32 H. M. BERNAKI). 



the supply might be less than the demand^ and the layer in 

 consequence thin away. The really important fact is that we 

 find ourselves compelled to assume a migration of nuclei 

 on a very large scale. Not only can it be shown that nuclei 

 travel outwards through the outer reticular layer in great 

 numbers to become rod nuclei, but that all the nuclei destined 

 for this function have, at least after the original supply has 

 been used up, to travel down from the edge of the retina 

 along the middle nuclear layer to their ultimate destinations. 

 In addition to these movements it can be shown that 

 nuclei of the so-called " ganglionic cell '^ layer occasionally 

 travel outwards through the thick inner reticular layer until 

 in old eyes {g.l., fig. 21) they may be almost entirely used up. 



Our investigations into the growth-processes of the retinas 

 of some score of frog- and toad-tadpoles having thus 

 eliminated all other possible sources for the nuclei of new 

 rods required by the central regions of the retina except this 

 immigration from the rim, it remains for us to see what direct 

 or indirect evidence there is for such an unexpected 

 phenomenon, not as an occasional, but as a normal growth- 

 process. It is hardly likely that such a movement could take 

 place without showing visible traces, — without leaving its 

 mark on the tectonics of the retina itself. We shall now see 

 that this surmise is fully justified. 



The divisions take place in early growth-stages along the 

 whole of the rim into the iris, but are most numerous in the 

 angle between the iris and the cup of the retina. To this 

 angle, as the iris becomes differentiated, they are usually con- 

 fined. They also take place chiefly, though not exclusively, in 

 the outermost row of nuclei, in what I have elsewhere called 

 the palisade layer. In the part Avhere the divisions are active 

 it is common to find the large, radially arranged, more or 

 less spindle-shaped nuclei attached either to the internal or 

 to the external limiting membrane by a frequently thick stain- 

 ing cytoplasmic strand. The nuclei are usually so numerous as 

 to obscure the sections, so that one cannot state that these 

 strands, each with its suspended nucleus, run distinct and 



