of the family that they are placed in different sub- 

 families than the Recent species, while the 

 Triacanthidae are represented by Protacanthodes om- 

 bonii, a somewhat intermediate form between the 

 Triacanthodidae and Triacanthidae and different 

 enough from the Recent triacanthids and those known 

 from the Oligocene and Miocene to be placed in a 

 separate subfamily. 



By contrast, of what little is known of them, the single 

 species of Eocene aracanid and ostraciid are relatively 

 modem enough in general countenance to be easily ac- 

 commodated in the same higher categories as the Recent 

 species. The diversifications of the families apparently 

 was rapid and in both cases with the derived families 

 specialized for a shallow-water existence coextant with 

 the ancestral families successful continuance in deeper 

 waters. 



Relationship to the Balistidae. — The derivation of 

 the ostracioids presents a slight enigma. Both families of 

 ostracioids appeared in the Eocene with genera not 

 markedly different from Recent genera, and paleon- 

 tology offers no clue as to what plectognath group the os- 

 tracioids are most closely related. As pointed out in the 

 historical section of the Introduction, the ostracioids 

 have been considered variously as a third major sub- 

 ordinal group of plectognaths, or as members of both the 

 Sclerodermi and Gymnodontes. Regardless of how they 

 have been classified, their relationships to the 

 triacanthoids and balistoids, with which they are usually 

 placed, have never been analyzed much beyond the sim- 

 ple statement that ostracioids have individual discrete 

 teeth protruding from the jaws and hence show an af- 

 finity with the scleroderms and that since they lack the 

 spiny dorsal and pelvic fins, they are probably related to 

 the balistoids, which have the spiny dorsal and pelvic 

 fins at least reduced in size or in number of elements in 

 relation to the triacanthoids. 



This is not to belittle the clue that the presence of dis- 

 crete protruding relatively normal teeth and lack of spiny 

 dorsal and pelvic fins in ostracioids affords, for I agree 

 that they are important indications of their relationship 

 with the balistoid scleroderms. Any such clues are impor- 

 tant, for ostracioids are so highly modified for life within 

 a shell that many of their features are unique. 



The divergence of many of the anal fin basal pterygio- 

 phores away from the midline is not found in fishes other 

 than ostracioids, while the fusion of several abdominal 

 vertebrae to the skull and the heavy ossification of 

 Baudelot's ligament are at least highly unusual features 

 among fishes. Other features of the ostracioids that are 

 less distinctive, but at least unique among the plectog- 

 naths, are the heavy lateral expansions of the ventral 

 edge of the parasphenoid forming a partial hard palate, 

 the articulation of the preoperculum with a groove on the 

 lateral surface of the hyomandibular, the complete fu- 

 sion of all epural, hypural, and parhypural elements into 

 a single largo plate, the displacement in ostraciids of the 

 haemal canal to one side or the other of the midline, and 



the great expansion in aracanids of the postcleithrum. 

 Beset with a multitude of such specializations one is for- 

 tunate to find even a few structures in ostracioids that 

 give evidence concerning the origin of the group. 



The forward extension of the prootics as a shelf under 

 the orbit is unusual among fishes, and in the plectog- 

 naths it is found only in the balistoids and ostracioids. 

 The interoperculum is a short and simple rod not ex- 

 tending posteriorly past the level of the epihyal only in 

 balistoids and ostracioids among the plectognaths, and 

 the operculum and suboperculum are about equally 

 reduced in the two groups. Only in balistoids and os- 

 tracioids is the hyomandibular supported dorsally by the 

 prootic and pterotic alone, without contacting the 

 sphenotic. The palatine in balistoids and ostracioids is 

 reduced in size in comparison to other plectognaths, al- 

 though it is articulated differently in these two super- 

 families. The shape and size of the inflexibly articulated 

 premaxillary and maxillary are very similar in balistoids 

 and ostracioids, as is the rotation of the upper jaw around 

 a laterally expanded and buttressed ethmoid-vomerine 

 region. The shape of the teeth of Recent ostracioids is 

 somewhat intermediate between that of triacanthodids 

 and that of triacanthids and balistids, but there is an in- 

 dication that the teeth of the Eocene ostracioids were 

 more balistidlike than at present (see description of 

 Eolactoria). 



The ostracioids do not show any such similarities with 

 any of the gymnodonts, and it is obvious from the above 

 that they have their closest anatomical affinity with the 

 balistoids. Since the ostracioids have completely lost the 

 spiny dorsal and pelvic fins, and the pelvis, one must 

 consider the possibility that they are, among the 

 balistoids, more closely related to the derivative mona- 

 canthids than to the ancestral balistids, for it is among 

 the monacanthids that the reduction in size and number 

 of dorsal fin spines (to the presence of a single and 

 sometimes weakly developed spine in a few genera) and 

 in the size of the pelvic fin (completely lost in many 

 genera) and pelvis (to a relatively slender shaft in a few 

 genera) has reached its extreme for the superfamily. 

 However, the monacanthids possess a number of 

 specialized features in comparison to balistids that show, 

 as discussed below, that the ostracioids are more closely 

 related to balistids than to monacanthids. 



The T-shaped palatine of balistids, with the foot of the 

 T articulated against the ectopterygoid and the crossbar 

 articulated between the maxillary and ethmoid-vomer- 

 ine region, becomes a simple rod in monacanthids repre- 

 senting only the crossbar of the balistid palatine, and it is 

 usually well removed from the ectopterygoid, even though 

 connected to it by a strong ligament. The columnar 

 palatine of ostracioids, with its ventral end sutured to the 

 ectopterygoid and its dorsal end firmly held to the eth- 

 moid-vomerine (especially to the latter) region is dif- 

 ficult to derive from the highly specialized and greatly 

 reduced in size monacanthid condition but easy to derive 

 from the balistid condition. The ostracioid palatine 

 represents basically the shaftlike foot of the T-shaped 



