actinosts are present in all plectognaths, except in 

 molids, which have three. 



The close connection of the scapula with the first ac- 

 tinost is such that Regan (1903a:291) drew attention to 

 the fact that "On a superficial examination there ap- 

 pears to be no scapula, and the pectoral fin to be sup- 

 ported by a series of four enlarged pterygials. In fact, the 

 united upper pterygial and scapula together resemble 

 one of the enlarged pterygials." Even after this state- 

 ment, others (e.g., Kaschkaroff 1914a:355; Awati and Bal 

 1933:91) continued to misidentify the actinosts and 

 scapula. On the other hand, Regan's (1903a) statement 

 that the scleroderms have the "pterygials (pectoral 

 basalia) not enlarged, movably attached by ligament to 

 the scapula and coracoid, three to the former and one to 

 the latter" (p. 286) while the gymnodonts have the 

 "lower three pterygials enlarged and immovable united 

 to the coraco-scapula" (p. 291) is only partially true (see 

 above). In molids what was the first actinost of other 

 plectognaths is either lost or fused with the scapula, so 

 that only three actinosts are present. Somewhat under- 

 standably, a number of workers have mistaken the 

 scapula of Mola for an actinost and have concluded that 

 Mola possessed four actinosts and no scapula. However, 

 Gregory and Raven (1934:148) described Mola as having 

 four actinosts (i.e., three actinosts and a scapula), as well 

 as a "vestigial scapula" represented by a "bone which is 

 crowded between the expanded first pterygial and the 

 posterior recurved border of the cleithrum." It is impos- 

 sible to state what it was that Gregory and Raven saw, 

 for their description is brief and there is no illustration of 

 the vestigial scapula. 



The structure of the pelvis and pelvic fin in plectog- 

 naths has been the cause of much confusion in the 

 literature, for the relatively normal pelvic apparatus of 

 triacanthoids is replaced by a highly modified and 

 specialized structure in balistoids that has been often 

 misinterpreted. Tyler (1962b) has surveyed in detail the 

 reduction and eventual loss of the pelvic apparatus in 

 plectognaths, and these features are used prominently in 

 the diagnoses given here, while Winterbottom (1970) sur- 

 veyed the muscles of the triacanthoid pelvic fin. The 

 locking mechanism of the triacanthoid pelvic spines 

 (continuous series of positions of erection in triacan- 

 thodids and two positions in triacanthids) is also treated 

 by Tyler (1962b). HoUard (1853:107) and Thilo (1896a, 

 1896b:327, 1898, 1899a, 1900) incorrectly described only a 

 single position of erection in Triacanthus, but the two- 

 position mechanism has been correctly described by 

 Sorensen (1884:69, 1897) and Monod (1959c). 



The true nature of the rudimentary pelvic fin element 

 in balistids was first made known by Monod (1959a), in 

 excellent style for Balistes forcipatus. 



As detailed in the accounts of each family given here, the 

 number of vertebrae in plectognaths varies from 16 to 30 

 (Table 2) but is usually less than 24 and normally in the 

 range of 17 to 21, while the structure of the vertebral 

 column is highly modified in some groups, especially the 

 ostracioids (Tyler 1963a) and molids. In aracanids the 

 first two, and in ostraciids the first four or five, ab- 



dominal vertebrae are at least partially fused to the 

 skull, and the whole vertebral column is relatively in- 

 flexible except in the caudal peduncular region ex- 

 tending outside of the carapace. Another unusual feature 

 of the ostraciid vertebral column, first pointed out by 

 Fraser-Brunner (1941c:307), is that the course of the 

 haemal canal is displaced to one side or the other from 

 the midline, while Tyler (1963a) called attention to the 

 uniquely divergent positions away from the midline of 

 the anal fin basal pterygiophores. Abe (1942) presented 

 an extensive study of the variation in the form and num- 

 ber of elements of the tetraodontid vertebral column. 

 Ford (1937) compared the balistid and zeoid vertebral 

 columns. 



The fact that true or pleural ribs are found in the 

 monacanthid Pseudaluteres and in the gymnodont 

 Triodon negates Regan's (1903a:285) statement that "the 

 feature of most importance in diagnosing the suborder 

 Plectognathi is the absence of ribs." 



The sequence of vertebrae to which the epipleural 

 bones are attached is a character of systematic impor- 

 tance, as first shown by Fraser-Brunner (1941b: 176) in 

 his differentiation of the triacanthodids from the triacan- 

 thids and of the balistids from the monacanthids. The 

 epipleurals in some monacanthids tend to become swol- 

 len (hyperostotic) and closely applied to the transverse 

 processes of the vertebrae, and it is not surprising that 

 they have been misinterpreted (e.g., Smith 1935). The 

 development of the epipleurals has been studied in 

 several monacanthids by Goeppert (1895) and Goette 

 (1879). 



The structure and reduction in number of elements in 

 the caudal fin supports is documented by Tyler (1970b), 

 while comments are given by Pope (1945), Okada (1950), 

 Randall (1964), and Tyler (1970d) on abnormalities of 

 vertebral structures. The work of Abe (1949b) should be 

 consulted for a detailed survey of hypural fusion in tetra- 

 odontids. The description given by Whitehouse (1910) 

 was one of the earlier accurate accounts of a balistid 

 caudal skeleton. Monod (1968) accurately surveyed and 

 illustrated the ural structures of a wide variety of plec- 

 tognaths, and agreed with Dareste (1850) and Le Danois 

 (1955) that the triacanthoids and balistoids are not close- 

 ly related to the ostracioids and gymnodonts (the Or- 

 biculates). 



Gosline (1961:269) pointed out that "there seems to 

 have been a general trend toward fusion of parts in the 

 caudal skeleton of teleostean fishes that has occurred 

 repeatedly and independently in various lineages." The 

 Plectognathi are certainly a perfect example of this, since 

 there are several uroneurals, one epural, and six hy- 

 purals in the primitive triacanthodids, but the number of 

 elements becomes reduced in the triacanthid-balistid- 

 monacanthid series and becomes even more reduced in 

 the ostracioids. The same reduction takes place in- 

 dependently in the gymnodonts. Gosline said that the 

 caudal skeletons of Zanclus, Acanthurus, and the plec- 

 tognath fishes form a series, but Monod (1959b:728) has 

 said that the caudal skeleton of Acanthurus and Balistes 

 are "toutefois profondement diff6rente" and thus the 



