FORCES ACTING IN MAGNETIC CIRCUITS. 205 
traction which could not be explained by unavoidable experimental 
errors. With short bars and high inductions necessitating the use 
of very strong fields, some induction is included by the testing 
coil which does not help the traction and which tends to make the 
calculated traction appear too large. When this source of error 
was eliminated no greater discrepancies were observed with short 
bars than with long ones. 
10. I conclude therefore : (i.) The traction produced by a given 
tube of induction when running out of air into iron and crossing 
the surface normally is independent of the nature of the iron or 
of its form. I had a difficulty in bringing myself to believe this, 
but the conclusion seems inevitable. 
Corollary (i.) The magnetic forces are independent of the stresses. 
in ether inside the iron. 
Corollary (ii.) Setting aside Prof. J. J. Thomson’s stresses, the 
ether stress in air is less than that in iron—assuming that Max- 
well’s “ Magnetic Material” sufficiently represents iron. The 
difference of tensions is 
B? 
3, 47 (BH — 3H’) 
ps 2 
or (ey 
8 ir 
This is an unbalanced stress, and if the lines of induction in 
the iron give rise to forces similar to those produced in air, this. 
must mean that the boundary tends to be pulled off the iron. 
Taking Prof. Thomson’s stresses into account, this effect may easily 
be reversed in any actual case. 
Referring to Prof. Thomson’s investigation, (Physics and 
Chemistry), I cannot avoid the impression that there still remains 
a set of stresses depending on the variation of elastic constants 
with temperature. This would further complicate matters. 
11. Each tube of induction is therefore a tube of force within 
the usual definition, but it does not follow that the only forces 
are those represented by the tubes of induction. If the tubes 
