To these we may add the more or less distinct classes of (5) lake fishes, inhabiting only waters which are deep, clear, and cold, as the various species of whitefish[85] and the Great Lake trout;[86] (6) anadromous fishes, or those which run up from the sea to spawn in fresh waters, as the salmon,[87] sturgeon,[88] shad,[89] and striped bass;[90] (7) catadromous fishes, like the eel,[91] which pass down to spawn in the sea; and (8) brackish-water fishes, which thrive best in the debatable waters of the river-mouths, as most of the sticklebacks and the killifishes. As regards the range of species, we have every possible gradation from those which seem to be confined to a single river, and are rare even in their restricted habitat, to those which are in a measure cosmopolitan,[92] ranging everywhere in suitable waters. =Characters of Species.=--Still, again, we have all degrees of constancy and inconstancy in what we regard as the characters of a species. Those found only in a single river-basin are usually uniform enough; but the species having a wide range usually vary much in different localities. Such variations have at different times been taken to be the indications of as many different species. Continued explorations bring to light, from year to year, new species; but the number of new forms now discovered each year is usually less than the number of recognized species which are yearly proved to be untenable. Four complete lists of the fresh-water fishes of the United States (north of the Mexican boundary) have been published by the present writer. That of Jordan and Copeland,[93] published in 1876, enumerates 670 species. That of Jordan[94] in 1878 contains 665 species, and that of Jordan and Gilbert[95] in 1883, 587 species. That of Jordan and Evermann[96] in 1898 contains 585 species, although upwards of 130 new species were detected in the twenty-two years which elapsed between the first and the last list. Additional specimens from intervening localities are often found to form connecting links among the nominal species, and thus several supposed species become in time merged in one. Thus the common channel catfish[97] of our rivers has been described as a new species not less than twenty-five times, on account of differences real or imaginary, but comparatively trifling in value. Where species can readily migrate, their uniformity is preserved; but whenever a form becomes localized its representatives assume some characters not shared by the species as a whole. When we can trace, as we often can, the disappearance by degrees of these characters, such forms no longer represent to us distinct species. In cases where the connecting forms are extinct, or at least not represented in collections, each form which is apparently different must be regarded as a distinct species.