SECTION IV: ANALYTICAL KEY TO SPECIES 245 



awl-shaped ; seeds pale brown, quite naked, rounded or 

 not angled 37. 0. taitense . . . . (p. 248). 

 t2 Leaf-stalks warted, rarely ciliate, lobes of the leaves arching 

 outwards ; calyx teeth 3-veined, but at most acute, never 

 tailed; seeds rounded, beaked, and often with very slight 

 fuzzy tuft on the apex 



38. G. purpurascens. . . . (p. 250). 



** Tiuigs faintly angled on the young growing parts and pedicels, 

 green or brown coloured ; leaves ovate-oblong, more or less cordate, 

 entire or 3-5 deeply palmately lobed ; flowers large, nearly twice 

 the length of bracteoles or more, petals pure yellow but with 

 conspicuous purple claws. 



t Leaves pilose-tomentose below, lobes somewhat irregular but 

 more or less ascending, broad, usually only 3, and the central 

 one often much longer than the laterals ; flowers frequently 

 three times the length of bracteoles, and the fruit 3-, 4-, often 

 5-celled ; seeds free, brown, glabrous 



39. G. vitifolimn (p. 255). 



t2 Leaves very nearly glabrous, half or more segmented into 



3-5 spreading oblong acuminate lobes ; flowers hardly 

 more than- twice the bracteoles, and fruit 3-, 4-, mostly 3- 

 valved ; seeds free, brown, almost glabrous 



40. 0. barbadense. . . . (p. 265). 

 t3 Leaves almost quite glabrous, and in herbarium specimens 



usually drying into a brown colour, deeply cordate, very 

 broad and large, cut into five palmately spreading acumin- 

 ate lobes, the central one the longest, and often from 4 to 

 8 inches in length ; flowers very large, and braeteoles some- 

 times maculated on the inner surface ; fruit 3-celled, and 

 seeds almost quite naked, striated, and united into a kidney 

 mass 41. G. brasiliense. . . . (p- 295). 



If the two assemblages of what have been called fuzzy -seeded Assort- 

 cottons manifest many perplexing problems as to their origin and ^"* y 

 the limitations that have to be placed on the species, varieties and 

 races known to exist, these difficulties are increased tenfold when 

 the naked-seeded cottons, that it is here contemplated to review 

 collectively, are taken into consideration. One point, however, 

 stands out fairly vividly, namely, that when a record of our 



