HOLMAN. — PYROMETRY. 235 



matiou holding withiu the limits of variable error of the instrument, 

 and probably well withiu the limits of uncertainty of the assumed 

 values of the melting points employed in calibration. 



For clearness, the usual method of calibration will be first stated, 

 and tlien the proposed method. 



Usual Method. 



The cold junction is kept at a temperature usually about that of the 

 room, and measured by a mercurial thermometer. The spot of light 

 is made to read zero with the circuit open. The hot junction is 

 then exposed successively to several known high temperatures, and 

 the scale readings are taken. These temperatures are, ordinarily, 

 the boiling points of naphthaliu (C10H3), and of sulphur, the melt- 

 ing points of aluminum, gold, and platinum, — selected as being 

 the most satisfactory in manipulation, and reliable in value. The 

 values assumed for these points differ, following the judgment of 

 the experimenter, Violle's figures 1775° C. for platinum and 1035° 

 or 1045° C. for gold, and Le Chatelier's 625° or 635° for aluminum, 

 being most commonly accepted. From measurements and considera- 

 tions elsewhere * discussed, the author recommends the following as 

 provisional numbers in preference to the foregoing values. 



Crafts,t Holman and Gleason.J 

 Callendar and Griffiths. § 

 Holman, Lawrence, and Barr.* 



A plot is then made with deflections (scale readings) as abscissas 

 (horizontal) and temperature differences (t — c) between the hot and 

 cold junctions as ordinates (vertical). The points thus obtained lie 

 along a curve which in the upper part approaches somewhat closely to 

 a straight line. But it is nowhere exactly straight, and the error of 

 assuming it so may amount to 10° or 15° or even more. On the 

 other hand the points lie so far apart, if plotted upon a scale com- 



* Holman, Lawrence, and Barr, ante, p. 219. 



t Crafts, Bull, de la Soc. Chim., XXXIX. 196, 277 (1883). 



t Holman and Gleason, Proc. Amer. Acad., XXI. 237 (1888). 



§ Calendar and Griffiths, Phil. Transact., CLXXXII. 119, 157 (1891). 



