This emendation of the Arcifera and Firmisternia was accepted by Cope in his synopsis of the families of Vertebrata (Amer. Natural. xxiii., 1890), except that he still retained his suborder Gastrechmia.
Since the publication of Boulenger's great work a number of forms have been discovered which, from the characters of their dentition, have necessitated the establishment of certain new families, namely, Ceratobatrachidae and Genyophrynidae; and Boulenger was the first to recognise that the taxonomic value of the mere presence or absence of teeth in the jaws had been overestimated. I therefore propose using it as a character distinctive of the sub-families only, thereby reducing the number of families, relying first (leaving the Aglossa aside) upon the firmisternal or arciferous condition of the pectoral arch, secondly upon the dilated or cylindrical shape of the sacral diapophyses, thirdly upon the dentition. Blindly consistent application of these principles would reduce the Phaneroglossa to four families only, namely Ranidae, Engystomatidae, Cystignathidae and a fourth family comprising all the Arcifera with dilated sacral diapophyses. This would obviously be wrong. We have therefore to resort to other additional characters or rather peculiarities. The opisthocoelous character of the vertebrae and the possession of distinct ribs, together with the disc-shaped tongue, separate the Discoglossidae and justify their retention as a family. The Hylidae are marked off by the claw-shaped terminal phalanges, but the remaining forms, comprising the Bufonidae and Pelobatidae, cannot be separated except by their dentition, and I plead guilty of inconsistency in retaining them as separate families.
After all, our classification may not represent the natural system, and it may be nothing but a convenient key.
When we have eliminated the characters of the vertebrae, the dentition, the claw-shaped phalanges and the adhesive discs, it may well be asked what characters remain. The firmisternal is a further, higher modification of the older, more primitive arciferous condition. The difference between the dilated and cylindrical shape of the sacral diapophyses is in not a few cases very slight, and there are various, most suggestive exceptions. The presence or absence, size and shape of the omosternum and metasternum are of very limited taxonomic value, not always applicable to all the members of the same family. The fact is, that the Anura are a very recent and a most adaptive, plastic group. The earliest known fossils are scarcely older than the Middle Eocene.
Almost every one of the greater families has produced terrestrial, arboreal, aquatic, and burrowing forms. Their habits have modified, and are still shaping their various organs, first of course those by which the animals come first and most directly into contact with their surroundings (e.g. adhesive discs, dentition, general shape of the body, length of limbs, wartiness of the skin, tympanic disc). These are the so-called adaptive characters, sometimes decried as merely physiological; as if habits, use, and requirements did not likewise influence and ultimately model every other organ (e.g. tympanic cavity, Eustachian tubes, vertebrae, ribs, coccyx, pectoral arch, etc.). There are true Toads, Bufonidae, which are as smooth, wartless, slender-bodied and long-legged as the most typical of "Frogs"; true Ranidae, like Rhacophorus, which by their green colour, large adhesive discs and arboreal habits may well put many of the Hylidæ to shame. Ceratohyla has developed the claw-shaped terminal phalanges which are otherwise typical of, and peculiar to, the Hylidae, but this genus reveals itself by various details as a close relation of the other Hemiphractinae; and these fall in with the Cystignathidae on the strength of their cylindrical, not dilated, sacral diapophyses.
In sketching the phylogenetic tree of the families of the Anura we have to proceed with great caution.
There is not much doubt about the Aglossa. They have retained some of the most primitive characters, but have by now been so much modified and specialised that they are to be looked upon as an early side-branch.
Among the Phaneroglossa the Discoglossidae are with certainty the oldest, but are now scarce in genera and species, and much specialised. The Pelobatidae connect them with the Bufonidae. The Cystignathidae form a rather ill-defined assembly which points downwards to the Pelobatidae, upwards to the Hylidae. There is no divergence of opinion about the Ranidae being the highest of all the Anura, and amongst them the Raninae the most typical, the Dendrobatinae the most specialised. If we assume that moderately dilated sacral diapophyses represent a more primitive stage than cylindrical processes, we shall naturally look to the Engystomatidae as the connecting link between the Ranidae and the Arcifera, through Bufonoid creatures still with teeth in both jaws. If, on the other hand, we take the dilatation to be a further development from more or less cylindrical processes, then the Ranidae can be considered as having sprung from Cystignathoid creatures, which have consolidated their pectoral arch into the firmisternal condition; and in this case the Firmisternia would not be a natural group, the Engystomatidae pointing, to the Bufonoid stock. This would, to a great extent, mean a reversion to Cope's idea.
Sub-Order 1. Aglossa.–The two diagnostic peculiarities of the few members of this group are: first, the absence of a tongue; secondly, the union of the Eustachian tubes into one median pharyngeal opening in the posterior portion of the palate.
