Strains Mechanically Applied. 167 



wire an arched structure. In steel, at least, there is a dense 

 shell surrounding an unusually rare core in such a way that 

 the density of the whole mass is materially lessened. Con- 

 ditions favourable to the retention of high-strain intensities, 

 are also conditions favourable to the occurrence of molecular 

 instability. Hence the marked loss of viscosity of a drawn 

 wire, as compared ccet. par. with a soft Avire. This premised, 

 it appears that in the case of a wire hardened by simple trac- 

 tion the strain retained after traction is of insufficient 

 intensity to be accompanied by marked molecular instability. 

 More rigorously : if the wire during traction has experienced 

 a strain S, which strain after the tensile stress is withdrawn 

 diminishes to S' (S > S'), then the wire need exhibit no change 

 of viscous quality. For in the above cases of thermal anneal- 

 ing and torsional motional annealing (§§ 2, 9), the greater 

 tensile strain S, supposing its action sufficiently prolonged, 

 has wiped out all motional instability for strains 8' < S ; so 

 it follows here that to evoke viscous deformation by aid of 

 tensile strains, the wire must be examined during traction, 

 and preferably under conditions of strain near the limit of 

 rupture. 



11. The apparatus used in these experiments is shown in 

 fig. 1, in which ah and cd are the two steel wires to be coun- 

 ter-twisted. The system is fastened above and below to two 

 massive torsion-circles, A and B, respectively. The inner 

 ends of the wires are joined by a strong brass rod be, carrying 

 a symmetrical circular platform near the lower end, on which 

 the lead scale-weights C C C" &c. (4 kilogs. to 5 kilogs. 

 each) may be supported. In order to facilitate quick work, 

 the connecting rod he is provided with a pair of cross-vanes 

 D, D', submerged in the water contained in an annular 

 trough, fgih fg'i'h'. Finally, the mirror m adjustably at- 

 tached to the rod he (readings by telescope and scale) 

 indicates the difference of viscous motion of the two wires in 

 consequence of a fixed rate of twist stored between A and B. 

 I may add that the rod he can easily be chosen light enough, 

 compatibly with strength, so as to introduce no viscous effect 

 of its own. 



12. The data to test the above are given in the following 

 tables, of which Nos. IV., V., VI. exhibit the behaviour of 

 some drawn steel wires. In each case the comparison is made 

 with my steel normal No. 15 (the lower wire in fig. 1, 

 An. 450° and twisted to permanent viscous qualities). The 

 drawn wire in Table IV. is in the moderately resilient bright 

 commercial state, very soft to the file. The wire in Table V. 



