49(3 



NATURE 



[March 23, 1899 



determination of the highest temperatures which can be observed 

 and measured in the laboratory with material thermometers. 

 There are other diflkulties which are peculiar to the determin- 

 ation of the law of radiation, but we are at present concerned 

 primarily wilh those relating to the measurement of temperature. 



There are two comparatively independent lines along which 

 research may proceed with advantage at the present time, (i) 

 The direct comparison of different arbitrary methods ; (2) the 

 extension of the range of the gas-thermometer. 



In order to secure consistency of statement and the reduction 

 of the results of different observers to a common standard, it is 

 in the first place desirable that the various methods available at 

 the present time for the measurement of high temperatures in the 

 laboratory should be directly compared iiilcr se, through the 

 greatest possible range. It is the custom at present for different 

 observers to reduce their results iiidiratly to the scale of the 

 gas-thermometer by reference to certain assumed values of the 

 boiling and freezing points of various substances. They gener- 

 ally assume different values for these fixed points, and adopt 

 different methods of calibration, which are undoubtedly respon- 

 sible for many of the discrepancies at present existing. 



To lake an illustration from the experiments already quoted, 

 the remarkable discrepancy between the experiments of 

 Bottomley, Paschen and Petavel, on the one hand, and those 

 of Wilson and Gray and Schleiermacher on the other, in the 

 determination of the intensity of radiation from polished 

 platinum, may be traced primarily to differences in the methods 

 of measurement adopted. Bottomley and Petavel measured the 

 electrical resistance of the radiating wire itself, and deduced the 

 temperature by the usual formula for the platinum scale. 

 Paschen calibrated his thermo-couple by reference to numerous 

 fusing and boiling points. Wilson and Gray adopted the meldo- 

 meter method based on the expansion of platinum, which they 

 found to be uniform. The vacuum in Schleiermacher's experi- 

 ments could not be measured, and was probably vitiated by gas 

 evolved from the heated platinum. 



" Platinum " Methods of Pyroinetry. 

 These and similar discrepancies might be in a great measure 

 removed, 30 far as they depend on the measurement of temper- 

 ature, by the direct comparison of the various methods of 

 measurement. The "platinum" methods are among the most 

 important and the most easily comparable by direct experiment. 

 These methods are founded on the characteristic stability and 

 infusibilily of the metals of the platinum group, properties which 

 are accompanied by an even more remarkable degree of con- 

 stancy in their less obvious electrical attributes. The two older 

 methods, based on (I) the expansion and (2) the specific heat of 

 platinum, are of comparatively limited application, but have 

 given very good results in the able hands of Joly and Violle. 

 The more modern electrical methods have the advantage of 

 much wider applicability and convenience. They are of two 

 distinct kinds : (3) the thermo-electric method, represented by the 

 I't-Pd. thermo-couple of Becquerel, the I'tlr. thermo-couple of 

 Barus, and the I't-Kh. thermocouple of Le Chatelier, and (4) 

 the platinum resistance pyrometer of Siemens. The third 

 method has been naturalised in this country, and brought to 

 great perfection by the work of Sir William Roberts- Austen. 

 The fourth method was that adopted by Bottomley, Schleier- 

 macher, and Petavel in the experiments above mentioned, and 

 has been applied with great success liy Heycock and Neville at 

 high temperatures, and by Dewar and I-"lcming at the other 

 extremity of the scale. 



Method 0/ Indireet Comparison. 

 The usual or indirect comparison of the foregoing methods 

 by means of the fusing points of various metals is illustrated in 

 the annexed table, which contains several of the most recent 

 results. The numbers given in brackets are now published for 

 the first time, and should be regarded as preliminary. 



Table \\.— Fusing Points hy '^Platinum" Methods. 

 .Method. Observers. Silver Gold. Copper. P.illa- Plali- 



dium num. 

 (945) (io«0 (1085) (1640) (1980*) 



(i) Expansion. 

 (3) Spec. heal. 

