July 28, 1898] 



NATURE 



307 



uncommon. Holborn and Wien have compared the Le Chatelier 

 couple with the air thermometer at the Reichsanstalt up to 

 1450° C. In Wied. Ann. vol. Ivi. p. 364, they say that the 

 readings of different thermo-elements may be expected to agree 

 within + 5° at 1000° C, while different observations with the 

 same instrument agree better than this. They also say that pro- 

 perly prepared thermo-elements have remained unaltered for 

 years, whether unemployed or subjected to frequent temperature 

 changes, always provided they are not exposed to certain sources 

 of contamination. 



In their more recent low temperature work, already referred to, 

 Holborn and Wien made further use of thermo-couples, but the 

 metals chiefly employed were apparently iron and constantan. 



In translating measurements of E. M.F., E, into air tempera- 

 ture, /, Holborn and Wien employ an ordinary algebraic 

 formula 



Here, as usual, a, h, c denote constants, which may be 

 determined by observations at three fixed temperatures. 



The question of the most suitable type of formula to be 

 applied to thermo-electric data is discussed very fully by Prof. 

 S. W. Holman in the Phil. Mag. for June 1896. The three 

 types he advances as most deserving of notice are 



E= (T-To)|a-f<5(T-fTo)K 



E = w(t" - To"), 

 E = 7nt". 

 In all E represents E. M. F, m and « constants to be determined 

 by reference to fixed points, t and t„ temperatures of hot and 

 cold junctions measured from alsolute zero, t ordinary Centi- 

 grade temperature of hot junction (the cold junction being 

 supposed in ice). The first or algebraic type, in a special 

 form, is usually associated in this country with the name 

 of Prof. Tait. The second type is called by Prof. Holman 

 the exponential, and the third the logarithmic (as lending 

 itself readily to logarithmic calculation). The three types 

 are applied by Prof. Holman to what he regards as the most 

 notable series of recent observations. In addition to the 

 high temperature observations of Holborn and Wien, already re- 

 ferred to, he considers a number of comparisons of platinum 

 and platinum-iridium couples with constant-pressure air 

 thermometers made by Barus in America, and less extensive 

 series by other observers in France and Germany. Of the three 

 types of formulae, the algebraic proved the least suitable for 

 application to a wide temperature range. 



In a later paper in the Phil. Mag. (vol. xlii., 1896, p. 37), 

 Prof. Holman, with Messrs. Lawrence and Barr, apply the three 

 above specified formula to observations of their own with 

 couples of platinum and platinum-rhodium at the melting points 

 of aluminium, silver, gold, copper and platinum. The con- 

 stants in the formulae were determined from the same three 

 fixed points, viz. the ice point, the boiling point of sulphur at 

 standard pressure, and the melting point of gold. For the 

 second point Callendar and Griffiths' value 444° 53 C. was 

 accepted, and for the third point Holborn and Wien's mean 

 result 1072° C. 



Observations were taken at the boiling points of water and 

 naphthalin, as well as at the melting points of the several metals. 

 The temperatures calculated from the three formula agree 

 closely for the copper point— which lies near the gold point— 

 and fairly closely for the silver point. For the naphthalin point 

 the calculated values differed from the true air scale temper- 

 atures by from 4° to 12°, and the errors in the calculated values 

 for the steam point were fully as large. If the authors are cor- 

 rect in their opinion, " so far ... as constant or variable instru- 

 mental errors are concerned, it is believed that no error beyond 

 o'-S to i°C. exists in the results," we must conclude that further 

 inquiry into thermo-electric methods is highly desirable.. 



Thermo-electric methods lend themselves fairly readily to the 

 study of gradual temperature changes, the spot of light reflected 

 by the mirror of the galvanometer measuring the E.M.F. being 

 thrown either on a screen or on a photographic plate actuated 

 by clockwork. Prof. Roberts- Austen {Koy. Sac. Proc. , vol. xlix. , 

 1891, p. 347) has inaugurated investigations by this method ' into 

 1 Since this article was written there has appeared in the Phil. Mag. for 

 July 1898, an interesting paper by Mr. A. Stansfield, describing improve- 

 mentsin Prof. Rob)erts-Austen's recording pyrometer, and discussing thermo- 

 eiectric results. Mr. Stansfield obtains E = ax -(- ^ log t -|- cr, for the relation 

 between E.M.F. and temperature, measured from the absolute zero. His 

 melting points agree well, on the whole, with the determinations of Hevcock 

 and Neville. ' 



NO. 1500, VOL. 58] 



the phenomena accompanying solidification of metals. Prof. 

