cells which are larger and more elongated than either the cells of the root or the ganglion. The condition of the anterior root at this stage is hardly altered from what it was; it is composed of very small cells, which with hæmatoxylin stain more deeply than any other cell of the section. A figure of it is given in I II. Horizontal longitudinal sections of this stage are both easy to make and very instructive. On Pl. 23, fig. K I is represented a horizontal section through a plane near the dorsal surface of the spinal cord: each posterior root is seen in this section to lie nearly opposite the anterior extremity of a muscle-plate. In a more ventral plane (fig. K II) this relation is altered, and the posterior roots lie opposite the hinder parts of the muscle-plates. The nerves themselves are invested by the hyaline membrane spoken of above; and surrounding this again there is present a delicate mesoblastic investment of spindle-shaped cells. Longitudinal sections also throw light upon the constitution of the anterior nerve roots (vide fig. K II, _ar_). In the two segments on the left-hand side in this figure the anterior roots are cut through as they are proceeding, in a more or less horizontal course, from the spinal cord to the muscle-plates. Where the section (which is not quite horizontal) passes through the plane of the notochord, as on the right-hand side, the anterior roots are cut transversely. Each root, in fact, changes its direction, and takes a downward course.