series over the upper edge of the orbit. In most species of 

 the subgenus CycUchthys the third spine has its exter- 

 nal process placed distinctly behind the eye, but in orbi- 

 cularis it is placed just above the rear margin of the eye, 

 while in the subgenus Chilumycterus, only affinis has the 

 external process just above the rear margin of the eye, 

 while in atinga and tigrinus it is placed just behind the 

 rear margin of the eye and in reticulatus distinctly 

 behind the rear margin of the eye. 



Some of the difficulties in this arrangement of genera 

 are that: 1) Diodon contains species such as jaculiferus 

 in which most of the spines have massive triradiate bases 

 and only a few just behind the pectoral fin base are es- 

 pecially elongate and erectile, the bases of these spines 

 being only biradiate, or with a triradiate base in which 

 one of the radiations (that normally directed anteriorly 

 when there is a triradiate base and nonerectile spine) is 

 shortened and directed strongly dorsally or ventrally 

 toward one of the two well-developed radiations; 2) Dio- 

 don contains species such as calori in which only the 

 spines on the forehead are two rooted and erectile; 

 3) those species of Diodon such as hystrix and holocan- 

 thus that have all or most of the spines essentially two 

 rooted and erectile in young stages have most of the 

 spines nonerectile in large adults, the base of the spine 

 tending to become massive and triradiate with the devel- 

 opment of an anteriorly directed root; 4) in Diodon at 

 least two species, jaculiferus and calori, sometimes do 

 not have two nostrils, the septum between the nostrils in 

 a minority of specimens (about lO'^^c in jaculiferus) ap- 

 parently being resorbed (and not damaged by tearing); 



5) in Dicotylichthys, just as in Diodon, there are species 

 such as pilatus and punctulatus in which most of the 

 spines are triradiate and nonerectile, those that are es- 

 sentially biradiate and erectile being, respectively, just 

 behind the pectoral fin base and on the head and belly; 



6) in Dicotylichthys at least one species, nicthemerus, 

 usually has a nasal tube with two nostrils in young spec- 

 imens, the partition between the two nostrils apparently 

 usually being resorbed in adults to produce a bilobed 

 tube or tentacle, and Le Danois (1959:227) reported a 

 specimen with both conditions, one on either side; 7) in 

 Chilomycterus there are two types of nasal apparatuses, 

 most species having two nostrils and a relatively smooth 

 or only slightly folded epithelium on the inner surface of 

 the tube, but three species, reticulatus, tigrinus, and af- 

 finis having the nasal apparatus as an open low cup (i.e., 

 one large nostril) with a pitted epithelium, perhaps at all 

 sizes, and another species, atinga, in which the nasal 

 apparatus is tubular with two nostrils and a relatively 

 smooth inner epithelium in young specimens, but with 

 the great majority of large specimens having converted 

 this into an open cup with a pitted surface by the resorp- 

 tion of the partition originally separating the two nos- 

 trils, the remains of the partition apparent even in large 

 adults as an unpitted ridge across the middle of the 

 pitted cup. 



Le Danois (1959) recognized Diodon for those species 

 with either two nostrils or a bifid tentacle and with at 

 least some of the spines two rooted and erectile, this be- 



ing the combination of Fraser-Brunner's (1943) Diodon 

 and Dicotylichthys, while Fraser-Brunner's Chilomyc- 

 terus, containing all the species in which all the spines 

 are three rooted and nonerectile, in Le Danois' system is 

 split into Chilomycterus for reticulatus (and tigrinus), 

 with the nasal apparatus an open cup with a pitted sur- 

 face, and Atinga for all the other species, supposedly all 

 with two nostrils and an unpitted inner epithelium. Le 

 Danois' system has all of the drawbacks of Fraser- 

 Brunner's in that Diodon contains species with a cover- 

 ing of almost as fully nonerectile, triradiate spines as in 

 Fraser-Brunner's Chilomycterus, and of two different 

 types of nasal tentacles, while Le Danois' Chilomycterus 

 contains a single species (she considered tigrinus and 

 reticulatus synonymous) with a pitted cup nasal appa- 

 ratus (as well as one poorly preserved Miocene species, 

 acanthodes, which seems to have had some two-rooted 

 erectile spines and an unknown nasal apparatus, and 

 would thus be assignable by her own system to Diodon) 

 and her Atinga contains species having a nasal tube with 

 two nostrils except for two species, affinis, which has a 

 pitted cup at all stages as far as known, and atinga, in 

 which adults develop a pitted cup nasal apparatus exact- 

 ly as found in Chilomycterus reticulatus, tigrinus, and 

 affinus. 



Neither of the above two systems are tenable, but 

 internal characters do little to clarify the situation, even 

 though a large number of species were examined for this 

 work. For purposes of discussion, the generic categories 

 of Fraser-Brunner are used below. The nomenclature and 

 diagnostic characters of many of the species of Chilo- 

 mycterus are confused, and one or two of the names used 

 here may be synonymous. For example, C. reticulatus, 

 the type-species of Chilomycterus, is known only from 

 very large adults (300 to over 600 mm), and C. tigrinus 

 may be the younger adult of it, since both are extremely 

 similar osteologically. They differ mainly in tigrinus hav- 

 ing fewer but larger dark spots and few if any larger dark 

 blotches. 



Before passing on to the osteological diversity of the 

 diodontids, one additional external feature is of syste- 

 matic significance. As outlined by Tyler (1970b:25), the 

 great majority of diodontids have 9 caudal fin rays, while 

 a few species of Chilomycterus (sensu Fraser-Brunner) 

 have 10. The four species presently known with 10 rays 

 are affinis, atinga, reticulatus, and tigrinus, all four of 

 which have the nasal apparatus as a pitted cup, at least 

 in adults. Since the ancestral tetraodontids and at least 

 some of the Eocene diodontids have 11 caudal rays, the 

 four species of Chilomycterus with 10 caudal rays are 

 more generalized in this respect than are the other spe- 

 cies of diodontids, all of which have the caudal rays fur- 

 ther reduced to 9, at least modally. Conversely, these 

 four species are more specialized than other diodontids in 

 their nasal apparatus, for at least the more generalized of 

 the ancestral tetraodontids, as well as the triodontids 

 and triacanthodids, have two nostrils to each side. 



One unusual internal feature is shared by the four spe- 

 cies of Chilomycterus with the combination of 10 caudal 

 rays, no spines on the forehead, and a pitted cup nasal 



