PITTIER TREATMENT OF THE GENUS CASTILLA. 255 



as in the common fig. It is usually provided at the base with a more 

 or less complete involucre of free or subconnate bracts, while the re- 

 mainder of the outer surface of the receptacle is covered with more 

 imbricating bracts, which, for convenience sake, we have called 

 " scales." On the whole, bracts and scales hardly seem to furnish us 

 with any good diagnostical characters. 



Of the six rubber-producing species of the Isthmus, half (C. lactifaa, 

 (', guatemalensis, and 0. panamensis) have emarginate or kidney- 

 shaped male receptacles, while the remainder do not present that 

 peculiarity, on account either of their irregular shape (C. costaricana, 

 C. elastica) or of their mode of dehiscence {C. nicoyensis) . 



The stamens are inserted on the inside surface of both lobes of the 

 receptacles. Most authors seem to have adopted the view that they 

 are distributed without any regular arrangement, and that the bract- 

 lets, scales, or floral leaflets that accompany them are all small and 

 isolated. In fact, the disposition of the stamens is dependent on the 

 continuous blades or lamella 1 that radiate from the bottom of the 

 receptacle and branch out so as to cover the increasing staminal area. 

 These blades vary in breadth and in the indentation of the margin 

 according to the species, being, for instance, small and irregularly 

 developed in C. guatemalensis, and rather broad and conspicuously 

 lobulate in C. elastica. The stamens are inserted singly along these 

 blades, or in clusters in the axils or at the ends of their ramifications. 

 The bractlets are very unequally developed in the same receptacle, 

 many stamens being destitute of them, others showing several, while 

 at times bractlets appear where there are no stamens at all. The 

 bractlets also seem to vary in size according to the species; they are 

 few and small in C . jallax, ('. nicoyensis, and C. panamensis, numer- 

 ous and almost exuberant in C. elastica. They are generally inserted 

 at the base of the stamen, but in the last species the filament is often 

 provided with an additional foliaceous, accumbent appendage, at- 

 tached somewhere below the anther. In the same specimens remark- 

 able transitions of a regressive nature occur between stamens and 

 bractlets. Sometimes the bractlet is reduced to an awkwardly long, 

 cylindrical filament, ending with the rudiments of an anther; in 

 other cases the still defective anthers appear on a foliaceous, bract- 

 like filament ; and besides these there are perfect anthers borne on a 

 flattened, broad filament, etc. All these anomalies have been noticed 

 in C. elastica, the other species investigated being mostly normal. 



The stamen itself consists of a rounded, always smooth filament, 

 ending in a claviform connective, on both sides of which are inserted 

 the narrow, elongated anther cells. The connective is in all cases 

 basifix and it is evidently a misinterpreted observation that makes 

 Trecul say that the anther is dorsifix or peltate. The variations in 

 57441°— vol 13, ft 7—10 2 



