40 



NATURE 



[November 14, 1895 



hundredth of the difference between its resistance at ioo° and 

 o°C. 



Hence, if R be the resistance at any temperature, R, the re- 

 sistance at 100° C, Ru at o° C, and/^ the temperature on the 

 platinum scale, then 



pt = iL^l^o X 100. 

 Ri - R„ 



The investigations of Prof. Callendar' estabUshed the relation 

 between/^ and ^ (where / is the temperature on the air scale) 

 over the range o° to about 6oc° C. for a particular sample of 

 platinum wire. 



This relation is given by the following equation. 



d=i-J,f=sf(-iy-J-\. . . .Eq. (i) 

 (\ioo/ 100 ) 



the value of S for Callendar's wire being I -57. 



If it was at all times possible to obtain platinum wires of 

 exactly the same degree of purity as Callendar's, we could at 

 once establish a standard platinum scale, which could be used 

 for purposes of reference independently of any assumptions as to 

 its relation to the air scale. The impossibility of securing uni- 

 formity in this respect, however, would, at first sight, appear to 

 be an insuperable impediment to the adoption of such a 

 proposal. 



Subsequent experiments by Callendar and myself led, how- 

 ever, to the following conclusion.^ 



That, although the value of 5 varies greatly according to the 

 purity of the sample of platinum, the relation given by the equa- 

 tion {d) holds true, provided the percentage of impurities is 

 small (this condition is ;sufficiently fulfilled by ordinary com- 

 mercial samples). 



This conclusion is an exceedingly important one, for (the t - pt 

 curve in every case being a parabola) it is only necessary to 

 determine the resistance at three different temperatures in order 

 to ascertain the appropriate value of 5, and thus to completely 

 standardise the thermometer. 



Much experimental work had to be accomplished before we 

 could venture to regard the above proposition as established ; 

 but I think that any impartial reader, who cares to study the 

 original papers ^ dealing with this matter, will admit that the 

 evidence is sufficient. 



The three temperatures selected for the purposes of standard- 

 isation were the melting-point of ice, steam at a pressure of 

 760 m.m., and the vapour of sulphur at the same pressure. 



Certain discrepancies between thermometers thus standardised 

 and others standardised by direct comparison with the air ther- 

 mometer, led to a redetermination (by means of an air thermo- 

 meter) of the boiUng-point of sulphur, when we found that 

 Regnaull's value (448° "34)1 was too high, our experiments leading 

 to the conclusion that 444''-53 was the correct value.'* 



Subsequent investigations by different observers have con- 

 firmed the accuracy of the above conclusions, which may now 

 be regarded as experimentally established over the range 0° to 

 600° C. 



There is, however, a large amount of indirect evidence which 

 indicates that formula {d) holds true over a far more extended 

 range. 



For example, the results obtained by Messrs. Heycock and 

 Neville {Chem. Soc. Trans., 1895) are entirely dependent on 

 the validity of the above conclusions. They find the freezing- 

 point of copper as io8o°*5 C, whereas Holborn and Wien, 

 using a platinum rhodium couple standardised by direct com- 



1 Phil. Trans. Roy. Soc. A, 1887. 

 I 'S A summary of these experiments is given in Science Progress, Sept- 

 ember 1894. 



3 Callendar, Phil. Trans. Roy. Soc. A, 1887 ; Griffiths, " Report of 

 Electrical Standards Committee, B.A. 1890; Heycock and Neville, C/l^w. 

 Soc.Journ. 1890 ; Griffiths, Phil. Trans. Roy. Soc. A, 1891 ; Callendar and 

 Griffiths, Phil. Trans. Roy. Soc. A, 1891 ; Callendar, Phil. Mag, July 

 1891 ; Griffiths and Clark, Phil. Mag., December 1892 ; Griffiths, Phil. 

 Trans. Roy. Soc. A, 1893 ; ibid., Proc. Roy. Soc. vol. Iv. 1894 ; ibid.. Science 

 Progress, 1894 ; Thorpe, " Dictionary of Applied Chemistry," article 

 "Thermometry" ; Heycock and Neville, Trans. Chem. Soc. 1895. 



4 In the_ last edition of Watts's "Dictionary of Chemistry," article 

 ^' Sulphur," I find that some doubts aie expressed (by Mr Pattison Muir) as 

 to the validity of this determination, owing to uncertainty as to the purity of 

 our sample of sulphur. 1 sub equently investigated the boiling-point of a 

 .specially pure sample by means of one of the platinum thermometers (ther- 

 mometer E) used during the original comparison of the air and platinum 

 thermometers in sulphur vapour, and I found no evidence of any difference 

 in the boiling-point Of the two samples. We may assume, therefore, that if 

 any impurities were present, they were not of such a nature as to influence 

 the temperature of the vapour. 



