These rods were originally kept in position by a pair of 
steel shackles (see fig. 2) which exerted a vertical pressure 
on the rod midway between the boss and the rim. The 
shackles were attached by set-screws to a cross-bar bolted 
on the underside of the boss (see figs 1 and 2), a little 
slackness being left in the screws, so that the shackles 
could move slightly through a small vertical arc. This 
cross-bar was attached to the boss by two set-screws pass- 
ing through circular grooves in the cross-bar, so that it 
could be moved round into any desired position. These 
shackles were found to be unsatisfactory, owing to the fact 
that their centrifugal force caused movement of the rod. 
In consequence, half-inch rubber bands were used in their 
stead, and were found to be quite satisfactory. The vertical 
pull of each band was eight pounds. The rods shewn in 
figs. 4 and 5, owing to their shallow section, were subject 
to great bending. This was found after a number of trials 
to mar the results, so that these rods were rejected in 
favour of the rod shown in fig, 3, which was much more 
rigid than the others. 
264 A. BOYD. 
In addition to these, the author has constructed a rod of 
aluminium (fig. 6), which was considerably lighter than any 
of the others, weighing only 7°25 ozs., whereas the rod, 
shewn in fig. 3 weighed 1 fb. 3°25 ozs. It consists of two 
aluminium tubes half an inch diameter united by a short 
steel bar. The rod is attached to the rim by a pair of 
aluminium pieces which are clamped to the rim and to the 
rod by means of set-screws. The shape of these pieces is 
shewn in the end view of the rod (fig. 6). an 
Pressure is applied to the rod directly over the prism by 
means of a spring held by a pair of set-screws passing through 
slots in the steel central portion of the rod. The set-screws 
are screwed into a short moveable cross-bar on the upper 
side of the boss. This rod was made towards the end of 
