THE PHANEROGAMS OF THE JUAN FERNANDEZ ISLANDS 125 



been bitten off by the goats; in fact, not one intact specimen was seen and 

 most of them were badly damaged, so that the existence of the species seems 

 to be seriously threatened. As far as I can find, R. caprarum has no relative 

 in America. Habitually, it belongs to the type of R. Lyallii Hook, fil., in- 

 signis Hook. fil. and nivicola Hook., all from New Zealand; it is remarkably 

 like the last mentioned, which differs in the less deeply cleft basal leaves and 

 in the unwinged achenes. But in another New Zealand species, R. Haastii Hook, 

 fil. (Fig. 8 g), which has a different habit, we find the same flat, winged achenes. 

 It is possible that R. hawaiensis A. Gray (Fig. 8 f) and R. mauiensis A. Gray 

 from Hawaii belong to the same group; their leaves are more divided with 

 stalked segments and the nut is barely winged, but there is a distinct trace of a 

 wing on the style. I regard R. caprarum as another addition to the old Pacific 

 element in the flora. 



Area of distribution: Endemic in Masafuera. 



Berberidaceae. 

 Berberis L. 



56. B. corymbosa Hook, et Arn. — Joi-iow, Estud. 115. — Fig. 9 c— d. 



Masatierra: not uncommon along the high ridges but never occurring in 

 considerable number nor forming thickets; sometimes also found in the open 

 forest. — Co Centinela, c. 400 m; El Pangal, near the entrance and on the walls 

 of the gorge, c. 200 m; Q. Damajuana, open forest on steep slope, 345 m; in 

 the gap between Damajuana and Yunque, 550—600 m (unr. fr. I8 /ia 16, no. 158); 

 high peak between Pangal and V. Colonial, 365 m; V. Colonial, C. Central, 

 570 m (also observed by JOHOW); Portezuelo de Villagra, a few small shrubs 

 near the Selkirk tablet, c. 600 m (fl. Via 16, no. 36); Q. del Monte Maderugo, 

 steep rocks; Q. Seca, open forest, c. 500 m; C. Salsipuedes (also observed by 

 Johow), 400 — 600 m, in brushwood and forest, on the ridge and on both slopes; 

 Pto Ingles, central ridge, 470 m; Q. Vaqueria, in open forest, c. 250 m; mountain 

 spur W of El Yunque, in forest, 530 m; Q. Villagra, higher parts, scattered. 



B. paniculata Phil, was retained as a species by Hemsley but reduced to 

 B. corymbosa by Reiche and JOHOW. After having examined the type in 

 Herb. Santiago I can only subscribe to the opinion of the latter. From the 

 description of Philippi (narrow leaves!) one would perhaps conclude that B. 

 paniculata is the same form that I have described below from Masafuera. This 

 is not the case; the leaves vary in B. corymbosa, but B. paniculata falls 

 within the limits if this. Bracts in the panicle are found in otherwise typical 

 corymbosa. 



Area of distribution: Endemic in Masatierra. 



*57. B. masafuerana nov. spec. — Fig. 9 a— b. 



Frutex metralis et ultra, ramis tenuibus cortice obscure rubro — atro-violaceo, 

 leviter sulcato in junioribus saltern lucido sicut vernicoso, ramellis distantibus. 



