546 THE AMERICAN NATURALIST. [Vol. XXXVII. 



to those actually seen, for the fibrils would be nearly at right 

 angles to the longitudinal axis of the rod. Nor do these reactions 

 favor the view held by Bernard (:oi) that the rods are proto- 

 plasmic vesicles filled with an amorphous refractive substance, 

 for the material is not amorphous but gives evidence of longi- 

 tudinal fibrillation. Patten's (98) hypothesis that the outer 

 segments are made up of minute fibrils at right angles to their 

 longitudinal axes is also inconsistent with these observations. 



Although the evidence I have advanced cannot be said to be 

 opposed to the generally accepted view that the rod is made up 

 of many disk-shaped plates. I am not inclined to place so much 

 emphasis on this as some have done. I have obtained abundant 

 evidence for the presence in fresh rods of transverse bands about 

 equal to each other in thickness and held together by an inter- 

 mediate substance of different optical behavior. But I have not 

 found the evidence for the disintegration of a rod into disks at 

 all convincing. There were certainly frequent instances of trans- 

 verse breaking, but it was seldom clear cut and there were often 

 signs of longitudinal splitting and of spreading at broken ends. 



I believe we have in the rod certain conditions analogous to 

 those of striped muscle fibres. Both bodies are positively refrac- 

 tive, both possess a transverse lamellar arrangement of optically 

 differing substances, and under the action of certain reagents 

 both are said to break into transverse segments. 



The structure of the muscle fibre is essentially fibrillar not- 

 withstanding its transverse fracture and 1 believe the structure 

 of the outer segment of the retinal rod to be in this respect like 

 that of the muscle fibre. 



Having given the evidence for the longitudinal fibrous struc- 

 ture of the rods as I have found it by the use of polarized light, 

 I wish to discuss some contradictory results already recorded as 



