FLORAL STRUCTURE AND AFFINITIES OF SAPOTACE.E. 67 



petal. In tlie adult flower these appendages are so arranged that 

 each appears to pan- with its neighbour of the next petal, so that 

 they form paria sepalis superposita. But in all the species of 

 these genera that I have dissected, fresh or diy, the true petals 

 were invariably and exactly alternisepalous, a relation seen by 

 Eichler, but contradicted or passed over by most other writers. 



Andr(ecium. — In all the species examined fresh, and probably 

 in a vast majority of the species of the Order, the androecium 

 is truly diplostemonous. Next the petals comes an equal simul- 

 taneous alternating whorl of stamens, and then is formed an 

 alternipetalous set internal to the former." In Mimusops and 

 Achras the alternipetalous set become sterile staminodes ; in 

 Chrysophyllum they appear as tubercles, but soon abort altogether, 

 leaving no trace. I have seen a third whorl in one or two adult 

 flowers of Bassia longifolia. 



In Lahourdonnaisia (d), where the corolla is biseriate, the two 

 sets of stamens continue the alternation of the petaline whorls. 

 In Payena [d) (including Cacosmcmthus and Ceratophorm), the same 

 is the case ; but then a thnd doubled inner whorl of stamens is 

 formed, alternating with the individual members of both sets f (see 

 formula below.) 



The androecium requires further study in certain species of 

 Bassia | and Pycnandra, Cryptogyne and Omphalocarpum. Through- 

 out the Order the petals soon become connate with one another 

 and the stamens by the upgrowth of a podium or " corolla-tube." 



Gyn^cium. — The carpels form a single verticil, and are usually 

 isomerous and alternate with the innermost whorl of stamens. 

 In Achras, however, (contrary to Eichler's conjecture) the carpels 

 alternate with the stamens and staminodes together ; and in 

 Chrysophyllum Caimito (/"), some of the intervals between the 

 stamens are occupied by two carpels, or rather some of the carpels 

 develope in the intervals between the still visible abortive staminal 

 tubercles and the fertile ones. A similar explanation probably 

 fits the pleiomerous pistil of Lucuma grandifiora, Labramia, &c. 

 Omphalocarpum needs working out. The oligomerous pistil of 

 certain species of Lucuma, &c., is most probably due to true 

 abortion. 



The special development of the pistil deserves further notice. 

 The carpels, at first hemispherical, soon become crescentic. The 

 cusps, which run inwards, soon unite with their neighbours, but do 

 not reach the centre of the receptacle. The united carpels then 

 grow vertically upwards to form a tube, entire externally, and 

 divided inside by the incomplete radial septa formed by the carpel- 

 lary commissures into as many elongated chamberlets opening into 



* In the other species it was easy to observe the successive appearance of 

 these -whorls; but in Mimusops I had to infer it from the inner set being 

 the smaller in the very youngest buds, though soou equalling or outgrowing the 

 others. 



+ I had conjectured this from an inspection of Wight's figure of Payena 

 lucida {Isonandra polyandra, Wt., Tcones t. 1589) ; and a dissection of every 

 species in the Kew Herbarium has confirmed tbis amply. 



+ Eichler conjectures that the pleiomery is here due to '^ dedoublement." 



