126 



THE ROYAL SOCIETY OF CANADA 



determinations of the period corresponding to variations in either of 

 these adjustments and then plot curves, one for each knife edge, 

 showing relation between the period and the varied item. The 

 point of intersection of these curves corresponds to the position of 

 the variable for equal periods and the value of the common period 

 may be read off from, the curve or obtained from the observed values 

 by the calculation corresponding to the determination of the said 

 point of intersection. Unless the periods are nearly equal these 

 curves are usually not rectilinear so that the calculation is complicated 

 or a large number of observations are necessary in order to obtain 

 an accurately drawn curve. 



2. In the pendulum to be described an attempt is made to com- 

 bine a num,ber of the desiderata indicated in the foregoing discussion. 

 The pendulum is sym.metrical in figure as required for considerations 

 (b) and (c). Its construction is shown diagramatically in Figure 1 



A\ — I 

 FIGURE 'I 



and its dimensions are given in the following table. CB is solid brass 

 rod and AB is aluminium tubing of the same external diameter as 

 the rod. In order to avoid correction for expansion by change of 

 temperature I should prefer that the whole be constructed of an Invar 

 of which the coefficient of expansion is a m.inim.um. at the tempera- 

 tures of ordinary use; the dimensions of the solid rod and the thickness 

 of the tubing could easily be chosen to fît into the further require- 

 ments indicated below. 



The knife edges are inserted rigidly at F, F. The ends are closed 

 and the joints m.ade air tight. Each end carries platinum, tips to 



