BIGIDITT OF BOCKS AND BYSTERESIS FUNCTION. 9 



the specimens were twisted to and I'm cyclically up to the amplitude 

 6 given below : — 



tf='2-68 ; =0-71 ; =10-07 ; = 1611 ; = 20T4 X 10 fi c.g.s. units. 

 The result of the experiments is plotted in PI. IV. It will be seen, 

 at a glance, that 



1. If the curve be divided into several sections at the positions 

 where the variation of couple changes its sign, each section, 

 except the one which starts from the origin, will be approxim- 

 ately straight, provided a comparatively small portion im- 

 mediately after the turning point is left out of consideration. 



2. None of the sections except the one which starts from the 

 origin, pass through the origin. Each section for which the 

 variation of couple is a positive, or on-curve, lies below the 

 origin, i.e. it passes through the fourth quadrant : while each 

 section for wdiich the variation is a negative, or off-curve, passes 

 through the second quadrant. 



3. Suppose a straight line to be drawn coinciding as nearly as 

 possible with each section. Of these representatives, that which 

 represents the section which belongs to a cycle of greater 

 amplitude lies further from the origin. 



4. That which lies further from the origin makes a greater angle 

 with the axis of couple. 



Of the above facts, the first shows that Hooke's law is approxim- 

 ately fulfilled, provided the change of couple always takes place in 

 one direction only ; and that there must be a certain disturbing 

 cause, prevailing especially in the neighbourhood of the starting point 

 and at the turning point of the course. The second shows that there 

 is certain dissipation of energy for each cycle, due either to the 

 imperfect elasticity of the twisted rocks or to the imperfection of 

 the twisting apparatus. Observe that the latter is a constant. The 



