2H" 



the circumpolar species are 



the endemic factor thus becomes ff , 



GEOGEAPHICAL DISTEIBUTIOX 



confined to the American section of the Circumpolar province 



i 



or 75*4 per cent 



dently a less natural result 



than 

 be a 



21 



24) 



87'5 per cent 



•> 



dded to the Jap 



the result in the table. If the Kurile-Kamstchatdale species 

 total of 12 with an endemic factor of ffc, or 56*9 per cent., 



htained-again a less natural r 



esult than the if, or 66*6 per cent 



of the table 



If 



the species 

 Siberio-Turkestan 



that occur in Arctic Siberia are considered along with those of the natural 



province, the total 



eached is 57, the total endemic 34. This gives an 



endemic factor of ff, or 59'6 per cent.-ag 



a less natural result. Finally 



> 



of the table 



i 



those 



addition to the 



per 



cent 



cent 



as compared with the f J, or 65 per cent., 



f the Arctic European species are added to 



the total is raised to 43 without making any 



This makes the endemic factor ff, or 76 '7 

 most clearly a less natural result than that of table, which is f£, or 85 per 



f the true Europ 



province 



i 



ruber of endemic species 



It is not necessary to discuss each of the provinces in such detail 



may 



be 



is sufficient to show how this 



the statistical tests that have in every case been applied 



added to the European, the conjoint total becomes 55 



One other example 

 ded to all the areas and to indicate 



If th 



C 



specie 



be 



the 



demic total 45, or no 



gain 



in 



numbers from the union. But the percentage for the whole area now becomes 



shown exists in the true European province. 



endemic 



bee 



Hi -8 in place of the 85-0 which it h 



If they be added to the Siberio-Turkestan species, the total becomes 65, the 



total 44, or only a gain of 



plac 



of the percentag 



they be 

 results. 



on the whole area. 1 This gives a percentage of 67*6 in 

 of 73-3 shown by the Caucasian province already. Finally, if 



total of 116 with an endemic total of 101 



added to the Himalayan species, a 

 This 



again causes no improvement over th 



whole area, but gives it 



percent 



age of 87-0 in place 



of the 89*1 already shown to exist in the Himalayan province 



Besides this brief review of the general distribution of the genus 



j 



it 



necessary to 



srive a more 



detailed account of the distribution in the Himalayan province in which all 



the species described in this paper are found 

 able within it: 



Three 



y 



natural divisions are recognis 



1. The Himalayan division proper 



This includes on the west the Hindu-Kush 



and ends in the east at the valley 



through which 



the Sanpo passes to 



2 



reach the Brahmaputra. It extends from 68° E. Long, to 95° E. Long., 

 passes insensibly into the tableland of Tibet to the north, and is abruptly 

 defined by the plains of India and the valley of Assam on the south. It 

 contains 61 species, 43, or 70"5 per cent., of which are endemic. 2 



s extends from the Brahmaputra on the west 



The Indo-Chinese division 



Th 



to II 



the east 



between 9 



-o 



E. Long. 



and 110° E. Long 



> 



and 



is somewhat abruptly separated from Ch 

 valley on the north, while it passes inse 



Prop 



by the Yang-tse-kiang 



insensibly into the uplands of Assam 

 and Burma on the south. It contains 47 species, of which 35, or 74 4 

 per cent., are endemic. 



— — 



1 Even this gain of one is doubtful, for (cf. Maxim., Mel. Biol, x, 120) it is not certain that the species in question 

 (P. achillea/olia) exists in both regions. 



This division of the Himalayan province is almost exactly equivalent to the Emodic or " Wallichian" region of 

 Schow's system of Phytogeography. 



