CELASTRACEiE. 293 



quincuncially imbricated, while, according to Kunth, they are valvate 

 in P. Quinduensis. This would seem to be the case in two flowers 

 which have been obligingly furnished me by M. Tulasne, taken from 

 an original specimen in the Paris Museum ; but they are much too 

 old to decide this point. Even if this be so, it would not be proper 

 to establish a new genus for the Hawaiian plant, unless the flowers of 

 the Peruvian species should prove to be truly hermaphrodite (which 

 I doubt), and the rhaphe normally ventral. As species, they are 

 abundantly distinguished by the subsessile flowers, the conspicuous sti- 

 pules, and the larger petals of P. Quinduensis, which are not at all 

 ciliate, and by the longitudinal sculpturing of its bony seeds.* These, 

 in the specimens of Humboldt and Bonpland (the only ones known), 

 are hollow and destitute of embryo, just as are those of our Hawaiian 

 plant ; probably from not having been fertilized by the pollen of male 

 flowers. The internal structure of the seed is therefore a desidera- 

 tum ; and until this is known, the affinities of the genus cannot be 

 positively determined. Most probably, however, it belongs to the 

 Celastracece, notwithstanding the total absence of any arillus. 

 According to a note on the ticket of Gaudichaud's specimen, this 

 . shrub is called Oraye or Olaye by the Hawaiians. 



Plate 24. — Perrottetia Sandwicensis. Fig. 1. A branch of the 

 fertile plant, in flower and fruit, of the natural size. 2. Flowering 

 branch of the male plant. 3. A small portion of the inflorescence of 

 the fertile plant, enlarged. 4. Diagram of the flower. 5. A sterile 

 flower. 6. A fertile flower, expanded. 7. A pistil, from the latter. 

 8. Vertical section of a flower through the pistil and disk. 9. Trans- 

 verse section of the base of the ovary, showing the position of the 

 ovules. 10. An ovule. 11. A fruit. 12. Vertical section of the 

 same. 13. A seed. — All the details magnified. 



* M. Tulasne informs me that, if he is not mistaken, the ovary of P. Quinduensis is 

 sometimes four-celled, with a single ovule in each cell. 



74 



