128 CALIFORNIA ACADEMY OF SCIENCES. 



rsomewhat less than one half of the species under considera- 

 tion. The others have in no instance the perianth-pattern 

 of either of those genera; are never, like them, merely tri- 

 androus; and their anthers are in no instance adnate. Along 

 with considerable variability in the shape of the perianth, 

 they display always six perfect stamens with versatile an- 

 thers. There is, moreover, a striking peculiarity in the way 

 in which the filaments are joined to the tube of the peri- 

 anth, and. that is of the following description : the filament 

 is slender and the upper part free, more or less; the adnate 

 portion inconspicuous down the upper part of the tube, 

 reappearing toward the base in the form of a thin but prom- 

 inent crest. The species, however closely agreeing in 

 habit and in the points of floral structure thus indicated, 

 are diverse to a troublesome degree in the relative propor- 

 tions of the tube and limb of the perianth, and more espec- 

 ially in the structure and attachment of the androecium. 

 The three or four species representing the very extremes of 

 this diversity were, singularly, those which fell first into the 

 hands of botanists, and each of these was very naturally 

 and, under the circumstances, quite logically taken to be 

 the type of a genus; and so there was Triteleia, seeming to 

 approach Brodiwa by its broadly tubular perianth: Seuher- 

 tia, in which the tube is attenuate below and the internal 

 crests very strongly brought out; Calliprova, in which the 

 cristiform reappearing of the filament quite fails, but is com- 

 pensated for by an alar dilation of the upper free part of that 

 organ; Hesperoscordum, in which the whole perianth is open 

 campanulate, and the filaments dilated and monad elphous 

 below. This last has, in my opinion, better claims than any 

 of the others to separate generic rank. A year ago I should 

 probably have insisted on its restoration. But the past 

 season's collecting has yielded us a second species whose 

 filaments are not at all dilated, but simply and singly adnate 

 to the perianth for one half their length. Morphologically 

 there is nothing in these two plants to keep them out of 



