Christian Monuments of Wiltshire. 55 



parallel. The length of the face of the shaft is only about two 

 and three-quarter times its breadth, so that there is not room for 

 three complete rings. The clever method of finishing the pattern 

 at the top, so as to get over this difficulty commands our admiration. 

 The circular medallions containing beasts may perhaps have suggested 

 the twist-and-ring pattern of the interlaced work to the mind of 

 the designer. Anyhow, the ornament on all four sides o£ this shaft 

 must have been set out in the same way, by drawing two complete 

 circles and three-quarters of a circle at the top. The nearest approach 

 to the pattern on the Ramsbury stone (B) is on a stone at Durham 

 Cathedral, but the bands are double, not triple. 



The only other ordinary knot which occurs on the Wiltshire 

 monuments is on the fragment of the coped stone (E) at Ramsbury. 

 Here the tongues of the serpentine creatures whose bodies form the 

 ridge and hips of the stone, are tied together in a Stafford knot^ 

 (Fig. 6). 



Pfe. 6. 



In my " Analysis of Celtic Interlaced Ornament " in the " Pro- 

 ceedings of the Society of Antiquaries of Scotland" (vol. 17, pp. 211 

 to 271), already frequently referred to, I have termed a particular 

 class of interlaced ornament "circular knotwork," on account of the 

 predominance of bands making circular curves throughout the 

 patterns of this description. I have no doubt in my own mind that 

 this particular development of interlaced work was originally in- 

 vented on the Continent, and probably in Italy, where so many 

 examples of it are to be found.* The Celtic artists improved upon 



» " Analysis," p. 242, knot A. 

 * Grado, in Dalmatia ; Milan and Como, in Italy ; and Grandson, in Swits^r- 

 land. 



