IN TORSIONAL OSCILLATION 115 



from the y axis of the line which the curve approaches asymp- 

 totically gives the value of a. If a wrong value of a be taken, 

 the points in the curve of log y against log (x+a) will not lie 

 in a straight line, a curve convex to the origin being obtained 

 if the value of a be too large, and a curve concave to the 

 origin if the value of a be too small. This is seen 1 to be the 

 case in figure 1, when with a=90 the curve is concave, and 

 with a= 1 10, convex. The value of a which gives the straightest 

 line is taken, and from the tangent of the angle included by 

 the line and the axis along which log y is measured n is found, 

 and 6 can then be got by substitution. 



METHOD OF CONDUCTING THE EXPERIMENT 



The wire under consideration was suspended from a clamp 

 attached to a torsion head, and at the other end was clamped, 

 symmetrically and horizontally, a heavy lead ring of large 

 moment of inertia. To the outer surface of this ring was 

 fastened a scale divided into millimetres. The vibrations 

 of the apparatus were damped out, and the torsion head then 

 carefully turned so that no pendulum oscillation should be 

 set up in the wire. Exterior disturbances were also, as far 

 as possible, avoided. Readings of successive maxima ranges 

 of oscillation were taken by means of a telescope with cross 

 wires inserted, the crossing point being fixed in the same 

 horizontal plane as the lead ring, at a distance of about 6 

 feet from the scale. It was found convenient to miss the 

 first reading, and to take readings at the end of every oscilla- 

 tion after the first until ten oscillations had been completed, 

 and thereafter to take readings after every fifth oscillation. 

 Except in the case of tin wire, in which case the oscillations 

 died down with extreme rapidity, the readings were extended 

 over a hundred oscillations. The zero of the scale was found 

 by taking successive readings to right and left at intervals, 



1 The scale readings y on the diagrams correspond to a rotation through 1 per '175 

 cm. of scale. 



