ON THE SO-CALLED BOW-PULLER OF ANTIQUITY. 161 



form. If any grounds exist for believing it to be a lamp 

 Utensil then the spines might be supposed to assist in hold- 

 ing the wicking material whatever it might be. The upward 

 bending of the rings might be supposed to fit the curving 

 surface of the oil saucer. Roman saucer-lamps are common 

 and would seem to necessitate some device for keeping 

 the wick submerged. The great variety in the objects 

 which are supposed to be bow-pullers, or related to them, 

 suggests the idea that they were not all used for the same 

 purpose. Whatever the first one was designed for it is 

 possible that objects for entirely different purposes were 

 made in imitation of the first form. Thus in our times a 

 paper-weight may be seen in the form of four cannon balls, 

 an inkstand in the form of a mortar, a bronze pen-rest in 

 the shape of a cavalry saddle, or a horseshoe turned up 

 on end (a remarkable Greek vase in the British Museum 

 is beautifully modelled in the form of a horse's hoof) and 

 so on. Thus it may be that some of these objects may 

 have been used as a weight to hold down the lamp wick. 1 

 It is possible that the long-spined ones were strapped 

 or bound to a horse or man to keep a load from shifting 

 or swinging. This use is suggested on account of their 



O O CO 



manner of wear and breakage. 



A friend of mine has suggested that the object might 

 have been bound to the hand to enable a chariot driver to 

 hold the reins more firmly in driving ; a curb, in fact, but 

 held in the hand and not attached to the horse's head. 

 This idea is strengthened by the uniform length of the ob- 

 ject, and the upward inclination of the rings corresponding 

 respectively with the width of the palm and its hollowing 

 shape. Nearly all the bow-pullers fit naturally into the 

 palm of the hand ; the occasional bending of the spines and 



1 1 have already called attention to the evidence of, at least, two distinct types 

 of bow-pullers, not including such forms as those shown in Figs. 10 and 12. 

 With sufficient material these types may be found to run into each other ; but with 

 the objects thus far examined the differences seem to hold good. 



