logenetic relationship to the short-based soft dorsal and 

 anal fins of ostracioids. Nevertheless, considering the 

 reductive tendencies in the spiny dorsal fin in the plec- 

 tognaths, it is not inconceivable that a group like the 

 spinacanthins at one time early in their history greatly 

 enlarged the spiny dorsal fin, an experiment which even- 

 tually proved a failure and gave way to the more usual 

 trend of reduction, leading to a spinacanthin line with 

 the spiny dorsal fin reduced in size and eventually lost, 

 leaving a hypothetical form much more suitable for con- 

 sideration as an ancestral line to the ostracioids than are 

 the two presently known species. 



The scales in the tetraodontid Ephippion guttifer are 

 relatively normal prickles in the young, with two or three 

 radiations from the basal plate supporting the projecting 

 spine. By about 100 mm SL, however, only the prickles 

 on the belly retain the normal shape, those of the body 

 having the basal plate enlarged and elongate, and the 

 projecting spine reduced in size. By 200 mm SL the basal 

 plates of the scales between the levels of somewhat 

 behind the pectoral fin base and the soft dorsal and anal 

 fin origins are further enlarged into irregularly rounded 

 plates, with irregularly granular surfaces, closely held 

 together by numerous interdigitations along their edges 

 of contact. The scale plates anterior and posterior to the 

 girdle of sutured plates are progressively more distantly 

 spaced from one another and elongate rather than round- 

 ed. The scales of the belly, below the carapace girdle, re- 

 main as normal prickles. At larger sizes, as illustrated for 

 a 325 mm specimen, the scales of the girdle become 

 progressively larger and thicker and the girdle itself ex- 

 tended further anteriorly and posteriorly. The surface 

 sculpturing of the plates becomes more regularly 

 granular and the firmness of the interdigitation between 

 the plates along all their edges of contact stronger and 

 less flexible, the largest and heaviest plates mostly in the 

 ventral region of the girdle. The plates have relatively ir- 

 regular outlines, but many range from triangular to hex- 

 agonal. 



This girdle of interdigitated thickened scale plates in 

 Ephippion adults is fully as strong and solid as that of 

 ostracioids, and only slightly more flexible, mainly 

 because it does not completely enclose the body, the bel- 

 ly always retaining normal prickles, and the girdle not 

 extending onto the head. In four specimens, the size of 

 the largest scale plates increased as follows: 3.6% SL at 

 101 mm; 4.3% SL at 232 mm; 7.1% SL at 325 mm; 9.2% 

 SL at 391 mm. In ostracioids the largest scale plates 

 range from about 6 to 13% SL relatively independent of 

 specimen size beyond juvenile stages. 



Among the molids the individual scale plates are 

 relatively small. In Mola and Masturus the basal plates 

 are more or less rounded and flexibly articulated with 

 one another by delicate denticulations. In Ramania, 

 however, the basal plates become thicker with in- 

 creasing specimen size and the extent of the interdigita- 

 tion also increases so that adults have nearly the entire 

 body covered by an only slightly flexible carapace of 

 small (in two specimens, 0.8% SL at 65.1 mm and 1.2% 

 SL at 493 mm) minutely denticulated and irregularly 



geometrically arranged basal plates with only low and in- 

 conspicuous surface sculpturing, ranging from 3 to 12 

 sided, but with some distinctly hexagonal. This carapace 

 of scale plates never attains the thickness of that of os- 

 tracioids or Ephippion. 



In balistids the thick scale plates are usually rhom- 

 boidal with slightly overlapping edges (plates not over- 

 lapping in Canthidermis) and usually with low surface 

 sculpturing of wide variety, while some plates (es- 

 pecially on the caudal peduncle) may bear higher spiny 

 processes. Although usually rhomboidal, the anterior 

 and posterior apices sometimes are flattened, the plate 

 then having a decidedly hexagonal shape. Scale plates 

 such as found in balistids could easily be ancestral to 

 those of ostracioids, by an increase in thickness and sur- 

 face area, more usual hexagonal shape, and conversion 

 from slightly overlapping edges to apposed inter- 

 digitated edges. 



The strongest reason, however, for not considering the 

 spinacanthins as ancestral to the ostracioids is that the 

 osteological evidence from Recent species indicates that 

 the balistids and ostracioids evolved from a common 

 triacanthid ancestral stock, as discussed fully under the 

 relationships of the ostracioids, but basically because 

 balistids and ostracioids have a well-developed prootic 

 shelf, the hyomandibular supported by the prootic and 

 pterotic but not by the sphenotic, an expanded 

 parasphenoid, and similarities in the shape or size of the 

 interoperculum, operculum, suboperculum, premaxil- 

 lary, and maxillary. 



Thus, for the present at least and until more is known 

 of the internal anatomy of the spinacanthins, it seems to 

 me reasonable to tentatively consider the spinacanthins 

 an extinct line of early triacanthodid experimentation 

 not ancestral to any of the other known subgroups of 

 plectognaths. 



In summary, there is good evidence that the HoUardi- 

 inae are ancestral to the Triacanthidae and the Eoplec- 

 tinae to the Tetraodontoidei, while probable that the 

 Triacanthodinae are derived from the hollardiins and 

 possible that the Spinacanthinae were an evolutionary 

 dead end. 



In fact, the Eoplectinae are so obviously ancestral to 

 the Gymnodontes that the eoplectins could be con- 

 sidered as the most basal family of that suborder, even 

 more generalized than the Triodontidae, rather than as a 

 subfamily of the Triacanthodidae with highly specialized 

 jaws. Similarly, the Protacanthodinae are so obviously 

 ancestral to the Recent Triacanthidae and intermediate 

 between them and the Triacanthodidae that the 

 protacanthodins could be considered a specialized sub- 

 family of the Triacanthodidae rather than as the most 

 generalized subfamily of the Triacanthidae. It is on the 

 basis of my own personal assessment of the overall 

 similarities and differences between most triacan- 

 thodids, Eoplectus, Zignoichthys, and triodontids and 

 between triacanthodids, Protacanthodes, and most 

 triacanthids that I subjectively feel that Eoplectus and 

 Zignoichthys are best placed in the Triacanthodidae and 

 Protacanthodes in the Triacanthidae. 



