688 MAMMALS 



Placentalian cones are all alike in certain respects: they are all only 

 single, without paraboloids, and without oil-droplets. These similarities 

 are negative, and really mean that placental cones are cones reduced to 

 their lowest structural terms. Naturally they would be alike, even if 

 those of the tree-shrews, the higher primates, the duplex descendants 

 of the pre-creodonts (i.e. carnivores, artiodactyls) , and the duplex de- 

 scendants of the Cretaceous pro-ungulates (/. e. hyracoids, proboscideans, 

 perissodactyls) all represent independent productions of new cones in 

 erstwhile pure-rod retinae. 



The absence of the paraboloid in placentals is no proof of an identity 

 of placental cones with those of monotremes and marsupials. The latter 

 groups have lost the paraboloid, to be sure; but the cones of placentals 

 would not be expected to have evolved them even if those cones are 

 'new'. Paraboloids occur only in the cones of groups which have retinal 

 photomechanical changes, and the paraboloid has been claimed to be a 

 reserve of food which furnishes the energy for the activity of the cone 

 myoid. The cones of lampreys and elasmobranchs naturally have never 

 produced them, nor have the cones of snakes, which are certainly 'new' 

 cones. 



If the placentalian cone represents the reptilian droplet-bearing single 

 cone, then one can understand its lack of the oil-droplet icf. opossums) ; 

 but what has become of the reptilian double cone, so stubbornly per- 

 sisting in even the most strongly nocturnal of the lower mammals except 

 where all cones have been lost (Tachyglossus)? Elsewhere above the 

 fishes,* double cones have never been either discarded, or transmuted 

 into rods, without the matching single cones also undergoing discard 

 or transmutation. 



It seems highly significant that the placentalian cone has no con- 

 sequential capacity for color vision except in the primates, where color 

 vision has evolved within the group (see pp. 518-21). If the duplex 

 placental mammals had had continuously duplex retinae ever since the 

 placentals originated, then all such mammals, and not the simians alone, 

 should have as complete a color-vision system as that which character- 

 izes the Sauropsida; for, they should have retained that same system — 

 having retained the same cones. 



All in all, it seems most probable that at one time the only living 

 placentals had no cones, but only the rods which we see in the lower 



*The chondrosteans and Neoceratodus have apparently lost ancestral double cones 

 Plate I. 



