Wedgwood's Pyroscope. Frontispiece^ The Measurement of High Temperatures BY G. K. BURGESS * i BUREAU OF STANDARDS AND H. LE CHATELIER MEMBRE DE L'lNSTITUT . . THIRD EDITION. REWRITTEN AND ENLARGED FIRST THOUSAND NEW YORK JOHN WILEY & SONS LONDON: CHAPMAN & HALL, LIMITED 1912 .11 COPYRIGHT 1901, 1904, 1912, BY G. K. BURGESS Entered at Stationers' Hall, London Stanhope ftress F. H. GILSON COMPANY BOSTON, U.S. A. PREFACE I. THE main purpose of this preface is to recall the origin of the volume Mr. Burgess and I present to the reader. For a long time, all precise, scientific investigations at high temperatures were made impossible by the absence of suitable methods for the measurement of these temperatures. Wedgwood, more than a century ago, had already insisted on the capital importance of the carrying out of high-temperature investigations, and devised for this purpose his pyrometer, which is but an arbitrary-com- parison apparatus. The question was taken up later by many scientists, but with little success, until I called attention defi- nitely to the precision to be obtained by the judicious use of thermoelectric couples. Pouillet about 1830, and afterwards Edmond Becquerel, had made some measurements with the gas thermometer provided with a platinum bulb. This method, however, was completely discredited following the discovery, by Henry Sainte-Claire Deville, of the permeability of platinum to hydrogen; and it is only since the very recent employment of platinum-wound elec- tric resistance furnaces, free from all combustible material, that it has been possible to obtain accurate measurements with the platinum-bulb gas thermometer; but its complexity and the dif- ficulty of manipulation limit its use to the standardization of other measuring apparatus. Henry Becquerel, later his son Edmond, and also Pouillet, advocated the employment of thermoelectric couples; but their use did not spread, and Regnault emphatically condemned them after finding serious irregularities in their behavior. These anomalies, the cause of which he did not then recognize, were due, as I showed later, to the use of iron as one of the 236339 VI PREFACE I metals of the couple, and to imperfections in the methods of electrical measurements. Violle, following Regnault and Sir William Siemens, proposed the calorimetric method with platinum as the heated substance, instead of iron used by Regnault and copper by Siemens, in their industrial pyrometers. This complicated method, of deli- cate and slow manipulation, did not come into general use. Mention should also be made of other isolated and even more restricted attempts, the application of which hardly exceeded a series of observations by a single experimenter. Sir William Siemens proposed the electrical resistance pyrometer; before this, Edmond Becquerel had suggested a radiation pyrometer; finally, several observers sought to apply to determinations of the sun's temperature certain heat-radiation methods. In 1885, when I attacked the problem of the measurement of high temperatures, it is fair to say there existed nothing defi- nite available on this important question; we possessed only qualitative observations for temperatures above 500 C. En- gaged at that time in industrial studies relative to the manu- facture of cement, I sought a method which above all would be rapid and simple, and decided on the use of thermoelectric couples, intending to determine the order of magnitude of the sources of error noticed by Regnault. The readings of even a crude galvanometer might be very useful in technical work, pro- vided the limitations of its accuracy were appreciated. I soon recognized that the errors attributed to this method could easily be eliminated by discarding in the construction of the couples certain metals, such as iron, nickel, and palladium, which give rise to singular anomalies; and indicated a simple test for rec- ognizing the suitable metals. One takes a stretched wire of the metal, the ends of which are connected to the terminals of a sufficiently sensitive galvanometer, indicating at least T^^O vo ^> and the wire is then heated from point to point with a Bunsen flame, which is carried back and forth beneath the wire, when no electric currents should be produced. Now, iron and pal- ladium, the two metals advocated by Becquerel and Pouillet, PREFACE I vii give rise to large and variable parasite currents which diminish the accuracy of the measurements. Among the different metals and alloys studied, pure platinum and the alloy of platinum and rhodium which are still used to-day, gave the most satis- factory results. Finally, I called attention to the importance, overlooked by Regnault, of employing only galvanometers of high resistance, to avoid the influence of variations in resistance of the wires of the couple when heated. I recommended also the calibration of the couples, not against the air thermometer directly, as Bec- querel had tried to do, but in terms of the fixed points of boiling or fusion of certain pure substances, in such a way that, when these temperatures should be known more exactly, as is the case since my earlier researches, the results could be corrected with certainty. Some months later, at the request of Sir Robert Hadfield, director of the Hecla Steel Works, I developed an optical py- rometer, and calibrated it by comparison with the thermoelectric couple. By means of these two instruments, I determined a large number of temperatures, in the laboratory and in the in- dustries, and rectified, often by several hundred degrees, the numbers previously admitted in terms of fantastic estimations. From this date, the measurement of high temperatures came rapidly into general use in the laboratory as well as in the in- dustries. A few years later, in a course of lectures delivered during the year 1898 at the College de France, I thought it use- ful to give a summary of the progress accomplished. These lectures, gathered into book form with the aid of my assistant, Mr. Boudouard, formed the first edition of this work. Mr. Burgess, who had followed my lectures, took the trouble to translate it into English; but, while there was little demand for the French edition, the English translation was soon exhausted. Mr. Burgess wrote a second edition, considerably improved and enlarged by him; this is again exhausted. This time Mr. Bur- gess has rewritten anew the whole book, so that it is no longer a translation but an original work which we present to the reader. VU1 PREFACE I For several years past, my studies have taken me into other fields of investigation, and I have been unable to follow the con- siderable progress realized in the measurement of temperatures. Mr. Burgess, on the contrary, has been actively interested in these new researches and to him is due an important part in the more recent advances. Consequently, this book is much more his work than mine, which enables me to praise it as it deserves, and state that this publication will render great service both to investigators and engineers. H. LE CHATELIER. PARK, February 15, ign. PREFACE II. SINCE the appearance in 1900 of Mesure des Temperatures Elevees by Messrs. Le Chatelier and Boudouard, the theory and practice of pyrometry have grown greatly, and methods which at that time were in a tentative stage of development have been improved in accuracy and convenience, and adapted by means of new instruments both to technical and scientific re- quirements. In gas pyrometry, accurate measurements may be said to have been initiated at the Reichsanstalt by the publication in 1900 of a series of metal freezing points, by Holborn and Day, constituting what is still known as the Reichsanstalt tempera- ture scale. Again, it is only since 1900 that the significance of the applica- tion of the laws of radiation to pyrometry has been appreciated. The theoretical work of Wien, Planck, and others closely con- temporaneous with the experimental verifications of Paschen, Lummer and Pringsheim, and many others, was soon followed by the optical and radiation pyrometers of Wanner, Fery, Morse, and Holborn and Kurlbaum, and by many applications to technical and scientific uses. In thermoelectric and electric resistance pyrometry there has been in recent years an unparalleled improvement in the design of electric measuring instruments, millivoltmeters, potentiom- eters, galvanometers, Wheatstone bridges and the like, suitable for use in temperature measurements, either in the works or in the laboratory. There have also been executed since 1900 several exact experimental investigations in pyrometry with such apparatus, notably at the several national laboratories and at the Geophysical Laboratory of Washington. Furthermore, X PREFACE II the subject of automatic temperature recording has received a great deal of attention, resulting in many new instruments. In so far as practicable in the following pages, we have dwelt less upon particular types of instrument than on the principles underlying them. We have, however, consulted nearly all the manufacturers of pyrometers as to their practice, and have drawn very freely on the material they have been so kind as to put at our disposal, material that in several instances is other- wise unpublished, and for which we express our obligation. We have kept in mind three classes of readers: the student, to whom the historical aspect and fundamental principles are of prime interest ; the engineer, who is interested mainly in adapting some method or instrument to his particular technical opera- tion; and the investigator, who has an intensive interest in accurate methods of measurement and their adaptability to his needs. We realize that one book cannot meet satisfactorily all these requirements. If the wants of the investigator have been somewhat neglected, he has ready access to the literature, a summary of which is given in the Bibliography. We are indebted to Dr. C. W. Waidner for many suggestions; to Dr. R. B. Sosman for reading the chapters on Gas and Ther- moelectric Pyrometry; and especially to Dr. A. L. Day, from whose criticisms of the manuscript we have been able to profit greatly. GEORGE K. BURGESS. WASHINGTON, August 24, ign. CONTENTS. INTRODUCTION. PAGE Thermometric Scales 2 Fixed Points 5 Pyrometers .- 9 CHAPTER I. STANDARD SCALE OF TEMPERATURES. Laws of Mariotte and Gay-Lussac 13 Gas Thermometers 14 Constant- volume Thermometer. . . , 14 Constant-pressure Thermometer 15 Thermometer of Variable Pressure and Mass 15 Volumetric Thermometer 15 Experiments of Regnault 16 Results Obtained by Chappuis 20 Normal Scale of Temperatures 21 Thermodynamic Scale 26 Approximate Expression 26 Second Approximation 28 Gas-scale Corrections 30 The Ice Point 34 CHAPTER II. GAS PYROMETER. Introduction 37 Standard Gas Thermometer 38 Formulae and Corrections 44 Constant-volume Thermometer 44 Constant-pressure Thermometer 50 Volumetric Thermometer 51 Substance of the Bulb 53 Platinum and its Alloys 54 Iridium 56 Iron 56 Porcelain 57 Glass 59 Quartz 6a xi xii CONTENTS PAGE Early Experimenters 61 Pouillet 61 Ed. Becquerel 62 Violle 63 Mallard and Le Chatelier 65 Barus 66 Recent Experimental Investigations 67 Holborn and Day 68 Jaquerod and Perrot 70 Calendar's Constant-pressure Thermometer 70 Holborn and Valentiner 75 Day, Clement, and Sosman 75 Comparison of Results 79 Suggestions for Future Experiments 80 Methods 80 Bulb 81 Gas 82 Manometer 82 Industrial Air Pyrometers 82 Indirect Processes 83 Methods of Crafts and Meier 83 Methods of H. Sainte-Claire-Deville 85 Method of D. Berthelot 85 CHAPTER III. CALORIMETRIC PYROMETRY. Principle 89 Choice of Metal OI Platinum OI Iron 02 Nickel Q3 Copper 03 Calorimeters O4 Industrial Calorimeter Q5 Siemens Calorimeter 0,5 Precision of Measurements Q7 Conditions of Use oo CHAPTER IV. THERMOELECTRIC PYROMETER. Principle IOI Experiments of Becquerel, Pouillet, and Regnault 101 Experiments of Le Chatelier and of Barus I0 2 Choice of the Couple IO5 Electromotive Force IO ,- Absence of Parasite Currents IO 5 Chemical Changes I0 - CONTENTS xiii PAGE Thermoelectric Formulae 108 Thermoelectric Power 109 Formulae m Platinum and its Alloys in Variation of E.M.F. with Composition 116 The Base-metal Couples 116 Methods of Measurement of Temperature 116 Galvanometric Method 118 Resistance of Couples and Galvanometer 118 Pyrometer Galvanometers 120 Types of Suspended-coil Galvanometers 124 Pivot Galvanometers 129 Temperature Coefficient of Galvanometers 131 Galvanometer Requirements for Industrial Practice 133 The Galvanometer Method in the Laboratory 135 Potentiometric Methods 135 Apparatus Required 136 Principle of the Method 137 Potentiometers for Use with Thermocouples 139 The Potentiometer Indicator 140 Precision Requirements 141 Types of Thermocouple Potentiometer 143 The Thermocouple Circuit 147 Junction of the Wires 148 Annealing : 149 Insulation and Protection 149 Cold Junction. '. 154 The Cold-junction Correction !55 Elimination of Cold-junction Changes 156 Constancy of Thermocouples 159 Measurement of Inhomogeneity 162 Reproducibility of Thermoelectric Apparatus 164 Base-metal Thermocouples 165 Nickel-copper 167 Nickel-iron 168 Complex Alloy Couples 170 The Noble Metals: GeibePs Data 171 Special Couples 174 Silver-constantan 174 Silver-nickel 174 Iridium-ruthenium 174 Compound Thermocouples 175 Calibration of Thermocouples 178 Precision Calibration 178 Crucible Method 179 Wire Method 183 xiv CONTENTS PAGE: Use of Boiling Points **7 Technical Calibrations l88 Industrial and Scientific Applications iQ 1 Conditions of Use J Q2 CHAPTER V. ELECTRICAL RESISTANCE PYROMETER. Introduction *94 Work of Early Investigators: Siemens *94 Callendar and Griffiths iQ5 Holborn and Wien - 196 Law of Variation of Platinum Resistance 19? Nomenclature 201 Construction of the Platinum Thermometer 202 Choice of Size of Wire 206 Precautions in Construction and Use 206 Methods of Measurement 207 Compensation for Pyrometer Leads 208 Three-lead Thermometer 208 Four-lead Thermometer 211 Wheatstone Bridge Method 211 Precision Bridges 212 The Potential-terminal Thermometer 214 The Kelvin Bridge 215 Sensibility 216 Direct-reading Thermometers 218 Harris Resistance-thermometer Indicator 218 Logometer and Ratiometer 221 Cambridge Deflectional Instrument 223 Leeds and Northrup Indicators 224 Calibration 224 Reduction Tables 226 Some Results Obtained 226 Use as a Standard 227 Sources of Error in Accurate Work 228 Heating by the Measuring Current 228 Lag of the Platinum Thermometer 229 Insulation 229 Compensation for Resistance of Leads 230 Conduction along Leads 231 Use of Impure Platinum 231 Changes in Constants 232 Use of Metals other than Platinum \ 233 Conditions of Use of Resistance Pyrometer 234 Industrial Installations and Checking 234 CONTENTS XV CHAPTER VI. THE LAWS OF RADIATION. PAGE -General Principles 238 Temperature and Intensity of Radiation 238 Emissive Powers 239 The Black Body 239 Experimental Realization 239 Black-body Temperature 242 Kirchhoff's Law 243 Stefan's Law 245 Laws of Energy Distribution 246 Wien's Laws 248 Applications to Pyrometry .253 CHAPTER VII. RADIATION PYROMETER. Principle '. 261 Early Investigators: Temperature of the Sun 262 Pouillet's Pyrheliometer 262 Violle's Actinometer 263 Work of Rosetti 265 Modern Radiometric Apparatus 267 The Thermopile 268 The Radiomicrometer 271 The Bolometer ; 273 The Radiometer , 275 Standard Pyrheliometers 276 Pyrometric Telescopes 277 The Fery Pyrometer 277 Fery Mirror Telescope . . 281 Fery Spiral Pyrometer 283 Other Radiation Pyrometers 283 Some Experimental Results 285 Conditions of Use 286 Calibration 288 Computation 290 CHAPTER VIII. OPTICAL PYROMETER. Principle 291 Properties of Monochromatic Radiation 291 Methods of Temperature Measurement 293 Measurement of Total Luminous Intensity 293 xvi CONTENTS PAGE Measurement of the Intensity of a Simple Radiation 295 Optical Pyrometer of Le Chatelier 296 Photometer 296 Adjustment of Apparatus 299 Measurements 299 Details of an Observation 300 Emissive Power 301 Measurements of Intensity 302 Calibration 302 Evaluation of Temperatures 305 Calibration in Terms of Wien's Law 306 Precision and Sources of Error 306 Modifications of the Le Chatelier Pyrometer 311 Shore Pyroscope 311 Fery Absorption Pyrometer 311 Wanner Pyrometer 315 Calibration 316 Sources of Error. 320 Range and Limitations 322 Instrument for Low Temperatures 324 Holborn-Kurlbaum and Morse Pyrometers 324 Holborn-Kurlbaum Form 325 Morse Thermogage 329 Henning's Spectral Pyrometer 330 Calibration of Optical Pyrometers 331 The Wide-filament Comparison Lamp 332 Use of Wedge-shaped Cavities 333 Monochromatic Glasses 334 Extension of Scale 336 Some Scientific Applications 338 Temperature of Flames 338 Temperature of Glow-lamp Filaments 340 Temperature within Furnaces 342 Melting Points of Microscopic Samples 343 Conditions of Use ,44 Some Industrial Uses 34^ Measurement of the Relative Intensity of Different Radiations 346 Use of the Eye ; 34 6 Use of Cobalt Glass. 347 Pyroscope of Mesure" and Nouel 34 g Crova's Pyrometer 3 _ Use of Flicker Photometer 3S2 Stellar Pyrometers Action of Light on Selenium CONTENTS XVil CHAPTER IX. VARIOUS PYROMETRIC METHODS. PAGE Wedgwood's Contraction Pyroscope 357 Expansion of Solids 360 The Joly Meldometer 361 High-range Mercury Thermometers 362 Fusing-point Pyrometry 365 Metallic Salts 365 Sentinel Pyrometers 366 Fusible Cones 368 Recent Investigations on Seger Cones 372 Wiborgh's Thermophones 376 Dilution Pyrometers 377 Transpiration Pyrometers 378 Vapor-pressure Pyrometers 380 Other Pyrometric Methods 380 CHAPTER X. RECORDING PYROMETERS. Forms of Temperature Records 381 Types of Cooling Curves 383 Methods of Recording 384 Recording Gas Pyrometer 385 Electrical Resistance Recording Pyrometer 386 Callendar's Slide- Wire Recorder 386 Deflectional Recorders 390 Leeds and Northrup Recorders 393 Carpentier's Electrothermal Recorder ; 394 Thermoelectric Recording Pyrometer 395 Temperature-rate Recorders 396 Le Chatelier's Experiments 396 Dejean's Apparatus 399 Temperature-time Recorders 401 Apparatus of Sir Roberts-Austen 401 Autographic Recorders 408 Semiautomatic Recording 413 Differential Curves 414 Use of a Neutral Body 414 Saladin's Apparatus 418 Registration of Rapid Cooling 420 Le Chatelier's Experiments 420 Benedicks' Experiments 421 Recording Radiation Pyrometers 423 xviii CONTENTS PAGE Recording Accessories 4 2 ^ Range Control 426 Multiple Records and Circuits 428 Furnace Control and Thermostats 43 CHAPTER XI. STANDARDIZATION OF PYROMETERS. Thermometric Scales 43 2 Fixed Points 433 Sulphur 433 Zinc 437 Gold 439 Silver 44 Copper 44i Palladium 442 Platinum 442 Rhodium 444 Iridium 444 Other Metals melting below 1100 C : 445 The Iron Group 445 Metals melting above 2000 C 446 Typical Freezing-point Curves 446 Boiling Points: Water, Aniline,, Naphthaline, Benzophenone 447 Metallic Salts 450 Alloys: Eutectic Points . 450 Reproducibility of Freezing Points. 451 Temperature of the Arc and Sun. . . ..^ 452 Table of Fixed Points 453 Standardization of Pyrometejs 4156 Standardizing Laboratories - . ' 457 Metals and Salts of Certified Melting Points 457 Electrically Heated Furnaces 458 Crucible Furnaces 4 59 Vacuum and Pressure Furnaces 461 Bibliography 465 Appendix : Tables 489 Index 499 HIGH TEMPERATURES INTRODUCTION. WEDGWOOD, the celebrated potter of Staffordshire, the inven- tor of fine earthenware and of fine china, was the first to occupy himself with the exact estimation of high temperatures. In an article published in 1782, in order to emphasize the importance of this question, he considered at length certain matters a study of which would be well worth while even to-day. "The greater part of the products obtained by the action of fire have their beauty and their value considerably depreciated by the excess or lack of very small quantities of heat; often the artist can reap no benefit from his own experiments on account of the impossibility to duplicate the degree of heat which he has obtained before his eyes. Still less can he profit from the experi- ments of others, because it is even less easy to communicate the imperfect idea which each person makes for himself of these degrees of temperature." Joining example to precept, Wedgwood made for his personal use a pyrometer utilizing the contraction of clay. This instru- ment, for nearly a century, was the only guide in researches at high temperatures. Replaced to-day by apparatus of a more scientific nature, it has been perhaps too readily forgotten. Since Wedgwood, many have undertaken the measurement of high temperatures, but with varying success. Too indifferent to practical requirements, the early experimenters above all regarded the problem as a pretext for learned dissertations. The novelty and the originality of methods attracted them more than the precision of the results or the facility of the measure- ments. Also, up to the past few years, the confusion was on the " HIGH TEMPERATURES increase. The temperature of a steel kiln varied according to the different observers from 1500 to 2000; that of the sun from 1500 to 1,000,000. More recently there has been great im- provement in methods. First of all, let us point out the chief difficulty of the problem. Temperature is not a measurable quantity in the strict sense of the term. To measure a length or a mass, is to count how many times it is necessary to take a given body chosen as a unit (meter, gram) in order to obtain a complex system equivalent either as to length or mass of the body in question. The possibility of such a measurement presupposes the previous existence of two physical laws: that of equivalence and that of addition. Tem- perature obeys well the first of these laws; two bodies in tempera- ture equilibrium with a third, and thus equivalent with respect to exchanges of heat in comparison with this third body, will also be equivalent, that is to say, equally in equilibrium with respect to every other body which would be separately in equilibrium with one of them. This law allows determination of temperature by comparison with a substance arbitrarily chosen as a thermo- metric body. But the second law is wanting; one cannot, by the juxtaposition of several bodies at the same temperature, realize a system equivalent, from the point of view of exchanges of heat, to a body of different temperature; thus temperature is not measured, at least insomuch as one considers only the phenomena of convection. In order to determine a temperature one observes any phenom- enon whatever varying with change of temperature. Thus for the mercury centigrade thermometer the temperature is denned by the apparent expansion of mercury from the point of fusion of ice measured by means of a unit equal to T ^ of the dilatation between the temperature of the fusion of ice and that of the ebulli- tion of water under atmospheric pressure. Thermometric Scales. For such a determination there are four quantities to be chosen arbitrarily: the phenomenon meas- ured, the thermometric substance, the origin of graduation, and the unit of measurement; while in a measurement properly INTRODUCTION so called there is but one quantity to be arbitrarily chosen, the magnitude selected as unity. It is evident that the number of thermometric scales may be indefinitely great; too often experi- menters have considered it a matter of pride for each to have his own. Here are some examples of thermometric scales chosen from among many: Author. Phenomenon. Substance. Origin. Unit. Fahrenheit Dilatation Mercury ( Very cold \ winter i/i8-> ice to B. P. p ^ Dilatation Ice 1/80 ice to B. P. Celsius Dilatation Mercury Ice i/ioo ice to B. P. Wedgwood {Permanent ) contraction ) Clay Dehydrated i/24ooinit. dimens. Pouillet Dilat. at const, p. Air Ice ( Normal ther.) (Therm odynamic scale) Dilat. at const, v. ( Reversible heat ) 1 scale j Hydrogen Anything Ice Heat =o i/iooice to B. P. Siemens Electric resistance Platinum Ice The enormous differences above mentioned in the estimations of high temperatures are much more the result of the diversity of the scales than due to the errors of the measurements them- selves. Thus the experiments on solar radiation which have led to values varying from 1500 to 1,000,000 are based on measure- ments which do not differ among themselves by more than 25 per cent. To escape from this confusion it was first necessary to agree upon a single scale of temperatures; that of the gas thermometer is to-day universally adopted, and this choice may be considered as permanent. The gases possess, more than any other state of matter, a property very important for a thermometric substance, - the possibility of being reproduced at any time and in any place identical with themselves; besides, their dilatation, which defines the scale of temperatures, is sufficient for very precise measurements; finally, this scale is practically identical with the thermodynamic scale. This last is in theory more important than all the other properties because it is independent of the nature of the phenomena and of the substances employed. It gives, too, a veritable measure and not a simple comparison; its 4 HIGH TEMPERATURES only inconvenience is for the moment not to be experimentally realizable, at least rigorously, but this will probably not always be the case. The adoption of the scale of the gas thermometer does not in any way imply the obligation to use this instrument actually in all measurements. Any thermometer may be taken, provided that in the first place its particular scale has been standardized by comparing it with that of the gas thermometer. According to the case, there will be advantage in employing one or another method; practically also one almost never employs the gas ther- mometer by reason of the difficulties inherent in its use, which result principally from its great dimensions and the onerous manipulation required. For the estimation of very high temperatures the gas scale can be used only by an indirect extrapolation in terms of some property of matter whose variation has been studied within the range of the gas scale attainable experimentally and which vari- ation is assumed to obey the same law at temperatures beyond which control cannot be had with the gas thermometer. The fact that certain of the radiation laws, to which resort must be had for the estimation of the highest temperatures, have a thermodynamic basis and may therefore be considered an extension of the thermodynamic scale, is of the greatest impor- tance in the extrapolation for temperatures above the attainable limit of the gas scale. There are several series of temperature measurements on the gas scale in good agreement to 1100 C., and two series reaching nearly to 1600 C. which differ by 25 at this temperature. Be- yond 1600 C. the most infusible substances permanently alter their properties, and we are forced to measure temperature in terms of the radiations coming from heated bodies for the reason that we have not been able to find any other than the radiating properties of such excessively heated bodies whose variations can be measured without destroying or permanently altering either the substance used as pyrometer or the substance examined. Perhaps also chemical methods may be employed eventually. INTRODUCTION 5 It is in the realm of the laws of radiation and their applications to pyrometric methods that some of the most recent and impor- tant advances in high-temperature measurements have been made, so that, with certain restrictions which will be treated in the chapter on the laws of radiation, it is possible to measure on a common scale the temperatures of bodies heated to the highest attainable limits. It is our purpose, in this introduction, to pass in review rapidly the different pyrometric methods (that is to say, thermometers utilizable at high temperatures) whose employment may be ad- vantageous in one or another circumstance; we shall then de- scribe more in detail each of them, and shall discuss the conditions for their employment. But in the first place it is necessary to define within what limits the different scales may be compared to that of the normal gas thermometer; it is the insufficiency of this comparison which is still to-day the cause of the most im- portant errors in the measurement of high temperatures. Fixed Points. The standardization of the different pyrom- eters is the most frequently made by means of the fixed points of fusion and ebullition which have been determined in the first place by means of the gas thermometer; the actual precision of the measurements of high temperatures is entirely subordinate to that with which these fixed points are known; this precision was for a long time most unsatisfactory because these fixed points could only be determined in an indirect manner with the gas thermometer, and some of them only by aid of processes of ex- trapolation, always very uncertain. Recent researches, however, by various observers, in which improved methods of heating have been used, as well as greater purity of materials and more carefully constructed and calibrated apparatus, have led to much more concordant results, in the determination of fixed points, even by most varied methods. Violle was the first to make a series of experiments of consider- able temperature range, which up to the last few years were our most reliable data on the question. In a first series of researches he determined the specific heat of platinum by direct comparison 6 HIGH TEMPERATURES with the air thermometer between the temperatures of 500 and 1200. He made use indirectly of the relation thus established between specific heat and temperature to determine by compari- son with platinum the points of fusion of gold and silver; then, by extrapolation of this same relation, the points of fusion of palladium and of platinum. ( Ag* Au Pd Pt Jl n ...................... \954 1045 1500 1779 Finally, in a second series of experiments, he determined by direct comparison with the air thermometer the boiling point of zinc. Boiling point... . Barus, when physicist of the United States Geological Survey, determined the boiling points of several metals by means of thermoelectric couples standardized against the air thermometer. ( Cd Zn Boiling point ........... \ 7?2 o and ?84 o 9260 and Q3i o Mean ................... 77# 928.5 Callendar and Griffiths, by means of a platinum resistance thermometer calibrated up to 500 by comparison with the air thermometer, have determined the following points of fusion and ebullition: Sn Bi Cd Pb Zn FUSi n \2 3 2 2 7 J2I' Boiling point under ( Aniline Naphthaline Benzophenone Mercury Sulphur 760 mm 1 184.1 217.8 305.8 356.7 444.5 These last figures may be compared with Regnault's, and Crafts' previous determinations with the gas thermometer: Naphthaline Benzophenone Mercury Sulphur 218 306.1 357 445 Heycock and Neville, employing the same method, but with extrapolation of the law of resistance for platinum established at that time only up to 450, determined the following points of fusion: 232 419 633 629.5 654.5 960.5 1062 1080.5 We shall use figures in italics for all fixed points determined in terms of the gas thermometer without extrapolation. INTRODUCTION 7 Also using the platinum thermometer calibrated at o, 100, and 444.7 C. (the sulphur boiling point), Waidner and Burgess more recently at the Bureau of Standards find the following: FREEZING POINTS Sn Cd Pb Zn Sb Al Ag 3 -Cu 2 Ag Cu-CuzO Cu 231.9 321.0 327.4 419.4 630.7 658.0 779.2 960.9 1063.2 1083.0 BOILING POINTS Naphthaline 218.0 Benzophenone 306-0 Jaquerod and Wassmer, using a hydrogen thermometer of 66 c.c. bulb, find: Boiling Naphthaline. . . .217.7 Benzophenone. . .305.4 One of the most important standardizing temperatures is the boiling point of sulphur to which the value 444-6 C. should be assigned from the work of some half-dozen investigators. At the Physikalisch-Technische Reichsanstalt the question of establishing a temperature scale has received deserved attention. In the early nineties Holborn and Wien, using a thermocouple calibrated in terms of a porcelain-bulb nitrogen thermometer as far as 1400, found the fusing points: Ag Au Pd Pt Fusi n } 9 70 1072 1580 1780' These results for Ag and Au were subsequently found to be high by Holborn and Day, who, after trying porcelain, worked with a platinum-iridium-bulb nitrogen thermometer and ther- mocouple, employing electric heating, two improvements that greatly increased the accuracy. In fact, the return to metal bulbs in place of porcelain, and the introduction of electric heat- ing in place of gas, may be considered the inauguration of modern gas pyrometry. Holborn and Day determined the following on the constant- volume gas scale: . ( Cd Zn Sb Al Ag Au Cu ' 1 321.7 419 630.5 657-5 96l-5 1064 1084 This scale, commonly known as the Reichsanstalt scale, was extended to 1600 by Holborn and Valentiner, who, using extrap- 8 HIGH TEMPERATURES olation by a spectral radiation method, obtained the further fixed points: Fusion. Palladium = 1575 Platinum = 1782 The most recent determinations (1911) of boiling and freezing points are due to Holborn and Henning: Naphthaline = 217. o 6 , Benzophenone = 303.89, Sulphur = 444 .51 Tin = 231. 8 3 , Cadmium = 320.9*, Zinc = 419.4$ Day, Clement, and Sosman, working at the Geophysical Lab- oratory of the Carnegie Institution, have further improved the constant-volume gas thermometer by eliminating the pressure of the gas on the bulb and substituting a Pt-Rh for a Pt-Ir bulb and so reducing the contamination of the comparison thermo- couples caused by evaporation of iridium. Their final tempera- tures of fusion are for the metals they studied: Cd Zn Sb Al Ag Au Cu Ni Co Pd Pt 320.0 418.2 629.2 658.0 960.0 1062.4 1082.6 1452 1490 1549 1755 Mr. Daniel Berthelot, in a series of most skillfully executed in- vestigations extending over several years, has calibrated thermo- couples by comparison with a special form of gas thermometer, making use of the variation of the index of refraction with density. In this way he has found the points: ( Ag Au Fusion \o62 1064 Se Cd Zn Ebullition 600 778 018 Besides these primary measurements there are some very im- portant secondary determinations, which will be discussed later. We may call attention, however, at this point to some of the estimations of very high temperatures obtained by extrapolating the Reichsanstalt scale by means of the radiation laws. The palladium and platinum melting points have been so determined by Nernst and Wartenberg: Palladium = 1541 Platinum = 1745 Again, Wartenberg, using a vacuum tungsten resistance furnace, finds by the same method: Ir Rh 90 Pt-io Rh W 2360 1940 1830 2900 INTRODUCTION 9 Finally, Waidner and Burgess, also using optical methods, find: Pd Pt Ta W I546 1753 2910 3050 Even the boiling points of some of the refractory metals have been determined, although the melting points are to be preferred as high-temperature fixed points. We may cite the very skill- fully executed boiling-point measurements of Greenwood made with an optical pyrometer : Al Sb Bi Cr Cu Pe Pb Mg Mn Ag Sn 1800 1440 1420 2200 2310 2450 1525 1120 1900 1955 2270 From all the results at hand we may conclude that the fixed points possessing the greatest reliability for the indirect stand- ardization of the various thermometric scales and thus for the calibration of pyrometers are the following: Sn Zn Sb Al Ag Au Cu Pd Pt Ir W Fusion j 2 ^ 2 ~ 4I y. 03J . 05Q . 90I - I00 y jotfj- 1550" 1755" 2300" 3000 ( Naphthaline Benzophenone Sulphur Ebullltlon \ 218.0 306.0 444.6 We may consider the high-temperature scale as known with an accuracy better than: 0.5 , between 200 and 500 C. 2. 3- 15- 25- So. 100. 500 800 1 100 1600 2OOO 24OO 800 1 100 1600 2000 2400 3000 A more detailed discussion of the determination of fixed points and their reliability and ease of reproduction will be found in Chapter XI on Standardization. Pyrometers. There have been a great number of pyrometric methods proposed, among which we shall dwell only upon those which have had considerable use or promise to be useful. Gas pyrometer (Pouillet, Becquerel, Sainte-Claire-Deville, Barus, Chappuis, Holborn, Callendar, Day). Utilizes the measurement of change in pressure of a gaseous mass kept at constant volume. Its great volume and its fragility render 10 HIGH TEMPERATURES it unsuitable for ordinary measurements; it serves only to give the definition of temperature and should only be used to stand- ardize other pyrometers. Calorimetric Pyrometer (Regnault, Violle, Le Chatelier, Sie- mens). Utilizes the total heat of metals, platinum in the laboratory and nickel in industrial works. Is to be recom- mended for intermittent researches in industrial establishments because its employment demands almost no apprenticeship and because the cost of installation is not great. Radiation Pyrometer (Rosetti, Langley, Boys, Fery). Utilizes the total heat radiated by warm bodies. Its indications are influenced by the variable emissive power of the different sub- stances. Convenient for the evaluation of very high tempera- tures which no thermometric substance can withstand (electric arc, sun, very hot furnaces), or when it is not convenient to approach the body whose temperature is wanted. Can be made self-registering. Optical Pyrometer (Becquerel, Le Chatelier, Wanner, Hol- born-Kurlbaum, Morse). Utilizes either the photometric meas- urement of radiation of a given wave length of a definite por- tion of the visible spectrum, or the disappearance of a bright filament against an incandescent background. Its indications, as in the preceding case but to a much less degree, are influenced by variations in emissive power. The intervention of the eye aids greatly the observations, but diminishes notably their pre- cision. This method is mainly employed in industrial works for the determination of the temperatures of bodies difficult of access for example, of bodies in movement (the casting of a metal, the hot metal passing to the rolling mill). Can be used to estimate the highest temperatures and is the best method for use above 1700 C. in laboratory and works. Electric Resistance Pyrometer (Siemens, Callendar, Waidner and Burgess). Utilizes the variations of electric resistance of metals (platinum) with the temperature. This method permits of very precise measurements to 1000 C., but requires the em- ployment of fragile apparatus. It merits the preference for very INTRODUCTION II precise investigations in laboratories. As a secondary instru- ment for the reproduction of a uniform temperature scale through- out the range in which the platinum resistance thermometer can be used, to 1000 except in very heavy wire, it is unsurpassed in precision and sensibility. It is also now constructed in conven- ient form for industrial use. Thermoelectric Pyrometer (Becquerel, Barus, Le Chatelier). - Utilizes the measure of electromotive forces developed by the difference in temperature of two similar thermoelectric junctions opposed one to the other. In employing for this measurement a Deprez-d'Arsonval galvanometer with movable coil, one has an apparatus easy to handle and of a precision amply sufficient for industrial and many scientific uses. With a potentiometer, an instrument is obtained of the most considerable precision, available for use to 1600 C., or even to 1750 with proper pre- cautions. This pyrometer was used for a good many years in scientific laboratories, before it spread into general industrial use, where it also renders most valuable service. Contraction Pyrometer (Wedgwood) . Utilizes the permanent contraction that clayey materials take up when submitted to temperatures more or less high. It is employed to-day only in a few pottery works. Fusible Cones (Seger). Utilize the unequal fusibility of earth- enware blocks of varied composition. Give only discontinu- ous indications. Such blocks studied by Seger are spaced so as to have fusing points distant about 20. In general use in pottery works and in some similar industries. There are a number of other pyrometers which have been found suitable in special cases or which for one reason or another have been found convenient in some particular line of work. Some of these we shall mention, among them being the mel- dometer (Joly), interesting to the chemist or mineralogist for determining fusing temperatures of minute specimens; the various industrial instruments based on the relative expan- sion of metals or of a metal and graphite used in air blasts and metal baths; and, finally, pyrometers based on the flow or 12 HIGH TEMPERATURES on the pressure of air or vapor (Hobson, Uhling-Steinbart, Job, Fournier). Recording Pyrometers (Sir Roberts- Austen, Callendar, Le Cha- telier, Siemens and Halske). Finally, we shall describe in some detail the application of registering methods to pyrometry both for technical and laboratory installations, a field that has been cultivated very intensively in recent years. CHAPTER I. STANDARD SCALE OF TEMPERATURES. WE have seen that temperature is not a measurable quantity: it is merely comparable with respect to a scale arbitrarily chosen. The normal or ideal standard scale is the thermodynamic scale ; but as it is impossible to realize rigorously this scale, it is neces- sary to have a practical one. In the same way that, besides the theoretical definition of the meter, there is a practical standard, a certain meter kept at the Bureau International des Poids et Mesures, there exists, besides the ideal scale of temperatures, a practical scale, which is that of a certain gas thermometer which we are going to study. We shall first discuss the gas laws in so far as is necessary for our purpose, and then show how exactly these laws are obeyed by the actual gases that may be used in defining the temperature scale in the several possible ways. Laws of Mariotte and Gay-Lussac. The laws of Mariotte (or Boyle) and of Gay-Lussac are the basis for the use of the dilatation of gases for the determination of temperatures. These two laws may be written Pl'Ol _ I +<*/! / \ poV I+crfo' in which we may assume for the present that the temperatures are being measured with the mercury thermometer from o C. a is a numerical coefficient, the same for all gases, at least to a first approximation, and its value is about a = 0.00366 = - 273 when it is agreed that the interval between the temperatures of melting ice and boiling water is 100. But instead of considering the formula (i) as the expression of 13 14 HIGH TEMPERATURES an experimental law joining the product pv to the temperature denned by the mercury thermometer, we may require of experi- ment merely the law of Mariotte and write a priori the formula in question, giving a new definition of temperature approximating that of the mercury thermometer. This new scale has the ad- vantage that it adapts itself to the study of very much higher temperatures. The use of this process suggested by Pouillet was carefully studied by Regnault and has since become the most common method of defining temperatures practically. The expression for the laws of Mariotte and Gay-Lussac can be put in the form . ..... (2) by calling n the number of units of quantity (this unit may be either the molecular weight or the gram); R the value of the expression for unit quantity of matter taken at the temperature of melting ice and under atmospheric pressure. Gas Thermometers. The equivalent expressions (i) and (2), which arbitrarily by convention give the definition of temper- ature in terms of the elastic properties of a gas, may be utilized, from the experimental point of view, in various ways for the realization of the standard thermometer. i. Constant-volume Thermometer. In the thermometer desig- nated by this name, the volume and the mass are kept invariable. The expression (2) then gives between the two temperatures t and to the relation A, from which (3) STANDARD SCALE OF TEMPERATURES 15 2. Constant-pressure Thermometer. In this case the pressure and the volume of the heated mass remain constant, but the mass is variable; a part of the gas leaves the reservoir. The expression (2) then gives from which It would be much more logical, instead of the classic expressions constant- volume thermometer or constant-pressure thermometer, to say thermometer of variable pressure, thermometer of variable mass, which describe much more exactly the manner of their action. 3. Thermometer of Variable Pressure and Mass. The action of this apparatus combines those of the two preceding types. A part of the gas leaves the reservoir, and the pressure is not kept constant. The expression (2) gives i p _ n a _ po n i ' --Ho from which 4. Volumetric Thermometer. There exists a fourth method of the use of the gas thermometer which was suggested by Ed. Becquerel, and presents, as we shall see later, a particular interest for the evaluation of high temperatures. We keep the name for it given by its inventor. The determination of the temperature is obtained by two measurements made at the same temperature, and not as in the preceding methods by two measurements made at two different temperatures one of which is supposed known. T 6 HIGH TEMPERATURES The mass contained in the reservoir is varied, and the ensuing change of pressure is observed. The expression (2) gives (,-,') R from which This necessitates a preliminary determination of the constant R. In the particular case in which p' = o, which supposes that a complete vacuum is obtained, the preceding relation becomes simpler and is I i + <.| ....... (7) a n R The definitions of temperature given by these different ther- mometers would be equivalent among themselves and with that of the mercury thermometer if the laws of Mariotte and Gay-Lussac were rigorously exact, as used to be held, and if the expansion of mercury in glass were linear. The only advantage of the gas thermometer then would be to extend to high tempera- tures the scale of the mercury thermometer. In this way it was employed by Pouillet, Becquerel, and Samte-Claire-Deville. Experiments of Regnault. The very precise experiments of Regnault caused a modification in the then admitted ideas con- cerning the mercury thermometer as well as the gas thermometer, and led to the definite adoption of the gas thermometer as standard. In the first place these experiments established that different mercury thermometers are not comparable among themselves on account of the unequal dilatation of the differing glass employed STANDARD SCALE OF TEMPERATURES in their construction. Thus they cannot give an invariable scale for the determination of temperature. In comparing them from o to 100 they do not present between these extreme tempera- tures very great differences, 0.30 as a maximum, but at tempera- tures above 100 these differences may become considerable and reach 10 to 20 or more. (See also Chap. IX.) Constant-vol. Mercury thermometer in 0o=76o. Crystal. White glass. Green glass. Bohemian glass. 100 + 0.00 +0.00 +0.00 +0.00 150 + 0.40 O.20 +0.30 +0.15 2OO + I.2S 0.30 +0.80 +0.50 250 + 3-oo +0.05 + I-85 + 1-44 300 + 5-72 +1.08 + 3-50 350 + 10.50 +4.00 The numbers figuring in this table indicate the quantities by which it is necessary to increase or diminish the temperatures given by the air thermometer in order to have them correspond with those which were observed with the different mercury thermometers. It was thus impossible to define the practical scale of tempera- tures in terms of the mercury thermometer. The use of the gas thermometer became necessary. But Regnault recognized that it was not possible to take a single coefficient of dilatation a independent of the nature of the gas, of its pressure, and of the mode of dilatation utilized. The coefficient of expansion at con- stant volume ()8) and the coefficient of expansion at constant pressure (a) are not identical. This follows from the fact that the law of Mariotte is not vigorously exact; we have in reality pv = p v + c, e being a very small quantity, but not zero. The experiments of Regnault permitted him not only to detect but to measure this variation of the coefficient of expansion. Here are, for example, the results which he found for air between o and 100. 1 8 HIGH TEMPERATURES Volume constant. Pressure constant. Pressure ft i ft Pressure a I a 266 0.003656 273.6 7 6o 0.003671 272.4 7 6o 3655 272.8 2525 3694 270.7 1692 3689 271 2620 3696 270.4 36SS 3709 269.5 For air at 4.5 Rankine obtained, from the experiments of Regnault, the formula pv = p Vo + 0.008163 CO pv w being the atmospheric pressure. These coefficients vary also from one gas to another, as is shown by the following table, taken also from Regnault's experi- ments: MEAN COEFFICIENT BETWEEN O AND IOO. Volume constant. Pressure constant. Pressure. R i Pressure i mm. tt mm. Air, 760 0.003665 272.8 760 0.003671 272.4 3655 3709 269-5 2620 3696 270.4 Hydrogen. 760 3667 272.7 760 36613 273.1 2545 36616 273.2 Carbon Monoxide. 760 3667 272.7 760 3669 272.5 Nitrogen. 760 3668 272.6 Carbonic acid. 760 3688 271.2 760 3710 296.5 3589 3860 259 2520 3845 259.5 Sulphurous acid. 760 3845 259.5 760 3902 253.0 980 3980 251.3 These experiments show that the easily liquefiable gases have coefficients quite different from those of the permanent gases. For the permanent gases the coefficients for constant volume differ much less among themselves than those for constant STANDARD SCALE OF TEMPERATURES pressure; for the former the extreme deviation does not exceed ToVo 5 f r the latter it is three times as great. Setting aside air, which is a mixture and which contains more easily liquefiable oxygen, the coefficients for constant volume of H 2 , N 2 , and CO are identical. Finally, for hydrogen the coefficient of expansion does not vary appreciably with the pressure. The inequality of the coefficients of expansion, however, does not prevent us from taking any gas whatever to define the scale of temperature, provided we apply to it the proper coefficient determined by experiment between o and 100. The scales are identical, if the coefficients of expansion do not vary with the temperature. This is the conclusion to which Regnault came from a comparison of thermometers at constant volume, differing by their initial pressure or the nature of the gas. Here are the results obtained, starting from the fixed points o and 100, by the aid of the following formulae: pv = nRT, p Q v = nRT , J_ ICO p-po r- r r,oo - r r AIR THERMOMETER. P=75i mm. p=i486 mm. Degrees. Degrees. 156.18 156.19 2S9-50 259-4I 324-33 324.20 PRESSURE = 760 MILLIMETERS. Air thermometer. Hydrogen thermometer. Air thermometer. CO 2 thermometer. Degrees. Degrees. Degrees. Degrees. 141-75 141.91 I59-78 l6o.OO 228.87 228.88 267.35 267.45 325-40 325.21 322.8 322.9 20 HIGH TEMPERATURES The deviations do not exceed 0.2, a value that Regnault estimated not to exceed the limits of error of his experiments; he concluded from this that one gas may be used as well as another, and he took air for the normal thermometer. Nevertheless his experiments on sulphurous acid had shown a very marked variation of the coefficient of expansion of this gas with the temperature. The following table gives the mean coefficient at constant volume between o and t for this case : 98.0 0.0038251 102 .45 ' 38225 185-42 37999 257.17 379 2 3 299.90 37913 310.31 37893 By analogy it is permissible to suppose that a similar effect ohould take place with the other gases; but the differences were then too small, and the degree of precision of the methods of Regnault insufficient to detect it. Results Obtained by Chappuis. This effect has been demon- strated by experiments of very great precision made at the Bureau International des Poids et Mesures, at Sevres. Chappuis has found, between o and 100, systematic deviations between thermometers of hydrogen, nitrogen, and carbonic acid, filled at o under a pressure of 1000 mm. of mercury. Hydrogen ther. N ther.-H ther. N ther.-COj then - 15 -0.016 -0.094 oo o + 25 +O.OU +0.050 + 5 +0.009 +0.059 + 75 +O.OH +0.038 + 100 o o In this table, taking as definition of the temperature the hydrogen thermometer at constant volume, the numbers in the last two columns indicate the deviations observed with the thermometers of nitrogen and carbonic acid; it is certain that these deviations are systematic. These results allow of the determination of the mean coefficients of expansion: STANDARD SCALE OF TEMPERATURES 21 t (hydrogen) o ioo 0.00366254 /3 (nitrogen) /8 (carbonic acid) 0.00367698 0.00373538 367466 372477 Thus the coefficients decrease with rise of temperature, while remaining higher than that of hydrogen, to which they tend to approach. The more recent work of Chappuis and Harker and others in the establishment of a normal scale of tempera- tures for high temperatures will be discussed in the following sections. In the interval o to 100, the values given above, calculated from Chappuis' data of 1888, may not be absolutely exact, but they are probably very nearly correct. Some of the later results are given below; those marked Callendar are calculated by him from the data of Kelvin and Joule, using a modified formula; Chappuis' results are from his latest determinations (1902), while those of Lehrfeldt and Rose-Innes are calculations involving special thermodynamical assumptions. DIFFERENCE BETWEEN SCALES OF NITROGEN AND HYDRO- GEN THERMOMETERS. ~ tfa v l- = const., 100 cms. Temp. Cent. Callendar. 1903. Chappuis. 1902. Rose-Innes. 1901. Lehrfeldt. 1898. + 20 + .006 + .005 + .002 + .011 +40 + .009 + .008 + .002 + .017 +50 + .009 -j-.oio + .002 + .OIQ +60 + .008 + .009 + .002 + .019 -I- 80 + .005 + .004 + .001 + .015 Normal Scale of Temperatures. It results from these ex- periments that the different scales furnished by the various gas thermometers are not rigorously identical; the deviations between o and 100 are very small, but their existence is certain. It becomes necessary, therefore, in order to have a scale of tem- perature rigorously defined, to make a choice of the nature of the gas, of its manner of dilatation, and of its initial pressure. 22 HIGH TEMPERATURES The normal thermometer selected by the Bureau International des Poids et Mesures to define the practical scale of temperatures, and everywhere adopted to-day, is the hydrogen thermometer, operated at constant volume and filled with gas at 1000 milli- meters of mercury at the temperature of melting ice. For high temperatures this definition is inadmissible, because we would reach such pressures that the apparatus could not with- stand. The use of the method at constant volume, that is to say, at invariable mass, is besides bad on account of the per- meability of the coverings at high temperatures. It would be of great advantage to be able to employ a gas other than hydrogen and operate the thermometer at variable mass. Practically, it has been the custom, in most of the recent work at high tem- peratures, to use nitrogen gas at reduced pressure, 150 to 300 mm. of Hg at o C., although there has been, as yet, no formal agreement as to the gas or type of thermometer to use in defining the high temperature scale. In the actual state of experimentation at high temperatures, it has been impossible as yet to obtain results exact to better than i, and indeed, practically, we are far from arriving at this accuracy for the highest temperatures measureable. It is very likely that we can, under these conditions, employ indifferently for the construction of the normal thermometer any permanent gas whatsoever that does not diffuse into or through the contain- ing bulb. According to the preceding experiments, all the gases would have a dilatation slightly greater than that for hydrogen, and their coefficient of expansion, which decreases with rise of temperature, would approach that for hydrogen. For deter- mining experimentally the error possible with a normal thermom- eter thus modified, we possess the following experimental data. Crafts compared in the neighborhood of 1500 the expansion at constant pressure of nitrogen and carbonic acid, and found for this latter the mean coefficient 0.00368 in assuming 0.00367 for nitrogen. The experiments were made by displacing in a Meyer's tube nitrogen by carbonic acid, or carbonic acid by nitrogen. STANDARD SCALE OF TEMPERATURES 23 10 cc. N 2 displace 10 cc. CC>2 displace io.o3ofCO 2 9-95ofN 2 10.01 9-9 1 10.00 9-98 10.03 9-93 9-95 10.09 Mean 9.94 Mean 10.02 The two measurements give positive and negative differences of the same order of magnitude; but it should be noticed that the observed deviation (^innr on an average) hardly exceeds the possible error of observation. However it may be, carbonic acid, which differs much from the permanent gases at ordinary temperatures, no longer so differs in an appreciable degree at 1500. Violle made some comparative measurements on the air pyrometer used at constant pressure and constant volume in his determinations of the specific heat of platinum. Volume constant. Press, constant. Difference. . , 1171 . 1165 6 1169 II66 3 H95 "92 3 There was on an average a deviation of only 4 between the two modes of observation, and the greater part of this deviation should be laid to accidental variations of the gaseous mass resulting from the permeability of the coverings. Chappuis has made an exhaustive experimental study of the divergences of gases from the normal scale at comparatively low temperatures and he finds that the coefficient of nitrogen (at constant volume) gradually diminishes as above stated (p. 21), but that at about 75 C. it reaches a limiting value equal to = 0.00367330, and it may be assumed that above this temperature the gas is in a perfect state. The mean coefficient at constant volume for this gas between o and 1 00 is 180-100 = 0.00367466 24 HIGH TEMPERATURES and the limiting value for an initial pressure P = o is /3 Po=0 = 0.0036617. This follows from the divergence that Chappuis and Harker found for the constant-volume nitrogen-thermometer from the normal scale of temperatures, in terms of the initial pressure; their experiments gave aft = i.28.io~ 8 per mm. change in pressure. dp It is to be remembered that when the volume or pressure coeffi- cient is found for any pressure, that value is, by definition, the one to use in computing the normal scale of temperatures. The experiments of Chappuis and Harker were carried out at the International Bureau of Weights and Measures and in- cluded also a comparison of the platinum-resistance and nitrogen thermometers up to 500 C. and a determination of the sulphur boiling point, to which questions we shall return. Such a normal scale of temperature for the nitrogen thermom- eter is given by finding the coefficient 0, at o C. for a pressure PQ which the gas would have, supposing it to remain perfect in the range o to 100. If P = 100 cm., PI OO = 136.7466 cm.; whence JP ' = 100.0086 and = Pm ~ T f / = 0.00367348, if 100 P Aim = 0.00367330 as stated above. Nitrogen at constant pressure gives, according to Chappuis, 8a = 1.19.10 8 per mm. dp and . a p=Q = 0.0036612. The divergences from the normal scale for this gas are about double those at constant volume, and the divergences between the uncorrected scale and the theoretical scale of the constant- volume thermometer, whose constants are given above and which represents the normal scale of temperatures, are proportional to the temperature measured from 100 and have the following values : STANDARD SCALE OF TEMPERATURES At 100. 200 . 300 . 400 . o.ooo .023 .047 .070 These deviations are evidently very slight and are entirely negligible within this range for practically all pyrometric uses. We shall see, however, that at 1000 this correction may assume a certain importance. For hydrogen, the limiting values given by D. Berthelot are: /3 P=0 = 0.0036625, a p=0 = 0.0036624, and the deviations of this gas from the normal scale are imma- terial. The latest results of Chappuis on the elastic properties of the various thermometric gases are given in the following table: EXPANSION COEFFICIENTS OF THERMOMETRIC GASES ACCORDING TO CHAPPUIS. ioX Hydrogen. Nitrogen. Air. CO,. 367? o 3733 ^ 00 40 2671; 4 372O O 3662 the whole length of the two isotherms. Thus -C Po' Po" . / . / PI po or 7" = 3T^* pi Po 2 g HIGH TEMPERATURES Equation (2) then takes the very simple form that is to say, the ratio of the absolute thermodynamic temperatures is equal to the ratio of the absolute temperatures of the gas thermome- ter; and if on the two scales it is agreed to take equal to 100 the interval comprised between the temperatures of melting ice and the vapor of boiling water, we have, at any temperature, the equality But this is only a first approximation, for we have employed relations that are but roughly so: the laws of Joule, Mariotte, and Gay-Lussac. 2. Second Approximation. Reconsider the problem by a more exact method. Since T differs very little from - > and . + t since the laws of Mariotte and Gay-Lussac are nearly true, we write, following a method of calculation indicated by Callendar, being a very small function of p and of T (thermodynamic temperature). We have, then, between the temperature of the gas thermom- eter and the thermodynamic temperature, the relation which will permit of passing from one scale of temperature to the other if we know the corresponding value of <. Consider, as before, Carnot's cycle, and let us determine the heat of isothermal expansion in a more exact manner, by utilizing STANDARD SCALE OF TEMPERATURES 29 the experiments of Joule and Thomson on the expansion through a porous plug, and those of Regnault on the deviations from Mariotte's law. We write for this that the changes in energy between two given isothermal states are the same, either for the reversible expansion or for the expansion of Joule and Thomson. being the very feeble change in heat of the gas accompanying its passage through the porous plug, in the experiment of Joule and Thomson. We get from this ft = A I v dp + / -^ dp (at constant temperature), (3) Jtf J dp for d (pv) = p dv + v dp. The relation pv = RT (i - 0) gives for the value of v RT ( ^ = (i -<*>)> which, substituted in equation (3), leads to Similarly, we have If we introduce these values in the expression for Carnot's cycle, after division by TI and T Q we should find an identity: - p YI dp --i-- p TV dp > HIGH TEMPERATURES The law of adiabatic expansion gives ^7t77 = J > lo * pipo = o. In order, then, that the expression reduce to an identity it is necessary that Referring to the experiments on air of Joule and Thomson, we have = 0.001173- A> po being the atmospheric pressure, and T the temperature of melting ice. This is still an approximate result, for we have depended upon the experiments of Joule and Thomson and on the law of adiabatic expansion; however, the approximation is more close. If it seems sufficient for air, it is certainly not so for carbonic acid. Neither is the formula rigorously exact for air. Gas Scale Corrections. Callendar has calculated the correction to make to the air thermometer readings by extrapolation up to 1000, and he found the following results: Readings of Volume constant. Pressure constant. thermometer. A/ " O.OOII73 0.001173 100 0.000627 O 0.000457 2OO 393 0.04 225 0.084 300 267 0.09 127 O.2O 500 H7 0.23 52 0.47 1000 54 O.62 12 1.19 The deviations of the air thermometer at high temperatures are thus very slight if concordance is established at o and 100, and we have seen that in the case of nitrogen the experiments of Chappuis and Harker have shown the same to be true for this gas. STANDARD SCALE OF TEMPERATURES 31 Callendar, in a more recent computation based upon the work of Kelvin and Joule and the experiments of Chappuis and others, arrives at the following values for the scale corrections for the best thermometric gases: SCALE CORRECTIONS FOR GASES, ASSUMING 8 = 273.10. Constant pressure, 76 cm. Constant volume, p 1 = ioo cm. centigrade. Helium. Hydro- Nitro- Air. Helium. Hydro- Nitro- Air. gen. gen. gen. gen. - ISO +0.073 +0.084 +0-945 +0.901 O.O26 +0.013 +0.195 +0.186 IOO + .030 + .022 + .328 + -3H .OI2 + -005 + .080 + .076 - 50 f- .009+ .006 + .090 + .086 .004 + .OO2 + .024 + -023 20 + .003+ .002 + .025 + .024 .OOI + .OOO + -007 + -007 + 20 .OOl6 .OOOQ .0141 .0134 + .0008 . 0003 .0043 + .0041 + 40 .0022 .0013 -.0195 - .0186 + .0011 .0004 -.0059 + .0056 ' + 50 .0022 .0013 .0195 - .0186 + .0011 .0004 -.0059 + .0056 + 60 .0021 .0012 .0180 .0172 + .0011 .0004 - .0054 + .0053 + 80 - .0013 .0008 .0113 .0108 + .0007 .0002 -.0038 + .0034 + 15 + .0054 + .0029 + -043 + .041 .0031 + .0010 + .0143 + .0136 + 200 + .0128 + .0068 + .101 + .096 .0076 + .0024 + .035 + 033 + 300 + .0332 + .0165 + -243 + -232 -.203 + .0059 + .088 + .084 + 450 + .071 + .034 + -495 + .472 .047 + .013 + .189 + .180 + IOOO + 243 + .104 + 1-53 + 1.46 -.187 + .044 + .646 + .616 The above table indicates that for the gases hydrogen and helium no attention need be paid to the thermodynamic correc- tion, for it is quite negligible for the whole temperature range for these two gases. All the gases are also seen to have a greater correction at constant pressure than at constant volume. Again it is to be noted that at small initial pressures these corrections will be proportionally reduced, and finally that it is only in the most refined work that this correction need be applied, as in the establishment of a fixed point in pyrometry as the gold fusing- point. D. Berthelot has indicated a simple method for calculating this thermodynamic correction for any gas. For a constant- volume thermometer: 373 273 32 HIGH TEMPERATURES T being the absolute temperature of melting ice (273.10), T the absolute temperature sought corresponding to the centigrade temperature / given by the gas thermometer in question at an initial pressure of one atmosphere. For other pressures p the correction to t must be multiplied by *- For the constant-pressure thermometer T - T Q = t\i- The value of a depends upon the critical constants of the gas and is 27 2 T c 3 = 6 4 R '77' where R is the gas constant (here Y T c and p c the critical pressure and temperature respectively. TABLE OF CRITICAL CONSTANTS. PC (. a Carbonic acid 72 .9 atm. + 31.3 2.188 Oxygen 50.0 -118 0.422 Air '. 39.0 140 342 Carbon monoxide . . 3.e Q 141 .363 Nitrogen . 33 6 146 343 Hydrogen 13 o 240 .016 Helium 3 -268 .009 The formulae of Berthelot give practically identical values for the thermodynamic corrections as found by Callendar. Buck- ingham has discussed in detail the departures of the temperature scales, both constant volume and constant pressure, given by the several gases, from the thermodynamic scale by a method similar to that of Berthelot 's, but using a somewhat simpler equation of state. The most interesting results relate to the behavior of nitrogen, which is now generally used as the thermometric gas in high temperature measurements, and in Fig. i are given the corrections of the nitrogen thermometer at PQ = 1000 mm. of Hg taken from Buckingham's paper. STANDARD SCALE OF TEMPERATURES 34 HIGH TEMPERATURES It should be noted that the calculated corrections to reduce the readings of any gas thermometer to the thermodynamic scale are extrapolations from data on the Joule-Thomson effect made at ordinary temperatures. This is probably not a serious source of concern, however, as both Buckingham and Berthelot show that the several gases, when treated by the method of corre- sponding states, that is reduced in pressure and temperature to the fractions of their critical constants, furnish data all lying on the same curve. Experimental science has now reached such a development that as above stated these corrections to the thermodynamic scale cannot always be neglected. The Ice Point. The experiments of Kelvin and Joule may also be used to determine the absolute temperature of the point of fusion of ice on the thermodynamic scale. Below are the results of a computation by Lehrfeldt made several years ago. Gas-ther. Thermodyn. ther. Hydrogen ................. 273.08 272.8 Air ....................... 272.43 273.27 Nitrogen ................. 273 . 13 273 . 2 Carbonic acid ........ .... 268.47 I 2?4 ' 8 i ( 273.48 (Natanson) The thermodynamic temperature of melting ice should be in all cases the same; the deviations come mainly from the uncertainties in the measurements of the heat of expansion, indicating the desirability of repeating Joule and Thomson's work with modern appliances. There have been several more recent computations of the temperature of fusion of ice on the thermodynamic scale ( = -=-) based on the experimentally found deviations of several of the real gases from the ideal state, account being taken of the Joule-Thomson effect as measured by various observers, the thermal expansion and the compressibility as determined by Chappuis and by Amagat, the computations requiring the use of a modified form of Van der Waal's equation of state. Some of these calculations are as follows: STANDARD SCALE OF TEMPERATURES 35 THERMODYNAMIC TEMPERATURE OF MELTING ICE (0 ) Author. Gases used in com- putation. D. Berthelot (1903) H, CO 2 , Air Buckingham (1907) H, N, CO 2 , Air Rose-Innes (1908) 273.11 273.174 273.131 273-I36 The following table gives Callendar's resume of the expansive properties of the thermometric gases. In the table is the ther- modynamic temperature of the ice-point as determined from hydrogen, and T this point on the various gas scales. EXPANSION AND PRESSURE COEFFICIENTS FOR = 273.10. Gas. Constant pressure, 76 cm. Constant volume, p ioocm. e -T To i/r, e -T r i/r, Helium -f-o. 10 - -135 + .70 + .71 273.00 273.235 272.40 272.39 .0036628 .00365985 .0036708 .0036709 +0.19 + .067 + -99 + -96 272.91 273-034 272. ii 272.14 .0036640 .00366254 .00367466 .00367425 Hydrogen . . . Nitrogen Air Chappuis' latest values give in the case of hydrogen- = 273.038 a and - = 273.033 for zero pressure on the hydrogen scale, as computed by himself, showing no sensible difference in the two hydrogen scales in the range o to 100 C, and, taken with the preceding tables, that the hydrogen and thermodynamic scales differ by about 0.10 C. at o C. Our knowledge of the thermo- dynamic scale, as realized by correcting the several gas scales, may be said to be in a very satisfactory condition. As we shall see in the chapter on the laws of radiation, the normal or thermo- dynamic scale of temperatures may be extended to temperatures indefinitely high in terms of the intensity of radiation total or monochromatic, which proceeds from a small opening in any enclosure at constant temperature throughout. We shall have realized, therefore, a single standard or normal temperature scale 36 ' HIGH TEMPERATURES independent of the properties of any particular substance, con- tinuous from the absolute zero to the highest temperatures that may be produced, and one that is practically reproducible for all technical and scientific purposes, by methods that are available in the several standardizing laboratories. CHAPTER II. GAS PYROMETER. Introduction. We have seen that the standard scale of tem- peratures adopted by the International Committee of Weights and Measures is given by a certain constant-volume hydrogen thermometer, namely that of the International Bureau at Sevres, which instrument, however, has not been used to measure tem- peratures above 100 C. The type of gas thermometer which is to be considered standard for higher temperatures has not as yet been agreed upon by any authoritative body, but for reasons which we shall develop, the constant-volume nitrogen thermome- ter appears to have the preference, at least for temperatures above 200 C. From what we have seen in the preceding chapter, it is practically immaterial in the definition of the high temperature scale what form of thermometer is actually used, as the indications of any of the gas thermometers may readily be compared with those of another by well established methods of computation and reduced with great accuracy to a common theoretical basis, that of the thermodynamic scale. It may be well to recall, at this point, in what consists the actual operation of the location of a temperature on the chosen gas scale, and point out, at the same time, some of the difficulties involved. The gas thermometer bulb must be brought through- out its volume to a sufficiently uniform temperature. To obtain a volume of 500 c.c. of gas, for example, constant in temperature to i at 1000 C. has not yet been attempted by any experimenter. Whatever the system of gas thermometry used, on account of the transient nature of the phenomenon measured, pressure on a manometer, a mass of displaced mercury, etc., it is also neces- sary, except in certain special cases as some boiling points, to bring to this same temperature some other body whose registra- 37 38 , HIGH TEMPERATURES tions are more permanent, such as a mercury, platinum resist- ance, or thermoelectric thermometer, or rarely a metal at its melting point, and finally it is practically necessary to transfer the readings of the gas thermometer by means of this auxiliary thermometer to a series of fixed temperatures such as freezing and boiling points. The gas scale, therefore, is found in practice to be finally a discontinuous one, or at best represented by con- tinuous interpolations in terms of some empirical law, not the gas law. We shall see that there are further and very serious limitations in the attainment of great accuracy with the gas ther- mometer; thus, the space containing gas between the hot and cold parts is at an unknown average temperature; the expan- sion of the bulb with heat must be corrected for; and the bulb must be of sufficient rigidity and impermeability at the highest temperatures. The gas thermometer, as we have seen above, need not of necessity be used for the measurement of temperatures; it suffices to make use of it for the standardization of the different processes employed in the determination of temperatures, but a priori there are not on the other hand any absolute reasons for discard- ing it in cases other than these standardizations. Indeed, it has often been so employed, although, as we shall see, it is usually more convenient to make use of some other method in practice. We shall describe first the standard gas thermometer, and then discuss in considerable detail the factors that enter into the construction and theory of gas thermometers suitable for high temperatures, and give an account of several of the various investigations in gas thermometry, and finally call attention to the requirements for future work in this domain. Standard Gas Thermometer. This thermometer, that of the International Bureau of Weights and Measures at Sevres, France, is a constant-volume thermometer filled with pure, dry hydrogen, under the pressure of i m. of mercury at the tempera- ture of melting ice. It consists of two essential parts: the bulb, enclosing the invariable gaseous mass, and the manometer, serv- .ing to measure the pressure of this gaseous mass. GAS PYROMETER 39 The bulb is made of a platinum-indium tube whose volume is 1.03899 liters at the temperature of melting ice. Its length is 1. 