EXPERIMENTAL STUDY OF HEAT LEAKAGE. 763 



rise of the boiling point during vaporization, tliough in a lesser 

 degree. 



It is doubtful, however, whether the hypsoxneter method of calibra- 

 tion would be satisfactory at temperatures as high as that of boiling 

 diphenylamin, even if the boiling point were perfectly steady. It was 

 found early in the work that even when naphthalin was used, the 

 temperature in the region occupied by the two thermometer bulbs was 

 not uniform, but that the apparent value of the normal resistance of 

 the differential thermometer could be changed by an amount of the 

 order of one or two hundredths of a degree on the scale of either 

 thermometer by merely interchanging the positions of the ther- 

 mometers. This effect was not accidental, but could be obtained re- 

 peatedly with great uniformity and definiteness; because of this, it 

 could wdth confidence be regarded as eliminated by taking the mean 

 of a number of measurements equally distributed between direct 

 and reversed positions of the thermometers. To facilitate this 

 method of observation, which was always used, the top of the hypso- 

 meter, from which the thermometers were suspended, was made capa- 

 ble of rotation about a vertical axis, with respect to which the two 

 thermometers were symmetrically arranged; thus the thermometers 

 could be made to exchange positions without withdrawing either from 

 the hypsometer or disconnecting it from the bridge. 



This non-uniformity in temperature of the vapor column in which 

 the two thermometers were immersed was first noticed in a much 

 simpler comparator than that which has been described. This earlier 

 comparator consisted merely of a large glass tube (3 inches in diameter) 

 of uniform bore, closed at one end and lagged on the sides with asbestos. 

 Y\ ith this apparatus, the apparent normal resistance of the differential 

 thermometer could be changed by an amount equivalent to 0°.16 C. 

 by merely interchanging the positions of the thermometers, — which 

 signifies a difference of 0°.08 C. in the temperature of the vapor (naph- 

 thalin) at two points at the same level and about 1.5 inches apart. 

 This difference was not accidental; in an experiment lasting over 

 three hours, the apparent differential resistance in either position of 

 the thermometers remained constant to within less than 0°.01 C. The 

 only evident asymmetry of arrangement which might account for this 

 rather remarkable effect consisted in the fact that the thermometer 

 bulbs were always on a line perpendicular to a brick wall near which 

 the apparatus was set up; but the phenomenon was not in the least 

 altered by the use or omission of radiation shields around the ther- 

 mometer bulbs, and has been noticed, in the Regnault comparator, 

 -even when the above-mentioned asvmmetr\' was avoided. 



