620 DE W. PEDDIE ON 



certain that increase of (l) must, at all angles, be, on the whole, the determining cause 

 of the change of configuration of the groups. Thus increase of stability, because of 

 increase of (l) and (2) together, is the only case requiring consideration. 



Numerical Value of the Critical Angle. 



If we regard the scale-unit, used above in the measurement of y, as the true unit 

 which gives nb constant, the dimensional data already given show that, in the wire 

 used, the critical angle corresponds to a twist of about 0'09 degree per centimetre of 

 length. But, in testing the constancy of nb, I frequently observed that the chosen 

 value of nb was near the lower limit of the range of values, outside of which it 

 could not be without causing distinct curvature in the system of points whose 

 co-ordinates were log. (x + a) and log. y. Yet, in other cases, the value of nb seemed 

 to be near the upper limit. Thus, when 7^= 1, 210 suits very well as the value of the 

 product. And, if we put Bk" = 210 when n = 1 and choose k = 0'9, we get B = 233. If 

 now n= 1*22, the product becomes 205 ; and these values very well suit, for example, 

 the cases, 22.7.95, 25.7.95 with a = 6, and 30.7.94 with a = 9. Again, when n = 07 

 we get 230 as the value of the product, and this suits the case 24.12.95 with a = 210. 

 Also, if n= 1*33, the product becomes 202, which agrees well with 23.7.95 and 

 26.7.95 (2). 



If we assume these values for k and B, the value of the critical angle is slightly less 

 than that corresj^onding to a twist of 0°*1 per centimetre of length. 



