30 Barus and Strouhal— Viscosity of Steel. 
rarely be such that the torsional yieldings continually equalize 
each other, no matter what relations of section may be chosen. 
The motion-curve of the bifilar-suspension will show maxima 
or even points of circumflexion, such as have been actually 
encountered in tables 28, 29, 29A, and the definition of rela- 
tive viscosities of the two wires will become correspondingly 
involved. 
Again, 4, for solids is not merely a function of time but very 
essentially a function of stress. Above we show (xxxil, p. 452) 
how by simply adjusting the lengths of the bifilar wires, vis- 
cosities may be compared at a given temperature, as time- 
functions with identical values of the parameters stress, strain 
(sectional area). We do not believe that viscous detorsion in 
this full relation has ever been rigidly investigated. The re- 
sults would lead to families of curves. 
Given a quadrifilar arrangement of four viscously identical 
wires. Let the twist ¢,, be stored between any two of them 
and then let the twist ¢,,,,, be additionally stored between 
these two as one, and the third and fourth wire as one. Such 
a device enables us in the above way to study viscosity in its 
simple dependence on strain, for all values of stress. We 
have not, after some searching, been able to find definite 
evidence of a viscous strain-effect. 
Viscosity and strain.—1. When we commenced the present 
research a comparison of the viscosity of glass and steel ap- 
peared desirable. In tables 27 and 28 such comparisons have 
been attempted, though we regret that our available time has 
not permitted us to pursue them further. In table 27, tlie 
sectional area of glass fiber is less than that of steel wire; 
nevertheless the viscous yielding of the fiber is so much more 
rapid than that of the wire, that we may reasonably infer 
degrees of viscosity of the same order in the two substances. 
In table 28, the sectional area of glass exceeds that of steel. 
Hence these data prove that the torsional viscosity of annealed 
steel is greater than that of glass. The viscosity of hard steel 
during the first ten hours of detorsion is certainly very much 
greater than that of glass. During the remainder of the time 
it is decidedly less. The curve passes through a maximum for 
which point the rates of viscous detorsion of glass and of glass- 
hard steel exactly coincide. Since the sectional area of glass 
is greater, we safely infer that the viscosity of glass is not uni- 
formly greater than that of glass-hard steel. It is well to call 
to mind, however, that the sum of the torsions is here only 
{.+¢,=90°. Moreover, since for equal couples and dimensions 
Lijbj=Hw, where H is Young’s modulus* of resilience for 
* To obtain an estimate it is sufficient to accept the same ratio of modulus of 
torsion to modulus of longitudinal resilience for each case. Poisson’s ratios for 
elass and steel are about as 26 to 30. 