10 m., and its outer diameter 0.036 m. It is attached to the manometer by a capillary tube of platinum of 0.7 mm. diameter. Fig. 2. Mounting of Thermometer Bulb. A diameter of 0.5 mm. is as small as can be allowed in the colder part of such capillaries on account of the lag in obtaining pressure equilibrium. This bulb is supported horizontally in a double box with interior water circulation. For the determination of the 100 mark, indispensable for standardization, the bulb can be placed in the same way in a horizontal heater supplied with steam and composed of several concentric coverings. Manometer. The manometric apparatus is mounted upon an iron support of 2.10 m. height, which is made of a railway rail firmly bolted to a tripod of wrought iron. The lateral parts attached to this rail, planed their entire length, carry sliding pieces to which are fastened the manometer tubes and a barom- eter. Fig. 3 represents, in a slightly modified form, the mano- metric apparatus. It is composed essentially of a manometer open to the air whose open arm A serves as cistern for a barom- eter R. The other arm B, closed half-way up by a piece of steel H, is attached to the thermometric bulb by the capillary tube of platinum C. The two manometer tubes, each of 25 mm. interior diameter, have their lower ends fixed into a block of steel S. They communicate with each other by holes of 5 mm. 40 HIGH TEMPERATURES diameter bored in the block. A stopcock E permits closing this connection. A second three-way cock F is screwed on the same block. One of its branches can serve to let mercury run out; the other, to which is attached a long flexible steel tube, puts the manometer in communication with a large reservoir of mercury Fig. 3. Manometer of Standard Thermometer. D which can be raised or lowered the length of the support, either rapidly by hand, or slow-motioned by means of a screw. The barometer which sets in the open branch is fixed at its upper part on a carriage G whose vertical displacement is regu- lated throughout a length of 0.70 m. by a strong screw. The latter is held at its two ends by two nuts which permit it to turn without longitudinal motion; it works in a screw attached to the GAS PYROMETER 41 carriage, and carries at its lower end a toothed pinion which works into a cogwheel. It suffices to turn this wheel by acting upon the rod which serves as axis in order to raise or lower the carriage with the barometer tube. This last has a diameter of 25 mm. in its upper part. The chamber is furnished with two indexes of black glass soldered to the interior of the tube at 0.08 m. and o. 1 6 m. from the end. The points of these indexes, convex down- wards, coincide sensibly with the axis of the barometric chamber. The part of the barometer which fits into the open manometer arm has a diameter greater than o.oi m., and ends below in a narrower tube curved upwards. The piece of steel which ends the closed arm at H is adjusted to this tube, leaving between itself and the tube but a very slight space, which is filled with sealing wax. It rests upon the upper rim of this tube, to which it is besides pressed by leather washers tightly screwed up. At its lower end it terminates in a perfectly smooth polished plane, which is adjusted to be horizontal. In the middle of this surface, near to the opening of the canal which prolongs the joining tube, there is fixed a very fine platinum point, whose extremity, meant to be used as a reference mark, is at a distance of about 0.6 mm. from the plane surface. Above this piece is a tube B of 25 mm. interior diameter, open above and connected below to the open arm of the manometer. Since the measurement of a column of mercury is more easily made and with greater precision when the menisci whose dif- ference of level it is desired to find are situated along the same vertical, the barometer tube R is bent so as to bring into the same vertical line the axis of the closed arm of the manometer and that of the barometer. Under these conditions, the communi- cation between the two manometer arms A being established through E, the total pressure of the gas inclosed in the reservoir of the thermometer is given by the difference of level of the mercury in these superposed tubes B and R. The measurement of the pressures is made by means of a cathetometer furnished with three telescopes, each of which is provided with a micrometer and level. The micrometer circle is 42 HIGH TEMPERATURES divided into 100 parts; at the distance from which the manometer is read, each division of the circle corresponds to about 0.002 mm. The method adopted for the measurement of pressures consists in determining the position of each mercury meniscus in terms of a fixed scale, hung near the manometer tubes, at the same distance as these latter from the telescopes of the cathetometer. One of the principal difficulties arising in the measurement of pressures is that of the lighting of the menisci. The method employed by Chappuis consists in bringing up to the surface of the mercury an opaque point until its image reflected by the mercury appears in the observing telescope at a very small dis- tance from that of the point itself. These two images being almost in contact, it is easy to set the micrometer cross-wire midway between them, at the precise point where would be the image of the reflecting surface. In order to have a very sharp image of the point, it is well to illuminate from behind by means of a beam of light passing through a vertical slit. The point and its image then stand out black on a bright background. The use of a stylus of black glass is preferable to that of a steel point on account of unchangeableness and of the greater sharpness of the edges. The method with stylus cannot be advantageously employed except in wide tubes, where the reflecting surface of the mercury which aids in the formation of the image does not have a sensible curvature. Dead Space. This consists of the space occupied by the gas : (i) in that part of the capillary tube which does not undergo the same variations of temperature as the thermometric bulb; (2) within the piece of steel forming the plug which caps the closed arm of the manometer; (3) in the manometer tube between the mercury and the horizontal plane in which ends the piece of steel. The mercury is supposed to just touch the stylus serv- ing as reference mark. The capacity of the capillary tube has been determined by mercury calibration; it was found equal to 0.567 c.c. The length of the capillary tube being i m., if we deduct from this capacity GAS PYROMETER 43 that of 3 cm. of the tube which are exposed to the same tempera- tures as the reservoir, that is 0.015 c.c., this leaves 0.552 c.c. The capillary tube fits for a length of 27 mm. into the piece of steel serving as plug. The total thickness of this plug is 28.3 mm.; thus the portion of the canal included between the end of the capillary tube and the lower face of the plug is 1.3 mm. in length. As its diameter is 1.35 mm., the capacity of this canal is 0.0019 c.c. The space included between a cross section of the manometer tube passing through the stylus and the plane surface of the plug is 0.3126 c.c. To have the total volume occupied by the gas it is necessary to add as well to this space the volume of the depressed mercury in the manometric tube caused by the curvature of the meniscus. The radius of this tube being equal to 12.235 mm., we find for this volume 0.205 c.c. We thus have as the total of the dead space the sum of the following volumes : C.c. Capacity of capillary tube 0.5520 Volume of canal in the plug 19 Capacity of the manometer tube between the stylus and the plane 3126 Volume of depressed mercury 2050 Total dead space i .0715 When the mercury does not just touch the stylus, we shall have to add to this value 0.4772 c.c. per millimeter separation of the stylus from the top of the meniscus. The expansion of the metal of the bulb was measured by Fizeau's method; this volume was found to have at different temperatures the following values: liters. 20 o 20 40 60 80 TOO .03846 .03899 .03926 .04007 .04061 .04117 .04173 44 HIGH TEMPERATURES The variation of the capacity of the bulb due to changes of pressure has also been studied; per millimeter of mercury it is 0.02337 nim. 3 ; or For o mm o mm: 3 ' 100 .2.3 ' 200 4-7 ' 300 7.0 ' 400 9-3 The zero is verified from time to time by bringing the bulb to the temperature of melting ice; there is absolute constancy even after heating to 100. The deviation is at the most 0.03 mm. for a pressure of 995 mm. Chappuis made a most careful calibration of four mercury in verre dur thermometers in terms of this standard gas thermometer in an apparatus such as shown in Fig. 2, and these mercury thermometers, with copies that have been made and distributed, represent to-day the practical standards of temperature in the interval 35 to + 100 C., with an accuracy of about 0.002 C. After a discussion of the formulae involved, we shall consider the question of the experimental establishment of the high temperature scale, a problem which has occupied a great many able investigators for many years, and which is by no means as yet conclusively solved, there being, as we shall see, embarassing outstanding uncertainties in determinations of temperatures, for example, of 0.5 at 500 C. and some 20 at 1600 C., due wholly to experimental difficulties. Formulae and Corrections. To illustrate the principles in- volved we shall cite as examples some of the earlier work with porcelain bulbs. As we shall see later, all of the errors here discussed have been greatly reduced in magnitude in the latest work with quartz and metal bulbs. i. Thermometer at Constant Volume. We must now render more precise the formula of the gas thermometer given in the preceding chapter by taking account of the variations of volume of the bulb, of the surrounding air temperature which changes the density of the mercury, and finally of the volume of the dead space. GAS PYROMETER 45 We have three series of observations to make in order to determine a given temperature: PoF = n RT , (i) (2) . . (3) Putting the first two series serve to determine - It is preferable, except in researches of very great precision, to take - from previously obtained results, and not to make the observations at 100, unless one does so to check his experimental skill. Dividing the third equation by the first, we have the relation PV ffApF nRT nT (} r> T7 ~ ~ E7 A 17 i J?T > ~ T> y ' ' * ' \4/ z o ' LlQi\V o flQ\.J. o where H and H are the heights of mercury, A and A the den- sities of this metal. For a first approximation let us neglect the differences between V and FO, n and n , A and A . We shall have then an approxi- mate value T' for the temperature sought: 7* - l . (^ r ' ' (5) a for Let us find now the correction dT to T' to obtain the exact temperature. In order to find this, take the logarithmic dif- ferential of (4) : dT AO AQ A = A -[l - k(k -ti)l k = 0.00018(^2 4), ^ = - 0.00018 (k - h). AO dV V - FQ F = Fo V = F [l + '(r- To)], &' (porcelain) = 0.0000135, - = 0.0000135 (r - r ), ^0 by neglecting the variations of volume of the bulb due to changes of pressure. _ dn _ x Zn Au Pd c.c. Holborn and Day j 286 276 80 Pt 20 Ir 90 Pt 10 Ir 208 196 .0042 .0046 ( 2 43 3 to 10 (419.0 1 0.5 1064.0 1.0 }.... Jaquerod and ( Perrot ( 195 to 230 | Quartz glass 43 .0180 70 3 2 I 1067.2 1.8 J.... Holborn and 1 Valentiner ) 147 137 80 Pt .20 Ir Iridium 208 54 0042 022 20 ( 90 to ( 110 35) J30J 3 to 60 | 1575 1 10 Day and Sosman . | 217 to 347 J 80 Pt 20 Rh 206 OOI5 3 to 6 4 6 I (418.2 1 0.3 1062 4 0.8 1549.2 2.0 Suggestions for Future Experiments. It is perhaps easier to criticize than execute experiments of precision, but from what has been said above it is evident that there is still need for more work with the gas thermometer before the high-temperature scale is established in an entirely conclusive manner. Thus, the outstanding uncertainty of nearly 0.5 at the sulphur boiling point should be eliminated; and while there is good agreement, better than 5 at 1100 C., a difference of 25 between observers exists at the palladium melting point (1550 to 1575) ; and at tem- peratures to which it now seems hopeless to extend the gas scale directly, this range of uncertainty increases, becoming about 100 at 3000 C., or at the melting point of tungsten. Methods. The constant- volume method has been preferred by almost all experimenters who have worked at high tempera- tures, and the results by this method also have smaller correc- tions to reduce to the thermodynamic scale. For the lower range of temperatures at least, in view of the outstanding discrepancies, it would be well to adopt the same instrument for use both at constant pressure and constant volume. To as high tempera- GAS PYROMETER 8l tures as possible, the bulb should be immersed in stirred liquid baths to insure uniformity of temperature; and in this range of temperatures, or to perhaps 900, the transfer of the gas scale can probably be made with the greatest accuracy by means of platinum-resistance thermometers, the wires of which can be made to integrate very exactly the bulb temperature.* The volu- menometric method has not been used in any recent work, although it appears to possess the smallest instrumental cor- rections. It labors under the disadvantage of having an uncer- tain thermodynamic correction, but this largely disappears at high temperatures, where the outstanding uncertainties largely exceed this small correction. It would, therefore, be worth while carrying out new experiments by this method, especially at very high temperatures. The method of Crafts and Meier (p. 83) is also worthy of further study at high temperatures. The work of Day and Sosman shows that, for the constant-volume ther- mometer, the deformation of the bulb may be eliminated, and the error due to the dead space reduced to an almost negligible amount. Their work also shows the importance of an exact determination of the coefficient of expansion of the bulb and an exact adjustment of temperature over it by the use of properly designed electric furnaces. The uncertainty in the temperature of the manometric parts of the apparatus gives rise to an appre- ciable error which may be eliminated in future work by water- jacketing. The Bulb. All recent work has shown the superiority of the met^al bulb when its coefficient of expansion is carefully deter- mined. The alloy 80 Pt 20 Rh is the material that has been used so far which best suits all the requirements for temperatures to 1600 C, namely, rigidity, impermeability, regular expansion, and small contaminations of the auxiliary temperature apparatus. The best form appears to be cylindrical with a reentrant tube. It may be possible to find refractory earths which are sufficiently impermeable to use with some modification of the Crafts and * Both of these improvements and others have been introduced by Holborn and Henning to 450 C. since the above was written. 82 HIGH TEMPERATURES Meier method at very high temperatures; or possibly metallic tungsten or one of its alloys may be adopted for use in a suitable atmosphere for pushing the gas scale to the highest limits. In all cases it is desirable to have the volume of the bulb as great as possible consistent with uniform temperature distribution. To 500, there should be no difficulty in using a 500 c.c. bulb. The Gas. Nitrogen has proved satisfactory in every respect, and this gas will probably be continued in use, although there would be some theoretical advantage, at least for the higher temperatures, in substituting one of the monatomic inert gases, such as argon or helium. It is questionable whether it is worth while to attempt to ex- tend the use of the gas thermometer above 1600 C., as the con- stants of the laws of radiation can be exactly determined in this range, and the radiation laws are eminently suited for extrapola- tion, as they give the thermodynamic scale directly. The Manometer. It would be well to eliminate the some- what troublesome and uncertain reference to a variable pressure, that of the atmosphere, by the entire elimination of the barom- eter. This can be done, whatever the type of gas thermometer used, by evacuating to zero pressure the space above the manom- eter column and sealing off the manometer tube provided with a suitable globe or bulb at the top.* Variations in the tempera- ture of the mercury columns of the manometer may be com- pletely eliminated by water-jacketing. The errors of the ma- nometer are then easily made negligible compared with those of expansion of the bulb, the temperature distribution over it, and the transfer to the comparison thermometer. In general, it may be stated that it is not worth while to carry out any further gas thermometer experiments unless the utmost precautions are taken to assure the highest accuracy possible with modern appliances. Industrial Air Pyrometers. There have been attempts to construct air thermometers suitable for industrial usage, the argument sometimes being advanced that a gas pyrometer is * See note, p. 81. GAS PYROMETER 83 per se better than any other. As we have seen, however, there is probably no physical instrument which is more difficult to employ satisfactorily, and any seeming gain in making direct use of an air thermometer for industrial use is wholly illusory. Other evident objections are fragility, uncertain correction due to the dead space, and the development of small and often unperceived leaks. Furthermore, an empirical calibration is necessary, so that such an instrument does not carry the gas scale about with itself. Among the instruments that have been considerably used is Wiborgh's air pyrometer, shown in Fig. 14. A lens-shaped WIBOR6H AIR PYROMETER Fig. 14. reservoir V' is open to the air before an observation is taken, but when a temperature is to be read this lens is closed to the outer air and collapsed by a lever Z,, thus adding a definite mass of air to the bulb V of the thermometer; the resulting pressure is transmitted to a dial as in an aneroid barometer; provision is made for automatically correcting for variations in the pressure and temperature of the atmosphere. The Bristol Company have also made industrial forms of gas pyrometer. Indirect Processes. We shall place in this list various experi- ments in which the laws of the expansion of gases have been used only in an indirect way, or have been extended to vapors. Method of Crafts and Meier. It is a variation of the method of H. Sainte-Claire-Deville and Troost, consisting in removing the gas by means of a vacuum. Crafts and Meier displaced the 84 HIGH TEMPERATURES gas of the pyrometer by carbonic acid or hydrochloric acid, gases easily absorbable by suitable reagents. Hydrochloric acid is the more convenient, for its absorption by water is immediate; but there is to be feared at high temperatures its action on the air with formation of chlorine ; it is preferable to employ nitrogen in place of air. The apparatus (Fig. 15) consists of a porcelain bulb, whose inlet is large enough to let pass the entrance tube of the gas, which reaches to the bottom of the bulb. This arrangement increases considerably the in- [\ p fluence of the dead space and consequently diminishes the precision of the determina- tions. This method is especially convenient for observations on the densities of vapors which are made by the same apparatus; it then allows of having an approximate idea of the temperatures at which the experiments are made. Crafts and Meier have in this way deter- Fig. 15. Method of ^^(1 t h e variations in density of iodine Crafts and Meier. vapor as a function of the temperature. Regnault had previously proposed a similar method, without, however, making use of it. 1. One fills with hydrogen an iron vessel brought to the temperature that one desires to measure, and the hydrogen is driven out by a current of air; at the outlet of the metallic reservoir the hydrogen passes over a length of red-hot copper, and the water formed is absorbed in tubes of sulphuric acid in pumice stone and weighed. This method, very complicated, is bad on account of the permeability of the iron at high tempera- tures. At the same time, he proposed the following method: 2. An iron bottle containing mercury is taken; the vessel, being incompletely closed, is heated to the desired temperature and then allowed to cool, and the remaining mercury is weighed. GAS PYROMETER 85 This method is also defective on account of the permeability of iron at high temperatures; the hydrogen of the furnace gases can penetrate to the inside of the recipient and drive out an equivalent quantity of mercury vapor. Methods of H. Sainte-Claire-Deville. i. This savant tried in the first place to measure temperature by a process analogous to that of Dumas' determination of vapor densities. He took a porcelain bulb full of air, and heated it in the inclosure whose temperature was wanted, and sealed it off by the oxyhydrogen flame. He measured the air remaining by opening the bulb under water and weighing the water that entered, or else he determined merely the loss in weight of the bulb before and after heating. Observations taken on the boiling point of cadmium gave 860. 2. In a second method, which has the advantage of replacing the air by a very heavy vapor, Deville returned to the idea of Regnault, consisting in utilizing the vapor of mercury; but he ran against a practical difficulty. He had replaced the permeable iron recipients by porcelain recipients; the mercury condensed in the neck of the pyrometer and fell back in cold drops which caused the bulb to" break. For this reason he abandoned mercury and replaced it with iodine; the return of a cold liquid was completely obviated by reason of the nearness of the boiling point of this substance (175) and its fusing point (i 13). A large number of observations were made by this method; the boiling point of zinc, for example, was found to be equal to 1039. This method is quite faulty, as the iodine does not obey the laws of Mariotte and Gay-Lussac. The vapor density of this substance decreases with rise of temperature, this effect being attributed to a doubling of the iodine molecule. This fact was established by Crafts and Meier and confirmed by Troost. Method of D. Berthelot. All the preceding methods are limited by the difficulty of realizing solid envelopes resisting temperatures higher than 1600. D. Berthelot has devised a method which, at least in theory, may be applied to any tempera- tures, however high, because there is no envelope for the gas, or 86 HIGH TEMPERATURES at least no envelope at the same temperature. It is based on the variation of the index of refraction of gaseous mass heated at constant pressure; the velocity of light depends upon the chemical nature and the density of this medium, but is indepen- dent of its physical state. A gas, a liquid, or a solid of the same chemical nature produces a retardation of the light dependent only upon the quantity of matter traversed; this law, sensibly true for any bodies whatever, should be rigorously exact for substances approaching the condition of perfect gases. This retardation is measured by the displacement of interference fringes between two beams of parallel light, the one passing through the cold gas, the other through the hot gas. In reality, Berthelot employs a null method; he annuls the displacement of the fringes in changing at constant temperature the pressure of the cold gas until its density is equal to that of the gas in the warm arm which is at constant pressure. There is a difficulty arising from the necessity of separating the light into two parallel beams, then reuniting them without im- parting a difference of phase which renders the fringes invisible with white light. This is done in the following way (see Fig. 16) : A beam of light ab falls on a mirror MM' , which breaks it up into two parallel beams, bf and cd\ in order to separate the beams so as to be able to place apparatus conveniently with respect to them, a prism P gives to the beam bf the direction gh\ one can thus secure a separation of 92 mm. A second prism PI brings the beam cd into Im, and after reflection from a second mirror, MiMi, the fringes are observed in a telescope focused for parallel rays. The tubes containing the gases are placed at T and TV It is evidently necessary that the prisms P and PI be perfectly made. A preliminary adjustment is made with yellow light, then it is perfected with white light. The tube at variable pressure is closed by two pieces of plate glass, as is also the warm tube; these four plates should be abso- lutely alike. The warm tube is heated by a vapor bath at low temperatures, by an electric current passing through a spiral at high temperatures. GAS PYROMETFJl 87 But there is a difficulty in that in the warm tube there exists a region of variable temperature between the warm zone and the cold atmosphere. To eliminate the influence of this variable zone, there are inside the warm tube two tubes containing running cold water, whose distance apart can be changed; it is assumed that the variable region remains the same, and that distance between the two tubes gives the warm column actually utilized. It follows that T 4 d . <& m \ 9 J p 1 9 I* : - Tfc Fig. 1 6. D. Berthelot's Method. the comparative lengths of the warm column and of the cold column (this latter remaining constant) are not the same; the formula to be used will be somewhat more complicated. n being the index of refraction of a gas and d its density, we have n i = kd. In the constant-pressure tube *L =L. do po To obtain the invariability of the fringes, it is necessary that (HI n ) L = (n f n ) I, 88 HIGH TEMPERATURES L being the length of the cold tube, and / the displacement of the warm tube. T / J 7 \ J- I /If 1 \ 1 K \Q,\ UQ) Li = K (d ttQ/ '> ifc-V-- an expression which gives a relation between the pressures and the temperatures. This method, employed for the control of the boiling points, has given the following results: Pressure Temperature Temperature observed. calculated. Alcohol 74i.5mm. 77.69 77.64 Water 740. i 99.2 99.20 Water 761.04 100.01 100.01 Aniline 746 . 48 183 . 62 183 . 54 Aniline 760.91 184.5 184.28 Berthelot has standardized by the same method thermocouples which he used to determine the fusing points of silver and gold, and the boiling points of zinc and cadmium: Silver, freezing 962 C. Gold, freezing 1064 Zinc, boiling 920 Cadmium, boiling 778 The numbers found are nearly identical with those which result from the best determinations made by other methods. We shall discuss further the determinations of fixed points in pyrometry in Chapter XI. CHAPTER III. CALORIMETRIC PYROMETRY. Principle. A mass m of a body, brought to a temperature T, is dropped into a calorimeter containing water at a temperature /o. Let ti be the final temperature of water and substance. M being the water equivalent of the substances in contact (water, calorimetric vessel, thermometer, etc.) which are raised from t Q to ti, L t the heat required to warm unit mass of the body from ti to r, we have L T t X m = M(h - t ). Taking as origin of temperatures the zero of the centigrade ther- mometer, the heat required to warm unit mass of the body to the temperature T will be The quantity L is easy to calculate, because the specific heats at low temperatures are sufficiently well known: The expression for the total heat becomes m ti and /o are the temperatures given by the direct readings of the thermometer. The value of the second member is thus wholly known, and consequently that of the first member which is equal to it. If previous experiments have made known the value of the total heat LQ for different temperatures, one may from the knowledge of LQ determine the value of T. It will be sufficient to trace a curve on a large scale whose ordinates are temperatures, and 89 90 HIGH TEMPERATURES N \ \ \ \ V i\ \ \ s. s . \ \ \ \ \ \ _> \ s \ \ \ \ \ \ \ \ \ \ > \ \ \ \ it \ \ \ \ s. \ \ \ \ s\ r \ \ \ \ \ \ V \ \ \\ \ s. \ \ \ TOTAL HEAT IN CALORIES vs TEMPERATURE CENTIGRADE for PLATINUM COPPER NICKEL IRON X N * \ \ \ s \ s \ \ \ x \ s ,Z \ \ \ N \ \ >s \ \1 \ \ ^ ^S, \ ^ z X \ ^ CALORIMETRIC PYROMETRY 91 abscissas total heats, and to find upon this curve the point whose abscissa has the value given by the calorimetric experiment. In Fig. 17 are given curves of temperature in terms of total heat from o C. for the several metals used in specific heat py- rometry. The values of the total heats are the means, for each metal, of the experimental results cited in the several tables which follow. Choice of Metal. Four metals have been proposed : platinum, iron, nickel, and copper. Platinum. This metal was first proposed by Pouillet, and taken up again by Violle. It is much to be preferred to the other metals; its total heat has been compared directly with the indi- cations of the gas thermometer. This metal can also be repro- duced identical with itself. Iridium, which commercial platinum often carries, has about the same specific heat. The high price of these substances is an obstacle to their use extensively in works ; for a calorimeter of a liter it is necessary to have at least loo grm. of platinum, or $100 in a volume of 5 c.c., easily lost or made away with. Violle determined the total heat of platinum from o to 1 200, and computed it by extrapolation to 1800. W. P. White has determined the specific heat of platinum to 1500, obtaining somewhat lower values than Violle. The differences cannot be accounted for by differences in tempera- ture scales. A few measurements by Tilden to 600 give values between the others, and Plato at 600 and 750 finds values agreeing closely with White's. TOTAL HEAT OF PLATINUM FROM o C. IN CALORIES. Temperature. Violle. White. Temperature. Violle. White. IOO 3-23 IOOO 37-70 35-45 2OO 6. 5 8 I IOO 42.13 39-32 300 9-95 9-49 1200 46.65 43-28 4OO 13-64 13.09 1300 (5I-35) 47-26 500 17-35 16.75 1400 (56.14) 49-22 600 21. 18 20.44 1500 (61.05) 55-20 700 25-13 24-15 1600 (66.08) (59-26) 800 29.20 27.88 1700 (7I-23) (63.45) QOO 33-39 31-63 1800 (76.50) 92 HIGH TEMPERATURES White's measurements on mean specific heat satisfy the equation 0.03193 + 3.4 io~ 6 t. Iron. Regnault, in an investigation made for the Paris Gas Company, had proposed, and caused to be adopted, iron, in attributing to it a specific heat of 0.126, instead of 0.106 at o. He used a cube of 7 cm. sides which was thrust into the furnaces by means of long iron bars. The calorimeter was of wood and had a capacity of 4 liters. Various observers have determined the total heat of iron; at high temperatures the accord is not perfect among the results, as shown in the following table: TOTAL HEAT OF IRON FROM o C. IN CALORIES. Temperature. Pionchon. Euchene. Harker. Oberhoffer. weiss an Beck. IOO ii .0 II .O 200 22.5 23.0 23.0 23-4 23.1 300 36.8 37-0 37-0 37-5 36.1 400 51-6 52.0 51-3 52.4 49-5 500 68.2 69-5 66.9 66.0 64.4 600 87.0 84.0 83-8 84.6 81.2 700 108.4 106.0 104.1 II3-4 IOI.O 800 135-4 131.0 127.8 135-2 124.2 900 157-2 151-5 148.0 152.1 148.5 IOOO 170.9 173-0 155-7 167.0 I IOO 168.8 182.6 1200 199.2 1300 215.8 1400 232.4 1500 250.5 The determinations of Pionchon and of Euchene are in terms of incorrect temperature scales, at least above 800 C., and al- though those of the other observers are in terms of approxi- mately the same scale, the one ordinarily used to-day, this agree- ment is far from satisfactory. Oberhoffer's results show an abrupt change in specific heat beginning at 650, and changes in specific heat corresponding to the allotropic forms of iron. The results of Weiss and Beck show an abrupt change in specific heat at 750, corresponding to the magnetic transformation point. According to Oberhoffer and Meuthen, the addition of carbon to iron increases the specific heat in the proportion of o.oon per each 0.5 per cent carbon added, at least for the temperature range o to 650 C. CALORIMETRIC PYROMETRY 93 In spite of its common use for this purpose, this metal is not at all suitable for calorimetric use, by reason in the first place of its great oxidability. There is formed at each heating a coating of oxide which breaks off upon immersion in water, so that the mass of the metal varies from one observation to the next. Besides, iron, especially when it contains carbon, possesses changes of state accompanied during the heating by a marked absorption of heat. By cooling in water, hardening takes place, which may irregularly prevent the inverse transformations. The use of electrolytic iron is therefore preferable, since the most marked transformation and the one at the lowest temperature is thus avoided, and the oxidation is less. Nickel. At the Industrial Gas Congress in 1889 Le Chatelier proposed nickel, which is but slightly oxidizable up to 1000, and which above 400 does not possess changes of state as does iron. The total heat of nickel has been determined by Pionchon, by Euchene, and by Weiss and Beck. The differences are due very probably in part to impurities that the nickel may contain, as well as to experimental and temperature-scale uncertainties. TOTAL HEAT OF NICKEL FROM o C. IN CALORIES. Temperature. Pionchon. Euchene. Weiss and Beck. 100 II .O 12 .O 200 22.5 24.0 23.1 300 42.0 37-0 36.2 4OO 52.0 50.0 50.0 500 65-5 63.5 63-2 600 78-5 75-0 76.6 700 92.5 90.0 90.0 800 107.0 103.0 104.9 9 00 123.0 II7-S IOOO I38-S 134-0 Copper is sometimes used, and although when pure it appears to possess no transformation regions, it oxidizes and scales very readily and cannot be used to as high temperatures as any of the 94 HIGH TEMPERATURES other metals proposed. In the following table are given values of the total heat of copper as computed from the experiments of Le Verrier and of Frazier and Richards. TOTAL HEAT OF COPPER FROM o C. IN CALORIES. Temperature. Le Verrier. Frazier and J. W. Richards. 100 2OO 10.4 20.8 9 .6 IQ-5 300 31.2 29.8 400 42.2 40.4 500 600 54-7 66.5 51-4 62.8 700 800 77-8 91.0 74-4 86.5 900 IOOO 103.8 115.6 99.0 111.7 Calorimeters. In laboratories a platinum mass is often employed with Berthelot's calorimeter, a description of which is given in various publications on calorimetry (Fig. 18). The thermometer used for the meas- urement of the rise in temper- ature should be very sensitive, so that a rise of from 2 to 4 be sufficient in order to render neg- ligible the cooling correction. If use is made, for instance, of a thermometer giving the hun- dredth of a degree, the mass of platinum should be about one- twentieth the mass of the water in the calorimeter. A form of water-inclosed cal- orimeter, with furnace, such as Fig. 18. Berthelot's Calorimeter. used by White in specific heat determinations, is shown diagram- matically in Fig. 19. This method of operation is also appli- cable to temperature estimations. The water cover is swung aside when the platinum mass is dropped into the calorimeter. CALORIMETRIC PYROMETRY 95 This type of calorimeter is to be preferred in exact calorimetric work where high temperatures are involved, as the uncertainties of radiation and evaporation are reduced to a minimum. The usual mercury thermometer may be replaced to advantage by some form of electric thermometer. Industrial Calorimeters (Fig. 20). In the arts, where the measure- ments are made with less precision, and where it is necessary to con- sider the cost of installation of the apparatus, nickel may be made use of, a thermometer giving tenths of a degree, and zinc calorimeter, which may be home-made. Such an in- stallation may cost as little as $5. A mass of nickel should be used equal to one-twentieth of the mass of water of the calorimeter. Fig. 20. Industrial Calorimeter. Fig. 19. White's Calorimeter. The calorimeters used by the Paris Gas Company are after the Berthelot pattern; they are also water- jacketed. Such an apparatus may consist of a cylindrical calorimeter A of two liters capacity, of zinc or of copper; a double cylindrical jacket B of the same metal, containing water, and which may be surrounded by felt on the outside. The calorimeter rests on this 9 6 HIGH TEMPERATURES n jacket by means of a wooden support C. There is preferably a metallic cover in good contact Fig. 21. Metal with the outside vessel. A thermometer grad- Camer. uated to fifths of a degree, having a small but quite long bulb, serves as stirrer. The thermometric substance is a piece of nickel of mass equal to one-tenth that of the water, or 200 grm., so as to have considerable rise of temperature easy to read by the workmen who make the measurements. As a general rule, one must avoid placing the thermometric substance upon the floor of the furnace. The piece of nickel, which is made in the form of small cylinders having from 15 to 25 mm. diameter and from 10 to 30 mm. length, rests so as to be insulated from the floor in a nickel crucible pro- vided with a foot and with two arms attached somewhat above the center of gravity. When it has been heated for a half-hour, an observer takes out the crucible with a forked rod, and another seizes this crucible with tongs to empty it into the calorimeter. Use is not made of an iron crucible because this metal oxidizes and lets drop scales, which falling into the cal- orimeter would vitiate the experiment. Fig. 21 shows a suitable arrangement for containing a nickel cylinder. Siemens Calorimeter. A convenient form of direct-reading calorimeter due to Siemens is shown in Fig. 22. Using Fig ' "' Siemens Calorimeter - always the same mass of water and a ball of given mass and CALORIMETRIC PYROMETRY 97 kind, the thermometer or an auxiliary scale may be graduated to read directly the temperature attained by the heated ball. Hollow copper cylinders are usually furnished with this appa- ratus. Precision of Measurements. Biju-Duval made a series of experiments to study the sources of error arising from the use of the industrial calorimeter by comparing its indications to those of the thermoelectric pyrometer of Le Chatelier. The observa- tions were taken by varying the following conditions: Use of thermometer graduated to | or to -$. Use of the old wooden gas-works calorimeter or of the water- jacketed calorimeter. Use of iron or nickel. I. Experiment. Old wooden gas-works calorimeter. Iron. Thermometer in fifths. P = 10,000 grm. p = 1031 grm. /o = 20.8 /i = 36-2 Qo = J 53-5 cal. Computed temperature: Mean specific heat of iron = 0.108 t = 1420 Mean specific heat of iron =0.126 / = 1210 Totai heat according to Biju-Duval / = 915 Thermoelectric pyrometer = 970 It is thus evident that the mean specific heats even with the correction suggested by Regnault give temperatures much too high. With the curve of total heats the temperature found is much too low on account of the following losses of heat: 1. Absorption of heat by the wooden walls; 2. Radiation from the iron cube during transfer; 3. Cooling of the water in the calorimeter, whose temperature exceeded by 16 the temperature of the surroundings. The following experiments were made with the thermometer 98 HIGH TEMPERATURES reading to -fa -, the piece of nickel was protected against radiation by a crucible. The two calorimeters were compared. II. Trial with the Wooden Calorimeter. T = 975 by the thermoelectric pyrometer P = 10,000 grm. p = 145 grm. to = 20.21 /I = 21.99 T Q L T = 125 cal. L 131.5 cal. from the curve at 975. The difference is 6.5 calories, or 5 per cent loss due to the jacket. III. Trial with the Water-jacketed Calorimeter. T = 985 P = 2000 grm. p = 48.4 grm. *o = 18.86 \= 2I '95 L = 130 cal. rp L Q = 133 cal. from the curve at 985. The difference is 3 calories, or a loss of about 2 per cent only when use is made of a carefully made calorimeter and of a ther- mometer giving -g*Q. This corresponds to an uncertainty of less than 20 in the temperatures sought. Some of this uncertainty may be due to the temperature assumed as correct in these measurements and to loss of heat during transfer. It is possible to work to better than 5 by the most refined methods, using a platinum mass. With the ^0 thermometers, necessitating a much greater rise of temperature of the water in the calorimeter, an uncertainty of 25 or more may exist. The relatively small mass of water used with a less sensitive thermometer is not necessarily a disadvantage, however, if the calorimeter is properly protected against heat loss and evaporation due to the greater temperature rise. CALORIMETRIC PYROMETRY 99 Conditions of Use. The advantages of the calorimetric pyrometer are: 1. Its low net cost; 2. The ease of its use, which allows of putting it in the hands of a workman. Its inconveniencies are : 1. The time necessary to take an observation, about a half- hour, except with the Siemens form; 2. The impossibility of taking continuous observations; 3. The impossibility of exceeding 1000 by the use of the piece of nickel ; 4. The deterioration of the balls used due to oxidation. Its use does not seem to be recommendable for laboratories, as there are continuous methods of greater accuracy readily available for such uses. In the laboratory, the calorimetric method is used ordinarily for the determination of specific heats at high temperatures rather than of these temperatures. In recent years, there have been introduced many refinements into calorimetric measurements, such as vacuum-jacketed calorimeters which nearly eliminate heat losses during the rise in temperature within the vessel; resistance thermometers and thermoelements of great sensitiveness and precision which give the rise in tempera- ture within the calorimeter more accurately than does a mercury thermometer; electric heating and vacuum furnaces for the pre- heating of the sample without contamination to the desired high temperature; and many other details of manipulation and construction, for descriptions of which the reader should consult the writings of Berthelot, Louginine arid Schukarew, Dickinson, Richards, White, Oberhoffer, and others. It is possible, for instance, to keep the total error due to the calorimeter to within i in 10,000. These improvements are of the greatest importance for the exact determination of specific and latent heats and similar constants at high temperatures, but have little interest from the purely pyrometric point of view, since much more delicate and accurate temperature-measuring methods exist which do not involve the transfer of heat. 100 HIGH TEMPERATURES The calorimetric or specific heat pyrometer is to be recom- mended for certain operations below 1000 C. in technical works where it is required to make only occasional measurements of moderate precision; in cases where there is not the personnel sufficiently skillful to use the more precise or delicate methods; and finally, where the importance of the measurements is not such as to justify the buying of more costly instruments. CHAPTER IV. THERMOELECTRIC PYROMETER. Principle. The junction of two metals heated to a given temperature is the seat of an electromotive force which is a function of the temperature only, at least under certain conditions which we shall define further on. In a circuit in- cluding several different junctions at different temperatures, the total electromotive force is equal to their algebraic sum. In a closed circuit there is produced a current equal to the quotient of this resultant electromotive force and the total resistance. Experiments of Becquerel, Pouillet, and Regnault. It was Becquerel who first had the idea to profit from the discovery of Seebeck to measure high temperatures (1830). He used a platinum-palladium couple, and estimated the temperature of the flame of an alcohol lamp, finding it equal to 135. In reality the temperature of a wire heated in a flame is not that of the gases in combustion; it is inferior to this. The method was studied and used for the first time in a systematic manner by Pouillet; he employed an iron- platinum couple which he compared with the air thermom- eter previously described (page 61). In order to protect the platinum from the action of the furnace gases, he inclosed it in an iron gun barrel which constituted the second metal of the junction. Pouillet does not seem to have made applica- tions of this method, which must have given him very dis- cordant results. Edm. Becquerel resumed the study of his father's couple (platinum-palladium). He was the first to remark the great importance of using in these measurements a galvanometer of high resistance. It is the electromotive force which is a function 101 102 HIGH TEMPERATURES of the temperature, and it is the current strength that is meas- ured, Ohm's law gives E = RI. In order to have proportionality between these quantities, E and 7, it is necessary that the resistance of the circuit be invariable. That of the couple necessarily changes when it is heated; this change must then be negligible in comparison with the total resistance of the circuit. Edm. Becquerel studied the platinum-palladium couple and made use of it as intermediary in all his measurements on fusing points, but he did not use it, properly speaking, as a pyrometer; he compared it, at the instant of observation, with an air ther- mometer heated to a temperature near to that which he wished to measure. He also tried to make a complete calibration of this couple, but this attempt was not successful; he did not take into account the irregularities due to the use of palladium; besides, he made use successively for this graduation of a mercury thermometer and of an air thermometer which did not agree with each other. He was led to assume for the relation between the temperature and the electromotive force a very complex expression; the formulae which he gives contain together twelve parameters, while with the parabolic formula of Tait and Ave- narius two suffice; thus which well represents the phenomenon for the couple in question to 1500. Regnault took up the study of Pouillet's couple, and he ob- served such irregularities that he condemned unreservedly the thermoelectric method. But these experiments were hardly con- clusive, for he does not seem to have considered the necessity of using a high-resistance galvanometer. Experiments of Le Chatelier and of Barus. The thermo- electric method possesses, nevertheless, very considerable practi- cal advantages for use in the laboratory as well as industrially, .such as: THERMOELECTRIC PYROMETER 103 Smallness of thermoelectric substance; Rapidity of indications; Possibility of placing at any distance the measuring apparatus. Le Chatelier decided to take up the study of this method, intending at the outset not to make disappear the irregularities which seemed inherent in the phenomena in question, but to study the law of these irregularities, so as to determine correc- tions which would permit of making use of this method, at least industrially, for approximate measurements. These investiga- tions showed in their turn that the sources of error observed could be suppressed; the principal one, and the only serious one, came from lack of homogeneity of the metals up to that time employed. Barus, whose work in this field dates from 1881, studied in great detail the thermoelectric measurement of high tempera- tures as well as the advantages and limitations of the various pyrometric methods. He was led from his researches to prefer the couple Pt, 90 Pt-io Ir. Iron, nickel, palladium, and their alloys were found to be un- suited for the exact measurement of high temperatures, because, heated in certain of their points, they give birth to parasite currents, sometimes relatively intense. D. Berthelot and others, however, have since used successfully in oxidizing atmospheres, thermocouples with palladium as one element. As an example of inhomogeneity, consider the electromotive forces observed by Le Chatelier in carrying a Bunsen flame along beneath a wire of ferronickel of i mm. diameter and 50 cm. long; the electromotive forces are expressed in microvolts (mil- lionths of a volt) : Distance 0.05 o.io 0.15 0.20 0.30 0.35 0.40 0.50 E.M.F 200 +250 150 looo 500 200 50 200 An electromotive force of 1000 microvolts is that given by the usual couples that we are going to study for a heating of 100. With such anomalies as above there could hardly be any meas- urements possible. 104 HJ GH TEMPERATURES These anomalies may sometimes be due to accidental varia- tions in the composition of the wires, but in general there is no preexisting heterogeneity; a physical heterogeneity due to the heating is produced. Iron and nickel, heated respectively to 750 and 380, undergo an allotropic transformation, incompletely reversible by rapid cooling. In the case of palladium, there may be produced, in a reducing atmosphere, phenomena of hydrogenation which change com- pletely the nature of the metal, so that a metal initially homo- geneous may become by simple heating quite heterogeneous and form a couple. Certain metals and alloys are quite free from these faults, notably platinum and its alloys with iridium and rhodium. The irregularities previously observed are thus due to the employ- ment of iron and palladium in all the couples tried. A second source of error, less important, comes from the annealing. In heating a wire at the dividing point between the hardened part and the annealed part, there is developed a current whose strength varies with the kind of wire and the degree of hardness. The twisting that a wire has undergone at a point suffices to produce a hardening. A couple whose wires are hard drawn throughout a certain length will give different indications according to the point of the wire where the heating ceases. Here are results in microvolts obtained by Le Chatelier with a platinum, platinum-iridium (20 per cent Ir) couple (platinum- indium alloy is very easily annealed) : 100 445 Before annealing noo 7200 After annealing 1300 7800 Difference 200 600 We shall now study successively: 1. The choice of the couple; 2. Thermoelectric formulae; 3. The methods of measurement; 4. The sources of error; 5. The standardization. THERMOELECTRIC PYROMETER 105 Choice of the Couple. We shall first reproduce the evidence and arguments which led Le Chatelier to prefer and introduce the thermocouple of composition platinum against its alloy with 10 per cent rhodium for temperature measurements in those cases for which the thermoelectric method is preferable or con- venient. We shall then give account of some of the later work of others in this domain. In the choice of the couple, account must be taken of the elec- tromotive force, the absence of parasite currents, and the inalter- ability of the metals used. Electromotive Force. This varies enormously from one couple to another. Below are several such electromotive forces given between o and 100 by metals that can be drawn into wires and opposed to pure platinum. Microvolts. Iron 2100 Hard steel 1800 Silver : 900 Cu + 10% Al 700 Gold 600 Pt 4- 10% Rh i Pt+io%Ir I S Cu + Ag 500 Ferronickel 100 Nickel steel (5% Ni) o Manganese steel (13% Mn) 3 Cu + 20% Ni - 600 Cu + Fe + Ni -1200 German silver (15% Ni) -1200 German silver (25% Ni) -2200 Nickel -2200 Nickel steel (35% Ni) -2700 Nickel steel (75% Ni) -3700 Barus studied certain alloys between o and 920; he obtained the following results against platinum : Microvolts. Iridium (2%) 79 1 Iridium (5%) 28 3 Iridium (10%) S7oo Iridium (15%) 7QOQ Iridium (20%) 93OQ Palladium (3%). 92 Palladium (10%) 93 Nickel (2%) 3744 Nickel (5%) io6 HIGH TEMPERATURES Here is another series made by Barus at the boiling point of sulphur with alloys of platinum containing 2, 5, and 10 per cent of another metal: Metals. Au Ag Pd Ir Cu 2% s 10 2% 5 10 2% 5 10 - 242 - 18 - 832 -105 -1225 -158 + 711 + 869 + 1127 + 1384 + 2035 +3228 +410 +392 + 257 Ni Co Fe Cr Sn Zn + 2166 +399 + 5095 + 26 -170 41 +3020 +3313 +3962 + 2239 +3123 +3583 + 261 + 199 + 151 +396 + 24 Al Mn Mo Pb Sb Bi + 779 +938 + 758 + 2206 + 263 + 1673 + 766 -268 +338 + H55 + 245 Of all these metals, the only ones to keep by reason of their high electromotive force are the alloys of platinum with iron, nickel, chromium, iridium, and rhodium. The following table gives, in microvolts, the electromotive forces of the 10 per cent alloys of these five metals up to the temperature of 1500: Temperatures. Fe Ni Cr Ir Rh 100 445 920 438 3.962 9,200 646 4,095 9,100 405 3,583 995 6,39 14,670 640 3,690 8,660 1500 19,900 20,200 26,010 15,550 Absence of Parasite Currents. The alloy with nickel gives parasite currents of great intensity, as do all the alloys of this metal. It is the same with iron. Chromium does not seem to present the same inconvenience: it forms an alloy difficult to fuse and, for this reason, difficult to prepare. With the alloys THERMOELECTRIC PYROMETER 107 of indium and of rhodium there is no considerable production of parasite currents if the metals are pure and the alloys homo- geneous. There remain, then, but three metals to consider: iridium, rhodium, and chromium. Of the alloys of these metals with platinum, that of iridium is the one which hardens the most easily. Chemical Changes. All the alloys of platinum are slightly alterable. Those of nickel and of iron, at high temperatures, assume a slight superficial brownish tint caused by oxidation of the metal. No test has been made to see if, after a long time, this attack would reach even to the interior of the wires. According to Le Chatelier the alloys of platinum, and plati- num itself, become brittle by simply heating them long enough, especially between 1000 and 1200; this is due without doubt to crystallization. The platinum-iridium alloy undergoes this change much more rapidly than the platinum-rhodium, and this latter more rapidly than pure platinum. It is questionable, however, if this effect, other than a slight crystallization, occurs in a strictly oxidizing atmosphere with couples containing only Pt, Rh, or Ir. But a much more grave cause of the alteration of platinum and its alloys is the heating to high temperatures in a reducing atmosphere. All the volatile metals attack platinum very rapidly, and a great number of metals are volatile. Copper, zinc, silver, anti- mony, nickel, cobalt, and palladium, at their points of fusion, already emit a sufficient quantity of vapor to alter rapidly the platinum wires placed in the neighborhood. These metallic vapors, that of silver and palladium excepted, can only exist in a reducing atmosphere. Among the metalloids, the vapors of phosphorus and of certain compounds of silicon are particularly dangerous. It is true that one is rarely concerned with these uncombined true metalloids, but their oxides in the presence of a reducing atmosphere are more or less completely reduced. In 108 HIGH TEMPERATURES the case of phosphorus, it is not only necessary to shun phosphoric acid, but also acid phosphates of all the metals and the basic phosphates of the reducible oxides; thus silicon, silica, and almost all the silicates, clay included, must be avoided if a reducing atmosphere is employed. The reducing flames in a fire-clay furnace lead little by little to the destruction of the platinum wires. It is thus indispen- sable to protect the couples against any reducing atmosphere by methods which will be indicated further on. In taking account of these different considerations, electro- motive force, homogeneity, hardness, alterability by fire, Le Chatelier was led to give the preference to the couple Pt Pt + 10% Rh, with the possibility of replacing the rhodium by irid- ium and perhaps by chromium. In all cases the wires should be annealed electrically to 1400 before using. The usual diameter of wire employed is 0.6 mm., but one of 0.4 mm. contains only half as much metal, and even for most in- dustrial purposes is of sufficient robustness. In the laboratory there is advantage, especially on account of heat conduction, to still further reduce this diameter. Thermoelectric Formulae. In spite of numerous attempts to solve the problem, it has thus far been impossible to deduce from purely theoretical grounds a satisfactory equation connect- ing the temperature and electromotive force of any thermo- electric couple. As we shall see, it is necessary to set up, for each type of couple, an empirical equation or a series of such equations which, sometimes within rather restricted temperature limits, represents well enough the desired relation. There is a great diversity of such formulae, and there has been in the past a considerable amount of indiscriminate and unwarranted extrap- olation of such empirical relations to temperature regions both high and low, in which the assumed formulae do not hold. From the very common use of the thermocouple as a temperature- indicating device, this practice has caused considerable confusion in the values to be assigned to high temperatures. We shall call attention to some of the formulae that have been used and THERMOELECTRIC PYROMETER I0 9 point out their limitations both for interpolation and extrapo- lation. In the construction of thermoelectric formulae it is customary to assume a constant temperature, usually o C., for the cold junctions, and to further assume that the only source of E.M.F. is the hot junction. The complete expression, however, for the total E.M.F. developed in a thermoelectric circuit requires ac- count to be taken also of (i) the Thomson effect, or the E.M.F.'s .generated due to differences in temperature along a homogeneous wire; (2) the Peltier effect due to the heating of the junction of two dissimilar metals anywhere in the circuit; (3) the Becquerel effect, or the E.M.F.'s developed by physical or chemical in- homogeneity in a single wire. The E.M.F. actually measured is the algebraic sum of all these quantities. Practically, the Thomson effect need not be taken account of separately in constructing a formula, as it is a function only of the temperature difference along the wires and of their nature. The undesirable Peltier and Becquerel effects, the former oc- curring often in the measuring apparatus, and the latter mainly in the thermocouple wires, cannot be taken care of numerically in any useful thermoelectric formula, and must therefore be eliminated by the use of materials and methods free from these effects. The following formulae, therefore, all assume thermoelectric circuits in which the only sources of E.M.F. are due to the dif- ference in temperature between the hot and cold junctions of the couple. Thermoelectric Power. By differentiating with respect to temperature the expression for the E.M.F.-temperature relation E =f(f) for any couple, we get a quantity known as its thermo- j -p electric power, , which we may designate by H. This quantity is a convenient one with which to compare the numerical be- havior at any temperature of two or more couples, or of one couple at different temperatures, as it gives the E.M.F. per de- no HIGH TEMPERATURES gree of temperature. For some couples H is practically a linear function of t over considerable ranges of temperatures, i.e., H = a -f bt and is a measure of the sensibility of any type of couple. We may cite the following as illustrations: THERMOELECTRIC POWERS OF THERMOCOUPLES. _,, , Thermoelectric Temperature power (microvolts). range. Pt, 90 Pt 10 Rh 4. 3 +o. 0088 t 0-1300 Pt, QoPt loir 11.3 -j- 0.0104 / o-iooo Pt, Ni 7. 8 + 0. 01325 300-1300 Cu, Ni 24.4 + 0.016 / 0-235 Cu, Constantan 42 . 3 + o . 058 t 0-320 Pt Fe (forged) 2 . 5 + o . 02 10 t 700-1000 Author. Le Chatelier Le Chatelier Burgess Pecheux Pecheux Le Chatelier It is usual to express H for a single substance in terms of lead as a standard at ordinary temperatures, but at high tempera- tures this becomes impracticable. The values of H for steels are of special interest in view of their use in many base-metal dE 18 10 loo c - 500 1000 Fig. 23. Thermoelectricity of Steels. couples. In Fig. 23, due to Belloc, are given the changes of H,. with temperature and carbon content for various steels against Pt, from which it is evident that thermocouples with steel or iron as one component have complex E.M.F.-temperature relations, and that the relation between thermoelectric power and tempera- ture is far from linear. THERMOELECTRIC PYROMETER III For some couples the thermoelectric powers of the component wires become equal and opposite in sign at some temperature known as the neutral point, beyond which the sign of the E.M.F. is negative. It is evidently of advantage to use couples in regions removed from their neutral point. As shown by Stansfield, the Peltier effect I T~ jis very nearly linear with the temperature for the Pt-Rh and Pt-Ir couples, but not the thermoelectric power. Sosman's observations on various Pt-Rh couples also bear out this statement. Formula. Avenarius and Tait have shown that up to 300 the electromotive force of a great number of couples is repre- sented in a manner sufficiently exact by means of a parabolic formula of two terms: The experiments of Le Chatelier on the platinum-palladium couple have shown that the same formula holds also for this couple up to the fusing point of palladium: e = 4-3 * - IOOO / = loo 445 954 1,060 i>55 e = 500 2950 10,900 12,260 24,030 Platinum and Its Alloys. This law fails completely, however, for couples made of pure platinum and an alloy of this metal. Here are three early series of determinations made with dif- ferent couples, giving an idea of the order of magnitude of the E.M.F.'s of thermocouples of types used very frequently, as determined by these observers. Barus. Pt - Pt 10% Ir. Le Chatelier. Pt - Pt 10% Rh. Holborn and Wien. Pt - Pt 10% Rh. I e t e J e 300 2,800 IOO 550 IOO 565 500 5. 2 50 357 2,770 2OO 1,260 700 7,900 445 3.630 4OO 3.030 900 10,050 665 6,180 600 4,920 I IOO 13,800 1060 10,560 800 6,970 155 16,100 IOOO 9,080 1780 18,200 I2OO 11,460 I4OO 13,860 l6oO 16,220 112 HIGH TEMPERATURES Holman showed that the results of Holborn and Wien may be expressed by a logarithmic formula containing only two param- eters and requiring, therefore, only two calibration temperatures. Le Chatelier showed likewise that his results could also be repre- sented by the Holman formula, and in general it may be said that for use below 1200 C. of the thermocouple made of platinum and its alloys with rhodium and iridium, the logarithmic formula satisfies the results of observations to 2 C., or well within the limits of all except the most accurate work. Holman's formula is as follows: (i) x_^ < where ^ e is the electromotive force of the couple for any tem- o perature / when the cold junction is kept at zero centigrade. The two constants are readily computed or evaluated graphically, and the resultirig plot serves indefinitely for the determination of any temperature with a given couple. The equation does not apply in the region in which the thermocouple is insensitive, that is, below 250 C. It may be written, for convenience in plotting and computation: (2) log 2/ e =n\Qgt + log m; so that if log e be plotted as abscissas and log / as ordinates, a straight line is obtained. This formula has been applied successfully to the above obser- vations of L^ Chatelier on platinum-rhodium couples and to those of Barus on platinum-iridium couples. Holborn and Day, in their very elaborate, direct comparison of the nitrogen thermometer with thermocouples made of the various platinum metals, in the interval 300 to 1100 C., found that if a precision of i is sought, a three- term formula is required to express the relation between E.M.F. and temperature. The formula (3) THERMOELECTRIC PYROMETER is the one they have used. The labor involved in computation with this form is considerable, and, unless a very great accuracy is required, Holman's formula is amply sufficient, when the uncertainty of the absolute values of high temperatures is con- sidered. Stansfield deduces from theoretical considerations the formula \4/ ffT J which may be written a form which satisfies the experimental results determined with pure platinum wires. This form possesses no practical advan- tage over that of Holborn and Day, unless it be its usefulness, by employing the graphical method, in detecting slight errors in fusing points. The values of at the points of fusion can be de obtained from the T vs. e plot, and the T vs. curve thus con- structed throws into prominence the experimental errors at these points. As the above formulae indicate, the curve for the plati- num metals constructed with T as abscissas and T - as al ordinates is a straight line. The errors of the method are less than 2 at 1000. The ordinary metals, on the other hand, with a few exceptions such as nickel and cobalt, give nearly a straight de line for the curve T vs. al A formula which has been used on account of its more con- venient form, than (3) for example, in the computation of tem- perature, is: (6) / = a + be - ce 2 . This formula satisfies the observations with platinum-rhodium and platinum-iridium couples in the range 300 to 1100 C. almost as well as (3). 114 HIGH TEMPERATURES We may compare these various formulae by computing their deviations at various fixed points, making use of the latest data with comparison of a thermocouple (90 Pt - 10 Rh-Pt) with the gas-thermometer scale, those of Day and Sosman, 1910. We shall assume as calibration temperatures for the three-term equations, (3), (5), and (6), the freezing points of zinc, antimony, and copper, and for Holman's equation (2), zinc and copper. COMPARISON OF THERMOELECTRIC FORMULA (Pt-QO Pt -ioRh) Substance. Freezing point. Observed (microvolts). Observed calculated temperatures. (2)' (3) (5) (6) Cadmium 320.0 2,502 O.2 -0.3 + 6.9 I.I Zinc 418.2 3.429 o Antimony 629.2 5.529 + 2.3 o o.i Silver 960.0 9,111 +2.5 + 0.2 + 2.2 - 0.9 Gold 1062.4 10,296 +0.4 + 0.2 O Copper 1082.6 io.535 o O O + O.I Diopside 1391 14,231 -6 + 10 10 +19 Nickel I45 2 14,969 -6 + 14 II +28 Cobalt 1490 15,423 -7 + 14 -13 +3i Palladium J 549 16,140 -5 + 20 14 +42 Platinum 1755 18,613 + i +42 -15 + 73 The numbers in parentheses refer to formulae on preceding pages. It is evident from the table that we have, therefore, as many thermoelectric scales as we have equations. The two formulae which best fit the region 300 to 1100 C., namely, (3) and (6), are clearly not suited for extrapolation without applying proper corrections. Of all the formulae, Holman's (2), which is also the simplest, is the best suited for general use throughout the whole range 300 to 1750, giving a maximum error of 2.5 below, and of 7 above, 1100 C. None of these equations is satisfactory for the most exact work, however. A cubic equation in / will satisfy the data more exactly, but this is extremely inconvenient to solve for t\ or two parabolas of type (3) may be used, the first from 300 to 1100, the second from 1100 to 1750. In 1905, Harker, using thermocouples of platinum against a 10 per cent rhodium and 10 per cent iridium alloy of plati- num, respectively, and extrapolating equation (3) from uooC., THERMOELECTRIC PYROMETER 115 obtained 1710 C. with both types of thermocouple as the uncor- rected value of the platinum melting point. This value, 1 7 10 C ., has been generally accepted in many quarters as the true melting point of this metal. Waidner and Burgess, however, demon- strated in 1907 that the value found for high melting points by extrapolating with thermocouples depends not only on the ther- moelectric relation assumed, but also on the nature of the couple. Some of their results for the palladium and platinum melting points are given below, the calibration equations and tempera- tures being the same as in the above. EXTRAPOLATION WITH VARIOUS THERMOCOUPLES. Tvnp nf rnimlp Equa- Palladium, Platinum, tion. MP = 1549- MP = I 7 5S. 4of Pt,9oPt - ioRh(approx.) ((3) 1521 to 1537 1698 to 1715 2 makers ................... } (2) 1536 to 1561 1717 to 1754 2 of Pt, 90 Pt 10 Ir ( (3) 1525 to 1528 1705 to 1710 2 makers .................. 1(2) 1516 to 1541 1697 to 1728 2 o f9 oPt- IO Rh,8oPt- 2 oRhj(3) J507 ,687 to ,7,0 It would appear from these data that the corrections to apply to a given type of thermocouple computed and extrapolated with a given formula are uncertain, the slight variations in composi- tion of the alloy wire from one couple to another apparently producing considerable differences in the computed temperatures. It is an interesting fact that the 10 per cent alloys of Rh and Ir with Pt, when treated by equation (3), give very exactly the same temperature scale to the melting point of platinum, although the actual shapes of the E.M.F. temperature curves are very different for those two couples, that for Pt Ir being the more nearly linear. It was an instructive case of two negatives not making an affirmative to assign the value 1710 as the true Pt melting point because both Ir and Rh couples led to the same result. Using Pt-Rh couples of i, 5, 10, and 15 per cent Rh, and cali- brating in terms of equation (3) at the melting points of copper, diopside, and palladium (see above), Sosman finds 1752 as a mean value for Pt with a range of only 7. Il6 HIGH TEMPERATURES Variation of E.M.F. with Composition. Sosman has also studied this for the Pt-Rh couples, and some of his results are given in Fig. 24. It will be noticed that in the region of the 10 per cent alloy, which is the one most commonly met with, or at least such is the nominal composition usually given, a change of i per cent in composition is equivalent to about 50 at 1000. The Base-metal Couples. The E.M.F.-temperature relation for some of these couples, of which there are a great many in use, is very nearly linear. For some couples, on the other hand, the E.M.F.-temperature relation is very complex; and in those cases in which there are allotropic or other transformations within the material, taking place over a temperature range or along the wire as the successive portions are heated or cooled, there some- times occur inflections in the curve, producing regions of con- siderable extent in which the couple is relatively very insensitive. When such inflections occur, there is usually no conveniently expressed relation between E.M.F. and temperature (see Fig. 23). We shall call attention later to some specific cases of base-metal thermoelectric formulae. Methods of Measurement of Temperature. Two methods may be used to measure the electromotive force of a couple: the method of opposition and the galvanometric method. From the scientific point of view, the first alone is rigorous; it is usually made use of in laboratories. The second method is simpler, but possesses the inconvenience of giving only indirectly the measure of the electromotive force by means of a measurement of current strength. This inconvenience is more apparent than real in the later forms of instrument, as will be shown. There are sources of error, however, inherent in the galvano- metric method, such as effects of lead resistance and temperature coefficients of leads and galvanometer, which, as we shall see, are difficult if not impossible of complete elimination even with the best apparatus available. The method of opposition, on the other hand, may be made, in so far as the measurements of E.M.F. are concerned, as exact as may be desired, or so that THERMOELECTRIC PYROMETER 117 /ll \l J f A 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1755 Temperature, Centigrade Fig. 24. E.M.F. of Pt-Rh Thermocouples. Il8 HIGH TEMPERATURES the only outstanding uncertainties are inherent in the thermo- couple itself. These uncertainties, such as inhomogeneity and conduction along the wires, variable zero, and actual change of E.M.F., are sometimes overlooked, giving rise to illusory accu- racy. We shall describe each of these methods and discuss their limitations, and also point out the sources of error most likely to \ be present with the various types of thermoelectric apparatus. Galva^on^etric Method. The measurement of an electro- motive force may be reduced to that of a current; it suffices for that to put the couple in a circuit of known resistance, and from Ohm's law we have *-:" If the resistance is not known, but is constant, the electro- motive force will be proportional to the current strength, and that will suffice, on the condition that the calibration of the couple is made with the same resistance. If this resistance is only approximately constant, the relation of proportionality will be only approximately exact. This method is the one used in practically all industrial prac- tice, and to-day galvanometers can be had satisfying all the re- quirements of which we shall treat in the following paragraphs. In many quarters the thermoelectric pyrometer has been dis- credited because instruments giving evidently unreliable results were used. With a better understanding of the requirements and the meeting of them by manufacturers, this prejudice is disappearing. \ Resistance of Couples and Galvanometer. The wires of the Wuple make necessarily a part of the circuit in which the Current strength is measured, and their resistance varies with increase of temperature. It is important to take account of the order of magnitude of this inevitable change of resist- ance. Barus made a systematic series of observations on the alloys of platinum with 10 per cent of other metal. The relation THERMOELECTRIC PYROMETER 119 between the resistance and the temperature being of the form R t = Ro (i + at), he obtained the following results: Pt (pure) Au Ag" Pd Ir Cu Ni Fe Cr Sn Specific resistance in mi- crohms (R) 1C. * 75 6 34.8 23 o 24 4 63 o ?? 