crushing beak of diodontids in his description of a species 

 of Diodon: "the true teeth appear on the surface of the 

 jaws like minute scales, as in several species of the genus 

 Scarus." 



It will be recalled that Regan (1903a) placed the genus 

 Triodon in the Sclerodermi, and, as a consequence of 

 this, it was necessary for him to state (p. 285) that "the 

 coalescense of the teeth in the jaws is a feature of little 

 importance, and has, as probably as not, originated in- 

 dependently in . . . [the Sclerodermi] . . . and in the 

 Gymnodontes." That view is not followed here, as the 

 placement of Triodon in the Gymnodontes indicates. 



For a discussion of the surface structure of plectog- 

 nath teeth in relation to sound production by stridula- 

 tion the reader is referred to Burkenroad (1931:22-24), 

 Fish et al. (1952:189-190), Fish (1954:62-77), Moulton 

 (1958:359-362), and Vincent (1963). 



The two dentaries are indistinguishably fused with one 

 another in Triodon, diodontids, and molids. The fibrous 

 tissue articulation of the two dentaries in tetraodontids is 

 strengthened by emarginations from the medial edges of 

 the two bones, just as is the case with the premaxil- 

 laries. 



In triacanthodids and molids the articular and den- 

 tary are firmly held to one another mainly by fibrous tis- 

 sue, but in other plectognaths there is normally an ex- 

 tensive interdigitation between the surfaces of these two 

 bones. It is not true, however, that the dentary and ar- 

 ticular are fused into a single piece, as was maintained 

 by Gill (1872:XL; 1885:412) for plectognaths in general, 

 and as Smith (1935:359) gave as a characteristic of the 

 balistoids. Regan (1903a:286) correctly stated that the 

 dentary and articular are not fused in the gymnodonts, 

 but, rather unaccountably, he agreed with Gill that these 

 two bones are fused in the scleroderms. In the descrip- 

 tions of the representative plectognaths given here no at- 

 tempt is made to differentiate between a dermal and an 

 endochondral portion of the articular, for this requires 

 histological study of the developing bone for consistant 

 accuracy. Only two papers in the literature treat this 

 subject in plectognaths. Rosen (1916a: 19) simply said, 

 without any elaboration, that in Sphoeroides the "ar- 

 ticulare develops as an autarticulare and a dermar- 

 ticulare." Haines (1937:14-21) gave a detailed descrip- 

 tion of the growth and fate of Meckel's cartilage and the 

 surrounding ossifications in Tetraodon, and concluded 

 that the endochondral portion of the articular was either 

 greatly reduced or absent. The sesamoid articular is pres- 

 ent in all plectognaths, except some tetraodontids and 

 molids. It has been described in a number of different 

 plectognath genera by Starks (1916:32-33), who said that 

 in Sphoeroides annulatus "the sesamoid articular more 

 evidently originates within the tendon than in any other 

 examples I have encountered." 



With the exception of Rumania, which has lost the 

 sixth branchiostegal ray, all gymnodonts have six such 

 rays, and statements in the literature giving a lesser 

 number probably reflect the loss of a ray during dissec- 

 tion, while Hubbs (1919:69) pointed out that "in Tetrao- 

 don the uppermost ray basally is an unossified liga- 



ment," which is true also of some other tetraodontids 

 where the sixth ray tends to be an especially slender 

 shaft. McAllister (1968) accurately surveyed the num- 

 ber of branchiostegal rays and the structure of the hyoid 

 arch in a wide selection of plectognaths, far more so than 

 ever previously attempted, and found them to be related 

 to the acanthopterygian type. 



Wellenbergh (1840) gave an accurate description and 

 figure of the six branchiostegal rays in Mala, but Kasch- 

 karoff (1914a:282) and Van Dobben (1935) stated that 

 there are only five branchiostegal rays present in Mola. 

 Roon and Pelkwijk (1939) and Roon (1942) have again 

 called attention to the fact that there are six branchio- 

 stegal rays in Mola. Thilo (1899a, 1914) first pointed out 

 that species such as Stephanolepis setifer and 

 Brachaluteres trossulus possess only five branchiostegal 

 rays. In contrast to Ranzania, however, it is not the sixth 

 branchiostegal that is lost; rather, it is the second 

 branchiostegal. The branchiostegal counts given by 

 Willen (1941, 1942, 1944, 1945, 1947, 1949) are not 

 reliable. He said that there were six branchiostegals in 

 Triacanthus, Monacanthus, Ostracion and Mola, but that 

 there were seven in Batistes and five in Tetraodon and 

 Diodon. The supposed "seventh ray" of balistids was 

 neither figured nor well described, and it is only possible 

 to guess that Willem may have mistaken the rodlike in- 

 teroperculum for a seventh branchiostegal. 



The branchial region is one of the most neglected 

 anatomical systems of fishes, for cleared and stained 

 specimens and especially delicate dissections are re- 

 quired, although Nelson (1969 et seq.) has made great 

 strides in rectifying this situation. In plectognaths the 

 extensive variation in branchial structure reported here 

 (Table 1) forms an important part of the diagnoses of the 

 higher categories recognized and of the phylogenetic inter- 

 pretations offered, and yet the only previous survey of 

 the plectognath branchial apparatus is the cursory one 

 by Kaschkaroff( 1914a). There are, however, a few papers 

 containing descriptions of the branchial or hyoid arches 

 of particular genera, such as Awati and Bal (1933) for 

 Tetraodon, Borcea (1907) for Batistes, Steenstrup and 

 Liitken (1898) for Mola, Supino (1905) for Balistes and 

 Mola, and Wellenbergh (1840) for Mola. 



The condition of the plectognath branchial ap- 

 paratus, with comments on the literature, is sum- 

 marized below. A basihyal is always present but the dor- 

 sal hypohyal is absent in many tetraodontids and a few 

 diodontids. The hypohyals are enlarged in ostracioids, 

 and it was probably the size of the dorsal hypohyal that 

 lead Horshelmann (1866:23) to say that Ostracion 

 possessed a tongue while the other plectognaths did not. 



The ceratohyal and epihyal are always present, while 

 the interhyal is absent in diodontids and most tetra- 

 odontids. The urohyal is well developed only in triacan- 

 thoids and balistoids, although it is present in a much- 

 reduced state in ostracioids and in Triodon. The 

 branchiostegal rays usually are present in the 2-1-4 

 arrangement typical of percoid fishes (Hubbs 1919), al- 

 though the second and sixth branchiostegal rays are ab- 

 sent in a few species of monacanthids and aracanids, and 



