IMPROVEMENT OF STOCKS 345 



G. barbadense, variety Sea Island (see 53, f. 11), and in G. peruvianum Bat- 

 (see 53, f. 20) and is met with also in Althaea. But there are many tres *?? d 

 modifications of it, or rather degrees of participation in the formations tion. 

 indicated. In G. arboreum the warts are small but numerous, 

 compact, distinctly striated, opaque, and the surface of the spore 

 conspicuously granular. (Plate 52, ff. 7 and 10.) In some species, on 

 the other hand as, for example, G. hirsutum and most of the Upland 

 Cottons the warts are often very obscure, or at least their striations 

 (and buttressings) are so, if indeed these structures might not be 

 described as entirely absent. (See Plates 52, f. 6, and 53, f. 13.) 



(&) The second chief condition is that in which warts and some- Palisade 

 times also spines are nearly or entirely absent. The latter, according cpndi- 

 to some observers, are caducous, and the exine is then seen to consist 

 of a uniform compact layer that, as Schacht observes, recalls a 

 palisade structure. Mr. Slater, however, seems to be doubtful whether 

 the spines are ever caducous, and in support of this opinion mentions 

 the circumstance that although he has examined many hundred 

 specimens of Gossypium he has never observed a detached spine. 

 The present condition is strikingly seen in G. drynarioides (Plate 52, 

 f. 1), where the palisade layer is very thick a condition that would 

 appear to be almost generic in Pavonia. In many of the races 

 attributed to G. Nanking (Plate 53, f. 14, and in G. Palmerii, Plate 

 53, f. 17) a remarkably thin palisade layer occurs which is also pre- 

 valent in Adansonia and Thurberia (of. p. 53, 63). In the roji cotton 

 of Gujarat an almost intermediate condition exists (Plate 53, f. 12) 

 in which a palisading layer appears like a union of the buttressings 

 of the spines. 



(c) A third condition is that in which both intine and exine are Hyaline- 

 distinct and hyaline, no trace of striation or palisading is visible, and ^^Q 0on 

 the long sub-linear spines seem as if they passed through the exine. dition. 

 This is well seen in G. Sturtii (Plate 52, f. 2) ; in G. Darwinii (f. 3) ; 



and to some extent is simulated in roji cotton (Plate 53, f. 12). 



(d) A fourth condition is that in which the warts are numerous, Opaque- 

 opaque, trapezoid in shape, suddenly tipped by long sharply pointed ^rTcon- 

 spines. This occurs in G. Stoclcsii (Plate 52, f. 4) ; in G. obtusifolium dition. 

 (ff. 8 and 9, also in Plate 53, f. 15) ; in G. Kirkii (Plate 53, f. 21) ; is 



to all intents and purposes the condition that is met with in 

 G. brasiliense (Plate 53, f. 18) ; and lastly seems the generic 

 peculiarity of the pollen-grains in Cienfuegosia. 



Thus the thickness, shape and size of the sculpturings or surface 



