Wesikopp — The Fort of Dun Aengusa in Inishmore, Aran. \V 



oval liugs into wliich the sea has cut for almost exactly half their extent. 

 Such a plan is possible, warranted by the fine triple-walled hill-town of 

 Moghane, but must not be too readily accepted. The fortress was probably 

 at first a simple oval ring-wall, like its neighbour tlie Dun of Eoghanacht 

 and many others. It was next strengthened by a second ring nearly equi- 

 distant from the first, like the forts of Dun Oghil on the same island, 

 Furmina on Inishere, Glenquin and others in Clare. Still later, a third wall, 

 either somewhat egg-shaped in plan or a crescent, abutting on the cliff, was 

 built. This was regarded as fiaal ; and an elaborate abattis of close-set pillar- 

 stones was made roimd it, extending to its foot, even where it crowned a 

 steep slope. Lastly (and probably at a far later period) a large irregidar 

 space, determined to the east by a low ridge, was enclosed with another wall, 

 defending the approach from the landing-places of Port Murvey. We do not 

 for the moment assert whether, or how far, the sea had cut into the hill when 

 these works were built. O'Flaherty seems to imply that the middle ring was 

 entire, like " the bawn of a castle," in 1684 ; but it then stood " on the brim 

 of a high cliff'."' Probably the old second wall was originally entire ; but this 

 is uncertain, for the stone fort of Cahercommaim, in Clare, on the edge of the 

 Corcomroes, has a central ring and two crescent walls.- It is strikingly like 

 Dun Aengusa. The central fort is even more massive ; but it overhangs a 

 dry valley rising at both ends, so is evidently in its original condition so far 

 as regards the plan. Another alternative is possible, namely, that (as at 

 Dim Conor and the Clare forts of Caherlisauiska and Langough) the 

 central fort was a ring with the outer enclosures looping in to meet its wall. 

 We have, howe^ er, only found this looping in forts on flat fields and low 

 i-idges — never at high cKffs or slopes, or even on a low shore when there is 

 deep water beyond it, as at the crescent fort at Cahernacalla on Ballycar 

 Lake.' The crescent wall,* therefore, does not necessarily prove a fall of the 

 cUif, for it is common inland in Ireland, and indeed all over central Europe 

 and in America. 



In Clare, besides Cahercommaim and Cahernacalla, with a central ring 



' Cliorographical description of "lar Gonnaught" (ed. Hardiman), p. 75. 



-A potm of Seanchan, daling about 640 (Book of Lecan, p. 17), mentions " the three mounds of 

 walled fortresses" in Burren, Co. Clare — perhaps this (the only triple-vralled) fort — as on the 

 Burren hills, though now included in Inchiquin. It is probably the fort of Caechan Boime, near 

 Incliiquin Lake, named, perhaps, about 800 (certainly <iii/e 1014), in the " Book of Rights." See 

 Journal R.S.A.I., vol. xxvi., p. 153, and I'roc. R.I. Acad., ser. iii., vol. vi., p. 430, for plans com- 

 parable with that of Dun Aengusa. 



»Proc. R.I. Acad., vol. x.Tvii. (C), p. 227. 



'The crescent type is so closely akin to the promontory fort that in some cases distinction cannot 

 be drawn save by rej:iriiing the existence of a headland behind the defences. Mr. .\llcroft classes 

 them loyelber iu '■ KaiiliWurk of Kugland," chapter iii. 



[2*] 



