Arctic Plants: Geographical Distribution 103 b 



farther south, for instance, in the Altai and Baikal mountains, Davuria, and 

 Kamtchatka. On this continent it does occur on the arctic coast, but only at 

 a very few stations; from there it extends south to the Rocky mountains in 

 Colorado. Although a near ally of S. tricuspidata, the geographical distribution 

 of these does not coincide, for the latter is common in Greenland and on the 

 islands of the arctic American archipelago; it has also been recorded from the 

 Appalachian mountains and the Hudson bay region but not from any stations 

 in the Old World. We have thus two closely allied species, one of decidedly 

 Siberian origin and another, confined to this continent, but both occurring in 

 the arctic region and, so to speak, meeting each other on the coasts of Bering 

 strait. The present distribution of S. tricuspidata, notably in Greenland, where 

 it abounds between 60° and 78°18' N.L. on the west coast, and from 60° to 76° 

 N.L. on the east coast, besides that it seems to be quite frequent on the arctic 

 islands, makes the suggestion very probable, that the centre may be sought there, 

 in the north eastern arctic corner of this continent. 



A very strange distribution is exhibited by S. aizoides, viz. : many stations 

 in Europe, from the arctic region (Spitzbergen and Scandinavia) south to the 

 Pyrenees and the Alps of Italy and Switzerland; in America from the arctic 

 coast south to 51° N. L. in the Rocky and Selkirk ranges, but absent frofn Asia 

 except a very few stations in Nova Zembla and Ural. As far as concerns the 

 arctic area covered by this species in Europe compared with the corresponding 

 latitude on this continent, including Greenland, the American area is consider- 

 ably Tyider. But the much larger area occupied by the species in temperate 

 Europe, besides the more frequent occurrence of the species, does indicate, I 

 believe, that the species originated there, and possibly in the Alps or Pyrenees. 



Being almost circumpolar, the only districts uncovered being arctic Scan- 

 dinavia and the Russian coast, Saxifraga flagellaris may nevertheless be con- 

 sidered as a truly arctic type which originated in the polar regions. It is widely 

 distributed on the arctic coast of this continent, including Greenland; it is 

 known from Spitzbergen and Nova Zembla, and from several stations on the 

 Siberian coast. Farther south it is known from the Rocky mountains (Colorado), 

 Caucasus, Altai and Baikal mountains, and the Himalayas. However, it is 

 true that most of the species of the section Trachyphyllum are Himalayan, as 

 pointed out by Engler (I.e.), and, moreover, this peculiar little species is in the 

 Himalayas associated with two others, S. Brunoniana Wall., and S. pilifera 

 Hook. fil. et Thoms., which exhibit exactly the same habit as S. flagellaris, viz. : 

 ' filiform stolons above ground being developed from the axils of the rosette- 

 leaves, and terminated by a ^all, leafy rosette. But it would be difficult to 

 imagine that these mountains, the Himalayas, should be the actual centre of a 

 distribution so enormously wide in the polar regions. I am more inclined to 

 beheve that S. flagellaris, like most of the circumpolar species, originated in the 

 arctic regions, and that it reached the Himalayas by way of the Altai and Baikal 

 mountains. The association with the two analogous types may simply indicate 

 a recent geographical centre of this interesting little group of species, extending 

 as far west as Caucasus. 



The remaining species of Saxifraga: S. rivularis, S. cernua, S. Hirculus, 

 S. nivalis, S. hieraciifolia, S. decipiens, and S. oppositifolia are circumpolar, 

 and of these S. Hirculus is the only one which has no immediate allies outside 

 the Himalayas, and the species itself is also widely distributed in these mountains, 

 ranging from western Tibet (4,500-5,000 m. above sea-level) to eastern Hima- 

 layas, at the same altitude. Considering the fact that the section Hirculus, 

 containing 11 species, has 10 of them endemic to these mountains, it is natural 

 to suppose that the Himalayas constitute an important, or perhaps the only 

 centre, where the section has developed further. But the actual centre of dis- 

 tribution of the species S. Hirculus may not have been located there; it seems 

 more natural to suppose that similar to most of the other circumpolar species 



