Permeability of Iron and Steel. 91 
attained. In this case A may be estimated from the formula 
w (5—8,)/ 
o(5—8,) —a(S—d)” 
where one accent refers to the first of two assumptions, two to 
the second. 
Before this way of estimating A was devised, the only way 
of approximating to it was to carry out the whole calculation 
of w from % for each assumed value of A, and then judge 
from the results. It was soon seen that the representations 
obtained differed in the position of the maximum of w. The 
following correspondences were thus obtained for ring H:— 
A ce (A/—A”) 
Assumed. Value of 1% for which p is 
A. a maximum. 
1 Between 7000 and 11,000. 
= About 7000, still too late. 
4 Before 6400, a little too early. 
An accurate judgment could be formed by the correspondences 
of uw before and behind the maximum. 
The application of the above method saved the great labour 
of these complete computations by way of trial. 
It is necessary to repeat the above processes until the 
resulting value of w(6—6,) is sensibly zero. The quantities f 
and — obtained in the computation which satisfies this con- 
dition are taken on for the final process. 
The next thing is to form the table connecting % and @, 
from equation (3). This is most conveniently done by form- 
O 
—wf 
ing a table of log ae once for all, for each degree from 
about 20° to 60°*. The addition of tliese logarithms to that 
of : gives the values of % corresponding to the degrees of w0. 
This table (%,0) is written out, with differences. The 
values of w§ corresponding to the experimental values of G 
are then found by interpolation. 
Multiplying @ by 7, we find 6 corresponding to each 
experimental value of %. 
With A, 6d, and G, pu is calculated from equation (1) for 
each experimental number %. ; 
The following are comparisons of this theory with experiment. 
'* For steel the table is formed for every 10” from 60° to 59° 59’, and 
then for every minute down to 59°. 
H2 
