132 PYROMETER of glass or silver of 4 cc. capacity connected with a vertical stem of thermometer tubing of 0'2 mm. diameter. This stem terminates in an open vessel o? mercury, and thus the pressure of the gas can be measured. Berthelot's instrument is graduated by reference to four fixed points, namely, the freezing-point and boiling-point of water, and the boiling-points of mercury and sulphur, lu order that the mercury index may move easily in the tube, extreme care must be taken in drying the tube, and only perfectly pure mercury cau be used. 4. The results obtained by any of the air-pyrometric methods just described maybe employed to express directly the temperature of the pyrometer in numbers agreeing closely with the thennodynamic scale. The other instru- ments to which we now turn our attention can only be regarded as intrinsic thermoscopes, which, in order to give intelligible numerical results, must be graduated by direct comparison with an air-thermometer. Some of them may indeed be used by extrapolation to give a numerical measure of temperatures outside the practical range of the air-thermometer, employing for that purpose a formula verified for temperatures within the range. A case in point is the determination of the temperature of fusion of platinum by the calorimetric method described below. These intrinsic thermoscopes are frequently much more convenient in practice than any of the modifications of the air-pyrometer. 5. Discontinuous Intrinsic Thermoscopes. The best ex- ample of the measurement of temperature by a discontinu- ous intrinsic thermoscope is that suggested by Prinsep. 1 He formed a series of definite percentage alloys of silver and gold and of gold and platinum. The melting-points of these alloys give a series of fixed temperatures lying be- tween the melting-points of silver and gold and of gold and platinum respectively. An observation is taken by exposing in the furnace, upon a small cupel, a set of small flattened specimens of the alloys, not necessarily larger than pin heads, and noticing which of them are fused. The temperatures of fusion of these alloys have been determined by Erhard and Shertel 2 ; their results are given in the following table, taken from Landolt and Bbrnstein's Physikalisch-chemischa Tabellen Table I. The Fusing-Points of Prinsep' s Alloys. 1. SILVER AND GOLD. Per cent of silver. Per cent, of gold. Fusing- point' Per cent, of silver. Per cent, of gold. Fusing- point. ].',.! 954 40 60 1020 80 20 975 20 80 1045 60 40 995 100 1075 2. GOLD AND PLATINUM. Per cent. Per cent of Fusing- Per cent Per cent, of Fusing- of gold. platinum. point.3 of gold. platinum. point 100 1075 45 55 1420 95 5 1100 40 60 1460 90 10 1130 85 65 1495 85 15 1160 30 70 1535 80 20 1190 25 75 1570 75 25 1220 20 80 1610 70 SO 1255 15 85 1650 65 35 1285 10 90 1690 60 40 1320 5 95 1730 65 45 1350 100 1775 50 50 1385 It is said, however, that some difficulty is met with in the use of Prinsep's alloys in consequence of the property possessed by silver of taking up oxygen when melted and ejecting it on solidify- ing and of molecular changes in the alloys which make it unadvis- able to use the same specimen more than once. A similar method lias recently been employed by Carnelley and Carleton Williams, 4 in which metallic salts with high fusing- points were employed instead of alloys, the fusing-points being initially determined by a calorimetric method. These methods recall an old empirical method sometimes employed in porcelain manufacture for estimating the 1 Phil. Trans., 1828, p. 79. 2 Jahrb. fur das Berg- und Hiltten-Wesen in Sachsen, 1879. 3 Determinations of temperature by a porcelain air-thermometer. Errors in general less than 20. 4 See Chem. Soc. Jour., 1876, i. 489 ; 1877, i. 365 ; 1878. temperature of a furnace. Certain "pyrometrical beads" or "trials" i.e., small hoops or gallipots of clay indicated the temperature by their tint much iti the same way as the proper temperature is indicated by the colour of steel in tempering. 