STRUCTURE AND AFFINITIES OF THE PLATYSOMID. 385 
bring together. The Pycnodonts are not heterocercal in the same sense as Platy- 
somus, Eurynotus, &c., but are nearly as homocercal as the Salmonide ; nor is 
the tail of Hurynotus, or of Mesolepis “ equilobate.” The anal fin of Hurynotus 
has not an elongate, but a short base, and the ventral fins, both in it and in 
Mesolepis, are of very respectable size. 
Other characters, whatever value they may have in distinguishing families and 
genera, are hardly admissible in the definition of a “suborder” of fishes, being 
merely part of the endless variations and coincidences in external form which 
the process of specialisation brings out in forms which may either be very dis- 
tantly related or closely allied. Such are the deep shape of the body (which 
here cannot be called “ rhomboidal” in every case), the length of the dorsal and 
anal fins, and the small size of the ventrals, even if these peculiarities of the 
two last named fins held good with all the genera, which is not the case. 
Similarly, I cannot look upon the form of the scales as being a character of 
prime importance, though it certainly is of greater value than the depth of the 
body, or the length or size of a fin. 
For, if we compare the scales of Palwoniscus (Pl. VI. fig. 16), or of Eury- 
notus (Pl. III. fig. 3), with those of Platysomus (Pl. VI. figs. 3, 4) and of 
Gyrodus (Pl. VI. figs. 14, 15), it becomes perfectly clear that the so-called 
“scale rib” or “ Lepidopleuron” of the last-named genera is no special or 
isolated phenomenon, but is, after all, nothing more or less than that vertical 
keel which is characteristic of the under surface of the scales in almost all 
rhombiferous Ganoids, and which ordinarily passes up into or ends a little in 
front of the base of the articular spine, besides being bevelled off inferiorly by 
the anterior margin of the little fossette which lodges the spine of the scale next 
below. This keel may be in some cases prominent, in others obsolete, in some 
more or less central, in others placed at or near the anterior margin. Of course 
in the true Pycnodonts these “scale ribs” form very prominent objects from 
the thinness, and in some cases the entire absence of the rest of the scale. But 
for my own part, I cannot understand how the mere marginal position of such a 
_ keel can ever carry with it so great a morphological importance as to entitle it 
to be used as a SUBORDINAL character, especially when contradicted by obvious 
facts of structure, cranial, or otherwise. But it is indeed hardly necessary, at 
the present day, when AGassiz’s system of classification of fishes, according to 
their scales, is a thing of the past, to dwell upon the fact that all attempts to 
found any large groups upon the mere external configuration of these append- 
ages must prove utterly futile. 
Again, the non-lobate nature of the paired fins, and the branchiostegal rays 
not taking the form of broad (jugular) plates, are characters shared also by the 
Amioid and Lepidosteoid Ganoids, and in many of the latter the notochord is 
also persistent, with well-ossified arches. It is here not meant that no 
