60 U.S. NATIONAL MUSEUM BULLETIN 284 



amined, except those of the striatus species group (Springer 1967b) 

 the attachment is similar to that described for E. nigricans. In the 

 striatus species group the attachment is to the pterotic. In the Ne- 

 mophidinae there does not seem to be a ligament in the position of 

 the ventral arm of the post-temporal. The presence or absence of an 

 intercalar or its relative size appears to be independent of whether 

 the posttemporal is forked or M^hether it attaches to the intercalar 

 or pterotic. 



The number of pectoral rays in blenniids varies from 11 to 18, 

 with 11 and 18 rays as uncommon intraspecific variants (counts of 

 17 and 18 rays are knoAvn only for Atrosalarias) . The most frequently 

 encountered modal numbers of rays for any species are 13 to 15. The 

 modal number of rays appears to be fairly constant within most 

 genera (some notable exceptions: Ebseniws, Enchelyurus, and Blennius). 

 The rays are unbranched, except that some large specimens of parti- 

 cular species may show a slight forking at the tips of one or two rays. 



The number of pectoral radials is invariably four, but the rela- 

 tionship of radials to scapula and coracoid is variable. The relation- 

 ships range from two radials articulating with the scapula, one with 

 both the scapula and coracoid (at the scapulo-coracoid joint), and 

 one with the coracoid (or radial formula 2-1-1, in some species of 

 Blenniini and Omobranchini) , to one radial articulating with the 

 scapula and three with the coracoid (1-0-3, in Ecsenius, AspidontiLS, 

 and Blennechis). The most common relationship is that of two radials 

 articulating with the scapula and two with the coracoid (2-0-2), 

 which occurs in most Salariini. The relationships grade into each 

 other somewhat because one radial may articulate with the carti- 

 laginous scapulo-coracoid joint area (which may be a relatively 

 broad area, rather than one that has the osseous portions of the two 

 bones impinging). 



Regan (1913) indicated that the most generalized perciforms had 

 a 2-1-1 radial formula. A perusal of Starks' (1930) paper on the 

 pectoral girdle as well as my own dissections of Kuhlia sandvicensis 

 (3-0-1, Kuhlidae) and Abudefduf abdominalis (3-1-0, Pomacentri- 

 dae), both of which are less specialized than blenniids — at least in 

 caudal structure (Gosline, 1961) — indicates that most of the generally 

 considered less-specialized perciforms do have a radial formula of 

 3-1-0 or 2-1-1. AU chaenopsids, tripterygiids, and clinids (except 

 Mnierpes and some species of TjihrisomuSy Malacoctenus, and Ophi- 

 clinus — all clinids- — with radial formulas of 3-0-1), which are l^s 

 speciaUzed than blenniids, have a radial formula of 2-1-1. If the 

 meager evidence presented herein is indicative of evolutionary trends 

 in the perciforms — and I think it is — then the primitive perciform 

 had three radials attached to the scapula and one probably in the 



