26. PEOCEEDUfGS OF THE 



coast and the mountains. I ought to repeat that I excluded 

 from this list of 69 plants, those Sooudreebun plants which extend 

 over the plains of Bengal ; and, of these generally distributed 

 plants, many reach the hills. 



As to the plants found in sandy places inland, we may regard 

 them as arenaceous (as well as maritime) species. 



I have found that one important point in any tabulation is to 

 leave oneself no options. The extent to which the personal 

 weakness of working up to a theory desired to be established 

 vitiates tables of geographic distribution is very large. 



Por example : — I will assume the Shepherd's Purse to be a 

 tolerably well-defined species. It will be at my discretion in 

 many areas to reckon this plant present or not. I may define 

 the word " wild " as I like : if I say that a plant is wild if brought 

 by a bird though the suitable cultivated soil has been prepared 

 by man, and though the plant may also have been introduced by 

 man, then it becomes an option with me in how many of the 

 large areas of the world I locate the Shepherd's Purse. The 

 case of variable and critical plants still further enlarges my 

 option. Take the Eubi : I may make the English species 7 or 

 70, which will entirely alter the numerical results of my tabu- 

 lation. Moreover, in these critical plants the distribution of a 

 species cannot be got from herbaria because only an adept in the 

 group recognizes or collects properly. Pocke got several species, 

 new to the South of England, in a three-mile walk out of 

 Bournemouth. Purther, the naming of these critical groups 

 differs among botanists ; so that attempts to correlate them from 

 the names in botanic lists only leads to confusion and illusion. 

 This source of error applies not only to critical, but to ditfieult 

 plants ; no two plants can be more definitely distinct specifically 

 to me than the JEleocliaris palustris, R. Br., var. uniglumis, and 

 the Eleocharis multicaulis, Sm. I put them in two difterent 

 sections ; but they are confounded in herbaria, and still more in 

 the books, in such a way that to enter them in tables of geo- 

 graphic distribution would merely mask the inferences at which 

 we should arrive by omitting them altogether, except, of course, 

 in so far as I confine myself to examples honajide collected and 

 numbered in accurately stated habitats. 



Among the 69 characteristic Soondreebun plants I have included 

 5 very critical sedges, because I happen to know them so well in 

 the field that 1 was confident that 1 could name the herbarium 

 examples accurately ; and their distribution in the table comes 

 out normal, i. e. satisfactory. 



It is evident that with accurate knowledge of the species the 

 number of 69 species which 1 have worked with may be greatly 

 increased. 



In investigating the geographic distribution of the English 

 Plora, I ventured at our last evening meeting to suggest that we 

 might in the present state of our knowledge get at some results 

 by working with a limited number of well-defined, generally 



