been placed in either the Sclerodermi or Gymnodontes, 

 and thus had always had a subordinate position until 

 Goodrich gave it equal ranking with the two Cuvierian 

 divisions. However, Goodrich's (1909:439) remarks on 

 why Triodon had been so elevated were only that this is 

 "a family intermediate between the first and third sub- 

 groups, whose exact position it is difficult to deter- 

 mine." Thus, the new third category was only a matter of 

 convenience. In his (1930) "Studies of the Structure and 

 Development of Vertebrates," Goodrich followed exactly 

 the same classification. 



The handling of the acanthurid-plectognath line in 

 Gregory's (1933) "Fish Skulls" was modified from that of 

 Goodrich. Gregory recognized a Balistoidei as one of the 

 23 subdivisions of the Acanthopterygii. The Balistoidei 

 contained five large assemblages: acanthurids, zan- 

 clids, siganids, teuthids, and plectognaths (Gregory 

 hesitated to give any of them a precise rank). His 

 Balistoidei was therefore equal to the expanded "Plec- 

 tognathi" (or Division B of the Subtribe Chaetodonti- 

 formes) of (Goodrich. Goodrich's "Squammipennes" 

 (Division A of the Subtribe Chaetodontiformes) became 

 the Chaetodontoidei of Gregory, a group equal in rank 

 with the Balistoidei. Gregory felt that the chaetodon- 

 tids, acanthurids, and plectognaths formed a natural 

 series, but he split apart Goodrich's Chaetodontiformes 

 because it was his opinion (1933:281) "that the cleft 

 between the typical chaetodonts and acanthurids is 

 much greater than that between the latter and the plec- 

 tognath stem as represented by the balistids." Just as 

 with Dareste, one wonders why the comparison was not 

 made between acanthurids and triacanthoids, which are 

 the most generalized of the plectognaths, rather than 

 between acanthurids and balistids, which are obviously of 

 triacanthoid derivation. Gregory even illustrated the 

 skull of a triacanthid (1933:282, fig. 160), but no men- 

 tion at all was made of it in the text. Highly useful as 

 Gregory's plectognath figures (1933:figs. 160-172) are for 

 obtaining a general idea of cranial construction, they 

 lack detail. 



The tentative classification as seen in Gregory's (1933) 

 "Fish Skulls" was a great change from that Gregory 

 (1907) used in his much earlier "Orders of Teleostomous 

 Fishes." In the earlier work Gregory recognized the Plec- 

 tognathi as one of the 10 orders of the Superorder 

 Acanthopteroidei. The Plectognathi were subdivided in 

 the same way as Regan had done, with the exception that 

 the monacanthids were recognized as two separate 

 families (with the Alutera-like species as a distinct fami- 

 ly) rather than being placed in the Balistidae. One other 

 paper should be mentioned in connection with Gregory. 

 Gregory and Raven (1934) discussed the anatomy and 

 relationships of Mola and supported previous workers' 

 placement of the molids in close association with the 

 diodontids. 



During the period 1912 to 1916 a number of papers of 

 great importance to the study of plectognaths were pub- 

 lished by Rosen in Sweden and Kaschkaroff in Russia. 

 Rosen produced five successive articles in his "Studies on 

 the Plectognaths" which dealt, respectively, with the 



anatomy of the blood-vascular system (1912), the air- 

 sac and intestine (1913a), the integument (1913b), the 

 body muscles (1913c), and the skeleton (1916a). In the 

 introduction to the series, Rosen (1912:1) explained his 

 purpose to be the study of a number of anatomical sys- 

 tems, since "to base the building up of a phylogenetical 

 tree solely on one anatomical system is a great error 

 which far too many anatomists commit." He then 

 promised to "end this series of studies on the Plectog- 

 naths with a section in which all the phylogenetical 

 results attained independently from the study of each or- 

 gan will be discussed together" (1912a:2). Unfortunately, 

 such a last summary article on phylogeny never ap- 

 peared, the last of the series being the 1916 paper on the 

 skeleton, in which almost no phylogenetic conclusions 

 were drawn. One can go to his earlier papers for his ten- 

 tative views on the phylogeny of plectognaths, but he cer- 

 tainly would not have drawn his final conclusions until 

 after the osteological paper had appeared. The ten- 

 tativeness of the conclusions that will be quoted from his 

 earlier work should be stressed, for the series is a good 

 anatomical survey of the plectognaths, but his phylo- 

 genetic remarks based on the blood-vascular, intestinal, 

 and integumentary systems do not seem to me to be 

 substantiated by the osteology of the order. Rosen's ideas 

 were most clearly expressed in his paper on the air-sac 

 and intestine (1913a:19) in which he stated that, "The 

 Molids and the Ostraciontids, which possess no such sac, 

 must have developed independently of the other 

 families. The inquiry given above on the air-sac has 

 clearly shown that we have every reason to believe that 

 the families Triacanthidae, Balistidae (including the 

 genus Monacanthus), Diodontidae, and Tetrodontidae 

 form a continuous series of development. This is an evi- 

 dent proof against the present classification of the Plec- 

 tognaths into two subgroups, Sclerodermi and Gym- 

 nodontes, as proposed by Regan." Rosen's work on the 

 integument confirmed his opinions and he (1913c:25-26) 

 added that "The Molids and Ostraciontids are primitive 

 Plectognaths, both groups separately specialized." The 

 fourth paper in the series, like the fifth, contained no 

 phylogenetic remarks. 



Regardless of how the above statements would have 

 been modified in a summarizing article, Rosen's series 

 was highly useful as a pioneering survey of the soft, as 

 well as the osteological, anatomy of the plectognaths. His 

 first four papers are presently the only publications that 

 even attempt to treat as a whole all of the main non- 

 osteological systems of the plectognaths, although Win- 

 terbottom's (1974) superb myological analysis is a major 

 step in that direction. Rosen added much new infor- 

 mation and reviewed the then existing knowledge. 

 Without going into details, a few of the leading struc- 

 tural peculiarities summarized by Rosen in his first 

 paper of the series should be mentioned as examples of 

 the type of information with which he dealt. He was par- 

 ticularly impressed by the fact, known since the time of 

 Cuvier, that Mola has a larger, and supposedly rather 

 primitive, number of valves in its conus arteriosus and 

 atrioventricular region than do the other plectognaths, 



