A REVISION OF THE GENUS BERTIERA 113 



None of the Madagascar species, for instance, have been found 

 in the other Mascarene Islands, nor vice versa. One species, 

 B. gonzaleoides, is endemic m Cuba, but the genus is absent from 

 the other islands of the West Indies, and, except for one 

 Nicaraguan species {B. oligosijerma), elsewhere in Central 

 America. A species described by Bentham, B. angustifolia, was 

 collected in the isolated Cocos Island, on the Sulphur expedition. 

 The American B. guianensis, the most widely spread of any 

 of the species, occurs from Panama to Trinidad and Guiana 

 in the north to Bolivia and Barra in the south and south-east, 

 respectively. 



On the African continent the genus occurs as far north as 

 Senegambia {B. labiata), but not south of the Congo ; Upper 

 Guinea is the area richest in species; and the most easterly 

 points of distribution are Niamniam-land (B. cBthiopica) and 

 Uganda {B. montana). 



The Species. 



I have arranged the species in four sections, in accordance 

 with the type of branching in the inflorescence. In the first 

 section (DivaricatcB) the branching is dichasial, and no definite 

 main rachis is differentiated. In the second and third a pro- 

 nounced median rachis occurs; this bears lateral cymose branches, 

 which may be relatively long and pedicellate, the flowers being 

 more or less laxly arranged {Laxa) ; or the lateral rachides may 

 be quite short or obsolete, the inflorescence being therefore spicate 

 or subspicate [SincatcE). In the fourth section {Capitatce) all the 

 flowering branches tend to suppression, and the inflorescence is 

 capitate or subcapitate. 



These sections are by no means sharply difl:'erentiated, although 

 the species typical of each are readily distinguished. Transitional 

 species occur ; and perhaps the most interesting is B. laxissima, 

 which links DivaricatcB with Laxce, and gives a hint as to how 

 the prominent median rachis of the latter section may have been 

 derived from the dichasial branching of the former. Tracings 

 made from the actual plants are reproduced in the appended 

 diagrams. No. 1 shows the dichotomous inflorescence of B. brevi- 

 flora (the ultimate branches are not shown at their full length, 

 which may be very considerable). No. 2 is a full tracing of the 

 inflorescence of B. laxisswia, and it will be readily seen that the 

 main rachis doubtless owes its ill-defined and sinuous character to 

 its obvious origin from a succession of dichotomies. No. 3 is a 

 similar diagram of the inflorescence of B. Batesii, revealing a 

 more definite main rachis, but the latter retains clear evidence of 

 its similar origin. No. 4 is a full tracing of the thyrsus of 

 B. guianensis, typical of the section Laxce, with a straight and 

 well-defined median rachis ; but it will be seen that dichotomy 

 still plays an important part in the branching. 



The Divaricates, then, will be regarded in the present arrange- 

 ment of the species as the relatively primitive section. From 

 these the Laxce were derived in the manner suggested above. 



