He points out especially the peculiarity of layer (2) (2, Fig. 41), the inner molecular, in which two rows of nuclei are arranged with great regularity, the one row closely touching the _membrana limitans interna_, the other at the inner boundary of the middle third of the {96}molecular layer. Of these two rows of nuclei, he describes the innermost as composed almost entirely of large nuclei belonging to ganglion cells, while the outermost is composed mainly of distinctly smaller nuclei, which in staining and appearance appear to belong not to nerve-cells but to the true reticular tissue of the molecular layer. He also draws special attention to the remarkable layer (5) (5, Fig. 41), which is not found in the retina of the higher vertebrates, the cells of which, in his opinion, are of the nature of ganglion-cells. W. Müller, in 1874, gave a most careful description of the eye of Ammocoetes and Petromyzon, and traced the development of the retina; the subsequent paper of Kohl does not add anything new, and his drawings are manifestly diagrams, and do not represent the appearances so accurately as Müller's illustrations. In the accompanying figure (Fig. 41) I reproduce on the right-hand side Müller's picture of the retina of Petromyzon, but have drawn it, as in Langerhans' picture, at the place of entry of the optic nerve. From his comparison of this retina with a large number of other vertebrate retinas, he comes to the conclusion that the retina of all vertebrates is divisible into _A._ An ectodermal (epithelial) part consisting of the layer of the visual cells, and _B._ A neurodermal (cerebral) part which forms the rest of the retina. Further, Müller points out that the neuroderm gives origin throughout the central nervous system to two totally different structures, on the one hand to the true nervous elements, on the other to a system of supporting cells and fibres which cannot be classed as connective tissue, for they do not arise from mesoblast, and are therefore called by him 'fulcrum-cells.' In the retina he recognizes two distinct groups of such supporting structures--(1) a system of radial fibres with well-marked elongated nuclei, which extend between the two limiting layers, and form at their outer ends a membrane-like expansion which was originally the outer limit of the retina, but becomes afterwards co-terminous with the _membrana limitans externa_, owing to the piercing through it of the external limbs of the rods. This system, which is known by the name of the radial Müllerian fibres (shown on the left-hand side of Fig. 41), has no connection with (2) the spongioblasts and neurospongium, which form a framework of neuroglia, in which the terminations of the {97}optic ganglion and of the retinal ganglion ramify to form the molecular layers.