2G6 DR E. G. COKER ON 



this is sufficiently suppressed by the frictional grip of the collars, thereby causing the 

 parts to act as one rigid whole for setting the instrument on the bar. 



Both main pieces are chucked by set-screws, and with a little experience this can be 

 eifected as accurately as by self- centring chucks.* 



In nearly all machines for applying torque some bending is also present, and it is 

 therefore necessary to eliminate any possible errors due to this cause. 



If the bar is bent in the plane containing its centre line and the observation wire, it 

 has the effect of causing new parts of the wire to come opposite the scale, but no error 

 in reading is caused thereby. If, however, the bar is bent in a plane a,t right angles to 

 this, the effect of the bending will be read as an addition to, or subtraction from the 

 twist. Bending in any other plane may be resolved into components in these two 

 planes, and it is therefore only necessary to eliminate the error due to bending in a 

 plane perpendicular to the plane of the paper. 



The error may be got rid of by using two reading microscopes set opposite to one 

 another, and a mean reading taken, but as this doubles the labour of observation it is 

 inconvenient. Another plan is to arrange the wire mid-way between the sections 

 gripped by the set-screws : then if 



2a be the length under measurement, 

 = angle of bending at first section, 

 <p = angle of bending at second section, 



error in reading becomes a (sin 6 + sin </>) ; 



and if and <p are equal and opposite the error vanishes. A specimen stressed by two 

 equal and opposite couples applied at its ends bends into the arc of a circle, and fulfils 

 the necessary condition for the equality of and (p, and in the application of torque 

 this condition has been fulfilled. As a matter of precaution the observation wire is 

 always set in the plane of bending. 



In order to test the accuracy of the instrument, torsion tests were made (I.) with no 

 bending, (II.) with bending moment of known amount. As an example the following 

 may be quoted : — 



Turned bar of rivet steel . . Torsion arm= 15'00. 



Diameter 0*662 . . . Calibration. 



Length under test . . .1 min. = 54*4 divisions. 



* hoc. cit. ante. 



