Barus and Strouhal— Viscosity of Steel. 27 
hardness may be typitied by sealing-wax ; soft steel by tallow.* 
Nevertheless the continuous variations of these properties ex- 
hibited by steel is as unique as it is striking. Indeed we felt 
diffidence in reporting this result and have taken pains to sub- 
stantiate it. 
The motion of the bifilar body of the above apparatus can 
be interpreted from two standpoints: Hither it is due to the 
torsional couple and the result of viscous yielding of the harder 
wire relatively to the softer; or it may be due to the bifilar 
and flexural couple and is then the result of viscous yield- 
ing of the softer wire relatively to the harder (xxxii, p. 466). 
But the bifilar and flexural couples have been proved to be 
zero and to produce a zero effect. Hence the inference above 
italicized is alone admissible. Again, in the bifilar apparatus 
where steel] is twisted against glass (table 28) the soft steel is 
demonstrably more viscous than glass. Hard steel, as shown 
by its behavior with the same fibre, is less viscous than soft 
steel. We do not wish to say that it is less viscous than glass 
because the sectional area of the latter fiber is larger. Again, © 
in table 31, which contains the data obtained with the tubular 
apparatus, eoft steel yields viscously at about the same rate as 
the brass tube of more than four times the sectional area of the 
steel wire. Hard steel under the same circumstances yields at 
a very much greater rate than brass. 
Our results for degrees of hardness higher than “Annealed 
100°, 10°” are to be regarded incomplete because of the mag- 
netic importance of those degrees. As steel passes in hardness 
from “Annealed 450°” to “Annealed 1000°” (soft), it probably 
marches through pronounced maximum viscosity. This result 
is pretty clearly indicated by table 83 and figure 3. Here also 
the results are to be regarded incomplete because of the mag- 
netic importance of the (soft) degrees of hardness in question. 
Our methods of annealing between 500° and 1000° are not as 
yet satisfactory. 
2. If we compare the results of table 33 or of fig. 3 with the 
known thermo-electric behaviort of steel wires, we detect a 
very striking similarity in contour and position of correspond- 
ing members of the viscous and thermo-electric families of 
curves. Both phenomena practically subside in the first 
phase of annealing; the effect of temperature becomes rapidly 
less as higher degrees are approached. 
3. The relations between hardness and viscosity here eh- 
countered may perhaps be conceived somewhat as follows: 
Suppose stress to be so distributed in a solid that its applica- 
*These examples (tallow, sealing wax) are given by Maxwell: Heat, p. 296, 
Appleton, N. Y., 1883. 
+ U.S. Geol. Survey, Bull. 14, pp. 54, 55. 
