scarcely developed at all in inermis and lunaris. The 

 neural and haemal spines of the last few caudal vertebrae 

 are shorter in scleratus than in the other species, 

 associated with the depressed caudal peduncle. 



A feature unique to Lagocephalus among the tetra- 

 odontids is the posterior prolongation of the distal ends of 

 the last basal pterygiophores of the dorsal and anal fins 

 as prongs above and below the neural and haemal spines 

 of the vertebrae just behind the bases of these fins. The 

 prongs are moderately developed in inermis and lago- 

 cephalus, slightly longer in laevigatas and lunaris, and of 

 greatest length in spadiceus (dorsal basal pterygiophore 

 only, the anal basal pterygiophore having a prong of only 

 moderate length) and in scleratus. 



All of the species of Lagocephalus have two or three 

 rounded trituration teeth in a single series to each side of 

 the midline of the upper jaw but none in the lower jaw, 

 and all possess an interhyal, while none of the abdominal 

 vertebrae possess complete haemal arches. The first ray 

 in both the dorsal and anal fins of Lagocephalus is excep- 

 tionally short (slightly longer in young than adults) and 

 sometimes lacks cross-striations, and it is shorter than 

 the first ray of any of the other tetraodontids examined. 

 In all of the species of Lagocephalus the form of the 

 epural is characteristic. In other tetraodontids the epural 

 varies from a more or less square block of bone to an 

 elongate rod, but the orientation of the bone is more or 

 less vertical or obliquely anteroventral to posterodorsal, 



Figure 261 .—Dorsal views of skulls of: 



left, Sphoeroides testudineua, 68.5 mm SL, Venezuela; 



right, S. annulatus, 174 mm SL. locality unknown. 



while in Lagocephalus the epural is dorsoventrally flat- 

 tened and prolonged anteroposteriorly, in which direc- 

 tion it is oriented, an apparently specialized condition 

 unlike that of any other gymnodont. The presence of 

 two lateral lines on the body in Lagocephalus can also be 

 considered a specialization, since only a single lateral 

 line is present in triodontids and the basal triacan- 

 thodids. 



Fraser-Brunner (1943) believed that Lagocephalus is 

 the most primitive tetraodontid and that Canthigaster 

 was derived from it, but he gave no reasons for so think- 

 ing. As discussed earlier, Canthigaster probably evolved 

 from a Carinotetraodon-like line. I find no particular 

 similarities between Canthigaster and Lagocephalus. 

 Fraser-Brunner thought that the skull of molids was 

 "almost exactly similar to that of Lagocephalus, even 

 possessing the postero-lateral limbs of the frontals 

 characteristic of that genus." In fact, the frontals of 

 molids do not have prominent posterolateral wings and 

 to my eyes the skulls of molids show no particular 

 similarities to those oi Lagocephalus, but, rather, as dis- 

 cussed under the Molidae, they do show similarities to 

 that of Triodon. One of the few similarities that at least a 

 few species of Lagocephalus have to Triodon not found in 



