536 



NATURE 



[April 7, 1892 



gested in 1826 by Becquerel, and adopted by Pouillet in 

 1836,^ who advocates the use of iron in conjunction with 

 platinum ; but of all the varied combinations of metals 

 and alloys which have been tried from time to time, that 

 proposedby H. Le Chatelier possesses many advantages, 

 on which I have elsewhere dwelt." It consists of a plati- 

 num wire twisted at its end with a wire of platinum 

 alloyed with 10 per cent, of rhodium. Such a couple 

 may be used for some time without change of zero, 

 and if the junction becomes injured it may be cut off, 

 and the severed ends of the wires may be twisted to- 

 gether again. I am satisfied that it can afford comparative 

 results which are accurate to 1° at temperatures of over 

 1000°. The diagrams given later (Figs. 4, 5, and 6) show 

 the disposition of the apparatus. The spot of light 

 indicating the deflections of the galvanometer needle 

 is caused, for the illustrations of this lecture, to fall 

 onto a graduated scale 45 feet long on the wall of the 

 theatre. The thermo-j unction has been calibrated with 

 the aid of certain known temperatures, and the long scale 

 is inscribed after the manner of the old thermometer 

 scales, with certain fixed points, which are, of course, far 

 higher than those it was possible to indicate by the ex- 

 pansion of mercury in a glass tube. [These fixed points 

 were: "water boils" (100°), "lead melts" (326°), "zinc 

 boils " (940°), "gold melts" (1045°), " palladium melts " 

 (1500°), " platinum melts " (1775''). On heating the 

 thermo-junction to bright redness in a Bunsen flame, the 

 spot of light moved rapidly to the point marked " zinc 

 boils,"] For laboratory experiments the scale is a short 

 transparent one, rigidly fixed in relation to the galvano- 

 meter. 



In leading up to the experiments which follow, in the 

 course of which metals will be exposed to high tempera- 

 tures, I would remind you that if an ordinary thermometer 

 be plunged into water which is gradually losing its heat 

 to a cold environment, the mercury will fall until the water 

 begins to freeze, but directly this happens the mercury 

 remains stationary until all the water is frozen ; so that if 

 the rate of fall be measured with a chronograph, there 

 will be a steady fall to the freezing-point of water, then a 

 long arrest, followed by a renewed fall. If these readings 

 be plotted, a well-known time-temperature curve will be 

 obtained. Exactly the same effect is produced when a 

 fluid metal " freezes," and before proceeding further it 

 may be well to determine experimentally the freezing- 

 point of gold. Beneath this little mass of pure gold, 

 A (Fig. 2), a thermo-junction, B, is protected by a very thin 



layer of clay from the metal. The oxyhydrogen flame is 

 made to play on the gold, there is a rapid movement of 

 the spot of light over almost 25 feet of the scale, there is 

 a diminution in the rate of rise near the point marked 

 1045°, the melting-point of gold; and then, when the 

 metal becomes fluid, the temperature rapidly rises as 

 more heat is given to the little mass. The source of heat is 

 now removed, the temperature falls, there is an arrest 

 just at 1045° C, the freezing-point of gold, and then 

 the spot of light resumes its course as the gold cools down 

 to the temperature of the room. The melting-point and 

 freezing-point of palladium, 1500° C, were then shown in 

 the same way. It should be observed, however, that when 

 a small fragment of palladium is fused in the naked flame 

 of the oxyhydrogen blow-pipe, hydrogen appears to be ab- 



' Comptes rendus, vol. iii. p. 782, 1836 



° British Association Lecture, Nature, vol. xli., 1889, pp. 11-32; 

 Report Inst. Mech. Eng., Oct. 1891, p. 543. 



NO, I 171, VOL. 45] 



sorbed by the metal ; and this absorption of gas lowers the 

 freezing-point materially, and makes it far less steady 

 than when a fresh piece of metal, cut from a large mass , is 

 fused for the first time. 



When the spot of light is allowed to fall on a sensitized 

 plate in a suitable camera,^ the time-temperature curve 

 traced on a moving plate will be of the form shown in 

 Fig. 3- 



It may be useful tp show the method by which these auto- 

 graphic curves are obtained : the following diagram, Fig. 4, is 

 therefore added. 



The arrangement consists in inclosinj? a galvanometer of the 

 Deprez and d'Arsonval type in a large camera ; atixed mirror, F, 

 being placed below the movable mirror, M, of the galvanometer, 

 so that the light, from the lime cylinder L, reflected in the 

 mirror H, passes to both mirrors, F and M, and is reflected in 

 the direction of a fine horizontal slit, A b, behind which a sen- 

 sitized photographic plate, c, is drawn vertically past the slit, 

 by means of gearing, D, driven by clockwork. The ray from 

 the fixed mirror is interrupted periodically by the vane e, and 

 a beaded datum line is given, which enables any irregularity in 

 the advance of the plate to be detected. 



The amount of divergence from its datum line of the spot of 

 light reflected by the movable mirror at any given moment 

 bears a relation (which can readily be found by calibration) to 

 the temperature to which the thermo-junction X is heated, and 

 the variations of temperature are recorded by a curve which is 

 the resultant of the upward movement of the plate and the hori- 

 zontal movement of the spot of light. A crucible, c, which may 

 be filled with molten metal, is provided with a tubulure, T, for 

 the insertion of the thermo-junction. The crucible is suspended 

 by wires in a double jacket of tin plate, a b. 



It will have been evident that the thermo-junction of 

 platinum and platinum-rhodium could not be used for 

 measuring temperatures higher than the melting-point of 

 the platinum of which it is made. Metals with higher 

 fusion-points than platinum are, however, available ; thus 

 iridium will only just melt in the flame produced by the 

 combustion of pure and dry hydrogen and oxygen. By 

 the kindness of Mr. Edward Matthey, a thin rod of 

 iridium has been prepared with much labour, and it can 

 be used as a thermo-junction with a similar rod of iridium 

 alloyed with 10 per cent, of platinum. The junction may 

 be readily melted in the electric arc, and by this means a 

 temperature may be registered which careful laboratory 

 experiments show to be close to 2000°, and this agrees 

 with the estimate of the melting-point of iridium which 

 Violle^ deduced from calorimetric experiments. [This 

 experiment was shown, a different scale being employed 

 for the screen, as the thermo-electric constants of the 

 iridium, and iridium-platinum couple, are different from 

 those of the platinum and rhodium one previously used.] 



It is interesting to remember that within a year in this 

 Institution temperatures ranging from - 200° to -f 2000° 



I Proc. Roy. Soc, vol. xlix., 1891, p. 347. 

 - Loc. cit. 



