OSTEOLOG Y OF HEMIBRA NCHIA TE FISHES— STARRS. 623 



diverge very widelj^ and have man}- characters radically different, as 

 may be seen by referring to the diagnosis here included/ 



It is more difficult to see just where the family Macrorham})hosid8e 

 comes in, whether it is higher than the Aulostomid fishes or has not 

 departed so far from its parent stem. It is an otfshot in another 

 direction. It certainly is not so highly specialized as the Aulostomid 

 families wnth their anterior vertebrte and parapophyses ankylosed; 

 their weak or absent spinous dorsal and their peculiarly modified 

 basioccipital condyle. It seems well placed above them, having a 

 well-developed spinous dorsal and the anterior vertebra?, but slightly 

 enlarged and normally articulated. 



The family Centriscidte shows evidence of having come from some- 

 where along the Macrorhamphosidte stem, but so far back and along 

 such different lines that the evidence is not satisfactory. It is much 

 more highly specialized. 



DIAGNOSIS OF THE HEMIBRAXCHIL 



Opisthotics absent; parietals usually absent; exoccipitals never 

 meeting over surface of basioccipitals; myodome usually al)sent or 

 rudimentary, sometimes present; post-temporal never typically forked, 

 sometimes a ganoid plate with an inner process, sometimes united by 

 suture to cranium; a portion of the hypocoracoid sometimes ganoid, 

 and appearing externally as a separate element ('' interclavicle"); 

 supraclavicle usually absent, small when present; postclavicle some- 

 times absent, composed of a single element when present; superior 

 pharyngeals and usually elements of branchial arches reduced in 

 number; inferior jiharyngeals present, not united; four anterior ver- 

 tebrae more or less elongated, sometimes united; bony dorsal buck- 

 lers coinciding with vertebrae developed anteriorily (hidden b}- scales 

 in MacTorliain])ho8UH)\ transverse processes developed on all abdomi- 

 nal vertebrje, usually largest anteriorly (in the Aulostomoidea united, 

 and forming lateral shelves); snout more or less produced and tube- 

 like, with a small mouth at its end; ventrals abdominal, sometimes 

 anteriorly placed; hypocoracoid foramen sometimes formed partly by 

 clavicle, always bordered by clavicle as seen from outer side, though 

 hypercoracoid may or may not entire!}^ inclose its foramen as seen 

 from within. 



GASTEROSTEOIDEA. 



Parietals present; pterotic normally placed above prootic exoccipi- 

 tal suture; condyle of basioccipital normal, concave; basis cranii or 

 shelf covering myodome incomplete, not nearly reaching to anterior 

 edge of prootics, but completed in cartilage; basisphenoid absent; 



^ It is this element of unstability in the coherence of characters that has made it so 

 difficult to assign many forms to their proper groups. We do not know in what 

 direction characters may be depended upon to show relationships. 



