edge closely in contact with several vertebrae) which 

 leads to incomparably high postanal counts. In lacto- 

 phrysins the more generalized group, Lactophrys = 

 Rhinesomus, have three postanal vertebrae and the more 

 specialized group, Acanthostracion, four, while Ostra- 

 cion and Rhynchostracion would thus be considered to 

 have a more specialized number of postanal vertebrae. 

 Trifid neural spines on some of the postdorsal 

 vertebrae being a specialization found in all ostraciins, it 

 seems reasonable to assume that the greater the number 

 of vertebrae with trifid neurals, the greater the degree of 

 specialization in this character. The greatest number, 

 four, is found in L. cornuta, with L. fornasinii having 

 three, and Tetrosomus, Ostracion, and Rhynchostracion 

 either two or three. 



While the number of vertebrae involved in the fusion 

 complex with the rear of the skull in all species of the 

 subfamily is four, except five in L. fornasinii, the lack of 

 fusion between the fused first-second vertebrae and the 

 fused third-fourth vertebrae in 0. tuberculatus even in 

 large adults and of the sometimes only partial fusion in 

 this region in 0. lentiginosum seem more generalized 

 than having at least the basal regions of the four 

 elements fully fused, as in Rhynchostracion, Lactoria, 

 and Tetrosomus. 



As seen above, no one genus has a relatively full com- 

 plement of either the specialized or generalized charac- 

 teristics discussed, all of the genera having a difficult to 

 interpret mixture of both. What follows is a highly specu- 

 lative interpretation of the evidence. 



The probably triangularly or pentagonally carapaced 

 ancestral group of the Ostraciidae had no more than four 

 vertebrae involved in the anterior fusion complex, the 

 haemal spine of the penultimate vertebra autogenous, 

 the last dorsal and anal fin basal pterygiophores without 

 shelves for carapace support, and the neural spines of the 

 more posterior vertebrae undivided. This ancestral group 

 gave rise on the one hand to the Lactophrysinae, which 

 remained relatively unchanged from the ancestral condi- 

 tion except for subsequently involving the fifth vertebra 

 in the fusion complex and having the penultimate 

 haemal spine become fused to the centrum in the most 

 specialized genus, in which the number of postanal 

 vertebrae was also increased. On the other hand the 

 ancestral group gave rise to the Ostraciinae, which 

 retained the ancestral condition of the anterior vertebral 

 fusion complex but which specialized in a number of 

 ways mostly involving the simplification of the postanal 

 haemal spines and of the development of carapace sup- 

 porting structures on the last dorsal and anal fin basal 

 pterygiophores and on some of the neural spines of the 

 postdorsal vertebrae. 



The line immediately ancestral to the Recent ostra- 

 ciins is envisioned as having the last dorsal and anal fin 

 basal pterygiophores moderately well developed (i.e., 

 larger than in the lactophrysins), with moderate shelves 

 (probably less well developed than in their greatest 

 development in the Recent species), and with only two or 

 three of the postdorsal vertebrae with trifid neural 

 spines. This hypothetical ancestral line leading from the 



immediate ancestry of the lactophrysins would probably 

 have had a triangular or pentangular carapace, and it is 

 postulated to have diverged into two lines of Recent 

 species, one line containing basically triangular forms 

 with many carapace spines {Lactoria and Tetrosomus) 

 and the other basically rectangular forms with a rela- 

 tively spineless carapace (Ostracion and Rhyncho- 

 stracion). 



In the Lactoria- Tetrosomus line of ostraciins, a form 

 close to L. cornuta, but with a still only moderately 

 enlarged last dorsal fin basal pterygiophore and only two 

 or three trifid neural spines, was probably ancestral, L. 

 cornuta being the only species to have the haemal canal 

 pierce the fused haemal spine of the penultimate 

 vertebra and enter into and exit from the caudal plate 

 and to have a relatively moderate lateral shelf on the last 

 anal fin basal pterygiophore. The L. cornuta-like (with 

 the exceptions noted) ancestral line probably gave rise to 

 L. cornuta on the one hand by the further enlargement of 

 the last dorsal fin basal pterygiophore and its shelf and 

 by the trifid division and enlargement of additional post- 

 dorsal neural spines. 



On the other hand the pre-L. cornuta ancestor is seen 

 as having given rise to L. fornasinii with no increase in 

 the number and size of trifid neural spines, the reduction 

 of the size of the haemal arches of the postanal vertebrae, 

 and the elimination of their foramina so that they are 

 solid and not pierced by the haemal canal which also 

 does not pierce the caudal plate, an enlargement of the 

 moderate shelf on the last anal fin basal pterygiophore 

 into a more substantial anterolateral shelf, a reduction in 

 the size of the neural spine of the vertebra (13th) between 

 that which supports the base of the last dorsal fin basal 

 pterygiophore and the most anterior of the three 

 vertebrae (14th to 16th) with trifid neural spines, and a 

 slight reduction in the size of the last dorsal fin basal 

 pterygiophore and its posterior carapace shelf so that the 

 basal portion of the pterygiophore is rodlike rather than 

 anteroposteriorly expanded while the posterior end of the 

 distal shelf does not make contact with any neural 

 spines. 



Probably a close derivative of the immediate ancestor 

 of L. fornasinii is Tetrosomus. Tetrosomus shows the 

 same tendencies, but more strongly, as L. fornasinii does 

 in the reduction of the size of the neural spines below the 

 last dorsal fin basal pterygiophore and in the size of the 

 latter. In Tetrosomus the size of the neural spines of the 

 12th to 14th vertebrae is greatly reduced, and that of the 

 14th is no longer trifid. The trifid neural spines of the 

 15th and 16th vertebrae are of about the same moderate 

 size as in L. fornasinii, the anterolateral shelf on the last 

 anal fin basal pterygiophore just as massive, and the 

 solid haemal spines of the postanal vertebrae just as 

 reduced in size and not pierced by the haemal canal, 

 which likewise does not enter the caudal plate. Thus, L. 

 fornasinii shares many peculiarities with Tetrosomus, 

 and a form like it is very likely to have been ancestral to 

 Tetrosomus. The higher crested carapace in Tetrosomus 

 could be either an enlargement of the lower crest in Lac- 

 toria or a hold over from the line leading from the pre- 



