Secondly, as the resistance is lowered in the central organs corresponding to the end organs of the retina upon which the stimulation falls, the image tends to assume its true form, but irradiation has been, and probably still is, present through the layers of the retina, so that certain favorably located portions of the image secure reënforcement by means of this irradiation, in the manner described, and these portions appear in consciousness sooner than the others. This reënforcement, in the case of the travelling oblong image, will make it appear convex. Moreover, since the resistance of the synapses corresponding to the centre of the oblong images will be less than those corresponding to the ends, there will be a certain tendency to "drain off" the stimulation from the rest of the image, a sort of reënforcement of the reënforcement, which will also help in making the image appear curved. Of course all the conditions which we found to modify the curvature of the images will still hold good, these conceptions being used only to describe the course of events which causes the image to appear convex. Thus a very weak or a very intense or a very long or an excessively short image will not appear curved, owing to a lack of difference in intensity between the ends and the centre great enough to produce perceptible curvature. As to the third kind of irradiation, that which proceeds from the ends of the moving image over the unstimulated portions of the retina, and which has the appearance of long streamers of light extending outward and backward from the moving image, this may be regarded as being in certain respects a form of the first and very rapid kind of excitation. It may well be that all the outrunning excitation which occurs immediately upon stimulation does not find its way to the central organs through those nerve-paths which correspond to the directly stimulated portions of the retina, even after the form of the image may be very clearly determined, but that some excitation proceeds outward from one retinal element to another, arousing fainter and fainter excitation as it proceeds. This being the case, we should expect to find these streamers of light from the ends of the image extending outward and backward over the retina. Of course the faster the image moved and the more intense it was, the longer then would be these streamers. For if the image moved very fast, very much less of the excitation would be "drained off" through the directly stimulated portion, and thus more of the excitation would be left behind, so to speak, by the image when it moved along rapidly, and this would appear to drag farther and farther behind. Of course these streamers being curved backward would appear more curved the faster the image moved, and if the pulsative processes occurred with these stimulations which occur in the course of other retinal stimulations, we should have Charpentier's "palm-branch" phenomenon.