38 PAPERS, ETC. 



most common, and I decidedly prefer it. Cannington, 

 however, is a fine specimen of the pointed form. 



The tie-beam roof is, as far as I remember, confined to 

 the churches whlch have the clerestoiy, but the reverse 

 rule will not hold good, as is shown by the cases of Yatton, 

 BanweU, and Congresbury ; but these three lie so close 

 together that this is probably a localism within a localism. 



I must here not omit to mention some rieh roofs of later 

 data, which seem to be a cinque-cento variety of the old 

 coved roof. That of the nave of Bath Abbey is well 

 known ; but finer ones, to my mind, with tracery, pen- 

 dents, etc., occur at East Brent and Axbridge, and even in 

 the poor little church of Biddesham. That at Axbridge 

 bears the date of 1636. 



The ordinary arrangement of the clerestory windows I 

 have already considered; I have now to speak of the 

 connexion of the clerestory with the roof and the arcades. 

 To bring an elevation into complete harmony, the vertical 

 division into baya, and the horizontal division into arcade, 

 clerestory, and triforium, (if there be any,) should both be 

 marked in the decorative construction. There should at 

 least be a string running over the arches; and the clerestory 

 should be divided by shafts supporting the roof, either 

 rising direct from the ground, or corbelled off over the 

 piers. Where these are not found, as at Long Sutton 

 and St. John's at Glastonbury, the intei'ior has an un- 

 fiuished look, and can hardly aspire to the name of an 

 architectural design. When they occur, a spandril is 

 formed by the pier arch and the i'oof shafts, and a fur- 

 ther spandril is left between the roof and the clerestory 

 window. To fiU these up is a further development. 



Two principal forms of vertical division occur in the 

 great Somersetshire churches. At Wrington and Yatton 



