METEORIC AND ARTIFICIAL NICKEL-IRON ALLOYS. 
99 
gaps in the magnetic circuit at 15° C. which are absent when the 27 per cent, alloy 
is magnetic—as it is when the upper curve is obtained. Consequently, during 
heating in the first case, the loss of magnetism by the reversible alloys merely causes 
increase of size in gaps already present, while in the second case the gaps are absent 
initially and are produced by the loss of magnetism of these reversible alloys. 
Now the effect upon the induction through a circuit of the formation of a small 
transverse gap is very much greater when no gap is present initially than when the 
gap formed merely increases the size of one already in existence. Thus the induction 
B (in a given field), when there is a transverse gap equal to the fraction 6 of the 
circumference in a ring of permeability /x, is 
B = B 0 /{l + d(/x — 1 )}, 
where B 0 is the induction when the gap is absent. And since, from the above 
expression, 
clB/dd = — B (p,—1)/{1 + 0 (p,—l)}, 
it is seen that the effect of a small increase, dd, in the gap is greatest when 9 — 0, 
and, assuming /x — 1 = 100 (which is approximately the case in the present experi¬ 
ments), the percentage variation of B with 6 is twice as great when 9 = 0 as when 
9 = 1/100. 
The above expressions apply only when the permeability of the material which loses 
its magnetism is assumed to be the same as that of the rest of the material. If the 
permeability of the former is less than that of the latter, the effect is of the same 
character, but less pronounced in proportion as the difference between the permeabilities 
increases; but the above computation is sufficiently accurate for the present purpose. 
It shows, when compared with the numerical data in the tables, that the effects 
observed are of the order of magnitude theoretically deducible, and confirms the view 
that they are due to the formation of breaks in the magnetic circuit in the way 
described. 
§ 10. The thermo-magnetic data seem thus to point to the correctness of the view 
that the taenite is composite, that it is transformed by successive heatings into a 
relatively more homogeneous alloy (containing between 25 to 30 per cent, of Ni) and 
that the transformation takes place by the gradual interdiffusion of its richer and 
weaker constituents (containing probably not less than 6 per cent, of Ni in the one 
case and probably not more than 40 per cent, of Ni in the other). 
Fig. 28 summarises the evidences for these conclusions. The full curves indicate 
the condition of the meteoric iron after it had been heated four times above 800° C. 
They have been plotted from the numbers given in the table on pp. 42, 43, using the 
constants of the 1st winding given on p. 39. The dotted curves indicate the state of 
the material after it had been heated about twenty times above 800° C., including 
once above 900° C. They were plotted from the data on pp. 43 to 45, using the constants 
o 2 
