METEORIC AND ARTIFICIAL NICKEL-IRON ALLOYS. 
87 
the case of the artificial nickel steel. Thus, considering only the curves given by 
ITopkinson and the corresponding curves of fig. 18, it would appear that all the iron 
had become magnetic when, during cooling, the temperature had fallen to about 550° C. 
But the later measurements show that the material is not in a state of final 
equilibrium at temperatures below 600° C. until it has undergone successive alternations 
between that temperature and 15° C., and that, when this state is considered, the 
behaviour of the steel is analogous to that of the meteoric iron (cf. § 12 below, p. 92). 
§ 6. If the cooling is interrupted at some temperature below that at which 
magnetism appears and if the temperature is then varied between this and some 
higher temperature (cf. fig. 26, I.), the amount of magnetic iron should increase at first 
relatively rapidly and finally very slowly, as already explained. The slope of vw 
(figs. 26, I., and 27, IV.) should be greater than that of xy, and that of wx should 
exceed that of yz. In the same way the permeability change corresponding to vw 
should exceed that corresponding to xy, and that corresponding to u'x should be 
greater than that corresponding to yz. If diffusion can only take place slowly, the 
increase in permeability will become imperceptible after a few alternations ; xy and yz 
will be practically horizontal, and the temperature coefficient of permeability between 
the temperatures 6 X and d 2 (fig. 27, IV.) will become practically identical with that 
observed when all the iron is magnetic. 
Examination of the data given in the tables shows that after the first alternation 
between given temperature limits in the meteoric iron and in the nickel steel, the 
change produced by subsequent alternations (during the time occupied by the 
experiments) is practically inappreciable, i.e., xy, yz, &c., are practically superposed. 
An example of this is given in the data 30 to 34, meteoric iron, 2nd winding, fig. II. 
A similarly rapid approach to a steady state is shown in the data 82 to 86, fig. 13. 
Besides these examples of the process vwxyz, there are also numerous examples of 
the process vwx in the tables, which incidentally lead to the conclusion that if y and z 
had been observed in these cases the results would have been the same as in the 
experiments above mentioned. For the temperature coefficient of permeability over 
the range represented by wx has, within the limits of experimental error, in each case 
reached almost the same value as that over the same range when the material is 
fully magnetic. This is seen by comparing Experiments 51 and 52, 1st winding, with 
8 and 9, and also Experiments 27 and 28, 75 and 76, 79 and 80, &c., 2nd winding, 
with 92 and 93. Also 906 and 91, 95 and 96, nickel steel, with 64 and 65. 
§ 7. The observations upon the nickel steel when fully magnetic between 350° C. 
and 600° C. lie very approximately on a parabola in which the relation between the 
constants is such that l/y. dy/d0 has a maximum value at about 570° C., but varies 
so slowly with change of 6 that its value is ‘0041 at 500° C. and ’0042 at 650° C.— 
the maximum value being about ‘0044. On account of the peculiarities of the 
meteoric iron curves a comparison cannot be obtained over this range, but the 
observations 92 and 93, fig. 13, correspond with a mean value of l/y. dy/dd equal to 
