Relations of the Alloys of Platinum. 431 



result under a somewhat involved form, was well justified in 

 computing by means of it the conductivity of a pure metal 

 from data found for metals slightly impure. This computation 

 premises the truth of Matthiessen's other principle that with 

 certain distinct exceptions, the electrical temperature cofficients 

 of all pure metals are the same (cf. § 6). 



4. After Matthiessen and Yogt the curious relation in ques- 

 tion seems to have failed to enlist further attention, and I 

 believe that the next systematic investigation is that made by 

 Dr. V. Strouhal and myself in studying the electrics of the 

 iron-carburets.* We did not, however, in the former paper 

 reduce our results from the curvilinear form in which they 

 appear when temperature coefficient is expressed in terms of 

 resistance, to the curves of the present linear character ; and 

 hence I may expediently make this reduction here. If tem- 

 perature coefficient (y) be expressed as a function of specific 

 resistance (a?), and if a curve be passed through all the points 

 investigated for iron, steel, cast-iron, then the following prin- 

 cipal coordinates obtain : 



x— 15 45 70 



2/=0-00420 0-00166 0'00130 



Interpreted by an hyperbolic equation of the form 

 (x+l) (y+m)=n, these data lead to constants 



l= — 3-73, m— — 0*000706, w = 0-0394, 



which do not reproduce the graphic curve satisfactorily. 

 Neither is I by any means negligible, so that the reduction to 

 linear forms is out of the question. At first sight this seems to 

 prove that iron, steel and cast iron do not here form a unique 

 series ; that the resistance variation due to the change of car- 

 buration from iron to cast-iron is in its nature different from 

 the resistance variation observed on passing from soft to hard 

 steel. It appears below, however, that the favorable position 

 of iron here unduly influences the result. I will temporarily 

 withdraw both iron and cast-iron from the series. For steel 

 alone we found (using the graphic method already referred to) 



a; = 15-9 28'9 45'7 



y = 0-00423 0*00244 0-00161, 



which data interpreted by the equation (x+l) (y+m)=n now 

 lead to the constants 



1=0-18, m= -0-0001435, ra=0 , 0682. 



Here the constant I is small, being only about five per cent of 

 the smallest steel value of x admitted. Inasmuch as a result 



* Cf. Wied. Ann., xx, p. 525, 1883 ; Bulletin U. S. G. S., No. 14, p. 15 to 25, 

 1885. 



