August 21, i8gi.] 



SCIENCE. 



105 



the salmon and herring. These have all numerous vertebrse, 

 small in size, and none of tliem in any notable degree modi- 

 fied or specialized. In the northern seas Isospondyli still 

 exceed all other fishes in number of individuals. They abound 

 in the depths of the ocean, but there are comparatively few 

 of them in the tropics. 



The Salmonidce vphich inhabit the rivers and lakes of the 

 northern zones have from 60 to 65 vertebrse. The Scopelidoe, 

 Stomiatidce, and other deep-sea analogues have from 40 up- 

 wards in the few species in which the number has been 

 counted. The group of Clupeidce is probably nearer the primi- 

 tive stock of Isospondyli than the salmon are. This group is 

 essentially northern in its distribution, but a considerable 

 number of its members are found within the tropics. The 

 common herring ranges farther into the Arctic regions than 

 any other. Its vertebrse are 56 in number. In the shad, a 

 northern species which ascends the rivers, the same number 

 has been recorded. 



The sprat and sardine, ranging farther south, have from 

 48 to 50, while in certain small herring which are strictly 

 confined to tropical shores the number is but 40. Allied to 

 the herring are the anchovies, mostly tropical. The north- 

 ernmost species, the common anchovy of Europe, has 46 

 vertebrae. A tropical species has 41 segments. There are, 

 however, a few soft-rayed fishes confined to the tropical seas 

 in which the numbers of vertebrae are still large, an excep- 

 tion to the general rule for which there is no evident reason 

 unless it be connected with the wide distribution of these 

 almost cosmopolitan fishes. In a fossil herring-like fish from 

 the Green Eiver shales, I counted 40 vertebrae ; in a bass-like 

 or serranoid fish from the same locality 24, these being the 

 usual numbers in the present tropical members of these 

 groups. 



The great family of Siluridce or catfishes seems to be not 

 allied to the Isospondyli. but a separate ofi'shoot from an- 

 other ganoid type. This group is represented in all the fresh 

 waters of temperate and tropical America, as well as in the 

 warmer parts of the Old World. One division of the family, 

 containing numerous species, abounds on the sandy shores 

 of the tropical seas. The others are all fresh-water fishes. 

 So far as the vertebrae in the Siluridm have been examined, 

 no conclusions can be drawn. The vertebree in the marine 

 species range from 35 to 50; in the North American forms 

 from 37 to 45, and in the South American fresh-water spe- 

 cies, where there is almost every imaginable variation in 

 form and structure, the numbers range from 28 to 50 or 

 more. 



The Cyprinidce, confined to the fresh waters of the north- 

 ern hemisphere, and their analogues, the CharacinidcB of 

 the rivers of South America and Africa, have also numerous 

 vertebrae, 36 to 50 in most cases. I fail to detect in either 

 group any relation in these numbers to surrounding condi- 

 tions. 



In general, we may say of the soft-rayed fishes that very 

 few of them are inhabitants of tropical shores. Of these few, 

 some which are closely related to northern forms have fewer 

 vertebras than their cold-water analogues. In the northern 

 species, the fresh-water species, and the species found in the 

 deep sea, the number of vertebras is always large, but the 

 same is true of some of the tropical species also. 



Among the spiny-rayed fishes the facts are more striking. 

 Of these, numerous families are chiefly or wholly confined 

 to the tropics, and in the great majority of all the species 

 the number of vertebrae is constantly 24, 10 in the body and 

 14 in the tail (10 -|- 14). In some families in which the pro- 



cess of ichthyization has gone on to an extreme degree, as in 

 certain Plectognath fishes, there has been a still further re- 

 duction, the lowest number, 14, existing in the short inflexi- 

 ble body of the trunkfish, in which the vertebral joints are 

 movable only in the base of the tail. In all these forms, the 

 process of reduction of vertebrfe has been accompanied by 

 specialization in other respects. The range of distribution 

 of these fishes is chiefly though not quite wholly confined 

 to the tropics. 



A very few spiny-rayed families are wholly confined to 

 the northern seas. One of the most notable of these is the 

 family of viviparous surf fishes, of which numerous species 

 abound on the coasts of California extending to Oregon, and 

 Japan, but which enter neither the waters of the frigid nor 

 the torrid zone. These fishes seem to be remotely connected 

 with the Ldbridce of the tropics, but no immediate proofs of 

 their origin exist. The surf fishes have from 32 to 42 verte- 

 bra, numbers which are never found among tropical fishes 

 of similar appearance or relationship. 



The case of the Ldbrid(B, in which the fact was first no- 

 ticed, has been already mentioned. Equally striking are the 

 facts in the great group of Cataphracti, or mailed-cheek 

 fishes, a tribe now divided into several families, diverging 

 from each other in various respects, but agreeing in certain 

 peculiarities of the skeleton. Among these fishes the family 

 most nearly related to ordinary fishes is that of Scorpcenidce. 

 This is a large family containing many species, fishes of local 

 habits, swarming about the rocks at moderate depths in all 

 zones. The species of the tropical genera have all 24 verte- 

 br£e. Those genera chiefly found in cooler waters, as in Cali- 

 fornia, Japan, Chili, and the Cape of Good Hope, have in all 

 their species 27 vertebrse, while in the single arctic genus 

 there are 31. An antarctic genus bearing some relation to 

 Sebastes has 39. 



Allied to the Scorpcenidce, but confined to the tropical or 

 semitropical seas, are the Platycephalidoe, with 27 vertebrffi, 

 and the Cephalacanthidce, with but 22. In the deeper waters 

 of the tropics are the Peristediidce, with 33 vertebrae, and 

 extending farther north, belonging as much to the temperate 

 as to the torrid zone, is a large family of the Ti'iglidoe, in 

 which the vertebraj range from 25 to 38. 



The family of Agonidce, with 36 to 40 vertebra;, is still 

 more decidedly northern in its distribution. Wholly con- 

 fined to northern waters is the great family of the Cottidce, 

 in which the vertebrfe ascend from 30 to 50. Entirely polar 

 and often in deep waters are the Liparididce. an offshoot 

 from the Cottidce, with soft, limp bodies, and the vertebrae 

 35 to 65. In these northern forms there are no scales, the 

 spines in the fins have practically disappeared, and only the 

 anatomy shows that they belong to the group of spiny-rayed 

 fishes. In the Cyclopteridce, likewise largely arctic, the 

 body becomes short and thick, the backbone inflexible, and 

 the vertebras are again reduced to 28. In most cases, as the 

 number of vertebrae increases, the body becomes proportion- 

 ally elongate. As a result of this, the fishes of arctic waters 

 are, for the most part, long and slender, and not a few of 

 them approach the form of eels. In the tropics, however, 

 while elongate fishes are common enough, most of them 

 (always excepting the eels) have the normal number of ver- 

 tebr£e, the greater length being due to the elongation of their 

 individual vertebra; and not to their increase in number. 



In the great group of bleuny-like fishes the facts are equally 

 striking. The arctic species are very slender in form as 

 compared with the tropical blennies, and this fact, caused by 

 a great increase in the number of their vertebra?, has led to 



