206 



The latter, as well as tropical Arabia, are far better explore 

 Somaliland, and this may account sufficiently for a greater n 



tropical Arabia finds a striking expression in the number of 

 grasses common to those areas. Out of the 67 non-endemic species 

 of British Somaliland no fewer than 45 (67 per cent.) have been 

 found in all three countries, whilst 49 (73 per cent.) occur in 

 Eritrea and 9 more in the adjacent parts of Abyssinia and Nubia, 

 and 48 (72 per cent.) are recorded from tropical Arabia. If we 

 finally consider the grasses which occur in British Somaliland and 

 at the same time appear either in Eritrea and the immediately 

 adjoining districts or in tropical Arabia, we find that no fewer 

 than 60 (91 per cent.) out of 67 non-endemic species come under 

 the heading. There is therefore no doubt that, so far as the grasses 

 are concerned, Somaliland proper and Eritrea with Eastern Nubia 

 on the one, and tropical Arabia on the other side of the Red Sea form 

 a natural unit. Similar conditions also prevail in other respects, 

 marking out that area as a distinct phyto-geographical province 

 which might appropriately be called the Erythraean Province. It 

 would naturally include, although as a well characterised sab- 

 division, the Island of Socotra. It is true that of the total of 

 50 non-endemic grasses known from Socotra only 20 have been 

 recorded from British Somaliland, and indeed from the whole of 

 the Somali Peninsula ; but on the other hand the island has 34 

 (68 per cent, of the non-endemic element) in common with tropical 

 Arabia and as many, and mostly the same species, with Eritrea. 



number 



of Socotra grasses having been found there than in the Pro- 

 tectorate. 



If we go beyond the limits of the Erythraean Province and take 

 a broad survey of the grass flora of Somaliland with respect to 

 more distant affinities, we find that its character quite fits in with 

 the position of the Erythraean Province as a division of the great 

 steppe and desert belt, which extends from the Cape Verd Islands 

 and Senegambia to the Indus, exhibiting a stronger leaning to- 

 wards India than towards the West. The number of grasses 

 common to British Somaliland and India is 40, 1 1 of which do 

 not occur west or south of the Erythraean Province. On the 

 other hand, 36 extend to Senegambia or the Cape Verd Islands ; 

 most of the latter are, however, species having a wide distribution, 

 or at least ranging eastwards and southwards of the Somali 

 Peninsula. 



The relationship with India will no doubt appear still more 

 accentuated when the species which Socotra has in common with 

 Arabia and Eritrea but not, so far as our present records go, with 

 British Somaliland, may have been discovered there, a contingency 

 that is to be expected as they are practically all members of the 

 Indian flora. 



As might be expected, a considerable number (31 and 25 

 respectively) of the species extends southwards into East Africa 

 and even South Africa, but here a; «in most of them are species of 

 a wide range, only five or six being limited to the Erythraean 

 Province and East Africa. The isolation of the Somali Peninsula 

 from the African Hylaea region is very strikingly shown by the 

 small number of species which extend into it. If we except the 

 grasses of more or less cosmopolitan character or of at least general 