7 N6 42 3O 1000 a 2.2 I 0.7 1.2 1.2 O.2 O.Q 0.4 0-5 0.7 Other tests gave the figures below: 5% Al & 10% Mo s pf 2% Sb $ % 5% Zn Ro lOOOtt 22 I t; So o 4 I 7 .6 I O 7-7 i 8 29.5 I 16.6 2 47.8 O 3 25 j i The coefficient a is taken between o and 357 (boiling point of mercury). The experiments of Le Chatelier, for the couples that he used, gave the following results: For platinum, R = ii. 2 (i + 0.002 i) between o and 1000. For platinum-rhodium (10% Rh), R = 27 (i + 0.0013 /) between o and 1000. Holborn and Wien found for pure platinum, R = 7.9 (i + 0.0031 1) between o and 100, R = 7.9 (i + 0.0028 /) between o and 1000. Very commonly couples are made of the platinum metals of wires i m. in length and 0.5 mm. in diameter; their resistance, which is about 2 ohms cold, is doubled at 1000. If use is made then of a galvanometer of a resistance of 200 ohms, and if the variation of the resistance of the couple is neglected, the error is equal to T ^ . In general this error is still less except in certain 120 HIGH TEMPERATURES industrial uses. Thus in the laboratory the length heated is often less than 10 cm., and then the error reduces to y^Vo- We may calculate the effect of resistance in the electrical cir- cuit, including that of the couple and galvanometer, on the read- ing of the pyrometer galvanometer in the following way: If E is the true E.M.F. generated by the thermocouple and E' the E.M.F. indicated by a galvanometer of resistance R, in series with the couple and leads, of resistance r and r' respectively, then In the case of certain industrial installations, where the galva- nometer is at a distance from the couple, the value of r' , the resistance of the copper wires connecting the couple to the gal- vanometer, may be of as great importance as that of the couple wires, r. The value of v' can of course be kept down, however, by increasing the size of wire used. Although in the case of platinum couples, which on account of cost, high specific resistance, and temperature coefficient of the materials necessarily have an appreciable resistance and therefore require a relatively high resistance galvanometer, it should be noted that, with base-metal couples of large cross section and consequently low resistance, galvanometers of very much lower resistance, and therefore of a more robust type, in general, may be allowed here. For example, if the couple has a resistance of o.i ohm and the connecting leads a negligible resistance, as may readily happen with certain types of pyrometer rod, the gal- vanometer may be a millivoltmeter of only 10 ohms without introducing errors over T ^, or 10 at 1000 C., due to this cause. Pyrometer Galvanometers. It may still be of interest to recall the historical development of this phase of the subject, as it offers a good illustration of the influence of one field of activity on another, and from the fact that the difficulties encountered and the precautions to be taken in the construction and use of these instruments are not yet sufficiently well appreciated by THERMOELECTRIC PYROMETER 121 some manufacturers as well as by many experimenters and other users. The earliest measurements, those of Becquerel and of Pouillet, were made with needle galvanometers controlled by terrestrial magnetism. Such apparatus, sensible to jarring, requires deli- cate adjustment, and the readings take a long time. The use of these instruments would have prevented the method from be- coming practical. It is only thanks to the use of movable-coil galvanometers of the Deprez-d'Arsonval type that the thermo- electric pyrometer has been able to become, as it is to-day, an apparatus in current usage. This apparatus, in one of its earlier forms (Fig. 25), is composed of a large horseshoe magnet between whose poles is suspended a movable frame through which the cur- rent passes. The metallic wires, which serve at the same time to suspend the coil and bring in the current, undergo then a torsion which is opposed to the displacement of the coil. The latter stops in a position of equilibrium which depends both on the strength of the current and the value of the torsion couple of the wires. To these two forces is added, in gen- eral, a third, due to the weight of the coil, which causes disturbing effects often very troublesome. We shall speak of this further on. The measurement of the angular displacement of the coil is made sometimes by means of a pointer which swings over a divided scale, more often by means of a mirror which reflects on a semitransparent scale the image of a wire stretched before a small opening conveniently lighted. These movable-coil galvanometers were for a long time con- sidered by physicists as unsuited for any quantitative measure- ments; they were only employed in null methods and made accordingly. In order to render them suitable for quantitative Fig. 25. Moving-coil Galvanometer. 122 HIGH TEMPERATURES measurements of current, it was necessary to attend to a series of details of construction, previously neglected. Here are the most important among these, as noted by Le Chatelier for suspended- coil galvanometers: 1. The movable coil should possess a resistance as little variable as possible with the surrounding temperature in order to avoid corrections always very uncertain. The coils of copper wire ordinarily used to augment the sensibility should be absolutely discarded; use should be made of coils of German silver or of sim- ilar metal with small temperature coefficient such as manganin. 2. The spaces which separate the coils, from the poles of the magnet, on the one hand, and from the central soft-iron core on the other, should be sufficiently great to avoid with certainty any accidental friction which would prevent the free movement of the coil. The rubbings to look out for do not come from the direct contact of the frame with the magnet: these latter are too visible to escape unseen. Those which are to be guarded against come from the rubbing of filaments of silk which stand out from insulating covering of the metallic wires, and from the ferruginous dust which clings to the magnet. It is here, it would seem, that the most serious source of error is met with in the use of the movable-coil galvanometer as measuring instrument. There is no warning indication of these slight rubbings which limit the displacement of the coil without, however, taking from it its apparent mobility. 3. The suspending wire should be as strong as may be to support the coil without being exposed to breaking by shocks; on the other hand, it should be very fine, so as not to have too great a torsion couple. Two different artifices help to reconcile somewhat these two opposed conditions: the use of the mode of suspension of Ayrton and Perry, which consists in replacing the straight wire by a spiral made of a flattened wire, or more simply the use of a straight wire flattened by a passage between rollers. The first method offers the greatest security from shocks; it is, on the other hand, more difficultly realizable; minute precau- tions should be taken to prevent any rubbing between adjoining THERMOELECTRIC PYROMETER 123 spirals. The second method allows more easily having the large angular displacements which are indispensable when it is desired to take readings upon a dial. The most essential property necessary for the wires is absence of permanent torsion during the operations. These torsions cause changes of zero which may render worthless all the obser- vations if account is not taken of this, which complicates matters considerably if such correction has to be made. This result is reached by using wires as long as possible, having not less than 100 mm. length, and by avoiding giving to them an initial torsion, a precaution that should be kept constantly in mind, which it often is not. When one wishes to adjust the coil to the zero of graduation, one turns often haphazard either one of the wires; it may be then that each of the wires is given an initial torsion of considerable magnitude and of opposite sign. If the two wires are not symmetrical, as is ordinarily the case, the "permanent deformation resulting from this exaggerated torsion will cause a continual displacement of the zero which may last for weeks and months, increasing or decreasing during the observations accord- ing to the direction of displacement of the coil. This torsion is easy to obviate at the time of construction, but it is not possible to verify later its absence in the case of round wires or spirals except by dismounting the apparatus. On the contrary, by the use of stretched flat wires it is very easy upon simple examina- tion to determine the existence or absence of torsion. This is another reason for employing them. Finally, use must be made of wires having a very high elastic limit. For that it is necessary that the metal has been hardened, and besides that the metal does not undergo spontaneous hard- ening at ordinary temperatures. Silver, generally employed as suspension wire, is worthless. A metal, as iron, which even after annealing possesses a high elastic limit, would be perfect if it were not for its too great alterability. One cannot be sure of having uniform hardening, because the soldering of wires, indispensable to assure good contacts, anneals them throughout a certain length. German silver is the metal the most frequently used 124 HI GH TEMPERATURES in galvanometer suspensions destined for pyrometric measure- ments. The alloy of platinum with 10 per cent of nickel seems preferable; after annealing it has a high elastic limit, and possesses a tenacity much higher than that of German silver. Its dis- advantage is to possess a limit of elasticity twice as great, which reduces by one-half the deflections of a given cross section of wire. Phosphor bronze also gives good results. 4. Installation of the apparatus for the galvanometers, in which the coil is carried by two opposed stretched wires, necessitates special precautions. In the first place, it should be located beyond the influence of jarrings of the ground, which render reading impossible; then it is necessary that its position remain rigorously fixed. If, in fact, the two extreme points of suspension of the wires are not exactly in the same vertical, and if the center of gravity of the coil is not exactly in the line of the two points of suspension, two conditions which can be never rigorously realized, the apparatus constitutes a bifilar pendulum of great sensibility. . The slightest jarring suffices to provoke very considerable angular displacements of the coil. To avoid them, the apparatus should rest upon a metallic support attached to a wall of masonry. When the apparatus is placed, as is often the case, upon a wooden table resting upon an ordinary wooden floor, in order to obtain a deflection of the coil, and in consequence a displacement of the zero, it suffices to walk around the table, which causes the floor to bend slightly, or to provoke a current of air, which, in chang- ing the hygrometric state of the legs of the table, causes it to tip somewhat. Coils freely suspended from above have not these disadvan- tages. Types of Suspended-coil Galvanometer. A series of galvanom- eters have been studied especially in view of pyrometric meas- urements; we shall pass them rapidly in review. For laboratory researches the usual swinging-coil galvanom- eter as made by Carpentier is often used in France. One must make sure that these instruments satisfy well the indispensable THERMOELECTRIC PYROMETER 125 conditions which we have mentioned, which is not always the case when these instruments have been constructed with refer- ence to the ordinary experiments of physics. This laboratory apparatus, the only one which existed at the time of the first investigations of Le Chatelier, was not trans- portable, and could not be arranged for experiments in industrial works. It was then necessary to devise a special model of gal- vanometer easy to carry about and to put in place. The appara- tus (Fig. 26) is composed of two parts, the galvanometer and the transparent scale with its light. The two parts are symmetrical,. Fig. 26. Le Chatelier's Thermoelectric Galvanometer. and, for transportation, may be fixed back to back on the same plank carrying a handle. For observations they are fastened to a wall by means of two nails driven in at a suitable distance apart. The suspension wires, in case of breakage, may be immediately replaced. They carry, soldered to their two ends, small nickel spheres, which one has only to slip on to forked pieces attached to the top and bottom of the coil, and to the supports of the apparatus, respectively. The mirror consists of a plano-convex lens, silvered on the plane face, which gives much sharper and brighter images than the ordinary small mirrors with parallel faces. Carpentier has also made for the same purpose a galvanometer 126 HIGH TEMPERATURES in which the readings are taken by means of a microscope. It is an easily transportable apparatus and very convenient. It has the fault to be subject to a displacement of the zero resulting from the unsymmetrical heating of the body of the microscope by the small lamp which lights the reticule. The stretched wires are replaced by large spirals which offer an absolute resistance to rupture by shock during transportation. Fig. 27. Keiser and Schmidt Outfit. The use of this apparatus necessitates an arrangement which permits, during the observations, putting the galvanometer on open circuit so as to verify the zero reading. In the three preceding galvanometers the measurement of the deflection of the coil is made by optical means; in the following, the measurement is made by means of a needle swinging over a scale. After a study made by Holborn and Wien at the Physikalische THERMOELECTRIC PYROMETER 127 Reichsanstalt in Berlin of the Le Chatelier thermoelectric pyrom- eter, the firm of Keiser and Schmidt devised a needle galvanom- eter (Fig. 27) which works fairly well, although the early forms of this instrument were of too low resistance for many industrial purposes and its temperature coefficient is unduly high. It has the disadvantage of being somewhat fragile. The suspending wire of the coil does not seem to have more than ^V mm - diameter ; the mounting of the apparatus is quite complicated. Repairs cannot readily be made either in the laboratory or works. Fig. 28. Siemens and Halske Pyrometer Galvanometer. The firm Siemens and Halske has also devised an excellent model of needle galvanometer suitable for temperature measure- ments (Fig. 28). Its resistance is 340 ohms, or 400 ohms in the later forms; the scale has 180 divisions, each corresponding to 10 microvolts. There is also a second graduation which gives the temperature directly with the couple sold with the apparatus. Commutators allow of putting the apparatus successively in communication with different thermoelectric couples, if it is desired to take simultaneously several sets of observations. This instrument is provided with a good level, and has a small temperature coefficient. Hartmann and Braun also manufac- 128 HIGH TEMPERATURES ture excellent instruments of this type. Their wall pattern is shown in Fig. 29. Pellin of Paris has made, from designs of Le Chatelier, a needle galvanometer of simple construction which can be repaired where it stands. The very long suspension wire is of 10 per cent nickel platinum; it has T V mm - diameter and is drawn out flat. The lower wire is made of a spiral of the same wire of ^ mm. diameter, which is situated in the interior of the iron core so as to Fig. 29. Hartmann and Braun Wall Type. insure uniformity of temperature. When the spirals of the sus- pension are unequally heated by radiation from the room or for other reason, there results considerable displacement of the zero. A spirit level permits of rendering the apparatus vertical, but it is prudent, by reason of the length of the suspension wire, to make sure directly of the absence of rubbing on the coil. For this a slight jar is given to the apparatus; the point of the needle should take up and keep for a long time a slow oscillatory move- ment in the direction of its length; the transverse oscillations ceasing rapidly indicate friction upon the coil. THERMOELECTRIC PYROMETER 129 Fig. 30. Paul's Unipivot Mounting. v Pivot Galvanometers. The development of satisfactory piv- oted moving-coil electrical instruments with spring control, whose indications as given by a pointer on a scale do not change with time, is very largely due to Weston. It is only recently, however, that pivoted milli- voltmeters of sufficiently high resist- ance and range to use with platinum couples have been made. The charac- teristics of the design of one type of such instruments are shown in Fig. 30, illustrating Paul's mono-pivot construc- tion. The indicator is a low-reading mov- ing-coil voltmeter, : the circular coil of which is pivoted at the center of a spherical iron core, and carefully balanced so that its position is unaffected by vibration or want of exact leveling. The pivot works in a finely polished jewel, from which it is completely lifted on depressing a plunger projecting through the top of the instru- ment, thus rendering the apparatus proof against rough handling in transit. A moving coil of low resistance is used in conjunc- tion with a large resistance of negligible temperature coefficient included in the instrument, "so that any error due to change of temperature of the indicator is thus reduced to a minimum. The movement of the coil is controlled by a spring, and the index may be set to zero, should this be necessary, without opening the instrument, an external adjustment being provided for this purpose. The unipivot principle entirely eliminates the delicate suspensions previously used, which frequently caused trouble by accidental breakage, this necessitating the entire re- adjustment of the apparatus. In general, we may point out that for this type of galvanom- eter the case must be dust-free to avoid the collection of particles in the very small clearance of the moving coil, mounted between the pole pieces of a powerful permanent magnet. When well 130 HIGH TEMPERATURES designed the magnetic circuit does not change its qualities appre- ciably in years, and such instruments are very little affected by extraneous magnetic fields, and are very robust, being capable of standing relatively rough handling. They require no leveling and several types have no adjustment whatever. It is usually well, however, to have a set screw to lock the pointer or coil system, unless, as is sometimes the case, lifting the instrument clamps the pointer. It is also convenient to be able to adjust readily the zero position of the galvanometer, and also to be able to eliminate mechanically the effects of temperature change on the readings of such an instrument. There are many milli- voltmeters on the market of sufficient range and sensibility for the thermoelectric measurement of temperature, but only a very few of them are properly designed for such usage, and great care should be taken, in purchasing a pivot galvanometer, to find out if the instrument in question is suited for the work in hand. It has been the custom of some dealers in pyrometric apparatus to make use, for example, of pivot milli voltmeters of absurdly low resistance in connection with relatively high resistance thermo- couples (see page 119). A milli voltmeter may, therefore, be suitable for use with one type of thermocouple and not with another. In order to enjoy the practical conveniences of the pivot type of galvanometer, at least when using platinum thermocouples, some sacrifice of precision, range or sensibility has to be made. It appears to be as yet impracticable, for example, to make open- scale instruments with a range of 18 millivolts and increase the resistance above 170 ohms, and the range of the best makes is from 90 to 1 60 ohms. In this case, as we have seen, the read- ing of the galvanometer will depend somewhat upon the length of leads and upon the depth of immersion of the couple in the heated space. There are a great many manufacturers of low-resistance pivot millivoltmeters, some of which are suitable for use with base- metal couples of sufficiently low resistance. Among the manu- facturers of pivot instruments suitable for use with platinum THERMOELECTRIC PYROMETER 131 couples are Paul of London, Siemens and Halske, and the Cam- bridge Scientific Instrument Company. The first makes a uni- pivot galvanometer, and the others double-pivot instruments of the West on type. Temperature Coefficient of Galvanometers. It is desirable that the readings of indicating galvanometers be as little affected as possible by temperature changes in the instruments themselves. In the earlier pyrometer galvanometers this matter was generally overlooked, but in many of the newer instruments provision is made for eliminating this effect. Some of the instruments most commonly used in pyrometric practice have temperature coeffi- cients ranging from 0.03 per cent to 0.25 per cent per degree C., depending on the type and maker. They all read too low for an increase in temperature. That this is a serious source of error is evident from an example. If an instrument having a temperature coefficient of o.i per cent per degree C. is calibrated at 15 C. and is used at 35 C., as may readily happen in practice, its readings will be low by 2 per cent, or 20 C. at 1000 C., due to this cause alone. The simplest method in theory for the elimination of this effect is to use wire having no temperature coefficient, such as man- ganin, for the coil and auxiliary resistance of pyrometer galva- nometers. Manganin has the further advantage that its Peltier effect against copper is almost zero. It appears, however, to be difficult to get sufficient sensibility in this way due to the high specific resistance of manganin. There are various other devices for cutting down or eliminat- ing this effect, some based on the choice of materials for and the ratio of the coil and balance resistances, and others on the variation of the strength of the magnetic field between the pole pieces, effected either by hand or automatically. In the single-pivot indicators of 100 ohms total resistance, of R. W. Paul of London, for example, the resistance of the copper moving coil is only 10 ohms, the balance being of manganin, reducing the temperature variation in the resistance of this gal- vanometer to the order of 0.047 P er degree C. 132 HIGH TEMPERATURES The use of an adjustable magnetic shunt for the elimination of this temperature correction may be illustrated as follows: The deflection D of the galvanometer may be considered pro- portional to the product of the flux F of the magnet by that, /, of the moving coil, or D = kFf. But / is directly proportional to the electromotive force e to be measured and inversely to the resistance of the circuit, whence fu [!+(< -I S )]' where ^15 is the resistance of the circuit at 15 C., a its tempera- ture coefficient, and t its temperature. We have, therefore, D = kk'F [i+a(/- 15)] Since F remains sensibly constant with temperature, it follows that in order to have the same deflection for a given value of e y it is sufficient to cause F to change proportionally with the re- sistance of the circuit. This is realized in practice, as in the instruments of Chauvin and Arnoux, by the use of a small bar of soft iron which may be brought nearer to or farther from the poles of the magnet, which operation produces a change in the magnetic flux through the movable coil. The motion of the iron bar is controlled by a screw whose head is graduated in degrees of temperature. The temperature of the auxiliary thermometer embedded in the gal- vanometer case is read and the magnet-control screw set to the indicated temperature, when the galvanometer readings are then corrected for temperature coefficient. An automatic magnetic balancing of the increase in resistance of the galvanometer coil with temperature has been introduced into the Thwing galvanometers, as shown in Fig. 31. The coil rotates about one of its ends in a uniform field between two plane pole pieces. The two magnets that are connected in parallel by these pole pieces differ from those ordinarily used in being thin and therefore flexible. These magnets are pressed together somewhat by the long arm of a strong lever, the short arm of THERMOELECTRIC PYROMETER 133 which rests upon a post which is part of the aluminium case. The fulcrum is a bar of invar. Changes in temperature expand or contract the aluminium part, closing or opening somewhat the gap between the poles of the magnet, and the whole may be adjusted so that the change in flux through the coil may balance its change in resistance. Dial Needier, y Lever Fig. 31. Thwing's Compensating Device. The Siemens and Halske method of temperature compensation is by a suitable combination of series and shunted resistances of copper and manganin in the swamping resistance of the instru- ment. It should perhaps be emphasized at this point that the elimi- nation of the temperature coefficient of the indicating galva- nometer does not do away with making proper corrections for changes in temperature of the cold junctions of the thermo- couple (see page 155). Galvanometer Requirements for Industrial Practice. In many industrial operations it is desirable to be sure of temperature measurements to within 10 C., often over a very considerable temperature range. This accuracy can be obtained with certain forms of the pyrometer galvanometer both with platinum couples and with some of the base-metal couples, but only when certain conditions are fulfilled by the maker and the user of the instru- ment. We may emphasize some of the desirable and necessary features of the galvanometer, as follows: The instrument, if of the moving-coil pointer type, should be dust-free, of sufficient sensibility and range, and at the same time it should have an open, nearly equidistant scale which is well marked and easily read, without parallax, for example, by means 134 HIGH TEMPERATURES of reflection of the pointer in a mirror alongside the scale. The deflection should be aperiodic or deadbeat, and the open-circuit reading should remain constant even after large deflections long maintained. There should be, in the case of suspended-coil instruments and in some pivot types, a suitable leveling device which has been accurately adjusted, and in these instruments particularly the case and other supporting parts should be free from warping. The zero position of the pointer should be readily adjustable. The instrument should either be free from errors due to changes in its temperature or else some form of compensation provided, and there should be no possibility of thermoelectric effects in the wiring within the instrument. The effects of jarring due even to shocks of considerable intensity and changes in the surrounding magnetic field should be with- out material influence on the readings. For pivot instruments particularly, it should be noted that the same E.M.F. always gives the same deflections. Finally, as we have before stated, the resistance of the galvanometer must be sufficiently high for the type of couple with which it is to be used. The effect of varia- tions in the temperature of the cold ends of the thermocouple will be treated later. When a new couple is substituted, it should be noted that the E.M.F. scale of the galvanometer will still be correct, barring the effect of change in resistance of the circuit, but unless the new couple is identical in its electrical properties with the old, the temperature scale of the instrument will no longer hold. One instrument may often serve for use with several couples of the same or of different types. It is then very important to avoid bad contacts in switches, and with very low-resistance outfits considerable errors that are not readily detectable may creep into the measurements. A galvanometer suitable for use with a Pt-Rh couple may very properly be used with a low- resistance base-metal couple of higher E.M.F. by putting addi- tional resistance into the circuit if necessary, but a galvanometer suitable for use with the base-metal couple may be totally unfit for use with one of Pt-Rh. THERMOELECTRIC PYROMETER 135 The Galvanometer Method in the Laboratory. On account apparently of its relatively low cost, and also because of its speed of operation, the galvanometer method of measuring temperatures with the thermocouple has been used frequently in scientific investigations of considerable delicacy. It should be borne in mind, however, that, even with the best pointer instruments carefully calibrated, which are much used in metal- lurgical and physiochemical researches, an accuracy of 5 is barely attainable with Pt-Rh couples, and this only by paying attention to the numerous sources of error we have emphasized above. A sensitive d'Arsonval galvanometer read by reflection upon a straight graduated scale, or by means of a telescope and scale, has also been a favorite method of working. In this way the sensitiveness over the pointer method may be increased greatly, but in general the accuracy will not be very materially improved, as practically all of the troubles inherent to the galvanometer method are usually still present, whatever the method adopted for reading the deflection of the galvanometer coil. By slight modifications, the exactness of the galvanometer method may be increased, as for instance keeping the cold junctions at a definite and known high temperature and depending on the sensitive galvanometer for a smaller temperature interval; or better, by opposing the greater part of the E.M.F. of the couple with a known E.M.F. furnished by a standard cell and resistance or volt box. This last, however., is the simplest case of the potentiometric methods which we shall now study. Potentiometric Methods. The fundamental principle on which the many potentiometric methods are based is the adjust- ing of the electric circuit so that no current flows through the thermocouple. This is accomplished by balancing the E.M.F. generated in the thermocouple by an E.M.F. whose numerical value may be varied at will and measured. Since the two E.M.F. J s are in opposition, the measurements may be made to have all the advantages of a null or zero method, which is usually desirable in precision work. 136 HIGH TEMPERATURES Apparatus Required. A complete installation for work to i C. consists of: i. A standard cell, which should not have any current pass through it, and serves to determine, as term of comparison, a difference of potential between two points of a circuit through which there is a current given by an accumulator. The cell used may be a Latimer-Clark, whose electromotive force for small changes in temperature is e = 1.433 v l ts 0.00119 (t 15). This cell is made up as follows : zinc, sulphate of zinc, mercurous sulphate, mercury. The zinc sulphate should be perfectly neu- tral; for that, the saturated solution of the salt is heated to 40 or more with an excess of zinc oxide to saturate the free acid, is then treated with mercurous sulphate to remove the excess of zinc oxide dissolved in the sulphate, and finally crystallization is produced at o; one thus obtains crystals of zinc sulphate which can be immediately used. This element is very constant. With a surface of zinc electrode equal to 100 sq. cm. and a resistance of 1000 ohms, the dropping off of the electromotive force of the cell in action does not reach nnnr with 100 ohms only, this would be %fa. Practically it is possible, with a resistance of 1000 ohms, to limit the surface of the electrodes to 30 sq. cm., and to do away with the use of accumulators. But then the theoretical advantage of the abso- lute rigor of the method employed is lost. There are other forms of standard cell which possess the ad- vantages of portability and small temperature coefficient, ren- dering them better adapted for ordinary use than the original Clark form. The Carhart-Clark cell is made with unsaturated mercurous sulphate and has the E.M.F. e = 1.439 ~ 0.00056 (f - 15). In the Weston normal cadmium cell, which has generally re- placed the Clark as a standard of E.M.F. , and has been officially recognized as the standard by the London Electrical Conference of 1908, cadmium and cadmium sulphate replace the zinc and THERMOELECTRIC PYROMETER 137 zinc sulphate of the Clark cell; its E.M.F. at 20 C. is 1.0183 and its temperature coefficient to two terms as found by Wolff is: E t = 20 0.04406 (/ 20) o.o 6 95 (* ~ 2 ) 2 - In the portable form of the cell the cadmium sulphate is unsatu- rated. This portable cell has no appreciable temperature coeffi- cient, so that no precautions as to temperature control have to be taken. This cell also recovers rapidly after maltreatment. Its E.M.F. is 1.0187 volts at 20 C., although different cells will differ slightly, i.e., by 0.0005 volt. Hulett has tried using a large-area cadmium cell simultaneously as a battery and stand- ard E.M.F. with considerable success. The values of the E.M.F.'s given above are in international volts, which are legal in the United States and used by the National Bureau of Standards, and are the values effective Jan. i, 1911, as recommended by the International Committee on elec- trical standards. The values previously used for the Clark were 1.434 volts at 15 C., and for the normal Weston, 1.0189 volts at 25, in the United States. 2. A resistance box, or one of the forms of potentiometer of which we shall treat immediately. The former includes a fixed resistance of about 1000 ohms and a series of resistances of o to 10 ohms, permitting by their combinations to realize in this in- terval resistances varying by tenths of an ohm. One may, for greater simplicity, but by sacrificing precision, replace this series of small resistances by a single Pouillet's rheostat having a total resistance of 10 ohms. This apparatus consists of two parallel wires of a meter in length and 3 mm. in diameter, made of an alloy of platinum and 3 per cent copper. 3. A sensitive galvanometer giving an appreciable deflection for 10 microvolts. Since it is placed in the circuit of the couple, and since this is a case of reduction to zero, use may be made here of a Deprez-d' Arson val galvanometer of small resistance. Principle of the Method. If we have an electric circuit con- sisting of a standard cell, or other source of E.M.F. of known value , and a suitable combination of resistances whose total 138 HIGH TEMPERATURES value is R for the whole circuit; and if the thermocouple in series with a galvanometer is connected across a portion r of R so that there is no deflection of the galvanometer, the E.M.F. of the couple is given by the expression A modification of this method eliminating the standard cell in actual work with the couple has its advantages. A storage cell at W (Fig. 32) is in series with a rheostat R and a series of coils or combinations of coils and bridge wire represented by AB. The E.M.F. of the standard cell at E is balanced against Fig. 32. Principle of Potentiometer. that of the battery W by varying R, the points of contact M and M ' being at A and B and the balance indicated by no current in the galvanometer. The standard cell is now replaced at E by the couple whose E.M.F. is to be measured; M and M' are then varied in position until a balance is again obtained; then MM' e = E AB This is the simplest form of potentiometer, of which there are many convenient forms now available for temperature measure- ments. Another modification of this method, eliminating the use of a potentiometer or carefully calibrated resistance box, but requir- ing a calibrated milliammeter and one or more well-known re- THERMOELECTRIC PYROMETER 139 sistances, was first used by Holman in thermoelectric work, and Fig. 33 illustrates the principle. If is a milliammeter and r a small (o.i co) known resistance, R a rheostat with fine adjust- ment, G the galvanometer, and T the thermocouple. The de- flection of G is brought to zero by varying R when the product of the current given by M and the resistance r gives the desired E.M.F. With a series of coils to substitute at r, the range of measurable temperature may be indefinitely extended. The pre- cision of this method is limited by that of the milliammeter M . Siemens and Halske sell a convenient form of this apparatus as devised by Lindeck of the Reichsanstalt. Fig. 33. Holman's Method. Various other special forms of apparatus for the exact measure- ment of thermocouple E.M.F.'s have been devised, but they are all modifications, more or less complicated, of the above. We shall treat of some of them under potentiometers. Potentiometers for Use with Thermocouples. Although the galvanometric method is suitable for many technical thermo- electric measurements of temperature, it is generally necessary to resort to potentiometric methods when an accuracy of 10 C. or better is required, as is the case in many laboratory operations. This exact work is usually best done with thermocouples of the 140 HIGH TEMPERATURES platinum metals, so that the problem of best design of poten- tiometers for temperature measurement is quite a definite one. The need of sufficiently sensitive and accurate devices for the measurement of small E.M.F.'s in thermoelectric pyrometry has acted as an incentive for the great improvement, in recent years, of apparatus suitable for this purpose, and there are now avail- able a considerable number of potentiometers meeting the re- quirements for very exact temperature measurements by this method, as well as less costly instruments giving an accuracy between that obtained with the galvanometer method and the more elaborate potentiometric installations. R R' BI 1 WWWW VWMM/j To Thermocouple scfc Fig. 34. Potentiometer Indicator Circuits. The potentiometer indicator of Leeds and Northrup, shown in Fig. 34, illustrates a type of instrument of intermediate precision, but without the disadvantages of the galvanometric method, it being possible to get results to about 3 C. with this apparatus, using Pt-Rh thermocouples. This indicator consists of a Weston standard cell, a secondary dry battery, and a galvanometer connected up as a potentiometer, the whole being mounted in a box of convenient size, making a portable testing outfit (Fig. 35). The dry cell is continuously on the closed potentiometer circuit ABCF, which includes the two regulating rheostats R and R' and a fixed resistance S. The current in the potentiometer circuit is adjusted by changing R THERMOELECTRIC PYROMETER 141 and R' with the key at SC until the galvanometer shows no de- flection. Pressing the key to TC, the pointer G is set on the slide wire DE, calibrated in millivolts, until again the galvanom- eter remains undeflected, indicating a balance in the thermo- couple circuit. Precision Requirements. Some of the requirements which must be met in potentiometer construction we may emphasize. For work to 0.1 C. with Pt-Rh couples, for example, we must Fig. 35. Potentiometer Indicator. have a sensibility of i microvolt (millionth of a volt) throughout the range of the instrument, which may be of 20 millivolts, neces- sitating an accuracy of i in 20,000 in all adjustments affecting the final value of the E.M.F. Contact or thermal E.M.F.'s, such as develop even for slight temperature differences in the various parts of such an apparatus, are to be avoided in the electric circuits, as far as possible, by proper choice of materials, design, and method of manipulation; for example, using thin metal contacts, putting sliding contacts in battery circuit, and working with the galvanometer circuit closed. In order to in- 142 HIGH TEMPERATURES crease the sensibility and permit the use of moving-coil galva- nometers of reasonably attainable behavior, it is necessary to keep down the total resistance of the potentiometer. This causes the contact resistances of the adjustable parts, such as the dials, to become of importance, and a very exact and somewhat com- plicated mechanical construction is required to eliminate this source of error. It appears to be practically necessary, in design- ing a potentiometer, to choose between some contact resistance or some thermal E.M.F. For rapid work, it is desirable that the potentiometer circuit be so designed that the standard cell may be checked up without disturbing the potentiometer circuit, and similarly it is advan- tageous to be able to change the range of the potentiometer without disturbing the regulating rheostats or rechecking the standard cell. Sometimes, also, the final figure in E.M.F. is obtained from the galvanometer deflection, in which case it is convenient to make provision for a constant galvanometer sen- sibility for all E.M.F.'s, which may be effected by auxiliary resistances in the galvanometer circuit. A very important matter is that of insulation, or the prevention of leaks from one part of the potentiometer circuit to another (internal leakage) and from the outside to or from this circuit (external leakage). The former becomes less important with low-resistance potentiometers. The latter effect becomes par- ticularly menacing when the thermocouple is immersed in an electrically heated furnace. It can be overcome by interposing wire-connected equipotential shields made of metal between the measuring system and all external sources of E.M.F., or by reversal of the heating or other suspected circuit and taking the mean of the potentiometer readings. Most of the potentiometers in use are, in part at least, slide- wire instruments, but for the very highest accuracy it is ad- visable to use the more costly dial construction throughout. As we shall see, potentiometers suitable for the thermoelectric or resistance measurement of temperature may now be obtained, provided with five dials and reading accurately to o.i microvolt, THERMOELECTRIC PYROMETER 143 or to considerably better than any thermocouple can be depended upon at high temperatures. An inconstant battery is troublesome, and in exact work it is necessary to pay particular attention to this point in addition to frequently checking against the standard cell. Accumulators of considerable volume, or several so connected as to give a minimum change of E.M.F. with time, should be used; and it is well, since so little current is taken from the battery, to have it constantly closed through its potentiometer circuit. The battery may also be advantageously inclosed and packed to obviate temperature changes which may cause fluctuations in its E.M.F. of sufficient magnitude to be troublesome in work of high pre- cision. Some care has to be exercised in the choice or design of the 'galvanometer to be used with precision potentiometers. For work to 0.1 C. with platinum thermocouples, it is necessary to have an appreciable deflection for i microvolt with the galva- nometer in circuit, and the design should be such that the de- flection is aperiodic when the galvanometer is used with a given potentiometer. For rapid work, as in taking cooling curves, the period of the galvanometer should be kept down; and if, besides, the last increment of E.M.F. is to be measured by the galva- nometer swing, it is desirable to have a period of not over five seconds. These requirements, combined with freedom from thermoelectric effects, are very severe for the swinging-coil type of galvanometer and can be met only by the more skillful con- structors of such instruments. Types of Thermocouple Potentiometer. The Cambridge ther- mocouple potentiometer, similar in design to that of Harker, is an instrument designed for measuring E.M.F.'s of 30 millivolts or less. By estimation, microvolts may be read, corresponding to about 0.1 at 1000 C. with Pt-Rh couples. The circuits of this potentiometer are shown diagrammatically in Fig. 36. The total resistance in the circuit is arranged to give a fall of potential of about i volt per 50 ohms, and the resistances B.C. (about 42.5 co) and s.c. (about 51 o>) are adjusted to give a fall of 144 HIGH TEMPERATURES potential from M to N on the slide wire ss equal to the E.M.F, of a cadmium cell C. This potentiometer is operated as follows: With N set at the known value of the standard cell C, and the key k thrown to cc r putting C in opposition with the storage cell B, the resistances RI and R 2 are adjusted until the galvanometer G shows no de- flection on tapping the key. The battery B is then substituted for the cell C by throwing k to the side xx for the determination of the unknown E.M.F., X. The balancing of X against B is made by setting the dial, or series of millivolt coils, MVC r and the pointer Q on the slide wire V V, until as before the galva- nometer shows no deflection on pressing the key. The value Fig. 36. Cambridge, SThehflocouple Potentiometer. ^\ of X is then given directly in millivolts by adding the readings of MVC and Q. This is effected by making the dial MVC of 29 coils each of 0.05 co, giving on the basis of i volt per 50 ohms a pressure diagram of i millivolt on each section. Similarly, the resistance of the wire VV being 0.06 ohm, the fall of potential along its length is 1.2 millivolts, or the maximum E.M.F. measurable is 30.2 millivolts. This range will take in most base-metal thermocouples as well as the usual platinum couples. In order to minimize thermal E.M.F. 's and tempera- ture coefficients, all coils are of manganin and all connections of copper. THERMOELECTRIC PYROMETER 14$ The Leeds and Northrup thermocouple potentiometer repre- sents another, if somewhat similar, solution of this problem to about the same degree of accuracy. The arrangement of circuits is shown in Fig. 37. By means of the plug at A the range of the instrument may be increased tenfold. The heavy slide wire possesses eleven turns and permits reading to better than i microvolt with a suitable galvanometer. The resistance of each, of the seventeen millivolt coils is 0.5 ohm, giving with the slide wire a total of about 9 ohms in the main circuit. Ba. I !> i O S C yj~j fflCr U QE.M-F.Q OcenO Fig. 37. Leeds and Northrup Thermocouple Potentiometer. In both of the above instruments, settings on the standard cell may be made without disturbing the battery circuit, and the range and sensibility of either may be increased at will by suit- able devices which may conveniently be built into the instru- ments. There are also numerous other potentiometers, such as those of Siemens and Halske, Carpentier, and Wolff, based on similar methods of operation. This type of instrument is not entirely free from internal thermoelectric forces, but these may be practically eliminated by proper reversals in the circuits. The Diesselhorst potentiometer, built by O. Wolff of Berlin, is based on quite different principles from the preceding and 146 HIGH TEMPERATURES represents an attempt to attain the highest accuracy possible in such apparatus, o.i microvolt being measurable with exact- ness. It is a five-dial instrument of very low resistance, and combines principles of construction suggested by several writers including Hausrath, White, and Diesselhorst. Thermoelectric effects in the main potentiometer circuit are eliminated by the design of the instrument, and temperature coefficient changes in the coils may be avoided by oil immersion. The effects of contact resistances are eliminated only by the excellence of construction. As constructed, this potentiometer possesses the disadvantages usually common to split-circuit instruments in which the range is altered by changing the? battery current, such as requiring the adjustment of the rheostat in the battery circuit and rebalancing against the standard cell whenever the range of the instrument is changed. This is prohibitive for the rapid intercomparison of considerably different E.M.F.'s such as is often required in temperature measurements, unless sensi- bility is sacrificed. White has developed a dial potentiometer suitable for thermo- electric work of high accuracy, in which, however, the last two dials are replaced by the galvanometer deflection, necessitating a construction, which has been realized, giving constant galva- nometer sensibility. White has also realized a double potenti- ometer permitting alternate and independent measurements of two rapidly varying E.M.F.'s with all the advantages of two instruments, but with the accessories of only one. Finally, Wenner has suggested a modification of the potenti- ometer circuit suitable for the measurement of low E.M.F.'s, consisting in shunting by a comparatively high resistance a part of the circuit including the potential point of a dial. By means of a double-dial switch both branch points between the shunt and the main circuit may be shifted in steps of equal resistance so as to introduce a larger or smaller resistance in the dial while keeping the resistance shunted constant. In Fig. 38 is shown a plan for this potentiometer for use with thermocouples. THERMOELECTRIC PYROMETER 147 The dial contacts are all in the battery circuit, each branch of which is of comparatively high resistance, so that the resistance of the contacts and thermoelectromotive forces due to the setting of the dials have only a very small effect. The compensation circuit, on the other hand, is of low and nearly constant resistance, which makes it possible to use a galvanometer having a high voltage sensibility and permits the reading of a small unbalanced electromotive force from the deflection of the galvanometer (G). Fig. 38. Wenner's Design. The effect of thermoelectromotive forces in the galvanometer is much reduced by keeping the circuit closed and the resistance approximately independent of the position of the galvanometer key. Under these conditions a change in the deflection of the galvanometer following a change in the position of the key (K) signifies an uncompensated electromotive force independent of any fairly constant electromotive force in the galvanometer. The question of best design of precision potentiometers for use with thermocouples may be said to be in a state of flux, and no single best instrument meeting satisfactorily all the conditions imposed above has yet appeared in practical form. The Thermocouple Circuit. For good working of the plati- num thermocouple there are certain practical precautions to be taken, which we shall consider. Most of these remarks apply even with greater force to the base-metal couples. 148 HIGH TEMPERATURES Junction of the Wires. The contacts of the different parts of the circuit should be assured in a positive manner; the best way is to solder them. Binding screws often work loose in time, or the metallic surfaces in contact become oxidized. The im- portance of this precaution varies with the conditions of the experiments; one can dispense with it for experiments that last only a few hours, because there is little chance that the con- tacts will become modified in so short a time; soldering is, on the contrary, indispensable in an industrial installation which may be used for months without being tested anew. But in any case, the soldering together of the two leads of the couple is absolutely indispensable. It is quite true that the electromotive force is independent of the manner of making contact. The two wires twisted together or soldered will give at the same temperature the same electromotive force. But under the action of heat the twisted parts are soon loosened, and there result bad contacts which increase the resistance of the whole circuit. In general, this accident is not noticed until the untying is almost complete, so that one may make before this a whole series of false measurements without being warned. The best method of soldering is the autogene junction by direct fusion of the wires of the couple; it is necessary, in order to effect this, to have oxygen at hand. One commences by twisting the two leads together for a length of about 5 mm., and they are then clamped above an oxy hydrogen blast lamp. Oxy- gen is admitted through the central tube, and gas through the annular space ; the oxygen is allowed to flow in normal quantity, and the gas in feeble quantity, then one opens progressively the gas cock. At a certain instant one sees the extremities of the wires melt, giving off sparks; the gas is then shut off. If one waits too long, the junction will melt completely and the two wires separate. With a little practice a good junction can be made by touching together, in the oxyhydrogen blast, the two untwisted wires held in the hand. In default of oxygen, the wires may be soldered with palladium, which can be melted by means of a blast lamp furnished with air, THERMOELECTRIC PYROMETER 149 taking care to reduce the action of radiation. A hole is cut in a piece of charcoal in which is placed the junction of the two wires twisted together after having wound about it a wire or a small strip of palladium, and the flame of the lamp is then directed upon the junction. In the cases in which the couple is not to be used above 1000, and only in these cases, the soldering may be done still more simply by the use of gold ; the ordinary Bunsen flame is sufficient to make this junction. Annealing. Before use, even with new couples which are usually hard-drawn, the wires of the couple should be rendered as homogeneous as possible by annealing them electrically. For the platinum couples of 0.6 mm. diameter, which are in com- mon use, a current of 14 amperes usually suffices. The cur- rent is kept on until the wires glow uniformly. In the case of couples that have been used, bad spots are easily detected in this way, and should be cut out if the glowing does not remove them. Insulation and Protection. The two leads should be insulated from one another throughout their length. For this, use may be made in the laboratory of glass tubes or pipestems, or of thread of pure asbestos wound about the two wires, by crossing it each time between the two (Fig. 60) so as to make a double knot in the form of an eight, each of the wires passing through one of the loops of the eight. This is a convenient method of insulation for laboratory use, although ordinary asbestos is likely to contain impurities which will damage the couple. The two wires with their envelope form a small rod of considerable rigidity which is easily slipped into apparatus. With this arrangement it is im- possible to go above 1200 or 1300, at which temperature asbes- tos melts. The most satisfactory insulation, however, is had by means of thin tubes of hard porcelain standing 1500 C. and of Marquardt mixture, 1600, obtained from the Royal Berlin Porcelain Works. For industrial installations, use may be made of small fire-clay cylinders of 100 mm. in length and 10 mm. in diameter, pierced HIGH TEMPERATURES Fig. 39. Parvillee's Mounting. in the direction of the axis by two holes of i mm. diameter, through which pass the wires, or hard porcelain tubes may be used. One or another of the other forms of insula- tor is added in sufficient numbers. They are placed, according to the case, in an iron tube or in a porcelain tube. The porcelain tube should be employed in fixed installations in which the temperatures may exceed 800. One may, as does Parvillee in his porcelain furnaces (Fig. 39), place the porcelain tube in the lining of the furnace in such a way that its end is flush with the inner surface of the lining. An open space of a decimeter cube is cut in the lining about this extremity of the tube. This method makes easier the establishment of temperature equi- librium without subjecting the tube to too great chances of breaking by accidental blows. The iron tube is used for temperatures not exceed- ing 800, in the lead baths serving to harden steel for example, and for movable couples which are exposed to heat only during the time necessary to take the observations. In this case the junction is placed some 5 cm. beyond the insulators and the iron jacket. The wires take up the temperature within 5 seconds, and the observation can be taken before the tube becomes hot enough to be burned, even in furnaces for steel whose temperatures exceed 1600, and before the wires have had time to be altered even in strongly reducing flames. The other extremity of the iron tube carries a wooden handle (Fig. 40) where are located, outside, the binding posts for the galva- nometer leads, and inside an extra length of wire for the couple to replace portions burned or broken off. The figure shows one arrangement of this handle. In all cases in which the furnace whose temperature it is de- Fig. 40. Opened Wooden Handle. THERMOELECTRIC PYROMETER sired to measure is under a reduced pressure, suitable precautions must be taken to prevent any permanent entrance of cold air by the orifice necessary for the intro- duction of the tube, as well before as during an observation; otherwise one runs the chance of having in- exact results. In the case of prolonged observa- tions in a reducing atmosphere or in contact with melted bodies, as the metals capable of altering the plat- inum, the couple should be protected by inclosing it in a covering imper- meable to the melted metals and to vapors. For fixed installations in industrial works, use should be made of a porcelain tube, or one of iron, closed at the extremity where the junction is located; in this case the dimensions of the tube are unim- portant. Quartz or porcelain tubes with an iron tube furnish oftentimes a very permanent and satisfactory sheathing. Fig. 41 shows one form of mounting for a protected couple attached to its galvanometer. For laboratory investigations it is often indispensable, on the contrary, to have around the wires a covering of as small diameter as possible If it is simply a question of protect- ing the couple against the action of non-volatile metals, the simplest way is to use, as did Roberts- Austen, a paste sold in England under the name of Purimachos, which serves to repair the cazettes employed in molding. We 152 HIGH TEMPERATURES have made an analysis of this which gave the following compo- sition after desiccation at 200: Alumina and iron 14 Soda 3.2 Water 2.6 Silica (by difference) 80 . 2 It is a very finely powdered quartz to which is added 10 per cent of clay, and diluted with a solution of silicate of sodium. To use it, the matter is diluted so as to form a thick paste, and the couple is dipped in it the required length, arranging the wires parallel to each other at a distance apart of about i mm. The whole may then be dried and calcined very rapidly, with- out fear of snapping the covering, as would happen with clay alone; but this covering is not sufficiently impermeable to protect the couple against the very volatile metals, as zinc. It is better, in this case, to use small porcelain tubes of 5 mm. inside diameter, i mm. thickness of wall, and 100 mm. long, straight or curved, according to the usage to which they are to be put. The couple insulated by asbestos thread, or by a small inner porcelain tube of i mm. inside diameter, as has been said pre- viously, is pushed down to the bottom of the tube. If one has not at hand such tubes of porcelain, and it is required to make a single observation at a temperature not exceeding 1000, as, for instance, a standardization in boiling zinc, one may use a glass tube. It melts and sticks to the asbestos, which holds a thick enough layer to itself to protect the platinum. But, on cooling, the tube breaks, and it is necessary to make a new set-up for each operation. This is not practicable for continuous observations. Fused quartz is now obtainable for insulating thermocouples and for containing sheaths. This material gradually crystallizes and crumbles above 1200, and in the presence of a volatile reduc- ing agent, as graphite or hydrogen, volatile silicides are formed above 1200 C., which will destroy platinum. Some types of industrial mountings used by Heraeus for platinum thermo- couples are shown in Fig. 42. THERMOELECTRIC PYROMETER 153 orcelain Tube Metal Porcelain Quautz Glass Double Steel Tube Graphite Fig. 42. Herseus' Thermocouple Mountings. 154 HIGH TEMPERATURES Cold Junction. In general, in a thermoelectric element, one distinguishes the hot junction and the cold junction. The latter is supposed kept at a constant temperature. In order to realize rigorously this arrangement, three wires are necessary, two of platinum and one of an alloy connecting two junctions. This theoretical arrangement is practically without interest, and the second junction is always dispensed with. If, in fact, the tem- perature of the whole circuit exclusive of the hot junction is uniform, the presence or the absence of the cold junction does not affect the electromotive force; if this temperature is not uniform, the second junction is not advantageous, for there is then in the circuit an infinity of other junctions just as important to consider: the junctions of the copper leads with the platinum wires, those of the galvanometer leads and of the different parts of the galvanometer among them- selves. One must satisfy himself as well as may be as to the uniformity of temperature. in the cold circuit, and rigorously of the equality of temperature between corresponding junctions, particularly those of the two platinum wires with the copper leads. These uncertainties in the temperature of the cold junctions are an important source of error in the measurement of temperatures by thermoelectric couples, but for ordinary practice they are easily eliminated. In order to realize exact measurements, pre- cise to i, for instance by the galvanometer method, it will be necessary to have completely homogeneous circuits, includ- ing the galvanometer, with the single exception of the junc- tions of the platinum wires with the conducting leads; these should be immersed in the same bath at constant tempera- ture. It would be necessary for this that the constructors of galvanometers limit themselves to the use of the same kind of wire for all parts of the apparatus, wires of the coil, suspending wires, leads, and parts of the coil. That is difficult to obtain. In the standardization of thermocouples for exact work, it is customary to immerse the cold junctions, i.e., the points THERMOELECTRIC PYROMETER 155 of contact of the copper leads and platinum-metal wires, in an oil bath in ice. With the potentiometer, irregularities due to other sources of E.M.F. in the circuit are eliminated by reversing simultaneously the battery current and the couple circuit. The Cold-junction Correction. In work of high accuracy with platinum couples and when the potentiometric method of measurement is used, the cold-junction correction should be experimentally eliminated by keeping the cold junctions at a constant temperature, most conveniently at o C. When the galvanometer method is used, it is often not convenient to keep the junctions of the couple to the lead wires of the galvanometer at a definite temperature, although the galvanometer itself may be so removed from the fur- nace that its temperature changes are slight. Except in the roughest kind of work, allowance has to be made for the cold- junction temperatures, which may be measured by an auxiliary thermometer. Calling / the cold-junction temperature for which the instru- ment reads correctly, / the observed temperature of the cold junction, the correction to apply to the observed temperature readings of the galvanometer, otherwise supposed to read cor- rectly for a given thermocouple, usually lies between f (/ / ) and (/ /o), depending on the type of couple and the tempera- tures of both hot and cold junctions. This question has been treated in detail for several types of couple by C. Otterhaus and E. H. Fischer. That this correction depends in general upon both hot and cold junctions is due to the fact that the E.M.F.-temperature curve is not a straight line (see Fig. 24). The correction factor by which to multiply (t to) is numerically equal to the ratio of the tangents of this curve for the hot- and cold-junction temperatures. As an example, we may compute the corrections to apply for a Pt, 90 Pt-io Rh Heraeus thermocouple using the E.M.F. data of Day and Sosman (page 114). 156 HIGH TEMPERATURES CORRECTIONS FOR COLD JUNCTION (Pt, 90 Pt-io Rh). Temperature of hot Correction factor for cold junctions near: junction. 20 40 100 C. 0.76 0.81 0.86 200 -65 .68 73 300 .60 63 .68 400 57 .61 65 500 55 59 63 600 54 57 .61 700 53 55 59 800 54 57 900 49 52 55 IOOO 48 50 53 I2OO 46 49 Si I4OO 45 -48 50 1600 45 .48 50 It is to be kept in mind that the E.M.F. indicated by a direct- reading galvanometer is a measure of the difference in tempera- ture between the hot and cold junctions. If the galvanometer needle be set at zero, which is a convenient way of working, this zero reading will correspond to the temperature of the cold junction at the start; therefore the true temperature is obtained by adding to the observed temperature reading a quantity corre- sponding in millivolts to the cold-junction temperature, obtained as already explained. The starting point / in the above is of course the temperature at which the cold junctions were kept during the original calibration, often o or 20 C. Thus, if the cold junction is at 25 C. and the hot at 500, this correction, from the above table, is +0.60 (25 o) = +15, if the couple was calibrated from o C., and the galvanometer read zero for a cold- junction temperature of 25. It is a simple procedure, and usually sufficiently exact when the temperature scale of the galvanometer corresponds approxi- mately to that given by the thermocouple, to set the pointer of the galvanometer at the position on its scale corresponding to the temperature of the cold junctions. The readings of the gal- vanometer, otherwise corrected, will then give temperatures. Elimination of Cold-junction Changes. The Bristol base- metal thermocouples are provided with extension pieces of the same composition as the fire end, permitting the cold junction to THERMOELECTRIC PYROMETER 157 be removed to a place of slight temperature change, as near the floor, and this arrangement also facilitates the convenient renewal of the short, heavy fire ends of these couples when they have to be discarded. Bristol has also devised an automatic compensator for cold-end temperatures, shown in Fig. 43, consisting of a small glass bulb and capillary tube partially filled with mercury, into which a short loop of fine platinum wire dips. This is inserted in the thermoelectric circuit close to the cold junction, Changes in Thermo-electric Couple t Compensator Fig. 43. Bristol's Compensator. temperature cause the mercury to expand or contract, cutting in or out resistance in the circuit. This acts in opposition to the change in E.M.F. with temperature at the cold end, so that a balance may be established if the parts are properly designed. In the Thwing instruments, the elimination of the temperature variations of the cold ends of the couple, where they can be brought close to the galvanometer, is affected by a device con- sisting of a compound strip of two metals having unequal coeffi- cients of expansion, so attached to the spring controlling the pointer that the reading of the galvanometer when no current is flowing is the temperature of the surroundings. 158 HIGH TEMPERATURES 1 In many industrial estab- lishments running water of practically constant tem- perature is available, and the cold end of the thermo- couple can then be water- jacketed and so kept sufficiently constant in temperature, as shown in Fig. 44, which represents an arrangement for this purpose as constructed by Hartmann and Braun. The movable arm can be swung out horizontally when the thermocouple is to be immersed. Paul provides an attach- ment by which an inexpen- sive supplementary couple, with one end water-cooled, is placed in series with and in opposition to the main thermocouple by means of nonreversible plugs which fit into sockets in the head of the pyrometer cane. The temperature difference then indicated by the in- strument is that between the fire end and water- cooled end. A Breguet spiral, to which one end of the con- trol spring of the milli- voltmeter is attached, has been devised by C. R. Darling and Fig. 44. Water-jacketed Cold Junction. THERMOELECTRIC PYROMETER sold by Paul. In this way the zero of the instrument is made to vary with its temperature. The Crompton Company provide their instruments with a multiple scale (Fig. 45), which allows for the cold-junction tem- perature variations. Finally, the cold end of the thermocouple may be buried in a box underground, for instance, and copper wires run to the galvanometer. We shall mention other such devices under the heading " Compound Thermocouples " and when discussing accessories to recorders. Fig. 45. The Crompton Scale. Constancy of Thermocouples. This matter is of the greatest importance in thermoelectric measurements both in the labora- tory and in the works, as there is nothing more aggravating than the gradual deterioration of a product due to insidious, and often unnoticed until too late, changes in the controlling apparatus. The behavior of thermocouples made of platinum and its alloys has been studied in great detail, from this point of view, by several observers, but the data are somewhat contradictory. If a thermocouple, however well protected, is heated for a long time at a high temperature, its E.M.F. will change. It is well for accurate work to have at least two thermocouples, one of which is kept as a standard and only occasionally heated, and never above 1200 C. In this way changes in the couple ordi- narily used may be readily detected. Holborn, with Henning and Austin, has made a very complete study of the effects of continued heating in various atmospheres on the loss of weight i6o HIGH TEMPERATURES and changes produced in electric and thermoelectric properties of the platinum metals. The following table shows the results of continued heating in air on the E.M.F. of the platinum couples ordinarily used: EFFECT OF PROLONGED HEATING ON E.M.F. E.M.F. AGAINST PT IN MICROVOLTS. Duration of heating, hours. 90 Pt 10 Ir. 700 C. 900 1100 1300 O 6 8 i 9 12 9460 9160 8840 16,540 15,450 14,780 14,300 19,740 18,530 17,640 17,050 12,450 H,930 11,560 90 Pt 10 Rh. 800 C. 900 1000 1100 7230 7250 7270 7280 7290 8340 8380 8400 8410 8420 9480 9510 9540 9540 9550 10,670 10,690 10,720 This investigation shows that the E.M.F. of a couple, and thus the indicated temperature, changes with continued heating, very considerably for a Pt-Ir couple and about 0.5 per cent for a Pt-Rh couple for ten hours' heating. The change is greatest during the first part of the heating. The observed in- crease in E.M.F. of the Pt-Rh is difficult to explain unless it be due to distillation of iridium from the heating coil, as shown from Day and Sosman's work. Before use, a thermocouple should be annealed by passing a current through it at a white heat, when future changes will be slight if used in an oxidiz- ing atmosphere. This annealing also will restore to very nearly its normal value the E.M.F. of couples which have been in contact with silicates. Changes in temperature distribution along the wire may also affect the apparent electromotive force of the couple, causing THERMOELECTRIC PYROMETER l6l apparent changes in temperature as great as 20 at 1000 C. with some wires. The less homogeneous the wires the more marked is this effect. In the most exact work, therefore, the same con- ditions of immersion must be followed throughout, or the result- ing changes in E.M.F. measured. It follows from all this, as Holborn and Day state, that the temperature scale, once established by means of the thermo- couple, can be maintained with certainty only with the help of fixed temperatures such as the melting points. Dr. W. P. White draws particular attention to the importance of that region of the wires passing through a steep temperature gradient and the great influence that inhomogeneity in this portion of the wire may have upon the temperature readings with thermo- couples. If we consider an inhomogeneous thermocouple composed of short segments each of which is supposed homogeneous, at any junction of two segments there is developed an E.M.F. pro- portional to the temperature t and their difference in thermo- electric power A/7, or for the whole circuit: E = (fc-AFi + fe -A# 2 + . / n -AT n ) = S*-A#. It is evident that those portions of the circuit at constant tem- perature and of homogeneous material (&H = o) do not con- tribute to the value of E; but in the regions of temperature gradient of an inhomogeneous wire, the errors due to inhomo- geneity depend also upon the temperature distribution along the wire. If an inhomogeneous thermocouple, therefore, is raised or lowered in a furnace at constant temperature, the reading of the couple will change. In view of these facts, it is important that those portions of the thermocouple wires pass- ing from cold to hot regions be chemically and physically of uniform properties. The effect of initial chemical inhomogeneity, for the platinum thermocouples, appears to be either negligible or very small, but may be considerable for base-metal couples. The region between hard-drawn and annealed wire is one of marked physical inhomo- 162 HIGH TEMPERATURES geneity. This can .be, and should always be, eliminated by annealing the wires of most couples including platinum, pref- erably with an electric current, or by hardening in the case of constantan. The most troublesome source of inhomogeneity, however, the most difficult to remove, and the source of greatest error, particularly with platinum thermocouples, is that due to contamination from evaporation and diffusion of metal vapors into the region of temperature gradient of the platinum wire. The oxide coating which forms on some metals is also a similar source of uncertainty in the regions of variable temperature. Regarding the effects of contamination of platinum wires,, carbon, illuminating gas, and other reducing agents appear to act only through their reducing action on other substances, such as iron and silicon, capable of injuring the platinum. In an oxidizing atmosphere, iron oxides and silicates produce little or no effect; but metals such as iridium and rhodium, particularly the former, which is very volatile even from a Pt - Ir alloy wire at 900 C., and capable of alloying with platinum, will, if present,, produce marked contamination of platinum wires. Amputation of the contaminated portions appears to be the only remedy in this case. Excessive local heating will also cause inhomogeneity to develop, especially in the alloy wire, probably due in part to evaporation and in part to crystallization. Measurement of Inhomogeneity. This is very easily and ex- actly effected, and should be carried out on any thermocouples to be used in work of high accuracy. Each of the wires is tested separately, its ends being kept conveniently at a constant tem- perature of o. The wire is in circuit with a sensitive galva- nometer graduated in microvolts and is passed through a short electric resistance furnace kept at a constant temperature, 1000 or 1400 C. The readings of the galvanometer are taken for different positions of the furnace along the wire. The furnace may be replaced conveniently by a short length of porcelain heated by a Bunsen burner. Using a sensitive recorder, a small furnace may be pulled automatically along the wire which records its own variations. Such a device is in use at the Bureau of THERMOELECTRIC PYROMETER 163 at j 'H '3 164 HIGH TEMPERATURES Standards. Another method is that shown in Fig. 47, which permits a point-to-point study of the phenomenon. In Fig. 46 are shown the homogeneity curves, taken by the first method, of the wires of two thermocouples, the one (Pt- Examination . Rh) new and of fresh, pure Fig. 47- Testing of Homogeneity. materialS) ^ other ( p t _j r) old and impure. These methods are easily sensitive enough to differentiate the various grades of platinum wire used in thermo- couple manufacture. We have already called attention to the inhomogeneity of base- metal thermocouples. We shall return to this question in a later paragraph. Reproducibility of Thermoelectric Apparatus. It is often of considerable convenience in all kinds of measurements, especially on a large scale with numerous working units of the same kind, to be able to duplicate or replace corresponding parts without having resort to new calibrations. This is equally desirable in temperature measurements, and in recent years there has been a large measure of success in the attempt to produce thermo- couples and manufacture pyrometer galvanometers which are interchangeable . As an example of the former, we may cite the case of the well- known 10 per cent platinum-rhodium normal thermocouples of Heraeus. They are reported to have maintained the following constancy for the past six years: AVERAGE E.M.F. AT 1000 C. . Year. Millivolts. 