COOK RELATIONSHIPS OF THE IVORY PALMS. 137 



phas than they have with each other. Cocoid palms like Attalea. 

 and the curious bag palms. Manicaria, are certainly nearer to Phy- 

 telephas than they are to the fan palms. Instead of having 1 the shells 

 united into one. as in the cocoid palms, each seed of Phytelephas and 

 Manicaria has its separate bony shell. The fruit of the Cocaceae is 

 really more peculiar than that of Phytelephas, since it departs fur- 

 ther from the structure found in other palms. Instead of rejecting 

 Phytelephas from among the palms because its fruits differ from 

 the Cocaceae it should be considered that in this respect. Phytelephas 

 serves to connect the Cocaceae with the other palms. Moreover. 

 Attalea and Manicaria are both natives of the same geographical 

 region as Phytelephas. 



Except for the limitation to three carpels, the fruit of Manicaria 

 seems to be entirely analogous to that of Phytelephas. In both cases 

 there is an outer warty coat of fibrous material and an inner shell of 

 bony tissue of a columnar structure. The association of Manicaria 

 with Geonoma has no apparent reason, for the inflorescence of Mani- 

 caria has more analogy with the cocoid palms than with Geonoma. 



The statement of Martins that the flowers of Manicaria are buried 

 in pits in the spadix has been repeated by Drude, but appears to 

 have as little warrant as some of the statements regarding Phytele- 

 phas. Bentham and Hooker were more nearly correct than Drude 

 in that their group of " Arecae dublae ajjinitatis " included Manicaria 

 and Leopoldinia as well as Phytelephas and Xipa. The fact that 

 Leopoldinia has now found an apparently natural association with 

 Geonoma, can not be considered as a reason for treating Manicaria 

 in the same way, in the absence of common characters. 



Manicaria is a less specialized type than Phytelephas, but the 

 specializations have taken much the same direction, both in the exter- 

 nal and internal characters. The number of stamens, 27, falls only 

 slightly below that of some of the species of Phytelephas. Mani- 

 caria may be considered, therefore, as a link connecting Phytelephas 

 with other palms, and especially with the cocoid series. 



One of the features of Phytelephas not taken into account hitherto 

 as an evidence of relationship is the method of germination. See- 

 mann figured the germination of the Phytelephas that he described 

 in Panama, but seems not to have observed the germination of 

 Attalea. Spruce does not appear to have considered the germina- 

 tion of either genus. 



The germination of Phytelephas is practically the same as in the 

 cocoid genus Attalea. The embryo does not develop directly from 

 the seed, but is carried out on a long cylindrical cotyledon. The 

 behavior of the cotyledon of the Phytelephas has not been described, 

 but in Attalea it burrows in the ground, carrying the embryo with it. 

 The effect is to plant the embryo from 3 to 6 inches below the surface 



