— 10 — 



and at least in the lower half of the spike are connate to a little above 

 their middle, and the coma-bracts which do not extend much beyond the 

 floral bracts. 



The rather long pouches give them a superficial resemblance to 

 C. Roscoeana but on the whole they do not differ so much from the Curcuma type. 



Studying the descriptions and plates of the numerous species of Curcuma 

 known till now 1 found that several of them are receding from what I consider 

 as the original Curcuma type in the same way as the above named species. 

 Thus the genus can be divided into two subgenera which I propose to call 

 Eucurcuma and Paracurcuma (= Hitcheniopsis Baker ex parte). Beyond the 

 species named above, C. aurantiaca, C.petiolata and C. cordifolia, here belong: 

 C. meraukensis Val. and C. latifolia Val. (Valeton, 1913, 924 tab. 79 B. and C). 

 In both these species but the most distinctly in C. meraukensis the spurs 

 are appendages of the base of the cells and in the latter they show even 

 much resemblance with those of C. alismatifolia Gagn.; for the rest^flower 

 and inflorescence agree with that of C. petiolata.: 



C. australasiaca, Hook. f. (1867 t 5620). 



C. montana, Roxb. (1807 8,355.), Roxb. Cor. pi. II (1798, t 151). 



These two species, as already said by Hooker, are very nearly related, 

 and both have a great resemblance to C. pe//o/a/a. Of the second there exists 

 a detailed figure by Griffith (1853, 3, 415, t. 352) and here, as seen in 

 fig. 3 the cell seems not to be limited by a wall below but continues on 

 the lower margin. Also the figure given by Roxb. shows a great resemblance 

 to C. petiolata. 



C. longa, Bentl. and Trim, (non aliorum) (1886, 269). 



Both the apparently rather good figure, drawn from a plant, flowering 

 in the botanical Gardens at Kew, and the description, point to a species 

 very different from C. longa, Koen. which latter according to Roxburgh (see 

 below) is the original plant of the "turmeric". The kew plant belongs 

 apparently to the species which according to B. and Tr. was introduced 

 into the English stove by P. Miller in 1759, and which might be seen at 

 that time (1886) in most botanic gardens. It may have given origin to much 

 confusion; and it seems to me quite probable that it is this species that 

 caused Hooker to insist on the resemblance between C. australasiaca and 

 C. longa. 



§ 3. Survey of the Javan species of Curcuma. 



A. Paracurcuma: Bracts often very numerous, connected at least partly 

 beyond the middle. Spike cilindrical, with comparatively short bracts of the 

 coma. Bracteoles small, staminodia straight, larger than the dorsal petal which 

 is somewhat cucuUate, obtuse or with a short concave top, not clasping 

 the staminodes, except in C. cordifolia Wall. Anthers attached near the 

 base, not or very shortly calcarate, spur not longer than a quarter of the 



