or WJ^STEKK SOUTH AMERICA. 149 



when siifncicnt rain lias fallen, are speedily withered, and even 

 perennials are often unable to survive the long intervals of 

 drought. On tlie other haud, the eJimatal conditions are emi- 

 nently favourahle for the preservation of the vitality of many 

 seeds which may be carried by the wind over wide spaces. One 

 result, which has been frequently verified by observations in 

 Northern Africa, is that at any given spot the vegetation is 

 subject to singular variation in successive seasons. Some species 

 disappear, while others, not before seen, exhibit themselves, and 

 after a few seasons again give place to the old residents or 

 to other new comers. The other result is that over a large area 

 the flora is mainly composed of relatively few identical species 

 which spread far and wide. Of the plants inhabiting the vast 

 zone of desert, extending from Beluchistan to the Atlantic coast 



of Marocco, south of the Great Atlas, the large majority are 



found at intervals throughout the entire area. It would be of 



some interest to ascertain how far the^^e conclusions are verified 



in the desert region of western South America, which extends 



northward, from about 27° S. latitude in Chili to about 4° S. 



lat. in Peru, through a distance of about 1700 English miles. 



The conditions are in some respects so different that we should 



expect tlie phenomena to be here less marked than in Norlh 



Africa or Australia, lix the first place the wind is much less 



powerful, rarely exceeding in force a gentle breeze, and it sets 



uniformly in the same direction from south to north. Hence 



this agency for the dispersal of seeds is less cfiicient, and wliile 



we should expect to find a good many species carried from Chili 



along the sea-board of Peru, this mode of dispersion cannot act 



in the opposite direction. Further than this, the desert zone is, 



for the most part, a narrow strip running parallel to the chain 



of the Andes, and many Andean types of vegetation have been 



modified to suit them to the climate of the we:*tern sea-board. 



From this source we shouldexpect in the coast-flora a less amonnt 



of uniformity than is usual in desert-floras; but it seems likely 



that the causes which give rise to a great degree of variation in 



the vegetation of each particular place will act here much as 



they do elsewhere. To test this conclusion it would-be well if 



travellers landing at various points on the coast would register 



the species observed at each place. Excepting the rare occasions 



when numerous plants enjoy a short-lived existence, the nnmb(?r 



of species to be observed is so small that the observation would 



give little trouble. 



m2 



