376 NATURAL HISTORY OF PLANTS. 



The disk linlnir the receptacle is sometimes thin and inconspicuous ; 

 a«Min, the «,dands, into wliich, as we have seen, its edge is split, may 

 become very prominent ; tliey may be quite free or united in pairs. 

 This disk usually stops abruptly below the insertion of the stamens; 

 but in several herbaceous species the stamens are inserted, not only 

 within its edges, but all over its surface, right down from the base of 

 the perianth to near the insertion of the gynajceum.' The number 

 and position of the carpels are very variable ; they may be equal in 

 number to the petals and superposed to them, as above; or double 

 their number, lialf being superposed to the sepals.* But, strange to 

 say, those in front of the petals may then disappear, leaving only 

 those superposed to the sepals/ Finally, their number may be in- 

 detinite, or it may diminish to less than four or five, and even be 

 reduced to one or two/ Rarely are the carpels not quite free from 

 each other, and sometimes the organic apex of the receptacle will even 

 rise up into a little cone and separate the ovaries/ But sometimes 

 the latter are all united for a variable height, so that a transverse 

 section of the lower part of the gynseceum will show a single several- 

 celled ovary with axile placentation/ The ovules are not invariably 

 indefinite, and horizontal or only slightly descending. There are 

 sometimes only two, or even a single one,' descending and completely 

 or nearly anatropous, with the micropyle upwards and outwards. 



any case so ill-marked that it is not worth while pendula. S. lohata has often eight or nine 



to take it into consideration. The floral bracts carpels, with a tetramerous perianth. 



are here carried up very high on their axillary •* In -S'. Lhidleyaixa, for instance, as pointed 



pedicels. out by llcEPEE. This also occurs in S. sorbi- 



' Thus, in S. ZoJate the stamens are inserted folia, the type of the section Sorbaria (Seb., 



as above, over the whole surface of the recep- loc. cit., 545), or Schizonotus (Lindl., Wall. 



tacular disk. In S. Ulmaria, Vacchee has de- Cat., n. 703). 



nied the existence of a disk, and supposed that ■* As in S. Arunctis, the type of the section 



the stamens were hypogynous. But they are ^n<»«;5 (See., ?oc. cjY., 545), which, however, has 



inserted on the periphery of the receptacular more frequently three or four ; also in the allied 



8ttc, and there is a yellowish disk with its frilled species, wliich are most probably only forms of this. 



etlge just internal to their insertion. In S. Fill- These plants have usually unisexual liowers. 



pendvla the stamens are placed at different » This is pretty well marked in <b'. Filipendula, 



lieights on the inner surface of the receptacle. deciimbem, &c. 



Hul the outermost are inserted a good way below ^ The species where this union takes place to 



the petals. The latter have articulated bases each its fullest extent is S. Lindleyana, whose ovary 



Hcatcd just in the bottom of the sinus between in this respect recalls that of VauqueUnia, while 



two adjacent sepals. The shallow recejitacle of its carpels are also alteruipetalous ; so that -S'. 



.S. Aniucus is lined by a glauduhir disk, whose Zindlei/ana links VauqueUnia and Spiraia. 



edges arc indistinctly lobed ; but there are none More frequently the carpels of the latter genus are 



of the prominent isolated or geminate marginal only united for a very little way above the base. 



glands found in most of the other species. In ? There are usually only two in .S". Filipen- 



A'. J«o/-/w/(///V/ the margins of the intra-receptacular dida,a\\CL they become almost superposed. S. 



glandular layer are nearly entire, as they often Arunctis. Icevigata, &c., have often two pairs of 



are in Frioifi/na. descending ovules. S. lolata has two ovules, or 



2 There arc as many jis fifteen in .S'. Fill- more rarely a single one. 



