216 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY 



The opposite palatines, which are exposed in the top view, do not converge toward 

 the front but stretch, as it were, the rostrum into a thin flattened surface. This lateral 

 stretching of the rostrum may somehow be connected with the loss of the mesethmoid and 

 the presence of the two well developed proethmoids, which are of unknown origin.' 



The long wedge-like cranium is strongly braced on its dorsal side by the immense flat 

 frontals, which extend from near the tip of the snout almost to the occiput. As the body is 

 long and low the supraoccipital crest is small, although the supraoccipital itself forms the 

 well-braced keystone of the occipital arch. The short wide parietals, separated slightly 

 by the supraoccipital above the occiput, form the roof of a pair of deep caverns extending 

 inward between the pterotics and the epiotics and in front of the posttemporal fossae of 

 isospondyls. 



The orbits are small and the myodome not large. The lacrymals are much produced 

 in front to cover the sides of the snout and there are five rather small suborbitals, a pre- 

 orbital and a dermal prefrontal. The lateral ethmoids (parethmoids) form diverging 

 braces for the palatopterygoid arches. 



Other osteological details of the cranium, as recorded by Starks (1904a, p. 256), are 

 chiefly of interest in contrasting the Esocidae with the Umbridae and Poeciliidae. 



Umbra. — The skull of Umbra krameri from Hungary (No. 1013, Brit. Mus. Nat. Hist.) 

 is singularly devoid of marked specializations beyond those noted by Starks (1904^). The 

 mouth is fairly small; the maxilla enters the posterior part of the gape. The quadrate is 

 produced forward beyond the middle of the fairly large orbit. As a whole this skull would 

 seem to make a good structural ancestor for the specialized skulls of the cyprinodonts. 



The otolith (sagitta) of Umbra crameri is described by Frost (1926c, p. 465) as being of 

 the salmonid type, but differing in certain details. On the other hand, it resembles that 

 of Esox in a number of important features that appear to indicate close relationship. 



Dallia pectoralis. — This is a very peculiar Alaskan fish for which Gill erected the order 

 Xenomi, based chiefly on certain peculiar characters of the shoulder-girdle with which 

 we are not especially concerned. Starks (1904f) subjected the osteology of this fish to a 

 thorough analysis and showed that it is related to the order Haplomi in the following char- 

 acters: (1) in having paired dermal ethmoids; (2) four separate superior pharyngeals on 

 each side, those of the anterior two arches toothless, the others with teeth; (3) upper limb 

 of posttemporal attached to the epiotic by a ligament. Characters 1-3 he found elsewhere 

 only in Umbra and Lucius. In addition to these he found that (4) the palatopterygoid arch 

 was reduced to a single element, as in the Poeciliidse; (5) that the splanchnic anatomy is 

 very similar to that of Umbra. The cranium was largely cartilaginous, many of the usual 

 bones being absent. He redefined the order Xenomi on the characters of its coracoids and 

 actinosts, but pointed out its close relationship to the Haplomi. 



MiCROCYPRINI (TOP-MINNOWS, ETC.) 



The ordinary Fundulus is a stout-bodied, vigorous little fish with broad, quick-darting 

 tail, large rounded head and small upturned bulldog-like mouth adapted for seizing food 



' Allis' suggestion that they may be the homologue of the mesethmoid is not accepted by Starks (1926a, p. 335) for the 

 reason that he found in OTmnus and Thaltichthys a median mesethmoid in addition to paired proethmoids. On the other hand, 

 Allis' suggestion (1909, p. 24) that these paired ethmoidal elements gave rise to the ascending processes of the premaxillas of 

 acanthopts has in its favor only an imperfect correspondence in the topographic relations of the respective elements (see p. 96 

 below). 



