196 THE HOLLOW CURVE OF DISTRIBUTION [pt. ii 



These results are shown on pp. 162, 237. The middle class for 

 New Zealand is rather high, but this is probably due, as I have 

 elsewhere shown, to the occurrence of Cook's Strait in the middle 

 of New Zealand, and it is worthy of notice, that New Zealand, 

 from which such strong evidence has been derived in support of 

 my contention that age is the main factor in distribution, shows 

 the most irregular curve that has as yet been met with in 

 examining many hundreds. The fact of the division of the islands 

 by straits, and the probable occurrence of several different in- 

 vasions of plants (127, 131) are likely enough to account for this. 



As both these curves agreed in type, and as the figures for the 

 endemics of Ceylon, though only estimates and not actual 

 measurements, seemed to hint at something of the same kind, 

 my attention was thus roused, and especially so when the next 

 figures that I obtained, those for the distribution of the endemics 

 in the outlying islands of New Zealand (129, pp. 329, 331 ), showed 

 the same curve, but in a reversed direction, the maxima being 

 upon the largest areas. It was next shown by the endemics of 

 the Hawaiian Islands, where 47 per cent, were confined to one 

 island, and 20 per cent, more to two (out of seven), and the num- 

 bers rapidly diminished upwards (p. 162, curve 3); then by the 

 species of Callitris in Australia (130) and their local distribution, 

 and afterwards by other things. 



Numerous instances of the hollow curve have been given above, 

 for example (in species first of all), in the distribution of the 

 species of Banunculus in New Zealand (pp. 153-6), in the 

 general distribution of the species of Cijvtandra and their local 

 distribution within the Hawaiian Islands (p. 160), the distribu- 

 tion of the Boraginaeeae in New Zealand (p. 161), of Olearia in 

 that country (p. 101), of Doona in Ceylon, of Cyanea and Pelea 

 in the Hawaiian Islands (p. 161 ), of the species of endemic and 

 non-endemic genera in the Hawaiian Islapds (p. 163), ot Exacwn, 

 Christisonia and Ebermaiera in India (p. 163), and so on. In- 

 numerable instances of its applicability could if necessary be 

 produced. Most of these curves are shown on p. 162. 



In the same way, the curve applies to genera, and instances 

 have already been given, for example, in the geographical dis- 

 tribution of the Poiemoniaceae (p. 171), the Cistaceae (p. 172), 

 the Menispermaceae (p. 172), and the Monimiaeeae (p. 174). 



The curve is thus a general feature of the distribution of 

 species by areas occupied, and goes to show that age is of enor- 

 mous importance in geographical distribution. In view of these 



