ON THE PERPENDICULAR OF SOMERSET. 35 



tion, but still tbe lozenge form has by no means completely 

 vanisbed. 



As tbe section of tbe piers is tbe most prevalent of tbe 

 Somersetsbh'e cbaracteristics, so it is tbe least distinctive ; 

 tbe otber points are seldom met with elsewbere, wbile this 

 lozenge section frequently is. For instance, tbe section of 

 tbe piers in St. Marj's at Oxford is only a more elaborate 

 form of tbat of Wrington and St. Stepben's ; but as soon 

 as we reacb tbe capitals and arcb-mouldings, tbe resem- 

 blance vanisbes. Tbe fact is, tbat wbat elsewbere is one 

 not uncommon form among others, becomes in Somerset- 

 sbire nearly universal. 



Exceptions are more common in tbe capitals. Tbe 

 departure from ordinary practice generally consists in 

 carrying tbe abacus all round tbe pier, instead of leaving 

 tbe sides of tbe lozenge to be continued uninterruptedly in 

 tbe areb. Sometimes, as at Mark, Wedmore, Dunster, and 

 St. Decuman's, tbe capitals foUow tbe section of tbe pier, 

 (wbetber tbe usual one or any otber) or some sligbt 

 modification of it, as at Trull. In otbers, all tbe sbafts are 

 gatbered togetber under one lozenge-sbaped capital ; tbis, 

 wbicb I believe is a Devonsbire custom, occurs in a rüde 

 form in tbe cboir of Dunster, and in a very elaborate one at 

 Lydiard St. Lawrence. It is a form well adapted to render 

 tbe capital a beautiful individual feature, but it is one 

 completely destructive of all Continuous efFect. Octagonal 

 capitals to individual members of tbe Cluster are by no 

 means common, but tbey also occur m some parts of 

 Dunster. 



In tbe cases wbere a more elaborate section of tbe pier is 

 eraployed, some dliference necessarily follows in tbe mould- 

 ings of tbe arcb. Some mouldings necessarily rise from tbe 

 subordinate sbafts, and even tbose rising from tbe principal 



