counted in any of them, although they presumedly num- 

 bered eight as in the Eocene Protacanthodes, the Oligo- 

 cene Cryptobalistes, and the Recent species of the family. 



On the basis of the configuration o( Acanthopleurus, 

 the Triacanthinae have changed very little from at least 

 the Oligocene. 



The other fossil species of Triacanthinae is the very 

 poorly known Marosichthys huismani (de Beaufort 

 1926), based on a single very incomplete specimen from 

 the Miocene of the Celebes comprising the head and up- 

 per anterior part of the body. The dorsal fin spines are 

 better developed than in Acanthopleurus, and the body 

 deeper, but Marosichthys can be considered to be at the 

 same level of organization as Acanthopleurus, as discuss- 

 ed by Tyler (1968:241-242). 



The Oligocene Cryptobalistes brevis (Rath 1859), the 

 only representative of the Cryptobalistinae, is known 

 from a single specimen (perhaps two, see Winterbottom 

 1974:96) from the same black schist of Canton Glarus, 

 Switzerland, in which the apparently thoroughly modem 

 Acanthopleurus is found. The specimen of Cryptoba- 

 listes was not reexamined for this work, for the holotype 

 cannot be located, and the following very briefly sum- 

 marizes the long phylogenetic discussion of the species 

 presented by Tyler (1968:243-249). 



Cryptobalistes is an enigma. In general appearance it 

 is rather balistoid, although it has (probably) 20 rather 

 than 18 vertebrae, 12 of which are caudal, at least four 

 dorsal spines and no locking mechanism of the first by 

 the second, no supraneural strut bracing the last spiny dor- 

 sal fin basal pterygiophore, a large pair of pelvic spines, 

 and a relatively wide, dorsally concave, basinlike pelvis. 



Although Cryptobalistes is in most ways an excellent 

 anatomical intermediate between the triacanthids and 

 balistids, the shape of its pelvis negates the other 

 evidence, and the balistoid appearance of Cryptobalistes 

 is presumed to be due to parallelism. Cryptobalistes 

 represents an extinct lineage, for in the same Oligocene 

 strata in which it is found there are also thoroughly 

 modem balistids (Balistomorphus) not much different 

 from genera alive today, and these Oligocene balistids 

 already posses a stout shaftlike pelvis with a dorsal 

 lobe, three dorsal fin spines with a locking mechanism, 

 a carina with a supraneural supporting strut, and no 

 pelvic spines. While Cryptobalistes represents an 

 evolutionary dead end, it greatly increases the known 

 anatomical diversity of the Triacanthidae and proves 

 that the triacanthids were capable to giving rise to 

 balistidlike forms. 



Generic relationships. — As discussed by Tyler 

 (1968:250), the four Recent genera fall naturally into two 

 groups on the basis of the structure of the posterior shaft 

 of the pelvis, Triacanthus and Trixiphichthys having a 

 relatively untapered pelvis and Pseudotriacanthus and 

 Tripodichthys a distinctly tapering pelvis ending in a 

 point. On the meager evidence of the better development 

 of the second dorsal spine and shorter anal fin base in 

 Pseudotriacanthus, conditions closer to those of the 

 ancestral triacanthodids, Pseudotriacanthus can be con- 

 sidered slightly more generalized than Tripodichthys. On 

 the basis mainly of the specialized reduced dentition and 

 of the elongate and narrow snout, Trixiphichthys is less 

 generalized than Triacanthus. The relationship between 

 the two more generalized genera in the two basic subdivi- 

 sions of the Recent species is not clear, but both are cer- 

 tainly closely related to the Oligocene Acanthopleurus, 

 with the Miocene Marosichthys of unknown relationship 

 to Acanthopleurus. The Oligocene Cryptobalistes is of 

 uncertain relationship with the other triacanthids, while 

 the Eocene Protacanthodes is clearly on the line directly 

 intermediate between the ancestral triacanthodids and 

 the derived triacanthids, as discussed in the preceding 

 section on the relationships of the Triacanthodidae. 



The relationship of the Triacanthidae to the Balistidae 

 is discussed under the latter, with the conclusion being 

 that Protacanthodes is not far removed from the line of 

 early triacanthids that gave rise to the balistids and 

 aracanids, and thus, respectively, to the monacanthids 

 and ostraciids as well. 



In spite of its name, Protriacanthus gortanii d'Erasmo 

 (1946:116), from the upper Cretaceous, is not a plec- 

 tognath (see Patterson 1964:429-432). 



Figure il .—Marosichthys huismani: lateral 

 view of anterior part of bod.v of holot.vpe, 

 ca. 36 mm head length, Miocene of the Celebes 

 (de Beaufort 1926; fig. 5 of pi. 5). 