6. The Calorimetric Method. This is a very conveni- ent method and is often practically employed for measur- ing the temperature of furnaces. The observation consists in determining the amount of heat given out by a mass of platinum, copper, or wrought-iron on cooling in water from the high temperature. The theory is simple. Let m be the capacity for heat of the calorimeter and of the water contained in it, M the mass of metal, T the tem- perature required, t the initial temperature of the water in the calorimeter, 6 the final temperature of the water after the introduction of the metal, and K the mean specific heat of the metal between the temperatures 6 and T. Then M ' . K The value of K, the mean specific heat of the metal between the temperatures occurring in the experiment, must be determined by precisely similar calorimetric experiments, in which the high tem- perature T is determined by the application of one of the air-pyro- meter methods. The following table (II.) gives the best-known determinations of the mean specific heat of platinum for different ranges of temperature. Table II. Mean Specific Heat of Platinum. Pouillet.s by platinum Veinhold,8 by porcelain reservoir air-thermo- j reservoir air -thermo- Violle,7 by porcelain re- servoir air - thermo- meter. 1 meter. meter. Range of Mean ' Range of Mean Range of Mean temp. spec. heat. temp. spec. heat. temp. spec. heat. 0- 100" 0-03350 10-2 - 99 -1 0-03287 0-100 0-0323 200 0-03392 16 -49-238 -5 0-03270 300 0-03434 16'9 -246-4 0-03520 400 0-03476 17'2 -256-8 0-03411 500 0-03518 23 -5 -476 0-03188 600 0-03560 24' '6 -478 0-03230 700 800 0-03602 0-03644 25'4 -507 20 -7 -705 0-03253 0-03333 0-7S4 0-0365 900 003686 j! 23 -6 -766 0-03381 1000 0-03728 22 -3 -934 0-03396 0-1000 00377 1100 0-03770 17 '3 -952 0-03333 0-1177 0-0388 Violle's results give, if c ' be the mean specific heat between and t, c '= 0-0317 + -000006^. Assuming this formula to hold beyond the verified limits, he obtains by calorimetric observations 1779 C. as the temperature of the melting-point of platinum. The tine specific heat of wrought-iron at temperature t is accord- ing to Weinhold (I.e.) given by the formula c t = c + at + pP, where c = 0-105907, = 0-00006538, /3 = 0-00000006647 7, and the^ total heat obtained from unit-mass of wrought-irou cooling from t to t is thereforey7 2 ( c o + at + P$)dt. The specific heat of copper does not appear to have been accurately determined for high temperatures. The determinations by Bede, quoted by Landolt and Bornstein (op. cit., p. 178) are 15-100 mean specific heat 0-09331 ; 16-172 0-09483; 17-247 0-09680. There are two obvious sources of error of considerable amount in the use of the calorimeter for pyrometrical purposes, viz., (1) the liability of the metal to lose heat during its passage from the fur- nace to the calorimeter, and (2) the evaporation of water from the calorimeter. With the small mass of platinum generally used, the former source of error is likely to be very important, for the temperature of a mass of 50 grammes of mercury at 100" C. may fall a full degree in being carried to a calorimeter 3 feet away. It does not appear that any estimates of the amount of loss which may be so produced in calorimetric determinations have been published ; but in order to reduce the loss Salleron 8 suggests the employment of a platinum or copper carrier in which to heat the mass of metal, and J. C. Hoadly" uses a graphite crucible for that purpose. The second source of loss is more easily disposed of. Weinhold (I.e.) uses a calorimeter closed by a lid and quite filled with water. This is provided with a broad tube passing nearly to the bottom of the calorimeter, and the latter is tilted while the platinum mass is being introduced ; whereas Violle 10 gets over the same difficulty by the use of a calorimeter provided with a platinum " eprouvette, " so that the heat is imparted more slowly to the water. In a calori- metric pyrometer for technical purposes, made by Messrs Siemens 6 C. R., iii. p. 786 (1836). 6 Pogg. Ann., cxlix. 7 Phil. Mag., [5], iv. p. 318. 8 Chem. News, xxvii. 77. 9 Jour, of Franklin Inst., xciv. p. 252. 10 Phil. Mag., [5], iv. p. 318.
