specialized than aracanids, and these are mostly con- 

 cerned with dentition and correlated with the coarser 

 diet of balistids. In balistids there are only three pharyn- 

 gobranchials, a toothless suspensory element followed by 

 two elements bearing prominent teeth, while in 

 aracanids the toothless suspensory element is followed by 

 three elements, the first of which is either toothless or 

 with minute teeth but the third and fourth elements with 

 teeth almost as well developed as in balistids but usually 

 of lesser number. The relatively conical teeth of 

 aracanids are more generalized than the heavier notched 

 incisors of balistids in the sense that they are closer in 

 structure to those of the more generalized 

 triacanthodids, the basal plectognaths, while those of 

 balistids are more similar to those of the triacanthids 

 derived from the triacanthodids. However, as indicated 

 previously, there is evidence that the structure of the 

 teeth in balistids and aracanids might not have been 

 much different in the Eocene. By the retention of an in- 

 ner series of teeth in the upper jaw, balistids are more 

 generalized than aracanids, while the variable presence 

 in aracanids of a fifth tooth in what is the outer series of 

 balistids, which always have only four in each half of 

 both jaws, is slightly more generalized than in balistids. 



The uppermost pectoral fin ray of aracanids is relative- 

 ly well developed and the two halves are of about equal 

 length, while in balistids this ray is greatly reduced in 

 size and the medial half is larger than the splintlike nub- 

 bin that represents the lateral half of the ray. Here again 

 the condition in aracanids is similar to that in 

 triacanthodids while that of balistids is similar to that in 

 triacanthids, although the lateral half of the ray in 

 balistids is even more reduced than in triacanthids. It is 

 presumed here that these few more specialized features 

 of balistids were acquired after the divergence of the 

 common triacanthoid phyletic line leading on the one 

 hand to balistids and on the other to aracanids, as dis- 

 cussed subsequently. 



The conversion of a balistidlike fish into an aracanid 

 requires mainly the following: 1) the complete loss of 

 the pelvis along with its rudimentary pelvic fin-ray ele- 

 ment and encasing scales; 2) the complete loss of the 

 three dorsal fin spines and of their three pterygial and 

 supraneural supports, with the possible exception of one 

 of them; 3) a rearrangement of the posterodorsal part of 

 the head that no longer supports the spiny dorsal fin, in- 

 cluding the development of a posteriorly projecting 

 supraoccipital crest, subsequently to be lost by the os- 

 traciids; 4) a flattening of the dorsal surface of the 

 supraoccipital; 5) the elimination of the articulation 

 foramen of the first basal pterygiophore of the spiny dor- 

 sal fin between the epiotic and supraoccipital, and a 

 reduction in the posteromedial region of the exoccipitals 

 so that they no longer are closely apposed to one another 

 in the midline on the posterior wall of the skull; 6) the 

 development of a dorsal flange on the parasphenoid into 

 the orbital septum to meet an anteroventral prolonga- 

 tion of the pterosphenoids; 7) an enlargement of the 

 vomer with a broader surface of suturing to the ethmoid 



and parasphenoid; 8) a reduction of the anterior cross- 

 bar of the T-shaped palatine and a firm suturing of the 

 foot of the T to the ectopterygoid and mesopterygoid; 9) 

 the loss of the inner series of teeth but the retention of a 

 slightly more generalized form and number in the outer 

 series than in Recent balistids; 10) a foreshortening and 

 deepening of many of the hyoid arch elements along with 

 the reduction in width of the first branchiostegal and the 

 development of a constant articulation between the last 

 branchiostegal and the suboperculum; 11) the retention 

 of a third toothed pharyngobranchial and the reduction 

 in size of the teeth on the first of the toothed elements; 

 12) loss of teeth on the fifth ceratobranchial and develop- 

 ment of gill rakers along its anterior edge; 13) great 

 reduction in the size of the urohyal; 14) great increase in 

 the size of the hyomandibular, posttemporal, postcleith- 

 rum, cleithrum, coracoid, epiotic, and pterotic; 15) 

 reduction in the size of the myodome and the elimination 

 of the posterior opening into it; 16) the ossification of 

 Baudelot's ligament; 17) the elaboration of the prootic 

 shelf including a recurved lateral wing; 18) a wider 

 separation of the pterotics on the ventral surface of the 

 skull; 19) the reduction in size and the fusion of the first 

 two vertebrae to the skull; 20) the loss of epipleurals; 21) 

 complete fusion of all the caudal fin supporting elements 

 of the last vertebral centrum; 22) an increase in the size 

 of the neural spines of the predorsal vertebrae and a great 

 reduction in length but a broadening of the haemal 

 spines of most of the caudal vertebrae; 23) a great reduc- 

 tion in the number of dorsal and anal fin rays and of their 

 basal pterygiophores and the development of divergent 

 anal fin basal pterygiophores; 24) concomitant with the 

 reduction in the dorsal and anal fins, an increase in the 

 number of vertebrae with neural or haemal spines an- 

 terior to the first basal pterygiophores of both fins (i.e., 

 an increase in predorsal and abdominal vertebrae); 25) 

 development of lateral flanges on many of the more cen- 

 trally located vertebrae; 26) loss of one branched caudal 

 fin ray; 27) the enlargement and thickening of the scales 

 into heavy, more or less hexagonal, plates over the head 

 and body anterior to about the level of the ends of the 

 dorsal and anal fin bases; 28) the enlargement of a supra- 

 neural element into a long strut projecting anteriorly 

 from the dorsal fin base; 29) the suturing of the actinosts 

 to one another and to the scapula and coracoid. 



Some of these above changes from the balistid to 

 aracanid type of organization are not shared by the os- 

 traciids. For example, ostraciids do not have: 1) the 

 posteriorly prolonged supraoccipital crest; 2) the es- 

 pecially elongate supraneural in front of the dorsal fin; 3) 

 the immensely enlarged postcleithrum; 4) the pos- 

 teriorly prolonged epiotic and generally more elongate 

 postorbital region of the head; 5) the elongate neural 

 spines of most of the predorsal vertebrae. This indicates 

 either that the ostraciids diverged from the aracanids at 

 a time before the ancestral line had developed these 

 latter differences with balistids or that the ostraciids at 

 any early stage had similar differences with balistids 

 which they have subsequently modified. I suspect, with- 



