of contact with the first vertebra. In balistids the first 

 basal pterygiophore is a highly complex, horizontally 

 elongate structure articulated high on the rear of the 

 skull by a short ventral shaft. The pterygiophore features 

 a concave dorsal surface in the region of the first two 

 spines that opens laterally through the wall of the pte- 

 rygiophore to admit the muscles controlling the locking 

 mechanism, there being nothing similar to this in tria- 

 canthids. 



In triacanthids the second basal pterygiophore, bear- 

 ing the third spine, is relatively small, has a relatively 

 flat dorsal surface, and is not sutured to the first basal 

 pterygiophore, while in balistids it is relatively large, 

 deeply concave dorsally, and is firmly sutured to the 

 first. The third basal pterygiophore of triacanthids, bear- 

 ing the fourth spine, is probably the pterygiophore 

 modified in balistids into the supraneural strut bracing 

 the rear end of the carina (formed by the first two pteryg- 

 iophores) against the tip of the neural spine of the fifth 

 vertebra and the anteroventral edge of the first basal 

 pterygiophore of the soft dorsal fin. 



The articulation of the carina high on the rear of the 

 skull in balistids has drastically changed the configura- 

 tion of the skull in this region, but it still bears evidence 

 of the triacanthid ancestry of the balistids. In both the 

 triacanthids and balistids the posterior surface of the 

 supraoccipital is concave, and while the supraoccipital of 

 balistids is so modified that it is no longer a simple dome, 

 it does exclude the epiotics from meeting one another on 

 the dorsal surface of the skull, and the epiotics contact 

 the frontals, just as in triacanthids and their ancestral 

 hoUardiin triacanthodids, and not as in the tria- 

 canthodin triacanthodids. The face that one Oligocene 

 triacanthid {Cryptobalistes) has the otherwise 

 triacanthidlike spiny dorsal fin articulated high on the 

 rear of the skull by a short ventral shaft, even though the 

 more numerous basal pterygiophores are small and do 

 not form a carina, and there is neither a supraneural 

 strut nor a locking mechanism between the first two 

 spines, shows the potentiality and preadaptation of the 

 triacanthids to give rise to the balistid-type spiny dorsal 

 fin and pterygiophores. 



In triacanthids the premaxillary pedicels are well 

 developed and help allow the protusibility of the upper 

 jaw, while the premaxillary and maxillary are movably 

 articulated to one another through fibrous tissue. In 

 balistids there are no premaxillary pedicels, the upper 

 jaw is not protrusible and the premaxillary and maxil- 

 lary are immovably held to one another, often by sutur- 

 ing. The palatine in triacanthids is a flat squarish bone 

 with an anterior prong articulating with the deeply 

 recessed dorsolateral surface of the maxillary, while in 

 balistids the palatine is T-shaped, the anterior end of the 

 cross bar articulated with a slight concavity on the dorso- 

 lateral surface of the maxillary. In triacanthids there are 

 usually five teeth in an outer series in each half of each 

 jaw, internal to which are one or two teeth in each half of 

 the upper jaw and one in each half of the lower jaw. In 

 balistids there are always four teeth in an outer series in 

 each half of each jaw, and three inner series teeth in each 



half of the upper jaw but none in the lower jaw. Thus, 

 balistids have slightly reduced the number of teeth as 

 found in triacanthids, and also modified their form. In 

 triacanthids the edges of the teeth are straighter than in 

 the typically notched teeth of balistids, and while the 

 size of the teeth in balistids is usually larger than that of 

 triacanthids, they are thinner and less massive, except in 

 most large adult balistids in which the teeth become just 

 as massive as in triacanthids. The teeth in the balistids 

 Canthidermis, and, to a lesser extent, Odonus and 

 Xanthichthys, do not tend to become so massive in large 

 adults. 



In triacanthids the pelvis is shaftlike only posterior to 

 the level of the pelvic spines in about the middle of its 

 length and cannot be rotated around its anterior ar- 

 ticulation with the cleithra. In balistids the pelvis is 

 shaftlike throughout its length and can be rotated in a 

 vertical plane around its anterior articulation with the 

 cleithra, while the pelvic fin is represented by a rudimen- 

 tary but complex structure representing the fused fins 

 from each side at the posterior end of the pelvis, sur- 

 rounded by enlarged encasing scales. Whether this com- 

 plex fusion product of the pelvic fins represents the 

 spines or rays of triacanthids is problematical. One sus- 

 pects that because the fin rays in triacanthids are 

 relatively rudimentary and often lost while the spines are 

 always well developed, that, in the process of reduction 

 of the pelvic fin from that of triacanthids to that of 

 balistids, the rays would be irrevocably lost early in the 

 process, leaving the spine alone to become further reduc- 

 ed in size and eventually fused with its opposite member 

 at the rear of the pelvis. As the spine became of relatively 

 minute size and less fully ossifed, it may have secondari- 

 ly become branched distally, as in many balistids, this 

 branching of the unstriated element perhaps not at all 

 homologous to that of rays. 



It is possible that the reduced spine migrated pos- 

 teriorly along the shaftlike posterior portion of the pelvis 

 eventually to arrive at its extreme posterior end while at 

 the same time the anterior half of the pelvis became 

 shaftlike and rotatable. This seems more likely to me 

 than to account for the position of the balistid pelvic fin 

 as a result of the loss of the shaftlike posterior half of the 

 triacanthid pelvis at the same time that the spine was 

 reduced and the anterior half of the pelvis was becoming 

 shaftlike and rotatable and also greatly elongate so that 

 the stationary spine was moved posteriorly to the anal 

 region by the growth of the anterior half of the pelvis 

 rather than by migration. 



The shaftlike form of at least the posterior half of the 

 pelvis in balistids, triacanthids, and hoUardiin 

 triacanthodids is additional evidence of the phylo- 

 genetic validity of their relationship. 



The vertebrae of triacanthids are 8 -f 12, while those of 

 balistids are 7 -(- 11. In triacanthids there are three tooth- 

 ed pharyngobranchials following the toothless suspen- 

 sory element, but only two toothed elements in balistids. 

 In triacanthids the hyomandibular is supported dorsally 

 by the prootic, pterotic, and sphenotic, but in balistids 

 only by the prootic and pterotic. In triacanthids there is 



133 



