INTEODUCTION. XI 



physoclists with abdominal ventrals (Catosteomi and Percesoces) is, I feel, much 

 in need of revision, and it may be found advisable to break up this group [Catosteomi] 

 into a greater number of sub-orders." 



I have paid a good deal of attention to these groups, and am of opinion that they 

 are unnatural and indefinable. 



After removal of the Selenichthyes, which I have recently been able to show are 

 related to the Tasniosomi, and of the Hypostomides, the remainder of the Catosteomi, 

 which corresponds to the Hemibranchii of Smith Woodward, is still a heterogeneous 

 assemblage which I find incapable of definition, and includes three well-marked but 

 probably related groups which should, in my opiuion, be given sub-ordinal rank. 

 These are : (1) Thoracostei *, which have on each side a large dermal plate, which 

 in the adult is co-ossified with the coracoid and suturally united to the clavicle ; 

 (2) SoLENiCHTHYES (nom, nov.), which have a considerable amount of dermal armour 

 but no plates similar to the ectocoracoids of the Thoracostei — the tubiform snout, 

 terminal toothless mouth, pectinate gills, and elongate anterior vertebrae with separate 

 transverse processes further characterize this group ; (3) Lophobeais^chii, difi'ering 

 , from the Solenichthyes in the lobate gills and normal anterior vertebra3. 



The resemblances between the Centriscidae and the Solenostomidse, respectively 

 the most generalized of the Solenichthyes and the Lophobranchii, seem to indicate 

 relationship, but are, no doubt, in great part due to similar modes of life. 



A few fishes belonging to the sub-order Lophobranchii are found in the rivers 

 of Mexico and Central America. 



Adding the OsphromenidcE, which should, in my opinion, be placed with the 

 Anabantidse, and after removing the Scombresocidas to the Haplomi, the Ammodytidte 

 and Champsodontidae to the Percomorphi, and the Chiasmodontidae {incertce. seclis), the 

 families included by Boulenger in the Percesoces may be arranged thus : — 



I. Pelvic bones remote from tlje clavicles; a separate spinous dorsal ; no supra- 

 branchial organ ; no cesophageal sacs AtJieriiiidce, MuyiUdm, 



Folynemidce, and ^phyrcenidm. 

 II. A suprabranchial organ OpMoceplialidcE, Anabantidce, and Osphromenidx. 



III. (Esophagus with a pair of lateral muscular sacs, with internal papilla which 



may be toothed Tetrayonurida, Stromateidoi. 



IV. Pelvic bones remote from the clavicles ; no fin-spnies ; no suprabranchial 



organ; no oesophageal sacs Icosteido'. 



In the second and third of these divisions we see the transition from abdominal 



* Swinnerton, Quart. Journ. Micr. Soi. xlv. 1902, p. 5S0, and xlix. 1905, pp. 369-380. 



h -1 