1904 9-52 9-53 1906 9 . 53 J 907 9-57 1908 9.59 9-55 Regarding the interchangeability of pyrometer galvanometers, the art of electrical instrument manufacture has so advanced in THERMOELECTRIC PYROMETER 165 recent years that equivalent instruments, in error absolutely and with respect to each other by less than 10 C. throughout their scales, are produced currently by several makers. Base-metal Thermocouples. There appears to be an insistent demand, on the part of many in charge of technical processes requiring temperature control, for inexpensive and robust measur- ing apparatus. For this reason, if for no other, the use of the base-metal thermocouple has become firmly established. Its success has been due to several causes, principal among them being the production of fairly satisfactory alloys of high E.M.F. with temperature change, which can be made into practically unbreakable pyrometric canes of very low resistance; and the simultaneous development of pivot millivoltmeters suitable for use as galvanometers with this type of couple. Fig. 48. Heavy Base-metal Welds. Such canes can be had, for example, with a resistance as low as 0.05 ohms cold, increasing by only o.oi ohm when heated to 1000 C. Even with a commercial milli voltmeter of only i ohm resistance, the calibration will remain constant, under these conditions, to 10 C. for any depth of immersion of the couple (see page 1 20) . It is indispensable in such cases that all junctions be of negligible resistance, and they are preferably soldered. In Fig. 48 are illustrated two types of weld for base-metal couples. A very considerable error, however, in the estimate of the temperature of regions into which are thrust thermocouples of considerable cross section and insufficient length, may arise from heat conduction along the pyrometer, the effect being to chill the l66 HIGH TEMPERATURES hot junction below the temperature of its surroundings. It is important that such pyrometers be calibrated to allow for this effect. The thermoelectric power of many of these base-metal couples is over 20 microvolts per i C., and some of them are 40 or more as compared with 10 microvolts per i C. for the ordinary plati- num-rhodium couple. The low-resistance pivot galvanometers of 20 or 40 millivolts range, suitable for use with the base-metal couples, may be made more cheaply and robust than instruments suitable for use with platinum couples for the same sensibility. As the base-metal couples receive hard usage, are cheap, and often require frequent replacing, it is of great advantage to make use of material of uniform thermoelectric properties, so that burned-out " fire ends " may be readily replaced as required without retesting. It is of course safer, and necessary in many instances, to calibrate each new fire end, either by comparison with a standard, or, as may be done conveniently even in tech- nical plants, by taking the reading in salt baths of known freezing points (see page 190). In the case of the use of alloys or metals possessing critical regions, accompanied by the absorption or liberation of heat, it should be emphasized that discordant results may be obtained on reheating, the actual E.M.F.- temperature relation depending upon the internal structure of the material of the couple, and this in turn upon the rate of heating or cooling through these critical regions. These effects are particularly marked in couples of considerable size, and are enhanced by varying the depth of immersion of the couple in the test bath or furnace. We shall mention later some specific instances of these effects. Although a very considerable number of base-metal thermo- couples have been put upon the market in recent years, there appears to be very little certain knowledge available as to the exact composition, thermoelectric properties, and behavior of most of them, some of which are quite complex alloys, as for example of Ni, Cr, Al, and Cu. It should perhaps be noted that our use with base-metal couples of the adjectives " constant," THERMOELECTRIC PYROMETER 167 " reproducible/' etc., is not to be taken in the same rigorous sense as for the platinum couples. Nickel-copper. Various combinations of these metals have been used and recommended for temperatures as high as 900 C. According to the investigations of M. Pecheux, the most satis- factory one seems to be that with the alloy constantan (known also as "advance"), 60 Cu 40 Ni, as one wire and pure Cu as the other. The E.M.F.-temperature curve for this thermo- couple approximates a fairly flat parabola, but appears to require an equation of the third or fourth degree in / to express the results with some exactness to 900 C.; or this interval may be divided into three, each of which is represented, to a fraction of a degree, by a parabola of the form E% = a + bt + ct 2 . Between o and 250 C. the numerical equation is roughly E* Q = 40 / + 0.03 / 2 in microvolts for the copper-constantan thermocouple. Those with a smaller percentage of nickel have less flat curves, lower thermoelectric powers, and guard their original calibration less well than does the copper-constantan couple. The limiting case in this series, that of pure nickel against pure copper, is of some interest, as it gives a good illustration of the effects of molecular transformation on thermoelectric behavior. Nickel undergoes such transformation between about 230 and 390 C., which causes both its electrical resistance and thermo- electric power to depart from their normal trend in this region. These effects are shown in Fig. 49, the data on resistance being from some measurements made by Somerville, on nickel wire, and on thermoelectric power of the Ni-Cu couple from the observa- tions of M. Pecheux. When, for such a couple, or any thermocouple in which nickel or any substance possessing regions of molecular transformation, its rate of heating or cooling is varied, the E.M.F. readings of the couple will not in general be the same for a given temperature within this region; and for rapid cooling, in some cases, especially for wires or rods of considerable diameter, the E.M.F.-tempera- ture relations may be changed at all temperatures below this region as well, due to the retardation or partial prevention of i68 HIGH TEMPERATURES the complete transformation by chilling. Reannealing and slow cooling will oftentimes restore the original annealed condition. The importance of annealing all such couples before their first calibration becomes apparent from the above considerations. The presence of impurities appears to be a further source of considerable uncertainty in the constancy of these couples with continued use, it being noted, by Pecheux for example, that couples with very pure nickel remained more constant in use than those with the less pure metal. NICKEL RESISTANCE -H--f THERMOELECTRIC POWER AGAINST COPPER o o o 500 420 340. 260- 180 100 100 300 300 400 500 600 700 800 900 1000 Temperature Centigrade Fig. 49. Resistance and Thermoelectricity of Nickel. The addition of zinc in any proportions to the copper-nickel alloy, giving the German silvers, appears to be detrimental in all respects. The upper limit of 900, assigned by Pecheux for the copper- constantan couples, is better replaced by 600, or even less, for continued use with any considerable precision ; and even at 600 both wires will oxidize and soon become fragile. Nickel-iron. We have already called attention to the be- havior of iron wires and the enormous parasite E.M.F.'s that they may develop. Nevertheless, a favorite industrial thermo- electric combination has been, and still is in some quarters, a THERMOELECTRIC PYROMETER 169 tube of soft iron inclosing a nickel wire. Nickel, associated with any other metal or alloy, will furnish a couple possessing anomalies in - - below 400 C., due to its molecular transforma- at tion region. Similarly, iron and also the various carbon- and alloy-steel wires have been found to introduce erratic thermo- electric behavior when used as one element in thermocouples. Harrison, Barrett, Belloc, and others, who have studied in detail the behavior of such thermocouples, find, for example, that there are considerable changes in E.M.F.'s due to oxidation, carburization, reheating, rate of cooling, the nature of the furnace atmosphere, maximum temperature reached, and time of heating at any temperature; and in general, within any region of molec- ular transformation, the E.M.F. on cooling will differ from that on heating, producing E.M.F. -temperature cycles or hysteresis. The couples containing iron are most unreliable above 800 C., and all such couples, as well as those with wires of steel, nickel, copper, and many of their alloys, become brittle above 700 C. The effect of reheating in producing hysteresis is shown in an experiment of Barrett on the change in temperature of the neu- tral point of a copper-steel couple : NEUTRAL POINT OF COPPER- STEEL. Heating. Second. Third. When heating 328 283 268 When cooling 258 241 241 The E.M.F. is not always higher during heating, however, the following alloy for example giving lower E.M.F.'s against Cu, Pt, or Fe on heating. This alloy (Fe = 68.8, Ni = 25.0, Mn = 5.0, C = 1.2), due to Sir Robert Hadfield, also pos- sesses the remarkable property of giving against a nearly pure iron an E.M.F. constant to within 4 per cent from 300 to 1050 C. The variation in electromotive force with composition of steels against platinum has been studied by Belloc, whose results 170 HIGH TEMPERATURES are shown in a somewhat idealized form in Fig. 23, page no, from which it is evident that any changes in composition due to heating or atmosphere will produce great changes in the E.M.F. of such couples below 350 C. and above 700 C. By heating a platinum, 1.2 per cent carbon-steel couple fifteen times to 1000, its E.M.F. per degree at 800 changed from n to 19 microvolts. On the other hand, for relatively low temperatures, very exact measurements have been made with thermocouples made of iron or nickel and a base-metal alloy. Thus Palmer, working in the range o to 200 C. with an iron-cons tan tan couple, gets a pre- cision of 0.04 per cent when the residual E.M.F.'s, including those due to mechanical strains, are eliminated. Complex Alloy Couples. Several manufacturers have sought to produce base-metal couples which are free from some of the defects usually present in this type. The work done so far gives promise that this field of investigation is worthy of further de- velopment. Most of the effort has been on the modification of the iron and nickel elements by the addition of other metallic components such as tungsten, copper, chromium, cobalt, silicon, and aluminium. We may mention the Bristol, Thwing, and Hos- kins couples as examples. One of the last, a robust nickel-chro- mium combination (Ni, 90 Ni 10 Cr), has an E.M.F. about four times that of the ordinary Pt-Rh couples, the E.M.F. - temperature relation is nearly linear to 1400 C. without any recalescence disturbances of sufficient magnitude to seriously affect the temperature readings in technical work, and after an- nealing this couple retains its readings sufficiently for many commercial uses, even when heated to over 1300 C. for short periods. For high temperatures, the Hoskins Company now use prin- cipally the couple nickel aluminium (2% Al) nickel chromium (10% Cr); and for low temperatures, nickel copper (65% Cu) nickel chromium (10% Cr). The characteristics of the mate- rials used in the Hoskins couples are given by M. A. L. Marsh, as shown in the accompanying table : THERMOELECTRIC PYROMETER 171 HOSKINS' THERMOCOUPLE CALIBRATIONS. COLD JUNCTION AT 25 C. Couple. Millivolts. Positive element. Negative element. 100 C. 232 C. 419 C. 657 C. 800 C. 1065 C. Nickel Nickel chromium, 10% Cr. 3.21 9.34 16.88 26.0 31.50 41.80 Nickel aluminium, 2%A1 Nickel chromium, 10% Cr 9.35 17.60 28.10 34.53 46.20 Nickel aluminium, 3$%A1 Nickel chromium, 10% Cr. 8.75 17.20 28.25 35.00 47.20 Nickel aluminium, 5% Al Nickel chromium, 10% Cr. 8.25 16.67 47.00 Cobalt Nickel chromium, 10% Cr. 3.09 9-93 20.06 33. 18 45.58 Nickel copper, , 65% Cu Nickel chromium, 10% Cr. 4.39 13.15 27.27 44.66 75-40 Copper Nickel chromium, 10% Cr. 1.48 4.15 7.73 11.27 13-57 ELECTRICAL RESISTANCE OF THERMOCOUPLE ELEMENTS. Resistance per foot, q npHfir _ Temperature Element. 0.40 m/m. wire, &pecihc resistance coefficient at 25 C. per degree C. Nickel 0.26 ohm 10.34 0.00415 Nickel aluminium, 3!% Al. . .63 25.0 .00274 Cobalt 356 14-15 Nickel copper, 65% Cu 99 39.3 Copper 043 1.75 . 00388 Nickel chromium, 10% Cr. . . i . 76 70.0 .00051 A difficulty met with in the manufacture of complex alloy couples is the reproducibility of the same E.M.F. -temperature relation from one casting to another. The identity of behavior is, however, highly desirable in cheap commercial couples which are frequently replaced, as it obviates the necessity of recalibra- tion or adjustment of the galvanometer scale for each couple. The Noble Metals: Geibel's Data. We have already dis- cussed the thermoelectric behavior of the platinum-rhodium alloys, page 116. Some of the platinum group metals and alloys have been studied by Holborn and Day, Rudolphi, Doerinckel, and others. The most thorough and reliable investigation, however, of the electrical and mechanical properties of the noble metals and their alloys, in view of their availability for tempera- ture measurement, has been made by W. Geibel in the laboratory of the Heraeus platinum works, using materials much purer than could be obtained by Barus or Le* Chatelier twenty-five years ago. The data of Geibel show wide differences from the earlier results of these observers. 172 HIGH TEMPERATURES PROPERTIES OF THE NOBLE Electromotive force (millivolts) against Metal or alloy. 100 200 300 400 500 600 700 800 900 Pd I I.I - i. 9 4.1 - 7.0 i .0 + 1.8 2.0 - 3-5 - 6.8 O.2 + 1.7 + 0.6 + 1.6 + i.S + 0.7 i.i - 1.8 - 2.9 - 6.4 -10.5 - 1.7 + 3-1 - 3-3 - 5-7 -10.7 + V.9 + 0.8 + 2.5 + 2.3 + 1.2 - 1.6 - 2.6 - 4.1 - 9.4 -15-0 - 2.5 + 4-5 - 5-o - 8.5 -15-3 - 0.3 + 4-4 + I.O + 3-5 + 3-3 + i.7 2.4 - 3-4 - 5-4 -12.5 -19.7 ~ 3-4 + 6.2 - 6.8 -II. 7 20.2 + Y.I + I.O + 4-4 + 4-4 + 2.3 - 3-3 - 4-6 - 6.9 16.0 -24.2 - 4-4 + 8.0 - 9.0 -15.2 -25-4 - 0.4 + 8.2 + 0.8 + 5-3 + 5-4 + 2.7 - 4-5 - 6.0 - 8.4 -19.7 -28.8 + 9-9 -ii. 4 -19.2 -31-0 + 10.6 + 0.6 + 6.2 + 6.5 + 3-2 - 5-8 - 7-5 IO.2 -23-4 -33-5 - 6.8 + 12. 14.2 -23.2 -36.5 - 0.5 + I3-I + O.I + 6.8 + 7-6 + 3-7 - 7-3 - 9.2 12. i -27.3 -38-2 - 8.2 +14.2 -17.0 -27.4 -42.1 - 0.6 + 15-9 - 0.6 + 7-4 + 8.5 + 4-2 +"9.7 + 14.0 +17.1 + 16.8 + 10.3 + n. 7 >lts at 4-7 4-5 6.6 6.8 Pd II -3'-8' -0.5 +0.8 -0.9 -1.8 -3-7 O.I +0.7 +0.3 +0.8 +0.7 +0-3 Pd-Au 10. . . Pd Au40. . . ^ Pd-Au 60. .. Pd-Au 80... Au Pd Ag 10. . . Pd Ag 20. . . Pd-Ag4o. .. Pd-Ag8o. .. Ag Pd-Ptio.... Pd-Pt 3 o.... Pd-Pt6o.... Pd-Ptpo.... Pt Pt-Ir 5.... Pt-Ir 10 Pt-Ir 25.... Pt-Ir 35.... Ir* Rh* + 1.1 + 1-3 + 1.2 + 1.1 +0.65 +0.65 ! +0.2 -0.4 + 2.1 + 2.6 + 2.6 + 2.5 + i-S + i-S Not cc + 0.4 - 0.8 + 3-2 + 4-1 + 4-3 + 4-1 + 2.5 + 2.6 >nstant + 0.7 - 1.4 + 4-3 + 5-8 + 6.2 + 5-9 + 3-6 + 3-7 varia ! 1.3 1 I.O t.o ( 2.1 + 5-4 + 7-4 + 8.2 + 7-9 + 4-8 + 5-1 tions a 1.8 i-5 2.6 2.8 + 6.5 + 9-1 + 10.4 + 9-9 + 6.1 + 6.5 s grea 2.4 2.1 3-5 3-7 + 7-6 + 10.7 + 12.6 + 12. 1 + 7-6 + 8.1 / as 2 3-i 2.8 4-5 4-7 + 8.7 + 12.3 + 14-8 + 14-4 + 9-1 + 9-9 millivc 3-8 3-6 5-5 5-7 Au-Ptio.... AU Pt20.... Au Pt4o.... Ag-Ptio.... Ag-Pt 3 o.... * Holborn and Day. Some of his results, taken from a very complete series, on E.M.F. against platinum, electrical conductivity, temperature coefficient, and tensile strength, are given in the accompanying table. Wires of 1.3 mm. were first glowed and then cold-drawn to i mm. Compositions are per cent by weight. The tensile strength may be taken as giving an approximate measure of hardness. One of the most satisfactory combinations for use as thermo- couple to say 1000 C. appears to be 40 Pd 60 Au Pt, which .at 1000 C. gives four times the E.M.F. of the ordinary Le Chate- lier couple. This Pd Au alloy also has a very low temperature THERMOELECTRIC PYROMETER METALS AND THEIR ALLOYS. (GEIBEL.) 173 platinum. Electrical conductivity Xio-2 w> w, 6 4 5 2 2 2 3 6 1531.0 1530.5 1530.0 I529-5 1530.0 1530.5 1530.0 1530.5 I530.I Horizontal furnace; bare wires. Horizontal furnace; porcelain tubes. < i Vertical furnace; see Fig. 57. M.P. on thermoelectric scale (equa- tion (3), p. 112). Mean = 1530. 2 THERMOELECTRIC PYROMETER I8 7 Boiling points, including those of the metals such as Cd and Zn, have frequently been used for the calibration of thermo- couples, but the boiling metals are much more difficult to manip- ulate than the melting, and there is far greater danger of con- tamination of the thermocouples, nor is there need of resorting to them. If desired, however, the freezing points of Sn, Pb or Cd, and Zn may be replaced by the boiling points 'of naphthaline, benzophenone, and sulphur respectively, none of which attack the couples ordinarily used. The standard form of boiling appa- ratus for an accuracy of 0.05 C. or better is shown in Fig. 169, except that for naphthalene and benzophenone a side condenser tube should be added ; or an air blast from a ring burner around the top of the boiling-tube may be used. For a somewhat less accuracy the smaller portable apparatus of Barus (Fig. 59) may be used for boiling points, including also water and analine. This consists of a tube of thin glass, similar to test tubes, of 15 mm. inside diameter, 300 mm. long, with a small bulb at 50 mm. below the open end. It is surrounded with a plaster muff of 150 mm. height and 100 mm. diameter which has been cast about the glass tube inside of a thin metallic cylinder forming the outside surface. The bulb is immediately above the plaster jacket, below which the tube, closed at its lower end, extends to a distance of 70 mm. As soon as the plaster has begun to set the glass tube is taken 'out, giving it a slight twisting motion. The cylinder is left to dry, and the tube is again put in place. This allows, F jf ar us when the tube is broken, of taking it out and replacing it, which would be difficult if it adhered to the plaster. A jacketed Victor Meyer tube may also be used. The lower free portion is heated by a Bunsen flame gently at first, then without any special precaution, once boiling sets in. The liquid at rest should occupy two-thirds of the height of the i88 HIGH TEMPERATURES free end of the tube. The heating is continued until the liquid coming from the condensation of the vapor runs abundantly down the walls of the bulb. The flame is then adjusted so that the limit of condensation of the liquid, which is very sharp, remains constantly midway up the bulb. There is then a very uniform temperature in the interior of the glass tube throughout the height of the plaster cylinder. The junction of the couple is inserted and the coil of the galvanometer takes up a fixed invariable position. It is well to prevent the liquid from run- ning down about the couple by placing a small cone of aluminium or asbestos above the junction. Electric heating may also be used. For the boiling point of zinc, Barus made small crucibles of porcelain very ingeniously arranged, but also very complicated, besides being fragile and costly. One can make use more simply of a porcelain crucible 70 mm. deep (Fig. 60), filled with melted zinc for 50 mm. of its depth, and, above, 20 mm. of charcoal dust. A cone pierced with a central hole lets pass a small porcelain tube containing the couple. The whole is heated until there is seen a small white flame of zinc escaping from the crucible. It Is indis- pensable that the openings for the escape of zinc vapor be large enough. They tend) indeed, to become clogged by a deposit of zinc oxide which solders at the same time the cover to the crucible, and this causes an explosion when there is no longer vent for the zinc vapors. A better form providing for the condensation of vapors was used by D. Berthelot. Technical Calibrations. The process of calibration is greatly simplified if an uncertainty of 5 C. or more may be permitted, as is the case for most technical operations. When the crucible method is used, smaller crucibles and furnaces may be allowed Fig. 60. Zinc-boiling Apparatus. THERMOELECTRIC PYROMETER 189 than for exact calibration, and metals and salts of less certain purity may be tolerated, although it is most certainly safer to use only very pure materials. Some metals, such as Al and Sb, have their melting points greatly influenced by small impurities, and are obtainable in sufficient purity with difficulty. Other metals, such as Sn, Pb, Zn, and Cu, can be trusted from almost any source of supply to give temperatures to within 2 C. of the melting points of the pure metals. The precautions of manipu- lation mentioned in the preceding paragraphs apply here, with somewhat attenuated emphasis, depending upon the accuracy desired. For example, it is often not convenient to maintain the cold junctions at o C., or one may desire to keep them during calibration at the average temperature to which they are sub- jected in use or attached directly to the pyrometer galvanometer. We have indicated elsewhere (page 155) how to make allowance for variations in cold-junction temperatures. The galvanometer and thermocouple may be tested either together or separately, but the former method is the more con- venient and eliminates the use of any auxiliary apparatus. Five or six points on the galvanometer scale are usually sufficient for a technical calibration of the couple and galvanometer together. These may be given by any of the freezing or boiling points we have mentioned. A convenient inexpensive series of the former, requiring the minimum of precautions in manipulation and suitable in crucibles of 50 or 100 c.c. in a small air-gas furnace when the couples are wires of small diameter, is the following : Sn 232 C. Al 657 C. Pb 327 NaCl 800 Zn 419 Cu CuO 2 1063 The last is copper saturated with its oxide. A piece of carbon steel (0.9 per cent C) bored with a small hole into which the couple is inserted will give on cooling a calibration temperature of about 700 C. If it is desired to use salts only, as may be the case with cer- tain base-metal couples of the cane type, which are plunged with- I QO HIGH TEMPERATURES out protection directly into the bath, there will be some sacrifice of accuracy even when large quantities of salt are used in nickel crucibles, due in part to our uncertainty of their melting points, in part to the great effects that slight impurities exercise on these temperatures, and above all to heat conduction along the leads, as the couple is practically short-circuited by the salt near its surface. This is even more emphatically true when a couple is plunged bare into metal, which is bad practice even when the couple will stand it. The following list of salts is suggested, with considerable reserve as to the numerical values, some of which may be 10 C. or more in error: NaNO 3 308 C. BaCl 2 950 C. KN0 3 336 K 2 S0 4 1060 Ca(NO 3 ) 2 550 Na 2 SiO 3 1088 KI 680 Li 2 SiO 3 1202 KC1 780 Diopside 1391 J NaCl 800 Anor^hite J 55O ^' Na 2 SO 4 885 Another sharply defined temperature which may be used is the transformation temperature of crystalline quartz at 575 C., obtained with a good quality of silica sand. Still another method of calibration is by the comparison of the pyrometer readings with those of a standardized instrument in the same furnace. With the platinum couples and a porcelain- tube platinum-resistance furnace of 1.5 cm. diameter and 30 to 60 cm. long, results to 2 degrees or better may be obtained if special precautions are taken to insure constant temperature, such as inclosing hot junctions within a short platinum cylinder and passing them through notches in a platinum disk. When base-metal couples are compared with those of platinum, it is usually necessary to protect the latter from contamination from the former by inclosing the platinum wires and junctions in glazed porcelain or other suitable tubes. A better temperature distribution for the comparison of base-metal couples than is found usually in tube furnaces may be obtained by using large crucible furnaces in which are placed baths of mixed salts which may be stirred and contained in long iron crucibles. THERMOELECTRIC PYROMETER 191 Industrial and Scientific Applications. The measurement of temperatures by thermoelectric couples has enhanced the accu- rate knowledge of a great number of high temperatures of which previously little or nothing was known. The earlier measure- ments were particularly numerous in the scientific and industrial investigations on iron. It was with the thermoelectric couple that Osmond and others, Roberts-Austen, Arnold, Howe, and Charpy made all their studies on the molecular transformations of irons and steels. The conditions of manufacture and of treatment of these metals have been improved by the introduc- tion into industrial works of this method of high-temperature measurements. We give below, as examples of the early use of the thermo- couple, a series of determinations made by Le Chatelier in a number of industrial operations. Steel. Siemens-Martin open-hearth furnace : Gas at the outlet of the gas generator 720 Gas at the entrance of the regenerator 400 Gas at the outlet of the regenerator 1200 Air at the outlet of the regenerator 1000 Interior of the furnace during refining 1550 Smoke at the foot of the chimney 300 Glass. Basin furnace for bottles; pot furnace for window glass: Furnace 1400 Glass in affinage 1310 Annealing of bottles 585 Rolling of window glass 600 ' Illuminating gas. Gazogene furnace: Top of furnace 1 190 Base of furnace 1060 Retort at end of distillation 975 Smoke at base of regenerator 680 Porcelain. Furnaces: Hard porcelain 1400 China porcelain 1275 To-day, the use of the thermocouple in the most varied indus- tries is so widespread that the above list could be indefinitely IQ2 HIGH TEMPERATURES multiplied. It is not merely, however, in the determination of the temperatures entering into a great number of industrial operations that the thermoelectric pyrometer has paved the way for itself and for other types of pyrometer, but it is, above all, in the ability that is thereby given by such temperature measure- ments to control the quality of the products of these operations depending on temperature, and so permit the exact reproduction, within as close limits as is wanted, of any desired result, and to increase thereby enormously the efficiency of many industrial plants. The use of the thermocouple in scientific investigations has been not less extensive or fruitful, and we have, for instance, what may be called a new science, or at least a new aspect of chemistry, namely, thermal analysis, which has grown up in recent years, based mainly on the interpretation of physicochemical phenom- ena at high temperatures by means of the indications of the ther- mocouple. In the development of scientific metallurgy, again, the thermo- couple has been almost the only temperature-measuring device which has been employed. These two generalizations are suffi- cient to indicate its secure position as an instrument of pyrometric research. Conditions of Use. Thermoelectric couples, as we have seen, may be divided for convenience into two general classes, the platinum-alloy couples and the base-metal couples, both of which are readily calibrated and easy to use. The former, by reason also of their small size and of the permanence and precision of their indications, are on the whole preferable to all other pyrometric methods for ordinary investigations, scientific or industrial, over the wide temperature range for which they are best adapted, or from 300 to 1600 C. We shall see, however, that when the highest accuracy is desired, or better than i C., the resistance thermometer of platinum may be given the preference to 900 C. from the very lowest temperatures, even over the thermocouple used with a potentiometer. The former also is somewhat more adapted for use with robust recording instruments. Above THERMOELECTRIC PYROMETER 193 1000 C., however, the platinum thermocouple is the only form of electric pyrometer which can be used with any considerable certainty; and attached either to a suitable direct-reading galva- nometer or to an automatic recorder, this instrument is proving of great utility in the industries. It may also be wired readily for use in multiple on a single distant recording instrument, each couple also having its separate indicator beside it. For temperatures below 600 C., there is gain in sensibility without serious loss in accuracy by substituting such couples as that of silver-constantan or copper-constantan, both of which can be kept small; and below 500 C. we reach the range of accurate mercury-in-glass thermometers. We have discussed elsewhere the precautions and methods of use both in work of precision and in technical work for the various types of couple. The use of the base-metal couple is limited to the technical field, and even here great discernment has to be used to satisfy oneself that the couple in hand, with its accessories, is suitable for the use to which it is contemplated to put it. There are few base-metal couples which can be used safely above 1000 C., and some of them are of very questionable utility for any purpose. We shall see that, even in the range for which it is best adapted, the thermocouple may in certain lines of work be replaced to advantage by still other methods, such for instance as the radia- tion and optical pyrometers. CHAPTER V. ELECTRICAL RESISTANCE PYROMETER. Introduction. In this method use is ordinarily made of the variations of electric resistance of a platinum wire as a function of the temperature; these variations are of the order of mag- nitude of those of the expansion of gases. The ratio of the resistances is 1.39 at 100, and 4.4 at 1000. As electrical resistances are measurable with great accuracy, this process of estimation of temperatures offers a very great sensibility, and applying exactly the law that connects the variation of resistances to that of temperature most excellent results may be obtained. The electric pyrometer was proposed by Siemens in 1871 (Bakerian Lecture); it rapidly came into use in metallurgical works on account of the reputation of its inventor, but it was soon abandoned for reasons which will be given later. This method of temperature measurement was revived twenty years afterwards by Callendar and Griffiths, and has been growing in favor ever since, both in the laboratory and in the industries, especially in England, and more recently in America. It is perhaps of interest to note that in Cambridge, England, the re- sistance thermometer was first brought into a satisfactory condi- tion as a physical instrument and its theory successfully worked out by Callendar and Griffiths; there it was first used in most delicate measurements of chemical phenomena by Heycock and Neville; and finally, the Cambridge Scientific Instrument Com- pany were pioneers in the manufacture of instruments suitable for industrial and scientific use. Work of Early Investigators. Siemens. The Siemens pyrom- eter consists of a fine platinum wire i m. long and o.i mm. in diameter, wound on a cylinder of porcelain or fire clay; the 194 ELECTRICAL RESISTANCE PYROMETER 195 whole is inclosed in an iron tube, destined to protect the instru- ment from the action of the flames. Siemens tried also, but without success, ceramic materials impregnated with metals of the platinum group. To measure the resistance he employed either a galvanometer, for laboratory experiments, or a voltameter, for the measure- ments in works. In this latter case the current from a cell divides between the heated resistance and a standard resistance at constant temperature; in each one of the circuits was placed a voltameter: the ratio of the volumes of gas set free gives the ratio of the current strengths and thus the inverse ratio of the resistances. Finally Siemens gave a formula of three terms connecting the electrical resistance of platinum to temperatures 'on the air thermometer, but without publishing the experimental data on which this graduation was based. Experiment soon showed that the apparatus did not rest com- parable with itself. A committee of the British Association for the Advancement of Science found that the resistance of platinum increases after heating. It would be necessary then to calibrate the apparatus each time that it was used. This change of resistance is due mainly to a chemical alteration of platinum, which is enormous when heated directly in the flame, less, but still marked, if placed in an iron tube, and which almost disappears if use is made of a platinum or porcelain tube. This increase of resistance may reach 15 per cent by repeated heatings to 900. Platinum being very costly and porcelain very fragile, it was impossible to use these two bodies in the industries, which alone at that time occupied themselves with measurements of high temperatures, and this method was abandoned completely during twenty years. Callendar and Griffiths. These savants revived this method for laboratory purposes; it seems the best for many kinds of work of precision to moderately high temperatures, on the condition of being assured of the invariability of the resistance of platinum. 196 HIGH TEMPERATURES Callendar found that clay helps to cause the variation of resistance, that the platinum wire becomes brittle on its support and sticks there; this action is probably due to impurities in the clay. With mica, on the other hand, which the wire touches only at the edges (the reel is made of two perpendicular slices of mica) , there is perfect insulation without cause of alteration ; but mica becomes dehydrated at 800 and then becomes very fragile. All metallic solderings should be proscribed, for they are "vola- tile and attack platinum. Pressure joints (screw or torsion) are equally bad, for they become loose. One should use only autogenous soldering by the fusion of platinum. Copper conductors should also be rejected, at least in the heated portions, on account of the volatility of the metal. A pyrometer with such conductors, heated during an hour at 850, showed an increase of resistance of J per cent. Holborn and Wien. These investigators made a very com- plete study of this alterability of platinum wires, in a comparison between the methods of measurement of temperatures by electric resistance and thermoelectric forces; they worked with wires of o.i mm. to 0.3 mm. diameter. They soon found that above 1200 platinum commences to undergo a feeble volatilization which suffices to increase notably the resistance of the very fine wires. Hydrogen in presence of silicious materials causes at about 850 a rapid alteration of the platinum. Below are the results relative to wires of 0.3 mm. of a length of 1 60 mm.: Wire a. R at 15. Wire 0. R at 15". At start o . 239 ohm At start o 24.7 ohm After heating red-hot: After several days in hy- . -^tf./ Twice in air at 1200. . 0. 238 ohm drogen at 15. . . o .246 ohm Once in vacuo at 1200 o , 240 ohm After heating in hydro- Once in H at 1200 .... o . 262 ohm gen to 1200. . . . 255 ohm Once in vacuo at 1200 0.253 ohm Wire 7. Rat 15. At start . . 0.183 ohm After heating in air to 1250 (three times) o. 182 ohm After heating in H to 1250 o. 188 ohm After heating in H to 1250 o. 195 ohm ELECTRICAL RESISTANCE PYROMETER 197 Wire 7 heated to 1350 in an earthenware tube and in hydrogen became brittle; this result may be explained by a siliciuration of the platinum, for there is nothing observed if the wire is heated by the electric current in the interior of a cold glass tube, even in hydrogen. Similar experiments were made by the same observers with palladium, rhodium, and iridium. We shall return to this question of the constancy of the resistance of platinum. Law of the Variation of Platinum Resistance. Callendar and Griffiths have compared the resistance of platinum with the air thermometer up to 550 C.; they found that up to 500 the relation could be represented at least to 0.1 by a parabolic for- mula of three parameters. In order to graduate such a pyrometer it would be sufficient then to have three fixed points: ice, steam, and boiling sulphur. They gave a special form to the relation; let p t be the platinum temperature defined by the relation . -K-100 ~~ -K-0 that is to say, the value of the temperature in the case in which the resistance varies proportionally to the temperature. They then placed It would appear as if this formula contained the single param- eter 5; but in reality p t includes two. Substituting for p t its value, we have . * WO Q . 2 an equation of the form R t = R Q (i + at - bf-), which it is sometimes convenient to use. Callendar and Griffiths used their pyrometer before having standardized it against the air thermometer. Not being able to compute t, they provisionally computed the approximate temperatures p t , and later determined 198 HIGH TEMPERATURES the correction between / and p t} after having sought the formula expressing the difference between these two quantities by means of a careful determination of the sulphur boiling point on the air thermometer. By extrapolation up to 1000 the points of fusion of gold and of silver were found quite near to those determined by other observers. Harker, working at the National Physical Laboratory, Eng- land, has compared the readings of platinum thermometers, when reduced to the gas scale by the use of Callendar's difference for- mula, with the readings of thermocouples calibrated at the Reich- sanstalt, and with the indications of an inglazed porcelain-bulb nitrogen thermometer at contant volume of the Reichsanstalt form. Specially constructed, compensated electric furnaces were used for heating. As shown by the accompanying table, taken from one series of Barker's measurements, the agreement between the scales of the platinum-resistance and thermoelectric pyrometers was within 0.5 C. throughout the temperature range up to 1000, although the gas pyrometer gave somewhat discordant results. COMPARISON OF PYROMETRIC SCALES BY HARKER. Temperature. Gas ther- Thermo- Pt ther- G-Pt. G-Th. P-Th. mometer. couple. mometer. 523-I 524.3 524.39 -i-3 1.2 -O.I 598.5 597-8 597.62 +0.9 +0.7 -0. 2 641.1 641 . 1 641 - 75 +0.6 +O.O -0.6 776.7 775-5 775.13 + 1.6 + 1.2 -0.4 820.0 818.4 818.31 + 1-7 + 1.6 o. i 875-0 875-4 875.24 O.2 -0.4 O. 2 959-8 956.0 955-47 +4-3 +3-8 -0.5 1005.0 1004.4 1004.37 +0.6 +0.6 o.o I A very careful direct comparison of the reduced indications of several platinum thermometers with the gas scale as furnished by the constant-volume nitrogen thermometer has also been made by Chappuis and Harker at the International Bureau at ELECTRICAL RESISTANCE PYROMETER 199 Sevres, and their results give further evidence that the indications of the platinum thermometer up to 600 C. can be sufficiently well expressed by Callendar's formula. There is another method of comparison of temperature scales which is capable of great accuracy, namely, the determination on the several scales of the freezing and boiling points of a number s of pure substances. This method has some decided advantages over the above method of comparison even in a most carefully compensated electric furnace. Heycock and Neville in England, and more recently Waidner and Burgess at the Bureau of Standards, have determined the freezing points of several pure metals in terms of the scale of the platinum thermometer stand- ardized at o, 100, and 444.70 C. (the boiling point of sulphur), and find that the freezing points so determined give tempera- tures on the gas scale as closely as the latter can be reproduced, as shown in the following table: GAS AND RESISTANCE TEMPERATURE SCALES: Gas scale. Resistance scale. Holborn and Day and Heycock and Waidner and Day. Sosman. Neville. Burgess. Cd 321 .7 320.0 320.7 321 .0 Zn 410 o 418.2 419.4 419.4 Sb 630.6 629.2 630 . i 630.7 Al 657 .0 658.0 658.0 Ag 061 ^ 960.0 961 .9 960.9 Cu.. 1084 . i 1082.6 1082.0 1083.0 These results confirm the view of the sufficiency of the Callendar difference formula for the most accurate work up to the upper limit of the safe use of the platinum-resistance thermometer. Holborn and Wien have shown that at very high temperatures the interpolation formula is certainly inexact. The resistance seems to become asymptotic to a straight line, while the formula leads to a maximum evidently inacceptable ; in their opinion it would be better represented by an expression of the form R.t = a + b(t + 273)-. 200 HIGH TEMPERATURES Here are the results of two series of their experiments made on the same wire: t R t R Degrees. Ohms. Degrees. Ohms. o 0.0355 o 0.0356 1045 1510 1040 1487 H93 1595 "44 1574 1303 1699 1328 1720 1395 T 787 1425 1802 1513 1877 1550 1908 1578 1933 1610 1962 Using the Callendar formula and platinum wires, Petavel found the melting point of palladium to be 1489, which Callendar and Eumorfopoulos found to be 1550. This latter number is in exact agreement with the best determinations of this tem- perature. Although the work of Holborn and Wien, as well as that of Tory and others, shows that the platinum-resistance thermometer made of fine wire cannot be depended upon to remain constant above 1000 C., yet, in the range from 200 C. to +1000 C., it serves as the most accurate, and, on the whole, most convenient method of measuring temperatures where great precision is required, and is particularly adapted for the delicate control of a given temperature. Dickson has proposed the formula (R + a)* = p(t + b), in which #, 6, and p are constants. It possesses the possible theoretical advantage over the Callendar formula of not requir- ing a maximum value for the resistance of platinum. This form, however, does not lend itself to the convenient graphical treat- ment applicable to the difference formula; and furthermore, for thermometers of pure platinum calibrated at three temperatures in the usual way, the Dickson formula does not reproduce the same temperature scale as the difference formula as shown by Waidner and Burgess, it giving, for instance, 1051 C. for ELECTRICAL RESISTANCE PYROMETER 2OI copper instead of 1083 C. for calibration in ice, steam, and sulphur vapor. Nomenclature. To determine a temperature by means of a platinum thermometer, if the instrument has not been calibrated already in degrees, it is necessary to know the difference coefficient 5 of the wire, which may be obtained by finding the platinum temperature pt at some known point, as the sulphur boiling point (S.B.P.), or by comparison with a calibrated instrument. Callendar has suggested the following notation which seems convenient for platinum thermometry: Fundamental Interval. The denominator RIQQ RQ in the formula IPO (R - RQ) ( , P t = ~T~ ~j?V> ...... W ^AIOO HQ) for the platinum temperature pt, represents the change of resist- ance of the thermometer between o and 100. Fundamental coefficient = c = mean value of temperature co- efficient of change of resistance between o and 100: _ (RiQQ RQ) IOO RQ Fundamental zero = pt Q = - = reciprocal of fundamental co- c efficient. It represents the temperature on the scale of the instrument itself at which its resistance would vanish. Difference Formula. The following form is the most con- venient for computation: -iV ..... (2) IOO / IOO Parabolic function expresses the vanishing at o and 100 of above formula, which becomes " S.B.P. " Method of Reduction. D is obtained very con- veniently by determining R", and thus pt" at l" = the boiling point of sulphur (= S.B.P.). 202 HIGH TEMPERATURES Resistance Formula. The parabolic difference formula is equivalent to assuming R_.. I+at + bt 2 () where / d \ , cd. a=c(i-\ ), o= > \ loo/ 10,000 or 6 = - - ' ' I0 a + o io j Graphic Method of Reduction. An easy way to reduce plati- num temperatures to the gas scale is to plot the difference t pt in terms of t as abscissas, and to deduce graphically the curve of difference in terms of pt as abscissas. This is most convenient for a single instrument up to 500. Other methods have been used by Heycock and Neville and by Tory. Difference Formula in Terms of pt. This formula is to be used only where a high degree of accuracy is not required. The value of d' may be determined from S.B.P., or approximately d d' = (i 0.077 Construction of the Platinum Thermometer. Callendar first devised a satisfactory and perhaps the most commonly used form of platinum thermometer, in which the platinum wire is wound on two strips of mica set crosswise. In Fig. 61 is shown a labora- tory form of Calendar's potential terminal thermometer used at the Bureau of Standards in precision work to 1100 C. The heavy copper head insures a minimum of thermoelectric effects at the platinum-copper junctions, and provision is made for air cooling of the head, which is an advantage for work at the highest temperatures. The junctions of the leads to the platinum coil ELECTRICAL RESISTANCE PYROMETER 203 are easily made by arc soldering, using platinum as one terminal and a graphite pencil as the other. No material other than platinum should enter into joints to be heated. Forms of mica supporting frame are shown in Fig. 62. ,,-Air Circulation Porcelain Tube > Mica Disks Serrated Mica Frame Mica Frame Fig. 62. Mica Frames. Various modifications of the above de- sign are used in the industrial forms, they being in general so arranged as to secure the maximum of protection and robustness to the platinum coil. Mica frames are sometimes replaced by steatite by Leeds and Northrup except for the highest temperatures. Industrial types of mounting used by the Cambridge Company are shown in Fig. 63. The outer containing tubes for indus- trial instruments are preferably of metal, such as nickel or iron, over a quartz Fig. 61. Resistance Py- n rometer, Laboratory Type, or porcelain tube, the actual material of the sheath depending, however, on the use to which it is- to- be put. 204 HIGH TEMPERATURES Platinum Coil Screwed Flange Porcelain Tube-- Mica Discs/" 4 Platinum Leads- SUPERHEATER THERMOMETER. K35mmH THERMOMETER FOR FLUE TEMPERATURES. Fig. 63. Types of Industrial Mounting. ELECTRICAL RESISTANCE PYROMETER 205 Porcelain, Tubes Steatite Disk Mica Partition Platinum -Coil Steatite Fig. 64. Freely Sus- pended Coil. For use at very high temperatures, Leeds and Northrup have designed the form of potential lead thermometer shown in Fig. 64. Heavy wire (0.6 mm.) is used in the coil, which is freely suspended and therefore not subject to strains on cooling. Due to its very low resistance, special precautions have to be taken in the temperature measurements to secure sensibility. Such heavy-wire thermometers will change their constants very much less than those of fine wire when heated to high temperatures. Thus Waidner and Burgess found I that heating them for several hours to 1200 or 1300 C. changed the zero reading by only a few tenths degree, after they had been once annealed at 1300 C. In order to secure an instrument of small vol- ume and at the same time satisfactorily protect and rigidly mount the platinum coil, Heraeus has devised the form shown in Fig. 65, in which the platinum coil is embedded in fused quartz glass. The behavior of this type of thermometer, with wires of 0.05 to 0.15 mm., has been studied at the Reichsanstalt. The effect of embedding in quartz is to decrease the value of a (equation (3), page 202) and increase the value of 5. As compared with wires mounted in the usual way, and receiving the same heat treatment, the change in the con- stants is very great for these thermometers. For the former, a decreased by 0.45 per cent and 8 by 0.65 per cent; for the latter, the changes were 1.7 per cent and 6.7 per cent respectively. Where very great rapidity of action is desired the form of construction shown in Fig. 66, due to Dickinson, may be used in certain cases, the metallic parts being Fig. 65. Mountings in Quartz. 206 HIGH TEMPERATURES preferably all of platinum where great permanence is desired, and the insulation of mica strips. Where platinum thermometers are to be used with a definite form of measuring apparatus, or where several such thermometers are to be used with a single bridge, recorder, or other indicating or reg- istering device, it is convenient to have them all adjusted to exactly the same resistance at zero and of the same fundamental interval, and so make them interchangeable. This is done by several firms by means of auxiliary manganin coils set into the thermometer head. Choice of Size of Wire. Re- garding the choice of the diameter of wire to use in constructing a thermometer coil of given resist- ance, there are several points to consider besides the current-carry- ing capacity without undue heat- ing of the coil, which is in favor of the heavy wire ; such as the greater temperature lag, heat conduction along the leads, and excessive size of the thermometer coil, which, together with the cost, are the main inconveniences of heavy wire; and liability to strains, fragility, and greater evaporation, which limit the use and pre- cision of too small wire. It is easy to get enough current sen- sibility, constancy of resistance, and robustness with wires of 0.15 to 0.20 mm. diameter except for very low-resistance pyrom- eters, 2 ohms or less, which are to be avoided, save for work at very high temperatures, as taxing too severely the sensitiveness of ordinary forms of measuring apparatus. Precautions in Construction and Use. The platinum ther- mometer, as usually constructed, is a fragile instrument in spite Fig. 66. Thermometer of Small Lag. ELECTRICAL RESISTANCE PYROMETER 207 of its appearance of robustness when encased in a metal tube, therefore careful handling is required. To avoid breaking from sudden heating when porcelain or similar containing tubes are used, the pyrometer should be installed in advance in the furnace, or preheated in a muffle if it is necessary to introduce it into the hot furnace. It is necessary, also, to heat a sufficient length of the stem in the furnace in order to avoid the effect of heat conductivity, which would prevent the thermometer spiral from taking up the temperature of the space in which it is immersed. Platinum is readily attacked and its resistance changed by con- tact with most substances, including many vapors and gases, so that the thermometer coil must be carefully shielded by materials impervious to the atmosphere in which it is placed, such as porcelain glazed on the outside. As platinum changes its nature with heating, and as the frame on which the coil is wound may permanently change its dimensions, especially if mica is used, the thermometer before calibration should be annealed at a tem- perature higher than that at which it is to be used. A plati- num thermometer will change its readings with time the more rapidly, the higher the temperatures at which it is used ; therefore, in order to control its constancy, it is necessary to take its reading occasionally at some known temperature, as the ice or steam point. Well-shielded, pure platinum wound on a frame that does not contaminate the wire will change its constants with use less than does impure platinum, so that it is highly desirable to use only the purest of platinum in the construction of pyrometers. Even with pure platinum, however, in work of great precision, it is necessary to recalibrate occasionally, and when temperatures above 1000 C. are measured frequently this operation becomes very onerous. Great care has to be exercised, and this should be especially emphasized for industrial as well as scientific installa- tions, to secure a proper insulation of all electrical circuits. Methods of Measurement. It is evident that most of the ordinary methods for the measurement of resistance may be used in platinum thermometry, but in practice only a few of these methods have been applied to temperature measurements, 208 HIGH TEMPERATURES although there is a tendency at the present time, in the solution of specific-temperature problems, to take advantage of the pecu- liarities of less usual methods both for work of high precision in the laboratory and for industrial applications. Thus, in addition to the ordinary slide- wire and dial Wheats tone bridge methods, the Kelvin double bridge is sometimes used with pyrometers of very low resistance, for which this method is particularly adapted. Potential terminal and differential galvanometer methods are also used in precision work, and for industrial practice several deflection methods have been developed for the direct reading of temperatures on a galvanometer scale. Compensation for Pyrometer Leads. There is one character- istic in the measurement of a resistance coil used as a pyrometer that distinguishes it from an ordinary resistance measurement, namely, that in the case of the pyrometer coil there is a region of great temperature gradient from the coil to the measuring apparatus, so that it becomes imperative to eliminate the variable resistance of the leads to the pyrometer coil a resistance that varies both with the depth of immersion of the coil and with its temperature. There are several ways of effecting the necessary compensation of this variable lead resistance, and they will be described under the several headings. Three-lead Thermometer. This was the form originally given to the instrument by Siemens in 1871, and it is used in the con- struction of apparatus suitable for industrial use by Siemens and Halske and by Leeds and Northrup. In the Siemens method (Fig. 67), the thermometer coil P forms one arm of a Wheatstone bridge, of which the others are r\, r 2 , and R, when from the principle of the bridge, if the galvanometer G remains undeflected, P = R , neglecting the leads. fi The compensation for the variable resistance of the thermom- eter leads is effected in the following manner: The lead aa', of the same material as the thermometer coil P to avoid thermo- electric effects at their junction, is constructed to be as exactly equal as possible electrically to the similar lead bb'. The lead ELECTRICAL RESISTANCE PYROMETER 2OQ ad is in the P arm of the bridge and the lead W is put in the R arm by means of the auxiliary lead c'b of the same material as P. a a' Fig. 67. Three-lead Compensated Thermometer. This lead c'b may be put in the battery circuit as shown, or in the galvanometer circuit if preferred. It is not necessary to adjust c'b to any particular resist- ance, so that fine wire may be used for it. With this arrangement, therefore, the resistance of the thermometer remains apparently constant for a given temperature whatever its depth of immer- sion and whatever the temperature gradient along the leads aa', bb', so long as it is the same for both. The three-lead compensated thermometer may also be used with a differential galvanometer. Fig. 68 shows the principle of such an arrange- ment for an instrument of Leeds and Northrup. The slider d is set on the slide wire i in such a Use of Differential position that the current from B divides equally Galvanometer, between the circuits b + R + gi and T + a + #2, of which g\ and gz are the two differential galvanometer coils. If the resistance Fig. 68. 210 HIGH TEMPERATURES R remains fixed, the changes in temperature of T, the ther- mometer coil, may be read directly in degrees on the slide wire if desired. The compensation by means of the leads #, b, c is Fig. 69. Thermometer of Siemens and Halske. effected as before. The arrangement used by Siemens and Halske is shown in Fig. 69. In Fig. 70 is shown a system of wiring for four thermometers of the Siemens type and used with a single indicator. For work of great precision, this method is of course capable ELECTRICAL RESISTANCE PYROMETER 211 of elaboration and refinements, as in the calorimetric measure- ments of Jager and Steinwehr, who, however, used a four-lead thermometer. Four-lead Thermometer. There are four ways in which the four-lead compensated thermometer of Callendar and Griffiths Fig. 70. Four Thermometers with One Indicator. has been used, namely, the Wheatstone and Kelvin ' bridge, the potential terminal, and the differential galvanometer methods. The Wheatstone bridge method is illustrated in Fig. 71, from which it is seen that the compensating leads are inserted in one arm R of the bridge and the thermometer leads in the other. It is necessary that all four leads be as nearly as possible of the same length, diameter, and material. For work of great accuracy it is necessary to take all the precautions which obtain in exact 212 HIGH TEMPERATURES resistance measurements, and in particular the elimination of thermoelectric effects and uncertainties in the exact value of the ratio coils. Precision Bridges. In Fig. 72 is shown diagrammatically the important features of a bridge designed and in use at the Bureau of Standards, constructed by Leeds and Northrup, and capable of measurements to i in 100,000, and connected, in the figure, for use with a four-lead thermometer. This bridge can also be used, however, with a three-lead thermometer. Some of its char- Compensating Leads Coil Leads. V Fig. 71. Four- lead Compensated Thermometer. acteristics are possibility of reversal of all circuits, the inter- changeability of the ratio coils, mercury contacts for the higher resistances to eliminate contact resistances, a device due to Waidner consisting in a split ohm shunted across three dials to give rapidity of setting for final adjustment, and the ability to test the bridge without other accessories. The bridge is oil- immersed and kept at constant temperature by thermostatic control, and all coils are of seasoned manganin, which for the very highest precision should be sealed air-tight separately to avoid effects of humidity even beneath the oil. As galvanometer, a very sensitive form of d'Arsonval, due to Weston, is used, and ELECTRICAL RESISTANCE PYROMETER as battery one to three dry cells. The thermoelectric key may be dispensed with and a single contact key put in the battery circuit, with a variable resistance to replace the usual galvanom- eter shunt for varying the sensibility. Another form of the Callendar and Griffiths self-testing bridge, designed primarily for use with resistance thermometers of a fundamental interval of one ohm, is constructed by the Cam- bridge Scientific Instrument Company To eliminate tempera- Griffith's Thermoelectric Key- Oil Immersed X 0.001 X 0.0001 X 0.00001 Fig. 72. Thermometer Bridge of the Bureau of Standards. ture variations in this bridge, not only the coils but the bridge wire and all contacts are oil-immersed; and it is capable of reading platinum temperatures, in the latest model, to better than 0.01 C. by direct reading on the scale of the bridge wire, when a galvanometer of suitable sensibility and resistance is used, such as a Broca instrument of 10 ohms. The principle of the construction and wiring of this bridge is shown in Fig. 73, in which RI and R% are ratio coils of 10 co each, which should be interchangeable, BC the balance arm, adjustable by nine manganin coils AR and the slide wire s, while DC is the 214 HIGH TEMPERATURES thermometer arm. P and C are the thermometer and compen- sating leads respectively. The unit of the bridge is one degree on the platinum scale (page 201), and this corresponds to o.oi co for a FJ. of i co in the thermometer. This bridge possesses many mechanical excellencies, such as a special form of combined plug and mercury contact, protection from mercury, and a convenient form of vernier and slide wire. Fig. 73. Callendar and Griffiths Bridge. The resistance of the potential terminal thermometer is deter- mined by sending the same current from a storage battery through the thermometer and a known resistance in series, and measuring the potential drop by means of a potentiometer (page 138), first across the known resistance and then across the thermometer coil. This method of measurement for accurate work is illustrated in Fig. 74, which shows a rheostat and milli- ELECTRICAL RESISTANCE PYROMETER 215 ammeter in the circuit for adjusting the measuring current. The mercury-contact resistance box may be adjusted to within o.oi ohm of the thermometer, thus eliminating potentiometer errors. This box may of course be replaced by a single-standard resist- ance, in which case an accurate calibration of the potentiometer is required. The current leads for this type of thermometer do not have to be adjusted to equality, and the potential leads may be of fine wire, as may also the current leads, but still keeping the thermometer sufficiently robust, so that errors To Potentiometer To Potentiometer- Milli Ammeter A Rheostat Mercury Contact Resistance Box .01 .01 .02 .02 .05 .1 .2 25 Current Leads Therm. Coil \ \ N Fig. 74. Potential Terminal Thermometer of Precision. due to heat conduction along the leads need not creep into the measurements. The Kelvin Bridge. The principle of this method of measur- ing resistances is shown in Fig. 75, in which 6 1 is an adjustable resistance, x the unknown, and the others are such that, by construction, 7 = 7^, when x = -S for no current in the eral- b 0i o vanometer. This method of bridge design, accompanied by a sufficiently sensitive galvanometer, permits the measurement of o.oi ohm to be made with about the same precision as 100 ohms by the usual bridge methods, and is therefore particularly well adapted for resistance thermometers which are to be used at very high temperatures, because such instruments must be made 2l6 HIGH TEMPERATURES of wire of large diameter, and therefore of low resistance, in order to avoid changes in their constants due to heating. The Kelvin bridge method permits cutting down the amount of platinum in the pyrometer, an advantage both in cost and in volume of the instrument. Leeds and Northrup made a potential point indicator (Fig. 75 A), with slide wire for use with a heavy-coil low-resistance thermometer carrying a current of 0.3 ampere. The extension coils and slide wire may be graduated in degrees of tem- perature for any given thermometer. The high values (520 ) of a and a' (Fig. 75), necessary to eliminate resistance changes in the potential 'leads, require that the galvanometer used Fig. 75. Principle of Kelvin Bridge. shall have a greater sensibility than can easily be gotten in a portable pointer instrument. The type of galvanometer is the same as that required for high-precision Wheatstone bridge work with proper adjustment of critical external re- sistance. Sensibility. The sensitiveness of the measurements in re- sistance thermometry is that of the very great precision attain- able in resistance measurements, or it may be better than i in 100,000, or about 0.001 C. for a high-temperature thermometer whose resistance at o C. is from 3 to 25 ohms, if proper pre- cautions are taken. The factors limiting the sensibility of re- sistance measurements in the Wheatstone bridge method, for example, and which are inherent in thermometric work, are the ELECTRICAL RESISTANCE PYROMETER 217 practical necessity of using a i : i ratio, required for the lead compensation; the need of keeping the current through the ther- mometer coil so low as not to raise its temperature unduly; and finally, the sensibility of the galvanometer. Due to the first and P, \^) i OvwwOwvwO Fig. 75 A. Potential Point Indicator. second of these conditions, the ordinary rules for the Wheatstone bridge do not apply without modification, and fortunately the limitations they impose may very largely be overcome by prop- erly choosing the constants of the thermometer and galva- nometer. It should be remarked that, with the d'Arsonval or 2l8 HIGH TEMPERATURES moving-coil galvanometers of very great sensibility and of prac- tically constant zero which are available to-day, the question of the realization of sufficient sensibility is of distinctly secondary importance in accurate work. In the case of recording instru- ments, when in general a less sensitive instrument must be used, some attention has to be paid to the matter, and particular care has to be taken here to so arrange as not to overheat the ther- mometer with the larger currents required by such galvanometers. For the maximum current through the galvanometer and the minimum through the thermometer coil, with a battery of neg- ligible resistance, the bridge should be arranged as follows, as shown by Callendar: " Connect the battery so as to make the resistance in series with the thermometer greater than the resist- ance in parallel." Direct-reading Thermometers. There have been in recent years a considerable number of direct-reading resistance py- rometers devised by several manufacturers. We shall be able to call attention to only a few typical instruments, which are, of course, of interest mainly in technical practice. A principle commonly made use of in some of its modifications is that of the ohmmeter, in which a variable resistance, that of the ther- mometer, is balanced against a fixed resistance by means of the deflection of a galvanometer coil carrying currents from circuits shunted around the two resistances in question. Such deflection instruments are constructed by Paul, Hartmann and Braun, Carpentier, Leeds and Northrup, and others. The Harris Direct-reading Resistance-thermometer Indicator, manufactured by Mr. Robert W. Paul of London, indicates tem- peratures directly by the movement of a pointer over a scale; moreover its accuracy is independent of the battery or supply voltage used. In principle it is a two-coil ohmmeter, or coil-controlled galvanometer; the requisite sensitivity to the small changes in resistance, which are utilized in platinum thermometry, is attained by making the action of the deflecting k coil differential. ELECTRICAL RESISTANCE PYROMETER 219 The differential windings of the deflecting coil are respectively connected in shunt with the platinum thermometer and a resist- ance equivalent to that of the thermometer at any desired tem- perature, dependent upon the part of the temperature scale at which it is desired to work. The control coil of the ohmmeter system is connected in shunt with a resistance suitably chosen to give the required sen- sitivity. These combina- tions are connected in series. Hence, on the passage of an electric cur- rent, the forces due to the windings are proportional to the resistances they respectively shunt. In the accompanying vector diagram, the plat- inum thermometer is assumed to have a Fun- damental Interval of one ohm and to be brought up to a resistance of three ohms at o C., by means of a resistance which has no temperature coefficient, suitably introduced into the circuit. This enables the thermometers to be made electrically interchangeable with each other. Considering first the case of an ohmmeter system without the differential winding of the deflecting coil, let AB represent the controlling force of the ohmmeter system (proportional to the control-coil shunt referred to above, and in this instance taken as one ohm); and let AC represent the deflecting force with a plati- num thermometer of one ohm Fundamental Interval at o C. The pointer of the instrument will then take up the position AF. AC-AE AC-AE 76 ' Vector Diagram for Ohmmeter. 220 HIGH TEMPERATURES Suppose the temperature of the thermometer is raised to 100 C.; the deflecting force now increases by the amount of CC' (pro- portional to one ohm) and becomes equal to AC, causing the pointer to set along AG. Similarly, if the thermometer drops to -iooC., the position taken up by the pointer is along AH, giving an angle 6 for 200 C. variation. If, however, the deflecting coil be wound differentially, and a current equal to that producing the deflecting force AC be passed through the other winding in such a direction that its effort is in opposition to AC, introducing the vector AE, the initial position of the pointer will be along AB, and a variation CC' in the vector AC gives a resultant AC" and causes the pointer to set along the line AJ. Similarly, should AC decrease by an amount equal to CC", the resultant deflection is on the opposite side of the initial position, and the pointer takes up the position AK, giving the large angle for the change in the thermometer resistance equal to that which only gave the angle 6 with a nondifferential ar- rangement. It will be noted that AE may be given a value equal to A C at any required temperature of the platinum ther- mometer, the position of the pointer at such temperature lying along AB and covering the same angles as before for the same ohmic variation in the thermometer resistance. By simulta- neously varying the vector AB the angle < may be kept true to gas-scale temperature. It is thus possible to construct a multiple- range instrument. The accompanying diagram, Fig. 77, shows the scheme of con- nections for such an instrument. One of the differential windings (X in the figure) is shunted with a platinum thermometer, the other winding 5 being shunted with a resistance s, which is made variable so as to equal the resistance of the thermometer at certain fixed temperatures. The control coil of the ohmmeter system is also shunted with the resistance d, the value of which is determined by the degree of sensitiveness required, and may be made variable with s. These shunted windings are connected in series, and the circuit is completed through a battery and switch. ELECTRICAL RESISTANCE PYROMETER 221 In this arrangement, since the currents in the ohmmeter system windings depend upon the resistance of the platinum thermom- eter, s and d respectively, the value of s may be taken as the Fig. 77. Harris-Paul Indicator. vector AE, increasing by steps equal to the rises in resistance of the platinum thermometer for each range of temperature. The platinum thermometer represents the vector AC, AC, etc., while d represents AB. This is made variable with 5 in order that the instrument shall read in gas-scale degrees on all ranges, and its values are calculated in accordance with Callendar's formula for the platinum ther- mometer. The Logometer and Ratiometer. Messrs. Carpentier and Joly have also proposed the construc- tion of a deflectional-resistance pyrometer based on the use of the logometer, an apparatus de- signed for the measurement of Fig. 78. Logometer Coil. the ratio of two currents. This is shown in plan in Fig. 78, where two oppositely wound coils, similar to those of a d'Arson- val galvanometer, are mounted by double pivot in the unsym- metrical field of a permanent magnet NS. If the two coils 222 HIGH TEMPERATURES have the same number of turns, we have iH = i' H f , and since the electromagnetic force of each coil is directed toward a weaker field, the final position of the coils will be stable and will depend only on the ratio of the two currents i and i' in the coils. For the measurement of resistance, the circuit, in one of its simplest forms, is arranged as shown in Fig. 79, the logometer coils being shunted, one about a manganin resistance r and the Logometer Coil nrtnro innnr Manganin lAAAAAAAAAA Platinum IAAAAAAAA/V Thermometer Fig. 79. Simple Logometer Circuit. other about the platinum thermometer of resistance />; when if s and s' are the coil resistances, their currents are and -^, s s where i = current in principal circuit, and their ratio is *- , a quantity whose variations depend only on p, the resistance of the platinum thermometer, neglecting any variations in the resistance of leads to the logometer coils. The logometer dial, over which moves a pointer attached to the moving coils, may therefore be graduated directly in degrees of temperature. The readings of the instrument are independent of the value or of variations in the current i. Measurements may be taken with alternating currents, and when the manganin and platinum coils are noninductively wound, the readings will be independent of changes in voltage and frequency. This instrument may be arranged to develop relatively powerful directing couples and is therefore readily rendered recording. ELECTRICAL RESISTANCE PYROMETER 225 Northrup's ratiometer, similar to the preceding, is also an adaptation of the deflection-ohmmeter principle to temperature measurements. Northrup makes use of the three-lead thermom- eter with connections as shown in Fig. 80, in which C\ and C 2 are two flat coils mounted on a damped movable system between the two crescent-shaped pole pieces of a permanent magnet. This instrument, which is made in a compact form and read by an attached microscope, can be made sensitive to about 0.1 C. and constant to better than 2 C. Fig. 80. Northrup's Ratiometer. The Cambridge Deflectional Instrument. The Cambridge Scientific Instrument Company also have recently devised a deflectional method for the measurement of temperature with resistance thermometers, in which the temperature is indicated by the " out-of -balance " current in a Wheats tone bridge, pro- vided with compensating leads; the arms of the bridge are all fixed resistances except the one which forms the resistance ther- mometer. As designed, provision is made for exactly setting the zero of the indicator for a balance of the bridge, for adjusting the current to give the required deflection for the temperature range, and is provided with an " ice coil " for balancing the ther- mometer at o C. 224 HIGH TEMPERATURES The Whipple indicator shown in Fig. 81 is a dial instrument using the Wheatstone bridge principle. The galvanometer needle has to be brought to rest by turning a knob, when the dial reading gives temperatures directly. The Leeds and Northrup Indicators. This firm has brought out several patterns of balance and deflection indicator instru- ments based for the most part on the use either of the differential galvanometer (page 209) or the Kelvin bridge (page 215). A very convenient deflection indicator with adjustable scale is shown in Fig. 82. The dial may be set to an}^ desired tern- Fig. 8 1. Whipple Indicator. perature, and the position of the deflector needle indicates how much higher or lower the furnace is than the required tempera- ture. The deflector has a very open scale, permitting readings to be taken from a distance. The workman has to concern himself only with deflections of the needle from the vertical. The accuracy is independent of voltage fluctuations, and the instrument may be run, if desired, from the ordinary lighting circuit. Temperature intervals as small as two degrees are readable. Calibration. For platinum thermometers which are to be used with some form of calibrated resistance-measuring appa- ratus, such as described above, it is only necessary, in order ELECTRICAL RESISTANCE PYROMETER 225 to calibrate the thermometer, to take its readings at three tem- peratures, as at the ice point, the steam point, and the boiling point of sulphur, when, if the wire is of pure platinum, the temperatures found by using Callendars method of computation (see page 201) will be correct to as close as they are known in terms of the gas scale to 1100 C. Fig. 82. Deflection Indicator of Leeds and Northrup. There is advantage in using a fourth calibration point, as the silver freezing point, or that of Ag 3 Cu 2 , in calculating the value of 5 (page 201) for impure wires that are to be used at high temperatures. For the whole range of temperatures with such a wire, both the sulphur and silver points may be obtained, when 8 takes the form a + bt. For thermometers to be used with direct-reading temperature indicators, it is necessary to compare their readings with those 226 HIGH TEMPERATURES of a standard at several temperatures, preferably in a resistance furnace of the Heraeus type. This second method of cali- bration is usually less accurate than the first. Methods of realizing experimentally the sulphur point and other fixed temperatures will be described in Chapter XI on Standardi- zation. The platinum thermometer may be, and should be, for techni- cal work, so constructed as to read directly in platinum degrees, or still better in degrees of temperature. This method saves much time and chance of mistake. The calibration curve, once made for an instrument, serves indefinitely, with occasional checking up if used at high temperatures; so that in spite of the appearance of complications in this method of measuring tem- peratures, actually in practical use the determination of a tem- perature on the normal scale by the platinum thermometer is the affair of a few seconds only. Reduction Tables. In the Appendix are given tables for the reduction of platinum temperatures to centigrade temperatures for wires of pure platinum, correction tables for wires of impure platinum, and other auxiliary tables. Some Results Obtained. There is a remarkable agreement among the fixed points obtained by several observers using the platinum thermometer, for observations extending over twenty years, as shown in the following tables, in which all the obser- vations were obtained by calibrating the platinum thermometer in ice, steam, and sulphur vapor, the temperature of this last being here taken as 444. 