GREGORY: FISH SKULLS 



165 



sagittal section of an Arapaima skull (Fig. 59) shows that even though the frontals extended 

 forward as a thin sheet for some distance beneath the posterior ends of the nasals, they were 

 still widely removed from the small ethmoids, which were lodged in a terminal notch be- 

 tween the nasals and the premaxillse. In view of the fact that in all other known primitive 



5cl^ 



crt- 



Arapaima ^i^as 



Fig 59. Arapaima gigas. 



teleosts the mesethmoids are larger and more important than the nasals, it would seem 

 that the opposite conditions in the osteoglossids, with the consequent resemblances to the 

 early ganoids, is due to convergence, like the sculpturing of the surface bones. And this 

 inference becomes more probable in view of the many other losses and specializations of 

 the Arapaima skull, for instance, the loss of the basisphenoid, of the supramaxillae, the 

 spreading of the symplectic, the marked reduction of the subopercular and the enlarge- 

 ment of the interopercular, which is seen only in the medial aspect of the preopercular. 



The osteoglossids, like other primitive isospondyls, have a well developed "scale bone" 

 (the "supratemporal" of Owen, Ridewood, Starks). In my specimen of Arapaima, how- 

 ever, there are two surface bones (Fig. 59), either one of which might be named the scale 

 bone or lateral extrascapula. The first lies in the usual position immediately below and 

 in front of the posttemporal; the second lies in front of the first and articulates with the 

 pterotic, with the enlarged fourth suborbital, with the preopercular and the opercular. 

 Comparison with Ridewood's figures of osteoglossids shows that he has applied the name 

 supratemporal, in all three genera, to the second of these two, the first not being shown in 

 the figures. Comparison with his figures of the supratemporal region of other isospondyls, 

 however, reveals the fact that in Elops the large "supratemporal" was nearly divided in 

 two by a deep posterior notch and that in the clupeoid genus Chanos there is, in addition 

 to the "supratemporal," a separate bone to which he applies the name "subtemporal," 

 and which has exactly the same position and connections as the above noted "second scale 

 bone" of Arapaima. Ridewood (1904^, p. 485) notes that this element in Chanos carries 

 a branch of the sensory canal that passes downward to the preopercular; also that this fact 

 "taken in conjunction with the position of the bone below the squamosal [pterotic] and 

 above the preopercular, points to the conclusion that the bone is the homologue of that 

 which, in the Salmon, Parker (Phil. Trans. Roy. Soc, clxiii, 1873, p. 99 and PI. \T, Fig. 1, 



