762 TRUEBLOOD. 



In spite of the fact that the calibration is not, in general, used 

 directly in calculating the temperature difference from the resistance 

 measurements, it is distinctly worth while, in the writer's opinion, for 

 the following reasons: 



First, it gives a valuable check on elimination by interchange; 

 second, it affords a very useful indication of the precision of the calori- 

 metric work; for the value of the normal resistance as determined by 

 interchanging thermometers in a Joule-Thomson experiment includes 

 the accidental errors of the experiment, so that the divergence of this 

 value from the calibrated value is an index of the magnitude of these 

 errors; third, if through some accident or oversight in setting up the 

 apparatus one of a pair of runs has to be rejected, the other may still 

 be used; for example, of a certain set of nine runs, the first four (with 

 thermometer No. 7 on the high-pressure side of the plug) had to be 

 thrown out because of a fault in one of the leads from the plug to the 

 bridge which was not discovered until after these runs had been com- 

 pleted. The value of the remaining runs, with No. 6 on the high side, 

 was not affected, because a calibration made immediately after 

 afforded the means of eliminating the normal resistance. 



In calibrating the differential thermometer, the chief difficulty lies 

 in getting the two thermometers simultaneously at the same tempera- 

 ture when this temperature exceeds that of the room by say 150° or 

 more. i\.t least two such points are necessary. It is probable that 

 the most satisfactory method of obtaining them would be to employ 

 a double oil bath, provided a sufficiently sensitive thermostat for the 

 higher temperatures could be devised. The method employed by the 

 writer made use of a large, specially-constructed hypsometer of the 

 well-known Regnault type, in which the two thermometers could be 

 exposed to the vapors of substances which boil at convenient tempera- 

 tures. The hypsometer was heavily lagged with asbestos except for 

 the reservoir containing the liciuid and for an air condenser by means 

 of which most of the vapor was liquefied and returned to this reservoir. 

 The substances used in this apparatus were water, cumol (about 

 165° C), naphthahn (218° C), diphenylamin (310° C). Water, of 

 course, gives no trouble whatever. Cumol, in spite of its compara- 

 tively low boiling point, has always been unsatisfactory, chiefly 

 because its boiling point rises continuously as vaporization progresses, 

 even if the liquid has been previously distilled. Naphthalin is very 

 satisfactory. Diphenylamin has been little used, as its boiling point 

 lies beyond the temperature range required in the work discussed 

 in this paper. It has the same disadvantage as cumol as regards 