KO. 1359, VOL. 53] 



parison with the porcelain air thermometer, find 1082° as the 

 value of the same constant.^ 



As illustrating the identity of the results obtained by the use 

 of thermometers having a very different value of 8, I quote the 

 following numbers from Table XII. of Heycock and Neville's 

 paper : — 



Freezing-point of gold. 



Results of this kind prove that even if the reduction does not 

 express the temperature accurately in the air scale, it at all 

 events gives us a constant scale in which all high temperatures 

 can be expressed, and it is further evident that this constant 

 scale differs but little (even at these high temperatures) from the 

 true air scale. ^ 



Indications are not wanting that the same relations hold true 

 at very low temperatures.^ 



Finally, a very careful comparison of the platinum and air 

 thermometers over the range 0° to 100° C, and also of the 

 platinum thermometer with the nitrogen standard of the Bureau 

 International, establishes the validity of the methods of observa- 

 tion and reduction at ordinary temperatures. 



As regards the constancy of platinum thermometers there 

 should now be little uncertainty. The prevailing doubt (amongst 

 those who have not used them) may be traced to the adverse 

 report of a British Association Committee in 1874, on another 

 form of the instrument, and I would refer those who may be in- 

 fluenced by that report to a letter by Prof. Carey Foster, F.R.S., 

 in Nature, August 23, 1894. 



An inspection of the voluminous tables given in Heycock and 

 Neville's paper {supra) will show, however, that when the 

 thermometers are repeatedly exposed to temperatures above 900° 

 or so, a slight permanent increase in FI (the Fundamental 

 Interval = Rj - R^) is observable. It is probable that this 

 change is due to a permanent thickening of the mica plates by 

 which the wire is supported, and thus, on cooling, the wire is 

 slightly strained. The change is small, and can always be 

 traced by repeating the determinations of Rj and R,,, and does 

 not appear to appreciably affect the values of 5. 



To show the order of magnitude of the change, I give the 

 following illustration, compiled from Table VHI. of Heycock and 

 Neville's paper. 



History of Pyrometer 1 3. 



On August 3, 1894, the fundamental interval was 100-64. 

 During the next few months this pyrometer was used for the 

 determination of the freezing-points of the following substances : — 



Number of 

 determinations. 

 , 10 



Substance. 

 Silver ... 



Aluminium ... 



Potassium sulphate 



Sodium sulphate 



Sodium carbonate ... 



Magnesium ... 



Antimony 



Tin 



B.P. of sulphur 



Also the pyrometer had been raised to a bright red heat in a 

 muffle furnace some scores of times, and the exterior porcelain 



1 The following example illustrates the importance of the alteration in the 

 boilii g-point of sulphur. In Table VI. of Heycock and Neville's paper 

 (supra) are given the details of an observation on the freezing-point of Cu 

 determmed by pyrometer No. 8. They are as follows: //^ = 421-29, 

 5 = I-5I7, rf = 159-3, < = 1080-7. If we assumed the validity of Regnault's 

 boiling-point of sulphur (448'-34), the above value of pt^ would change the 

 value of 5 to 1-729 ; this .<would give d ■= i8i'6 ; hence t = iio3°-o. In this 

 case the discrepancy between the results of Holborn and Wien, and Heycock 

 and Neville would be very marked — a difference of 2i°-o as against the 

 present difference of i°-3. 



^ Messrs. Heycock and Neville, in Table XVI. (Chem. Soc. Trans., 1895, 

 P- i95)> give Violle's value for the freezing-point of gold as 1035* C., and the 

 discrepancy between this number and that found by them (io6i°'7) is con- 

 sidi rable. A redetermination, however, by VioUe in 1892 (Comptes rendus, 

 92, p. 866) raised his number to 1045". Some recent experiments by Le 

 CkzxtW^x (Comptes rendus, Kv.%Vi%\.ii, 1895) lead that observer to the con- 

 clusion that Violle's later value should be further raised by about 15' (or at 

 all events by a number " not exceeding 20° "), i.e. to about 1060° C, a very 

 close approximation to the io6i''-7 found by Heycock and Neville in De- 

 cember- 1804. ■ ^ 



3 Griffiths and Clark, Phil. Mag., December 1892. 



