The teaching of Development is on this point of supreme importance. Unhappily there has not yet been a sufficient collection of systematic observations to enable us to speak very confidently as to the successive stages, but some negative evidence there is. The changes take place with great rapidity, and the earliest stages have hardly been observed at all. Although for several successive years I watched the development of tadpoles, the difficulties were so great, and the appearances so perplexing, that the only benefit I derived was that of being able the better to understand the more successful investigations of others. Four or five days after fecundation is the earliest period of which I have any recorded observation; at this period the cerebral substance appeared as a finely granular matter, having numerous lines of segmentation marking it off into somewhat spherical and oval masses, interspersed with large granules and fat globules. Here and there hyaline substance appeared between the segments. Similar observations have since been recorded by Charles Robin in the earliest stages of the Triton.[168] He says that when the external gills presented their first indications, nuclei appeared, each surrounded by a rim of hyaline substance, from which a pale filament was prolonged at one end, sometimes one at both ends, and this filament subdivided as it grew in length until it had all the appearance of an axis cylinder. This, however, he says, is a striation, not a fibrillation; he refuses to admit that the axis cylinder is a bundle of fibrils. He further notices the simultaneous appearance of amorphous substance; and as this is several days before there is any trace of a _pia mater_, or proper connective tissue, he urges this among the many considerations which should prevent the identification of neuroglia with connective tissue. In a very young embryo of a mole (I could not determine its age) the cortex of the hemispheres showed granular amorphous substance, in which were embedded spherical masses of somewhat paler color, which had no nuclei, and were therefore not cells. Besides these, there were nucleated masses (apolar cells, therefore) and more developed cells, unipolar, bipolar, and tripolar. Not a trace of a nerve-fibre was visible. In agreement with this are the observations of Masius and Van Lair, who cut out a portion of the spinal cord in a frog, and observed the regenerated tissue after the lapse of a month. It contained apolar, bipolar, and multipolar cells, together with “corpuscles without processes, for the most part larger than the cells, and appearing to be mere agglomerations of granules,”--these latter I suppose to have been what I describe as segmentations of the undeveloped substance. Gray fibres, with a few varicose fibres, also appeared.[169]