Septemher 30, 1897] 



NA TURE 



529 



gold, silver and platinum, and determined them very well. Other 

 alloys were afterwards substituted, and graded mixtures made of 

 quartz, chalk, kaolin and felspar for the purpose. Efforts to ob- 

 tain more accurate values are due to Becquerel, but the absolute 

 values most widely used until quite recently, namely, the melting 

 points of silver (958°), gold (1035°), copper (1054°), palladium 

 (1500"), platinum {1775'), iridium (1950 ), are due to the re- 

 searches of VioUe. 



Interest in high-temperature fusions has of recent date rather 

 increased than abated. The demand for more accurate data has 

 been met by the Reichsanstalt, and we have now a set of values 

 for silver, copper, gold, nickel, palladium and platinum in terms 

 of the air-thermometer standard of that institution. Data have 

 also been supplied by Callendar. Among these values there is as 

 yet considerable confusion, and the end is not yet. Long ago I 

 suspected that the Violle melting points were probably too low, 

 whereas the assumed zinc boiling point is probably too high. 

 This surmise has been partially borne out by the Reichsanstalt, 

 though Le Chatelier even now prefers \'iolle's values.' 



Thermoscopes based on a specific heat have an advantage over 

 fusion thermoscopes in not being discontinuous. They are quite 

 as " intrinsic " and much less convenient in practice. Guyton- 

 Morveau at the beginning of the century pointed out the pyro- 

 meter importance of specific heat, and a host of observers 

 followed him. But the critical discussion of the subject is due 

 to Pouillet (1836), who determined the thermal capacity of 

 platinum between 0° and 1200° absolutely, and found a value so 

 nearly constant as to place this method of pyrometry in a very 

 favourable light Other observers followed with new data, and 

 the bulk of our knowledge to-day is again due to Violle. Violle 

 used Deville and Troost's exhaustion air thermometer, and 

 determined the law of variations of specific heat and temperature 

 throughout a large pyrometric interval with a number of metals, 

 silver, gold, copper, palladium, platinum, iridium among them. 

 It was by prolonging this law as far as fusion that the melting 

 points of the metals, to which I have already alluded, were 

 obtained. This verges on extrapolation, but it is not extrapola- 

 tion gone mad. 



The importance of calometric high temperature measurement 

 has recently been accentuated in connection with the remarkable 

 high temperature accomplishments of Moissan. P\irnace 

 temperatures in the case of such technological operations as are 

 used in connection with iron, glass and porcelain manufacture, 

 rarely exceed 1400° ; except perhaps in the Bessemer {)rocess, 

 where the temperatures are wont to exceed 1600° and even 

 reach 2000°. In Moissan's furnace, which is essentially an 

 electric arc enclosed by nonconducting lime, a totally new order 

 of high temperatures is impressed. There was thus a call for at 

 least an approximate measurement of their values which was 

 answered by Violle, assuming that the specific heat of carbon 

 above 1000° approaches a limit of 0*5 calorie. The sufficiency 

 of this hypothesis is not unchallenged, however ; for instance, 

 Le Chatelier finds that, up to 1000°, the specific heat of carbon 

 continually increases having no certain limit. Admitting Violle's 

 results, Moissan's furnace temperatures exceed 2000° even at 

 30 amperes and 55 volts ; at 360 amperes and 70 volts tin and 

 zinc oxides melt and boil ; they exceed 3000" at 500 amperes and 

 70 volts, where lime melts, and often boils. Moissan, however, 

 went as far as icxxj amperes at 50 volts. 



The striking novelty of Moissan's work is rather of chemical 

 interest, and a large part of it is so fresh in our memory that in 

 view of Moissan's valuable Vjook (" Le Four electrique," par 

 Henri Moissan ; Paris, Steinheil), I need merely glance at it 

 A range of fusibilities, among which platinum lies lowest, while 

 chromium, molybdenum, uranium, tungsten, vanadium, follow 

 in order, and t)f ebullitions beginning with silica and zinc oxide, 

 is rather breath-taking. Finally his structural investigations on 

 the occurrence of carbon, and his long series of carbides, many 

 of them commercially valuable, have staggered even the sensa- 

 tional press. 



Leaving other intrinsic thermoscopes for the moment, I will 

 ask your attention in this place to the development of the only 

 fruitful method of absolute pyrometry which has yet been devised. 



