THE PHANEROGAMS OF 1HF. JUAN FERNANDEZ ISLANDS 



distance north of Buque Varado, c. 1230 m (fl.-fr. ' 3 17, no. 367 b); fell-fields 

 near the Correspondencia Camp, c. 1200 m; in moss among the rocks of Las 

 Torres, 1370 m (fl.-fr. 14 1 17. no. 367); C. del Batril, 1 1 10 m (fr. \ 3 17, no. 530); 

 Los Inocentes, 1375 m. 



The curious appendages do not appear to have been found in any other 

 species. Their nature is problematic. In the cases where each carpel ends in 

 such a tail-like process it would lie near at hand to regard them as enlarged, 

 persistent styles, but this explanation is impossible, as they are situated outside 

 the disc; inside this the traces of the styles will be found. In several instances 

 as many as three were found on one carpel (fig. 25 g): in two cases such an 

 appendage was found attached to the pedicel (fig. 25 h). Finally, many plants 

 did not have any at all. 



G. masafueranum is related to a number of Andine species, such as G. 

 fuegianum Hook fil., uncinulatum DC, canescens Kunth and andicola Krause. 

 The latter three are larger plants with many-flowered inflorescences and with 

 much longer hairs on the fruits and leaves: G. fuegiamim has larger leaves, 

 long pedicellate flowers and a glabrou> stem, and all of them have barbed 

 setae om the fruit. 



Area of distribution: Endemic in Masafuera. 



Campanulaceae. 



Wahlenbergia Schrad. 



In Monogr. Camp. (1830) 160 ALPHONSE DECANDOLLE described W.fernan- 

 deziana A. DC. P'our years later CoLLA (Mem. Accad. Torino XXXVIII. 118, 



tab. 351 described Campanula Larraini Bert. ined. COLLA stated that, to judge 

 from Decandolle's diagnose, C. Larraini was different. The material quoted 

 by DECANDOLLE and collected by Mrs. Graham, DOUGLAS and Scoi'LER 

 shows that his species comprised several forms, for Mrs. Graham collected 

 If". Grahamae Hemsl. and what I call W. femandeziana s. str. and Douglas 

 and Scouler collected W. Larraini. This explains why A. P. DECANDOLLE, 

 Prodr. VIII. 438 referred W. Larraini to femandeziana as a synonym. 



The question of nomenclature becomes complicated by the fact that there 

 are two types among BERTERO's plants, both under no. 1443. I do not know 

 if Colla did see both, anyhow, his description and plate answer very well to 

 one of them, and this I have retained here as If". Larraini. DC. Prodr., 

 HEMSLEV and Johow write Wahlenbergia Larraini Bert., Colla, but COLLA did 

 not admit the genus Wahlenbergia, but brought his species to Campanula. 



He.MSLEV, Challenger Report 45 rejects Larraini, but describes the new- 

 species Grahamae, based upon a part of Mrs. Graham's material. Johow 

 admits one species only, 11". femandeziana A. DC. For some reason he compares 

 the narrow-leaved species (my Larraini) with II". Berteroi. which is a quite 

 different thing; he declares that he found all sorts of transitions between Lar- 

 raini and Grahamae. It is true that there are forms apparently intermediate 

 between Grahamae and femandeziana s. str., but nobody who has seen these 



