harper: thermometric lag 207 



The values obtained in vigorously stirred water ranged from 15 

 to 30 seconds, while for a mercury thermometer under the same 

 conditions the usual value is between 2 and 6 seconds, except for 

 the very largest bulbs. The Dickinson-Mueller type of resist- 

 ance thermometer, an improved form of the instrument described 

 in 1907, was found to be very much faster than a mercury ther- 

 mometer. The only result of some attempts to measure directly 

 the lag when immersed in well stirred water, was to ascertain 

 definitely that it was but a small fraction of the lag of the gal- 

 vanometer employed, which was about 1^ seconds. 



The J aeger-Steinwehr method of computing the lag of a resistance 

 thermometer.- — Only a very meagre discussion of this method can 

 be attempted here. It is probably the best, and perhaps in many 

 cases the only possible way of determining the lag of such instru- 

 ments as are constructed in a form permitting of applying it at 

 all. The principle involved is the equating, in the equilibrium 

 state, of the heat supplied to the thermometer by a small current, 

 to the heat transferred to the bath because of the elevation of 

 the temperature of the coil slightly above that of the bath. The 

 lag and the heat capacity of the system whose temperature is 

 so elevated, enter the equation together with other quantities 

 which are readily measured experimentally. The possibility of 

 determining the lag depends therefore upon whether or not a satis- 

 factor}' estimate of the heat capacity factor can be made. In 

 this respect the paper suggesting the method is decidedly mis- 

 leading, stating for the proper value of the heat capacity that of 

 the wire alone, the formula being reduced to terms of the dimen- 

 sions, specific heat, specific resistance, etc., of the wire. Some 

 portion of the supports of the coil, and the covering if there be 

 one, must always be heated along with the coil, and in most cases 

 the fraction to be included has several times the heat capacity of 

 the wire in the coil. A thermometer of the Dickinson-Mueller 

 type has, in the region of the platinum coil, an amount of mica 

 whose heat capacity is about 0.15 calorie per degree. A prob- 

 able distribution of the temperature gradient from the coil to the 

 bath, places the portion to be included with the coil as about 

 one-third, i.e., 0.05 calorie per ° C. The heat capacity of the coil 



