XV CLASSIFICATION 381 
tinguish real marks of relationship from what is due 
merely to similarity of life and circumstances, or, to 
put it technically, to depend upon homologies and not 
upon mere analogies. The application of the true 
principles has caused the Horned Screamer, in spite 
of his arboreal habits, to be put near the Goose. In 
spite of his way of life and his long legs, which suggest 
that we should class him with wading birds, the 
Flamingo is allied to the Duck, as his webbed feet 
and his beak proclaim. Not only must structural and 
not functional characters, or mere habits, be studied 
for purposes of classification, but the concurrent 
testimony of a number of characters must in every 
case decide to which family a bird belongs. In 
botanical classification Linnzus made the mistake of 
taking into consideration nothing but the number of 
stamens. According to his system a_ wall-flower 
and a lily, a campanula and a dandelion, a buttercup 
and a rose, would belong to the same orders. The 
natural system produces results which may seem 
strange (eg. the buttercup is put in the same order 
as the Traveller’s Joy), but which will bear investiga- 
tion. And in the same way the scientific classification 
of birds, startling as its results may often appear, yet 
gains more and more adherents as true principles 
come to be recognised. But, though the right method 
has now been adopted, the difficulties have not 
vanished. Many systems have recently appeared 
which differ in very important particulars, though 
the constant tendency is towards the narrowing of 
the divergencies. In Mr. Howard Saunders’ Manual 
of British Birds the orders are not in all cases the 
