612 PHANER0GAM1A. 



represented in Forster's rude figures. Ovary strongly depressed, its 

 lower part adnate to the broad base of the calyx-tube, the free summit at 

 first flat, radiaiely 12-striate, after anthesis becoming convex, 12-celled 

 within, the thin dissepiments apparently evanescent after anthesis. 

 Ovides 2 in each cell, collaterally inserted on a central axis or column, 

 anatropous, at first horizontal, and with the upward growth of the 

 ovary after anthesis becoming pendulous. Fruit not seen in the 

 collection. Accordiug to Forster it is a hemispherical, many-striate, 

 superior, one-celled berry, containing several globular seeds. 



Our collectors did not meet with Crossostylis at the Society Islands, 

 nor indeed has it been detected since the time of Forster. It 

 appears, indeed, from the habitat, cited in Guillemin's Zephyritis 

 Taitensis, that Forster did not collect it on Tahiti ; but on the out- 

 lying island of Raiatea. Our specimens (unfortunately incomplete 

 ones, with only some very young flower-buds and a few detached 

 flowers) all came from Tutuila, one of the Navigators' Islands : but 

 they appear to belong to Forster's species, judging from the detailed 

 description reproduced by Guillemin, above-cited, and from the speci- 

 men of Lambert's herbarium, now in that of the British Museum, 

 which, however, consists of the foliage alone. But the toothed apex 

 of the petals is not mentioned by Forster. 



Crossostylis has always been one of the " genera incertas sedis." 

 As such it was placed at the end of his Genera Plantarum by Jussieu, 

 w r ho, however, with prophetic insight arranged it next to Cassipourea ! 

 and conjectured their affinity with the Salicariece. DeCandolle ap- 

 pended it, with much doubt, to the Myrtacea?, in which he was fol- 

 lowed by Endlicher. I am not aware that any Later conjecture has 

 been hazarded in respect to it; except that Dr. Pickering, in his manu- 

 script memoranda, queries whether it may not rather belong to the 

 PhUadelphiece. It is now abundantly evident, however, that we have 

 in Crossostylis an accession to that small group of genera which Mr. 

 Brown* long ago sagaciously appended to the Bhizophorece, and noted 

 as indicating a passage on the one hand to Salicareai, on the other to 

 CunoniaceoB ; and which, adopting the name of Legnotidece from Bart- 

 ling,-)* Dr. Blume has recently proposed as a separate natural order.J 



* Obs. Bot. Congo, p. 18. 



f Ordines Naturales Plantarum, p. 318. 



| Museum Botanicum Lugduno-Batavum, p. 126. — Apparently without valid reasons; 



