GERANIUM PURPUREUM VILL. AND G. ROBERTIANUM L. : I 165 



(Natural History). It is quoted by White (1912). As expected, it proved, on examination, 

 to be G. rohertianum, of which several forms occur at Cheddar. However, genuine speci- 

 mens of G. purpureum subsp. purpureum were collected by Miss Roper in 1918 and by 

 R. and K. Tuckins in 1938; they do not appear to have been recorded in the literature. 

 In 1945 I collected the plant growing on a ledge on the south side of the Gorge (and 

 have used this material in experimental studies) and, in 1946, saw the/ species again 

 growing on the roof of a greengrocers' stall in the Gorge (where seed had evidently dropped 

 from the cliffs above). Interested by this, Mr. N. Y. Sandwith kept a look-out for the 

 species and confirmed its presence. 



The single station in Gloucestershire lies across the Avon Gorge from Leigh Woods. 

 G. purpureum was first recognised on Clifton Down as recently as 1925 by C. Wall. Her- 

 barium specimens suggest that this population has suffered some introgression from G. 

 rohertianum. This also applies to the relatively recent Carmarthenshire record, for, 

 although it is referable to G. purpureum, the single specimen from Newcastle Emlyn has 

 several features reminiscent of G. rohertianum. A population-analysis in this locality 

 might be very rewarding particularly as this is a remarkably disjunct outpost for G. 

 purpureum. Similar suggestions of interspecific hybridisation apply in the Torquay 

 area, at Bere Ferrers, and in north Cornwall. 



Druce (1932) is in error in including the Isle of Wight and Breconshire in the range 

 of G. purpureum. 



Of the records from outside England and Wales, those from the Channel Isles are, 

 in general, well known. The record from Bordeaux Harbour is of interest in that subsp. 

 purpureum occurs there at the same station as the subsp. forsteri and, apparently, in 

 circumstances reminiscent of those at Clymping, Sussex. The two Irish records are of 

 long standing and are quite genuine although, surprisingly, they are not given by Druce 

 (1932), nor by Praeger in any of his lists. AtDungarvan, however, there is a suggestion of 

 natural hybridisation with G. rohertianum. The records from Kerry, Clare and Galway 

 (Moore & More, 1866; Colgan & Scully, 1898; Praeger, 1909, 1934a, 1934b; Scully, 1916; 

 Warburg, 1952; Webb, 1952) are all due to confusion with prostrate maritime G. 

 rohertianum. 



In the Scilly Isles, only G. rohertianum is reported in the literature but Curnow's 

 specimen in the National Museum of Wales appears to be G. purpureum. The literature 

 records of G. rohertianum (kindly supplied by Mr. J. E. Lousley) are all rather old and 

 may also turn out to be referable to G. purpureum. Further collecting is very desirable. 



G. PURPUREUM subsp. FORSTERI (Wilmott) comb. nov. 



G. purpureum Vill. sec. T. F. Forster (1830), Engl. Bot. Suppl, 3, 2648. 



G. rohertianum var. y Babington (1851), Man. Brit. Bot., ed. 3, 62. 



G. rohertianum var. purpureum (Vill.) sec. Syme, 1864, Engl. Bot., ed. 3, 2, 204. 



G. purpureum var. forsteri Wilmott (1921), /. Bot., 59, 95. 



Type locality : Clymping (Middleton), W. Sussex, England. 



Plants of this subspecies have a prostrate habit, for the leader soon branches and 

 the branches run at only a slight angle with the ground although the tips ascend. The 

 petioles of the rosette leaves are still shorter than those of subsp. purpureum. They grow 

 only in the stabilised area at the rear of certain shingle-beaches. In cultivation they 

 flower about a fortnight later than British material of subsp. purpureum. 



While the fruits of most British representatives of the subsp. purpureum are quite 

 distinctly ridged, any population of subsp. forsteri shows a considerable range from 

 typical purpureum to fruits very like those of G. rohertianum (Figs. 3-5) and both glabrous 

 and hairy fruits occur together. This makes the determination of herbarium materia] 



