586 HIGHER FISHES 



In nearly all teleosts there are three types of visual cells : rods, single 

 cones, and the twin cones which the teleosts monopolize (Fig. 170). 

 In a particular retina, or in particular regions in some retinae, only one 

 of these, or any two of them, may occur. The pure-rod teleosts include 

 the deep-sea forms and (according to Verrier) one siluroid, Clarias 

 batrachus* Twin cones alone occur in Pollachius pollachtus and some 

 Gadus spp.,"*" in Scorpcena porcus, Sebastodes elongatus, Alosa finta, 

 and in all but the extreme periphery in a number of others — particularly 

 flatfishes and swift surface forms (tunas, mackerels, mullets, etc.). The 

 relative numbers of twin cones (where they are mingled with singles 

 and with rods) decrease with an increase in the species' preferred depth 

 of swimming. Clearly, the twin cone is associated with exposure to 

 bright light. 



The origin of the twin cone cannot be traced with certainty. It usually 

 looks so much like two single cones fused together that this simplest 

 explanation is the one dictated by the law of parsimony (see Plate I). 

 But the holostean visual-cell assortment looks superficially much like the 

 teleostean. Here (Fig. 170b) there are rods, single and double cones. 

 The teleostean twin might have arisen from the holostean double through 

 an equalization of the latter's two members, involving the loss of the 

 accessory's paraboloid (the chief cone's oil-droplet being already long 

 since gone in Amia, and replaced functionally there by a yellow cornea). 

 Supporting this possibility is the fact that double cones, of sorts, do 

 occur in teleosts — that is, conjugate elements whose two members are 

 unlike in size and, to some extent, in structure. The oldest report of 

 such elements is that of Greeff, who described them for Rutilus rutilus 

 in 1900. The writer has found the conjugate elements of the goldfish 

 (Carassius auratus) to be of this same sort. Rutilus and Carassius are 

 both members of the minnow family (Cyprinidae), which stands rather 

 near the bottom of the malacopterygian division. The Salmonidse rank 

 about as low or lower; and Mile. Verrier and Miss McEwan have de- 

 scribed doubles, or unequal twins, for Salmo gairdnerii irideus and S. 

 trutta fario. The occurrence of so many instances of unequal twins 



*A dubious observation, for a few years later she reported rods and single cones, in equal 

 numbers, for Clarias dussumien. 



fTending to throw doubt upon the coryphsenoidid ancestry of the cods (see pp. 389-9), 

 since for the cods to have had pure-rod ancestors, and yet possess twin cones, would neces- 

 sitate believing that they had invented twin cones for themselves. Still, the absence in gadids 

 of accommodation and of scleral ossicles, together with the particularly easy 'splittability' of 

 their comese, suggests that these fishes may well have risen secondarily from the ocean floor. 



