BAR! s.j PYRO-ELECTRIC PROPERTIES OF ALLOYS. 161 
far as the data go. Referred to Matthiessen's datum they yield an in- 
ordinately high value for the temperature-coefficient of irou. Referred 
to Benoit's values for iron and for soft steel the agreement is better, 
but the computed results are nevertheless high. 
/(0)=12.1; /'(0): /(0) computed =0.0053; found (Benoit) 0.0045 
f(0)=10.9; /'(0): /(0) computed =0.0059; found (Benoit) 0.0050. 
The results of this laborious analysis for iron-carburets has not been 
so satisfactory as the corresponding results for platinum alloy. The 
auses for irregularities are, however, not far to seek. Oast iron is at 
best an exceedingly heterogeneous product and the passage from pure 
iron to pure cast iron must be indefinite because of the divers modifica- 
tions in which the carbon may occur in iron. The variations of occur- 
rence are not fully under control. On the other hand, steel in passing 
from its lowest to its highest resistance does so largely in consequence 
)f definite variations of structure. Finally, in working with iron- 
arburets the difficulties of measurement are of a serious kind. Hence, 
since data have here been coordinated, the changes of value of which 
ire due to causes intrinsically different, it is surprising to find results 
is harmonious as these prove to be. To return therefore to the inference 
)n page 157, with reference to the initial tangents of the loci of copper- 
silver, of silver-platinum, of gold-platinum, of gold-silver, and of the 
otality of platinum alloys, and of steel in all states of temper, it appears 
;hat all these lines show a common tendency to intersect in a small 
ield lying very near the origin of co ordinates. The loci here in ques- 
ion express the relation between electrical conductivity and tempera- 
ure. 
To give perspicuity to the remarks of this paragraph it will be desir- 
ible finally to construct a chart of the linear relations as they have been 
bund. Unfortunately it is difficult to reduce the Matthiessen and Vogt 
lata to the present standards. I have therefore withheld them from 
he larger chart now to be given. In Fig. 27 the individual data for 
platinum alloys and for steel and their computed mean linear loci are 
nserted. The disposition of points with reference to this line is very 
ilearly shown. The steel line has an obviously different slope from the 
ine of iron-carburets (marked Fe), while the latter (Fe) is not far from 
parallelism with the platinum line. It is worthy of note that the iron- 
arburet line contains pure iron, cast-iron, and glass hard steel, sub- 
tances in which the carbon is largely chemically combined with iron, 
whereas annealed steels and soft steels, each of which contains carbon 
incombined, has a linear locus of its own, differing from the iron-car- 
Duret locus. In the above, platinum alloys of nickel, cobalt, and iron, 
iertain irregularities have been discerned, which may in like manner be 
'eferred to imperfect alloying. 
General remarks.— It is not improbable that the relations sketched 
ffould hold indefinitely, provided alloying were not necessarily accom- 
Bull. 54 11' (815) 
