basic subgroups within the two families. His families are 

 now of subordinal rank, and his genera have become the 

 bases for most of the families now recognized. 



In only one group did Cuvier miss an important major 

 distinction, i.e., the ostracioids can be divided easily into 

 ostraciids and aracanids, but Cuvier could not have been 

 very familiar with the aracanids, for these mostly Japan- 

 ese and Australian deepwater forms were rarely collected 

 at that time and known primarily from a single species 

 (aurita, Shaw 1798 illustration and 1804 description). Of 

 the 10 families of plectognaths presently recognized, only 

 the triacanthodids and triodontids were unmentioned by 

 Cuvier, and they were not to be discovered for another 

 decade or more. Cuvier's description of Reinwardt's 

 specimen of the unique Triodon bursarius (= mac- 

 ropterus) appeared in the second edition of Cuvier's "Le 

 Regne Animal" (1829:370) as the fourth genus of the 

 family Gymnodontes, while the first triacanthodid was 

 described by Schlegel (1850). In summary, most of the 

 basic groupings of plectognaths were made known by 

 Cuvier, with the notable exception of the two sub- 

 divisions of the ostracioids. At about the same time that 

 his (1817) first edition of "Le Regne Animal" appeared, 

 Cuvier (1818) published a short paper on Diodon, de- 

 scribing some new species, but, more importantly, cor- 

 recting a number of anatomically erroneous statements 

 about diodontids made by such workers as Broussonet, 

 Plumier, and Bloch. 



The appearance of the classifications of Latreille 

 (1825) and Risso (1826) shortly after Cuvier's (1817) 

 "Le Regne Animal" set the precedent for the thereafter 

 nearly unanimous ordinal recognition of the Plectog- 

 nathi. Risso followed Cuvier rather closely, and as far as 

 the plectognaths are concerned, added nothing to their 

 classification. In his (1825) "Families Naturelles," 

 Latreille used Cuvier's definition of the Ordre Plectog- 

 nathes, but placed them in a different relationship with 

 other orders (with the sturgeons and lophobranchs). 



At about this time Geoffroy Saint-Hilaire (1827) pub- 

 lished his "Poissons du Nil." His description (p. 176-214, 

 pis. 1-2) of Tetrodon physa (= lineatus Linnaeus), the 

 common pufferfish of the Nile, was the finest descrip- 

 tion of a plectognath that had appeared up to that time, 

 describing the general anatomy, habits, distribution, and 

 nomenclature of the fish and gently setting aside the er- 

 rors made by Bloch and by Lacepede on the inflation 

 mechanism, showing the sac to be a diverticulum of the 

 oesophagus and describing its general structure and mus- 

 culature. 



The year 1833 saw the begiiming of the publication of 

 Agassiz's "Recherches sur les Poissons Fossiles" [vols. 1 

 (1844a) and 2 (1833, 1842, 1844b)], which was to make a 

 major (or at least temporarily so) change in the 

 classification of nearly all groups of fishes. Based on the 

 configuration and composition of the scales, Agassiz 

 hoped to establish a more natural grouping of orders than 

 had previously been available. Agassiz's system as a 

 whole did not stand the test of time, and his association 

 of the plectognaths with the ganoids was a major error. 

 His Ordre des Ganoides (1844a: 169) contained six 



groups: Acipenserides, Siluroides, Lophobranches, 

 Gymnodontes, Sclerodermes, and Sauroides. 



To Agassiz the plectognaths were ganoides for various 

 reasons. The scales of Ostracion were described (Agassiz 

 1844a:75) as consisting of a homy substance deposited in 

 strata and covered with a thick layer of dentine charac- 

 terized by ramifying calcareous tubes like those of the 

 teeth. The major references to plectognaths, however, 

 were contained in a section (1844b:248-267) entitled "De 

 la famille des Gymnodontes" where brief osteological de- 

 scriptions of a Tetraodon and a Diodon were given, along 

 with a comparison of the structure of the teeth and spines 

 of these two gymnodonts. Agassiz found the gymnodont 

 spines to be similar to the scales of scleroderms, in that 

 both possessed a dentine layer with calcareous tubes, 

 while the teeth of gymnodonts were said to be similar to 

 the teeth of sharks. Agassiz (1857) also published a brief 

 note on the structure of Mola. 



Agassiz (1833:1-2) listed the six families of his Ordre 

 Ganoides as: Famille Lepidoides (with Acanthodes, 

 Palaeoniscus, Osteolepis, Lepidotus, etc.); Famille 

 Sauroides (with Leptolepis, Sauropsis, etc.); Famille 

 Pycnodontes (with Placodus, Pycnodus, Microdon, etc.); 

 Famille Sclerodermes; Famille Gymnodontes; and 

 Famille Lophobranches (with Syngnathus, etc.). It is 

 strange to see the plectognaths associated in the same 

 order with such forms as the placoderm Acanthodes, the 

 crossopterygian Osteolepis, and the isospondyl Lep- 

 tolepis. Even stranger, however, was the assemblage in- 

 cluded in the Sclerodermes (1844b: 248- 267). Not only 

 were Balistes, Ostracion, the Balistes-like fossil Acan- 

 thoderma, and the Triacanthus-Vike fossil Acan- 

 thopleurus placed there, but also included in the 

 Sclerodermes were such obviously nonplectognath fos- 

 sils as Blochius (which Woodward, 1901:591, referred to 

 the Blenniiformes), Dercetis (which Woodward, p. 171, 

 referred to the Isospondyli), and Rhinellus (which Wood- 

 ward, p. 266, referred to the Isospondyli). Agassiz rec- 

 ognized in later years that his classification was highly 

 artificial (see Agassiz 1860). 



With the appearance of Miiller's (1844) "Ueber den 

 Bau und die Grenzen der Ganoiden," the Ganoides of 

 Agassiz were modified to a considerable extent. Miiller 

 purged the Ganoides of the plectognaths, lophobranchs, 

 and most of the other living species placed there by Agas- 

 siz. Muller pointed out that the true ganoids had, among 

 other characteristics, more than two valves in the conus 

 arteriosus, while the teleosts had only two. Among the 

 supposed ganoids of Agassiz that Muller had examined 

 for this characteristic were Balistes, Ostracion, and Tet- 

 raodon. The classification at which Muller arrived 

 (1844:85-88) recognized six subclasses of Pisces, with the 

 Subclass Teleostei containing the Order Plectognathi 

 and its three families: Balistini, Ostraciones, and Gym- 

 nodontes. 



Muller pointed out (1844:6) that whereas the true 

 ganoids are physostomes with abdominal pelvics, the 

 Plectognathi are physoclists and that when obvious pel- 

 vic fins are present, as in the triacanthoids, they are not 

 abdominal. He also disagreed (p. 7) with Agassiz's con- 



