STUDIES IN THE EETINA. 317 



Further, tlie passage of tlie filaments from these ganglionic 

 nuclei through the thick inner reticuhir layer to join the nuclei 

 of the middle layer can be demonstrated. This is interesting for 

 our future argument, for it shows that, in spite of the extreme 

 delicacy of the filaments, they can persist over considerable 

 distances, establishing organic connection between 

 nuclei or systems of nuclei which are far apart. 



It surprised me somewhat at first to find such delicate 

 filaments running through a layer so variable in its texture. 

 For, as we saw in Part V, the variations in the conditions of 

 this layer are very great. Its strands are close and finely 

 matted, forming a " punkt-substanz" in very young eyes, but 

 usually become very coarse and irregularly meshed in older eyes. 

 We have seen one cause for this in that its component threads 

 are sooner or later thickened by becoming the tracks along 

 which the absorbed pigmentary matter from the rods streams 

 through the retina (see Part V). At the outset it would seem 

 hopeless to endeavour to trace the nuclear connecting fibrils 

 through this matted layer, since they are but just visible in a 

 clear field, and this is so far true that I have never succeeded 

 in finding them in the coarser conditions of the layer. But 

 traces of them are occasionally visible when the layer is fine- 

 textured (a " punkt-substauz ^'), and these traces are quite 

 sufficient to justify our assumption of their presence as per- 

 manent structural elements. Very thin lightly stained sections 

 will sometimes show a fine regular radial striation of the 

 punkt-substanz. I have endeavoui-ed to reproduce an instance 

 in fig. 15. But the drawing is far too coarse. I had often 

 before noticed such a striation crossing the usual tangential 

 stratification of the layer at right angles, but had never 

 understood it. Its real cause is, however, clearly shown in 

 this figure, because it happened to be associated with nuclei 

 of the middle layer which had their connecting filaments 

 running into the reticular layer jKirallel with one another and 

 evidently continuous with a certain number of the stria3. 



When the layer is coarse-meshed, we must assume that the 

 filaments bend about in the walls of the meshes. We are at 