1 The following table contains a brief summary : 



Ag. (Violle) q°54 (Barus) 986- 985 (Callendar) 982 (H. & W.) 971 



Au. 1045 1091-1093 1037 1072 



Cu. 1054 1096-1097 1082 



Ni. — 1476-1517 1484 



Pd. 1500 1585-1643 1587 



Pt. 1775 1757-1853 J780 



NO. 1457, VOL. 56] 



I refer, of course, to the gas thermometer, or, in other words, to 

 i t'le manometric methods of measuring the thermal expansion o. 

 j gases. Efforts have indeed been made to use gaseous viscosity 

 I for absolute high temperature work. It has been definitely 

 pointed out, inasmuch as viscosity in gases is independent of 

 pressure, while both viscosity and volume increase with temper- 

 ature, that the transpiration rates of gases through capillary 

 tubes of platinum glazed externally would necessarily be an 

 exceedingly sensitive criterion of the variation of high tempera- 

 tures. The small volume of the transpiration pyrometer as 

 compared with the clumsy fragile bulb and appurtenances of the 

 air thermometer is further to the point. But modern kinetics 

 has as yet failed to fathom the law of variation of viscosity with 

 temperature, and even the researches of O. E. Meyer in this 

 direction do not seem to have quite touched bottom. Gaseous 

 transpiration pyrometry is thus still much in the air, surveying 

 the horizon of a glorious future. 



Returning from this digression to the air thermometer, we find 

 the first thorough-going piece of high temperature work carried 

 out by Prinsep (1829) by the aid of a reservoir of pure gold, to 

 which I have already alluded. Prinsep's manometer was filled 

 with olive oil, and the volume issuing at constant pressures was 

 found by the balance. In view of the pure olive oil, probably 

 available in 1829, these experiments must have been comfortably 

 appetising, and I dare say Prinsep's good humour in the matter 

 may have contributed to the remarkable excellence of his results. 

 Prinsep's researches were not superseded until Pouillet, in 1836, 

 published his paper on pyrometry. Pouillet constructed an air- 

 thermometer bulb of platinum, and was thus able to reach the 

 farthest pyrometric north of the day and long after. His results 

 are many-sided ; they contain the first definite data in radiation 

 pyrometry and in calometric pyrometry. His constant pressure 

 manometer, afterwards further perfected by Regnault, is the best 

 apparatus lor the purpose to-day. Pouillet did not suspect, in- 

 deed he remained quite unaware of, the permeability of platinum 

 to furnace gases ; perhaps for this and other reasons he failed to 

 detect the thermo-electric anomalies in the platinum-iron couple 

 which he has so carefully calibrated. 



It was thus a great step in advance when Deville and Troost 

 long after replaced platinum by glazed porcelain, availing them- 

 selves (1857-60) of Dumas' famous vapour density method for 

 measuring temperature. Unfortunately for the rapid progress of 

 pyrometry Deville and Troost used iodine vapour in their bulbs, 

 a heavy gas indeed, but a gas, as was afterwards found, whose 

 low temperature molecule dissociates at higher temperatures. 

 Thus they unwittingly committed an even greater error than 

 Pouillet in gliding over permeable platinum ; and their data for 

 the boiling points of zinc and of cadmium were about 100° 

 too high. In fact these results were challenged not long after by 

 Becquerel (1863), who had fallen heir to Pouillet's platinum air 

 thermometer, had used it to calibrate a platinum-palladium 

 thermo-couple of his own, and had found data for the boiling 

 points of zinc and cadmium upwards of 110° below those of 

 Deville and Troost. I cannot here enter upon the discussion 

 which thereafter arose between these active observers, further 

 than to state that in the course of it both parties frequently 

 repeated their measurements (Becquerel even substituting a 

 porcelain bulb for Pouillet's thermometer) without removing 

 the discrepancy between their values. 



Later researches have decided in favour of Becquerel's results, 

 and his original research, with its applications to fusion, to 

 radiation, to thermo-electrics, &c., is one of the noteworthy 

 accomplishments in the history of pyrometry. Nevertheless it. 

 must not be forgotten that to the investigation of Deville and 

 Troost our knowledge of the perviousness of iron, platinum, and 

 other metals to gases is due. We are also indebted to Deville 

 for the great discovery of dissociation, the very phenomenon 

 which he was here so loth to acknowledge. This is the case of 

 a man stumbling in his own footprints. Victor Meyer was, I 

 believe, the first to point out the probable dissociability of the 

 iodine molecule, suggesting a fruitful subject of research which 

 has since been extended to many other molecules. 



In 1863, Deville and Troost began a new series of high tem- 

 perature researches, the feature of which is the perfected form 

 of porcelain bulb. This was a hollow sphere and long capillary 

 stem adapted for use with Regnault 's standard air thermometer. 

 Great difficulties were encountered in the endeavour to glaze 

 the bulbs within. They were finally overcome by making bulb 

 and stem separately, and then soldering them together with felsjjar 

 before the oxyhydrogen blow-pipe. Elaborate measurements 



