t)uctile Materials under Combined Stress. 91 



ascertain directly the value of a scale-division in terms of 

 the axial strain. To do this the change of scale- reading 

 caused by one or more turns of the screw, the head of which 

 is graduated and the pitch of which is known ; then, by 

 taking account of the multiplication by the levers and of the 

 length of the specimen, the axial strain per scale-division is 

 easily found. The screw Q was compared directly with a 

 Brown and Sharpe micrometer, and the two were found to 

 agree to the j^^ inch in eight turns or 0*2 inch. A con- 

 sideration of the behaviour of the instrument under the 

 supposition that the specimen bends or twists will show that 

 these do not affect the reading, which will always represent 

 the mean axial elongation. One kinematic restraint, it will 

 be noticed, is wanting : this, however, would be supplied by 

 a wing attached to the length-bar and arrangements similar to 

 YZ, but would have no effect on the working of the instru- 

 ment ; it is therefore omitted. Figs, 5, 6, 7, 8, and 9 

 represent parts of the extensometer : fig. 6 being the top 

 grip, 7 the lever, 8 a trihedral hole, and 9 a side-view and 

 sectional plan of the mirror and its supports, on a rather 

 larger scale than the other parts. 



The instrument proved sensitive and trustworthy ; the tests 

 of it for sensitiveness made on the Norwegian iron specimen 

 previously referred to are given in Table IV. The scale was 

 placed at such a distance that the extension in 8 inches 

 corresponding to one scale-division was about the one hundred 

 thousandth of an inch ; the nearest scale-division is recorded, 

 except when the cross-hair of the telescope was apparently 

 midway between two divisions. 



34. Errors of Extensometer. — In a figure formed by 

 plotting out Table IV. it can be noticed that the hysteresis- 

 loop, due to the yield-point having been reached previously, 

 is well marked, and its width is considerably greater than the 

 instrumental error ; the errors, which are quite insignificant, 

 are about the value of one scale-division, and include any 

 error due to the want of sensibility of the testing-machine. 



35. A JSeio Twist-measuring Apparatus. — The apparatus 

 for measuring the twist consisted merely of a pair of mirrors 

 mounted on the grips 2 and 3, a scale placed horizontally 

 being read by light reflected at each of the mirrors. Two 

 arrangements occurred to me : one in which the mirrors had 

 their normals perpendicular to the axis of the specimen (see 

 fig. 12), and one in which they were inclined at 45° or so 

 (see figs. 10 & 11). In these figures the course of the ray 

 of light is indicated roughly by the broken line, the scale 



