186 Mr. II. B. Sharpe on the 



brids are often to be seen in collections*. As before stated, 

 Celebes possesses one peculiar species, C. temmincki. 



Secondly, we come to the Ground-Rollers, which I propose 

 to raise to the rank of a subfamily. They are distinguished 

 from the true Rollers by their long legs. So rare are all the 

 species that little is known of the habits of these birds. I am 

 the fortunate possessor of the three known species, each of which 

 is the type of a distinct genus. Mr. Crossley, who has lately 

 returned from Madagascar, and to whom I am indebted for the 

 specimens which adorn my collection, could tell me little of 

 their habits. Atelornis we know, from the observations of Mr. 

 Edward Newton, to be crepuscular in its habits ; and so Mr. 

 Crossley found it to be. The second species oi Atelornis, viz. A. 

 squamigera, I find, is not really congeneric with A. pittoides, but 

 belongs to a distinct genus, which may be called Geobiastes, the 

 characters of which are given in the Plate (PL VIII. figs. 1,2). 



Mr. Crossley never met with either this bird or Brachyptei'a- 

 cias himself. He tells me that the only specimens he obtained 

 were procured by the natives, who informed bim that they only 

 came abroad in the night-time, and sought their food on the 

 ground, this latter fact being confirmed by the earth which 

 always remained on their bills. Madagascar is the home of these 

 curious birds, which, though Coraciine in their affinities, are 

 among the most aberrant forms of which that wonderful island 

 gives us so many examples. I think that their habits and struc- 

 tural peculiarities warrant their separation as a distinct subfamily. 

 Along with the skins of Atelornis, Mr. Crossley sent two skele- 



* [We are not at all sure that these specimens in apparently interme- 

 diate plumage between the two perfectly segregated forms C. imlica and 

 C. affinis are hybrids at all, but consider that, in the absence of all ob- 

 served facts bearing on the point, the diversity of colouring is more rea- 

 sonably to be attributed to the imperfect segregation of the species in 

 the particular locality in which these varieties are found. The point has 

 been often discussed with reference to two Pheasants of the genus Euplo- 

 camus, E. Uneatus and E. horsfieldi {Cf. Sel. P. Z. S. 1863, p. 120), and also 

 in the case of the North- American Woodpeckers of the genus Colaptes — 

 C. aurahis, and C. mexicanus (Cf. Baird, Birds of N. Am. p. 122). Per- 

 haps Mr. Sharpe will allow us to suggest that it is not in a collection that 

 satisfiictory evidence concerning hybridism must be sought. — Ed.] 



