376 Holborn and Day — Expansion of Certain 



are considerably nearer together toward the ends than in the 

 center, but to compensate, the fall in temperature near the 

 two openings through which the observations were made, is 

 hardly possible without tremendously over-compensating the 

 adjoining sections. The reason lies in the fact that one whole 

 turn of wire is lost at the openings. 



The presence of the bar also contributed something toward 

 equalizing the temperature through its own conductivity. This 

 was especially noticeable in the case of silver (6 mm in diameter) 

 when compared with platinum and palladium (each 5 mm in 

 diameter), for example. 



Table I contains observations of the temperature distribution 

 in the two tubes. 



Nickel is included with this series for a special reason. 

 Between the two measurements with nickel to 750° a consider- 

 able interval of time intervened, during which the oven was 

 used for other purposes. These two series therefore show very 

 clearly that the distribution of temperature suffers little from 

 the oxidation of the oven coil with use. 



Table I shows the differences of temperature in microvolts 

 observed at eight points along the bar as compared with the 

 middle. The relation between the thermo-electric force and 

 the temperature for the element used is contained in Table 

 XVI of the original investigation (loc. cit., p. 186). The tem- 

 perature gradient is then obtained graphically and the mean, A, 

 algebraically added to the reading at the middle point to obtain 

 the mean temperature of the bar. 



The observations were always begun with a series of readings 

 at the temperature of the room, which was measured by intro- 

 ducing a mercury thermometer into the oven tube, then at the 

 points 250, 500, 750 and 1000° as nearly as the oven could be 

 regulated. In certain cases the intermediate temperatures 375, 

 625 and 875° were also observed. No point below 250° was 

 attempted on account of the diminished sensitiveness of the 

 thermo-element at the lower temperatures. For observations 

 in this region it would be better to substitute a bath for which 

 the arrangement of the apparatus and the heating arrange- 

 ments could be readily adapted. It would indeed be perfectly 

 possible to carry on these observations in a niter bath to above 

 600° with a perfectly uniform distribution of temperature 

 throughout the length of the bar by properly insulating the 

 oven coil. 



1. Platinum. 



The platinum bar, as well as the two immediately following, 

 were prepared from chemically pure material and placed at our 

 disposal through the courtesy of the tirm of Heraus in Hanau, 



