CLASSIFICATION OF DEVONIAN FISHES 453 
of dried specimens, I had even pitched upon the Siluroid genus Clarias 
as that most likely to help me to understand Coccosteius ; but it was 
not until my friend and former pupil, Mr. J. J. Monteiro, brought home 
for me from Congo some specimens of Clarias capensis preserved in 
spirits, that I was able to examine the osseous structure of that fish 
with sufficient care and thoroughness for the purposes of an efficient 
comparison. 
In fig. 20 a careful, reduced representation of the top of the skull 
of this fish is given, and it will be seen that, in everything but the 
minor details of form, it agrees with Coccosteus. The middle line of 
the skull is, asin the latter genus, occupied by three bones. S. O., the 
supra-occipital, is, in the recent form, pointed behind ; Fr., the principal 
frontal, is,as in the fossil, four-rayed ; it exhibits a considerable gap or 
fontanelle, but no median suture; Eth., the ethmoid, and Pmx., the 
premaxilla, correspond exactly in the two skulls, if we leave out of 
consideration the position of the suture seen in the fossil in this region. 
The bone Pr. F., which can be at once identified as the prefrontal in 
Clarias, and which sends down a process dividing the orbit from the 
nostril, obviously corresponds with the similarly related bone in Coc- 
costeus ; while in Clarzas the orbit is completed below by the spatulate 
suborbital bone, Sb. O., smaller in proportion and undivided, but 
otherwise similar to the bone z, s' of Coccosteus. The post-orbital bone, 
Pt. O., and the supra-temporal bone, S. T., of the former appear to 
have their homologues in the bones x and y of the latter fish. 
The space between the frontal, the supra-occipital, and the supra- 
temporal is occupied, in C/larias, by two bones, the anterior of which 
certainly represents the post-frontal ; while the posterior occupies the 
situation of no less than three distinct bones in the heads of ordinary 
fishes, viz., the parietal, the squamosal, and the epiotic. The reduction 
in the normal number of bones which obtains in the Siluroid has 
been carried a step’ further in Coccostews, where the plate lettered for 
shortness’ sake only Pa. Ep. is the only representative of the bones 
Pt. F. and Pa. Sy. Ep. of Clarias. 
Lastly, comes the bone S.s. naturally united in C/ardas to Pa. Sq. 
Ep. and to S. T., and which corresponds with the supra-scapula of 
ordinary osseous fishes, in which it is usually connected with the skull 
only by ligament. The Siluroids and Ganoids, however, coincide in 
always having this bone more closely united with the regular cranial 
bones, and Coccosteus, it will be observed, agrees with them. 
So much for the cranial shield. To comprehend the dorsal and 
ventral body shields we must have recourse, not to C/arzas, but to other 
Siluroids, such as Bagrus, Arius, &c. In these fishes, in fact, the an- 