70 on the constant-volume nitrogen scale, the value best representing the work of these observers except Holborn and Henning. SCALE OF RESISTANCE THERMOMETER. BOILING POINTS. Naphthaline. Benzophenone. Callendar and Griffiths (1891) 217.97 305.89 Travers and Gwyer (1905) 218.07 305.87 Holborn and Henning (1908 and 1911) * 217.96 305.89 Waidner and Burgess (1910) 217.98 306.02 * S.B.P. = 444.51 C. at constant volume. ELECTRICAL RESISTANCE PYROMETER 227 SCALE OF RESISTANCE THERMOMETER (Continued) FREEZING POINTS. Sn Cd Pb Zn Callendar and Griffiths (1891) 231.9 320.8 327.8 419.0 Heycock and Neville (1897) 231.9 419 .4 Waidner and Burgess (1909) 231.9 321.0 327.4 419.4 Holborn and Henning (1911) f 231.83 320.92 419.40 FREEZING POINTS. Sb Al Ag Au Cu Heycock and Neville 630.0 656* 961.9 1063.5 1082 Waidner and Burgess 630. 7 658 960.9 1083 * Containing 0.5 per cent impurities. t S.B.P. = 444. Si C.at constant volume. Use as a Standard. In 1899, Callendar, at a meeting of the British Association for the Advancement of Science, in view of the relative ease and great precision of resistance measurements and the great difficulties in the use of the gas thermometer, suggested that the platinum thermometer be adopted as a secondary standard, reducing its readings as above indicated, and assuming as calibration points, o, 100, 444.5, the last being the sulphur boiling point on the constant-pressure scale. All platinum thermometers could then be compared with one selected as standard and calibrated as above indicated. He also pointed out that as regards portability and ease of reproduction, it is sufficient to send a few grams of the standard wire in an ordinary letter, to reproduce the scale with the utmost accuracy in any part of the world. The work done in platinum and gas thermometry since 1899 abundantly justifies Callendar's suggestion of using the platinum thermometer as a secondary standard, since, as has been shown in preceding paragraphs, a resistance thermometer of pure platinum calibrated at three temperatures reproduces the gas scale with the greatest exactness to as high temperatures as the platinum thermometer can be used conveniently. It is not necessary, how- ever, to compare a platinum thermometer with another taken as standard, if means are at hand for an independent calibra- tion, since the characteristic constants of pure platinum are now known, and this metal can easily be had of sufficient purity 228 HIGH TEMPERATURES to satisfy them. It is perhaps better to take the sulphur boiling temperature on the constant-volume scale, as most of the recent work in the determination of fixed points has been in terms of this scale. For a platinum-resistance thermometer to serve as a secondary standard, therefore, provided its construction and use are other- wise correct, it is necessary and sufficient that its value of 6 = 1.50 (or 1.49 according to Holborn and Henning's scale) and of C= 0.0039, when calibrated in ice, steam, and sulphur vapor. (See chapter on Standardization.) Sources of Error in Accurate Work. Heating by the Meas- uring Current. It is evident that if a too large current is sent through an electrical resistance thermometer, the heating thus occasioned will cause the indicated temperatures to be high. The limiting value of the current Callendar has shown to be about o.oi ampere per o.oi degree with an average platinum thermometer of wire 0.15 mm. in diameter. If a galvanometer of sufficient sensibility is used this effect is negligible, and when a greater current has to be used on account of lack of galva- nometer sensibility, the heating effect may be maintained nearly constant by keeping the current constant by means of a rheostat in the battery circuit, since the resistance of the thermometer increases very nearly as fast as the rate of cooling, or a little faster than the temperature. Callendar also indicates that the heating effect is readily measured by using as current source two storage cells, connected first in parallel and then in series, the current heating correction being given by subtracting from the first reading one-third of the difference between the two readings. Waidner and Eurgess have also studied the heating effect of the measuring current and find that although this current may heat the coil to more than i C. above its surroundings, the value of the fundamental interval of the thermometer remains the same as when a current one-fifth as great is used. The effect of using different measuring currents with a ther- mometer of RQ = 348 co is shown below: ELECTRICAL RESISTANCE PYROMETER 229 HEATING EFFECT OF MEASURING CURRENT. Amperes. /? P.I. #444.33 Pts.B.f. * 2.5-IO- 3 3.48160 I.34H5 9.13220 421-33 !-503 io. o io- 3 3-48i74 I.34H3 9-J32I3 421.31 1-505 50.0 io~ 3 3.48705 1.34114 9.13608 421.21 1.511 loo.o-io- 3 3-50373 I-34I73 9- I 4832 420.69 1.545 HEATING OF PT COIL ABOVE SURROUNDING TEMPERATURE. Amperes. AT &T m 2.5 IQ s O.OOI O.OOI .... IO.O ID" 3 .Oil .OIO .... 5O.O IQ 3 .41 .41 .29 IOO.O IQ 3 1.65 1.69 1. 2O For a given small excess in temperature of the platinum coil above the temperature of its surroundings, the energy radiated in steam is 3.7 and in sulphur 52 times that radiated at o C. r assuming that the radiation from platinum is proportional to the fifth power of the absolute temperature. For constant measuring current, the energy supplied to the coil at the S.B.P. is only 2.6 times the energy supplied at o C. It follows that the greater part of the energy loss is by convection and conduction rather than by radiation. Lag of the Platinum Thermometer. Inclosed as it necessarily is for most work in a sheath of porcelain, and possessing besides considerable mass, the platinum thermometer does not immedi- ately assume the temperature of its surroundings. Put into a sulphur bath, it assumes an equilibrium condition in ten minutes. For small changes of temperature this effect is hardly perceptible and may be neglected in most practical work. Inclosed in a thin flat-sided metal case (see Fig. 66), the tem- perature lag is practically nothing. Insulation. Defective insulation due to moisture condensed in the tubes is sometimes a source of error in accurate work at the ice point and lower temperatures with thermometers of high resistance if the tubes are not sealed. This may be readily done, if the containing sheath is of glass, by sealing the platinum leads into the glass so that they terminate in cups. When the con- taining sheath is of porcelain, as for high-temperature work, this 230 HIGH TEMPERATURES sealing is not necessary, nor is it convenient; but running the leads into metal cups containing a fusible alloy still offers the readiest method of securing a good contact with the rest of the circuit. Compensation for Resistance of the Leads. It is necessary, in order to avoid thermal currents at the junctions with the ther- mometer proper and also evaporation and consequent change of resistance, to employ platinum leads from the thermometer to a point in the circuit at a constant temperature. Even if these leads are of relatively large diameter, there will still remain an error due to the varying resistance of these leads with change in temperature and with varying depth of immersion. It becomes necessary either to apply a "stem correction," which is trouble- some and uncertain, or compensate for this effect as described under methods of measurement. Nowadays most platinum ther- mometers sold for industrial and scientific purposes are compen- sated. Uncompensated thermometers with gold leads are also to be found. They are not to be recommended for work of high accuracy. Silver leads are to be avoided. The copper leads from the thermometer head to the measuring apparatus may be of appreciable resistance, and to render them flexible they are often stranded, when their resistance may vary somewhat. Thus Mr. F. W. Smith has found copper leads of -^Q co resistance to vary i or 2 per cent, giving 0.003 C. uncer- tainty at o C. In work of high accuracy it is evidently as im- portant to keep the copper leads as .constant as the platinum. It is now possible to obtain stranded wire in which each strand is enameled, and so eliminate the slip resistance. In addition to the potential lead method, there have been bridge methods devised for the complete experimental elimina- tion of all leads to the thermometer which is required in work of the highest accuracy, as it is extremely difficult, if not impos- sible, to make the compensation absolutely exact by adjustment in construction. These methods require, for the most part, rather elaborate experimental arrangements, for descriptions of which the reader is referred to the papers of Edwards, ELECTRICAL RESISTANCE PYROMETER 231 W. Jaeger, and F. W. Smith. In brief, such methods depend either upon devices for alternately throwing the thermometer leads in the two sides of the bridge, measuring these leads, or eliminating them by use of the Kelvin double bridge or some modification. Conduction along the Leads. The thermometer leads may be the seat of another source of error, which increases in importance with the diameter of the leads, their length immersed, and the temperature gradient, namely, the effect of heat conduction along the leads influencing the resistance of the thermometer coil. This effect is especially to be looked for in four-lead bridge ther- mometers, where all four leads are of relatively heavy platinum. The best way to eliminate this source of error is to so design the instrument that it is negligible. Its presence may be recognized and corrected for by varying the depth of immersion of the ther- mometer in a bath at constant temperature. Use of Impure Platinum. The value of the constant 5 in the Callendar formula (2), page 201, is a measure of the purity of the metal. For the purest platinum the value of 5 is 1.500, assuming the S.B.P. = 444.70, and for impure platinum the value of 6 increases with the impurity. Heycock and Neville made a study of the effect on the temperature scale obtained by using platinum of different degrees of purity, and concluded, erroneously it now appears, due to an incorrect method of calcu- lation, that thermometers having different values of d would give the same temperature scale when reduced by the parabolic for- mula of Callendar. It has been shown since that impure platinum does not obey the same resistance-temperature law as the pure metal, and Waidner and Burgess have indicated the corrections to be applied to temperatures obtained by the Callendar method, using impure platinum to reduce to the usual temperature scale. They find, for platinum of varying degrees of purity as indicated by the values of 5, the following values for fixed points, using the Callendar equation in all cases for computing the temperatures: 232 HIGH TEMPERATURES FREEZING POINTS FOR VALUES OF 5 INDICATED BELOW. 5= 1.505 1-570 1.803 Tin 231.90 231.82 Zinc 4I9-37 419-32 Antimony... 630.70 631.25 632.65 Ag 3 -Cu 2 .... 779-2 784-6 Silver 960.9 966.2 975-3 Copper 1083.0 1092.0 1106.0 In Table VIII of the Appendix are indicated the corrections to be applied when using thermometers of impure platinum. Wires with a large 5 are more liable to change with use, so that, although correct results may be obtained with them if properly reduced and checked up occasionally, it is preferable to use the purest platinum. Changes in the Constants. If platinum thermometers are re- peatedly heated to temperatures in the neighborhood of 1000 C., or are kept for very considerable periods of time at even lower temperatures, changes in the value of the constants R Q , Jf2i 00 , and d will develop, necessitating frequent recalibration in work of high accuracy. Pyrometers for use at high temperatures should not be inclosed in inglazed porcelain even if the glaze does not touch the metal, as deterioration of the latter will other- wise ensue. The mica supports undergo distortion on cooling from high temperatures, increasing in size, tending to stretch the wire and increase its resistance. For this reason it is probably better to use the constants determined before a measurement at high temperature, rather than those determined afterwards. Again, if the wire of the thermometer has not been well an- nealed at a temperature higher than it is to be used, irregular changes will occur, which are the most marked for the first few heatings. Waidner and Burgess find that for thermometers of pure platinum, the changes in their constants after the wires have been annealed are very much less than for those of impure platinum; thus, as shown in the accompanying table, which is typical, Ro changes only by a few tenths of a degree for pure platinum, but by several degrees for impure. These changes are ELECTRICAL RESISTANCE PYROMETER 233 CHANGES IN ZERO OF PLATINUM THERMOMETERS. Thermometer of pure platinum. Thermometer of impure platinum. R =3- 47971 at start; 6 = 1.503. ^0=3.48164 at end; diam.= .15 mm. RO= 21 .3476 at start; 5=1.570. Ro=2i .0617 at end; diam.= . 10 mm. Changes in zero. C. History of thermometer previously annealed at 1200. Changes in zero. C. History of thermometer previously annealed at 1200. -0.005 ' .001 -f- .007 .002 .050 + .013 + .138 + -144 After Zn P.P. 3 times. After Sb P.P. i time. After Sb P.P. 2 times. After 2 hrs. at noo+ C. After Cu P.P. i time. After Cu P.P. 2 times. After Cu P.P. 5 times. After Ag -Cu P.P. 5 times. -0.18 - .29 -2.27 -3-94 -4.66 -5-99 6. 20 -6.46 After Zn P.P. 10 times. After Sb P.P. 7 times. After 2 hrs. at noo+ C. After Cu P.P. i time. After Cu P.P. i time.' After Cu P.P. 2 times. After Ag P.P. 2 times. After Ag - Cu P.P. 4 times. least for pure platinum wire of large diameter and suspended free from strains. For impure platinum wire, the effect of high temperatures is to decrease R Q and to increase the fundamental coefficient, c\ that is, the effect is as if the wire became purer, possibly because of the evaporation of impurities, for example, indium. If the platinum is pure, the slight changes indicate a contamination of the wire and the effect of strains, as is evidenced by decrease in c and increase in R Q . The total change observed is the resultant of the effects of strains, of annealing, and of contamination and purification. Use of Metals other than Platinum. Holborn and Wien found that with palladium the absorption of hydrogen at low tem- peratures, giving the hydride, increases the resistance by 60 per cent; besides, the same effect of alteration as with platinum is noticed if the palladium is placed in hydrogen in the presence of silica. Palladium wound on mica and inclosed in porcelain has been shown by Waidner and Burgess to behave in a very similar manner as platinum to above ioooC.; the law of the change of resistance of palladium with temperature is, however, very different from the Callendar equation, and is an equation of the fourth degree between o and 1100 C. for a precision better 234 HIGH TEMPERATURES than 0.5 C., although up to 600 the Callendar equation is nearly satisfied. No very definite conclusion is to be drawn from the work of Holborn and Wien with iridium and rhodium, except that these metals assume their normal resistance only after being heated several times to a high temperature. Iridium evaporates so much more readily than the others that it would seem the least best adapted for temperature measurement by means of the metals of this group, and platinum is evidently to be preferred. Nickel is sometimes used in resistance thermometers, but it is not -to be recommended for temperatures above 300 C., due to the change in the resistance-temperature relation as the tran- sition temperature of nickel is approached and to oxidation at higher temperatures. Marvin has shown that for pure nickel the equation log R = a + mt holds approximately in the above limited range o to 300 C. Conditions of Use. The electrical resistance pyrometer of platinum seems, by reason of the great precision of the measure- ments which it allows, to be especially serviceable for laboratory investigations. It seems, on the other hand, to be too fragile for many of the industrial applications when there is rough handling, although it is very convenient in permanent installations when properly protected, and when it is desired to eliminate completely the often troublesome correction necessary for the temperature of the cold junction of the thermocouple. The relation between the platinum-thermometer scale and the gas scale is well established to 1100 C., which is beyond the limit above which it is not safe to use the platinum-resistance pyrometer without frequent checking of its calibration. The resistance pyrometer is the best instrument for differential work and for detecting small temperature changes as well as for controlling a constant temperature. It is also particularly adapted for use with recording instruments. Great care has to be taken that the platinum does not become contaminated. Industrial Installations and Checking. We have already called attention to the fragility of the fire end of a resistance ELECTRICAL RESISTANCE PYROMETER 235 thermometer and the necessity for protection of the coil from contact with furnace gases. In industrial installations it is preferable to so mount the py- rometer that it may not readily be damaged by the furnace op- erations or by the handling of the pyrometer when necessary to withdraw it. This can usually be done by a suitable arrange- ment of the pyrometer within the furnace and by providing a convenient mechanism for with- drawing and holding the py- rometer free of the furnace. A design of bracket by Leeds and Northrup is shown for use with a small oil- or gas-burning fur- nace in Fig. 83. A sliding and slotted collar L carries the py- rometer on the arm Nj the whole may be raised, caught, and Fig. 83. Bracket Mounting. Fig. 84. Mountings in Oven. turned on the pin M, permitting the removal of the pyrom- eter without shock and providing a resting place without 236 HIGH TEMPERATURES handling or fear of breakage. Proper designs for mounting py- rometers in furnaces, kilns and duct pipes are shown in Figs. 83, 84, and 85. If the pyrometer tube be inserted horizon- tally supported only at one end, there is danger of bending and breaking even when the outer sheath is of metal. The resistance pyrometer and its electrical circuit may be tested in place and the calibration verified without removal. An industrial installation should always be tested for proper insula- Fig. 85. Mounting in Duct. tion, not only when new but periodically, or when irregular be- havior occurs. The actual operations of checking out the insula- tion, lead, and contact resistances will depend upon the design of the instrument and the voltage for which it is intended. It is safe to say that the resistance between any wire and the ground or thermometer case, or between two disconnected wires of the system, should be over i megohm per 100 volts. Some of the manufacturers provide outfits for the automatic checking of thermometer indicators and coils. Thus, Leeds and Northrup furnish an equipment consisting of a series of coils cor- ELECTRICAL RESISTANCE PYROMETER 237 responding to definite temperatures on the indicator, and another coil equal to the resistance of the thermometer at room tempera- ture. The Cambridge Company also furnish " ice bobbins " by means of which the thermometer resistance may be checked at o C. The use of the resistance pyrometer industrially is also greatly facilitated by the practice that is becoming general among makers of constructing the instruments and all parts so that they are interchangeable. This is particularly necessary for multiple circuits using the same indicator or when using a single automatic recorder in connection with a number of indicators. These questions are considered in Chapter X. CHAPTER VI. THE LAWS OF RADIATION. General Principles. The temperature of bodies may be estimated from the radiant energy they send out, either in the form of visible light radiation or of the longer infra-red waves that are studied by their thermal effects. For the estimation of temperature in this way use is made of the laws of radiation. Temperature and Intensity of Radiation. When we consider the enormous increase in the intensity of radiation with rise in temperature, this method appears especially well adapted to the measurement of high temperatures. Thus, for example, if the intensity of the red light (X =0.65 /*) emitted by a body at 1000 C. is called i, at 1500 C. the intensity will be over 130 times as great, and at 2000 C. over 2100 times as great. The rapid increase of the photometric intensity of the light in comparison with that of the temperatures is shown by the follow- ing table, from Lummer and Kurlbaum, for light emitted by incandescent platinum. If I\ and 7 2 are the intensities of the total light emitted at the absolute temperatures T\ and Tz (not differing many degrees from one another), then if we write with Lummer and Pringsheim the values of x at various absolute temperatures (T C. 4- 273) are as follows : T abs. x. 900 30 1000 25 IIOO 21 I2OO IQ 1400 18 1600 15 1900 14 238 THE LAWS OF RADIATION 239 From this table it will at once be seen that at 1000 absolute (727 C.) the intensity of the light increases twenty-five times as rapidly as the temperature; at 1900 absolute (1627 C.) fourteen times as rapidly. The product r# = 25,000 as shown by Rasch seems to express the relation between T and the exponent x. Emissive Powers. It would therefore appear that a system of optical pyrometry based on the intensity of the light emitted, by incandescent bodies would be an ideal one, inasmuch as a comparatively rough measurement of the photometric intensity would measure the temperature quite accurately. This, however, is only partly true ; it is limited somewhat by the fact that different bodies, although at the same temperature, emit vastly different amounts of light. Thus the intensity of the radiation from in- candescent iron or carbon at 1000 C., for example, is many times greater than that emitted by such substances as magnesia, polished platinum, etc., at the same temperature. Consequently, if any conclusions were drawn as to the temperatures of these bodies from the light that they emit, it might lead to large errors. Thus at 1500 C. this difference in the intensity of the light emitted by carbon and by polished platinum would lead to a difference in the estimated temperature of these bodies of about 1 00 C., and less at lower temperatures. The "Black Body" Kirchhoff in one of the most important contributions to the theory of radiation was led to the important conception of what he termed a "black body," which he defined as one which would absorb all radiations falling on it, and would neither reflect nor transmit any. He further pointed out clearly the important fact that the radiation from such a black body was a function of the temperature alone, and was identical with the radiation inside an inclosure all parts of which have the same temperature. Various expressions are in use for the " black body," such as "integral radiator," "full radiator," etc. Experimental Realization. The first experimental realization of a black body as a practical laboratory apparatus was made by Lummer and Wien, by heating the walls of a hollow opaque inclosure as uniformly as possible and observing the radiation 240 HIGH TEMPERATURES coming from the inside through a very small opening in the walls of the inclosure. No substance is known, however, whose surface radiation is exactly that of a black body. The radiations from such substances as carbon and iron oxide approximate fairly near to black-body radiation, while such bodies as polished platinum and magnesia, etc., depart very far from it. Black-body radia- tions, corresponding to temperatures from that of liquid air or lower, up to 2500 C. or higher (if suitable materials are chosen), are now available in the laboratory. For temperatures up to 600 or thereabouts, this is realized by immersing a metallic or other vessel in a constant-temperature bath (liquid, gas, vapor, or fused salt) and observing the radiation from the interior through a small opening in the walls. At higher temperatures it is very difficult to heat the walls of the inclosure uniformly, especially with gas flames. Lummer and Kurlbaum have very satisfactorily overcome this difficulty in their electrically heated black body, which is shown in section in Fig. 86. The central porcelain tube is wound over with thin platinum foil through which an electric current is sent which can be ad- justed to maintain any desired temperature up to 1500 C. This tube is provided with a number of diaphragms to minimize the disturbing effects of air currents. To protect this inner tube from external influences and to diminish unnecessary heat losses, it is surrounded by several porcelain tubes and air spaces, as shown in the figure. The radiation from the uniformly heated region near the center, and which passes out through the end of the tube at 0, is a very close approximation of the ideal black- body radiation of Kirchhoff. The temperature of this central region is measured by means of one or more carefully calibrated thermocouples. By adding supplementary heating coils at the ends the temperature distribution may be improved. Waidner and Burgess were able to obtain a constancy of i C. throughout the greater part of the length of such an apparatus. The cali- bration of optical and radiation pyrometers is carried out by means of such a black body. For higher temperatures special furnaces are used, which we shall describe later. THE LAWS OF RADIATION 241 3 a 242 HIGH TEMPERATURES As has already been stated, if magnesia, porcelain, platinum, iron, etc., are heated to the same temperature, they will emit vastly different amounts of light. If, however, these bodies* are heated inside a black body, they will all emit the same radia- tion, and on looking into the small opening all details of. their contour will be lost, the whole region being of uniform brightness. Thus, in the black body described above, before the heating has become uniform, the platinum wires of the thermocouple can be seen as dark lines against the brighter background, but when the heating current has been maintained constant for some time, so that the heating has become uniform in the inner central chamber, the wires of the couple almost completely disappear, notwith- standing that, of all substances, platinum and the black oxide of the radiating walls differ most widely in their radiating powers (emissivities). Realization in Practice. Fortunately, in pyrometric practice it is often easy to realize very nearly the conditions of a black or totally absorbing body. Thus the interior of most furnaces, kilns, and ovens approximates this condition, or the bottom of a closed tube of any material thrust into any space heated to in- candescence. Again, iron and coal observed in the open are not far removed in their optical properties from the black body. Black-body Temperature. The term " black-body tempera- ture " has come into quite extensive use and is of great conven- ience in the discussion of pyrometric problems. The tempera- tures indicated by a radiation pyrometer that has been calibrated against a black body are known as black-body temperatures. Thus, were a piece of iron and a piece of porcelain both at 1200, the optical pyrometer, which used the red light emitted by these bodies, would give, as the temperature of these bodies, 1140 and 1 1 00 respectively. This means that iron and porcelain at 1200 emit red light of the same intensity as is emitted by a black body at 1140 and nooC. respectively. The " black- * It is here assumed that the radiation is purely thermal and that no part is due to luminescence, as the laws of radiation are only directly appli- cable where such is the case. THE LAWS OF RADIATION 243 body temperature " of these materials for green light might differ quite appreciably from that for red light. It is at once evident that if the " black-body temperatures " of different bodies, e.g., carbon and platinum, are equal, their actual temperatures may differ considerably (180 C., or so, at 1500 C.). This violates our ordinary conception of equal temperatures, which is based on thermal equilibrium between the bodies if brought into contact. The term " equivalent temperature," suggested by Guillaume, is also used for " black-body temperature." Waidner and Burgess have suggested the notation: 1500 K&, meaning 1500 absolute centigrade as viewed with light of wave length 0.65/4. The temperature of any body, therefore, as measured by an optical or by a radiation pyrometer, will always be lower than its true temperature by an amount depending on the departure of its radiation from that of a black body. There is another source of error, however, that may act in the direction of making the pyrometer read too high, due to light reflected by the body whose temperature is being measured. This source of error may very often be eliminated, where the accessibility of the work permits, by running a tube down to the incandescent surface, which will cut off stray radiation from the surrounding flames. The magnitude of the error that may arise from light reflected from surrounding hotter objects may be quite considerable (several hundred degrees), depending on the temperature, area, and posi- tion of the surrounding hot objects and the reflecting power of the surface whose temperature is under observation. Kirchhoff 's Law. If we consider an opaque object and let radiation fall upon it, the relation between the proportions re- flected (r) and absorbed (a) is: r + a = i. For such objects, therefore, eliminating the effects of polarized light and angle of incidence and assuming we are dealing with matt surfaces and thermal radiation only, the determination of either the absorbing or reflecting power gives also the other. 244 HIGH TEMPERATURES The quantity a depends on the nature of the substance and is a function of the wave length and temperature only, or a =f(\,T). For a radiating body the emissive power e is a similar function of X and T. By definition, a black body absorbs all the radiation incident upon it, therefore in this case a = i and r = o for all values of X and T. The emissive power of a radiating black body is evidently fundamental to the theory of radiation, and the func- tion e = F(\,T) forms the basis of several of the radiation laws. KirchhofFs law as applied to monochromatic radiation may be stated in the following way: e = ae, or more completely: a a n The ratio of the emission to the absorption is for all bodies the same function of wave length and temperature, and is equal to the emission of a black body. There are a number of corollaries to KirchhofFs law, some of which we may emphasize as being of interest in temperature measurements, bearing in mind that we are here dealing only with radiation due to thermal causes. The emissive power of a black body is greater than that of any other body at the same temperature. Every body absorbs the same rays that it emits at a given temperature. It may also absorb other rays, but they will be among those that a black body does not emit at the given temperature. In general, the ratio e : a, which is the same for all bodies for given values of X and T, does not depend on the degree or kind of polarization of the radiation. The energy curves e =/(X) for each value of T lie wholly within the corresponding black-body curves e = /(X). In the case of composite radiations, that is, of spectral bands having for limiting case the whole spectrum, KirchhofFs law ap- plies only under special conditions. Thus KirchhofFs law holds for any composite radiations, taken between the limits Xi and THE LAWS OF RADIATION 245 ^2, if the total absorption is referred to the radiation from a black body at the same temperature as the bodies to be compared. Again, Kirchhoff's law holds for composite radiations, when the two given bodies are at the same temperature and when each of them serves as source to the radiation which measures the total absorption of the other. A corollary of considerable practical importance is the follow- ing: If two surfaces of any substances whatever at the same temperature radiate only on each other, the radiation from each is equivalent to the emission of a black body; from which it follows that, within an inclosed space at constant temperature, all bodies emit radiation identical to that of a black body. And finally, the radiation from a small opening in such an inclosure at constant temperature is black-body radiation, and depends only on the temperature. It is worthy of remark that in the measurement of radiation from such a black body the receiver should also be a black body, or at least its coefficient of absorption should be known for the kind of radiation to be studied, if the radiation laws as applied to a black body are assumed to hold, as is often the case. Stefan's Law. Naturally the first numerical relation sought between intensity of radiation and temperature was one for the total energy of radiation sent out by a body, as it required less delicate instruments for measurement than the study of the spectral distribution of energy. Numerous attempts to express such a relation were made by Newton, Dulong and Petit, Rosetti, and others. These attempts, however, merely resulted in empiri- cal expressions that held only through narrow ranges of tempera- ture. The first important step was made by Stefan, who ex- amined some of the experimental data of Tyndall on the radiation of incandescent platinum wire in the interval 525 C. to 1200 C., and was led to the conclusion that the energy radiated was. proportional to the fourth power of the absolute temperatures. This relation seemed to be further supported by the best experi- mental data of other observers, at least to within the limit of accuracy of their observations, being strictly true, however, only 246 HIGH TEMPERATURES for the energy of total radiation from a black body. This rela- tion received independent confirmation from Boltzmann, who deduced it from thermodynamic reasoning. The conditions im- posed by Boltzmann in his discussion on the nature of the radi- ation were such as are fulfilled by the radiation from a black body. This relation, which has now come to be generally known as the Stefan-Boltzmann radiation law, may then be stated as follows : The total energy radiated by a black body is proportional to the fourth power of the absolute temperature, or (B) when E is the total energy radiated by the body at absolute temperature T to the body at absolute temperature T, and a- is a constant depending on the units used. Usually T Q is small compared with T, so that practically we may write (B r ) E = oT 4 . This law has received abundant experimental support from the researches of Lummer, Kurlbaum, Pringsheim, Paschen, and others, throughout the widest range within which temperature measurements can be made. An illustration of the experimental evidence in support of this law is given in the table taken from the experiments of Lummer and Kurlbaum. E r* - r r n Black body. Polished platinum. Iron oxide. 372.8 290.5 108.9 492 290 109.0 2.28 33-1 654 290 108.4 6.56 33-1 795 290 109.9 8.14 36.6 1108 290 109.0 12. 18 46.9 1481 290 110.7 16.69 65.3 1761 290 19.64 THE LAWS OF RADIATION 247 It will also be seen from this table that while the intensity of the total radiation of iron oxide is 4 or 5 times that of polished platinum, it is still considerably less than that emitted by a black body. The total radiation from objects other than a black body increases more rapidly than the fourth power of the absolute temperature, so that as the temperature is raised the radiation of all bodies appears to approach that of the black body. Whether or not there is a maximum limiting value of radiation due to purely thermal causes, is still an unsettled question, however. The numerical value of the constant a for a black body is of interest in absolute measurements and in checking the constants of radiation instruments. For the radiation per degree C. from i cm. 2 , expressed in gram-calories per second, the values found for a- range from less than i.o-io~ 12 by Bottomly and King to 1.52 io~ 12 by Fery. The following observers, however, have obtained results agreeing more closely: Kurlbaum, 1.277 I0 ~ 12 Valentiner, 1.286 Bauer and Moulin, 1.275 ,, _ 19 19 watts Mean = 1.279 10 5-34 * I0 2 ' cm. The last two series of measurements were carried out over very extended temperature intervals in the case of Valentiner to nearly 1600 C. His results correspond to the temperature scale, established by Holborn and Valentiner. It should be noted, however, that if compensating errors are present in the energy measurements or in the non-blackness of the radiator or receiver, it is possible to obtain a correct value of a > 5. The corresponding form assumed by Stefan's law for non-black bodies is E = c/r"- 1 , . (BO where a is the same as in the preceding equation. Lummer and Pringsheim found the following limits of tempera- ture as given by the Wien relation (la) : \ m Tmai. Train Electric arc O 7u 4200 abs. 37 corresponding to X = 460. For very small objects which would have to t>e placed very near, a supplementary objective is put in front of the telescope; the object is placed in the principal focus of this new lens, the objective of the apparatus being focused for parallel rays. The absorptive power of this supplementary lens is reckoned as T V ^ Details of an Observation. The first operation to make is the determination of the absorption coefficients of the absorbing glasses. For that, an object of suitable brightness is viewed once with the tinted glass before the cat's-eye and then without this glass. Let N be the aperture of the cat's-eye without tinted glass, and N f the aperture with such a glass. The coefficient k of absorption is -(I)' The following observations furnish data for the determina- tion of the absorbing powers of different glasses employed in the course of studies relative to the radiations from incandescent mantles. OPTICAL PYROMETER ABSORBING GLASS PLACED BEFORE THE SOURCE TO BE STUDIED. Temperature. Aperture of cat's-ye. Red. Green. Blue. 1270 1270 (-f- 1 glass) . IQ-5 5-5 21.2 7-9 35 II .1 (no glass) k r = 12.5 kg = 7 2 5 = 9.9 ABSORBING GLASS PLACED BEFORE THE STANDARD LAMP. 1170 ( i glass) 2 Q tr .QC IO.2 1170 (no glass) 4 16.1 31 .5 k r = 10. 5 kg =7 -3 *.-,.s Emissive Power. Before being able to establish the relation which exists between the intensity of radiation of incandescent bodies and their temperature, it is necessary to know the emissive powers of these bodies (see page 293). For this measurement use was made by Le Chatelier of the principle stated above, that the interior of fissures in bodies may be considered as inclosed in an envelope at uniform temperature. The emissive power is thus, at the temperature considered, equal to the ratio of the luminous intensity of the surface to that of the bottom of deep fissures, with the condition, evidently, that the aperture of the fissures be sufficiently small. The body to be studied was placed in the state of a paste, as dry as possible, on the end of a couple previously flattened so as to take the form of a disk of 2 or 3 mm. diameter. The drying was very slow, so as not to have any swelling of the mass, and one obtained in this way a coating possessing fissures; the conditions described above are then satisfied. The end of the couple thus prepared is heated either in a Bunsen flame or a blast lamp, and the temperature of the junction is noted, while, simultaneously, readings are taken with the optical pyrometer. In order to 302 HIGH TEMPERATURES obtain a temperature as constant as possible, it is necessary to guard against currents of air and use a flame of small size. Here are some results obtained: I. COUPLE COVERED WITH A MIXTURE CONTAINING 99 PARTS OF THORIUM AND 1 OF CERIUM. Temperatures. 0^0 ( i glass). Red. (I) (2) 16 o ... Green. Blue. (I) (2) (I) (2) 21 .O I4.O 23 .O .... II .O 9.O 12. 12. 0- 4-5 3-2 3-5 3-5 2.O 2.0 1.9 1.9 5-0 4-0 . 15.5 9.0 3-0 2.0 6.0 137"? 7.0 1525 3.2 1650 (+ i glass) 8.3 II. MAGNESIA. 1340 ( i glass). .. ................. 12.2 4.0 18.5 6.7 19.0 9.0 1460 ( i glass) .................... 4.9 2.5 8.2 3.1 7.7 4.1 1540 ( i glass) .................... 2.4 1.3 3.1 1.8 3.2 2.1 The numbers give the divisions of the cat's-eye; those of column (i) refer to the surface, and those of column (2) to the bottom of the fissures. The indications ( i glass) and (+ i glass) mean that the absorbing glass is placed either before the standard lamp or before the source studied. Measurements of Intensity. The following table gives an idea of the order of magnitude of the intensities of different lumi- nous sources, the measurements of brightness being made in the red. Unity is the brightness of the axial portion of stearine- candle flame. Carbon beginning to glow (600) ................. o.oooi Silver melting (960) ............................ 0.015 Stearine candle ...... ] Gas flame ........... > ......................... i.o Acetate of amyl lamp J Pigeon lamp, with mineral oil .................... i . i Argand burner, with chimney. . '. ................. 1.9 Auer burner ..................................... 2.05 Fe 3 O melting (1350) ............................ 2 . 25 Palladium melting .............................. 4.8 Platinum melting . . ............................. 15 . o Incandescent lamp .............................. 40 Crater of electric arc ........................... 10,000 Sun at midday .................................. 90,000 Calibration. Le Chatelier made a first graduation of his. optical pyrometer by measuring the brightness of iron oxide heated on the junction of a thermoelectric couple, and admitting that, for the red, the emissive power of this substance is equal OPTICAL PYROMETER 303 to unity. He found a law of variation of the intensity of the red radiations as function of the temperature, which is well represented by the formula 3210 / = io 6 ' 7 T T , in which unit intensity corresponds to the most brilliant axial region of the flame of a candle. (T is absolute temperature.) This formula has been shown by Rasch to be equivalent to (B), page 294, for red light, in which a = 13.02. It is therefore a derivative from Wien's law (page 251). The table below gives, for intervals of 100, the intensities of red radiations emitted by bodies of an emissive power equal to unity. These numbers were calculated by means of the inter- polation formula given above. Intensities. Temperatures. Intensities. Temperatures. 0.00008 600 39 ,1800 .00073 700 60 1900 . 0046 800 93 2000 . 020 900 1800 3000 .078 1000 9,700 4000 .24... 1 100 28,000 5000 .64... 1200 56,000 6000 1.63 1300 100,000 7000 3.35 1400 150,000 8000 6.7 1500 224,000 9000 12.9 1600 305,000 10,000 22.4. 1700 These results are represented graphically in Fig. in. After having determined the value of the diaphragm opening d , which gives equality of brightness of the standard candle with that of the comparison lamp, and the absorbing power k of the tinted glasses, one may, as was said before, prepare a table which gives directly the temperature corresponding to each aperture of the cat's-eye. With an apparatus for which do = 5.2, k = iV, the following table is obtained, in which the plus sign refers to tinted glasses placed before the objective, and the minus sign to those before the comparison lamp. This graduation applies to all bodies placed in an inclosure at 34 HIGH TEMPERATURES 5 4 3 2 ^ 1 1. -1 -2 -3 ^-' ^ *'''' ,S 4 /' . / / / / ' / / / "2.9 3 3.1 3.2 3.3 3.4- 3.5 3.6 3.7 3.8 3.9 4 Log. (2+273) Fig. in. Intensities in Terms of Temperatures. the same temperature, in the interior of furnaces for example, and to black bodies whatever the temperature surrounding them; for example, it applies very closely for a piece of red-hot iron exposed to the free air. For bodies whose emissive power is inferior to unity, as platinum, magnesia, lime, it is necessary, when they are exposed to the air and not surrounded by an in- closure at the same temperature, to make a special calibration. TYPICAL CALIBRATION TABLE FOR A LE CHATELIER OPTICAL PYROMETER. Temperatures. 2 glasses, i glass. o glass. +i glass. +2 glasses. 7oo 17-3 800 6.Q 23.0 QOO II. O .... IOOO 5.6 18.6 IIOO , 10.5 1 200 6.5 1300 4-0 13-6 1400 9.4 1500 6.6 I6OO 4-8 1 700 3-6 12. O I800 9.1 IQOO 7-3 2000 5-9 OPTICAL PYROMETER 305 Le Chatelier and Boudouard made a series of measurements on radiations of different wave lengths. The junction of a ther- moelectric couple was placed in a small platinum tube, to realize approximately an inclosed space. By taking as unity the bright- ness of melting platinum, the results obtained are the following for the red, green, and blue radiations: TEMPERATURE VS. BRIGHTNESS (IN TERMS OF MELTING PLATINUM). / Log(* + 2?3) I r Log/ r I, Log/, I b Log/6 900 3.0707 o . 0009 4-95 0.00018 4-25 0.00002 5-3 1180 3.161 .0024 3.88 .0087 3-94 .0015 3-17 1275 3.190 -075 2.78 037 2-57 .013 2. II 1430 3-230 23 1-36 .16 1.6 7 .058 2. 7 6 1565 3-265 -72 1.86 47 i .20 24 I". 38 1715 3-300 i .69 0.23 i-45 o. 16 9 0.95 Evaluation of Temperatures. Finally, Le Chatelier has used his optical pyrometer to determine the very highest temperatures realized in some of the most important phenomena in nature and in the industries. These results, quite different from pre- vious determinations, were at first regarded with considerable reserve; they are admitted to-day as reasonable, at least within the limits of precision, and in terms of the temperature scale used by him. Here are some of the figures obtained: Siemens-Martin furnace 1490 to 1580 C. Furnace of glass works 1375 to 1400 Furnace for hard porcelain J 37o Furnace for new porcelain 1250 Incandescent lamp 1800 Arc lamp 4100 Sun 7600 This determination of the temperature of the sun, generally believed to be low at the time it was found, was confirmed by the experiments of Wilson and Gray (page 271) by a totally dif- ferent method. Later determinations of the sun's temperature, using the more recently established laws of radiation (Chapter VI), give values between 5500 and 6500. A series of measurements were made with the same apparatus in ironworks. Here are some results: 306 HIGH TEMPERATURES BLAST FURNACE SMELTING GRAY PIG. Opening before the tuyere i93 C. Tapping the pig iron, beginning 1400 Tapping the pig iron, end iS 2 BESSEMER CONVERTER. Pouring the slag 1580 Pouring the steel into the ladle 1640 Pouring the steel into the molds 1580 Reheating of the ingot 1200 End of the hammering 1080 SIEMENS-MARTIN FURNACE. Flow of the steel into the ladle, beginning 1580 Flow of the steel into the ladle, end 1420 Flow into the molds 149 Calibration in Terms of Wien's Law. As approximately mono- chromatic radiation is used, the Le Chatelier optical pyrometer may be calibrated in terms of Wien's law (III) (page 251) by sighting upon a black body (Fig. 86) whose temperature is given by means of a thermocouple. For this purpose Wien's law may be written: kg / - 1 ^- kL where / is the intensity of light, in terms of the center of the Hefner flame for example, and T is the absolute temperature. This method of graduation has the advantage that only two points are required to completely calibrate the instrument, for the relation between log 7 and is linear, so that these quantities being plotted give a straight line which may evidently be ex- tended to lower and higher temperatures, since Wien's law has been shown (page 250) to hold over the widest temperature interval measurable, provided the light used is monochromatic and the bodies observed approximate blackness and are not luminescent, that is, their light not produced by chemical or electrical excitation. The value of 7 is given by the equation of page 300, and for a given absorption glass and focus is propor- tional to J 2 . Precision and Sources of Error. We shall give in some detail a discussion of the factors which in the use of the Le Chatelier OPTICAL PYROMETER 307 optical pyrometer may influence the photometric settings and so affect the accuracy of temperature determinations, as results of such a discussion are illustrative of what may be expected from optical pyrometers in general. The results are taken from those of Waidner and Burgess, who have made an experi- mental comparison of all the available optical pyrometers. The sources of error of this instrument may be those due to the standard Hefner amyl-acetate or other standard of constant photometric intensity or temperature placed before the cat's-eye when adjusting the pyrometer, the oil comparison lamp, the focusing system, the nature of the red glass used, and the co- efficients of absorption of the tinted glasses. The first of these affects only comparative results with different instruments, while the others, if they exist, may be of considerable importance in work with a single instrument. We shall consider them in the order named. As only the central portion of the amyl-acetate flame is used, variations in height and fluctuations in total intensity due to various causes such as moisture and carbonic acid in the atmos- phere and changes due to differing samples of acetate become almost, if not quite, insignificant in this method of comparison; so that, when using only a small central area of the amyl-acetate flame, it is a very perfectly reproducible standard under the most varying conditions of burning. Again, the effects of any slight fluctuations in light intensity are further greatly reduced when transformed into temperature changes, as has been shown (page 238). Thus, the effect of varying the height of the Hefner flame by i mm., which amounts to 10 per cent of the total in- tensity when the whole flame is used, causes a change of less than i per cent in the intensity of light from the central area, which is equivalent to less than 0.5 C. change in temperature at 1000 C. Although used intermittently as above indicated, the Hefner serves well enough as an ultimate standard by means of which the indications of all photometer pyrometers may be reduced to a common basis, yet the Hefner is not suited for use as HIGH TEMPERATURES comparison lamp in the pyrometer itself, as has been previously stated. In a study of the constancy of the comparison lamp the fol- lowing arrangement was adopted: In order to obtain a perfectly constant source of light with which to compare the flame, a 32-c.p. incandescent electric lamp was placed in a fixed position before the objective of the pyrometer and a glass diffusing screen inserted before the objective. The voltage across the lamp terminals was kept rigorously constant, thus giving an arbitrary but invariable standard of illumination. The concordance of results obtained by different observers set- ting the gasoline flame and observing is shown below: WITHOUT ABSORPTION GLASS. Cat's-eye scale readings 7-4 7-4 7.2 7-8 7-9 7-7 7-6 7-8 7.6 7-3 7.0 8.0 7-8 7-7 7-8 7.8 7-7 7-7 7-7 7-8 7-4 7-1 8-3 8.0 Means 7.55 7.73 7.65 7.60 Observers Nos. 2 and 4 had no previous experience in the use of the instrument. WITH ABSORPTION GLASS. Observer i 3 Cat's-eye scale readings. 25.7 25.8 24.0 24.8 23.6 26.0 24.1 25.8 25.4 24.8 24.8 24.9 24.8 25.3 Means 24.63 25.34 Here the greatest variation corresponds to less than 3 degrees in temperature at 1000 C. To control accurately the flame height in the gasoline lamp, a sight was inserted consisting of a horizontal scratch 2 mm. above the window before the flame, and a very fine platinum wire in the same horizontal plane but in a collar behind the flame. With OPTICAL PYROMETER 309 this improvement an observer can set and control the flame height to 0.2 mm. Such provision, however, is not necessary except in the most refined work, for experiment showed that for most purposes changes of over 2 mm. may be made in the flame height with unimportant changes resulting in the temperature estimation. Considering the time effect of burning upon the flame height and intensity due to local heating and change of depth of oil, it was found that the flame ceases creeping up after ten minutes and will then remain at constant height to within 0.5 mm. until the oil is used up, in three hours; and during all this period the brightness of the flame does not change by an amount correspond- ing to more than 5 degrees in temperature. It might be expected that oils of different grades would give widely differing results, but an examination of this possible source of error showed that different samples of gasoline and gasolines mixed with several per cent of a heavy kerosene gave identical results. This is of great importance in the practical use of the instrument, as it shows that a calibration made with a given sample of gasoline remains good for any other gasoline. From the above it is clear that variations in brightness of the comparison flame due to all possible causes need not produce errors in temperature measurement of over 5 C. at 1000 C., that is, within the experimental limits of making the photometric setting. Considering now the sources of error due to focusing and sighting upon the object whose temperature is sought, it is first to be noticed that there is a minimum distance from the object at which the pyrometer can be focused, this distance being somewhat over a meter, depending, of course, upon the focal length of the objective and length of drawtube. There is also a minimum area which can be sighted upon and give an image of sufficient size to completely cover the desired photometric field; this minimum size of object is about 6 mm. on a side when the instrument is at its least distance; for greater distances a larger area must be viewed. 310 HIGH TEMPERATURES The drawtube can easily be set to 2 mm. when focusing, and as the image is over 20 cm. from the objective in all cases, the resulting error in intensity due to focusing is not greater than 2 per cent. This corresponds to i C. in temperature, showing that an error of even 5. mm. in focusing the drawtube will not produce an appreciable error in temperature estimation. Often, in use, the distance of the instrument from the objects studied needs to be changed considerably, and in rapid work it is not always convenient to ref ocus ; a change in this distance of a fourth of its value, i.e., from 120 cm. to 150 cm., will produce an apparent change in intensity of only 9 per cent, or about 5 C. in temperature. That these errors of focusing are so small when interpreted into temperatures, showing that no unusual precautions are needed, is evidently of great convenience in the use of the instrument. The nonmonochromatism of the red glass in the eyepiece produces no considerable error in temperature measurement up to 1600 C., although if this glass is not very nearly monochro- matic the differences in hue in the two adjacent photometric fields from the comparison lamp and other sources are very troublesome, and the strain on the eye in matching them is con- siderable. For the best work at high temperatures a better glass than is usually furnished with the instrument must be used (see page 335)- There remains to be considered the error introduced due to uncertainty in the knowledge of the. coefficient of absorption of the absorbing glasses. If an observation (N f ) is taken with, and then one (N) without, an absorption glass, we have ATA 2 k = ' N ^ so that the accuracy in determining k depends directly upon the precision of setting and reading the cat's-eye opening. Errors of over 5 at 1000 C. can hardly occur from this cause, although the determination of k is the most difficult and uncertain of all the operations in optical pyrometry. OPTICAL PYROMETER 311 Modifications of the Le Chatelier Pyrometer. For use in technical works and other places where there are sure to be strong drafts of air causing unsteadiness of the flame of the oil compari- son lamp, the Le Chatelier pyrometer might be improved by the substitution of an electric incandescent lamp of low voltage (six) placed before a uniformly ground diffusing glass screen, which, illuminated by the incandescent lamp, becomes the con- stant-comparison source. The electric lamp may be mounted in a vertical arm which serves at the same time as a handle, and then the instrument becomes nearly as portable as 'an opera glass. The reliability of such a method of producing a comparison light of invariable intensity will be discussed when describing the Wanner instrument. Other modifications will be discussed under the Fery and Wanner optical pyrometers. The Shore Pyro scope. In this instrument (Figs. 112, 112 A) the principle used is similar to that of the Le Chatelier optical pyrometer, the parts being arranged in a slightly different man- ner. The temperature is read directly off a scale controlled by the diaphragm; the telescope is movable about a horizontal axis, and the lenses are protected by easily removable cover glasses. In taking an observation, the diaphragm is turned until the object sighted upon and the flame, viewed by reflection, are of the same brightness. Fery Absorption Pyrometer. This is similar to Le Chatelier's instrument, except that a pair of absorbing-glass wedges p, p' replaces the iris diaphragm; and the 45 mirror G, with parallel faces, is silvered over a narrow vertical strip, giving a photometric field of form shown at ab, when looking at a hot crucible. The instrument also has a fixed angular aperture, so that no correction has to be made for focusing or for varying distance from furnace. The comparison light L plays the same role as in Le Chatelier's pyrometer, and the range of the instrument may be similarly extended by the use of auxiliary absorbing glasses A, A 1 . Fery has in addition made his instrument movable about a horizontal axis, which is a convenience in sighting. The calibration is equally simple. If x is the thickness of the 3 I2 HIGH TEMPERATURES Fig. 112. Shore Pyroscope. wedges, read off on a scale, when the light from the comparison lamp and furnace is of the same brightness, then the relation between brightness I and thickness of wedge is where k is the coefficient of absorption of the glass of the wedges for the red light used and c is a constant. OPTICAL PYROMETER 313 Kerosene Lamp R = Comparison Reflector Fig. 112 A. Shore Pyroscope, Section. 314 HIGH TEMPERATURES A ' V Fig. 113. Fery Absorption Pyrometer. But by Wien's law III (page 251), assuming it to apply here, or combining these two equations we have B whence Ce kx = Ae T , kx + C ' = Thus it follows that the thickness of the wedge is inversely pro- portional to the absolute temperature, so that the calibration OPTICAL PYROMETER 315 may be effected by finding the thickness of wedge for two tem- peratures only and plotting a straight line and constructing a table giving 7 and T respectively in terms of x. It is questionable if there is any gain in substituting the wedge for the cat's-eye in the desire to extend the range over which the instrument may be used without employing the auxiliary absorb- ing glasses, for thereby the sensibility is somewhat reduced, and, more important still, the wedge instrument cannot be used at such low temperatures as the original Le Chatelier form, nor is there any gain in simplicity of calibration and ease of manipula- tion. The shape of the photometric field, the use of an aperture of constant angle, and making the instrument movable about a horizontal axis, however, are improvements over the Le Chate- lier instrument; and the Fery instrument enjoys the further advantages that it may more conveniently be sighted on small objects, and fewer absorption glasses are needed. Wanner Pyrometer. Wanner, making use of the polarizing principle discarded by Le Chatelier, has brought out a photom- eter pyrometer which is a modification, suited to temperature measurements, of Konig's spectrophotometer. The comparison light is a 6-volt incandescent lamp, illuminat- ing a glass matt surface; monochromatic red light is produced by means of a direct- vision spectroscope and screen cutting out all but a narrow band in the red, and the photometric comparison is made by adjusting to equal brightness both halves of the photometric field by means of a polarizing arrangement. The slit Si is illuminated by light from the comparison source, a small 4- volt electric lamp (Fig. 115), not shown in the Fig. 114, reaching Si after diffuse reflection from a right-angled prism placed before Si. Light from the object whose temperature is sought enters the slit $2. The two beams are rendered parallel by the lens LI, and each dispersed into a continuous spectrum by the direct-vision prism P. Each of these beams is next separated by a Rochon prism R into two beams, polarized in planes at right angles. Considering only the red light, there would now be four images formed by the lens L% and distributed HIGH TEMPERATURES about the slit 5 4 . In order to bring two red images oppositely polarized exactly before this slit, a bi-prism B is interposed whose angle is such as to effect this for two images only, at the same time increasing the number of images to eight. There is now in the field of view before the Nicol analyzer, A, two contigu- ous red fields composed of light oppositely polarized, the light of one coming from Si alone, and of the other from S% alone. All the other images are cut off from the slit S^. If the analyzer is at an angle of 45 with the plane of polarization of each beam, and if the illumina- tion of 5*1 and 5 2 is of the same brightness, the eye will see a single red circular field of uniform brightness. If one slit receives more light than the other, one-half of the field will brighten, and the two may be brought to equality again by turning the analyzer carrying a graduated scale, which may be calibrated in terms of temperature. If the analyzer is turned through an angle to bring the two halves of the field to the same p brightness, the relation between the two inten- . sities from Si and Sz is 1 = tan 2 0. Calibration. Since monochromatic light is used, and the comparison beam and that from the object examined undergo the same optical changes, Wien's law III may form the basis of the calibration. As constructed and generally used, the 45 position of the analyzer when setting on the standard corresponds most conveniently to some intermediate arbitrarily chosen position on the graduated scale (Fig. 115) OPTICAL PYROMETER 317 of the instrument. This reference position or "normal point" is the scale reading to which the instrument must be adjusted, by varying the current through the comparison lamp or its distance from the slit Si, when sighting upon the standard amyl- acetate flame. The positions of the flame and pyrometer are fixed mechanically (see Fig. 115). The flame height must be carefully adjusted and the lamp should burn some ten minutes before standardizing. Fig. 115. Wanner Pyrometer. If 7 is the intensity of light from the standard amyl-acetate lamp, TQ the corresponding equivalent temperature absolute, and $o the reading in degrees on the scale of the instrument for the " normal point," and /, T, and are the intensity, apparent temperature, and scale readings when sighting upon the object whose temperature is sought, we have tan tan 2 (a) assuming the circle to be uniformly graduated and the optical parts in adjustment. Also Wien's law III (page 251) gives us 318 HIGH TEMPERATURES Since the constant c 2 = 14,500 for a black body and X = 0.656 /* as the instrument is usually constructed, a knowledge of the apparent black-body temperature of the standard source, to- gether with the reading of the analyzer scale at the normal point when / = /o, for such an instrument, is all the data required for its calibration, as any temperature may then be calculated by means of equations (a) and (b) in terms of the scale readings. The apparent temperature T of the amyl acetate may be taken as 1673 aDS - or 1400 C. This instrument may also, of course, be empirically calibrated in terms of the readings of a thermo- couple, using a black body to sight upon (see p. 241). The actual computation involved in a calibration is very simple, and is readily done graphically in a manner similar to that suggested for the Le Chatelier optical pyrometer. From equations (a) and (b) above we have log tan = a + ft., ...... ( c ) so that if log tan is plotted in terms of , we have a straight line of which b is the tangent and a the intercept on the log tan < axis. If a and b are known for the type of pyrometer used, a single calibration temperature suffices, otherwise two observa- tions of T and are required to completely solve (c) . It is safer, however, to take several temperatures and draw the line best representing the observations according to (c). A table or a curve of vst ( = 7^ 273) may then be constructed for practi- cal use. It is evidently necessary to be able to always reproduce exactly the standard intensity 7 . Now, the brightness of an electric lamp will vary with the current through it, so it is necessary to check frequently the constancy of illumination of the slit Si against a standard source of light. The amyl-acetate lamp and ground-glass diffusing screen are placed before the slit S 2 , thus reproducing the standard light required. The analyzer is then set at the previously determined normal point and the distance of the electric lamp from Si adjusted or the current through the OPTICAL PYROMETER 319 lamp changed by a rheostat, until the two fields appear of the same brightness. In the latest form of this instrument the details of its mechanical construction have been improved, and it has been made direct-reading by providing a second scale on the instru- ment graduated in temperatures, corresponding, of course, to a definite normal point and for a source approximating a black body. The amyl-acetate standardizing lamp may be eliminated wholly or in part in the use and calibration of the Wanner pyrom- eter. If the electric comparison lamp be fixed in position, the reading of the instrument sighted on a black body at a single temperature, as the gold point (adjusting the scale to a conven- iently located "normal point"), or better, at a series of known temperatures, may be taken, for a definite current through the comparison lamp. If this same current is always maintained in the use of the instrument, this calibration will hold as long as the lamp does not change nor the optical parts of instrument become deranged. This last method of use is preferable in exact work where calibrating apparatus is available. The normal point, however, may still be that given by the amyl acetate if so desired. Or, the amyl-acetate standard with its corresponding normal point may be retained, but used only occasionally for checking and adjusting the constancy of the pyrometer, whose uniformity of indications is maintained in the meantime by keeping the current constant in the comparison lamp when tak- ing measurements, and keeping the comparison lamp in a fixed position. According to Nernst and Wartenberg, it may also be necessary to correct the circle readings by a constant fraction; thus they found that for a certain Wanner instrument the ratio tan 2 - was constant, where m is nearly unity. 320 HIGH TEMPERATURES Sources of Error. A study of a Wanner instrument by Waid- ner and Burgess has led them to the following conclusions. The sensibility of this pyrometer varies with change in the angle, and is so adjusted as to be the greatest between 1000 and 1500 C.,, and is about as follows: o.i scale div. o i C. at 1000 C. o.i scale div. =c= 2 C. at 1500 C. o.i scale div. ^ 7 C. at 1800 C. The reproducibility of the brightness of the amyl-acetate flame as viewed through the ground-glass diffusing screen is a measure of the ability of the instrument to repeat its indications. It is very important that this diffusing screen be always placed in exactly the same position relative to the flame and slit S^ and further, that it be free from dust and finger marks. These re- quirements can only be satisfactorily met by protecting this screen by a cover glass and providing an adjustment for setting it exactly in place between the flame and slit. The constancy of the amyl-acetate flame as used with this pyrometer under ordinary conditions of burning is illustrated by the following set of observations, during which the current through the electric comparison lamp was kept rigorously con- stant by means of a milliammeter and rheostat: Reading of instrument. Deviations. 39-9 0.28 39-9 -0.28 40.1 -0.48 39-9 0.28 +0.52 39-2 +0.42 39-8 -O.l8 39-0 +0.62 39.6 0.38 This shows that the flame can be relied upon to give an intensity of illumination whose constancy expressed in terms of tempera- ture is 0.5 per cent. Variations in height of the flame, if they do not exceed 2 to 3 mm., together with fluctuations in atmospheric conditions, will not produce errors in temperature estimation ex- ceeding i per cent. OPTICAL PYROMETER 321 The uncertainty of setting the Nicol, due to lack of sensitive- ness of the eye to exactly match the two halves of the photometric field, is also about i per cent, or slightly better with practice. The adjustment of the electric lamp to standard intensity at the point on the scale chosen as normal point can be made, when proper care is taken regarding the diffusing screen, to i per cent expressed in temperature change. This source of error does not affect relative results in any one series for one setting to the normal point. The most serious source of error, except when special pre- cautions are taken, is the variation in brightness of the electric comparison lamp due to variation in the current furnished by the three-cell storage battery. With the lo-ampere-hour battery furnished with the Wanner instrument, after making circuit the electromotive force drops by about 2 per cent in two minutes and then falls off slowly, but nearly recovers the original voltage after remaining on open circuit even for a very short time. When the battery is in good condition the variation in three hours at normal discharge (0.75 ampere) is about 0.08 volt, and somewhat less for the current (0.55 ampere) taken by the lamp; with the battery in poor con- dition these changes are much accentuated. The following table illustrates the effect of slight variations in current through the lamp on apparent temperature of the amyl-acetate flame, for the small battery of 10 ampere hours furnished with the instrument. The apparent change in tem- perature is calculated from the current change: SMALL BATTERY. Time. Minutes. Wanner scale. Current in amperes. Per cent change in current. Apparent change in temperature. I r 31 .2 0.5645 2O 31-8 0.5640 O.I iC. 27 32.7 0-5550 i-7 10 37 34-6 0.5400 4-3 2 5 36 Disconnected battery two minutes. 40 32.5 0.5610 0.6 3 42 3 J -7 0.5570 !