the sixth absent in one of the molids. There are a num- 

 ber of diagnostically important differences given here in 

 the manner by which the branchiostegal rays are articu- 

 lated to the ceratohyal and in the form of the modified 

 first branchiostegal ray of many gymnodonts. 



Thilo (1899b, 1914) drew attention to the obvious cor- 

 relation between the presence of an inflation mechanism 

 in many gymnodonts and the enlargement of the first 

 branchiostegal ray, and described the pumping action of 

 that platelike element. Thilo (1914) contended that when 

 the distensible diverticulum of the oesophagus is not 

 well developed (as in Fugu rubripes and Lagocephalus 

 scleratus, according to Thilo) the first branchiostegal ray 

 is as small as it is in most balistoids. Thilo is guilty of 

 exaggeration, for the first branchiostegal ray in those two 

 species is much larger than it is in any balistoid, even 

 though it is somewhat reduced in size from that seen in 

 more typical gymnodonts. 



There are always three basibranchials (except in the 

 monacanthid Psilocephalus with two), three pairs of 

 hypobranchials, five pairs of ceratobranchials, and four 

 pairs of epibranchials (except in Psilocephalus with 

 three), but the number of pharyngobranchials is variable 

 and of much diagnostic interest. There are always at 

 least two pharyngobranchials, those of the second and 

 third arches, except that in diodontids they are of the 

 first and second arches. The pharyngobranchial of the 

 first arch is absent in monacanthids, molids, and some 

 ostraciids, while that of the fourth arch is absent in 

 balistids, monacanthids, ostraciids, tetraodontids, and 

 diodontids and that of the third arch absent only in 

 diodontids. The dentition of the pharyngobranchials is 

 also of great diagnostic use, for teeth are present on the 

 pharyngobranchial of the first arch in diodontids and 

 some tetraodontids, on that of the second arch in all but 

 a few aracanids, on that of the third arch in all but some 

 ostraciids and diodontids, and on that of the fourth in 

 triacanthodids, triacanthids, aracanids, triodontids, and 

 molids. The fifth ceratobranchial is toothed in triacan- 

 thodids, triacanthids, balistids, triodontids, and, with 

 minute teeth, in a few diodontids. 



In Canthigaster the first three ceratobranchials are 

 more compressed and deeper than in other tetraodon- 

 tids, while in molids the ceratobranchials are so deep 

 that in Mola they have been called (Barnard 1935:657) 

 "knife-like bones." These support the enormous gills of 

 molids, which were well illustrated as long ago as Ales- 

 sandrini (1839). The first pharyngobranchial in triacan- 

 thoids, balistoids, and triodontids is usually a toothless, 

 often rodlike, suspensory element. 



Grieb (1935) has said that teeth are present on the 

 lower pharyngeals of "Spheroides sp.," but that has not 

 been found to be the case in any tetraodontids examined 

 here. The reader is referred to Al-Hussaini (1947) for an 

 interesting discussion of the correlation between diet, 

 length of intestine, and development of pharyngeal teeth 

 in a number of plectognaths. Iwai (1964) has carefully 

 surveyed the taste buds on the gill rakers and gill arches 

 of Fugu and Rudarius. 



In the pectoral girdle the position of the supracleith- 



rum varies from more or less vertical in scleroderms to being 

 obliquely placed in relation to the skull in triodontids 

 and tetraodontids, while in diodontids and molids it is 

 almost horizontal to the skull. Regan (1903a) used the 

 position of the supracleithrum as one of the distinguish- 

 ing features between the Sclerodermi and Gym- 

 nodontes. 



It was probably Gill (1872, 1885) who made best known 

 that there is some rudiment of a posttemporal present 

 between the supracleithrum and pterotic in at least some 

 plectognaths, and much the same was said by Regan 

 (1903a:285), Rosen (1916a:21), and Berg (1940:495). 

 Even though relatively small and closely sutured and in- 

 terdigitated to the skull, careful examination reveals its 

 presence in all scleroderms except one monacanthid, and 

 its absence in all gymnodonts. 



Siebenrock's (1901) descriptions of the pectoral girdle 

 in a number of plectognaths are so erroneous that they 

 are useless, and Regan (1903a:291) called attention to the 

 incorrectness of Seibenrock's contention that the supra- 

 cleithrum of Mola was really a posttemporal. Sjirensen's 

 (1883) mainly myological description of the pectoral 

 region of Tetraodon is highly detailed. 



Both the dorsal and the ventral postcleithra are pres- 

 ent in most plectognaths as distinct elements, hut in 

 monacanthids, diodontids, and molids the two elements 

 tend to indistinguishably fuse. Fraser-Brunner (1941c:307) 

 used size of the postcleithra as one characteristic to distin- 

 guish aracanids from ostraciids. The suture between the dor- 

 sal and ventral postcleithra is difficult to trace in many 

 ostracioids, and it is possible that the two elements oc- 

 casionally fuse together. The anteriorly directed process, 

 which lies just lateral to the actinosts, of the fused post- 

 cleithra of molids is a diagnostic feature of that group. 

 Kuronuma (1943) has used the shape of the ventral post- 

 cleithrum in tetraodontids as a systematic character and 

 Parr (1927) gave an interesting discussion of the func- 

 tion of the postcleithra in gymnodonts, while Winterbot- 

 tom (1971) has thoroughly surveyed the matter. Thilo 

 (1899b) was surely in error in saying that the post- 

 cleithra could be moved through an arc of 90° . The role 

 played by the postcleithra and cleithra in sound produc- 

 tion has been discussed by Mobius (1889), Cunningham 

 (1910:117), and Fish (1954:69). For a general survey of 

 the endochondral bones of the plectognath pectoral gir- 

 dle the reader is referred to Starks' (1930:206-210) excel- 

 lent work on that subject. Sound production by pectoral 

 fin drumming is reviewed by Salmon et al. (1968), and by 

 this and other methods by Schneider (1961). 



The scapula completely encloses the scapular foramen 

 in most scleroderms and in Triodon, but in tetraodontids 

 and diodontids the foramen is incomplete, being enclosed 

 by both the scapula and the cleithrum. In molids the 

 scapular foramen is essentially absent or highly incom- 

 plete, being represented only by a small hole in the sheet 

 of fibrous tissue holding the bottom half of the scapula to 

 the cleithrum. In scleroderms and in Triodon the dorsal 

 edge of the scapula always bears an emargination to 

 which is attached the first pectoral fin ray, but in the 

 other plectognaths no such emargination is present. Four 



