REPORT OP MEETINGS FOR 1902 273 



In the widest part of the wall there is a double arrangement 

 of cells, one large chamber measuring over 18 feet by about 

 8 feet. Beyond this, a door opening through a wall 3 feet in 

 thickness connects with a smaller chamber over 6 feet square. 

 Outside this there still remains about 12 feet of building. 

 These chambers are certainly formed in the thickness of the 

 wall, and have a greater similarity to the cells formed in 

 the walls of a broch than to those seen in the ordinary British 

 fort. Possibly the builders here had seen a broch. The 

 general character of the mason work is that usual in stone 

 forts the building of which, considered as mason work, is 

 inferior to that seen in the broch. There is a tendency to 

 set any long stones they have had upright in the ground 

 with the broadest face outwards. In many forts these 

 stones so set, earth-fast, remain to indicate the line of the 

 wall after the smaller stones used as packing have been 

 removed. I have formed the opinion that the race who built 

 thus had formed their ideas of building construction in a 

 forest country, where wood was the material used, and these 

 upright stones are substitutes for wooden piles or posts. 

 No doubt there were originally a greater number of cells 

 in the wall than can now be traced. Mr M. Home's earlier 

 plan shows more than I can now do. Iti the interior of the 

 enclosure none of the hut circles remain. There is by the 

 south gate a guard chamber, and there are some enclosures 

 there that may have been used at some time for cattle. 

 The diagonal mound running straight across the interior is a 

 common feature in forts in exposed positions. At Hillhouse 

 Fort, in upper Lauderdale, a large mound runs across, and 

 close against it shelters a line of hut circles. In Parkhill 

 Fort (see Plate VII. in paper on the "Heads of Bowmont 

 Water," in Club's Proceedings for 1897, p. 191) two similar 

 mounds occur, where the hut circles still remain. 



But a notable circumstance at Haerfaulds is that to its 

 eastern gate a branch of Herrits Dyke runs up. Herrits 

 Dyke is the Berwickshire name for a work of a widely 

 spread type. A hollow with a mound on one side is the 

 common form, but sometimes there is a mound on both 

 sides, and occasionally there is a hollow without any mound. 

 The Oatrail in Roxburghshire, the Deil's Dyke in QftUoway, 



