ON THE MAGNETIC QUALITIES OF IRON. 
1017 
independent. It is interesting to notice also that the easy slope of some of the soft 
iron curves in fig. 22 allovs them to be crossed by curves for steel, even under 
moderate magnetizing forces. For instance, a force of 13 c.g,s. induces more 
magnetism in the rather hard steel of Fing II. than in the very soft iron of Ring V. 
A comparison of permeabilities in respect to any such force as this would furnish no 
criterion of magnetic softness. It may, however, be added that the specimens which 
are soft in the sense of having small hysteresis losses excel in permeability in the 
initial stage of the magnetizing process. Compare, for instance, the relative 
permeabilities of the rings when B is say 1000 : it will be seen that the order in 
which they stand in regard to smallness of hysteresis loss, and in regard to 
permeability, is at this stage the same. If softness is to be judged by the B-H curve 
it is clear that the early portion of the curve is the best guide. 
In a paper read before the American Institute of Electrical Engineers'^ on the 
“ Law of Hysteresis,” Mr. C. P. Steiivmetz has discussed certain experiments on 
the relation of |H dl to B, and contends that the empirical formula 
|h dl = 
where rj is a constant factor, is in good agreement with the results of experiment. 
From the figures Mr. Steinmetz gives, and from those which our own experiments 
supply, it appears that within a certain range of values of B this formula may be 
taken as giving a fairly close approximation to the real value of JH dl. As an 
empirical formula of the kind is of use to designers of transformers, we have been 
at some pains to examine how nearly and within what range a formula of this 
type may be taken to represent the facts. It will suffice to refer to one example in 
some detail. 
Taking Bing IV. (thin sheet iron) for which we have a numerous and very 
consistent series of determination of JH dl, extending from B = 43 to B = 14,720, 
we have tested the constancy of the index e in the formula 
|h dl = t^B^ 
by plotting log B in relation to log JH dl. 
From B = 2000 to B = 8000, the curve obtained in this way is a good straight line 
giving the value 1'475 for the index e, and O'Ol for g. When B is about 8000 the 
inclination of the line changes, and from B = 8000 to B = 14,000 we again obtain 
very nearly a straight line giving e = 1'70 and g = 0'00134. 
Again, the measurements obtained from the small-cycle tests of Bing IV., show 
that from B = 200 toB= 500 the curve of log B and log JH dl is very closely a 
straight line, giving e = 1’9. 
Where B is above 500 the gradient of the line gradually changes. Taking the 
* ‘ Transactions of the Institute,’ Jan. 19, 1892, vol. 9, p. 3, 
MDCCCXCIII.—A 6 O 
