ACOUSTICS AND GRAVITATION. 83 



may be prolonged and terminate in the thin metallic plates p, p'. These are 

 to be surrounded by the narrow cases c, c', to secure air-damping. The device 

 is now much more efficient than in the former apparatus, for there are no loose 

 connections or scale-pans and the cases c, c' may fit rather closely to the parts 

 p, p' of the beam, with but a small opening in front, for ^dbration. The counter- 

 poise W, by which the torsion of the beam is to be maintained, is adjustable 

 but rigidly attached at some convenient point. 



The cases c and c' were each placed on three leveling-screws below, and 

 held in place by a stout Vertical spring above, pressing downward. In this 

 way it was easy to free the plates p, p', particularly as the cases were adapted 

 to slide (with friction) on the leveling-screws. 



The new method thus depends on the degree to which the viscosity of the 

 wire vanishing at a retarded rate through 

 infinite time may be adequately treated 

 as a correction. The experiments showed, 

 however, that it also depends to an alarm- 

 ing extent on the thermal coefficient of 

 rigidity, if ordinary metals are used, though it remains to be seen in how far 

 the Guillaume alloys will meet the conditions. 



72. Observations on the permanently twisted wire. The readings were 

 made two or three times daily, as a rule. They were adequately recorded in 

 the graphs, figures 106 and 107, for a period of about a month. The ordinates 

 show the displacement of the micrometer A N, the divisions of the graph being 

 in steps of A7V=o.ooi cm. As the twist produced by the excess weight of 

 about i gram was here (as above) about 50, we have dg/g= i.6X io~ 4 for each 

 of the divisions (o.ooi cm.) of AjV. The laboratory temperature was very 

 variable from day to day, and the effect of this in the graphs is astonishing, 

 but contributes essentially to the interpretation. In figure 106, the average 

 rate of yield may be estimated as AW = 0.00091 cm. per day, as the result of 

 viscosity. This makes dg/g vary 0.00014 per day, a quantity in itself too 

 large to be used as a correction. Even a more serious consideration is the ther- 

 mal increase of rigidity. It is this feature which makes the graph so exceed- 

 ingly jagged. If we take the large drop between the eighth and tenth days 

 to estimate this effect, the data would be 



Temp. t = 22.0 14.8 19.8 

 io 3 A]V= S.i 2.6 8.7 



which is equivalent to A7V/A* = 0.00095 centimeter per degree, or dg/g = 

 i.6Xio~ 4 for each degree of fall of temperature as the equivalent of in- 

 creased rigidity. * 

 The graph, figure 107, is actually worse; for here, between the twentieth and 



* Excellent observations on the torsional rigidity of iron and steel, due to Pisati, are 

 given in Landolt and Boernstein's tables, 1905, p. 44. The coefficients are 2.1 xio~ 4 fo r iron 

 and 2.3 X io~ 4 for steel, therefore even larger than the above estimate. Moreover, the above 

 method could easily be modified to measure this constant accurately. 



