54 



GEOGRAPHICAL DISTRIBUTION 



from the American and European meridians as well. The presence of a few forms on 

 the Siberio-Caucasus meridian, which at first appears to invalidate this conclusion, on 

 closer examination confirms it. For when we examine the distribution of the individual 

 species, it is found that the one which extends westward to Persia (P. siphonanthaf is an 

 example of the few species that have become sub-Alpine as well as Alpine, and as such 

 has the zonal distribution characteristic of its class, and the four that extend to Siberia 

 are clearly examples of forms attracted to adjacent mountain flanks during a period of 

 retreat. More striking still, only one of these (P. longiflora) has yet overflowed the 



d 



I 



liffei 



mountain barrier to the slopes and plains beyond. 



The Riiyncholoph,e are, as diagram 7 shows, very generally distributed, the most 



striking features of this distribution being 



(1) the high percentage in the American meridian; 

 2) the uniform percentage along the whole extent of the Siberio-Caucasus 



the Japan-Himalayan meridians; and 

 (3) the low percentage of the Siberio-Caucasus meridian. 

 The uniformity along the Siberian and Chinese meridians, coupled with the marked 



between the two, corroborates the whole of the other evidence as to the 



distinctness of these meridians. The high percentage of the American meridian, coupled 

 with the fact that it is on this meridian that the highest type of Rhyncholophous corolla 

 occurs, seems to indicate that here this section took origin. Its presence both in Europe 

 and in the Chinese meridian in proportions which, while not quite equal to that in 

 America, are nevertheless high, strongly supports this view, and we have already seen 

 that, so far at least as the Himalayan province is concerned, there appears to have been 

 during one or other of the periods of furthest advance, zonal communication between the 

 equatorial end of the American meridian and it. But the high percentage in the Cir 

 cumpolar province shows that it is probable that some at least of the Rhyncholophous 

 forms were evolved at a comparatively remote date and were conducted to the neighbour- 

 hood of the pole during a period of retreat anterior to the last, and that, issuing thence 

 during the last southward migration, they were not only redistributed in America but in the 

 European meridians as well. The ancient cleavage already alluded to more than once 

 as existing between Greenland and Alaska made this impossible so far as the Chinese 



> 



- 



meridian is concerned; it is therefore a very striking corroboration of the general 

 hypothesis to find that the European Rhyncholophae are all of comparatively simple 

 type {Bostratce), as from their polar derivation they might be expected to be, while in 

 the Himalaya there are but two or three examples of this type, though many are almost 

 as highly specialised as the American forms themselves, and are, moreover, closely allied 

 to them. The low percentage in the Siberian meridian seems to show that the last 

 southward migration commenced before many forms had extended eastward by Lapland 

 to Arctic Siberia, while the barriers of an Aralo-Caspian Sea to the west, and of an 

 unbroken mountain chain with a more rigorous climate along its summit to the east, 





6 v,*^~o w^ ~ 



prevented zonal communication with Europe and China respectively. 



The distribution of the ADUNC-ffi as a whole, which is shown in diagram 8, does 

 not afford evidence that is either direct or conclusive. This is because the two sections 

 that compose them, besides having been, like the sections of LONGIROSTRES, evolved in- 

 dependently, have, unlike Siphonanth^e and Orthorrhynch^:, originated at a different 



1 P. tiphonantha extends from Elbarz on the west through Afghanistan, Kashmir, Kamaon, Nepal, Sikkim, and 

 Cbumbi to Szetsehuan. 



