8 TTIE GEXrS MAXETTIA 



I have ventured to suggest that Ivubiacete, with the alhed famiUes 

 that together compose the order Rubiales, reflect a general tendency 

 to close aggregation of flowers into a continuous surface, thereby 

 facilitating cross-pollination by exposing a large number of flowers to 

 the effects of a single insect- visit — the so-called " umbellittoral " 

 tendency (op. cit. xi. 222, 382, etc.). 



From this aspect Manetfia may be regarded as relatively primitive, 

 reflecting but little the i-ealization of this tendency ; for in most of 

 the species the flowers are mingled more or less indiscriminately 

 among the leaves, and are not aggregated into very deflnite units of 

 inflorescence. This primitive arrangement — or rather lack of arrange- 

 ment's readily observable in the Pyrrhanthos-section, in which the 

 flowers are relatively " few and far hetween." But even in this ease 

 the leaves in the neighbourhood of flowers tend to become much 

 smaller than those composing the general vegetative bulk, and they 

 tend also to a uniform shape — often approaching the orbicular. The 

 flowering branches, too, tend to be much abbreviated, and also to 

 aggregation. These points are clearly shown in the herbarium- 

 specimens of M. cordifolia and its varieties — and, indeed, throughout 

 the Pyrrhanthos-section, in which insect-adaptation is attained chiefly 

 through enlargement of the individual flower (see New Phyt x. 79, 

 etc.) accompanied by reduction of neighbouring leaves — the foliage 

 being cleared, so to speak, to make way for the flowers. The general 

 transition from vegetative leaves to bracts and bracteoles is clearly 

 discernible in a comparative study of these forms ; although in some 

 cases the bracteoles are connate into small sheaths about the pedicel, 

 being probably of stipular nature. 



Still more primitive in this regard are most of the Heterochlora- 

 species. In these there is little or no sign of floral aggregation, and 

 the evidence of leaf- reduction is very faint. The principal indications 

 of any tendency to insect-adaptation are 1st, enlargement of indi- 

 vidual flowers, culminating in the inflated corollas of J/. JRojasiana ; 

 and 2nd, lengthening of the flowering stalks, the flower being carried 

 thus more or less free of the leaves ; this feature is also well reflected 

 in M. Rojasiana. 



M. calycosa, which in many ways represents the transition between 

 § Heterochlora and § Lygistum, display's a very decided tendency to 

 floral aggregation. The relatively small flowers are borne, several 

 together in the axils, in an abbreviated, more or less umbellate 

 inflorescence. 



In the large section Lygistum the flowers are invariably aggi*e- 

 gated, two or three together at least, into axillary inflorescences. In 

 the species that are least advanced in this respect the inflorescence 

 is considerably exceeded by the leaves. Such is the case, for example, 

 in M. Lygistum itself and its nearer allies. This covering of the 

 floral by the vegetative parts, and the consequent interference with 

 seed-dispersal set up by the evergreen leaves, goes some way to account 

 for the remarkable limitations in the distribution of nearly all the 

 species to which we shall draw attention later ; and this in spite of the 

 efficient wing-mechanism possessed by the seed for transference over 

 a considerable distance. 



