760 TRUEBLOOD. 



change in the lead resistance. ' Solid leads are much to be preferred, 

 wherever it is possil^le to use them. 



g. Calibratio7i of differential thermometers. 



The measurement of a change of temperature in continuous flow 

 work involves the use of two thermometers, which must be as nearly 

 alike as it is possible to get them, if the benefits of a direct measure- 

 ment of the difference of their resistances (as against the separate 

 measurement of the resistance of each) are to be completely retained. 

 Since it is not possible to make two thermometers exactly alike, there 

 wall always exist a difference in their resistances when they are at the 

 same temperature. It will be convenient to speak of the combina- 

 tion of two thermometers used in this way as a differential thermometer, 

 and to call the difference in their resistances at a common temperature 

 the normal resistance of the differential thermometer. A particular 

 differential thermometer will be referred to by means of a number 

 made up of the numbers of the component thermometers, the number 

 of the component thermometer having the higher resistance coming 

 first: e. g., 67 is the differential thermometer composefl of the indi- 

 vidual thermometers 6 and 7. The normal resistance of this ther- 

 mometer will be denoted by R&y, etc. 



In order to get the difference of temperature corresponding to a 

 measured difference of resistance, the normal resistance of the differ- 

 ential thermometer must be eliminated from the latter. This may 

 be done by calibrating the differential thermometer, determining its 

 normal resistance as a function of the temperature, and subtracting 

 from a measured difference of resistance the normal resistance of the 

 thermometer as calculated for the temperature in question; or, the 

 experiment in which the measured difference of resistance is obtained 

 may be repeated with the component thermometers interchanged. 

 In most of the work described in this paper, both methods have been 

 used; but, except where for some reason the interchange method was 

 not used, the actual elimination has been eflfected by this method, 

 the calibration serving merelv as a check. There is little evidence 

 to indicate that the value of the normal resistance as given by one of 

 these methods is different from that given by the other; thus, the 

 mean of six values of {R&^ by calibration — i?67 by interchange) at 

 165° C. is 0.0000.3 ohm or about 0°.00l5 C. with an average de\'iation 

 of 0.00025 ohm or about 0°.012 C. The time interval involved in 

 this work is about a vear and a half. Of the six calibrations of the 