-5 7 45 32.5 0.5560 2-5 15 47 33-i 0.5505 4-i 24 322 HIGH TEMPERATURES A battery of 75 ampere hours gave similar results. The above results give abundant evidence of the need of maintaining the current through the lamp quite constant in work of precision. A series of experiments has shown that in the range 1000 to 1500 C. one division on the Wanner scale corresponds to about 0.009 ampere, or i C. apparent change in temperature is produced by a fluctuation of 0.0012 ampere through the lamp; hence to obtain a precision of 5 the current must be kept constant to o.oi of its value. The above table shows that this is by no means effected by using the battery without regulating the current, for even with the battery in the best condition the current increases by 2 per cent in the first eight or nine minutes of discharge and then falls off i per cent in the next twenty minutes. The temperature coefficient of the battery would produce only insignificant changes. The table shows further that breaking the circuit and then making it again may cause an apparent temperature change of over 20 C. For work of precision, therefore, it is essential to keep the current constant by means of a milliammeter and rheostat, otherwise un- certainties of over 25 C. will occur in the temperature measure- ments. These will increase with the battery in poor condition. Range and Limitations. The above description of the Wan- ner pyrometer has shown the great loss of light due to the optical system employed. This prevents measuring temperatures below about 900 C. (1650 F.) with this instrument. There is no method of sighting this pyrometer exactly upon the spot desired, except by trial, as no image of the object examined is formed in the eyepiece, but this inconvenience is in part compensated by not having to focus with varying distance from the object. There is another limitation which may in certain cases become a serious source of error: light from incandescent surfaces is, in general, polarized, and, as the Wanner instrument is a polarizing pyrometer, care must be taken to eliminate this source of error when it exists. If an incandescent object is viewed normally, the amount of polarized light is very small, but, as the angle of incidence in- OPTICAL PYROMETER 323 creases, the proportion of light polarized becomes greater and greater. Besides varying with the angle of incidence, the amount of polarized light emitted varies widely with different substances, being greatest for polished platinum and very much less for iron, glass, etc. In some measurements made with the Wanner pyrometer on the temperature of an incandescent platinum strip in the neighborhood of 1350 C., Waidner and Burgess have found a maximum difference in the readings of op C. for positions of the instrument at right angles to one another in azimuth and for an angle of incidence of 70 with the normal to the surface. This introduces, under these conditions, the possibility of an error of 45 C. in the temperature measurement. This source of error can be eliminated by taking the mean of four readings for azimuths 90 apart. The magnitude of the error arising from this cause is entirely negligible for all practical purposes for many substances, such as iron, porcelain, etc. A considerable area is needed to sight upon with this pyrometer, which is a dis- advantage when small objects are viewed from a distance. Due to the relatively large surface required in sighting the Wanner pyrometer, there is a tendency to bring the instrument too close to the furnace or object viewed, and this practice carried to excess may readily damage the instrument, deranging its optical parts and altering the calibration by very considerable amounts. Warning of overheating is sometimes given by the change in color of the field. Placing a water jacket between the furnace and instrument or otherwise screening the latter will evidently obviate this difficulty. Where an attempt is made to sight on very small or distant areas, such as wires or narrow strips which fill only a small part of the photometric field, there may be produced diffraction effects, as noticed by Hartmann. A review of the sources of error and limitations of the Wanner pyrometer shows that they may exert a relatively great effect on the temperature measurements, and it was, therefore, thought worth while to emphasize them; but, on the other hand, they may all be practically eliminated with reasonable care, and the instru- 324 HIGH TEMPERATURES ment then becomes one of great precision and convenience, for those measurements for which it is adapted. We shall see later how its range may be extended to the highest temperatures. Instrument for Low Temperatures. In order to render his pyrometer available for temperatures below 900, Wanner has brought out a modification suitable for use from 625 to 1000, with two ranges, 625 to 800 and 800 to 1000 C., which gives a very open scale and renders the instrument available for a great many industrial operations that were hitherto inaccessible to it. In this low-temperature form, shown in Fig. 116, the light from Fig. 116. Wanner Outfit for Low Temperatures. the furnace does not pass through the polarizing system, and the direct-vision prism is replaced by a red glass in the eyepiece, by which elimination light of much feebler intensity than with the high-range apparatus can be observed. The apparatus is very compact and easy to manipulate. It requires as accessories a 4- volt storage battery, milliammeter, and amyl-acetate standard. Holborn-Kurlbaum and Morse Pyrometers. If a sufficient current is sent through the filament of an electric lamp, the fila- ment glows red at first, and as the current is increased the fila- ment, getting hotter and hotter, becomes orange, yellow, and white, just as any progressively heated body. If now this filament is interposed between the eye and an incandescent OPTICAL PYROMETER 325 object, the current through the lamp may be adjusted until a portion of the filament is of the same color and brightness as the object. When this occurs this part of the filament becomes invisible against the bright background, and the current then becomes a measure of the temperature as given either by a ther- mocouple or in terms of the intensity of illumination. This principle appears to have been first used by Morse and inde- pendently developed by Holborn and Kurlbaum. An absolute match of both color and brightness cannot be made unless mono- chromatic light is used or unless the lamp filament and viewed object radiate similarly. 45'Mirror Absorbing Screen Secfion on A-C Fig. 117. Holborn-Kurlbaum Pyrometer. Holborn-Kurlbaum Form. A small 4-volt electric incandes- cent lamp L with a horseshoe filament is mounted in the focal plane of the objective and of the eyepiece of a telescope provided with suitable stops D, D, Z), and a focusing screw S for the objective. The lamp circuit is completed through a two-cell storage battery B, a rheostat, and a milliammeter. The determination of a temperature consists in focusing the instrument upon the incandescent object, thus bringing its image into the plane AC, and adjusting the current by means of the rheostat until the tip of the lamp filament disappears against the bright background, when a previous calibration of current, in terms of temperature for the particular lamp used, gives the temperature by reading the milliammeter. 326 HIGH TEMPERATURES As the temperature of the filament increases, the effect of irra- diation or too great brightness becomes blinding, and the photo- metric comparison is then rendered possible at these temperatures by the introduction of one or more monochromatic red glasses before the eyepiece, giving as well all the advantages of photom- etry of a single color. Below 800 C. the measurements are more easily made without any red glass, as the filament itself is then red, and the lowest temperatures are, of course, reached with the least interposition possible of absorbing media. The lower limit of the instrument is very nearly 600 C. Two red glasses are required for temperatures above 1200 C., and for very high temperatures, above 1500 or 1600 C., it is necessary, in order to avoid overheating the lamp filament by the current, to put absorbing glasses or a double-prism mirror (Fig. 121) before the objective; and they also, of course, require calibration. At very high temperatures, unless a strictly monochromatic glass is used, the pyrometry becomes difficult, the filament never disappearing completely. The eye is particularly sensitive in recognizing equality of brightness of two surfaces, one in front of the other, and this pyrometer, therefore, provides a very delicate means of judging temperatures, since the light intensity, as has been shown (page 238), varies so much faster than does the temperature. The precision attainable with this pyrometer is illustrated by the following series of observations which are indicative of the ordinary performance of the instrument: Temperature from H.-K. pyrometer. Temperature from thermocouple. Temperature from H.-K. pyrometer. Temperature from thermocouple. 1347 i347C. 632 634 C. I3SI 1347 634 633 1343 1343 633 633 1333 1332 633 632 1342 1342 Different observers do not differ by any appreciable amount in their readings, and at low temperatures the same values are obtained whether a red glass is used or not. OPTICAL PYROMETER 327 For the calibration of the instrument, it is necessary to find empirically the relation between the current through the lamp and the temperatures for a number of temperatures, and then interpolate either analytically, or more conveniently, graphically. The calibration will evidently be an independent one for each lamp used. The relation between current and temperature is sufficiently well expressed by a quadratic formula of the form C = a + bt + ct 2 . That this formula gives satisfactory results is shown by obser- vations of Holborn and Kurlbaum for a lamp satisfying the equation C io 3 = 170.0 + 0.1600 1 + 0,0001333 t 2 , when sighted on a black body (page 239), the temperature of which is given by a thermoelectric pyrometer calibrated at known melting points. C amp. io- 3 . t obs. t calc. A*. 340 686 679 -7C. 375 778 778 o 402 844 850 +6 477 1026 1032 -f-6 552 1196 1196 o 631 1354 1354 o 712 1504 1504 o We may also cite the behavior of one of the several standard pyrometer lamps of the Bureau of Standards. This lamp satis- fies the equation C = o.i 68 1 + 0.03 1482 t + o.o 6 i7oo/ 2 . C in amps. fobs. /calc. At o . 4486 920 921 -1 .5305 1087 . 5 1087.5 O 3357 650 649 + 1 .6023 1221 1221 .3525 692 692.5 -0-5 6393 1285 1285 o 5309 1089 1088.5 +0-5 No appreciable change in the readings of this lamp could be detected over a five-year period, the lamp being used very fre- quently during that time to temperatures as high as 1500 C. 328 HIGH TEMPERATURES Pirani and Meyer have shown that, for carbon and metal filament lamps, log C = a + b log T where C = current and T = absolute temperature. This per- mits of a calibration with two temperatures only. Mendenhall suggests that this pyrometer and the same is true of all the optical instruments using monochromatic light may be calibrated for all temperatures in terms of a single known temperature, such as the palladium melting point, by means of a series of sectored disks each of a different aperture, giving, by the application of Wien's law (see page 250), a corresponding series of effective temperatures. The sectors, of some 15 cm. di- ameter, may be rotated by means of a shaft attached to a small motor fixed near the middle of the outside of the pyrometer tube. Mendenhall has also made a direct-vision spectroscopic eyepiece for this instrument, and works with a field of about 25 A.U. width, giving X to about one-fifth per cent in the middle of the visible spectrum. Holborn and Kurlbaum as well as Waidner and Burgess have made a thorough study of the effects of aging. Lamps which have not been aged or burned for some time at a temperature considerably above that at which they will ordi- narily be used, undergo marked changes and are unreliable, but, if properly aged, they reach a steady condition, as indicated by the following table of results obtained by Holborn and Kurlbaum on these lamps. The current is given in each case for a tempera- ture of uooC. AGING OF LAMPS. Current. Lamp number I 2 3 After 20 hours burning at 1900 C . . . . 0.608 0.592 o 589 After 5 hours burning at 1900 C 613 .592 .592 After 5 hours burning at 1900 C .621 .597 .597 After 5 hours burning at 1900 C 622 . 599 .600 After 20 hours burning at 1500 C 622 . 599 .601 If a lamp is not aged its indications may change by as much as 25 C. with time, but after twenty hours' heating at 1800 it will undergo no appreciable further changes over a period of OPTICAL PYROMETER 329 time corresponding to many months if used in the shop, if not heated above 1500. This state of permanence is sufficient to satisfy the most rigid requirements of practice. By the substitution of tungsten for carbon filaments even greater permanence may be had, but the selective radiation of the metallic filament may then be a source of error or inconven- ience in certain cases. Morse Thermogage. In its original form, instead of a simple horseshoe filament, Morse used a large spiral filament in the lamp of his pyrometer, so that in sighting upon an incandescent body it was necessary to choose some particular spot of the spiral and try to make that spot disappear. This is fatiguing, as the spiral covers a large area and is of just sufficiently varying inten- sity to cause the eye to wander. This effect was aggravated by the fact that this instrument was not a telescope, possessing no eyepiece or objective, so that the eye had to accommodate itself back and forth between the filament and the object studied. Instead of the 4-volt battery for the Holborn-Kurlbaum lamps, the spiral lamp took a battery of 40 or 50 volts, requir- ing a costly installation unless the fluctuations of the ordinary no- volt lighting circuit were not too troublesome to use it with a suitable rheostat or shunt. The Morse instrument was designed for use in hardening steel, and, throughout the limited temperature range required in this process, in spite of the crudities of construction above noted, this pyrometer could be read to about 3 C. within this range. Above iiooC., however, it is very difficult, and it soon becomes impossible to make a satisfactory setting. Tests of these spiral filament lamps show that when aged at 1200 C. they will remain constant for several hundreds of hours within the range over which they are intended to be used. It is interesting in this connection to note the behavior of ordinary carbon incandescent lamps as to permanence. (See Fig. 118.) Later forms of the Morse thermogage are provided with lower voltage lamps with a single loop, red glass at the eyepiece, and 330 HIGH TEMPERATURES made into a telescope, following, in part, suggestions given to Morse by Waidner and Burgess. rfO 100 200 HOURS 400 425 500 Fig. 118. Behavior of Carbon Lamp. Henning's Spectral Pyrometer. In order to eliminate the uncertainties and corrections for the lack of monochromatism of colored glasses used with the Holborn-Kurlbaum instrument, and to permit temperature measurements with any colored light, Henning has devised a spectral pyrometer suitable for Fig. 119. Henning's Spectral Pyrometer. exact work in the laboratory from 1000 C. It is essentially a combination of the Holborn-Kurlbaum instrument with a spec- trometer as shown in Fig. 119. The collimator KLz, telescope FLi, carrying an observing slit D or an ocular, and Abbe prism P which can be set to give any wave length by means of the mi- crometer M NA, together with the slit E adjustable in width by OPTICAL PYROMETER 331 the screw U, constitute the spectrometer. An image of the in- candescent body is superposed on the lamp G by the lens Z, 4 , and both are seen in colored light with the observer's eye before D. The screen B carries a series of suitable stops. The micrometer scale A is calibrated in wave lengths by means of light from standard sources, as helium and mercury vacuum tubes. The instrument may also be arranged for use as a spectrophotometer. Henning has used his spectral pyrometer in a study of metal- filament lamps and for the determination of absorption and reflecting coefficients of metals. He has shown that, for a series of metals, the equation = const., in which 5 and So o oo are the absolute black-body temperatures for wave lengths X and X , holds over a wide range of temperatures; and that the absorption coefficients remain practically constant with change of temperature. Calibration of Optical Pyrometers. We have already called attention to the fact that the most accurate method of calibrating an optical pyrometer to about 1600 C. is to take its readings when sighted into an experimental black body (page 239) whose temperature is best given by two or more thermocouples which have in turn been calibrated by determining their E.M.F.'s at the freezing points of three or more pure metals. These calibra- tions are, in general, best left to a properly equipped standard- izing laboratory. However, it is often desirable to be able to calibrate, at least approximately, one's own optical pyrometer, even if not in the possession of a complete standardizing equip- ment. A fair substitute for the black body is a resistance-tube fur- nace of the Heraeus type with a diaphragm, say a piece of graphite, inserted at its center, or a little back of this, and on which the optical pyrometer is sighted. The temperature of this diaphragm may be obtained with a calibrated thermocouple or optical pyrometer. Sighted into such a furnace, whose total length is some twenty or thirty times its diameter, an optical pyrometer will read some 5 to 15 C. too low. 332 HIGH TEMPERATURES The following method may also be used, and this requires no auxiliary pyrometer, but does require from one to three or more deep crucibles of substances of known melting points, preferably the pure metals, such as Al or Sb, Cu, Ni, or Fe. The optical pyrometer is sighted on the bottom of a porcelain tube, preferably blackened inside, and which is thrust into the melted metal, and the reading of the pyrometer taken at the freezing point of the metal. Where several optical pyrometers are in use in the same estab- lishment, it is well to have at least one of them calibrated care- fully and kept as a standard. The others are readily calibrated by comparing their readings with that of the standard when sighted on any convenient incandescent source whatever, pro- vided the pyrometers all use the same colored light; otherwise it is safer to use a furnace as source, although graphite or iron (oxide) will answer in most cases. The criterium of a satisfactory comparison source for pyrom- eters using different colors is to view the source, when this is possible, with different colored glasses applied in succession to one pyrometer. If the same reading is obtained for all red, yellow, and green, for example the source is satis- factory. The Wide-filament Comparison Lamp. A very convenient and rapid method of standardizing one optical instrument in terms of another is shown in Fig. 120, which was devised by Waidner and Burgess for the determination of incandescent lamp filament temperatures and the melting points of very refractory metals. Fig. 120 illustrates the use of a carbon strip C mounted in vacuo for the former purpose. The standard pyrom- eter L and the lamp F whose filament temperature is sought are both brought to the same brightness as C, and the currents in L and F give a measure of their temperatures, which are assumed equal if the color of the glass G is the same as that used before L and if the filaments F and L are of the same material. The lenses E and make the readings of F more convenient and equalize the two optical systems. Evidently any type of optical OPTICAL PYROMETER 333 pyrometer may be substituted for the lamp F and calibrated in a similar manner. These carbon-strip comparison lamps may be used intermit- tently to temperatures as high as 1800 C. or even 2000 C. If used only at comparatively low temperatures, they may them- selves be calibrated in terms of current vs. temperature and then serve as a secondary standard, replacing the black body. Such lamps of this type as are at present available change pretty rapidly with even short burning, so that it is better to keep a filament lamp or other optical pyrometer as the standard and Fig. 120. Calibrating Method of Waidner and Burgess. use the wide strips merely as comparison sources. For extend- ing such comparisons to higher temperatures, it would be de- sirable to replace the carbon with tungsten strips, when probably 2500 C. or more could be realized. Other comparison sources are available, however, for these very high temperatures, such as the Arsem vacuum furnace (Fig. 176) with which temperatures of nearly 3000 C. may be attained, and, moreover, black-body conditions are completely realized. Use of Wedge-shaped Cavities. We have already seen that in the- calibration of his optical pyrometer Le Chatelier took advantage of crevices in heated materials surrounding a thermo- couple to obtain approximately black-body conditions. Fery has called attention to the necessity of the measuring instrument also being black in the Kirchhoff sense, at least when absolute 334 HIGH TEMPERATURES measurements are made, and he developed the use of conical receivers. Mendenhall, studying the relation between true and apparent temperatures of metals by means of the optical pyrometer, shows that if a thin metal strip is bent into a wedge of small angle, the radiation from within the wedge, heated electrically, as is a lamp filament, is very nearly that of a black body; so that simul- taneous readings with a calibrated pyrometer on the outside and inside of such a wedge give a measure of the selective properties of its substance. The wedge may also replace the black body for the comparison of one optical pyrometer with another. Assuming specular reflection and the wedge angle Z,, the number of reflections perpendicular to the edge of the wedge is n = ; J^j if the reflecting power of its material is r, that of the wedge is r n . For many metals r is of the order of 0.7 for red light, when for a lo-degrees wedge r n = 0.0016 and e = ae = 0.998 e, corresponding to a temperature difference from a black body of the same bright- ness of only 0.5 C. at 1600. For matt surfaces the departure from blackness is greater. The difference in temperature be- tween the inner and outer surfaces of the wedge is less than i C. for metals of 0.04 mm. or less in thickness. By burning out such wedges of platinum, Mendenhall and Faryther obtained a value for the platinum melting point only 8 degrees lower than the figure of Waidner and Burgess (1753 C.). Monochromatic Glasses. In order to use Wien's law with exactness and convenience, and especially when extrapolation on the temperature scale is resorted to, it is highly desirable that there be no change in the color of the light used in an optical pyrometer. With those pyrometers in which the monochro- matic light is produced by means of colored glasses, there may be an error introduced due to the lack of homogeneity of the light transmitted and to the consequent shift with temperature in the position of maximum intensity of the light. For such inhomoge- neous glasses this is equivalent to introducing a continuous change of wave length with temperature in Wien's law (page 251). OPTICAL PYROMETER 335 The behavior of certain Jena glasses, which are among the best in the smallness of this effect, as found by Waidner and Burgess, is shown in the following table: MONOCHROMATISM OF COLORED GLASSES (JENA). Glass. Thickness in mm. Temper- ature of source (C). Xinax. Limits of transmission band. Red, No. 274.C } 04 ( IOOO < I2?O 0.645/x 6^0 0.698^1-0. 6 1 0/A .731 - .602 Red, No. 274.C 6 o 1 ? f 1450 I4CO .656 661 .772 - .598 . 7C3 - .608 Green, N 3. 43 1 m . 6 18 i H50 547 .602 ~ .532 Blue, No. 3086 4-32 ( 145 ( 1320 ( 1470 .546 .462 .462 .631 ~ .468 .500 - .421 .511 - .408 The position of the optical center of gravity (X max . in the table) is seen to remain stationary for the green and blue, but to shift slightly to longer wave lengths for the red glass, with increase in temperature. An error of 0.005^1 in the estimation of the equivalent wave length for a colored glass corresponds to an error in temperature estimation of about 5 C. at 1750 C. For some of the newer monochromatic Jena glasses the follow- ing data on the transmission coefficients have been issued by Schott and Genossen: TRANSMISSION COEFFICIENTS (>) OF JENA GLASSES FOR 1 MM. THICKNESS. Glass. Fraction transmitted for wave lengths (in M). Type. Name. X = 0.644 0.578 0.546 0.509 0.480 0.436 F 4SI2 Red filter O O4. o o^ F 2745. . . . Copper-ruby O.72 O. 3Q 0.47 0.47 0.45 0.43 F43I3-..- F435I-... F4937-.-. F 4930. . . . Yellow glass, dark . . . Yellow glass, medium Yellow glass, light. . . Green filter 0.98 0.98 I .OO o 17 0.97 0.97 I. 00 o so 0-93 0.96 1. 00 o 64 0.83 0-93 0.99 o 62 0.09 0.44 0.74 o 44 o.iS 0.40 F3875 Blue filter o 18 o so 0.73 F 3 8i5.... Neutral black o 35* 0.35* o 37* 0.35* * 0.34 0.30* For a thickness of o.i mm. 336 HIGH TEMPERATURES The fractional transmission D x for any other glass thickness x x is given by the expression D x = D x , where D is the transmission for i mm. as given in the table. Extension of Scale. All of the optical pyrometers based on the use of a single wave length, such as the Le Chatelier, Wanner, and Morse, may have their scales indefinitely extended by the use of neutral absorbing glasses (such as Jena Rauch Glas) , reflecting mir- rors, or prisms of black glass (see Fig. 121), or sectored disks, placed between the furnace, or other source whose temperature is to be meas- ured, and the pyrometer. The same principle for the com- puting of temperatures with the screen in place applies for all of these screens and for any of these pyrometers. It is only necessary Fig. 121. Absorption Mirrors. to find the absorption coefficient of the screen for the colored light used with the pyrometer. This absorption coefficient may be calculated by making use of Wien's law (page 251) and from observations at one or more temperatures. Thus, if K is the absorption factor, that is, the reciprocal of the absorption co- efficient, TI and Tz the apparent temperatures in degrees abso- lute given by the pyrometer, sighting on a black body first without and then with the absorbing screen, then Wien's law III gives ^ , /i c 2 l( logio A = log : when ^2 = 14,500 for a black body, and X is the wave length in M ( = o.ooi mm.) of the light used by the pyrometer. Applied to the high-range Wanner and Henning spectral pyrometers, the above formula applies exactly to the highest attainable tempera- tures if the absorbing screen has a constant coefficient for all OPTICAL PYROMETER 337 brightnesses; but for those pyrometers using colored glasses, which are never strictly monochromatic, there will be an error entering into the extrapolations, which can, however, for the most part, be eliminated by the calibration in wave length vs. temperature of the colored glasses used, as shown in the pre- ceding paragraph. That these corrections can be made satisfactorily is shown by the following from the data of Waidner and Burgess on the determination of the melting point of platinum by means of a Holborn-Kurlbaum pyrometer using red, green, and blue glasses and provided with different kinds of absorbing screens. The metals were melted in an iridium-tube furnace approximating very closely a black body. The observations of Nernst and Wartenberg with a Wanner pyrometer using yellow light ' are also included, for comparison, their results being reduced to the same optical basis, i.e., for c Rauch glass 35-4 35-4 147 o-547 0.462 0.5896 6 4 4 1748 2 1749 3 1750 5 For most materials heretofore used as absorbing screens, either of the mirror or transmitting glass type, there is a rapid variation in absorbing factor with wave length of the incident light (see page 335 and above table). Schott and Genossen of Jena now furnish a " neutral black " glass (F 3815) of an absorb- ing factor which remains very constant throughout the visible 338 HIGH TEMPERATURES spectrum. The fractional transmission for this glass is given in the table on page 333. The use of a sector disk is preferable for exact work in the laboratory where the intensity of the source observed has to be cut down, for this form of screen has a constant absorption factor which may be determined geometrically with great exact- ness. The absorbing glasses are usually more convenient to use than the reflecting mirrors and are equally as good, or better. Some Scientific Applications. Our knowledge of phenomena occurring at very high temperatures has been increased greatly in the past few years, largely due to the availability of convenient and precise optical pyrometers using monochromatic light. We shall pass briefly in review some of the uses to which this type of instrument has been put in the laboratory as illustrations of what may be accomplished in high-temperature measurements by optical means. Temperature of Flames. Any substance inserted in a flame will take up a lower temperature than that of the flame itself, due to conduction, radiation, and diminished speed of the gas stream around the body. E. L. Nichols, by using thermocouples of progressively finer wires, sought to determine true flame temperatures by extrapolating for a wire of zero diameter. The uncertainty of this method is considerable, although it gives consistent results, which are probably low. The radiation methods have been employed by several experi- menters. The temperature as given by an optical pyrometer will depend on the thickness and density of the flame as well as upon its reflecting and absorbing powers. The reflecting power of a flame is small and probably varies with the kind of flame; the results as yet obtained are quite discordant on this point. Kurlbaum interposed a flame between a black body and the eye and assumed that the two were of the same temperature when the flame disappeared against its background. This method gave results lower than those obtained by Lummer and Pringsheim (page 252). Kurlbaum and Stewart both claim that OPTICAL PYROMETER 339 the carbon in the flame departs more widely from a black body than platinum, and the latter gets 2282 for the value of A in Wien's displacement equation \ m T= A, assuming Nichols's value 1900 C. for the acetylene temperature. Fery has shown, however, that the brightness of the sodium line, measured with a spectrophotometer, is not increased by passing obliquely a beam from an electric light across the flame studied, seeming to indicate that the diffusing power is nil for the light coming from carbon. This would imply a value of A of the order of 2800, or of 2400 C. for the acetylene flame, assuming X w = 1.05. Fery's method of measuring flame temperatures is to produce the reversal of a metallic line by means of light emitted by a solid body brought to the proper temperature. The image of the filament of an incandescent lamp is thrown by a large-aperture lens onto the narrow slit of a spectroscope. The rays from the filament pass through the flame to be studied, which contains sodium or other metallic vapor. When the filament is raised in temperature the D line, say, is ultimately reversed, and at the moment of disappearance the filament and flame are assumed to have the same temperature, which may be measured by sighting an optical pyrometer on the filament. Some of Fery's results are as follows: ( Open 1870 C. Bunsen j Half-open 1810 ( Shut 1710 Acetylene 2550 Oxyhydrogen with illuminating gas and oxygen 2200 Oxyhydrogen with H2 + O 2420 For this determination Fery used his absorption pyrometer. The results obtained may be slightly high, but hardly by more than 100 C., as a fine wire of platinum may be melted in an open Bunsen. There have been other estimations of apparent temperatures of flames by various optical methods based on the radiation laws, some of which have given values greatly below the true tempera- tures, as measured by the ability of these flames to melt refrac- tory materials of known melting point. 34O HIGH TEMPERATURES Making use of Wien's displacement law in form X max T = 2940, Ladenburg found 1405 for the Hefner and 1842 for the acety- lene flame. Becker, by a spectrophotometric method, obtained 1395 for the Hefner. Kurlbaum and Schulze, by a method similar to Fery's, found apparent variations in Bunsen flame temperatures when colored with different salts; but E. Bauer, using the same method, showed that by using a definite part of the flame no such differ- ences exist from one salt to another nor from one color to another. For the oxyhydrogen flame Bauer finds 2240 by applying Planck's law, and 2200 to 2300 by the reversal of the D line, using an electric arc as source of light in Fery's method. Bauer found from 1660 to 1850 for various portions of the Bunsen flame, using several optical methods. All of the above methods assume that flames are nonlumi- nescent, otherwise the results obtained are too high. Absurd results will also be obtained if the flames are colorless, i.e., con- tain no finely divided particles heated by the flame, as in an open Bunsen. Temperature of Glow-lamp Filaments. Since the observations of Le Chatelier with his optical pyrometer, and of Lummer and Pringsheim making use of the Wien relation \ m T = const., there have been numerous determinations of lamp temperatures by means of optical pyrometers. The first satisfactory obser- vations for a series of lamps were made by Waidner and Burgess in 1906, using their graphite-strip method of comparison (page 330), and the Holborn-Kurlbaum instrument, furnished with red, green, and blue glasses in succession before the eyepiece to enable estimations of true temperature to be made from the apparent temperatures, which last, of course, depend upon the selective radiation of the filament surfaces. They found that for platinum filaments inclosed in an evacuated glass bulb, add- ing the difference in temperature between the blue and red read- ings to the apparent temperature with blue light when sighted on the carbon strip, there was given very nearly true tempera- turesfor example, 1760 C. for the platinum melting point. OPTICAL PYROMETER 341 Assuming this empirical relation to hold generally, they found the following: NORMAL BURNING TEMPERATURES OF GLOW LAMPS. Type of lamp. Watts per candle power. Volts. Observed black-body temperatures (red). Maximum true temper- ature. Minimum true tem- perature. Carbon 4-0 50 1710 C. 1800 C. i'/55C. Carbon- 3-5 118 1760 1850 1805 Carbon 3-1 118 1860 1950 iQ5 Tantalum 2 O no 1865 2OOO IQ5C Tungsten I .O TOO 2135 2300 22I 5 In some of the other estimations no attempt has been made to correct for the lack of blackness of the filaments, and the results appear to be generally too low. We may cite the follow- ing determinations: NORMAL LAMP TEMPERATURES BY VARIOUS OBSERVERS. Observers. Grau Coblentz . . Carbon. 1660 (-1785 I 1570 Tantalum. Tungsten. 1850 Fery 1780 Pirani Joly ? 1650 to I JV^Ajr ....... -\ 1720 1 IQIO 1670 2000 1740 2O6O 1810 1875 2080 1810 Method and remarks. Iridium strip and Wanner py- rometer. \ m T=Cand graphite "black." (X m r=C and platinum "black." Temperature obs. of Waidner ( and Burgess with red light. f Combination of Wien and Ste- j fan laws; assumes W behaves i like Pt. Used absorption py- (^ rometer. Resistance and optical measure- ments. (Total photometric (Nernst); other methods gave lower val- f ues. These figures are not strictly comparable, as the ratings are not exactly the same; roughly, they are W = 1.25 , Ta = 1.5 c.p. c.p. A W and carbon = 3.5 c.p. 342 HIGH TEMPERATURES The use of the equation X m T = C (Coblentz) is questionable, as the form of the energy curves of lamp filaments is not that of the black body. The normal burning temperature of the Nernst filament has been measured several times, ranging from the absurdly low result of Hartmann of 1535 obtained with a thermocouple, to the value 2360 of Ingersoll by a luminous-efficiency method. Mendenhall and Ingersoll found that rhodium would melt on a Nernst filament below its normal burning, and that iridium would not, which places this temperature at about 2iooC.; an application of Wien's law gave them 2125 C. Temperatures within Furnaces. The optical pyrometer, espe- cially in its forms due to Wanner and to Holborn and Kurlbaum, has been of the greatest use in studying very high tempera- ture phenomena, including the formation, modification, and dis- sociation of many chemical products. Besides the numerous melting-point determinations described elsewhere, we may men- tion as illustrations the work of Nernst and his associates at Berlin on gaseous dissociation to temperatures above 2000 C., carried out in his type of iridium furnace; of Tucker and others at Columbia University on carborundum and other furnace products; of Thompson at the Mass. Inst. of Technology on a series of chemical reactions ; of Greenwood and of Prim at Man- chester, using a carbon vacuum and pressure furnace, on boil- ing points of the metals and on the temperature of formation of many chemical substances. In all of the above investigations the Wanner pyrometer was used, but where the furnace opening is small, as is usually the case, there is advantage in using an instrument requiring only a few millimeters area to sight on, as the Holborn-Kurlbaum type. This has been used at the Reich- sanstalt in comparing the optical and gas scales, and at the Bureau of Standards in most of the high-temperature work there, as well as at the Geophysical Laboratory. Using an Arsem furnace (Fig. 176), Dr. Kanolt with this pyrometer has been able to measure melting and freezing points of salts, alloys, and minerals to temperatures above 2100 C. by taking the OPTICAL PYROMETER 343 heating and cooling curves and making use of the latent heat of transformation. A few tenths of a gram of material are suffi- cient to give a very sharp point (see Fig. 122). 2100 2000 1900 1800 iroo 1600 1500 1400- 012 Minutes Fig. 122. Melting Points with Optical Pyrometer. Melting points of microscopic samples may also be obtained readily with the Holborn-Kurlbaum pyrometer by making use of the known departure from blackness, or the emissivity, of some substance such as platinum, on a strip of which, or other 344 HIGH TEMPERATURES suitable material such as indium, carbon, or tungsten, may be placed the substance whose melting point is sought. In Fig. 123 is shown the apparatus of Burgess used for the determination of the melting of points of the iron group (Chap. XI) in hydrogen, using samples of the order of o.ooi mg. melted on a platinum strip heated by a delicately adjustable electric current. The container is of brass blackened inside, and simul- taneous observations are taken through a mica window of the Fig. 123. Apparatus of Burgess for Microscopic Samples. melting of the sample with a microscope and of the temperature of the strip with the pyrometer. Conditions of Use. The optical pyrometer using mono- chromatic light, by reason of the uncertainty of emissive powers and of the relatively slight sensibility of the eye for comparisons of luminous intensities, cannot give quite as accurate results as the electric methods, although the accuracy attainable, since the satisfactory establishment of the laws of radiation through- out practically the attainable temperature range, is sufficient, as we have seen, when proper precautions are taken, for all industrial and most scientific needs. The range of this py- OPTICAL PYROMETER 345 rometer is from about 650 C. to the highest attainable temperature. The optical or radiation pyrometer is peculiarly well adapted for many cases in which other methods fail, as when contact with the object whose temperature is sought cannot be made or when for any reason the pyrometer must be placed at a distance ; for example, in the case of a moving body, as a rail passing into the rolling mill; in. the case of very high temperatures, as of the crucible of the blast furnace or that of the electric furnace; in the case of isolated bodies radiating freely into the air, as flames or wires heated by an electric current which cannot be touched without changing their temperature. We have also seen that it may give very exact results in such cases when the emissive properties of the substances sighted upon are known, as is often the case. It is also convenient in the case of strongly heated furnaces, as steel and porcelain furnaces. But in this usage care must be taken to guard against the brightness of the flames, always hotter than the furnace, and against the entry of cold air. The arrangement with the closed tube described in connection with the heat-radiation pyrometer is advisable if it is desired to ob- tain the most exact results. The optical pyrometer has the incon- venience to require active intervention on the part of the operator and can hardly be intrusted to a workman without oversight, while the set-up of the heat-radiation pyrometer may be made so that an observation reduces to a reading upon a scale. The latter pyrometer, however, is the more subject to error due to lack of blackness, flames, and furnace gases. Some Industrial Uses. The several forms of optical pyrom- eter using monochromatic light have been very generally intro- duced into industrial practice, where they are rendering most useful service, and for many operations they may advantageously replace the eye of the operator. Practically every furnace opera- tion can be controlled by this type of pyrometer with great pre- cision, with a resulting saving of fuel and a more uniform furnace product. A few of the types of furnaces for which such pyrom- 346 HIGH TEMPERATURES eters are adapted are the various steel-melting furnaces, blast furnaces, coke ovens, ceramic kilns, and glass weirs. In forging, annealing, hardening, and similar operations on steel, and in foundry practice in general, such pyrometers are equally useful. We have already called attention (page 305) to some industrial measurements made by Le Chatelier with his optical pyrometer. We may also mention some determinations with the Wanner pyrometer on a battery of six coke ovens: Oven. 123456 Over the retorts 1232 1264 1370 1464 1409 1436 Just over generator . 1409 1397 1464 1397 1296 1264 Fifth flue 1126 1002 1112 1104 1096 1119 Next to last flue 992 982 918 932 970 932 The Morse or Holborn-Kurlbaum type may be sighted on distant objects conveniently. It is possible to set up such an instrument in a foundry or forging shop and from one position measure temperatures of several furnaces, of pieces under the hammer, and of metal being poured into and from ladles. Measurement of the Relative Intensity of Different Radia- tions. It is on this principle that rests the eye estimation of temperatures such as are made by workmen in industrial works. Numerous attempts, none very successful, have been made to modify this method and make it precise. There is need to con- sider this mainly from the point of view of a rough control over the heating of industrial furnaces. Recently a modification of this method has been devised by Nordmann, which, as we shall see, is of interest in the estimation of the extremely high tem- peratures of stars. Use of the Eye. Pouillet made a comparison of the colors of incandescent bodies in terms of the air thermometer. The table that he drew up is reproduced everywhere to-day: POUILLET' S COLOR SCALE. First visible red 525 Dull orange 1100 Dull red 700 Bright orange 1200 Turning to cherry 800 White 1300 Cherry proper 900 Brilliant white 1400 Bright cherry 1000 Dazzling white 1500 OPTICAL PYROMETER 347 The estimation of these hues is very arbitrary and varies from one person to another; more than that, it varies for the same person with the exterior lighting. The hues are different by day from those by night; it is thus that the gas flame, yellow during the day, appears white at night. It is only in the reds that any accuracy can be had by the eye method. Workmen can sometimes guess to better than 25 C. up to 800 C. At 1200 errors of over 200 will be made. Use of Cobalt Glass. One may exaggerate the changes of hue in suppressing from the spectrum the central radiations, the yellow and green for example, so as only to keep the red and the blue. The relative variations of two hues are the greater the more separated they are in the spectrum; now, the red and the blue form the two extremities of the visible spectrum. It has been proposed for this purpose to use cobalt glass, which cuts out the yellow and green, but lets pass the red and blue. It must be remembered that the ratio of the radiations transmitted varies with the thickness of the glass as well as with their absolute intensities. Let I a and I b be the intensities of the radiations emitted, k a and k b the proportions transmitted by the glass through a thick- ness i. Through a thickness e the proportion transmitted will be which will vary with e in all cases that k a is different from k b . It results from this that two cobalt glasses, differing in thick- ness or in amount of cobalt, will not give the same results. So that if the cobalt glass habitually used is broken, all the training of the eye goes for naught. Besides, cobalt has the inconvenience of having an insufficient absorbing power for the red, which predominates at the more ordinary temperatures that we make use of. It would be possi- ble, without doubt, by the addition of copper oxide, to augment the absorbing power for the red. One would have better and more comparable results by 348 HIGH TEMPERATURES employing solutions of metallic salts or of organic compounds suitably chosen. But few trials have been made in this matter. Pyroscope of Mesure and Nouel. It is known that by placing between two nicols a plate of quartz cut perpendicularly to the axis, a certain number of the radiations of the spectrum are suppressed. This latter is then composed of dark bands whose spacing depends on the thickness of the quartz and the position of the angle of the nicols. Mesure and Nouel have utilized this principle in order to cut out the central portions of the spectrum ; this solution is excellent and preferable to the use of absorbing media. The apparatus (Fig. 124) consists essentially of a polarizer P and an analyzer A, whose adjustment to extinction gives the zero of graduation of the divided circle CC. This circle is gradu- Fig. 124. Apparatus of Mesure and Nouel. ated in degrees and is movable before a fixed index 7. Between the two nicols P and A is a quartz Q of suitable thickness, care- fully calibrated. The mounting M allows of its quick removal if it is necessary to verify the adjustment of the nicols P and A. The quartz Q is cut perpendicularly to the axis. A lens L views the opposite opening C furnished with a parallel-faced plate glass or, where desired, with a diffusing glass very slightly ground. The relative proportions of various rays that an incandescent body emits varying with the temperature, it follows that for a given position of the analyzer A the composite tint obtained is different for different temperatures. If the analyzer is turned while a given luminous body is viewed, it is noticed that the variations of coloration are much more rapid for a certain position of the analyzer. A very slight rota- OPTICAL PYROMETER 349 tion changes suddenly the color from red to green. Now, if the analyzer is left fixed, a slight variation in the temperature of the incandescent body produces the same effect. The transmission hue red-green constitutes what is called the sensitive hue. There are then two absorptions, one in the yellow and the other in the violet. This apparatus may be employed in two different ways. First fix permanently the analyzer in a position which gives the sensi- tive hue for the temperature that is to be watched, and observe the changes of hue which are produced when the temperature varies in one direction or the other from the chosen temperature. This is the ordinary method of use of this instrument. It is desired in a given manufacturing process (steel, glass) to make sure that the temperature of the furnace rests always the same; the instrument is adjusted once for all for this temperature. It suffices to have but a short experience to train the eye to appre- ciate the direction of the change of hue. The inventors have sought to make of their apparatus a meas- uring instrument; this idea is quite open to debate. In theory this is easy; it suffices, instead of having the analyzer fixed, to make it turn just to the securing of the sensitive hue and to note the angle which gives the position of the analyzer. But in fact the sensitive hue is not rigorously determinate and varies with the observer. A graduation made by one observer will not hold for another. It is not even certain that the same observer will choose always the same sensitive hue. At each temperature the sensitive hue is slightly different, and it is impossible to re- member throughout the scale of temperatures the hues that were chosen on the day of the graduation. There is even considerable difficulty to recall this for a single temperature. The following figures will give an idea of the differences which may exist between two observers as to the position of the sensi- tive hue: Temper- Angle of analyzer, ature. (i) (2) Sun 6000 84 86 Gas flame 1680 65 70 Red-hot platinum 800 40 45 350 HIGH TEMPERATURES The errors in the estimation of temperatures which result from the uncertainty of the sensitive hue will thus exceed 100. With observers having had more experience the difference will be somewhat reduced, but it will remain always quite large. Crova's Pyrometer. Crova endeavored to give to the method of estimation of temperatures based on the unequal variation of different radiations of the spectrum a scientific precision by measuring the absolute intensity of each of the two radiations utilized; but this method, from the practical point of view, does not seem to have given more exact results than the preceding ones. The eye is much less sensitive to difference of intensity than to difference of hue, so that there is no advantage in making use of observations of intensity. Crova compared two radiations, X = 676 (red), X = 523 (green), coming from the object studied and from the oil lamp used as standard. For this purpose, by means of a variable diaphragm, he brings to equality one of the two radiations emanating from each of the sources, and measures afterwards the ratio of the intensities of the two other radiations. The apparatus is a spectrophotometer. Placed before half the height of the flame is a total reflecting prism, which reflects the light from a ground glass, lighted by the radiations from an oil lamp, having first passed through two nicols and a diaphragm of variable aperture. On the other half of the slit is projected by means of a lens the image of the body to be studied. Before using the apparatus it is necessary to adjust the ex- treme limits of the displacement of the spectrum so as to pro- ject successively on the slit, in the focus of the eyepiece, the two radiations selected (X = 676 and X = 523). For this purpose there is interposed between the two crossed nicols a 4-mm. quartz plate which reestablishes the illuminations; for extinction again, the analyzer must be turned 115 38' for X = 523, and OPTICAL PYROMETER 351 65 52' for X = 676. The instrument is then so adjusted that the dark band produced by the quartz is situated in the middle of the ocular slit. The apparatus thus adjusted, in order to make a measurement at low temperatures, inferior to those of carbon burning in the standard lamp, one brings to equality the red radiations with the diaphragm, then, without touching the diaphragm again, the green is brought to equality by turning the nicol. The optical degree is given by the formula V = 1000 cos 2 a denoting by a the angle between the two principal sections of the nicols. For higher temperatures the operation is reversed; one brings first the green to equality by means of the diaphragm, then the red to equality by a rotation of the analyzer. The optical degree is then given by the formula N = I0 , and the rotation vary- cos 2 a ing from o to 90, the optical degrees vary from 1000 to infinity. This method, which is theoretically excellent, possesses certain practical disadvantages: 1. Lack of precision of the measurements. In admitting an error of 10 per cent in each one of the observations relative to the red and green radiations, the total possible error is 20 per cent; now, between 700 and 1500 the ratio of intensities varies from i to 5: this leads to a difference of ^ m 800, or 32. 2. Complication and slowness of observations. It is difficult to focus exactly on the body or the point on the body that one wishes to study. The set-up and the taking of observations sometimes require about half an hour. 3. Absence of comparison in terms of the gas scale. The a priori reason that had led to the study of this method was the supposition that, in general, the emissive power of sub- stances was the same for all radiations and that consequently its influence would disappear by taking the ratio of the intensities of the two radiations. The measurements of emissive power 352 HIGH TEMPERATURES given previously prove that this hypothesis is the more often inexact. Crova also suggested that the upper limit of the spectrum of an incandescent body might be used as a measure of its tem- perature, and Hempel has tried this method with a special form of spectroscope, using a luminescent screen for observing when the upper spectrum limit is beyond the visible radiations; but, as compared with the photometric and radiation pyrometers, only crude results can be obtained. Use of the Flicker Photometer. Lummer and Pringsheim have shown that the combination of a spectral apparatus with a flicker photometer permits of greatly increasing the accuracy of the method of comparison of the intensities of two colors, and also permits the use of Wien's law (page 251) in the calculation of temperatures. Sighting on a black body at the absolute temperature T, and measuring the two intensities /i and 7 2 corresponding to the wave lengths Xi and \2, we have from Wien's law: i/i i X2 , log - 1 = 5 log- 2 + / 2 AI in which T is the only unknown. Thiirmel has shown that the Purkinje effect does not vitiate the observations, and that results good to better than 2 per cent can be obtained, and that an observer will repeat his readings within this limit of error. Here are some of Thiirmers observations on a black body: TEMPERATURE WITH SPECTRAL FLICKER PHOTOMETER. i -- Temperature with -- Ratio of wave lengths. apparatus. Thermocouple. 660-480 1502 1477 660-500 1489 ____ 660-480 1742 1698 660-500 i 703 . .... Sighting on other objects than a black body will give incorrect temperatures, usually low, due to the difference in shape of the intensity curve from that of a black body, and on account of OPTICAL PYROMETER 353 the varying value of the absorption coefficient with wave length from one substance to another (see page 256). Stellar Pyrometers. Of recent years there has been an in- creasing interest among astronomers in the determination of the physical characteristics of the stellar bodies, resulting in the development and modification of physical instruments suited to their needs. Assuming that the ratio of intensities of two spectral colors, red and blue for example, varies according to Planck's law (page 251) for the terrestrial and celestial bodies sighted upon, M. Nordmann has recently constructed a heterochrome photom- eter and used it for the estimation of effective stellar tempera- tures. With this apparatus, which is still in a somewhat crude state of development, measurements are made, in the various parts of the spectrum, of the brightness of the star under observation referred to that of an artificial star realized by means of a second- ary electric standard, interchanging, in the path of rays common to the two stars, a series of monochromatic liquid screens. Consider measurements with red and blue light. If r, 7", T" , . . . are known temperatures of definite light sources, given, for example, by electric furnaces and the carbon arc, and if R, R', R", ... and B, B' ', B" , ... are the corre- sponding intensities of the images as measured through red- and blue-light filters respectively by means of the stellar photometer, 7? i then, according to Planck's law, the relation log vs. is a B T straight line. With the apparatus once standardized at known temperatures therefore, it is only necessary to measure the red and blue intensities for any light source as a star, in order to find its apparent or black-body temperature. The temperature found will approach the true temperature the more nearly the spectral-energy curve of the star approaches that of the black body. In the form of apparatus used by Nordmann, it is also necessary to correct for the shift of equivalent wave length with temperature of the monochromatic screens used. This could be avoided by transforming the apparatus into a spectrophotom- 354 HIGH TEMPERATURES eter in much the same manner as Henning's spectral pyrometer eliminates the colored glasses of the Holborn-Kurlbaum instru- ment. Some of the results found by M. Nordmann for effective stellar temperatures absolute are as follows: p Persei 2870 f Cephii 4260 Sun 5320 7 Cygni 5620 Polaris. 8,200 a Lyrae 12,200 d Persei 18,500 X Tauri 40,000 M. Fery has realized a form of stellar pyrometer which elimi- nates the use of colored screens. The principle of this apparatus, which is based on Wien's displacement law (page 249), consists in modifying the color of a comparison lamp by changing the Fig. 125. F6ry Spectral Pyrometer. ratio of the monochromatic intensities which it emits, so as to match this color with that of the star whose temperature is to be measured. In order to realize this principle, the light from the lamp L (Fig. 125), after passing through the slit F, is dispersed by the direct- vision prism P, and, by means of the lenses L and LI, forms a spectrum in the plane of a diaphragm D, to which we shall return. A third lens LZ forms, on the half -silvered mirror G, a white or undispersed image of the face of the prism P. The light from the star is concentrated by a telescope objective, whose tube is shown at T, and an image of the star is formed on the nonsilvered part of the mirror G, and may be examined by OPTICAL PYROMETER 355 an ocular simultaneously with the adjacent luminous area due to light from the comparison lamp. Fig. 126 gives the details of the diaphragm D of Fig. 125. The spectrum is formed between the two screens V and V'\ this last is semicircular and may be turned about A as an axis. This rotation of V causes the ratio of the intensities of J the extreme red and blue rays to vary u ^ l -'A and gives to the field as projected by the \ V ' lens L% (Fig. 125) the desired color. Both of the above types of apparatus may be calibrated for the lower tern. Fig. 126. Detail of i p * * r Diaphragm, peratures by means of an electric fur- nace, and for the higher stellar temperatures by taking the arc and sun temperatures as fixed points. Spectrophotometric measurements of the apparent tempera- tures of the sun and 109 stars have been made by Wilsing and Scheiner, using as comparison source an incandescent lamp cali- brated against a black body. Light of five wave lengths was used, and the observations were reduced in terms of Planck's law, using an equation similar to that of p. 255. Here are some of their results, assuming c% 14,600. WILSING AND SCHEINER'S STELLAR TEMPERATURES (ABS.). Sun 5130 to 5600 a Leoni 8700 f Pegasi 7900 a Lyrae 8100 a Pegasi 8700 7 Geminorum 6900 t ( We shall return to the question of the sun's apparent tem- perature in Chapter XI. Action of Light on Selenium. It has been known for a long time that light incident upon selenium changes the electric re- sistance of the latter, and pyrometers based on this principle have been devised. Light from an incandescent source whose temperature is sought falls upon a selenium cell forming part of an electric circuit in which are a battery and ammeter. As the light varies in intensity due to changes in temperature, the re- 356 HIGH TEMPERATURES sistance of the selenium varies, and the indications of the ammeter may be empirically calibrated in terms of temperature. As sele- nium is quite insensible to the invisible heat waves, the lower limit of this method is above incandescence. Selenium also requires some time to recover its original resistance after being acted upon by light, and this lag might prove troublesome. As a dial instrument is used, the method could readily be made recording. CHAPTER IX. VARIOUS PYROMETRIC METHODS. WHILE some of the several types of pyrometer which we have described in the preceding chapters have, by a process of elimina- tion, become generally recognized as meeting most requirements for high-temperature measurements, scientific and industrial, there nevertheless remain several methods, some of which are useful in special fields of investigation or practice, and others mark some important development in the history of pyrometry. We shall mention briefly a few of these methods. Wedgwood's contraction pyroscope, the oldest among such instruments, presents to-day hardly more than an historic inter- est, for its use has been almost entirely abandoned. It utilizes the permanent contraction assumed by clayey matters under the influence of high temperature. This contraction is variable with the chemical nature of the paste, the size of the grains, the com- pactness of the wet paste, the time of heating, etc. In order to have comparable results, it would be necessary to prepare simul- taneously, under the same conditions, a great quantity of cylinders, whose calibration would be made in terms of the gas thermometer. Wedgwood employed cylinders of fire clay, baked until dehy- drated, or to 600; this preliminary baking is indispensable, if one wishes to avoid their flying to pieces when suddenly sub- mitted to the action of fire. These cylinders have a plane face on which they rest in the measuring apparatus, so as always to face the same way (see the frontispiece). The contraction is measured by means of a gauge formed by two inclined edges; two similar gauges of 6 inches in length, one an extension of the other, are placed side by side; at one end they have a maximum separation of 0.5 inch, and at the other a minimum separation of 0.3 inch. Longitudinally the divisions are of 0.05 inch; each 357 358 HIGH TEMPERATURES division equals %\-$ of T % of an inch, or yaV^ inch, which corre- sponds to a relative contraction of y^Vo "*" iV = eTo" m terms of the initial dimensions. We then have the following relation between the Wedgwood degrees and the linear contraction per unit of length: Wedgwood ........... o 30 60 90 120 150 180 210 240 Contraction .......... o 0.05 o.io 0.15 0.20 0.25 0.30 0.35 0.40 Le Chatelier has made experiments to determine the degrees of the Wedgwood pyrometer in terms of the scale of the air thermometer by making use of clayey substances of different kinds, and in the first place of the cylinders from an old Wedg- wood pyrometer of the Ecole des Mines. The contraction which accompanies the dehydration is quite variable with the nature of the pastes. In these experiments the time of heating was half an hour. Centigrade temperature ............. 600 800 1000 1200 1400 1550 Wedgwood .......................... o 4 15 36 90 132 Argile de Mussidan ................. o 2 14 36 78 120 Limoges porcelain ................... o o 2 21 88 91 Faience de Choisy-le-Roi ............ o 2 5 12 48 75 Faience de Nevers .................. o o o 32 Melted Melted Kaolin ............................. 04 12 15 55 118 Clay .................. 25 ) Titanic acid .......... 75} .......... o 4 9 iQ 123 160 This table shows how variable are the observations; it is impossible, consequently, to compare the old measurements of Wedgwood and of his successors, because the manufacture of the cylinders has varied with the course of time. Wedgwood had given a graduation made by a process of extrap- olation which he has not explained, a graduation according to which he attributed 10,000 C. to 130 of his pyrometer, which corresponds to about 1550. One might still seek to reestablish the graduation by utilizing the determinations of the fusing points of the metals made by Wedgwood, but the results are too discordant to warrant any definite conclusion. According to Wedgwood, copper would be more fusible than silver, iron would not be far removed from silver; it is probable that these obser- vations were made with very impure metals, or at any rate were VARIOUS PYROMETRIC METHODS * 359 made with metals much oxidized before their fusion. In any case the cylinders which he made use of in his first experiments assume a much greater contraction than those of the pyrometer of the School of Mines whose graduation was given above. One might with considerable reserve indicate the following graduation for measurements made with the first cylinders employed about the year 1780: Wedgwood degrees o 15 30 100 140 Centigrade degrees 600 800 1000 1200 1400 The preparation of the cylinders was a most care-taking oper- ation. Molded in soft paste, they were necessarily somewhat irregular. After the first baking they had to be trimmed to bring them to a uniform size. To-day, in several pottery works where the method is still employed, a much greater regularity is obtained by using a very dry paste, 5 per cent of water for example, molding it under great pressure, about 100 kg. per square centimeter, in molds of turned steel. The precision of the measurements is increased by augmenting the diameter, to 50 mm. for example. It is necessary at the same time to reduce the thickness to about 5 mm., in order that the compression be uniform throughout the mass. This apparatus cannot be recommended in any instance as a true pyrometer, serving indirectly to evaluate temperatures in terms of the air-thermometer scale. The graduation is laborious and can only be made by means of the intermediary of another pyrometer; the use of fixed points is riot adapted for this gradua- tion because the curve of contraction of clay in function of the temperature is too irregular for two or three points to determine it; in no case do the indications of this instrument possess any considerable precision. But as simple pyroscope, that is to say, as an apparatus to indicate merely the equality or inequality of two temperatures, the Wedgwood pyrometer is very convenient. It has the advan- tage of costing almost nothing and it is easy to use and within the comprehension of any workman. It seems to be particularly recommendable for certain ceramic industries, in which the ordi- HIGH TEMPERATURES nary pastes found there may be used to make the contraction cylinders. It is necessary for this that the normal baking of these pastes be stopped at a temperature comprised within the period of rapid contraction. This is the case with fine faience and with ordinary earthenware. That would not be the case, however, for stanniferous faience nor for porcelain, because the baking of the first is stopped before the beginning of the con- traction, and that of the second after its finish. Expansion of Solids. Some of the earliest forms of indicating pyrometers were based on the relative expansion of two metals, or of a metal and graphite or fire clay. Some of these instruments have had and still enjoy a very wide use both in Europe and America, often under the name mechanical thermometers for the lower-range instruments, and some of them are suitable for certain industrial processes not requiring exact temperature determination or control. A common form of dial instrument is shown in Fig. 127. A tube of iron incloses a rod of graphite, and their differential expan- sion with change in temperature is com- Imunicated by levers to a pointer turning over a dial graduated in degrees. The upper limit of these instruments is about 800 C. (1500 F.), but they deteriorate rapidly when used at the higher tem- peratures. Their indications change with time, due to changes produced in the materials by continued heatings. Cor- recting the zero of such an instrument, which should be done frequently, does not completely correct the rest of the scale, as the expansion properties of the two materials change differently with heating. Varying depths of immersion will also change the readings. Fig. 127. Expansion Pyrometer. VARIOUS PYROMETRIC METHODS The Joly Meldometer. A modified form of this instrument was previously mentioned (page 271). As in its usual form it may be of great service to chemists, mineralogists, and others in determining the melting points and identification of minute speci- mens of minerals, salts, metals, and alloys, a further description may be of interest. H Platinum-Strip Air Current Shield Fig. 128. Joly's Meldometer. A platinum strip (Fig. 128) 10 cm. long, 4 mm. wide, and 0.02 mm. thick is held between two clamps C, C, and kept under a slight tension by the spring s. A storage-battery current con- trolled by a small step rheostat R is sent through the platinum strip whose length at any instant is given by the micrometer screw M, whose contact is made appreciable by the closing of the circuit of an electric bell. The platinum strip is calibrated preferably by means of salts of known melting points, as KNO 3 (399 C.), KBr (730), NaCl (800), and K 2 SO 4 (1060). Metals. 362 HIGH TEMPERATURES may also be used, but they tend to deteriorate the platinum. The upper limit of the instrument is about 1500 C., the Pd point being obtainable with difficulty. Permanent elongation sets in somewhat before this point is reached. The gold point (1063 C.) can be determined to 2 C., and only a few moments are required for an observation. To take an observation, a speck of the specimen whose melt- ing point is sought is placed on the middle of the strip under a low-power microscope magnifying about twenty-five times. The current is increased and at the instant of melting, as observed with the microscope, the micrometer is set to make contact and read; when by interpolation, most conveniently made graphically, the temperature is found corresponding to the length of strip observed. This instrument gives a nearly but not quite linear relation between length of strip and temperature. High-range Mercury Thermometers. Although mercury boils normally at about 356 C., yet this liquid subjected to high pressure may be kept from boiling and, suitably inclosed, may be used as thermometric substance to much higher temperatures. Compressed under an atmosphere of some inert gas, as nitrogen or carbonic acid, 'and inclosed in a very hard glass, the mercury thermometer can be used up to 550 C. (1000 F.). When a ther- mometer designed for only moderate temperatures, 200 C. or less, is sealed off gas free, there will be distillation of the mercury into the colder parts of the bore unless the column projects sufficiently above the heated region or the whole thermometer is immersed. There are two methods of producing the necessary pressure within the bore to prevent distillation and boiling of the mercury. In the one, there is a small bulb at the top of the bore, and the thermometer is sealed off at atmospheric pressure with the mer- cury at ordinary temperature; in the other, there is a large upper bulb, and the sealing off is done at increased pressure, making use of an auxiliary bulb. The second construction is preferable, as the internal-pressure change with rise of temperature, and con- sequent deformation of the main bulb containing mercury, is much less than with the first. VARIOUS PYROMETRIC METHODS Due to deformations in the glass, and consequent changes in readings, all high-range mercury thermometers should be fur- nished with some fixed point, preferably the ice point. This permits controlling conveniently the behavior of the thermom- eter due to changes in the volume of the bulb after the instru- ment has been calibrated. The bulbs of such thermometers should be carefully annealed, before filling, at a temperature higher than the instrument is to be used, and the thermometer should also be annealed after it is made and allowed to cool slowly, otherwise considerable and irregular changes in its in- dications will occur, amounting to several degrees. It is also advantageous to heat and cool slowly the thermometer a great many times before testing and using it. The zero reading of such a thermometer should be taken after every observation in work of precision. If a considerable length of stem emerges into the air when taking a reading, a very considerable error, 25 C. or so, may be introduced at high temperatures due to the differ- ence in temperature of the bulb and stem. This "stem correc- tion " varies slightly from one kind of glass to another and is very nearly: Stem correction = 0.00016 - n (T f) C., = 0.000088- n-(T /)F., where n = number of degrees emergent from bath; T = temperature of bath; / = mean temperature of the emergent mercury column determined by some auxiliary means, as the faden thermometer of Mahlke. Among the thermometric glasses for the construction of high- range instruments, and the upper limits to which they may be used safely, are: Jena i6 m normal, Corning normal, and the French verre dur, which reach 450 C. or somewhat higher; Jena 59, a borosilicate glass, although sometimes graduated to 550 C., should not be used over 520 C.; with special grades of combustion tubing 570 C. may be reached. If after proper annealing and preliminary heat treatment the zero of a ther- mometer falls, it is being used at too high temperatures. 364 HIGH TEMPERATURES Thermometers which are to be used as high-temperature pri- mary standards, or instruments which reproduce in themselves the temperature scale, should have both the ice and steam points, which permits calibrating the instrument in terms of the funda- mental interval o to 100 C. Due to the fact that the mercury- in-glass expansion varies from glass to glass, and is also different for all of them from the gas expansion on which the temperature scale is based, it is necessary to apply a correction to reduce the readings of a mercury-in-glass thermometer to the gas scale, unless the thermometer was originally " pointed " in terms of this scale. The relation between the scales given by Jena glasses and the gas scale is shown in the following table: VARIATION FROM GAS SCALE OF JENA-GLASS THERMOMETERS. Gas scale. Jena i6 m Gas scale. Jena 59 IU o o O o IOO IOO.OO IOO IOO.O ISO 149 . 90 2OO 200.7 200 200.04 300 304.1 220 22O. 21 325 330-9 240 24O . 46 350 358.1 260 260.83 375 385.4 280 281.33 400 412.3 300 301.96 425 440.7 450 469.1 475 498.0 500 527.8 If the bore of the thermometer is irregular, it should be calibrated by the use of a 5o-degree or loo-degree thread. Ordinary high-temperature thermometers are tested most con- veniently by comparison with a standard, or by taking readings at a series of known temperatures. High- temperature thermom- eters for a given limited range are kept of a reasonable length of stem and at the same time with an open scale by the insertion of intermediary bulbs which eliminate the undesired portions of the scale.* Therrnometric glasses and high-temperature thermometers are dis- cussed in Hovestadt's " Ja'ener Glas " (in German and English), Mathias' " Les Modifications Permanentes du Verre," and in the publications of the Bureau of Standards. Guillaume's ' ' Thermome'trie de Precision ' ' describes details of calibration and manipulation. VARIOUS PYROMETRIC METHODS 365 The glass of mercury thermometers has been successfully re- placed by quartz, which is almost an ideal thermometric envelope, possessing an insignificant expansion and no appreciable zero lag, and capable of being used at very high temperatures. Such mercury-in-quartz thermometers are now constructed by Siebert and Kuhn, and are graduated to about 700 C. Dufour has tried to substitute tin for mercury-in-quartz ther- mometers, thereby attaining a temperature of over 1000 C. Such thermometers have not yet, however, come into use. It is a difficult matter, not yet satisfactorily solved, to find a sub- stance suitable to use as thermometric fluid in quartz at high temperatures. Fusing-point Pyrometry. As long ago as 1827, Prinsep pro- posed to compare temperatures by means of the fusing points of certain metals and alloys. But the nonoxidizable metals are not numerous and all are relatively very costly: silver, gold, palladium, platinum. Use has, however, been made sometimes of these metals and their alloys, in admitting that the fusing point of a mixture of two substances is the arithmetical mean of the points of fusion of the components, which is not quite exact. The use of these alloys is entirely abandoned to-day, and with reason. In a sense, this method of pyrometry may be said to be still in use, since the temperature scales of the several standardizing laboratories are practically defined by the freezing temperatures of pure metals. By making use of metallic salts, among which a great number may be heated without alteration, one might construct a scale of fusing points whose use would be often very convenient; but this work is not yet done, at least not in a sufficiently precise manner. To the separate salts may be added their definite combinations and their eutectic mixtures which have perfectly definite fusing points. But any mixture whatever of two salts cannot be taken, because in general the solidification takes place throughout a large interval of temperature and in a progressive manner. 