— 10 — 



of the posterior portion of the trigemino-facialis chamber, and then turns backward in the bone, parallel 

 to that edge of the Chamber. At the anterior edge of the subtemporal fossa the canal turns inward 

 in the bounding wall of the fossa, still lying in the proötic, and, on one side of the head of the one 

 specimen sacrificed in this examination, there seemed to end blindly at the hind edge of the proötic. 

 On the other side of the head of the same specimen the canal opened into the subtemporal fossa, but 

 whether this opening was an artifact or not, I could not deternüne. What this canal is I can not 

 determine, but it may, perhaps, be a persisting remnant of the spiracular canal of Arnia. 



In Albula the temporal fossa is similar to that in Elops, but its anterior portion is much less 

 extensive; and I find no canal opening into the bottom of the fossa. No special description of it 

 seems necessary. 



In Esox, according to Vrolik's ('73) figures of transverse sections of the skull, there is, in the 

 temporal fossa, a mesial pocket which has the position of a preepiotic fossa. The deeper portion of 

 this pocket is bounded both above and below by portions of the supraoccipital bone, exactly as the 

 preepiotic fossa is, in Elops. The anterior end of the temporal fossa ends blindly in the thick dorso- 

 lateral edge of the chondrocranium, and is there partly surrounded by a thin layer of perichondrial 

 bone which apparentlv belongs to the pterotic though this is not so stated. The dorsal portion of the 

 alisphenoid is formed by two plates of perichondrial bone, one of which lines the external and the 

 other the internal surface of the cartilaginous cranial wall. The internal plate is apparently raised 

 into a ridge-like process, but whether it bounds a recess for the dorso- anterior end of the anterior 

 semicircular canal, or not, is not evident. The wall of the cranial cavity, however, quite certainly 

 continues backward, in the line of the ridge-like process, until it joins that flange of the supraoccipital 

 that forms the floor of the preepiotic pocket, as it does in Elops. A region corresponding to that oc- 

 cupied by the anterior portion of the temporal fossa of Elops is thus, in Esox, included in the cranial 

 wall, and the sphenotic, as in Elops, is excluded from bounding relations to the cranial cavity. Assume 

 that there is, on the dorsal surface of this part of the chondrocranium of Esox, an anterior diverticulum 

 of the temporal fossa similar to the one in Amia: a simple enlargement and deepening of this diver- 

 ticulum, and its partial confluence with the temporal groove of the fish would produce the conditions 

 found in Elops; and if the enlarged diverticulum did not become confluent with the temporal groove, 

 acquiring, instead, an independent opening to the exterior, on the lateral surface of the skull, it would 

 seem as if it must give rise to the conditions described by Ridewood ('04c, p. 473) in Engraulis. 



In the Characinidae, judging from Sagemehl's ('84b) descriptions and figures, the pterotic, 

 and that bone alone, forms the floor of the temporal fossa. This is however not true of Macrodon, 

 which I have examined in this connection. In this fish the fossa has anterior and posterior portions, 

 as in Elops, the two portions being separated by a low saddle. The floor of the posterior portion is 

 formed by the pterotic, as Sagemehl states, but that of the anterior portion is formed by the proötic, 

 the latter bone having a widely spreading dorsal edge. The anterior end of the anterior portion of the 

 fossa abuts against and is bounded anteriorly by the sphenotic. The fossa is thus less extensive than 

 in Elops, but much more extensive than in Scorpaena. It Las an extensive preepiotic pocket, and in 

 this pocket lie those fenestrations of the epiotic that are said to be so characteristic of the Characinidae. 

 The anterior wall of the labyrinth recess is formed by a strong flange that lies mainly on the proötic 

 but extends upwards slightly onto the cerebral surface of the sphenotic. Between this flange and the 

 cerebral wall of the anterior portion of the temporal fossa, in a deep groove, lies the anterior semi- 



