GEOPHYSICAL LABORATORY. 99 



both with temperature and with composition, is, within the limits of error, 



continuous over the entire range studied. The variation of thermo-electric 



de 

 power— — with the temperature and composition of alloys of platinum and 



CI V 



rhodium indicates the formation of solid solutions, but no compounds, from 

 o to 55 atomic per cent of rhodium. 



(14) The thermoelement as a precision thermometer. Walter P. White. Phys. Rev., 

 31, 136. 1910. 



The thermo-elements used in some of the calorimetric work of the labora- 

 tory have proved surprisingly convenient and accurate. The value of the 

 thermo-element for work of this kind has been but little appreciated; hence 

 a discussion is given of its construction, of some methods for increasing the 

 precision attainable, and of its value for different purposes. The following 

 summarizes the principal conclusions reached : 



( 1 ) The E. M. F. of a thermo-element, whether homogeneous or not, can be 



expressed as equal to I Hdd (H=thermo-electric power, 0=temperature). 



It follows that the effect of each portion of a thermo-element is proportional 

 to the magnitude of the temperature gradient in which it lies ; hence the vital 

 parts of a thermo-element are the parts along which the temperature varies ; 

 they mainly determine the calibration and the constancy; with them alone 

 need the maker or user of the thermo-element be concerned. In changing 

 them (for example, by varying the depth of immersion in a furnace) the 

 thermo-element itself is changed, if not homogeneous. Such changes in 

 regions of uniform temperature distribution (for example, at the junction) 

 have little effect on the temperature measurement. 



(2) Commercial constantan wire, good enough to make thermo-elements 

 which are accurate to about 0.05 per cent, is easily obtained and costs but 

 little trouble in the selection, but can not be relied upon without any test for 

 possible inhomogeneity. Compensated thermo-elements of higher accuracy, 

 reading consistently to 0.0001 , for use at ordinary and low temperatures, 

 can be made up in a day or two each. Improvement in the accessory appa- 

 ratus has shown that their constancy is about ten times that shown three 

 years ago. 



(3) The accuracy of 0.0001 reached by the thermo-element can also be 

 attained in its auxiliary apparatus (potentiometer, etc.). This accuracy, both 

 absolute and relative, and the sensitiveness are therefore quite comparable 

 with those of the best resistance thermometers now in use. 



(4) For the measurement of single temperatures, the resistance thermom- 

 eter is often more convenient. On account of the comprehensiveness of the 

 potentiometer and its indifference to external contact resistances, the thermo- 

 element is usually preferable where several different measurements are to 

 be made at once. The thermo-element is also usually superior for differen- 

 tial measurements, and its advantages in this respect can be made available 

 for calorimetric work by the substitution of a suitable comparison body for 

 the usual ice bath. 



(5) A very convenient wire-tester, a simple bath for testing and compar- 

 ing thermo-elements, and a convenient form of tabulation for calibration 

 curves are also described. 



