﻿Barus and Stroithcd — Viscosity of Steel. 451 



bifilar coefficient is less than 0*4 per cent of the combined tor- 

 sional and flexural coefficients, in very unfavorable values of 

 the chosen dimensions. This experimental result agrees with 

 the one calculated above. 



3. The following static method is better, being exhaustive 

 and final. Let the lower ends of the pair of identical steel wires, 

 A and B, be fastened with the bolt. Let rotation at this end 

 be temporarily checked. Then we may store like degrees of 

 torsion by twisting the upper ends of the wires in the follow- 

 ing three ways : 



a. Let the wire A be twisted +360° and then fastened to B. 

 Left to itself the upper end of the system will rotate —180° so 

 that the residual torsion of A is +180°; of B is -180°. The 

 total couple actuating the bifilar body then is of the form 



b. Let the wires A and B be twisted +180° and — 180 u 

 respectively and then fastened together. Left to itself the 

 upper end of the system will not move. The residual torsion 

 is therefore the same as in case a; but the total couple here 

 actuating the bifilar body is J/ T ±0. This is the case premised 

 in the analysis which introduces this paragraph. 



c. Let the wire B be twisted —360° and then fastened to A. 

 Left to itself the upper end will rotate +180°, so that the 

 residual torsions on A and B are the same as in cases a and b\ 

 but the total couple which in case of viscous motion actuates 

 the bifilar body has now the form ^pM^+Mr^Mf. 



This is the device: We are able to com mutate, as it were, 

 the combined flexural and bifilar couples relative to the tor- 

 sional couple of fixed sign. In the results below this test is 

 frequently applied and we will there show that for the chosen 

 dimensions the torsion couple need alone be considered. 



4. In the case of the tubular apparatus the present precau- 

 tions are at once superfluous. Here each wire is introduced 

 into the tube and compared with the next wire introduced 

 under like circumstances. The tubular apparatus is specially 

 adapted for the investigation of viscous detorsion in its depend- 

 ence on the applied torsional stress, as well as on time. But 

 the same work may also well be done with the bifilar. 



The viscous effect of the relative sections of the wires and 

 allied observations, with a bearing on viscosity proper, can 

 more expediently be made after the experimental data have 

 been communicated. It is convenient to insert a few remarks 

 on the rigidity of efficient parts of the above apparatus here. 

 If the tubular adjustment d be the diameter of the wire, d x and 

 D x the diameters of the tube, if L symbolize the effective length, 

 (p the amount of stored twist, G the rigidity, we have, in 

 view of the equal couples, 



