2 22 Proceedings of the Royal Irish Academy. 



fences and the hut-sites if such existed. The process has only just commenced 

 at Dun-Ooghaniska, where it has made some progress since 1909 owing to the 

 following very wet summer. At Dunnaglas it has destroyed about half the 

 fort, and at the nearly bare rocks of Dunmore and Dunnagappul near Doogort 

 the forts have vanished, and only their name clings to the rocks. In some 

 cases the destruction is very intermittent; vegetation grows again on the 

 bare surface and preserves it for some, even for many, years ; then from some 

 cause the plants are destroyed, and the bare bank again commences to crumble 

 till the grass again establishes itself. So far as I have examined the cliff- 

 forts of western Connaught and Munster, this is the most usual process of 

 destruction. It shows how many dozens or hundreds of entrenched head- 

 lands may have perished in the thousand years since the Norse wars, even 

 where the rock bases survive. The two Duns and the adjoining cliffs enclose 

 a series of beautiful picturesque little creeks looking out to Achill and the 

 hills near Mulranny. 



Ooghbeg (0. S. 75). — On a low headland, a mass of drift rapidly falling 

 away on every side, I found the remains of a massive wall. It is at present 

 66 feet long, but both ends are wasting. At 45 feet from the present 

 west end, is a long set slab suggestive of the jamb of a gateway. The wall, 

 instead of having been built on the surface, was made in a foundation trench 

 over 3 feet deep ; it was 12 feet thick, of very large blocks, and save for the 

 foundations, is nearly removed. It is very slightly convex to the land and 

 lies roughly east and west. Its masonry and massive character mark it as 

 ancient. The bank at it is 15 feet to 25 feet high, but it may be remembered 

 that the great fortification of the headland at Porth or Portanalbanach in 

 the Mullet abuts on even lower shores. The garth was once tilled, so no 

 traces of inner buildings remain. 



There are no marks of any early entrenchment or wall on the headland 

 at Grania Uaile's castle. 



Duncloak (0. S. 85). — This natural fortification lies westward from 

 the castle and bears considerable resemblance to the curious promontory 

 fort of Dunmore or Horse Island, near Loop Head, Co. Clare, and to the 

 Horse Island near Belderg on the north coast of Mayo. It is joined to the 

 mainland, however, by a broad grassy neck, 27 feet wide, with steep slopes 

 at either end, and low precipices on each side, the western at a curving 

 creek with two fine high natural arches ; the eastern at a crescent of 

 shingle beyond which are abrupt cliffs sheeted with masses of Osmunda 

 wherever a stream descends their faces. Like Dunallia, and the Clare Horse 

 Island it was only very slightly fenced by a mound of earth and large stones 

 at the summit of the neck, and eastward along the face, but probably it was 