366 HIGH TEMPERATURES In some metallurgical operations, it is often necessary to be certain that objects are heated above some definite temperature. Salt baths of known freezing points, and of materials not attack- ing the metals used, serve excellently both for heating such objects and automatically giving the minimum temperature allow- able. We may cite in this connection the investigations of Brearley and Morewood and of Grenet on pure salts and eutectic and iso- morphous mixtures suitable for this purpose. For the heat treat- ment of steels, Grenet recommends the following series of salts : GRENET'S SERIES OF SALTS FOR HEAT TREATMENT OF STEELS. Melting point. Melting point. K 2 S0 4 1070 C. KC1 775 C. BaCl 2 955 KBr 730 Na 2 SO 4 865 KI 682 5 K 2 S04+5Na 2 S0 4 ... 850 5.8 KC1+4.2 NaCl... . 655 3 K 2 SO 4 +7Na 2 SO 4 ... 830 3 NaCl+7KBr 625 2 K 2 SO 4 +8Na 2 S0 4 :.. 825 Ba(NO 3 ) 2 600 Na 2 CO 3 810 Ca(NO 3 ) 2 550 NaCl 800 The uncertainty of our knowledge of the numerical values of the melting points of some of the salts is illustrated in the table on next page and in Chapter XI. It would be worth while to carry out a careful series of deter- minations of the melting points of these and other salts, using the care and refinements of method that have been employed in recent work on metals, and employing large quantities of salt, 300 to 1000 grms. " Sentinel pyrometers " and pastes, such as those of Brearley (The Amalgams Company, Sheffield), are also useful in certain operations. The former are cast in the form of small cylinders from molecular mixtures of salts. For temperatures below 500 C. they are inclosed in glass tubes and therefore last indefinitely, and for higher temperatures are placed in saucers (Fig. 129). Two such " sentinels " may be used, for example, to control a furnace within any given temperature range, the one being liquid and the other solid. The paste, made from salts mixed with VARIOUS PYROMETRIC METHODS MELTING POINTS OF SALTS. 367 Date. 1896. 1894. 1903. 1905. Authors. Ramsay and Eumorfopoulos. McCrae. Ruff and Plato. Huttner and Tammann. Method. Meldometer. Thermoelectric. Thermoelectric. Thermoelectric. Calibration data. KN0 3 =339 K 2 S0 4 =Au+7 = 1052. Diphenylamine =304 SBP=445 Au=IO?2. Reichsanstalt scale. Sb= 630.6 Au=io64. Quantity in grms. O.OOI. Small. 20. 25-40. Na 2 S0 4 .. Na 2 CO 3 ... NaCl NaBr Nal K 2 SO 4 .... K 2 CO 3 .... KC1 . 884 851 792 733 603 1052 880 762 883 86 1 813 761 695 1059 893 800 880 820 765 650 1050 880 79 897 853 810 748 1074 894 778 KBr.. 777 746 75 740 KI 614 723 705 680 Li 2 SO 4 ... . Li 2 CO 3 . . . CaCl 2 SrCl 2 BaClj.... 815 618 710 796 844 802 854 916 780 960 859 734 paraffin, is smeared onto the metal to be heated, and melting of the paste is readily recognized. The range covered by the sen- tinels and pastes is to 1070 C. Fig. 129. Sentinel Pyrometers. Any method based on the use of fusing points alone, whether metals, alloys, or salts, is evidently a discontinuous one, and has- its main usefulness in processes where only a maximum or mini- mum temperature is required. 368 HIGH TEMPERATURES Fusible Cones. Instead of utilizing the fusion of crystallized substances which pass abruptly from the solid to the liquid state, use may be made of the progressive softening of vitreous matters, that is to say, of mixtures containing an excess of one of the three acids, silicic, boric, or phosphoric. It is necessary in this case to have a definite process for defining a type degree of softening; a definite depression of a prism of given size is taken. These small prisms, formed of vitreous matters, are known under the name of fusible cones. This method was first devised by Lauth and Vogt, who applied it in the manufactures at Sevres before 1882. But they did not develop it as far as was possible; they were content to construct a small number of fusible cones corresponding to the various temperatures employed in the manufacture of the Sevres porce- lain. Seger, director of a research laboratory at the royal pottery works of Berlin, published, in 1886, an important memoir on this question. He determined a whole series of fusible cones known as Seger cones, of intervals of about 25, including the interval of temperature from 600 to 1800. The substances which enter into the composition of these cones are essentially: Pure quartz sand; Norwegian feldspar; Pure carbonate of lime; Zettlitz kaolin. The composition of this last is: SiO 2 46.9 A1 2 O 3 38.6 FeO 3 0.8 Alkalies i . i Water 12.7 In order to obtain very infusible cones, calcined alumina is added, and for very fusible cones oxide of iron, oxide of lead, carbonate of soda, and boric acid. The shape of these cones (Fig. 130) is that of triangular pyra- mids of 15 mm. on a side and 50 mm. high. Under the action of heat, when softening begins, they at first contract without VARIOUS PYROMETRIC METHODS 369 change of form, then they tip, bending over, letting their apex turn downwards, and finally flattening out completely. One says that the cone has fallen, or that it has melted, when it is bent halfway over, the point directed downwards. The fusing points of these substances have been deter- Fig. 130. Seger Cones. mined at the Berlin porcelain works by comparison with the Le Chatelier thermoelectric pyrometer, previously described. The cones are numbered, for the less fusible, which were first adjusted, from i to 38; this last, the least fusible, corresponds to 1980. The second series, more fusible, and established later, by Cramer and Hecht, is numbered from 01 to 022 ; this last cone, the most fusible, corresponds to 590. If, instead of using the cones of German make, one wishes to make them himself in employing the same formulae, it is prudent to make a new calibration. The kaolins and feldspars from different sources never have exactly the same compositions, and very slight variations in their amounts of contained alkali may cause marked changes in the fusibility, at least for the less fusible cones. It is also well, on this account, to compare the behavior of new cones with old, even from the same maker. It is to be noticed that in a great number of cones silica and alumina are found in the proportions A1 2 O3 + 10 SiC>2. This is for the reason that this mixture is more fusible than can be had with silica and alumina alone. It is the starting point to obtain the other cones, the less fusible by the addition of alumina, and the more fusible by the addition of alkaline bases. The table on pages 371 and 372 gives the list of cones of Seger 's scale as they were originally issued. These cones may be classed in a series of groups, in each of which the compositions of different cones are derived from that of one of them, generally the most fusible, by addition in varying proportions or sometimes by substitution of another substance. 370 HIGH TEMPERATURES The cones 28 to 38 are derived from the cone 27 by the addi- tion of increasing quantities of A1 2 C>3. The cones 5 to 28 from the cone 5 by addition of increasing quantities of the mixture A1 2 O3 + 10 SiO 2 . The cones i to 5 from the cone i by substitution of increasing quantities of alumina for the sesquioxide of iron. The cones oio to i from the cone i by the substitution of boric acid for silica. 000 Fig. 131. Composition of Seger Cones. The cones 022 to oil from the cone 022 by the addition of increasing quantities of the mixture A1 2 3 + 2 Si0 2 . Fig. 131 gives the graphical representation of these data; the ordinates are temperatures, and the abscissae are values of x from the table. These fusible cones of Seger are pretty generally used in the ceramic industry; they are very convenient in all intermittent furnaces whose temperature has to increase constantly up to a VARIOUS PYROMETRIC METHODS 371 certain maximum, at which point the cooling off is allowed to commence. It is sufficient, before firing up, to place a certain number of fusible cones opposite a draft hole closed by a glass, through which they may be watched. In seeing them fall suc- cessively, one knows at what moments the furnace takes on a series of definite temperatures. In continuous furnaces, the cones may be put into the furnace during the process, but that is more delicate. It is necessary to place them on little earthenware supports that are moved into the desired part of the furnace by an iron rod. When, on the contrary, they are put in place at the start in the cold furnace, they are held in place by a small lump of clay. THE ORIGINAL SEGER CONE SCALE. Nos. D c g ' Composition. X Formulae. 38 1890 A1 2 2+I S0 2 9 36 1850 " + 1 5 8 35 1830 +2 34 1810 XA1 2 O 3 33 32 1790 1770 +4 . +(i-X)(Al,O l +10 Si0 2 ) 31 1750 +5 30 1730 +6 29 1710 +8 1 28 1690 + 10 27 1670 ( 0.3 K 2 |o.7CaO } +2o(Al 2 O 3 +io SiO,) 26 1650 +7.2 93 25 1630 " +6.6 24 1610 " +6 23 1590 " +5.4 22 1570 ** +4-9 21 1550 +4-4 20 1530 " +3.9 , 19 1510 +3-5 18 1490 " +3.1 X(Al 2 O 3 +ioSiO 2 ) 17 1470 " +2.7 . . Y /o 3 K 2 3J 16 1450 +2.4 79 -HI- A) ^ ; 7Ca 15 14 1430 1410 .. +2.1 +1.8 +o.s(A! 2 O 3 +ioSiO 2 )) 13 1390 41 +1.6 12 1370 " +1.4 II 1350 " + 1.2 58 10 1330 * + 1 9 1310 +0.9 8 1290 * +0.8 7 1270 * +0.7 6 1250 4 +0.6 5 1230 4 +0.5 4 1210 +o.sA! 2 O 3 +4SiO 2 i 3 2 1190 1170 i i +l"fiM}+^' X (0.5 Al 2 O 3 +4SiOj) + (i-X). (o. 5 Fe 2 +4SiO 2 +o.7CaO) I 1150 i + \o.l S&j+'ao, 372 HIGH TEMPERATURES THE ORIGINAL SEGER CONE SCALE (Continued}. Nos. D f Composition. X Formulae. 01 1130 1 { 0.7 CaO } + { 0.2 FejOa J ' + 3-95 Si0 2 o.osB 2 O 3 1.05 02 IIIO i " + + 3-9oSiO 2 o. 10 B 2 O 3 03 04 05 1090 1070 1050 i " + + i " + + 3.85 SiO 2 o. 15 B 2 O 3 3.8oSiO 2 0.20 B 2 O 3 3.75 Si0 2 o.2 5 B 2 3 1.25 (SiO 2 -B 2 O 3 ) 06 1030 i " +i " + 3.7oSiO 2 0.30 B 2 O 3 +d-x) ( ; 7C |oj 07 1010 i " +i " + 3.65 SiO 2 o.3S B 2 3 + {o > :2F(L 2 3 }+ 4Si0 ) j .. , (3.6oSiO 2 08 900 |o.4oB 2 3 3.55 SiO 2 09 970 i +i + 0.45 B 2 3 010 on 950 920 '{;ip-' *> + 3.5 SiO 2 0.5 B 2 3 13.6 SiO 2 [i.oB 2 O 3 5 0.62 012 890 I " +0.75 " + 3-5 SiO 2 i.oB 2 O 3 013 860 I " +0.70 " + 3-4Si0 2 i.oB 2 O 3 014 830 i " +0.65 " + 3.3Si0 2 i . o B 2 O 3 015 016 017 018 800 770 740 710 i " +0.60 " + i " +0.55 " + i " +0.50 " + i " +0.40 " + 3.2Si0 2 i . o B 2 O 3 3.iSi0 2 i.oB 2 O 3 3.oSiO 2 i.oB 2 O 3 2.8SiO 2 i.oBsO, 0.57 X(2 Si0 2 +Al 2 3 ) N ' | 2 Si0 2 iS + \ iB 2 o 3 |; 019 680 i " +0.30 " + i!oB 2 O 2 3 020 650 i " +0.20 " + |?'J!$. 021 022 620 590 I " +0.10 " + I " + 2.2SiO 2 i . o B 2 O 3 2.0 SiO 2 i.oB 2 O 3 Recent investigations on Seger cones, in view of their im- provement, have been carried out mainly by the staff of the Laboratorium fur Tonindustrie, and at the Reichsanstalt, and consequently there have been changes in their composition, melt- ing point, and numbering. The improvements have been mainly in increasing the sharpness of melting points, elimination in so far as possible of the lag due to rate of heating, and finding components that are uninfluenced by the usual ceramic furnace atmosphere. This has resulted in the elimination of lead and iron compounds. Cones Nos. 21 to 25 have been dropped, as their melting points were too close together; and four new VARIOUS PYROMETRIC METHODS 373 cones, Nos. 39 to 42, the most refractory of all, have been added. In the following table are given the melting points of the cones according to the Tonindustrie Zeitung circulars for 1910, together with their compositions when the latter differ from the original series. 1910 SCALE OF SEGER CONES. 42 2OOO A1 2 3 4i 1960 Al 2 O 3 o.i3 SiO 2 40 I92O Al 2 3 o.3 3 Si0 2 39 1880 Al 2 O 3 o.66 SiO 2 3 1850 No. Deg. C. No. Deg. C. 37 1825 28 1630 14 1410 36 1790 27 1610 13 1380 35 1770 26 1580 12 1350 34 1750 20 1530 II 1320 33 1730 19 1520 10 1300 3 2 I7IO 18 1500 9 1280 1690 17 1480 8 1250 30 1670 16 1460 7 1230 29 1650 15 H35 Composition No. 6a Deg. C. I2OO 0.013 Na 2 O 1 o. 2 88K 2 I o6o , A1( D (6.8oiSiO 2 o.68 S CaO f -93 A1 2 < 3 \ 0.026 B 2 Os 0.014 MgO J 0.028 Na 2 O ^ $a 1180 0.274 K 2 O 1 A* . j6. 5 6 5 SiO 2 o 666 CaO f ' J3 1 0.056 B 2 O 0.032 MgO j 0.043 Na 2 O 1 4& 1160 0.260 K 2 O 1 AI -, (6. 339 Si O 2 0.649 CaO rO.676Al 2 t J> I 0.086 B 2 O 0^048 MgO J 0.059 Na 2 O 1 3a 1140 0.244 K 2 O ! 6 A j < -^ (6. 083 Si O 2 o 630 CaO | / 2 3 / o. 119 B 2 Oa 0.067 MgO J 0.085 Na 2 O 1 2a II2O 0.220 K 2 O 1 A 2 AI ( -, ]5.687SiO 2 o 599 CaO f 2 3 ( 0.170 B 2 O S 0^096 MgO J o . 109 Na 2 O I la 1 100 0.198 K 2 O ! / AI i -^ 5.32oSiO 2 PflO r-39'"- 1 2' J * (0.217 B 2 O a 0.122 MgO 374 HIGH TEMPERATURES 1910 SCALE OF SEGER CONES (Continued). No. Deg. C. oia 1080 02 a 1060 03 a 1040 1020 1000 o6a 980 07 a 960 o8a 940 920 oioa 900 ona 880 OI2E 855 o.i34 Na 2 O o. 174 K 2 O 0.541 CaO 0.151 MgO > 0.625 A1 2 O 3 ( 4.931 SiO 2 (0.268 B 2 O 3 0.157 Na 2 O 0.153 K 2 O 0.513 CaO 0.177 MgO |-o.6ii A1 2 O 3 (4-572 SiO 2 ( 0.314 B 2 O 3 0.182 Na 2 O 0.130 K 2 O 0.484 CaO o . 204 MgO Lo.5 9 8Al 2 O 3 (4.199 SiO 2 1 0.363 B 2 3 0.204 Na 2 O o. 109 K 2 O 0.458 CaO 0.229 MgO I 0.586 A1 2 O 3 J3.86oSiO 2 (0.407 B 2 O 3 0.229 Na 2 O 0.086 K 2 O 0.428 CaO 0.257 MgO > 0.571 A1 2 O 3 (3.467810., 1 0.457 B 2 3 0.247 Na 2 O 0.069 K 2 O 0.407 CaO 0.277 MgO , > 0.561 A1 2 O 3 1 (3.197 Si0 2 (0.493 BaOa 0.261 Na 2 O * 0.055 K 2 O 0.391 CaO 0.293 MgO > ^ 0.554 A1 2 O 3 (2.984 SiO 2 (0.521 B 2 O 3 0.279 Na 2 O ] o.o38K 2 O ! 0.369 CaO j 0.314 MgO J ^0.543 A1 2 O 3 (2.691 S1O 2 0.336 Na 2 O ^ 0.018 K 2 O 1 0.335 CaO 0.311 MgO J ^0.468 A1 2 O 3 (3.o8 7 SiO 2 (0.671 B 2 O 3 0.338 Na 2 O I o.on K 2 O 0.338 CaO 0-313 MgO j -0.423 A1 2 O 3 (2.626 SiO 2 (o.67 S B 2 O 3 0.349 Na 2 O ) 0.340 CaO 0.311 MgO ) 0.4 A1 2 O 3 ( 2.38S1O 2 (0.68 B 2 O 3 0-345 Na 2 O ] 0.341 CaO 0.314 MgO ) 0.365 A1 2 3 | {l:SlSi VARIOUS PYROMETRIC METHODS 375 1910 SCALE OF SEGER CONES (Continued). No. Deg. C. f - Composition. 0.343 Na 2 O OI3a 83S OI4a 8l5 oisa 79 016 750 Bf.* 0.31 A1 2 3 I I' 61 017 730 Bf. 0.2 Al 2 018 710 Bf. o. 13 A1 2 3 | I 26 | 019 690 Bf . o .08 A1 2 3 | I l6 020 670 Bf. 0.04 A1 2 O 3 < 021 650 Bf. o.o 2 Al 2 O 3 j* >4 B 2 o 3 022 600 0.50 Na 2 O Bf= o.2sCaO 0.25 MgO It would seem to be well to replace, in so far as possible, these cones by pure compounds and eutectics having definite melting points, as the softening temperatures of the former are influenced considerably, in some cases 100 C. or more, by the rate of heat- ing, as has been remarked by several investigators. Dr. Kanolt of the Bureau of Standards has carried out a series of measure- ments on some of the" Standard Pyrometric Cones " of Professor Orton of the Ohio State University, corresponding to Nos. 25 to 36 of the Seger series, as well as on this Seger series. His work shows that heating the cones as rapidly as 5 C. per minute, when near their softening temperatures, will give too high values for these temperatures. The rate of heating in the experimental or calibrating furnace must be reduced to more 376 HIGH TEMPERATURES nearly that which obtains in kiln practice, in order to get a fair calibration of the cones for use in the ceramic industries. The Seger and Orton cones were found to agree very closely in their behavior. With slow heating, the cones of the series 25-36 were found to soften at temperatures lower by 40 to 70 C. than indicated in the table of 1910, page 373, agreeing closely in this with the results found by Heraeus. Melting in air or in vacuo gave the same results. Kanolt's measurements were made with an optical pyrometer whose scale is represented by Au = 1064, Pd = 1550, Pt = 1755 C. At the Reichsanstalt, Hoffmann and Meissner find (1911) sim- ilar differences between softening temperatures in ceramic kilns for a time of heating of about sixty hours and in the electric furnace. SOFTENING TEMPERATURES OF SEGER CONES. Cone In electric In ceramic rvffWo number. furnace. kiln. Qce ' 6 8 9 10 13 14 16 17 On the whole, it may be said that the Seger cone series give reliable relative temperatures to about 25 C. at the higher tem- peratures for any given method of procedure, but too much reliance should not be placed on the numerical values of the temperatures apparently measured. WiborgWs Thermophones. Another cheap, discontinuous py- roscope has been put on the market by Wiborgh. His thermo- phones are refractory earth cylinders 2.5 cm. long and 2 cm. in diameter, containing an explosive. A thermophone is quickly deposited in the region whose temperature is sought, and the time noted to the fifth of a second until the cylinder bursts. A table then gives the temperature. Very concordant results are 1225 1160 65 1260 1200 >6o 1285 1180 105 1305 v Galvanometer Wire n * "" ' " v/WWVWW/WAAWW^A | 1 1 I C Bridge Wire I LJ- yVWW\A/>^/WWWWVWNA/WV Battery Thermometer Bulb or Resistance to be Measured Fig. 137. Wiring Diagram, Callendar Recorder. 392 HIGH TEMPERATURES Siemens and Halske recorder (Fig. 149) may also be had with a very sensitive scale, i.e., 1.5 millivolt, for example, for a galva- nometer resistance of 10.5 ohms. Some form of platinum-thermometer bridge with a slide wire is required, and a rheostat capable of fine adjustment. This rheostat is connected in series with a storage cell and serves to adjust the scale of the record. One terminal of the recorder is connected at G between the ratio arms of the bridge BiG and and the other to the sliding contact on the bridge wire. Rh Plugs I, Slide Wire /wwvw Pt Fig. 138. Circuits for Deflectional Recorder. The compensator leads are connected in series with the balancing coils. With this arrangement the resistance S + C is nearly constant, so that if the galvanometer deflection is adjusted to be correct when the thermometer is at o C., or any convenient temperature, the scale will be very nearly correct when the ther- mometer is at any other temperature, the plugs R being suitably adjusted. A much greater sensibility can be obtained in this way than with thermocouples using such a recorder. Thus it is possible to get a scale of 50 mm. to i C. if desired. Such great sensibility, however, would only be necessary in extremely deli- RECORDING PYROMETERS 393 cate thermostatic control. The scale of the galvanometer is not strictly one of equal parts, and this has to be allowed for in exact work. The Leeds and Northrup Recorders. This firm has developed two types of resistance recorders suitable for temperature meas- urements, one having an electromagnetic control and the other mechanical. The former has been perfected to a definite instru- 1 IliiliillillllillilllllillllliliilillliJllii.ilulllaimliuiii Fig. 139. Leeds and Northrup Recorder. ment and the latter is still undergoing improvement in details. Both are of the balance type and give pen records proportional to temperatures in rectangular coordinates; in both the record- ing mechanism is entirely independent of the galvanometer system, and the accuracy uninfluenced by changes in the galva- nometer constant. The general appearance of the recorder with electromagnetic control is shown in Fig. 139. As usually constructed, this re- 394 HIGH TEMPERATURES corder is made for any range over 100 F. to 1800 F. with a sensibility of ^ per cent and a constancy of repetition of f per cent. The pen will follow temperature changes at the rate of ^ per cent of the total range per second, giving a continuous terraced record. The complete recorder consists of four essential parts: a dif- ferential galvanometer, with a plunger magnet attached to it, for operating periodically the contactor; a clock which drives the paper and operates the contacting plunger magnet; a sim- ple mechanism consisting of a screw and two electromagnets of the plunger type for traveling the pen and the balancing con- tact; and a paper drive, moved by the clock, which feeds a con- tinuous band of paper off a roll and over an apron carrying it under an ordinary stylographic pen. The mechanical recorder is of the same general design, except that the moving mechanisms are all driven mechanically, with the added factor that, according as the galvanometer needle is more or less deflected, the pen mechanism will be driven faster or slower. This permits following much more rapid tempera- ture changes than with the electromagnetically controlled re- corder. Both of these instruments may also be arranged for use with thermocouples; and the mechanical recorder has besides been adapted to record directly the differential-temperature curve (page 382) by means of a double galvanometer system, the paper moving proportionally to the temperature of the sample and the pen proportionally to the difference in temperature between the sample and neutral. Carpentier's Electrothermal Recorder. This system gives rec- tilinear coordinates by the use of a cylindrical frame contain- ing the paper, which advances continuously and against which the pen carried by the galvanometer boom impinges periodically by means of the following electrothermal mechanism: The galvanometer boom carries a fine wire of platinum-silver through which an electric current, either direct or alternating, may be .sent from an auxiliary, or the ordinary lighting circuit. The RECORDING PYROMETERS 395 recording pen and wire are so attached to the end of the boom by a spring that when no current is passing in the wire the pen is free of the paper; but the passage of a current weakens the tension of the wire, due to its expansion, and the spring then pulls the pen gently against the paper without shock. With a suit- able commutator, as many as four contacts per second may be had on the paper; and where several records are to be taken on the same sheet for as many separate thermometers, the same principle of electrothermal control may be applied also to the commutator, giving, when the contact surfaces are so spaced, dots or dashes of different length for each circuit. This type of recorder and commutator may evidently be used with any kind of pyrometer whose readings may be taken with a galvanometer. Thermoelectric Recording Pyrometer. We shall group the thermoelectric recording methods according to the type of curve which they register, temperature-time, differential, and tem- perature-rate, all of which may be realized both with photo- graphic and autographic registering apparatus. Due perhaps to the general convenience of the thermocouple for temperature measurements, there appears until very recently to have been more attention paid to the development of recording methods for this type of pyrometer than for any other, and this in spite of the fact that the thermocouple labors under the disadvantage, as compared with the resistance pyrometer, for instance, in having intrinsically very little energy available for the mechanical operation of a recorder. It would seem to be mainly for this reason that the early thermoelectric recorders were photographic instruments; and, in fact, it is only in recent years that satisfactory autographic ther- moelectric recorders have been devised, and they employ various artifices to maintain the sensitiveness of the registering galva- nometer, necessitating usually an intermittent or dotted record, at least with Pt-Rh thermocouples. The difficulty of the problem, in this case, is emphasized when it is recalled that only very weak currents can be had; thus for a precision of 10 C. an 396 HIGH TEMPERATURES apparatus sensible to ^ oTo"o vo ^ * s necessary, and as the galva- nometer should have at least 100 ohms resistance, as previously explained, when the deflection method is used, the corresponding current will be only a millionth of an ampere. In those cases where it is permissible, as in many industrial measurements, to use certain alloys of the base metals which develop large E.M.F.'s, and whose resistance is so low that the galvanometer resistance may also be reduced, it is possible to realize pen recorders giving a strictly continuous record curve of a precision sufficient for many technical operations. Temperature-rate Recorders. As we have seen, it is some- times of interest, more particularly in the laboratory, to measure directly the speed of the occurrence of certain phenomena, as the rate of cooling or chemical transformation. We require an apparatus that will give the rate of change of temperature of the sample in terms of its temperature. Le Chatelier's Experiments. Le Chatelier used this method in 1887 m hi s study of the properties of clays. He was also the first to employ a photographic apparatus for the recording of cooling- or heating-curve data, using an arrangement, to be de- scribed, in which the photographic plate remained stationary. A luminous beam reflected by the galvanometer mirror falls periodically at regular intervals, of a second for instance, upon a fixed sensitive plate. The distance apart of two successive images gives the variation of temperature during unit time, that is, the rate of heating or of cooling; the distance from the same image to the image corresponding to the beginning of the heating will give the measurement of the temperature. In all cases of photographic recording it is well to replace the ordinary galvanometer mirrors, which give images quite insuffi- cient as to definition and brightness, by special mirrors made of a plano-convex lens, silvered on the plane surface. These mir- rors are slightly heavier than parallel-face mirrors, but have two important advantages: the absence of extra images reflected by the front surface of the mirror, and a greater rigidity, which obviates accidental bendings of the mirror arising from the attach- RECORDING PYROMETERS 397 ments to its support. One may easily get good mirrors of this type of 20 mm. diameter, and with more difficulty of 30 mm. diameter. These last give nine times more light than the mirrors ordinarily employed. It is easy to so choose the lens as to give a mirror of desired focal length. A plano-convex lens whose principal focus by transmission is i m. will give, after silvering the plane surface, an optical system equivalent to a spherical mirror whose radius of curvature would be i m. Le Chatelier used discontinuous recording. In this manner of recording, the luminous source should possess periodic varia- tions; one of the simplest to employ is the electric spark between two metallic points. The interruption of the current is pro- duced by a pendulum at definite intervals of time. In order to have a spark sufficiently bright, it is necessary to use an induction coil so worked as to give freely sparks of 50 mm., and to reenforce it by a Leyden jar which reduces the length of these sparks to 5 mm.; it suffices for this to use a jar of i to 2 liters. The choice of metals for the points is equally important; zinc, aluminium, and especially magnesium give sparks that are very photogenic. These metals possess the disadvantage of oxidizing quite rapidly in the air, so that it is necessary from time to time to clean the points with a file. The metallic sticks may have 5 mm. diameter, and the dis- tance apart of the points is 2 mm. One might without doubt, using mercury, which gives sparks as photogenic as does magnesium, construct an inclosed ap- paratus in which the metal would be preserved unchanged. To produce the interruption, there is attached to the pendulum (Fig. 140) a 11. r i i i j j Fig. 140. Clock vertical platinum fork which dips into interrupter two cups of mercury covered with alcohol. It is useful, in order to reduce to a minimum the resistance that the immersion of the fork opposes to the motion of the o 398 HIGH TEMPERATURES pendulum, to place this fork in the same horizontal plane as the axis of rotation of the pendulum. In this way one avoids the translatory movements in the mercury which cause the most trouble. The only refinement with this intermittent lighting is to obtain, with a spark much too large and irregular to be photographed directly, the illumination of a very narrow slit. It is not sufficient to place the spark behind the slit and at a small distance away, because the slightest displacement of the spark would cause the luminous beam to fall outside of the mirror of the galvanometer. This difficulty is overcome by a well-known artifice. A lens is placed between the electrodes and the mirror (Fig. 141) ; the posi- tion of the electrodes is so adjusted that the image of the mirror is formed between them. With a distance apart of the electrodes Fig. 141. Focusing Device. of 2 mm., a lens of 100 mm. focal length and a mirror of 25 mm. diameter, the image of the latter will touch the two points; the spark then necessarily crosses the image of the mirror, and the radiations passed by the lens will fall certainly upon the mirror. One is thus sure in placing before the lens a fine metallic slit that all the rays transmitted will reach the mirror and will be sent to the photographic plate, and that whatever may be the position of the slit in front of the lens. To save time, it is advantageous to take several sets of observa- tions on the same plate; this is easily done by arranging the plate so that it may be displaced vertically between two series, or in adjusting the slit so that it may be moved similarly before the lens. The diagram (Fig. 142) is the reproduction of negatives relative to the action of heat on clays. The first line gives the graduation RECORDING PYROMETERS 399 of the couple; it has been drawn from several different photo- graphs which have been grouped to economize space. The following lines are reproductions of negatives made photographi- cally without any intervention of the hand of the engraver. The second line, for example, represents the heating of an ordinary clay. A slight contraction of the lines between 150 and 350 indicates a first phenomenon with absorption of heat; it is the vaporization of the inclosed water. A second cooling much more marked between 550 and 650 shows the dehydration, properly so called, of the clay, the liberation of the two molecules of water in combination. Finally, the considerable spacing of the lines at 1000 shows a sudden setting free of heat correspond- ing to the isomeric change of state, after which the alumina be- S Se 20 100 445 665 I nil ffiiiimiiimiiiiiiiiiMiii iMMiiiiiiiimJ HI miiiiYiiiiiiiM^ Fig. 142. Heating Curves of Clays. comes insoluble in acids. The other rows refer to the heating of other varieties of clay, the third row to kaolin, the fifth to steargillite. Dejean's Apparatus. Another method of recording the rate of heating or cooling in terms of the temperature has been de- vised by Dejean. The new feature of this method, which gives a continuous record, is the use of an induction galvanometer or relay which may be inserted in the circuit of the more sensitive galvanometer G\ of the Saladin system (Fig. 160). The principle of the apparatus is shown in Fig. 143. The induction relay is a modified d'Arsonval galvanometer having an electromagnet and a movable coil, the latter consisting of two distinct insulated windings, one of which is connected to a thermocouple. Heat- ing or cooling one junction of this couple causes the coil to be 400 HIGH TEMPERATURES deflected and its motion in the field of the electromagnet induces an E.M.F. in the second winding of the coil which is proportional to its angular speed and hence to the rate of change of E.M.F. of the couple, or approximately to the rate of cooling or heating, i.e., to . The induced E.M.F. is measured by joining this at winding to the sensitive galvanometer G\. The galvanometer deflection passes through a minimum when the heating or cooling passes through a minimum, that is, for a region in which there is an absorption or evolution of heat. A second thermocouple in series with the other galvanometer Gz of the Saladin system gives the temperature of the sample. We have, therefore, on the plate Induction Galvanometer Fig. 143. Dejean's Apparatus. P (Fig. 134), when the record is taken photographically, the tem- peratures as abscissae and the rate of cooling as ordinates. Dejean has used this method in the study of steels and has also investigated with it the copper-cuprous oxide system. The transition temperatures are very sharply marked. If desired, direct reading may be substituted for the photographic recording, with an increase in precision. Unless the temperature is chang- ing rapidly, however, this method lacks sensitiveness. It is evidently a perfectly general method for recording the rate of change of E.M.F. ~ In neither Le Chatelier's nor Dejean's arrangement can dif- ferences in the rate of heating or cooling due to the substance itself be distinguished from those due to external causes, since no neutral piece is used (see page 382). RECORDING PYROMETERS 4OI Temperature Time Recorders. There have been a great many types of instruments constructed for recording temperatures directly in terms of time for use with thermocouples. The early forms were for the most part photographic, giving continuous records, while many of the more recent ones are autographic, which usually give discontinuous results. We can mention only a few which illustrate sufficiently well the principles involved. The Apparatus of Sir Roberts- Austen. On account of its his- torical interest as well as its intrinsic usefulness, we shall first describe, with some of its modifications, the photographic appa- ratus of the late Sir Roberts-Austen, director of the royal mint at London. A vertical slit lighted from a convenient source projects its image, by means of the galvanometer mirror, on a metallic plate pierced by a fine horizontal slit, and behind this slit moves a sensitive surface plate or paper which receives the luminous beam, defined by the intersection of the horizontal slit with the image of the vertical slit. If all were at rest, the impression produced by this luminous beam would be reduced to a point. If the plate alone is moved, a vertical straight line will be had; if the galvanometer mirror alone turns, a horizontal line. Finally, the simultaneous displacement of the plate and mirror gives a curve whose abscissae represent temperatures, and whose ordi- nates, time. The illumination of the slit and the motion of the sensitive surface may be realized in many different ways. Regarding the lighting of the slit, there are two quite distinct cases to consider, that of laboratory researches by rapid heating or cooling, which last only a few minutes, and that of continuous recording of temperatures in industrial works, which may last hours and days, that is to say, periods 100 times to 1000 times longer. The rate of displacement of the sensitive surface, and consequently the time of exposure to the luminous action, may vary in the same ratio. The luminous source necessary will be therefore quite different, depending upon the case. For very slow displacements it is sufficient to use a small kerosene lamp with a flame of 5 to 10 mm. high. For more rapid displacements 402 HIGH TEMPERATURES use may be made of an ordinary oil lamp, an Auer burner, or an incandescent lamp; finally, for very rapid displacements of the sensitive plate, 10 mm. to 100 mm. per minute, one may advan- tageously employ the oxyhydrogen flame or the electric arc. For oxyhydrogen light the most convenient is the lamp of Dr. Roux, with magnesium spheres ; it consumes little gas and is inclosed in a metallic box which prevents all troublesome diffusions of the light. In more modern apparatus the Nernst lamp is often used. The electric arc gives much more light than is needed, and the rapid wearing away of the carbon, by displacing the positions of the luminous point, renders difficult the permanence of suitable illumination of the slit. For very short experiments one may very conveniently use the mercury lamp in vacuo (Fig. 144) or the arc playing between two mercury surfaces. In order to run it, 3 amperes at 30 volts are requisite. Its only disadvantage is its liability to go out after running a few minutes on account of the evaporation of the mercury in the central tube. It suffices, it is true, to give it a slight jar to make it go again, by causing a small quan- tity of mercury to pass from the outside annular space into the central tube. Special forms of mercury lamp exist, however, which are free from this trouble. Whatever the luminous source employed, the slit may be always lighted by means of a lens arranged as was indicated for discontinuous recording, that is, projecting upon the galvanometer mirror the image of the luminous source. When this is large enough, it suffices to place the slit before the luminous source, bringing it up close enough so as to be sure that some of the luminous rays, passing through, fall upon the mirror. But there is danger here of so considerably heating the slit that it may be altered: for this reason one is led to use more voluminous light sources than would otherwise be necessary. In the case of the use of a lens, the useful luminous intensity is as great as in placing. Fig. 144- Mercury Lamp. RECORDING PYROMETERS 403 the slit immediately next to the luminous source, so long as the image of the latter is greater than the galvanometer mirror; now with the ordinary dimensions of the sources employed this condition is always fulfilled without any special precaution. Instead of a slit lighted by a distinct luminous source, use may be made of a platinum wire, or better, as does Charpy, employ a carbon filament of an incandescent lamp heated by an electric current. In order that the line traced by the recorder be very fine, it is necessary that the two slits, the luminous slit and the horizontal slit, be equally fine. Skillful mechanicians can cut such slits in metals. But it is easier to make them by taking a photographic plate of bromide gelatine that has been exposed to the light, de- veloping until completely black, then wash and dry. By cutting the gelatine with the point of a penknife guided by a ruler, one may get transparent slits of a perfect fineness and sharpness. For sensitive surfaces, use is made of plates or films of bromide gelatine. Professor Roberts- Austen employed exclusively plates which permit more easily the printing of a great number of posi- tive proofs. Charpy, in his researches on the hardening of steel, made ,use of sensitive paper, which permits a much more simple installation. For industrial recording, paper would allow of the employing large rolls lasting several days, as in the recording magnetic apparatus of Mascart. But in general one wants to have quickly the results of the record; this is always the case in laboratory investigations, and almost always in industrial studies. It is thus preferable to be content with quite short bands of paper rolled on a cylinder. "There exists such a model quite well known and easy to use: the recording cylinders with an interior clock movement of the firm Richard, Paris. They may be ordered from the maker with any desired rate of rotation; unfortunately, this rate cannot be changed at the pleasure of the operator, a desideratum in laboratory investigations. In the apparatus used by Charpy, or in its very elaborate form as constructed by Toepfer of Potsdam, for Kurnakow, the ver-. 404 HIGH TEMPERATURES tically moving plate is replaced by a rotating cylinder wound with the sensitized paper on which the deflections of the galva- nometer are registered. This form of recorder had also been used and discarded by Roberts-Austen. Fig. 145 represents the installation of the recording pyrometer used by Charpy in his researches on the quenching of steel. To the right is the galva- nometer, to the left the Richard recording cylinder, and in the middle the electric furnace used for heating the samples of steel. It is interesting to note in passing that Charpy was the first to use electric heating in this kind of work. Kurnakow's apparatus, which must be placed in a dark room, is furnished with an Fig. 145. Charpy 's Apparatus. auxiliary telescope and scale system using red light, so that the experiment may be controlled during the taking of a record. As constructed, five speeds may be given to the cylinder; and there is provided an E.M.F. compensating system for main- taining the maximum sensibility over a Series of temperature ranges. There is another device, used by C. L. A. Schmidt, by which the experiment may be watched while a photographic record of a cooling curve is being taken. It consists in shunting the sensi- tive photo-recording galvanometer G (Fig. 146), in series with a high resistance R, across a direct-reading milli voltmeter V. If the resistance of R + G is great compared with that of V, the RECORDING PYROMETERS 405 readings of the millivoltmeter will not be altered appreciably by this operation. Schmidt moves the photographic plate, mounted as in the apparatus of Roberts-Austen, by means of a screw driven by a small motor. In this way any desired speed may be given to the plate. R Fig. 146. Schmidt's Device. If plates are used, they may be placed in a movable frame regulated by a clock movement; this is the first arrangement employed by Professor Roberts- Austen (Fig. 147). But this in- stallation, somewhat costly and complicated, has the same dis- advantage as the recording cylinders, in that but a single speed can be given to the sensitive surface. In order to drive the plate > Fig. 147. Apparatus of Roberts-Austen. Roberts-Austen later used a buoyed system in which the rate of rise of level of the water is controlled at will by the agency of a Mariotte's flask and a simple water cock. The plate is kept in an invariable vertical plane by means of two lateral cleats whose friction is negligible on- account of the mobility of the float. The sketch (Fig. 148) gives the arrangement of a similar appa- 4 o6 HIGH TEMPERATURES ratus made by Pellin for the laboratory of the College de France. It carries a 13 by 18 cm. plate which is attached to the float by means of two lateral springs not shown in the sketch. Neither are the two guides of the float, immersed in water, indicated ; the play next the cleats is only two-tenths of a millimeter. The uncer- tainty that this play can cause in the position of the plate is quite negligible. The curve (Fig. 149) is the reproduction of an experiment made with such an arrangement by Roberts-Austen on the solidification of gold. During the whole period of freezing the temperature remained stationary, then lowering of temperature was produced at a regularly decreasing rate as the temperature of the metal approached that of the surroundings. 1065 C. I ' ' r 3- ft - -1 i U I/ Fig. 148. Plate-holder. 12C. Fig. 149. Record with Apparatus of Roberts-Austen. It is indispensable to trace, on each sensitive surface on which is to be recorded a curve, the line corresponding to the surround- ing temperature, or at least a parallel reference line. This is very easy in the case of the guided plate or of the paper rolled on a cylinder. It suffices, after having' brought the couple to the temperature of its surroundings, to displace in the opposite RECORDING PYROMETERS 407 direction the sensitive surface; the second curve traced during this inverse movement is precisely the line of the zero of the graduation of the temperatures. But this is a dependence that may be evaded by registering at the same time as the curve a reference line by means of a fixed mirror attached to the galva- nometer in the path of the luminous beam which lights the mov- able mirror. Roberts-Austen likewise made use of the luminous beam reflected by the fixed mirror to inscribe the time in a precise manner. A movable screen driven by a second pendulum cuts off at equal intervals of time this second luminous beam. The reference line, instead of being continuous, is made up of a series of discontinuous marks whose successively corresponding parts are at intervals of one second, as is shown in Fig. 149. The curves once obtained must be very carefully examined to recognize the points where the gradient presents slight anomalies, characteristic of the transformations of the body studied. Gen- erally these irregularities are very insignificant, and it would be well, in order to recognize them with certainty, to obtain curves traced on a much greater scale. Practically this magnification is not possible without auxiliary devices which limit either the range or the sensibility; thus the sensitiveness of the galvanom- eter may be increased, and so the deflection, but then for the greater range of temperature the luminous image would fall off the sensitive plate. In practice it has been found difficult to realize conveniently a sufficiently steady motion of the plate in the Roberts-Austen system of recording, and attempts have been made to devise methods in which the photographic plate remains fixed in posi- tion. This has been successfully accomplished by Saladin, whose apparatus (Fig. 160, page 419) has been modified by Wologdine to give the temperature-time curve by removing the prism M and substituting for the second galvanometer GI a plane mirror turning about an horizontal axis. This mirror may be controlled by an hydraulic system as in Roberts-Austen's apparatus, or by clockwork as in the model constructed by Pellin of Paris. The deflection of the galvanometer G\ gives to the beam of light an 408 HIGH TEMPERATURES horizontal motion over the plate proportional to the tempera ture r while the vertical motion of the beam of light is given by the mirror turning at a uniform rate, and is therefore approximately proportional to the time as registered on a flat plate. Autographic Recorders. To obtain a satisfactory autographic or pen record with platinum thermocouples without sacrifice of sensibility of the galvanometer, it is necessary to eliminate the friction of the pen or stylus upon the paper. This has been accomplished by the use of mechanisms which cause the pen or stylus at the end of the galvanometer boom to make only momen- tary contact with the moving paper.* Fig. 150. Siemens and Halske Recorder. In the Siemens and Halske form of instrument (Figs. 150 and 151), the paper P is driven forward by the same clockwork that controls the pressing down, by means of the arm B, of the stylus N, which imprints dots periodically on the paper by means of a typewriter ribbon running across and beneath the record sheet. This system permits of taking a record continuously over very * There are a considerable number of thermoelectric recorders. Among the manufacturers of these instruments are: Siemens and Halske, Berlin; Hartmannand Braun, Frankfort a. M.; Pellin, Chauvin and Arnoux, Carpen- tier, and Richard, Paris; Leeds and Northrup, the Thwing Instrument Com- pany, and Queen of Philadelphia; the Scientific Instrument Company of Cambridge, England, and Rochester, N. Y.; the Bristol Company, Water- bury, Conn. RECORDING PYROMETERS 409 long periods of time. In mos<: of the other recorders the paper is wound upon a drum, and various devices are used to obtain the record ; thus in the Hartmann and Braun type a silver stylus makes sulphide dots on a prepared paper, and in the Cambridge thread recorder rectangular coordinates are obtained by having the galvanometer boom strike an inked thread which rui parallel to the drum (Fig. 152). Fig. 151. Principle of Recorder. A Siemens and Halske drum recorder with pivot galvanometer, suitable for technical work, and which is inclosed in a dustproof metallic case, is shown in Fig. 153. It may be adjusted for seven-day records. As previously stated, these autographic instruments all give intermittent records for the platinum thermocouples, and are limited to one or two speeds; and although they may be made very sensitive they are not adapted for the detection of trans- formations which take place very rapidly, since the recording 4io HIGH TEMPERATURES Fig. 152. Thread Recorder. Fig. 153. Drum Recorder. RECORDING PYROMETERS 411 interval cannot readily be shortened much below 10 seconds, and in most instruments this interval is greater than 15 seconds. In other words, they can be used advantageously only for slow cooling or heating. A continuous pen record may be obtained with galvanometers suited for use with the base-metal couples developing high E.M.F.'s, such as the Bristol, Hoskins, Thwrng, etc. In order to eliminate the effect of irregularity of outside con- ditions which influence the rate of cooling, a method commonly Time Fig. 154. Furnace and Charge Temperature Curves. used when endeavoring to detect small transformations consists in placing a second thermocouple in the furnace, but sufficiently removed from the substance studied to be uninfluenced by its behavior. Alternate readings on the temperature of the test piece (0) and of the furnace (0') are then taken, preferably at definite time intervals. The data are most readily discussed by plotting the two temperature-time curves side by side as shown in Fig. 154, or by plotting the difference in temperature 6' against the temperature of the test piece. This method may be made recording either by using two in- 412 HIGH TEMPERATURES struments or by modifying one of the above-mentioned auto- graphic recorders so as to trace the curves of two thermocouples on the same sheet. In practice, however, this method is usually resorted to only when great sensibility is desired, as in detecting minute internal-energy changes, when the potentiometer com- bined with the deflection galvanometer is the most sensitive and quick- working arrangement for taking the measurements. It is convenient to use thermocouples of the same composition so as Fig. 155. Brearley Curve Tracer and Accessories. to have readings of both the temperature of the sample and of the furnace given by the same potentiometer setting, and so depend upon the galvanometer deflections for measuring the residual parts of 6 and d' '. ' Regarding the precision of this method, it is to be noted that the quantity it is really desired to measure is 6 6' in terms of 6, and this is accomplished by measuring 6 and 0', hence the sensi- bility of 6 6' is no greater than that of 6 or 0'. In other words, the method requires the maximum refinement of measurement RECORDING PYROMETERS 413 to obtain the quantity sought, as well as the maximum of com- putation or plotting to reduce the observations. Semi-automatic Recording. The Brearley curve tracer, man- ufactured by the Cambridge Company, is a semi-automatic ap- paratus for registering the time-temperature curve. As shown with accessories in Fig. 155, a small tube furnace is connected to an electric supply main so as to heat the specimen within the fur- nace; a platinum-iridium couple, whose hot junction is within the specimen, is connected in series with a resistance and moving coil galvanometer of adjustable sensibility; a Nernst lamp furnishes -5" -4" -3" -2" Fig. 156. Curves with Brearley Apparatus. illumination and gives a sharp image at M on the scale G. The ro- tating drum L is surmounted by a sliding carriage N carrying two pointers, one of which, M, is fixed, and the second, immediately below, carries a pen and is depressed every second on the paper wound on the drum. The pointer M is made to fojlow the spot of light by the operator by turning a handle at the end of a long screw on which N runs. There is electromagnetic clock control of the drum and of the pen. The record is. therefore a series of dots one second apart. The temperature scale may be made as open as desired, and a complete heating and cooling curve for a steel sample may be obtained in a few minutes. This instru- 414 HIGH TEMPERATURES ment is also now made to give a continuous record. A sample curve is shown in Fig. 156. Another method of working, in which the apparatus is com- pletely autographic for relatively short temperature intervals, and at the same time very sensitive, is to use a recording galva- nometer in connection with a potentiometer. This requires the operator to adjust the dials of the potentiometer to step from one temperature interval to the next, these intervals varying in length with the galvanometer sensibility, which should be capable of adjustment to give longer or shorter temperature intervals. Differential Curves. The method of page 411 may readily be modified so as to give 6 0', the difference in temperature be- tween the test piece and furnace, by direct measurement instead of by computation, with the added advantage that the precision Fig. 157. Method of Burgess. of B 0' may be made very great as compared with that of 0, the temperature of the sample. This may be accomplished, for example, by placing a commutator, which may be driven by a clock mechanism, in the thermocouple circuit at A, Fig. 157, so that alternate measurements on and 0' may be taken in terms of the time. Evidently the connections may be made so that either the galvanometer G 2 of the same direct-reading or potentiometer system that measures 0, or a separate instrument GI, as shown in the figure, may be used to measure 0' '. Both galvanometers may be photographic or autographic recorders. Use of a Neutral Body. Accidental variations in the indica- tions of the auxiliary thermocouple giving 0', the furnace tem- perature, may largely be eliminated by placing this couple within a blank or neutral substance. The material of the neutral body should be such that it undergoes no transformations involving RECORDING PYROMETERS 415 an absorption or evolution of heat within the temperature range studied, such as a piece of platinum, porcelain, or even in some cases nickel or nickel steel. It is also desirable that the sample and neutral have as near as may be the same heat capacities and emissivities. The sample to be studied and the neutral piece are placed near together and arranged symmetrically with respect to the temperature distribution within the furnace. To Roberts-Austen again was due the credit of first devising a sensitive differential method using the neutral body. He also modified his photographic recorder (Fig. 147) so as to give, by means of a second galvanometer, the 6 0' vs t curve on the same plate with the vs t curve, from which a curve giving 6 B f in Fig. 158. Use of Neutral, Roberts- Austen. terms of could be constructed. His arrangement of the direct- reading and differential thermocouple and galvanometer circuits is shown in Fig. 158 in which S is the sample or test piece, and N the neutral body possessing no transformations; the galvanometer G 2 measures the temperature 6 of the sample/and GI measures the difference in temperature 6' between the sample and the neutral. Curves for steels and alloys were usually taken with the samples in vacuo. It is evident that Roberts-Austen's final photographic appa- ratus, although very sensitive, was also complicated and very delicate of adjustment, and in practice it took great skill in its use, requiring for instance some three or four successive exposures adjusted to the proper adjacent temperature ranges, to take the cooling curve of a steel from 1100 to 200 C. 4i6 HIGH TEMPERATURES Most of the recent exact work employing the principle of this method has been done by taking the observations of 6 directly on a potentiometer and 6 0' on the same or an auxiliary galva- nometer. In this case of direct reading, the simpler arrangement of thermocouples indicated in Fig. 157, due to Burgess, may ad- vantageously replace Roberts- Austen's (Fig. 158), or the modifica- tion shown in Fig. 159, such as used by Carpenter and others. The first dispenses with one thermocouple and the drilling of a second hole in the sample. This method is evidently capable of attaining maximum sen- sitiveness, since the galvanometer connected to the differential thermocouple, giving 9 0' vs t, may be made as sensitive as c c Fig. 159. Arrangement used by Carpenter. desired independently of the 6 vs t system. There is the further advantage that no limits are set to the range of temperatures over which a given precision in d' may be had. There is, however, a limitation on the certainty of interpretation of results by this method, especially when the rate of cooling is rapid, due to the fact that it is practically impossible to realize the ideal condition of having 6 tf = a constant, or keeping the cooling curves of the test piece and neutral parallel for temperature intervals within which there are no transformations of the test piece. The rate of cooling, and hence the value of 8 d', is influenced by several factors, among the most important of which are the mass of each substance, the unknown and the neutral, its specific heat, conductivity, and emissivity, as well as the relative heat capacities of the furnace and inclosed samples. The tf vs t RECORDING PYROMETERS 417 line is, however, always a smooth curve, except for the regions in which there are transformations in the substance under study. The autographic system of recording may also be used, and it is possible to construct an apparatus by means of which both the vs t and 6 d' vs t curves shall be recorded simultaneously on the same sheet by the same galvanometer boom. In order to accom- plish this, we have made use of a Siemens and Halske recording millivoltmeter having a total range of 1.5 millivolts and a resist- ance of 10.6 ohms. The E.M.F. generated by the differential thermocouple, proportional to 6 0', is recorded directly by this instrument. i C. corresponds to from 16 to 19 microvolts between 300 and 1100 C. for a platinum-iridium couple, or to about 1.8 mm. on the record paper. In series with the Pt-Ir thermocouple giving temperatures is a suitable resistance, about 200 ohms in this case, so that the galvanometer boom may be kept within the limits of the paper when recording values of 9. The circuit is made alternately through the direct and the differ- ential thermocouple circuits in series with the recorder by means of a polarized relay actuated by the same battery that depresses the galvanometer boom when the mark is made on the paper. The thermocouple circuits may be those of either Figs. 157, 158, or 159, but with the galvanometer G 2 indicating temperatures suppressed. It is evident that by recording the two curves, 6 f vs t and vs t, on the same sheet there is some sacrifice in the ability to de- tect small and rapid transformations, since the spacing is doubled. Usually also, with such an arrangement, the galvanometer will not be completely aperiodic for one or the other system. On the other hand, it is of great advantage to have the curves to- gether and obtained independently of inequalities in clock rates, which are a serious source of error in locating transformation points exactly when two separate instruments are used. The same result may be effected by shunting the galvanometer when on the temperature side. This of course cuts down very greatly the resistance of the thermocouple circuit, a disadvantage unless a sensitive galvanometer of high resistance is used. Such gal- 418 HIGH TEMPERATURES variometers suitable for mechanical recording are not yet avail- able. In Thwing's recording pyrometer, two galvanometers, one giving temperatures and the other differences, impress their records on a single chart driven by one clock. When it is desired merely to detect the existence of a transfor- mation without measuring its temperature exactly, the sensitive form of recording millivoltmeter may be connected directly to the differential thermocouple without other accessories, as was done by Hoffmann and Rothe in studying the transformations of liquid sulphur. Salaam's Apparatus. It is sometimes of advantage to be able to record and discuss the data independently of the time, and so express 6 8', the difference in temperature between sample and neutral, directly in terms of 0, the temperature of the sample. This may evidently be accomplished by replotting the results obtained from the curves of the previous differential methods which involve the time. It was reserved, however, to Saladin, engineer of the Creusot Works, to invent, in 1903, a method that would record photographically the 6 vs B 0' curve directly, thus obviating any replotting. His method possesses also the advantage of having the photographic plate fixed in place. The forms of curve obtained in this way are illustrated in Fig. 134. The arrangement of the apparatus in its simplest form, due to Le Chatelier, is shown in Fig. 1 60. Light from the source S strikes the mirror of the sensitive galvanometer G\ whose deflections measure the differences in temperature (6 B'} between the sample under study and the neutral body. The horizontal deflections of the beam of light are now turned into a vertical plane by passing through the totally reflecting prism M placed at an angle of 45 degrees. A second galvanometer G 2 , whose deflec- tions are a measure of the temperature of the sample and whose mirror in its zero position is at right angles to that of Gi, reflects the beam horizontally upon the plate at P. The spot of light has thus impressed upon it two motions at right angles to each other, giving, therefore, on the plate a curve whose abscissae are ap- RECORDING PYROMETERS 419 proximately proportional to the temperature 6 of the sample and whose ordinates are proportional to 6 0'. The sensitiveness of the method depends upon that of the galvanometer Gi, which may readily be made to give 5 or 6 mm. for each degree cen- Pt. Rh. Fig. 1 60. Saladin's Apparatus. tigrade. The arrangement of the thermocouple circuits is the same as in Figs. 158 or 159. If so desired, the time may also be recorded by means of a toothed wheel driven by a clock and placed in the path of the beam of light. Compact forms of this apparatus, which is used considerably in metallurgical labora- 420 HIGH TEMPERATURES tories, are made by Pellin, Paris, and by Siemens and Halske, Berlin. The lens between G\ and G 2 may be suppressed. When steels and metallic alloys in the solid state are being investigated, advantage may be taken of the thermoelectric behavior of the sample itself to record the critical regions with Saladin's apparatus. Thus Boudouard measures 0' by means of platinum wires set into crevices at each end of the sample, taking advantage of the fact that the transformation will usually be progressive along the sample. This modification eliminates the neutral piece and one platinum or alloy wire, but, as Le Chatelier has shown, is less accurate than the usual form of Saladin's apparatus; and its indications may even be indeter- minate or ambiguous, as the reaction may start midway between the embedded wires or at either end. Saladin's method, it should be noted, is a perfectly general one for recording the relations between any two phenomena which may be measured in terms of E.M.F. or as the deflections of two galvanometers. The Leeds and Northrup Company have recently modified their autographic recorder, p. 393, to trace the 6 vs 6 &' curve, using several differential couples in series in order to obtain the required sensibility. Registration of Rapid Cooling. None of the experimental arrangements so far described is adapted for measuring the very rapid cooling, i.e., several hundred degrees in a few seconds, met with in such processes as quenching or chilling. The develop- ment of methods for measuring rapidly varying temperatures will undoubtedly be of great use in the solution of many physical and metallurgical problems involving products whose properties depend on cooling velocities. Only a few preliminary investi- gations into this field, however, have as yet been made. Le Chatelier's Experiments. Le Chatelier, in an investigation of the quenching of small samples of steel, and the effect of various baths, made use of a galvanometer having a period of 0.2 second and a resistance of 7 ohms, whose deflections, produced by the current from a thermocouple inserted into the specimen under- going the quenching, were recorded on a photographic plate RECORDING PYROMETERS 421 moving vertically at a speed of 3 mm. per second. A half- second's pendulum vibrating across the path of the beam of light, from a Nernst glower as source, gave a measure of the time. He succeeded in recording satisfactorily temperature intervals of 700 C. in 6 seconds, using as samples cylinders 18 mm. on a side, and obtained results of great interest to the theory and practice of hardening steel samples by quenching in baths of various kinds of liquids. Le Cha teller recognized the desira- bility of increasing the precision and sensitiveness, and of improving the technique, of this method, and suggested the advantages of using for the registration an oscillograph arrange- ment, or a string galvanometer of very short period such as Enithoven's, in which the displacements of a silvered quartz fiber of high resistance in an intense magnetic field are measured photographically. Benedicks' Experiments. Following the suggestions of Le Chatelier, Benedicks has carried out a series of researches on the cooling power of liquids, on quenching velocities, and on certain constituents of steel. The errors in cooling curves of metals have also been studied recently by Hayes. Benedicks' apparatus, as arranged for taking the time-tem- perature curve of steel samples during quenching, is shown in Fig. 161. The principles here applied may evidently be used in other kinds of experimentation involving rapid cooling. The specimen A is heated in a small electric furnace B, which is provided, in its lower part, with a narrow opening parallel to the longitudinal axis, through which passes a holder C, which turns about an horizontal axis, being given a definite torque by a spiral spring D, and maintained vertical by an electromagnetic control E. Through a bore in C a thermocouple is led into the interior of A, and the cold junction is contained in the ice box F, from which the wires are led to a commutator G by means of which either the thermocouple A or the calibrat- ing apparatus b, c, etc., may be connected to the measuring instrument /, this being a small string galvanometer by Edel- mann of Munich. The light from the arc lamp K, passing 422 HIGH TEMPERATURES through the microscope of the galvanometer, fitted with a projection eyepiece, gives an image of the movable string on the registration apparatus L, provided with a rotating cylinder which carries the sensitive paper. Fig. 161. Apparatus of Benedicks. Finally, the electromagnet release E is connected with an accumulator N and a contact T on the shutter before the cylin- der L. The process of registration is therefore as follows : The cylinder L is set in rotation, and as the edge of the sensitive paper passes RECORDING PYROMETERS 423 the window T the shutter rises. At the same instant the cir- cuit of E is closed, releasing the arm C. This automatically and quickly quenches the specimen A in the cistern of- water M beneath. Precautions have to be taken in insulating the thermocouple wires leading into the specimen and in insuring good contact of the couple junction against the specimen with a water-tight joint. Capillary tubes of fused quartz, which will also stand sudden temperature changes, were used for insulating, and water was prevented from entering joints by means of compressed air introduced into the containing tube C of the thermocouple wires. The calibrating apparatus consists of a sliding commutator c, b, the blocks of which are connected to fixed points on a slide wire r in such positions as to give electromotive forces corre- sponding to definite temperatures, 400, 600, etc., of the thermo- couple when the resistances R, RI, and R 2 are properly adjusted; variations in the battery a are controlled by the standard cell and resistance R. This arrangement allows calibrating the gal- vanometer immediately before each experiment and give the calibration data on the same sheet as the quenching curve. The string galvanometer had a resistance of 6700 ohms; its sensitiveness may be adjusted to follow that of the thermocouple, although this is not necessary. The time correction of this gal- vanometer is such, fortunately, that the directly registered curve is simply a parallel curve to the one which would be obtained if the deflections were absolutely instantaneous; or in other words, no correction for the inability of the instrument to respond in- stantaneously is necessary. There remains, of course, a small unknown time correction due to the lag of the thermocouple with respect to the test specimen. In Fig. 162 are shown curves for a steel of 0.42 carbon quenched both from 850 and 950 C.; the time r is taken to 100 C. The calibration curves are also shown in the figure. Recording Radiation Pyrometers. Any phenomenon whose magnitude may be measured by the deflection of a galvanometer may be rendered self-registering by optical means. Total or 424 HIGH TEMPERATURES \ \ \ \ \ $ 3 u bo I o> g RECORDING PYROMETERS 425 monochromatic radiation falling on the exposed strip of a bolom- eter may, therefore, be made to record its intensity, which is, as we have seen, a function of the temperature of the radiating body. Langley, in 1892, rendered his bolometer a recording instrument, the records being taken photographically. This system of recording has been used mainly for the mapping of solar spectra, and incidentally for the estimation of the sun's, temperature and in other astrophysical investigations; and al- though it might be used in laboratory investigations in record- ing high temperatures in terms of either total or monochromatic radiation, it has not come into any general use for such purposes. The experimental arrangements are necessarily very elaborate H 5 6 78 9 10 11 Noon 1 2 3 4 5 ft Fig. 163. Solar Radiation Record. and delicate, for descriptions of which the reader should consult the Annals of the Astrophysical Observatory of the Smithsonian Institution. Callendar has applied also his slide-wire method of recording electrical resistances to Langley's bolometer. The curve of Fig. 163 gives the record of solar radiation for a day. The radiation pyrometers of the Fery type are readily made self-registering, it being only necessary to substitute for the indi- cating galvanometer a suitable deflection-recording instrument, such as the Cambridge thread recorder or a Siemens and Halske recording milli voltmeter of the required range and sensitiveness. In Fig. 164 is shown the record of the temperature of a pottery " biscuit " kiln as taken with a Fery radiation pyrometer and Cambridge thread recorder. A Callendar slide-wire recorder 426 HIGH TEMPERATURES could also be used to register very high temperatures if too great sensitiveness be not demanded. The Morse or Holborn and Kurlbaum instruments may be made semi-recording; that is, a registering ammeter may be put in the lamp circuit and made to record each temperature to which the pyrometer is set by the observer. This method would have some advantages in the control of those industrial operations for which this type of pyrometer is best adapted. c 1300 nool 1000' 900 C 800 C 700 1 8 9 10 11 Midnight 12345678 Fig. 164. Temperature Record of Pottery Kiln. Recording Accessories. We may mention, finally, a number of auxiliary pieces of apparatus and methods which are useful in special cases. Range Control. It is sometimes desirable to limit the range of the recorder to some restricted temperature interval and thereby gain greater sensibility with a more open temperature scale. This may be done in several ways. We shall use as illustrations the scale-control box of Peake as applied by the Cambridge Company to their thread recorder. In its more complete form this device is shown in Fig. 165, for use with thermocouples provided with Peake's compensating leads (page 176). RECORDING PYROMETERS 427 A 6-volt accumulator passes a current through a series of fixed resistances, RZ, RI, R 7 , and a portion of a variable resistance R 2 , the potentiometer circuit. A second circuit, the pyrometer circuit, consisting of the couple, leads, R& and R 5 , and the recorder, is connected to tap onto the ends of the coil R* in the potentiometer circuit. Thus the poten- tial drop in RI, due to the current from the accumulator, is opposed to the electromotive force of the couple, and, therefore, at some particular temperature, say 750 C, the two just balance, and no current will flow through the pyrometer circuit. If now the Accumulator hennocouple Switches 81 i 82 shown in running position; dotted lines show test petition Fig. 165. Peake's Scale Control Box. temperature of the couple falls, a current will flow in one direction through the recorder, whilst if it rises a current will flow in the reverse direction. Thus the zero or undeflected position of the recorder pointer may be made in the center of the scale, and will correspond in the above case to 750 C., whilst the resistance R$ may be so adjusted that one end of the scale will correspond to 600 C., and the other to 900 C. The accuracy of the arrangement depends upon the current in R* being maintained constant, and to secure this a Clark cell is connected across the resistances R 3 and R*. When the accumu- lator voltage is normal, this cell does not give any current, but if 428 HIGH TEMPERATURES accumulator voltage falls slightly, the Clark cell gives a slight current and tends to keep the voltage across the terminals of R z and RI nearly constant. The change in E.M.F. of the Clark cell with temperature also balances very nearly the corresponding changes in the compen- sating leads, so that the behavior of the apparatus is nearly independent of cold-junction temperature fluctuations. The arrangement may be simplified but rendered less exact by dis- pensing with the Clark cell. Automatic Commutator Pyrometer Galvanometer Hand Commutator Recording Galvanometer Fig. 1 66. Recorder and Indicator with Four Thermocouples. Multiple Records and Circuits. There are various devices for taking several records on a single sheet by means of one gal- vanometer. They practically all reduce to some type of auto- matically driven commutator and are often so constructed that the several records may be distinguished by the spacing or length of dots and dashes. While it is proper to record simul- taneously quite different temperatures, it is usually good practice not to try and so record temperatures that frequently overlap RECORDING PYROMETERS 429 on the sheet, as its interpretation may then become doubtful. It is also often convenient to have on one circuit a recorder, which may be in an office, together with one or more indicating instruments. An arrangement of four thermocouple circuits is shown in Fig. 166, whereby a recorder gives a continuous record for all the couples, and the temperature reading of each of them may also be taken by means of an indicating galvanometer. Thermometers Fig. 167. Recorder and Indicator with Four Resistance Pyrometers. A switchboard may also be used permitting interchangeability of pyrometers, as illustrated in Fig. 167, where four resistance- thermometer circuits are provided for, to be used with various instruments as required. See also Fig. 70. In general, it may be said that almost any industrial require- ment of combination of pyrometer circuits, recording and indi- cating instruments and alarms may be solved satisfactorily in practice. 