 Callendar, on the other hand (Trans. Roy. Soc. of Canada, 

 1897. P- .34). has recently applied the platinum thermometer in 

 the continuous registration of the changes of earth, water and 

 air temperatures ; and, unless my memory deceives me, bare 

 wire resistances have been used previously for the last-mentioned 

 purpose 



Thermo-electric and electrical resistance methods are also 

 specially applicable to the measurement of minute temperature 

 differences. As examples of this application, we may take the 

 bolometer of Langley and the radio- micrometer of Boys {Phil. 

 Trans., 1889, p. 159). The former instrument in its most 

 approved shape— as in use at the Astrophysical Observatory at 

 Washington — has been described very recently by Prof. Langley 

 himself (T",*^ Smithsonian Institution, 1846-96, pp. 419-442, 

 Washington, 1897). Its main use at Washington is in mapping 

 out the infra-red spectrum and determining the intensity of the 

 solar radiation at different wave-lengths. It is simply an 

 electrical resistance thermometer, the resistance being that of a 

 " metallic tape usually about ^-inch long, but narrower and far 

 thinner than a human hair . . . this, at present, may be arranged 

 to record changes of temperature as small as one-millionth of a 

 degree." Prof. Langley has devised the means of producing a 

 uniform relative motion of the bolometer and solar spectrum, 

 and obtaining an exact photographic record of the varying 

 heating effect ; and in this way he has apparently enormously 

 reduced the labour of mapping the spectrum. 



The radio-micrometer, on the other hand, consists essentially 

 of a thermo-electric circuit— the principally effective junction 

 being that of bismuth and antimony — which is suspended by a 

 quartz fibre and is capable of rotation in an intense magnetic 

 field. It is especially suitable for measuring the radiation from 

 a distant or feeble source of heat, the radiation being received 

 on a metal surface in immediate connection with the bismuth 

 antimony junction. The delicacy of the instrument varies 

 greatly with the shape of the circuit and the fineness of the 

 quartz fibre suspension. According to Mr. Boys, it would be 

 possible with the most approved pattern to detect with certainty 

 " a temperature difference of less than one two-millionth of a 

 degree Centigrade. " Whether this has been realised in practice, 

 I do not know. 



A differential radio-micrometer was employed some years ago 

 by Mr. W. E. Wilson and Mr. P. L. Gray (Phil. Trans. A, 1894, 

 p. 361) in experiments in which solar radiation was balanced 

 against the radiation from a strip of platinum heated to various 

 known temperatures. The object of the research was to deter- 

 mine the mean effective temperature of the sun. The method 

 is one which would seem capable of numerous useful applica- 

 tions. 



As already stated, thernio-electric and electrical resistance 

 methods are by no means the only ones, in addition to gas 

 thermometry, for which high accuracy is claimed in high tempera- 

 ture measurements. There is, however, only one other method 

 to which I shall refer here, viz. the expansion of solids. This is, 

 of course, a very old method, and is generally employed only for 

 commercial purposes for which high accuracy is not aimed at. In 

 1891, however. Dr. Joly applied the principle in a new special 

 form of instrument, the meldometer, for which high accuracy is 

 apparently claimed as a means of determining melting points. 

 The essential part of the instrument is a thin strip of platinum, 

 kept stretched by a spring and heated as required by an electric 

 current. A minute quantity of the substance under examination 

 is placed on the strip, whose temperature is raised until the 

 substance melts. There is delicate means of measuring the 

 lengthening of the strip, and the corresponding temperature 

 is deduced with the aid of a preliminary calibration, based 

 on observations at two or three known melting points. The 

 meldometer has been used by Prof. Ramsay and Mr. 

 Eumorfopoulos (Phil. Mag., vol. xli., 1896, p. 360) in 

 determining the melting points of a large number of salts, 

 and these observers seem to think highly of it. A meld- 

 ometer strip was also the source of heat, whose radiation was 

 compared with that of the sun in the experiments of Messrs. 

 Wilson and Gray already referred to. Having had no personal 

 experience of the meldometer, I can only say that I should 

 hardly expect it to rival in accuracy either the thermo-couple or 

 the electrical resistance thermometer ; but the smallness of the 

 quantity of material required, is unquestionably a recommend- 

 ation to its use in determining the melting points of rare or 

 precious substances. C. Chree. 



