Classificdliou of the Sconihroi'd Fishes. 67 



of them can be given, and, as a matter of fact, none of the 

 characters used to separate them from the Perciformes is 

 really distinctive. 



The discovery that the Trichiurid genera Lepidopus and 

 Evpleurogrammus do not conform to Boulenger's definition 

 of the Acanthopterygii, as the pelvic bones are remote from 

 the cleithra (clavicles) and only connected to them by a long 

 ligament, has led me to look into the anatomy of the 

 Scombriform fishes. 



As a result I find that the Carangidae, Rhachicentridae, 

 Coryphffinidse, Bramidse, and Menidse show no special affinity 

 to the Sconibridse and may be placed with the Percoids, but 

 the remainder of Boulenger's Scombriformes is a natural 

 group, which still includes so great a diversity of forms as to 

 be with difficulty definable. 



It is worth notice that none of the five families mentioned 

 above as rejected from the Scombriformes is known before 

 the Upper Eocene, but that the true Scombroids were 

 abundant in the Lower Eocene (London Clay), from which 

 forms quite as specialized as any living at the present day 

 are known. 



This early specialization of the Scombroids makes it 

 necessary to consider whether the indirect attachment of 

 the pelvic bones to the cleithra in the Trichiuridse may not 

 be a primitive feature. That this is not the case is shown 

 by the fact that in all the members of the group w^hich have 

 well-developed pelvic fins, the pelvic bones are directly 

 attached to the cleithra — i. e. their anterior extremities are 

 firmly imbedded in the ligament which connects the cleithra 

 above the symphysis *. In Lepidopus and Eiqjleurogrammus 

 the pelvic fins are reduced to a pair of scales and the pelvis 

 to a small spicular bone, connected, by a long ligament with 

 the symphysis ; this condition may be regarded as secondary 

 and due to the degeneration of tlie fins. 



Consequently the Scombroids may be regarded as an early 

 offshoot from the Percoid stem, agreeing with the Perches 

 in most characters, but differing in certain features of 

 specialization. 



Before proceeding to the classification of the true 



* Epinnula, the most generalized Trichiuroid, is not represented in 

 the British Museum. Dr. Th. Gill very kindly examined the specimen 

 of E. magistralis in the Smithsonian lustitution and wrote to me, " the 

 pelvic bones are attenuated forwards and terminate in a ligament con- 

 necting with the 'clavicles' at their symphysis." These words describe 

 the condition I observe in T/ii/)-sites, hut on dissecting away the ligament 

 in which the pelvic bones terminate I find that the latter extend forward, 

 enclosed in a ligamontous sheath, to between the cleithra. 



