( " ) 



sought in the changed conditions under which such birds exist. In the first instance 

 the absence of enemies on the insular areas inhabited bj- them leads to the loss of 

 the power of flight through the reduction of the wing by disuse, or by the natural 

 selection of the less strongly flying individuals; by this profound change of habit 

 the organism is thrown out of equilibrium with its environment, and variations, 

 eliminated under ordinary circumstances, survive. The reason why these variations 

 tend towards an increase of bulk may be that the expenditure of energy is much 

 less. It must, however, be pointed out that very great ditTerences of size also 

 occur among ordinary birds within the limits of a species. In an important 

 paper* published in 1871, J. A. Allen has shown that individual variation in the 

 size of the body to the extent of from fifteen to twenty per cent, often occurs in 

 many species of North American Passeres, while in particular measurements the 

 range may be much greater. In different skulls of Dicrphorcqjieryx the beaks are 

 commonly dissimilar, not only in size but also in form (degi-ee of cm-vature, etc.), 

 but since in the above-quoted paper somewhat similar differences are shown to 

 occur in such species as Mniotilla varia and Quiscalus versicolor, there can be 

 no excuse for establishing new species on the strength of such variations. 



Vertebrae. — No complete vertebral column has been found, the most perfect series 

 only consisting of sixteen vertebrae, and of these some may not belong to the same 

 individual as the rest ; there is, therefore, some doubt as to the precise number in 

 the various regions of the column, but, partly from examination of a large series of 

 odd vertebrae, and partly from comparison with Ocydromus, I conclude that there 

 were 13-14 true cervicals and 2 cervico-dorsals. Of the number of dorsals it is 

 impossible to be certain ; none were anchylosed with one another. 



The atlas is very similar to that of Ocydromus, but the neural arch is propor- 

 tionately much broader and the hypapopbysis is larger. 



The axis has a much larger and more massive neural spine than that of Ocydromius, 

 and the hypapopbysis, instead of forming a sharp keel-like outgrowth, is expanded 

 at its hinder end into a X-shaped structure. In the ne.xt vertebra (3rd) the median 

 bar of the X is much reduced, and the two lateral processes come oft" from the hinder 

 portion of the centrum almost independently, being only united by a trans\erse ridge 

 which represents the remnant of the median hypapopbysis. In the next vertebra 

 (4th) the paired hypapophyses rise quite sejiarately from the hinder end of the 

 centrum, and project nearly vertically downwards, enclosing a channel for the 

 carotid (?), just as the catapophyses on the 7th, 8th, and 9th cervicals do. In all the 

 smaller rails the hypapopbysis on the anterior cervicals is simple ; in Aplornis on 

 the 2nd, 3rd, and 4th it is j_-shaped, but in no case are there two distinct 

 hypapophyses, as in the present species. 



On the fourth cervical there is a well-developed interzygapophysial bar enclosing 

 a large interzygapophysial foramen. This does not occur in any of the other 

 vertebrae. 



On the fifth the catapophyses appear, and in the succeeding five or six vertebrae 

 they gradually increase in size, and approach one another till they nearly completely 

 enclose a carotid canal. In the last vertebra in which they occur they are reduced 

 in size and rise from a common base, and in the following vertebra they are replaced 

 by a median forwardly projecting hypapojihysis similar to that seen in Ocydromus. 

 In the succeeding vertebrae this process is gradually reduced in size. In the last 

 cervical, the two (?) cervico-dorsals, and first one or two dorsals, in addition to the 

 * Hidl. Mm. Ciivij). XooUigij. Ilnrvtivd, Vol. II. p. ISii. 



