460 



CONTRIBUTIONS FROM THE NATIONAL HERBARIUM. 



Fig. 84. — Sideroxylon mas- 

 tichodendron, floral details- 

 a, Petal (auriculate); 6, 

 utaminodes; c, stamens; d, 

 filament; e, pistil. Scale 3. 



Pedicels glabrous. Corolla 3.5 to 4 mm. long. Anthers slightly apiculate. 

 Staminodes hardly above 1 mm. long. Seeds ovoid (pi. 96, c); umbilical 

 area circular; glabella distinctly biscutellate. Leaves long-petiolate, 

 pouchless. — Porto Rico (Sintenis 5001, 3950; Heller 507); Florida (Garber; 

 Chapman 11, Curtiss 1759). 

 On account of the incomplete descriptions of the older authors, it is difficult to 

 identify these forms with their intended specific names. It is, however, safe to 

 admit that the Porto Rico and Florida specimens represent the original S. masticho- 

 dendron of Jacquin and Gaertner (fig. 84), even though 

 Jacquin does not mention the characteristic "glabella ovata 

 bipartite." which seems to be a constant feature of the 

 seeds in the Porto Rican and Floridan specimens and which 

 is clearly shown in plate 202 of the Supplementum Carpo- 

 logiae" of the last author. An obscure detail of this ques- 

 tion is that the origin of the specimen made use of in 

 Gaertner 's description is not given, while the names acoumu 

 jaune ou abricot dcs bois would seem to indicate a product 

 of the French West Indies. 



The other forms may represent entirely new species, or 

 correspond to some of the older ones (S. obovatum Ga-ertn., 

 S. acouma A. DC, S. foetidissimum Jacq.) {fig. 85). In 

 regard to their status, we may repeat here what Jacquin 

 said with respect to the whole genus: "Donee omnium 

 fructificatio plene cognita sit, frustra tentabimus." Only the careful comparison of 

 large series of complete materials will enable us to decide, and meanwhile Urban's 

 step in condensing forms a few of which are to all appearances specifically character- 

 ized must be considered as premature. 



The presence of a Sideroxylon of the mastichodendron group on the continental 

 part of Central America is indicated by a specimen collected by Gaumer (no. 763) 

 near Izamal (Yucatan). It was identified by Dr. Mills- 

 paugh b as the true 5. mastichodendron, but is certainly 

 quite distinct from the specimens quoted above from 

 Florida and Porto Rico, as well as from the Cuban tree, 

 specimens of which were collected by Wright. I now pro- 

 pose to name it Sideroxylon gaumer \, its description being 

 given below. 



Under the much-used name of Sideroxylon mastichodendron 

 Jacq., several collectors have distributed another Central 

 American representative of the genus, widely known 

 among the natives as tempisque or tempixque. It will be 

 conclusively shown that this is in reality a very distinct 

 type, more closely related to a Mexican species with many 

 names, Sideroxylon capiri, which is also described and 

 discussed in this paper. 



Explanation of Plate 90.— Seeds of Sideroxylon. a, Cuba, Wright 1324; b, Guadaloupe, Duss 2915; 

 r, Florida, Curtiss 1759— these representing three forms included by Urban under 5. fori idissi mum; d, 

 Sideroxylon capiri; e, Sidtroxylun tempisqur. From photographs taken in Washington by Pittier & Doyle. 

 All natural size. 



Sideroxylon gaumeri Pittier, sp. nov. Figure 86. 



A tree 25 to 30 meters high. Uranehlets rather thick, with a glabrous iron-gray 

 bark. Younger growth glabrous. 



Leaves crowded at the end of the branchlets, glabrous, petiolate. Petioles 3.5 

 to 4 cm. long, slender, obscurely sulcate above. Leaf blades coriaceous, 8 to 10 

 cm. long, 3 to 5.5 cm. broad, ovate, rounded and shortly decurrent at base, rounded 



"Gaert. f. Fruct. & Sem. 3:123. pl.ZOt. 1805-1807. b Field Mus. Bot. 1:313.1896. 



Fig. 85.— Sideroxylon foe- 

 tidissimum (?), floral de- 

 tails, a, Petal; b, Btami- 

 nodes; c, stamens; d, pis- 

 til. From Wright's No. 

 1H24 (Cuba). Scale 3. 



