SECTION III : G. MEXICANUM 229 



and 1499), named G. barbadense, are in reality pictures of G. mexi- 

 canum, and are thus neither G. barbadense nor G. purpurascens. 

 I reproduce by three-colour photography one of Eoxburgh's pictures 

 of what I take to be this plant in Plate No. 40. 



In the botanical department of the British Museum there is 

 a long series of MS. coloured drawings also of this plant, prepared 

 apparently at Saharanpur Botanic Gardens, but of these no history 

 exists. It is thus quite evident that Eoxburgh, and subsequently 

 the authorities of the Saharanpur Gardens, had grown the true 

 G. mexicanum in India. It has moreover often been alluded to by 

 recent Indian authors : see Wight's papers (' Agri. Hort. Soc. of 

 India,' old series, vol. vi., 189-196; vn., 194-215). Dr. Spry, in 

 fact, considered it as specially suited for cultivation in Behar, but 

 there is no record of success hitherto attained anywhere in India. 



It is a very distinctive plant, and in recent years (as represented 

 by the Upland cottons) has become perhaps the most important of 

 all stocks. But while every herbarium possesses examples of it 

 I have been unable, as already stated, to discover a record of its 

 having been gathered from a wild or acclimatised source. All the 

 specimens seen by me are stated to have been cultivated plants, and 

 it would not surprise me very greatly were they proved in the future 

 to have originated recently and to be alone maintained under cultiva- 

 tion. They are hybrids derived very possibly from G. punctatum or 

 G. hirsutum as the one ancestor, crossed with G. purpurascens for 

 one series of forms and G. vitifolium or G. barbadense for another. 

 There are many groups recognised by the growers (which I shall 

 presently discuss), but in general terms these may be assorted into 

 pilose forms, where the G. hirsutum element predominates, and into 

 glabrescent states, where the G. purpurascens has become the 

 dominant influence. It is the latter group of hybrids that has come 

 to bear the name G. mexicanum, Tod. But taking another conception 

 they may be assorted into broad, compact leaved states, as shown 

 in Plates Nos. 40 and 41, or more deeply-lobed conditions, as in 

 Plate No. 42, the last being due very possibly to a strong strain of 

 G. barbadense. 



I have on more than one occasion warned my readers that when Muta- 

 dealing with highly cultivated forms of cotton it would be most 

 unwise to suppose the existence of definite species, in the more rigid 

 botanical acceptation. Whether this variability be primarily a con- 

 sequence of hybridisation (as I am strongly disposed to believe may 



