MEASUREMENT OF TORSIONAL STRAINS. 209 



manipulation easier, and this is especially true of large 

 specimens, but it lacks the most important desideratum 

 simplicity. Any multiplication of parts only adds to the 

 possibilities of error, and makes such errors as do exist 

 more difficult to detect. 



The method of supporting the specimen in Deeley's 

 machine appears to be good, both in the provision made for 

 allowing a longitudinal variation in the specimen, and in 

 the frictionless support at the lever end. Of course this 

 latter arrangement only admits of a very limited angular 

 movement of the lever, and makes very delicate handling 

 necessary if a weight is used in place of the spring. 



117. The Measurement of Torsional Strains. Some of 

 the principles which should be observed in taking measure- 

 ments of the longitudinal strains of test bars in tension are 

 equally applicable to the measurement of torsional strains. 

 In the first place, the measuring apparatus should, if possible, 

 be attached to, and carried by, the bar itself, so that the actual 

 relative rotation of one cross-section of the bar with respect 

 to some other, may be measured. The observation of this 

 rule is absolutely necessary when only the elastic strains are 

 to be determined. In some torsion machines the plan is 

 adopted of observing the relative movement of the parts of 

 the machine which hold the two ends, and, assuming this 

 to be the same as the twist of the shaft itself ; this cannot 

 be done without introducing an error, because the 

 mechanism of the machine and the holding bushes and 

 keys must undergo a certain amount of deformation them- 

 selves, and this is, of course, registered as a strain of the 

 specimen. If the worm wheel and its fittings are of heavy 

 design, and accurately made, and care is taken that all 

 backlash is eliminated before any observations are made, 

 the error due to the deformation of the machine will be 

 small compared with the strain of the specimen itself. 



For strains beyond the limit and for the rougher kind 

 of tests this method of strain measurement may be safely 

 employed, but for elastic strains it is desirable to take the 

 measurement from the shaft itself, and not from any piece 

 of the machine to which it may be attached. 



The author* has made use of the above method in 

 making a series of tests of solid and hollow shafts of the 



* Popplewell and Coker on Shafts. Proceedings of Institution of Civil 



Engineers, vol. cxxii. 

 P2 



