28 TRANSACTIONS OF THE [Sess. lxxx 



regular transference of seeds direct from range to range is 

 not in accordance with the main facts of distribution on the 

 ranges : still less will it account for any peculiarities in that 

 distribution — for instance, the remnant flora, the greater 

 specific variety on the Mekong-Yangtze divide, and the occur- 

 rence of species peculiar toone range {e.g. FriinulaFrnnchetii, 

 Pax., FritiUdria SoulU^i, Franch., Cai^siope 'palpehrata, 

 W. W. Sm., on the Mekong-Salween divide; Gentiaria 

 shio-ornafa, Balf. f., Saxifraga nigruglanduloxa, Engl, et 

 Irmscher, Meconopsis integr if alia, Fra.nch., on the Mekong- 

 Yangtze divide ; see footnote, p. 27). Moreover, if wind and 

 birds could be relied on to transport seeds from range to 

 range with some degree of regularity, the floras should be 

 more similar than they actually are, especially in the alpine 

 region, where, as we have seen, the actual climates are 

 not very different. The floras would be adjusted to the 

 circumstances of distribution much more rapidly than either 

 could change owing to changes of climate. But the fact is, 

 even if we assume that a similar flora once clothed all the 

 divides owing to tlie dispersal of seeds across them, we are 

 still unable to dispense with the theor}^ of successive uplift 

 and formation of rain screens, as this alone would account 

 for the retreat of the ice and the remnant flora. From this 

 we are justifled in concluding that the flora has not ti'avelled 

 acro.'-s from ranofe to ranoe, and therefore that it has 

 ti-avelled either down or up the ranges (or both ), and hence 

 has been derived from a common source. As it stands, 

 the theory is suflicient to account for all the facts of dis- 

 tribution so far as I know, without dragging in the highly 

 im])i-obable idea that the Mekong valley is not a physical 

 Ijarrier to plant migration. 'Die gradual desiccation of the 

 Mekong-Yangtze divide would bring about changes in the 

 flora, particularly in the forest belt, rain being, as already 

 pointed out, a greater controlling factor in the case of forest 

 than it is with a herbaceous iloi-a, and it accounts readily 

 erifjugh for the remnant flora. It also accounts for a peculi- 

 arity alluded to above, namely, the greater specific variety 

 met with amongst many alpine genera on the Mekong- 

 Yangtze divide, e.g. Mecouojfsls, Geididua, Saxifraga 

 R/iof/of/endron, Fcdicida.riM, etc., a variety greater than 

 anvthiiif; met with on th(; Mekong-Salween divide ; for as 



