17'S Mr. G. A. Boulenger on the Suborders and 



III. No suborbital stay ; basis cranii simple ; scapula and coracoid 

 more or less reduced, sometimes vestigial ; pterygials large, only 

 one or two in contact with the coracoid ; ventral tins thoracic. 



V. Gobiiformes. 



IV. No suborbital stay ; basis cranii simple ; a perforate scapula ; 

 three pterygials in contact with the coracoid ; ventral fins thoracic ; 

 a suctorial transversely laminated disk on the upper surface of the 

 head VI. Discocephali. 



V. A suborbital stay, the second suborbital bone being more or less 



produced on the' cheek or joining the pra?operculum ; ventral tin3 

 thoracic VII. Scleroparei. 



VI. No suborbital stay ; ventral fins usually jugular or mental, or, if 

 thoracic, structure of the pectoral arch differing from that ascribed 

 to the first five divisions of this Synopsis. 



Pectoral fin with vertical or subvertical base ; 



anal fin usually elongate, rarely small VIII. Jugulares. 



Pectoral fin with horizontal or subhorizontal 



base ; body exceedingly compressed ; dorsal 



fin with all the rays simple ; anal fin absent 



or very small IX. TiENlOSOMl. 



Division I. PerciformeS. 



No bony stay for tlie proeoperculum. Basis cranii double. 

 Spinous dorsal usually well developed. None of the epi- 

 pleural bones attached to the centra of the vertebras in the 

 prafcaudal region. Pectoial arch with well-developed scapula 

 and coracoid, the former pierced by a foramen or fenestra ; 

 pterygials longer than broad, more or less regularly hour- 

 glass-shaped, 4 or 5 in number, one or two of which are in 

 contact with the coracoid. Ventral fins thoracic. 



This large group, consisting chiefly of marine forms, has 

 members in all parts of the world, with the exception of the 

 Arctic and Antarctic regions, and was already represented by 

 numerous Berycidai and a few h>erranida3 and Scorpididse in the 

 Upper Cretaceous. The division into families, capable of 

 rigid definition, is a task of considerable difficulty, and the 

 necessities of a linear arrangement result in the breaking 

 up of seme natural sequences. Thus it: appears highly 

 probable that the Scorpidida?, themselves derived, together 

 with the Serranidse, from the Berycidas, lead to the Carangida? 

 in the division fecombriformes, whilst a nearly perfect 

 passage can be traced between the Acanthuridaj of this 

 division and the Balistidre among the Plectognaths. 

 Thirty-six families : — 



I. Gills four, a slit behind the fourth. 

 A . Two nostrils on each side. 



1. Vtntrals with one spine and 6 to 13 soft rays. 



1. Berycidte. 



