70 
MR. S. W. J. SMITH ON THE THERMO MAGNETIC ANALYSIS OF 
comparatively restricted and the argument is based partly upon a process of 
extrapolation. 
It is known that there is no appreciable change in the density of nickel when it 
passes from the non-magnetic to the magnetic state, and hence the metrological method 
used by Guillaume (depending upon observation of the change in length of a bar of 
the nickel iron) affords no information concerning the changes that may take place in 
the nickel during the crystallisation of the iron by which it is accompanied in 
the alloy. 
§ 6. The marked change in the magnetic properties of nickel-iron alloys when the 
percentage of nickel passes through a value between 25 and 30 is shown clearly in 
the data given by Hopkinson (‘ R,oy. Soc. Proc.,’ vol. 48, 1890), and confirmed by the 
work of subsequent observers (c/!, e.g ., Barrett, ‘Trans. Roy. Soc. Dubl.,’ 1902-1905). 
The change in the magnetic properties corresponds closely with a well-marked 
change in crystalline structure which has been studied in detail by Guillet 
(‘ Metallographist,’ vol. 6, 1903). This change takes place, according to Guillet, 
when the percentage of nickel is 27, if the percentage of carbon in the alloy is small. 
Other data of a similar kind might be cited in support of the hypothesis that the 
permeability is intimately related to the crystalline structure, and of the deduction 
therefrom that observation of the permeability variation may be valuable as a means 
of tracing continuously variations in the state of crystallisation of the material. 
It will be seen that the experiments described in this paper bear directly upon this 
point; but before considering the evidence of the validity of the theory of solid 
solutions, which they convey, it seems well to indicate how, as it appears to me, the 
value of the theory as an explanatory hypothesis can be exhibited by consideration 
of certain data already known. 
The method of presentation of the facts enumerated and discussed in §§ 7-17, below, 
was suggested to me by a consideration of those thermomagnetic data of Section IY. 
which are discussed in Section VII. 
§ 7. As already stated, the positions of the dotted curves AEFCI) in fig. 23 
above are in the first instance purely qualitative. The values of GE and GF, for 
example, which represent the compositions of the two components of the supposed 
eutectic are not known from direct experiment upon artificial alloys ; but a study of 
the structure of nickel-iron alloys of natural origin (“meteoric” and “terrestrial”) 
strongly supports the view that the equilibrium curves are of the general form given 
in the figure. 
The results of tire chemical analysis of meteorites (see, e.g., Cohen, ‘ Meteoriten- 
kunde,’ 1894-1903) show that the amount of nickel (together with cobalt) found in 
meteoric iron is seldom less than 6 per cent, or more than 11 per cent. It is probably 
for this reason, as will be seen, that Widmanstatten figures are usually regarded as a 
characteristic of meteoric iron. But while the occurrence of large and almost 
perfectly symmetrical figures may, in the present state of knowledge (since the means 
