114 Prof. Bichard Threlfall on the 



used the method was more accurate. The range was from 

 about 80° C. to 36° C. The first experiment gave 3a= 

 •000060616, and the second, 3a = -000060666. Whence we 

 deduce a = -0000202. 



Experiments on " NacliwirkungT 

 The fibre used was pulled out by hand so as to leave the 

 ends thick for attachment. The length of the " uniform " 

 part may be taken as about twelve inches, and its thickness at 

 '0102 centimetre. One of the thick ends of the fibre was 

 rigidly clamped, and the other carried a fine mirror ground 

 and polished in the laboratory for galvanometer-purposes. A 

 cross bar was fastened to the lower thick portion below the 

 mirror, and this was twisted in the usual manner by a glass 

 cell. The subjective method was used. One complete rota- 

 tion corresponds to 2500 scale-divisions, and the scale was 

 easily read with certainty to tenths of a division. At first the 

 twist w T as put in by successive additions of five complete 

 turns at a time. The fibre was usually left twisted for five 

 minutes and then slowly allowed to come back to its original 

 position. No distinct" Nachwirkung " was observed till the 

 fibre had been twisted through fifteen turns (i. e. 37,500 scale- 

 divisions) and left in its place for 16 hours 40 minutes. On 

 releasing the thread it seemed to take 3 hours 40 minutes to 

 come back to near its zero position, and thus untwisted by 3*7 

 divisions. This creeping back seemed regular and normal. 

 We deduce from this and other experiments that one may 

 safely twist a thread of *01 centim. diameter through one 

 turn per three centimetres of length, and leave a good margin 

 of safety, since Nachwirkung first appeared when the rate of 

 twist rose to about one turn per two centimetres of length. 

 Since the intensity of the strain is proportional to the diameter 

 for a given rate of twist, this result allows us to estimate the 

 maximum twist that may be used with safet}^ with fibres of any 

 thickness. Number of turns per centimetre must not be 

 greater than .qi \ 



— -. : : X ~ . 



Diameter in centimetres 3* 



Torsional Fatigue. 

 After vibrating the thread that had been used for the 

 temperature-coefficient of the rigidity, continuously (in the 

 daytime) for a week, Mr. Adair failed to discover any signs 

 of torsional fatigue. The period of vibration remained the 

 same as ever. The experiments, however, were too rough to 

 enable us to say more than that torsional fatigue, if it exists, 

 is too small to afiect the readings of any instrument in which 

 threads are likely to be used. 



