PHYLUM VIII VERTEBRATA 263 



significant record of the ancient avian faunas. Yet as far as the record shows, 

 Tertiary ornithic faunas were essentially like the modern, and it is only as we 

 direct attention to the Mesozoic that signal differences are to be observed. In 

 Archaeopteryx, the oldest known bird, the tail had not become atrophied, nor 

 the teeth lost in the adult ; the pelvis was not fully developed, and the verte- 

 brae had not yet acquired the saddle-shaped articulations characteristic of 

 post-Jurassic forms. 



The Classification of Birds. — Classification, as here employed, is an attempt to 

 place together related species and to arrange the groups thus formed so as to 

 indicate the extent to which each has been modified, or has departed from 

 what is conceived to be the most generalised or lowest type. Further than 

 this a linear or tabular arrangement cannot go, and it is obviously impossible 

 to express in such a manner the interrelationships of the various groups ; 

 moreover, such a series cannot be so arranged that we may pass by regular 

 gradations from the lower to the higher forms. 



The difficulties attending the classification of birds are at once their great 

 general similarity of structure and their numerous adaptive modifications, 

 sometimes slight, sometimes so great as to obscure characters of real value. 

 There are, besides, a certain number of aberrant forms whose exact position is 

 a matter of uncertainty, and others in which there are departures more or less 

 pronounced from the general structure of the group in Avhich they should 

 obviously be placed. For it must be constantly borne in mind that in 

 palaeornithology we are not dealing with the entire class of birds, but only 

 with a certain portion of it, since the number of known fossil birds is very 

 small, and it is consequently impossible to trace the lines of descent of existing 

 species ; we do not even have broken lines to guide us, but merely isolated 

 dots to indicate their probable existence. For the proportion of fossil to 

 existing birds is small indeed, about 500 extinct to 12,000 living species, and 

 most of these are from the Miocene or later horizons ; they are easily refer- 

 able to existing families and often to existing genera, so that they throw little 

 light on the phylogeny of modern birds. 



The reasons for the remarkable dearth of fossil avian remains are obscure, 

 and those usually adduced, such as the imperfection of the geological record, do 

 not seem altogether satisfactory, the more so since in some favoured localities, 

 such as Allier in Southern France, and Fossil Lake, Oregon, bones of birds 

 have been found in considerable numbers. The palaeontologist is further 

 hampered by having to restrict himself to characters offered by the skeleton 

 alone, and while these are of primary importance, much valuable evidence 

 may be gathered from the muscles, viscera, and plumation. And birds must 

 be classified by the resultant of all their characters, not by any one set, for 

 the exceptions to any general rule are nowhere more numerous among verte- 

 brates than in this particular class. Finally, there is the personal equation, 

 or the individual opinion of the classifier, as to the relative values of the 

 characters on which we must rely for uniting or separating species. For these 

 reasons no two systems will be found to agree in all their details, certain 

 birds or groups of birds being particularly liable to shifting about at the hands 

 of the taxonomist. 



The divisions here employed are practically those of Stejneger,^ although 



^ Standard Natural History, vol. III. — Birds. Boston, 1885 (afterwards changed to the Eiverside 

 Natural History). 