 (3) Thermo-couple: 



(C&E) 

 Violle (1879) 

 Bcciiuerel (1863) 960" 

 Barus (1892) 985* 



,, (1894) 906" 



Holhorn & Wieu 



(1895) 968' 



H. & N. (1895) 961- 



NO. 1534. VOL. 59] 



The results above given for the expansion method (l) were 

 obtained by assuming the expansion to be uniform, and taking 

 the F.P. of gold as 1061°. The results of Violle by the 

 specific heat method (2) were deduced by assuming a linear 

 formula for the specific heat of platinum. The discrepancies; 

 of the various results obtained by the thermo-electric method 

 (3) are partly due to errors of observation, and partly to 

 extrapolation, i.e. to differences in the formulae of reduction. 

 The high value founil by Becquerel for the F. P. of copper as 

 compared with gold and silver is probably to be explained by the 

 use of a much thicker wire in the case of copper. The very 

 accurate and con.sistent experiments of lleycock and Neville 

 leave little doubt that the F.P. of pure copper is at least 20° 

 above that of gold. The much smaller difference of 4° to 5°, 

 given by Barus, may possibly be explained by contamination 

 with oxygen or other impurity. In the case of silver and gold, 

 Messrs. Ilolborn arid Wien adopted the Becquerel method of 

 observing the fusion of fine wires. In the ca.se of copper, they 

 adopted the much more accurate method of observing the 

 freezing point of a large mass of metal in a crucible, which had 

 been employed by the writer in 1892, and w,as u.sed by Heycock 

 and Neville throughout their researches. The Becquerel method 

 is very liable to give results which are too high. 



The determination of the higher fusing points of palladiurr> 

 and platinum is necessarily attended with greater uncertainty 

 because it involves extrapolation, and is therefore more de- 

 pendent on the particular formula of reduction assumed, in 

 addition to the experimental difficulties of the higher temper- 

 atures. Considering all the obstacles to be encountered, it would 

 be unreasonable to expect such different methods to give any 

 closer agreement at these points. 



Advantages of Direct Comparison. 



Whatever the origin of these discrepancies, there can be no 

 question that they greatly retard the progress of research and 

 discovery at high temperatures. With the object of helping to 

 remove these obstacles, the writer has recently been engaged, in 

 conjunction with Mr. Eumorfopoulos, in a direct comparison ot 

 methods (i), (3) and (4), which are the simplest and most 

 generally applicable. The advantages of the direct method of 

 comparison are very great, (i) The comparison may be ex- 

 tended continuously throughout the scale, and is not confined to 

 a few arbitrarily selected points. (2) It is easy to apply the 

 electric method of heating, which is of all methods the most 

 easily regulated. {3) It is easy to arrange the experiments in 

 such a way that there can be no question of difference of tem- 

 perature between the thermometers under comparison, which is 

 the most insidious source of error in high temperature measure- 

 ment. 



Comparison of the Expansion and Kesistancc Scales. 



In the comparison of the scale of the expansion of platinum 

 (l), with that of the platinum resistance thermometer (4), it is 

 simply necessary to observe simultaneously the expansion and 

 the electric resistance of a platinum .strip, tube or wire main- 

 tained at a steady temperature by means of an electric current. 

 The expansion may be measured, as in the meldomcter of Joly, 

 by means of a micrometer screw ; but for lecture purposes it is 

 preferable t<i adopt the method of the optical lever employed by 

 Laplace in his experiments on expansion a century ago. By 

 employing a direct reading ohmmeler to indicate the changes of 

 electrical resistance, it is thus possible to exhibit the difference 

 between the two methods by the simultaneous advance of two 

 spots of light on a single scale. If the two instruments are 

 adjusted to read correctly at 0° and 1000 C , the resistance 

 thermometer will be in advance at temperatures below iooo°, 

 but will lag behind at higher temperatures, because the rate ol 

 expansion increa,sesas the temperature rises, whereas the rale of 

 change of resistance diminishes. As the result of ihese experi- 

 ments, it appears that the two scales (1) and (4) differ from that 

 of the gas-thermometer to a nearly equal extent, but in opposite 

 directions. 



The resistance of ]ilatinum at its melting point is more than 

 six limes as great as at o" C, whereas the whole expansion 

 amounts to only one-fiftieth part of the length. The electrical 

 method is for this reason by lar the most accurate and .sensitive. 

 It also possesses in a very striking degree the merit of pliability 

 and adaptability to the needs of each (xarticular problem. For 

 this reason the scale of the platinum resistance thermometer has 



ll 



