196 DR CARGILL G. KNOTT ON THE 



First, we notice that palladium is very similar to platinum in the manner 

 of its changings, tending, however, to diminish in rate of change as compared 

 with the platinum at higher temperatures. Secondly, we see at a glance 

 that the behaviour of the nickels is very peculiar. About a temperature 

 of 180° or 200° C. the rate of growth of resistance of a given wire with 

 temperature undergoes a marked increase, and experiences a more evident 

 decrease at a temperature somewhat above 300° C. Throughout this range 

 of temperature the comparatively great slope in the resistance curve is very 

 striking. 



Thirdly, there seems to be a similar increase in the rate of growth of resist- 

 ance of iron wire, occurring at a temperature a little below 600° C. It was 

 unfortunately impossible to attain a much higher temperature with the means 

 at our disposal ; but in the very highest readings obtained there was sometimes 

 an indication of a decrease setting in, as in the case of nickel. In the curves 

 as drawn the peculiarities of the iron are not very distinctly shown. It was 

 thought better, however, to draw the curve to the same scale as the curves for 

 the other metals than to proportion the co-ordinates to make it well-conditioned. 

 The curves indicate at once the extremely great increase of resistance in 

 iron as compared with other metals. This is in accordance with former experi- 

 ments ; and the results here obtained agree fairly well with Von Walten- 

 hofen's results for steel (see Wiedemann's Electricitdt, vol. i. p. 525). The 

 measurements made by other experimenters do not agree nearly so well — as a 

 rule, a much smaller increase has been found. # 



In this paper, however, no special emphasis is laid on the results for iron, 

 except that they cannot be represented by any ordinary empirical formula, such 

 as C. W. Siemens has given. So far as they go, they bear out our result of 

 twelve years ago, that the rate of growth in resistance of iron experiences a 

 marked increase at a temperature of a dull red heat. This peculiarity has now 

 been proved to exist in the case of nickel, occurring however at a much lower 

 temperature. The further peculiarity, so distinct in the case of nickel — namely, 

 the subsequent decrease in the rate of growth — probably exists also in the case 

 of iron. Indeed, on Von Waltenhofen's authority, the continued increase of 

 resistance of steel as the temperature rises from a red heat to a white heat 

 tends to evanescence. This bears out the statement made above. Thus, it 

 appears that iron and nickel agree in a certain peculiarity in the rise of their 

 resistance with temperature. This peculiarity may be thus described. Within 

 a certain range of temperature, the resistance of a given wire increases at a 

 more rapid rate per temperature degree than at temperatures above or below 

 this particular range. For nickel this range lies between 200° and 320° centi- 

 grade ; for iron between a dull red and a bright red heat. Now, it is exactly 

 within these ranges respectively that the thermoelectric peculiarities of nickel 



