638 MR JAMES RUSSELL ON 



are necessarily higher than what might be called the effective average induction which 

 might be fairly assumed in order to arrive at the theoretical shielding ratios. Trial shows 

 that if the ordinates of the induction curves fig. v. are reduced one seventh, the theoretical 

 shielding values based thereon very closely approximate to the experimental ratio 

 curves of fig. vi. The dash lines give a fair idea of the approximation. 



We therefore conclude that when the transverse field is increased to a maximum of 

 130 C.G.S. units, the shielding ratio g' approximates to 



1 d KB 

 " = 2 R ■ "H + ] 



where B is the maximum induction and K is a constant somewhat less than unity. Thus 

 the shielding ratio, minus unity, is approximately proportional to the ratio permeability 

 and not to the differential permeability. 



I. (a) Experimental Shielding Ratios. — Transverse Field 

 decreased by steps from a maximum. 



§12. Table VI. gives the measurements obtained for both shields under the above 

 conditions when the transverse field is decreased from a maximum of about 80 C.G.S. 

 units, and also when it is carried through zero to a negative maximum. In figs. vm. 

 and ix. the values of the weakened shielded field (H) w (full line curves) and of the 

 negative residual field (H) r (dotted line curves) given in Table VI. are plotted against 

 the corresponding values of H, for shields A and B respectively. It will be noticed 

 that (with the exception of small values of H ( immediately after changing sign), the 

 measurements of the residual field (H) r are of the opposite sign when compared with 

 the measurements of the transverse field (H) t . This is obviously due to the residual 

 magnetisation in the sides of the iron shields completing its circuit partly within the 

 shields, when the transverse force sustaining this magnetisation has been withdrawn. 

 With the first shield experimented with, viz., shield B, measurements of the residual 

 field were taken throughout, but they have not been retained in these tables, with 

 the exception of those given in Table VI., as results did not appear to justify the 

 extra complication introduced, and they moreover made the work unnecessarily laborious, 

 more especially when the circular field was introduced, as after each determination 

 the iron shield had to be demagnetised in order to retain the same magnetic sequence. 

 With the second shield A, therefore, the residual field (H) r was not measured except with 

 the decreasing values of H t given in Table VI. 



In fig. ix. the value of (H)„, (Table V.) (faint full line curve), and the values of (H) r 

 (not given in these tables) (faint dash line curve) have been added when the transverse 

 field is increased from zero. Dealing therefore with shield B in the first place (fig. ix.), 

 we observe that with values of H 4 decreasing from about 80 C.G.S. units, the field 

 within the shield in the neighbourhood of H, = 50 becomes zero ; while with an increas- 