430 HIGH TEMPERATURES Furnace Control and Thermostats. In certain operations, for example, in taking heating and cooling curves, it is advanta- geous to raise or lower the temperature of the electric furnace continuously at a uniform rate. This is readily accomplished for an alternate-current supply by the use of a salt-water rheostat fed on heating from a Mario tte bottle. The metal electrodes may be cut to shape to favor the uniformity of rise in tempera- ture; during the cooling the water is siphoned off. The whole apparatus may be made completely automatic, if desired, so that a series of heating and cooling curves may be taken at any desired rate without the intervention of the observer. It is well to keep the temperature of the rheostat down by circulating water through it in a coil of pipe. It is often desirable to maintain a furnace at constant tem- perature. The method used will depend largely upon the tem- perature in question. In the range over which liquid baths may be used, to 350 C. with suitable oils, they are satisfactory when properly stirred and provided with thermostatic control. With a sensitive gas regulator a constancy of 0.05 C. may be maintained, and with electric control a somewhat better uniformity. A uniform temperature over a large volume may be estab- lished by means of a vapor in equilibrium with its liquid. This system is not available for high temperatures, and it is difficult to maintain a constant temperature over long periods of time. For high temperatures air baths only can be employed, the most usual form being the electric resistance tube furnace. Special windings and delicate control are required if it is desired to maintain a considerable volume at constant temperature. Various devices have been suggested for the automatic control of furnace temperatures, based usually on the use of relays actuated either electrically or optically. Most of the recorders we have described may be fitted with such an accessory. We may also cite the optical regulator of Kolowrat which will keep an electric furnace constant to 2 or 3 at 1000 C., and may be applied to either thermoelectric or resistance measurement of RECORDING PYROMETERS 43 ! temperature. The light from a powerful source, a Nernst lamp, is reflected from the galvanometer mirror onto a scale repre- senting temperatures. When in adjustment an increase in tem- perature of the furnace throws the spot of light onto a thermopile which operates a series of relays cutting in resistance to the heat- ing circuit and cooling the furnace slightly. This resistance is cut out when the spot leaves the ^ thermopile. Among the many electric ther- mostatic controls we may men- tion that of H. Darwin, which may also be used as an alarm (Fig. 1 68). When the galvanom- eter needle GV is deflected from the stop F, due to a rise in temperature, the needle engages the wheel W driven by clock- work and makes a circuit at L, which may be either that of an alarm, as shown, or that of a regulating circuit by means of relays. Fig. 168. Darwin Temperature Alarm. CHAPTER XI. STANDARDIZATION OF PYROMETERS. Thermometric Scales. The generally recognized standard temperature scale is that of the gas thermometer, which, as we have seen, has been realized in the form of the constant-volume nitrogen thermometer to 1550 C. This scale is fixed by the determination of certain reference temperatures, such as melting or freezing and boiling points. It would be desirable to define temperatures in terms of the normal or thermodynamic scale, which is independent of the properties of any particular substance. At the present time, however, the limit of accuracy attained in gas pyrometry does not exceed the departure of the constant- volume gas scale from the thermodynamic scale; and the scale as defined by various gases is also practically identical, so that for most practical purposes we may speak in terms of either scale interchangeably. Above the range of the gas thermometer, we are compelled to resort to extrapolation in terms of some phenomenon varying with the temperature. For this purpose, use is usually made of the radiation laws based on the relations which have been found to exist at lower temperatures between the intensity of total and monochromatic radiation and temperature. Just as the thermodynamic scale of temperature is independent of the thermal properties of any particular substance, but would be reproduced exactly by an ideal gas, and is very nearly realized by the thermal properties of ordinary gases : similarly, the radia- tion scale of temperature is independent of the radiating prop- erties of any particular substance, but would be reproduced exactly by the radiation from a black body, and is very nearly realized by the radiation from an almost completely closed, clear furnace at a uniform temperature. The radiation scale, then, 43 2 STANDARDIZATION OF PYROMETERS 433 may be, and in practice is, so denned as to be the thermodynamic scale, so that we have in reality a single, continuous-temperature scale from the lowest to the highest attainable temperatures. Unfortunately, there is not as yet a sufficiently good agree- ment among the few temperatures above 1200 C. determined with the gas thermometer, so that there is still considerable uncertainty in the values to assign to the constants in the radia- tion laws and therefore to fixed points in the higher ranges. Fixed Points. As the scale determined by the gas thermom- eter is the one universally recognized, it is necessary, in order to calibrate a pyrometer, to express its indications in terms of the gas scale. In general, it is not feasible to compare the readings of a pyrometer directly with those of the gas thermometer. The use of the latter becomes restricted mainly to the establishment of certain constant, reproducible temperatures or fixed points such as are given by freezing points and boiling points of the chemical elements and of certain compounds. The accuracy at- tainable in pyrometric researches is, therefore, limited by the exactness of our knowledge of these reference temperatures, and their determination has been and still is of the most fundamental importance in pyrometry. There have been a great many tem- jratures suggested for this use, but the actual number available is very small. Preference should, in general, be given to those deter- minations made with the gas thermometer itself, although there are others made indirectly in terms of the gas scale, as with thermocouples, optical pyrometers, and resistance thermometers, which are of considerable weight; and in fact the more common practice, when working with the gas thermometer, is to compare its readings in a furnace or bath with those of some more con- venient instrument and then transfer the gas scale by means of the latter to the melting or boiling points by interpolation. We have already called attention to many of these determi- nations of fixed points, among which the following may be con- sidered in greater detail: Sulphur. (Boiling) 444.6 C. on the constant- volume scale of nitrogen, or 444.5 on the constant-pressure scale; correspond- 434 HIGH TEMPERATURES ing to about 444.7 on the thermodynamic scale, under a pressure of 760 mm.; with a variation of 0.090 per millimeter change of mercury in the atmospheric pressure. The boiling point of sulphur has been the object of several series of distinct observations, among which we may cite the following, distinguishing between direct and strictly independent determinations with the gas thermometer and indirect ones by italicizing the former. BOILING POINT OF SULPHUR. cop Corr. to Observers. Method and remarks. observed const, vol. />o=i at. Regnault Const, vol., about 447-5 Crafts Const, vol 445 .... Callendar and Griffiths Const, press 444 . 53 444 . 74* Reichsanstalt Const, vol. Wiebe and Botcher scale 444 . 5 444-5 Chappuis and Harker . Const, vol. corr. from 445 . 2 . . 444 . 7 444 . 7 Holborn Const, vol. extrapolated Pt. resistance 444-55 444-55 Rothe Const, vol. Hg thermo. P.T.R. scale 444.7 444-8 Thermocouples 445 . o .... Eumorfopoulos Const, press 444-55 444-76 Holborn and Henriing Const, vol 444-51 444-51 Best value from above series . . . . 444. 6 Regnault's figure was obtained by plunging the reservoir of the thermometer in the liquid sulphur; but this liquid will super- heat, and so gives too high a value. The other eight very con- cordant results were obtained in the vapor. The result first published by Chappuis and Harker, using a constant- volume thermometer, was 445.2, but this difference from Callendar and Griffiths' result was shown probably to be due mainly to an incorrect value assumed for the expansion co- efficient of the porcelain bulbs used by the former. Eumorfo- poulos first published the value 443.7, which was recognized at the time to be uncertain, as it depended upon the unknown ex- pansion coefficient of mercury, which has since been determined by Callendar and Moss to high temperatures. STANDARDIZATION OF PYROMETERS 435 Callendar and Griffiths as well as Eumorfopoulos worked with a constant-pressure air thermometer, and it is of interest to note that the outstanding difference between several of the experi- mental determinations by the constant-volume and constant- pressure methods is of the order of difference to be expected between the two gas scales, constant- volume and constant- pressure, as seen from Callendar 's table (page 31) and from Fig. i . In fact, in work of the highest precision it will probably soon be desirable to reduce observations to the thermodynamic scale. At the Reichsanstalt a new determination of the S.B.P. has recently been carried out by Holborn and Henning, using sev- eral gases and bulbs of glass and quartz. Their result is about 0.2 C. lower than would be expected from the measurements at constant pressure. In order to reproduce the exact value of the sulphur boiling point, however, it is not sufficient to plunge the protected ther- mometer into the sulphur vapor, but it is necessary to guard it against superheating by radiation from the liquid and lower walls, on the one hand, and cooling by liquid sulphur condensed on the thermometer case and radiation from the thermometer, on the other hand. Unless proper precautions are taken, varia- tions of i C. may be found. Sulphur boils very smoothly with- out bumping, and, in a properly constructed apparatus, condenses in a very sharp line near the top of the boiling tube. A conical or cylindrical aluminium shield with an umbrella cap fitting close about the thermometer stem serves the double purpose of shield- ing the instrument from radiation and condensed sulphur. A sulphur boiling apparatus with the protected thermometer in place is shown in Fig. 169, with which measurements consistent to about 0.03 may be obtained. Gas or electric heating may be used, and the boiling tubes may be of hard glass, porcelain, or aluminium. A study by Waidner and Burgess of the various forms of sulphur apparatus used by previous experimenters showed that they give the same temperature to a few hundred ths of a degree. Commercial sulphur gives the same boiling point 436 HIGH TEMPERATURES .Asbestos Aluminum ELECTRICALLY HEATED 8.B.P. APPARATUS GAS HEATED 8. B. P. APPARATUS Fig. 169. Types of Sulphur Boiling Apparatus. STANDARDIZATION OF PYROMETERS 437 as *he best sulphur obtainable. A criterium of satisfactory realization of the S.B.P. is the constancy of reading when a ther- mometer with accessories is displaced several centimeters in the vapor. Waidner and Burgess have also shown that, measured in this way, the column of vapor above boiling sulphur is con- stant to about 0.03 C. In spite of the most excellent agreement of the observations in the above table, the determinations with the hydrogen ther- mometer by Jaquerod and Wassmer of the boiling points of naphthalene and benzophenone, and those by Day and Sosman with the nitrogen thermometer of the freezing points of zinc and cadmium, are not consistent with the value cited above for sulphur. As shown by Waidner and Burgess, using the platinum thermometer, the sulphur point as quoted would be nearly one degree too high in terms of the work of the observers mentioned above. In view of the almost universal use of the sulphur point as a calibration temperature, it is of prime importance to finally fix its value to at least better than 0.1 C. The several determinations of the change of boiling point of sulphur with pressure are in very close agreement. For exact work, the two- term formula of Holborn and Henning, or that of Harker and Sexton, is to be preferred. t = /?6o + 0.0912 (H 760) 0.0442 (H 760) 2 . Zinc. (Freezing or melting) 419.4 C. Freezing points un- dergo unappreciable changes with variations in atmospheric pres- sure, and their experimental determination is somewhat easier than for boiling points if a thermocouple is used. The direct determination of a metallic freezing or melting point with a gas thermometer is beset with almost insurmountable experimental difficulties, so recourse is always had to some auxiliary pyrometer whose indications have been exactly calibrated by direct com- parisons with a gas thermometer. Zinc is easily obtained in sufficient purity. Some recent de- terminations of this point are: 438 HIGH TEMPERATURES Heycock and Neville 419.4* Stansfield 418.2 Holborn and Day 41 9 . o Day and Sosman 418.2 Waidner and Burgess 419 .37 Holborn and Henning 4 19 . 40 The first and next to the last values were obtained with the resistance pyrometer, assuming the value for the S.B.P., 444.70; Stansfield's observation was obtained with a record- ing thermocouple, and the other is by direct transfer with thermocouples or resistance thermometers from the nitrogen- gas thermometer. Zinc. (Boiling) 920 C., with a variation of 0.15 for a change of i mm. in the atmospheric pressure. The boiling point of zinc has been the object of a great many determinations, and yet it is one of the least known and con- sequently the most unreliable to try to use, and is not to be recommended. It has been the object of so much study, un- doubtedly, as it was apparently the one point near the upper limit of the early experiments with the gas thermometer which could be determined directly by this instrument; but super- heating effects in vapors at such high temperatures and an un- even temperature distribution are very difficult to obviate even with electrical heating. Some of the results obtained are shown by the following table : E. Becquerel 930 and 890 C. Sainte-Claire-Deville 915 to 945 Barus 926 and 931 Violle . 930 Holborn and Day (two observations) 910 and 930 Callendar 916 D. Berthelot 918 The value 930 as given by Voille's and Barus' results was generally accepted until recently, but the more recent deter- minations indicate 930 to be over 10 high. The value adopted, 920, is probably not in error by over 5 C. * The value 419.0 is obtained if an observation on an admittedly too small sample be included. STANDARDIZATION OF PYROMETERS 439 Gold. (Fusion or freezing) io6f C. This point is to-day one of the best-known fixed points, and gold possesses the ad- vantages of being obtainable in very great purity, is not oxidiz- able in air, nor is it readily attacked by the silicious materials used in crucibles. Its cost is its only drawback for use in con- siderable quantities, but methods have been devised, as insert- ing a short length of wire between the leads of a thermocouple, requiring only very minute amounts of gold. These wire methods give on the average the same results as the crucible method, as shown by Holborn and Day and by D. Berthelot, although their precision is slightly less. The early determinations of the gold point were quite dis- cordant, but the later ones where electric heating was employed are in excellent agreement. Pouillet 1180 C. E. Becquerel 1092 and 1037 Violle 1045 Holborn and Wien 1070 to 7075 Heycock and Neville 1062 D. Berthelot 1064 Holborn and Day 1064 Jaquerod and Perrot 1067 Day and Sosman. 1062 Violle 's value was long quoted as the best for the gold point, but the later determinations show it to be some 20 low. Hol- born and Wien's high value was obtained with a porcelain-bulb thermometer and is to be considered as replaced by Holborn and Day's value, to obtain which nitrogen in a Pt-Ir bulb was used, together with a thermocouple. The agreement of their results when working under various conditions is shown from the follow- ing observations: Gold, sample i 1064.0 0.6 (crucible method) Gold, sample 2 1063 . 5 (crucible method) Gold, sample 2 1063 .9 (wire method) Not less than 300 grams was used for observations in both graphite and porcelain crucibles, while by the wire method 0.03 gram of the metal suffices. Berthelot used his optical gas pyrometer in connection with 440 HIGH TEMPERATURES thermocouples and considers his result to be in error by less than 2 degrees. Heycock and Neville's result was obtained by extrap- olation above the sulphur point of the platinum-resistance for- mula, while Jaquerod and Perrot's value was obtained in terms of a quartz-bulb constant- volume thermometer filled with various gases, the results agreeing to a few tenths of a degree. They used a modified form of the wire method, which consisted in making a small piece of gold wire a part of an alternating electric circuit, melting of the gold being noted by cessation of sound in a telephone. Day and Sosman used their nitrogen thermometer previously described. A preliminary determination with the same appa- ratus by Day and Clement gave 1059 for the gold point on a sample found subsequently to contain iron. It was unfor- tunate that Holborn and Valentiner, in their gas-thermometer work to 1600 C., did not repeat the gold point. An examina- tion of their thermoelectric data shows a discrepancy of about 5 degrees at this temperature from the value here cited as most probable. Berthelot has called attention to the fact that the later deter- minations are sufficiently concordant to warrant reducing them to the thermodynamic scale (see page 26). Observer, Gas. Corrections. Observed D. Berthelot ............... Air 76 cm. + i .36C. 1064 1065.6 Holborn and Day .......... N 29 cm. 0.27 1064 1064.3 Jaquerod and Ferret ....... j OCO \ 23 cm ' ' 21 1067.2 1067.4 Day and Sosman ........... N 21 cm. 0.21 1062.4 1062.6 Silver. (Freezing or melting) 961.0. The freezing point of silver is not a constant temperature except in a reducing atmos- phere, and this metal is volatile, thus making it unsafe to use under conditions in which its vapors may attack platinum wires, as of a thermocouple whose electric properties silver alters very considerably. Many determinations of this point have been made, but it is only the recent observations that take into account" the effects STANDARDIZATION OF PYROMETERS 441 of oxidizing and reducing atmospheres. Some of the determi- nations of the silver point follow: Pure Ag. In air. Pouillet I000 c. E. Becquerel 060 and 016 Violle gj 4 Holborn and Wien 070 Heycock and Neville 060 < occ D. Berthelot 9 6 2 957 Holborn and Day 961 .5 955 Day and Sosman . 960.0 Waidner and Burgess 960.9 953 to 957 Melted silver exposed to the air gradually absorbs oxygen, which lowers the freezing point, and this latter is not a definite temperature, varying with the rate of cooling, mass, and sur- roundings. This lowering may reach 20 degrees or more. The wire method gave 953 .6 =t 0.9 as found by Holborn and Day. The freezing point of pure silver may be obtained in a graphite crucible in an atmosphere of nitrogen or of CO, or covered with powdered graphite, i.e., in conditions preventing oxidation. The melting or freezing point is equally sharp, and on account of the ease of getting very pure silver its use is strongly recommended as a fixed point. Copper. (Freezing or melting) io6f in air, io8f pure. Whether the gold or the copper point was the higher was long an open question in pyrometry. The great advantage in prac- tice of copper is its cheapness, but the fact that copper appar- ently has two freezing points does not possess the same disad- vantages as with silver, for both of the copper points are very definite, the higher one, 1083, being that of the pure metal, easiest obtained with a graphite crucible, the metal being protected from the air by a layer of powdered graphite. The lower value, 1063, is given by the wire method, and copper may replace gold in this way. Values intermediate between 1063 and 1083 will be obtained in crucibles for incomplete protection from air, the effect being due to the formation and solution of cuprous oxide, saturation of the copper with the oxide giving the eutectic point 1063 for about 3.5 per cent CuzO. The presence of the eutectic temperature will usually be detectable, whatever the percentage 442 HIGH TEMPERATURES of Cu 2 O present, and this fact may be used to check the purity of copper in a crucible. We may note the following determinations of the copper point: Heycock and Neville 1080 . 5 Stansfield 1083 Holman Io86 Holborn and Day 1084.1 Day and Sosman 1082.6 Waidner and Burgess. 1083 The values obtained by Holborn and Day, and by Day and Sosman, are the only ones determined directly in terms of the gas thermometer. The difference of 20 C. between the Cu and Cu-Cu 2 O points has been determined by various observers. Palladium. (Fusion) 1550. This temperature marks the present upper limit of the gas thermometer. The following are some of the recent determinations of the palladium melting point: Observed p -.}_-,* Observers. Method. melting 5J5? Nernst and v. Warten- berg Optical; Wien's law, c 2 = 14, 600 1541 1546 Waidner and Burgess Optical; Wien's law, c 2 = 14, 500 1546 1546 Holborn and Valenti- Nitrogen gas, thermocouples, ner and optical c 2 =i4,2Oo 1575 T 5 60 Day and Sosman Nitrogen gas and thermocouples 1549 Palladium may be melted in air by the wire method and there- fore is a convenient control temperature for thermocouples. (See p. 186.) It will be noted that the gas- thermometer deter- mination of Day and Sosman appears to be equivalent to a value of 2 = 14,450 in the Wien equation. Platinum. (Fusion) 1755. Above the palladium point, resort must be had to extrapolation. There have been a great many experimental estimates made of the platinum melting point, some of them based only on extrapolation of purely empirical formulae from temperatures below 1100 C. Such, for instance, are the thermoelectric estimates based on the formula E = a -j- bt + ct* on data which satisfy this equation only in the range 300 to 1200 C. In view of the great importance of this temperature as the best one for reference in the upper part STANDARDIZATION OF PYROMETERS 443 of the scale, all the determinations of which we are aware that have been made are included in the table. The values found prior to the year 1900 are in terms of incorrect values of the basal temperatures, and cannot therefore be correct except by accident. EXPERIMENTAL DETERMINATIONS OF THE MELTING POINT OF PLATINUM. Published Reduced Date. Observers. Method. melting to common point. scale. * 1877-1879 Violle Calorimetric ^775-1779 1892 Barus Thermoelectric 1757-1855 Q j Holborn and 1895 } Wien > Thermoelectric 1780 o 6 { Holman, Law- 9 { rence and Barr | Thermoelectric 1760 1898 Petavel Total light from Pt 1766 1903 Nernst j Total light from black ( body \ 1782 ( Holborn and Thermoelectric and opti- j 1905 \ Henning cal ( 1729 . j Holborn and I9 5 i Henning | Thermoelectric 1710 1755 1905 Harker Thermoelectric 1710 J 755 ^ ( Nernst and 1900 ) Wartenberg Optical; Wien's law (c 2 =i4,6oo) | 1745 I75i ( Waidner and Optical; Wien's law j I7"?3 1907 i Burgess ( 1753 ( Holborn and 1907 i Valentiner Optical; Wien's law (2=14,200) j 1782 1763 ( Waidner and Monochromatic radiation j 1907 i Burgess from Pt ( 1750 1750 ( Waidner and 1907 i Burgess Thermoelectric 1 (two formulae) j 1706-1730 1753 f Monochromatic radiation T7> ' I from Pt: 1909 Fery 1 Oxidizing atmos. 1690 I Reducing atmos. 1740 .... 1910 Sosman j Thermoelectric from I Pd=i 54 9 1752 1755 1910 Ruff Optical about 1750 Best Value 1755 * This scale is that for which c^= 14,500 in Wien's law III, p. 251. The published thermoelectric determinations involving extrap- olation on the thermoelectric scale (equation (3) , page 112) from low temperatures have little or no weight. The method used by Nernst in 1903 is not capable of great accuracy. Fery's as well as Ruff's measurements appear to have been crude, and the differences noted by the former may be due to the surface prop- 444 HJ GH TEMPERATURES erties of the platinum in the different parts of a gas flame and not to the oxidizing and reducing atmospheres as such. All the optical measurements by the other observers were taken in an ox- idizing atmosphere and are at least 50 degrees higher than Fery's oxidizing- atmosphere values. The outstanding uncertainty of the platinum point is mainly attributable to the difference as- signed to Cz in the Wien formula and to the different gas scales in terms of which the extrapolations are made. The value here assigned to the platinum point, 1755, is in terms of the Day and Sosman gas scale (Pd = 1549), the optical determinations of Nernst and v. Wartenburg, and Waidner and Burgess, and the mean differences between the palladium and platinum points as found by them and by Holborn and Valentiner, thus: Observers. Pt-Pd. Nernst and v. Wartenberg 204 C. Holborn and Valentiner 207 Waidner and Burgess 207 Rhodium. (Fusion) 1940. The other members of the plat- inum group have had their melting points less well determined than palladium and platinum. For rhodium the following esti- mates, among others, have been made: Mendenhall and Ingersoll (Pt= 1755) 1932 V. Wartenberg (using a tungsten furnace) 1940 Iridium. (Fusion) 2300. Although it is questionable if temperature of 2000 C. and over can be determined in terms of the gas scale, it may, nevertheless, be found desirable to deter- mine as exactly as may be one or more fixed temperatures in this range by other methods, as specific heat and the laws of radiation. Iridium and tungsten seem to be the most suitable for this purpose. Hardly any limit of accuracy can as yet be placed upon such determinations. For iridium the following values have been found: Violle 1950 C. Veder Weyde 2200 Nernst 2200 to 2240 Rasch (computed from Nernst' s data) 2285 Mendenhall and Ingersoll 2300 v. Wartenberg 2360 STANDARDIZATION OF PYROMETERS 445 V. Wartenberg's determination was made in a tungsten furnace in vacuo; that of Mendenhall and Ingersoll of a bead on a Nernst glower. The recent development of furnaces suitable for use at these extreme temperatures will undoubtedly enable us to more sharply define other points in this part of the scale. It is interesting to note that at the extreme temperature of the electric arc, 3600 C., the various radiation methods and the specific-heat method give results agreeing to about 100 C. Other metals melting below 1100 C., such as cadmium, lead, antimony, and aluminium, have also been used in the attempt to determine fixed points, and some of the results are given in the accompanying table for metals melting below 1100 C. TABLE OF FREEZING POINTS TO 1100 C. Observers... . \ Date Stansfield. 1898 D. Berthe- lot. 1898-1901 Heycock and Neville. Callendar. 1895-99 Waidner and Burgess 1910 Holborn and Day. 1900-1901 Day and Sosman. 1910 Instrument. . . Recording thermo- couple. Optical interference. Electrical resistance. Electrical resistance. Nitrogen ther. and thermo- couple. Nitrogen ther. and thermo- couple. Calibration ( data j IOO 444-53 Expansion of air. IOO 444.53 IOO 444-70 Pt-Ir-bulb nitrogen thermometer Pt-Rh-bulb nitrogen thermometer Sn 2*2 1 2^1.9 2^1.9 Bi 268 4 269 2 Cd 320. 7 321 .0 321 . 7 320 Pb 32S .O 327. 7 327.4 326 .9 Zn Sb Al 418.2 640 2 419.0 629.5 645.5 419.4 630.7 658.0 419.0 630.6 657.0 418.2 629.2 658.0 Ag3~ Cll2 778 s 779- 2 Ag (in air) nee o 9 [ > t ? -O Ag (pure).. Au 961.5 1062 7 962 1064 960.7 1061 . 7 960.9 961.5 1064 . o 960.0 IO62 .4 Cu-Cu 2 O. 1063 . 2 1064 . 9 Cu 1083 . o 1080 . 5 1083.0 1084 . i 1082.6 The Iron Group. A fixed point that has been frequently used is the melting point of nickel (1450). The thermoelectric determinations based on empirical formulae gave values varying 446 HIGH TEMPERATURES from 1484 to 1427. Day and Sosman find 1452 with the gas thermometer, and Ruer 1451 with the thermocouple, assuming palladium = 154 1. The former find for cobalt 1490, and, from the measurements of several observers, iron would have a melt- ing point of about 1520 on the same scale. These metals are readily oxidized and usually contain sufficient impurities to in- fluence their melting temperatures somewhat. They are most readily worked in an atmosphere of hydrogen. Microscopic samples melted on platinum in hydrogen, as measured by Burgess with an optical pyrometer (see p. 343) , gave Ni = 1435, Co = 1464, and Fe = 1505. Metals Melting above 2000 C. Above the platinum point, there have been recently a considerable number of attempts to locate fixed points. With the exception of iridium and rhodium, which have already been mentioned, it appears to be necessary to work in vacuo all the elements available in this very high temperature region. A very convenient way to mount them is as filaments or strips as in incandescent lamps. Of the elements that have been so studied, only tantalum and tungsten have been determined with a fair agreement by several observers; and tungsten is the only one which melts without excessive evaporation, and, having the highest melting point yet measured, appears to be the best adapted for an extreme fixed point. For tungsten the following values have been found: "Waidner and Burgess (1906-1910) (2=14,500). . . . 3250-3050 C. V. Wartenberg (1907-1910) (^2=14,600) 2800-2900 Pirani (1910) (c 2 = 14.500) 3250 The value 3000 C. is probably correct to 100 C. For tantalum we have: V. Bolton (1905) 2250-2300 Waidner and Burgess (1907 and 1910) 2910 Pirani (1910) 3000 Pirani and Mayer (1911) 2850 Measurements of the melting points of osmium, molybdenum, titanium, and other very refractory elements have also been STANDARDIZATION OF PYROMETERS 447 made, but none of them gives promise of being as serviceable as the above for fixed points in pyrometry. Melting Points of the Chemical Elements. In Table II of the Appendix is given a list of the melting points for the chemical elements with some indication of our knowledge of their exactness. Fig. 170. Freezing of Copper. Typical Freezing-point Curves. The freezing-point curves of copper, antimony, silver, and aluminium are shown in Figs. 170 to 173, from data obtained at the Bureau of Standards, in which time in minutes is plotted as abscissa and E.M.F. of a 90 Pt-io Rh 448 HIGH TEMPERATURES .10 47.000 .900 I- \ \ > 1787 A P.P. Sb F.P.=46.8367o>o6307 C. \ \ A \ 7^ 4= =^f- =*= =F= =*= =F^ [ F= H , -> --^ % Resistance in l_i__i_JL_J \ 1 \ \ \ 1 \ 1 \ \ \ V \ \ V 619^6 C. 1H.18'20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 12H.O'02 04 06 Time in Minutes Fig. 171. Freezing of Antimony. 21 W .98 #1787F P.P. Ag F.P.=12.1622O960?72 C. *478F F.P.Ag F.P.=21.8902.96075 C. 3 hrs. 02 m. 04 06 08 10 12 It 16 18 20 22 24 26 28 30 Fig. 172. Freezing of Silver. STANDARDIZATION OF PYROMETERS 449 thermocouple as ordinate for copper and aluminium, and the resistances of a platinum thermometer for antimony and silver. An inspection of the copper curve shows why this metal is desir- able to use, as it gives a very flat curve. With aluminium rapid Fig. 173. Freezing and Melting of Aluminium. cooling would be fatal to an exact determination. The slant here observed in the curve at the transition point is characteristic of the presence of impurities and of low conductivity and latent heat. For this metal the melting curve is also given, showing the melting and freezing points to differ somewhat, apparently. 450 HIGH TEMPERATURES Antimony undergoes great undercooling, depending on the rate of cooling, and may reach over 30 C.; but the maximum is a very definite temperature for moderate undercooling, and for a quick- acting thermometer in a charge of metal that is not too small. For silver, two rates of cooling are shown. Boiling Points. Sometimes it is desired to calibrate a pyrom- eter down to room temperature, even if in this case the use of a mercury thermometer is usually to be preferred. Use may be made of the boiling points of water, aniline or naphthaline, and benzophenone, or of the tin freezing point, 231.9. Water. 100 by definition, with a variation of 0.04 for a. change of i mm. in atmospheric pressure. Aniline. 184.1, with a change of 0.05 per millimeter. This value is probably correct to 0.1. Aniline, however, oxidizes readily. Naphthalene. 218.0, with a change of 0.058 per millimeter. This point has been very carefully determined by several ob- servers (see page 226), and naphthalene is cheap and readily obtained of sufficient purity, best tested by taking its freezing point, which should be 80.0 C. Benzophenone. 306.0, with a change of 0.063 P er millimeter. Although expensive and difficult to get pure (melting point = 47.2), this substance appears to be the only satisfactory one so far found possessing a sufficiently constant boiling point between 218 and 445. The sulphur boiling-point apparatus (Fig. 169) may be used for both naphthalene and benzophenone if provided with an auxiliary condensing tube. Both these boil- ing points are easily kept constant to better than 0.05. Metallic Salts. The different fixed points that have been mentioned are not all of a very convenient use. It would be preferable to have in the place of the metals, metallic salts for the determination of the fixed points if they can be shown to be satisfactory otherwise. These salts fortunately are for the most part without action on platinum, which is of great advan- tage for the standardization of thermocouples and resistance thermometers. There are few, however, whose fusing points, STANDARDIZATION OF PYROMETERS 451 have been determined up to the present time in a sufficiently precise manner. Among the salts whose freezing or melting temperatures have been carefully determined, and which may therefore be used for calibration purposes, are: NaCl (melting) by W. P. White 801 C. NaCl (freezing) by G. K. Burgess 800 Na 2 SO 4 (melting) by W. P. White 885 Diopside (CaMg(SiO 3 )2) (melting) by Day and Sos- man 1391 Anorthite (CaAl2Si2O 8 ) (melting) by Day and Sosman 1549 Lithium metasilicate (Li 2 SiO 3 ) (melting), F. M. Jaeger 1202 Sodium metasilicate (Na 2 SiO 3 ) (melting), F. M Jaeger 1088 The transformation points of some salts can only be obtained satisfactorily on heating due to great undercooling when it is attempted to take their freezing points. This is true, for in- stance, of diopside, anorthite, and the silicates, the values for which apply only for chemically pure salts prepared artificially. Where stirring is practicable the undercooling can largely be avoided in taking freezing points of both metals and salts. In general, salts give a less sharp melting point than metals, due mainly to low conductivity and heat of fusion of the former, and of course impurities will act in the same way. There is no difficulty, however, in keeping the melting-point curve flat to within i C. for some pure salts such as NaCl and N2SO 4 . There are undoubtedly other salts which might be studied to advantage, such as: Melting point. i MolNaCl+i MolKCl about 650 Pb 2 O 5 2 Na 2 O about 1000 MgSO 4 about 1150 K 2 S0 4 has been used to some extent, but it appears to possess several dimorphous varieties with different melting points, like sulphur, so that the actual point observed may be uncertain. On page 367 is given a list of salts and the status of their melt- ing points as determined by relatively less precise methods than the above. Alloys : Eutectic Points. In the case of certain alloys there are well-defined transition points which may be used as fixed 452 HIGH TEMPERATURES temperatures to advantage in those temperature intervals in which there is no conveniently located and suitable metal freezing point. The most sharply denned of such transformations are the temperatures of freezing of'eutectics, when, if the components are pure and the alloy is of very nearly the eutectic composition, the evolution of heat and the constancy of temperature during the transformation compare favorably, in some cases, with the freezing of a pure metal. Such a suitable eutectic in a desirable location is that of sil- ver and copper, which happens to have the composition Ag 3 -Cu 2 , whose freezing temperature has been found by Heycock and Neville to be 779.0, and by Waidner and Burgess 779.2. There are probably a considerable number of such transformation temperatures that could be used as fixed points to advantage. Thus the eutectics of aluminium or of antimony with the members of the iron group are probably more sharply defined temperatures than those of the commercial metals often used in standardizing thermocouples. Another well-known and fairly reproducible transformation temperature on cooling in the solid state is the recalescent point of steel (iron-carbon), of maximum effect for C = 0.9% at about 705 C. for slow cooling and some- what lower for fast cooling. Reproducibility of Freezing Points. It is of great importance in pyrometry of precision and in the calibration of instruments to be able to reproduce exactly the fixed temperatures of boiling and freezing or fusion. We have seen that the materials ordi- narily used for boiling points can easily be had in sufficient purity to reproduce these temperatures to within 0.05, and that their freezing points are a delicate test of purity. There have been several intercomparisons of the thermal repro- ducibility of some of the metals whose freezing temperatures are used as fixed points. Thus, Day and Allen in 1904, using thermocouples, found that the metals used in the establishment of the Reichsanstalt scale could be purchased in America, with the exception of antimony, to give the same scale to within i C. Waidner and Burgess, using both thermocouples and platinum- STANDARDIZATION OF PYROMETERS 453. resistance pyrometers, the latter of which is capable of the much greater sensitiveness and reliability, have made recently an ex- haustive study of the reproducibility of several of the metal freezing points, as shown in the following table, in which the samples were purchased from reliable American and German firms as their best product: Metal Sn Cd Pb Zn Sb Al Cu Number of samples 5 3 4 3 6 3 4 Reproducibility* in degrees C. 0.06 .26 .10 .06 2.3 1.2 i.o There was one or more carefully analyzed sample of each metal, and this table shows that, with the exception of Sb and Al, it is very easy to get these metals pure enough from several sources. The only sufficiently pure antimony was " Kahlbaum," and the best aluminium was from the Aluminium Company of America. The uncertainty noted in the copper point is due mainly to oxida- tion and the uncertainties of measurement. Of the other metals often used, but not cited in the above table, silver and gold are readily obtained of the highest purity; palladium and platinum less readily so, but their purity as wires is easily tested by meas- uring their temperature coefficients (see Chapter V). Temperature of the Arc and Sun. In certain problems in- volving extremely high temperatures and as comparison sources for apparent stellar temperatures, the positive crater of the carbon arc and the sun's disk may be used, although the actual values to assign to their temperatures are still somewhat in doubt. In the case of measurements in terms of the radiation laws, it is to be remembered that, other things being equal and barring luminescence, the values found will be low, due to the selective radiation of carbon and of the sun's disk, or the de- parture of their radiation from the laws of the black body. We should expect, also, that measurements made by total-radiation methods would give lower temperatures than by spectral- radia- tion methods if the arc and sun have energy distributions differ- ing from those of the black body at the same temperatures. * The reproducibility is defined as the average deviation of the freezing points of the metals from their mean freezing point. 454 HIGH TEMPERATURES The following measurements have been made on the positive crater of the arc: TEMPERATURE OF THE CARBON ARC. Observers. Le Chatelier. . . Violle . . Wilson and Gray Petavel.. . Date. 1892 1895 Temper- ature centigrade. 4100 3600 3330 3830 Wanner Very Fery Fery 1900 3430-3630 I 1899 333~373 348-3930 3490 Waidner and Burgess 1902 1904 1904 ( 3343 3420 3450 Method. Photometric; intensity of red light. Calorimetric; specific heat of carbon. Total radiation of copper oxide, em- pirical relation for. Total light from Pt; empirical for- mula. (Varying with carbons used) photo- metric in terms of Wien's law, page 251- Wien's displacement law. Wien's displacement law. Total radiation; Stef an-Boltzman law. Photometric; Wien's law. Total radiation. Holborn-Kurlbaum pyrometer (red and green light). Wanner pyrometer. Le Chatelier optical pyrometer. Wien's law. It seems probable that the temperature of the arc is not over 3600 C., and the value 3500 C. appears to best represent the results. Considerable changes in current produce less effect on the apparent temperature than do variations in the kind of carbons used. In taking observation in the arc, it is convenient to mount the positive terminal horizontally and the negative vertically. Care should be taken to have a sufficient area at the maximum temperature, especially when using total-radiation methods. This can only be accomplished by using heavy car- bons, 1.5 cm. or more in diameter, and correspondingly high currents. Observations in terms of several of the radiation laws have been made on the apparent temperature of the sun's disk. Ob- servation shows also that the apparent temperature falls off from the center to the limb, due to absorption in the outer layers, from which it is deduced that the photosphere has a temperature some 500 hotter than the observed value for the center of the disk. STANDARDIZATION OF PYROMETERS 455 (A) If Stefan's law is assumed to hold, and if the solar con- stant / as well as the coefficient a are known, the formula / = oT 4 , with a proper choice of units, gives us T (absolute) directly; or a calibrated total-radiation pyrometer may be used if the absorp- tion of the earth's atmosphere is corrected for. (B) If the position of the wave length of maximum energy is known and if the spectral-energy curve for the sun resembles that of the black body, Wien's displacement law \ m T= C may be used if C is known. (C) Similarly, the relation E m T~ b = const, may also be used. / c, \-i (D) Finally, Planck's equation, I = Ci\- 5 (^ T - i J , gives us still another method if c% is known. If these methods, including measurements with several wave lengths by (D), all gave the same temperature for the sun, using the constants characteristic of the black body in the several equations, it would follow that the apparent temperature found would be the true temperature of the sun. The spectral methods in general, however, appear to give relatively high values, indicating that the true temperature of the sun, except for luminescent effects, is higher than any of the observed values. Some of the recent observations are given below for the appar- ent mean value of the temperature of the sun's disk. SOME RECENT ESTIMATES OF THE SUN'S APPARENT TEMPERATURE. Mean Observers. temperature Method. centigrade. Millochau and j ( (A) With actinometer (solar constant=2.8 to Fery ......... j" 59-539 o } 2 55 ) an( j total-radiation pyrometer. Scheiner ....... 593 (A) Wilsingand ) 5130-5600 (D) Using 5 wavelengths, c 2 = 14,600. M , v_^ J Modification of (D) with heterochrome pho- Nordmann ..... 5050-5630 j tomete r, assumes arc = 3 343 C., c 2 = 14,600. Abbot and Fowle 6160 (B) X m = 0.433 M I C= 2930. ( (A) Solar constant = i. 95 ;Kurlbaum's value of Abbot and Fowle 5 5?o { > (page ^ 5460 or j (D) Using several wave lengths and for a, r I4j200 456 HIGH TEMPERATURES Goldhammer has shown that (B) is probably the least reliable method, and (D) the one least subject to objection if several wave lengths are used. For measurements corrected for the earth's atmosphere, the value 6000 C. would seem to be a fair one for comparison with other celestial sources. Table of Fixed Points. In the actual state of our knowledge, the fixed points to which we should give preference are summa- rized in the table below, in which temperatures below 1600 C. are expressed in terms of the scale of the nitrogen constant- volume thermometer, which has given fairly consistent results to 1100 C., as we have seen, in the hands of several experimenters. Between 1100 and 1600 C. the results of Day and Sosman are followed, and above 1600 temperatures are expressed in terms of Wien's law (page 251), in which c 2 is taken as 14,500, as best representing the data at hand. Estimates of the accuracy with which these fixed points are known, and also of their reproduci- bility from a known source of supply, are given in the table. The uncertainty of some measuring device is of course included under reproducibility. TABLE OF FIXED POINTS. Boiling points. Accuracy. Naphthalene ...................... 218.0 0.2 .02 Benzophenone .................... 306 . o 0.3 .03 Sulphur ........................... 444-7 - 5 3 Freezing points. Tin ............................... 231.9 0.2 0.03 Cadmium ......................... 321 0.3 0.05 Lead ............................. 327 0.3 0.05 Zinc .............................. 419 0.5 0.05 Antimony ......................... 631 1.5 0.3 Sodium chloride .................. 800 2.0 i . o Silver ............................ 961 2.0 0.3 Gold ............ . ................ 1063 3.0 0.5 Copper ........................... 1083 3 i Lithium metasilicate .............. 1202 5 2 Diopside .......................... 1391 10 5 Nickel ............................ 1450 15 10 Palladium ........................ 1550 15 5 Platinum ......................... 1755 20 10 Tungsten ......................... 3000 100 25 Carbon arc ....................... 3500 150 50 Sun ............................... 6000 500 100 STANDARDIZATION OF PYROMETERS 457 Standardization of Pyrometers. The above discussion has shown that we possess a number of fixed points which have been established with sufficient accuracy to use them in the stand- ardization of pyrometers. For such standardization, two courses are open besides direct comparison with a gas thermometer, a proceeding usually out of the question, and furthermore rendered superfluous by the establishment of these fixed points in terms of the gas scale. When its construction permits, a pyrometer may be calibrated by finding its indications at two or more of the fixed points, or may be compared with another which has been so calibrated. The latter method is the one used for ordi- nary purposes, as in the graduation of industrial instruments, but for pyrometers which are to be used as standards the former method should be used when possible. We have discussed at some length, in their respective chap- ters, the methods of calibration for the various pyrometers, and it is unnecessary to dwell further on this matter, except to say that it cannot be assumed that a pyrometer once standardized is standardized for all time, especially if it is subjected to hard usage. Standardizing Laboratories. Recognizing the importance of establishing, preserving, and disseminating a common and author- itative temperature scale and of providing means of having pyrometers and other instruments certified as to their accuracy, some of the governments have established laboratories, such as the Physikalisch-Technische Reichsanstalt in Germany, the National Physical Laboratory in England, the National Bureau of Standards in the United States, and the Laboratoire d'Essais and the Laboratoire Central d'Electricite in France, whose functions are not only testing instruments but carrying on researches as well. The German institution, the oldest of these laboratories, has been one of the most potent factors in the development of excellency in German instruments, and has been of immense service to the industries as well as to the interests of science; and the other national laboratories are fast assuming a position of equal importance in their respective countries. 458 HIGH TEMPERATURES Metals and Salts of Certified Melting Points. It would often be of great convenience, when one has to calibrate his own pyrom- eter, and in cases of dispute between individuals as to their respective temperature scales, to have available metals or salts the melting points of which had been certified by a standardizing laboratory. The Bureau of Standards is preparing to issue such certified metals and salts of sufficient range and number to meet the ordinary requirements of pyrometer calibration. Electrically Heated Furnaces. For the standardization of pyrometers as well as in many other high-temperature problems, it is necessary to preserve a constant temperature for a consider- able time and to be able to reproduce a given temperature very exactly. Electrically heated resistance furnaces best serve these ends, and great improvements have been made in their construction in recent years. Furnaces wound with nickel wire of i to 2 mm. diameter on porcelain have been used considerably, but they are slow in heat- ing up and their upper limit is about 1200 C., if the furnace is to be used frequently, although for a single heating 1400 C. may be attained with care. Platinum wire has been used to attain .higher temperatures, but the use of this material in wire form is very expensive for heating. Heraeus has made electric heating to 1300 or 1450 C., depend- ing on size of furnace, generally accessible by the substitution of platinum foil for the wire, weighing about 1.5 grams per square centimeter or having a thickness of about 0.007 mm. This re- duces the cost of a platinum furnace very greatly, and has the further advantages of giving slightly greater uniformity of heat- ing and attaining at a somewhat greater speed high temperatures than with wire- wound furnaces. Above 1500 C. chemical action sets in between the platinum and material of the tubes usually .employed, so that although, as far as the platinum is concerned, 1700 C. could be maintained for a short time, yet the present safe upper limit for long periods of heating is 1400 with foil furnaces. STANDARDIZATION OF PYROMETERS 459 Their greatest weakness is the cracking of the porcelain tubes on which the foil is wound and the evaporation of the platinum when not covered with a suitable paste. For very high temperatures, up to 2100, the iridium-tube furnaces of Heraeus may be used, as they have been with success by Nernst and others in the study of vapor pressures at these temperatures, as well as in melting point and physiochemical investigations. In Fig. 174 is shown the furnace and accessories used by Waidner and Burgess for the determination of the palla- Fig. 174. Iridium-tube Furnace. dium and platinum melting points. For lower temperatures, tubes of platinum or of a platinum alloy may be used with great saving in cost. Crucible Furnaces. A suitable form of electrically heated crucible furnace for freezing and melting point determinations to 1100 C., such as used at the Bureau of Standards, is shown in Fig. 175. The double winding with platinum ribbon gives a very delicate temperature control if connected in parallel through sep- arate rheostats on the same battery. This furnace is designed to carry a crucible of 300 c.c. capacity, so as to give ample im- 460 HIGH TEMPERATURES mersion of the thermometer at a constant temperature. If the freezing or melting of a metal is made to take twenty minutes or more, the form of the freezing or melting curve becomes a very 10 cm Fig. 175. Double- wound Crucible Furnace. sensitive check on the purity of the sample. Another form of crucible furnace used successfully at the Carnegie Geophysical Laboratory to above 1600 C. is shown in Fig. 52. The char- acteristic of this furnace is the inwound platinum-wire heating STANDARDIZATION OF PYROMETERS 461 coil. Although of high first cost, it is a very durable furnace as constructed. Both types of furnace may be arranged for use with any desired atmosphere. Less satisfactory results will be obtained with ordinary gas furnaces to 1300 C. SECTION A-B Fig. 176. Arsem Vacuum Electric Furnace. Vacuum and Pressure Furnaces. Mr. Arsem of the General Electric Company has developed a type of vacuum furnace which is convenient for certain melting points, such as fireclays, re- fractory bricks, and ashes, and chemical investigations to 2500 C. or higher. In one of its ordinary water-cooled forms, shown in Fig. 176, the heating is produced by passing an alternating cur- 462 HIGH TEMPERATURES rent of low voltage through a graphite spiral. The highest tem- peratures may be reached in a few minutes. The interior is observed through a mica or glass window. Temperatures are measured with an optical pyrometer. By taking the heating curve, the transformation points for only a few tenths grams of material are easily observed. (See page 342.) Such a furnace has been in constant use at the Bureau of Standards for several years. V. Wartenberg has successfully constructed a tube resistance furnace of tungsten mounted in vacuo (Fig. 177), and with it Pump Fig. 177. Wartenberg's Tungsten Furnace. determined the melting points of a number of refractory elements melting above 2000 C. Messrs. Hutton and Petavel of Manchester, England, have constructed a pressure furnace for work at high temperatures, the essential parts of which are shown in Fig. 178. A vertical carbon tube was electro-coppered at the ends, soldered into brass castings, and provided with water circulation at A and B. Temperature readings were taken down the side tube of carbon, fixed into a brass tube with a window at the end, a current of hydrogen being admitted at C. The whole furnace was packed in crushed wood charcoal, while a thin walled graphite crucible STANDARDIZATION OF PYROMETERS 463 contained the metal to be studied. This furnace has been used by Greenwood for the determination of the boiling points of Fig. 178. Graphite Furnace of Hutton and Petavel. some of the metals and their variation with pressure. Other types of furnace are described in Chapters II, IV, and V. BIBLIOGRAPHY. GENERAL WORKS. (The heavy figures refer to volumes.) Weinhold. Principles of construction of pyrometers. Pogg. Ann., 149, p. 186; 1873- C. H. Boh. Die Pyrometer, Eine Kritik der bisher construirten hoher Temper- aturmesser in wissenschaftlich-technische Hinsicht, 1888. C. Barus. Die physikalische Behandlung und die Messung hoher Temperaturen, 1892, Leipzig. Bull. U. S. Geological Survey No. 64, 1889; No. 103, 1893. Kayser. Handbuch der Spectroscopie, Bd. 2, 1902. Summary of radiation methods. Callendar. Measurement of Extreme Temperatures. Nature, 69, 1899. C. W. Waidner. Methods of Pyrometry. Proc. Eng. Soc. Western Pennsylvania, Sept., 1904. Waidner and Burgess. Establishment of High-temperature Scale. Phys. Rev., 24, p. 441; 1907. Burgess. Estimation of High Temperatures. Proc. Am. Electroch. Soc., 11, p. 247; 1907. Chivolson. Traite" de Physique. (Instruments, Theory, Bibliography.) Leber. Ch. xiv in Geiger's Handbuch der Eisen- und Stahlgiesserei, 1911. NORMAL SCALE OF TEMPERATURES. Carnot. Reflections on the motive power of fire. Lippmann. Thermodynamics, p. 51. Thomson and Joule. Philosophical Transactions of the Royal Society, 42, p. 579; 1862. Thomson (Lord Kelvin). Collected Papers, 1, p. 174. Lehrfeldt. Philosophical Magazine, 46, p. 363; 1898. Callendar. Phil. Trans., 178, pp. 161-220; 1888. Regnault. Account of his investigations, 1, p. 168; 1847. Chappuis. Studies of the gas thermometer. Trav. du Bureau International des Poids et Mesures, 6, 1888. Chappuis and Barker. Trav. et M6m. du Bureau Int. des Poids et Mesures, 1900. Phil. Trans., 1900. Schreber. Absolute Temperature. W. Beibl., 22, p. 297; 1898. D. Berthelot. Reduction of gas-thermometer readings to the absolute scale. Trav. et Mem. du Bureau Int., 13, 1903. C. R., 138, p. 1153; 1904. Boltzmann. On the determination of absolute temperature. Wied. Ann., 63, p. 948; 1894, 465 466 HIGH TEMPERATURES E. Mack. Theory of Heat. Callendar. Thermodynamical correction to gas thermometer. Phil. Mag., May, 1903. Rose-Innes. Phil. Mag. (6), 2, p. 130; 1901. Pellat. C. R., 136, p. 809; 1903. Chappuis. ]\. de Phys., 3, p. 833; 1904. Ann. des Poids et Mesures, 13, p. a3; 1907. Berthelot. Ann. des Poids et Mesures, 13, p. b3; 1907. Buckingham. Bull. Bureau Standards, 3, p. 237; 1907. Rose-Innes. Phil. Mag., 14, p. 301; 1908. GAS PYROMETER. Prinsep. Ann. Chim. et Phys., 2d Series, 41, p. 247; 1829. Poidllet. Treatise on Physics, 9th ed., 1, p. 233; 1858. Comptes Rendus, 3, p. 782; 1836. Ed. Becquerel. C. R., 57, pp. 855, 902, 955; 1863. Sainte-Claire-Devitte and Troost. C. R., 90, pp. 727, 773; 1880. 45, p. 821; 1857. 49, p. 239; 1859. 56, p. 977; 57, pp. 894, 935; 1863. 98, p. 1427; 1884. 69, p. 162; 1864. Ann. Chim. et Phys. (3), 58, p. 257; 1860. Rupert. Chim. Appl., p. 326; 1863. Violle. Specific heat of platinum. C. R., 85, p. 543; 1877. Specific heat of palladium. C. R., 87, p. 98; 1878. 89, p. 702; 1879. Boiling point of zinc. C. R., 94, p. 721; 1882. V. and C. Meyer. Density of halogens. Ber. D. Ch. Ges., 12, p. 1426; 1879. Earns. Bull. U. S. Geological Survey No. 54, 1889. Phil. Mag. (5), 34, p. i; 1892. Report on Pyrometry, Congress at Paris, 1900. Regnault. Account of his experiments, 1, p. 168; Paris, 1847. Mem. de ITnstitut, 21, pp. 91, no; 1847. Ann. Chim. et Phys. (3), 68, p. 89; 1861. Holborn and Wien. Bull, de la Soc. pour 1'encouragement (5), 1, p. 1012; 1896. Wied. Ann., 47, p. 107; 1892. 56, p. 360; 1895. Zeits. fur Instrum, p. 257;. 1892. Crafts and Meier. Vapor density of iodine. C. R., 90, p. 690; 1880. Langer and V. Meyer. Pyrochemical Researches (Brunswick), 1885. Joly. Pogg. Ann., Jubelband, p. 97; 1874. Randall. Permeability of platinum. Bull. Soc. Chim., 21, p. 682; 1898. Mallard and Le Chatelier. Ann. des Mines, 4, p. 276; 1884. /. R. Erskine Murray. On a new form of constant-volume air thermometer. Edinburgh Proc., 21, p. 299; 1896-97. Journ. Phys. Chem., 1, p. 714; 1897. /. Rose-Innes. The thermodynamic correction for an air thermometer, etc. Nature, 58, p. 77; 1898. Phil. Mag. (5), 45, p. 227; 1898. 60, p. 251; 1900. Proc. Phys. Soc. London (i), 16, p. 26; 1898. Chappuis. Phil. Mag. (5), 50, p. 433; 1900. (6), 3, p. 243; 1902. Report for Paris Congress, 1900; Jour, de Phys., Jan., 1901, p. 20. D. Berthelot. On a new method of temperature measurement. C. R., 120, p. 831; 1895, Ann. Chim. et Phys. (7), 26, p. 58; 1902. BIBLIOGRAPHY 467 Holborn and Day. Wied. Ann., 68, p. 817; 1899. Am. Jour. Sci. (4), 8, p. 165; 1899. Zeitscher. Instrum., May, 1900. Am. Jour. Sci. (4), 10, p. 171; 1900. Drude's Ann., 2, p. 505; 1900. Chappuis and Harker. Trav. et Mem. du Bureau Int. des Poids, 1900, 1902. Phil. Trans., 1900. Callendar. Phil. Mag., 48, p. 519; 1899. P roc - Rov - Soc., 50, p. 247; 1891. Callendar and Griffiths. Phil. Trans., 182, 1891. D. Berthelot. On gas thermometers- and the reduction of their indications to the absolute scale. Trav. et Mem. du Bureau Int., 13, 1903. Travers. Studies in Gases (Macmillan). Proc. Roy. Soc., 70, p. 485. Kapp. Ann. der Phys., 5, p. 905; 1901. Wiebe and Bb'ttcher. Gas thermometry, Berich. Berlin Akad., 44, p. 1025. Inst'kunde, 1888. Jaquerod and Perrot. Various gases in quartz. C. R., 138, p. 1032; 1904. Day and Clement. Am. Jl. Sci., 26, p. 405; 1908. Day and Sosman. Scale 400 to 1550 C. Am. Jour. Sci. (4), 29, p. 93; 1910. Pub. 157, Carnegie Institution of Washington, 1911. A. L. Day. Met. and Chem. Eng., 8, p. 257; 1910. Holborn and Valentiner. Ann. der Phys. (4), 22, p. i; 1907. Berthelot. Value of gas constant. Zs. Elektroch., 10, p. 621; 1904. Jaquerod and Perrot. Expansion coefficients. C. R., 140, p. 1542; 1905. Holborn and Henning. Ann. der Phys. (4), 35, p. 761; 1911. CALORIMETRIC PYROMETER. (See also Specific Heat.) Violle. Boiling and fusing points. C. R., 89, p. 702; 1879. Le Chatelier. Sixteenth Congress of the Societe" technique de 1'industrie du gaz, June, 1889. Euchtne. Thermal relations in the distillation of oil. (Monograph.) Ferrini. Rend. Lomb., 35, p. 703; 1902. Berthelot. Calorimetry. Ann. Chim. et Phys. (4), 20, p. 109; (5), 5, p. 5; (5), 10, pp. 433, 447; (5), 12, p. 550. Hoadley. Trans. Am. Inst. Mech. Engs., 2, No. 23, p. 42; 3, No. 65, p. 187. THERMOELECTRIC PYROMETER. Becquerel. Ann. Chim. et Phys. (2), 31, p. 371; 1826. Pouillet.Traite de Physique, 4th ed., 2, p. 684; C. R., 3, p. 786; 4, p. 513; 1836. Ed. Becquerel. Annales du Conservatoire, 4, p. 597; 1864. C. R., 65, p. 826; 1862. Ann. de Chim. et de Phys. (3), 68, p. 495 1863. Tail. Thermoelectric power. Trans. Roy. Soc. Edinb., 27, p. 125; 1872-73. Regnault. Account of investigations on heat engines, 1, p. 240. C. R., 21, p. 240; 1847. (Fe-Pt couple.) Knott and MacGregor. Graphical representations. Trans. Roy. Soc. Edinb., 28, p. 321; 1876-77. Seebeck. Fundamental principles. Pogg. Ann., 6, pp. 133, 263; 1826. 468 HIGH TEMPERATURES Gumming. Inversion points. Trans. Cambridge Soc, 2, p. 47; 1823. Draper. Phil. Mag. (3), 16, p. 451; 1840. Mousson. Arch. d'Elec. (Geneva), 4, p. 5; 1844. (Inhomogeneity.) Magnus. Pogg. Ann., 83, p. 469; 1851. (Inhomogeneity.) Thomson. Trans. Edinb. Soc., 21, p. 123; 1854. Phil. Mag. (4), 11, pp. 214, 281, 379, 433; 1856. C. R., 39, p. 116; 1854. Phil. Trans., 146, p. 649; 1856. Avernarius. Formulae and various couples. Pogg. Ann., 119, p. 406; 1863. 122, p. 193; 1864. Kleminic and Czermak. Pt-Cu-Ni-Fe couples. Wied. Ann., 50, p. 175; 1893. Le Chatelier. Thermoelectric pyrometer. C. R., 102, p. 819; 1886. Journal de Phys. (2), 6, Jan., 1887; Genie civil, March 5, 1887; i6th Congress of the Societe technique de 1'industrie du gaz, June, 1889; Bull, de la Societe de I'encouragement, 1892; Bull. Soc. Chim. Paris, 47, p. 42; 1887. Barus. Washington, 1889, Bull, of the U. S. Geological Survey, No. 54, and No. 103 (No. 54 contains a very complete historical account of the whole subject of pyrometry). Phil. Mag. (5), 34, pp. 15, 376; 1892. Am. Jour. Sci., 36, p. 427, 1888; 47(3), p. 366; 48, p. 336; 1894. (Pt- Pt-Ir couples; formulae.) Holbornand Wien. Wied. Ann., 47, p. 107; 1892. 66, p. 360; 1895. Zeit. des Vereines deutscher Ingenieure, 41, p. 226; 1896. Stahl und Eisen, 16, p. 840. (Pt Pt-Rh couples.) Roberts-Austen. Recent progress in pyrometry. Trans. Am. Institute of Mining Engineers, 1893. (See also Recording Pyrometers.) E. Damour. Bull, de 1'Assoc. amicale des anciens eleves de 1'Ecole des Mines, March, 1889. H. Howe. Pyrometric data. Engineering and Mining Journal, 50, p. 426; 1890. Metallurgical laboratory notes (expts. with thermocouples), 1902. Holborn and Day. Pt alloys. Ann. d. Phys. (4), 2, p. 505; 1900. On the melting point of gold. Ann. d. Phys. (4), p. 99; 1901. Am. Jour. Sci., 11, p. 145; 1901. On the thermoelectric properties of certain metals. Sitz. Berl. Akad., p. 69; 1899. Am. Jour. Sci. (4), (8), p. 303; 1899. Mitth. Phys.-tech. Reichsanst, 37, 1899. Stansfield. Pt alloys, formulae. Phil. Mag. (5), 46, p. 59; 1898. Holman. Exponential formulae. Phil. Mag., 41, p. 465; 1896. Proc. Am. Acad., 31, p. 234. Holman, Lawrence, and Ban. Phil. Mag., 42, p. 37; 1896. Proc. Am. Acad., 31, p. 218. Schoentjes. Arch, de Phys. (4), 5, p. 136; 1898. Noll. Thermoelectricity of chemically pure metals. Wied. Ann., p. 874; 1894. Steinmann. Thermoelectricity of certain alloys. C. R., 130, p. 1300; 131, p. 34; 1900. Belloc. Thermoelectricity of steels. C. R., 131, p. 336; 1900. Ann. Chim. et Phys., 30, p. 42; 1903. C. R., 134, p. 105; 1902. D. Berthelot. On the graduation of couples. C. R., 134, p. 983; 1902. Thiede. Freezing-point apparatus. Zs. Angew. Ch., 16, p. 780; 1902. Lindeck and Rothe. The standardization of thermoelements for high-temper- ature measurements (as carried out at the Phys. Tech. Reichsanstalt). Zs. Instrument'k., 20, p. 285; 1900, and following years. BIBLIOGRAPHY 469 Nichols. Temperature of flames, etc., with thermocouple. Phys. Rev., 10, p. 234; 1900. Harker. Gas vs. thermoelectric scales. Phil. Trans., 203, p. 343; 1904. Chauvin and Arnoux. Compound couple. Bull. Soc. d'Encourag., 109, p. 1171; 1907. Crompton. Fe-Ni, Fe-Cu. Lond. Electrician, 66, p. 808; 1906. Bristol. Pyrometers. Am. Soc. Mech. Eng., 27, p. 552; 1906. Am. Machinist, p. 201; 1906. Electrochem. and Met. Ind., 4, p. 115; 1906. Pecheux. Le pyrome'tre thermoele"ctrique, Paris, 1909. (Ni-Cu, etc., couples.) C. R., 139, p. 1202; 1904. 148, p. 1041; 1909. 149, p. 1062; 1909. Broniewski. Thermoelectric properties of alloys (with a bibliography, 1822-1909). Rev. de Metallurgie, 7, p. 45; 1910. C. H. Wilson. Electrochem. and Met. Ind., 7, p. 116; 1909. Barrett. Phil. Mag. (5), 49, p. 309; 1900. (Fe-Ni-Mn to 1000 C.) Harrison. Fe-Ni-Cu alloys to 1050 C. Phil. Mag. (6), 3, p. 177; 1902. Boudouard. Steels o-i2ooC. Rev. d. M6tallurgie, 1, p. 80; 1904. Hevesy and Wolf. Ni-Ag couple. Phys. Zs., 11, p. 473; 1910. Stupakoff. Pamphlets on Industrial Practice. Schneider. Thermoelectric powers of heated wires. Elektrotech. Zs., 25, p. 233; 1904. Hirschson. C-Ni couple. Zs. Chem. Apparatk., p. 622; 1907. Sosman. Pt-Rh. Am. Jour. Sci., 30, p. i; 1910. Day and Allen. Pt-Rh to 1600 C. Phys. Rev., 19, p. 177; 1904. Palmer. Fe-Cu 0-200 . Phys. Rev., 21, p. 65; 1905. A. Campbell. Composite thermocouples. Phil. Mag., 9, p. 713; 1905. White. Constancy of thermoelements. Phys. Rev., 23, p. 449; 1906. Melting- point methods. Am. Jour. Sci., 28, p. 453; 1909. Geibel. The noble metals. Zs. Anorg. Ch., 69, p. 38; 1910. 70, p. 240; 1911. Waidner and Burgess. Comparison of thermoelectric and resistance scales. Bull. Bur. Standards, 6, p. 182; 1910. Memmler and Schob. Mitt. K. Material priifungsamt, 28, p. 307; 1910. (Use in testing machine.) CORRECTIONS FOR COLD JUNCTIONS. C. Ojferhaus and E. H. Fischer. Electrochem. and Met. Ind., 6, p. 362; 1908. R. Vogel. Zs. Anorg. Chem., 46, p. 13; 1905. Schultze and Koepsel. Centralb. Accumulatoren, 8, p. 102; 1907. POTENTIOMETERS FOR TEMPERATURE MEASUREMENTS. Fuessner. Zs. f. Instrumentenkunde, 10, p. 113; 1890. Lehrfeldt. Phil Mag., 6, p. 668; 1903. Varley. Brit. Assoc., 36, p. 14; 1866. Raps. Zs. f. Instrumentenkunde, 16, p. 215; 1895. Elektrotech. Zs., 16, p. 507; 1895. 24, p. 978; 1903. Franke. Elektrotech. Zs., 24, p. 978. Zs. f. Instrumentenkunde, 24, p. 93; 1904. Harker. Phil. Mag., 6, p. 41; 1903. 470 HIGH TEMPERATURES Holman. Phil. Mag., 42, p. 37; 1896. Lindeck. Zs. f. Instrumentenkunde, 20, p. 293; 1900. Hausrath. Ann. der Phys., 17, p. 735; 1905. Zs. f. Instrumentenkunde, 27, p. 309; 1907. Diesselhorst. Zs. f. Instrumentenkunde, 26, pp. 173, 297; 1906. 28, p. i; 1908. White. Zs. f. Instrumentenkunde, 27, p. 210; 1907. Phys. Rev., 25, p. 334; 1907. Wenner. Phys. Rev., 31, p. 94; 1910. ELECTRICAL RESISTANCE PYROMETER. Muller. Pogg. Ann., 103, p. 176; 1858. W. Siemens. Proc. Royal Soc., 19, p. 351; 1871. Bakerian Lecture, 1871. Trans- actions of the Society of Telegraph Engineers, 1879. British Association, p. 242; 1874. Benoit. C. R., 76, p. 342; 1873. Callendar.PhiL Trans, of R. S., 178, pp. 160-230; 1888. Proc. Roy. Soc. Lond., 41, p. 231; 1886. Phil. Trans., 1887. Phil. Trans., p. 119; 1892 (with Griffiths'). Platinum pyrometers. Iron and Steel Institute, May, 1892. Phil. Mag., 32, p. 104; 1891. 33, p. 220; 1892. Proposals for a standard scale of temperatures. B. A. Report, 1899; Phil. Mag., 47, pp. 191, 519; 1899, is a r6sume of the question. Phil. Trans., 199, p. i ; 1902. Steam temperatures. Br. Assoc. Rept., p. 422; 1897. (Ibid, with Nicholson) Proc. Inst. C. E., p. 131; 1898. Gas-engine Cylinders, with Dalby. Engineering (Lond.), 84, p. 887; 1907. Heycock and Neville. Determination of high temperatures. Jour, of Chem. Society, 67, pp. 160, 1024; 1895. Phil. Trans., 189, p. 25; 1897. 202, pp. 1-69. Earns. Amer. Jour. Sci. (3), 36, p. 427; 1888. Holborn and Wien. Ann. der Phys., 47, p. 107; 1892. 56, p. 360; 1895. Bull, de la Soc. d'encouragement, 5th Series, 1, p. 1012; 1896. Chappuis and Harker. A comparison of platinum and gas thermometers made at the B. Int. des Poids et Mesures. B. A. Report, 1899; Trav. et Mem. du Bureau Int. des Poids et Mesures, 1900, 1902; Phil. Trans., 1900; Jour, de Phys., 10, p. 20; 1901. Proc. Roy. Soc., 65, p. 377; 1899. Appleyard. Phil. Mag. (5), 41, p. 62; 1896. Dickson. Formulae Phil. Mag. (5), 44, p. 445; 1897. 45, p. 525; 1898. Wade. Wied. Beibl., 23, p. 963; 1899. Proc. Cambr. Soc., 9, p. 526; 1898. Waidner and Mallory. Phys. Rev., 8, p. 193; 1899. Barnes and Mclntosh. New form of platinum thermometer. Phil. Mag., 6, P- 3535 1903- Tory. Br. Assoc. Rpt, p. 588; 1897. Phil. Mag., 50, p. 421; 1900. Whipple. Temperature indicator, etc. Phys. Soc. (Lond.), 18, p. 235; 1902. Chree. Platinum Thermometry at the Kew Observatory. Proc. Roy. Soc., 67, p. 3. Harker. On the high-temperature standards of the National Physical Labora- tory. Proc. Roy. Soc., 73, p. 217; 1904. Jaeger. Precision of Resistance Methods. Zs. Instr'kunde, 26, p. 69; 1906. BIBLIOGRAPHY 471 Burstall. Rapidly varying temperatures. Phil. Mag., 40, p. 282; 1895. Prbc. Inst Mech. Engrs., p. 1031; 1901. Hopkinson. Gas-engine Temperatures. Engineering, 81, p. 777; 1906. Phil. Mag., 13, p. 84; 1907. H. Edwards. Contr. Jefferson Lab., 2, p. 549; 1904. Proc. Am. Acad., 40, No. 14. Marvin. Nickel to 300 C. Phys. Rev., 30, p. 522; 1910. Direct-reading Bridge for Ni and Pt. Jour. Frankl. Inst., 171, p. 439; 1911. Travers and Gwyer. Scale 444 to 190. Proc. Roy. Soc., 74, p. 528; 1905. A. Campbell. Direct Reading. Phil. Mag., 9, p. 713; 1905. Waidner and Burgess. Pt and Pd to 1 100 C. Bull. Bureau Standards, 6, pp. 149- 230; 1909. (Contains bibliography of about 135 titles.) Holborn and Henning. Pt w. Gas Therm, to S.B.P. Ann. der Phys. (4), 26, p. 835; 1908. 36, p. 761; 1911. Clark, Fisher, and Wadsworth. Pyrometer Bridge. Electrician, 60, p. 376; 1906. Harris. Deflectional Bridge Pyrometer. Electrician, 62, p. 430; 1908. Cambridge Co. Callendar and Griffiths Bridge. Electrician, 60, p. 477; 1906. Bruger. Hartmann and Braum Instrument. Elektrotech. Zs., 27, p. 531; 1906. E. F. Northrup. Measurement of Temperature by Electrical Means. Am. Inst. Elec. Eng., 25, pp. 219, 473; 1906. LAWS OF RADIATION. EARLY WORK. Newton. Opuscula Mathematica, 2, p. 417. Prevost. Sur 1'equilibre du feu, Geneva, 1792, 1809. Dulong and Petit. Ann. Chim. et Phys., 7, pp. 225, 337; 1817. Kirchhojf.- Pogg. Ann., 109, p. 275; 1860. Ann. Chim. et Phys., 69, p. 124; 1860. B. Stewart. Edinburgh Trans., 1858; Proc. Roy. Soc., 10, p. 385; 1860. Prowstaye and Desains. Ann. Chim. et Phys., 1860-1865. Draper. Am. Jour. Sci., 4, p. 388; 1847. Becquerel. C. R., 55, p. 826; 1862. Clausius. Pogg. Ann., 121, p. i; 1864. F. Rosetti. Phil. Mag. (5), 7, pp. 324, 438, 537; 1879. Violle. C. R., 88, p. 171; 1879. 92* PP- 866, 1204; 1881. 106, p. 163; 1887. Weber. Wied. Ann., 32, p. 256; 1887. Tail. Edinb. Proc., 12, p. 531; 1884. Tyndall. Phil. Mag., 28, p. 329; 1864. The laws of radiation and absorption. Me"m. by Prevost, Stewart, Kirchhoff, and Bunsen, edited by D. B. Brace. Am. Bk. Co., 1902. Ritchie. Mutual radiation. Pogg. Ann., 38, p. 378; 1866. RECENT WORK. Kayser. Handbuch der spectroscopie, 2, 1902. (Contains an admirable sum- mary of the work on radiation.) Drude. Theory of Optics (Engl. trans, pub. by Longmans). 472 HIGH TEMPERATURES Stefan. On the relation between heat radiation and temperature. Wien. Ber., 79, B. 2, p. 391; 1879. Schleirmacher. On Stefan's law. Wied. Ann., 26, p. 287; 1885. Boltzmann. Deduction of Stefan's law. Wied. Ann., 22, p. 291; 1884. Paschen. On the emission of heated gases. Wied. Ann., 50, p. 409; 1893. 51, p. i; 1894. 52, p. 209; 1894. On the emission of solids. Wied. Ann., 49, p. 50; 1893. 58, p. 455; 1896. 60, p. 662; 1897. Astrophys. Jour., 2, p. 202; 1895. On black-body radiation. Wied. Ann., 60, p. 719; 1897. Berl. Ber., p. 959; 1899. Ann. der Phys., 4, p. 277; 1901. 6, p. 646; 1901. Paschen and Wanner. On a photometric method, etc. Berl. Ber., p. 5; 1899. Wanner. Photometric measurement of black- body radiation. Ann. der Phys., 2, p. 141; 1900. Fery. Radiation from oxides. Ann. Chim. Phys. (7), 27, p. 433; 1902. Petavel. Heat dissipated by platinum, etc. Phil. Trans., 191, p. 501; 1898. 197, p. 229; 1901. Milliken. Polarization of light from incandescent surfaces. Phys. Rev., 3, p. 177; 1895. Langley. Distribution of energy in solar spectrum, etc. Am. Jour. Sci., 31, p. i; 1886. 36, p. 367; 1888. Phil. Mag. (5), 26, p. 505; 1888. Wilson and Gray. Temperature of arc and sun. Proc. Roy. Soc., 66, p. 250; 1894. 68, p. 24; 1895. W. Michelson. Theoretical study of the distribution of energy in the spectra of solids. Jour, de Phys. (2), 6, p. 467; 1887. Phil. Mag. (5), 25, p. 425; 1888. Jour. Russian Phys.-chem. Soc., 34, p. 155; 1902. Violle. The radiation of incandescent bodies. Jl. de Phys. (3), 1, p. 298; 1892. C. R., 114, p. 734; 116, p. 1273, 1892. Radiation from refractory bodies heated in the electric furnace. C. R., 117, p. 33; 1893. St. John. On the equality of light emissivities at high temperatures, etc. Wied. Ann., 66, p. 433; 1895. Kurlbaum. On the new platinum light unit of the Phys. Tech. Reichsanstalt. Verh. Phys. Ges. (Berlin), 14, p. 56; 1895. Larmor. On the relations of radiation to temperature. Nature, 62, p. 562; 1000. 63, p. 216. Planck's eq. Proc Roy. Soc., 83, p. 81; 1909. Theory. Phil. Mag., 20, p. 353; 1910. Guittaume. -The laws of radiation and the theory of incandescent mantles. Rev. Ge"n. des. Sci., 12, pp. 358, 422; 1901. Wien. Black- body radiation and the second law of thermodynamics. Berl. Ber., P- 555 1893. Temperature and entropy of radiation. Wied. Ann., 62, p. 132; 1894. The upper limit of wave lengths in the radiation of solid bodies, etc. Wied Ann., 46, p. 633; 1893. 62, p. 150; 1894. On the partition of energy in the emission spectrum of a black body. Wied. Ann., 68, p. 662; 1896. On the theory of radiation from a black body. Ann. der Phys., 3, P- 53o; 1900- Paris Cong. Rpts., 2, p. 23; 1900. Wien and Lummer. Method of demonstrating the radiation law for an abso- lutely black body. Wied. Ann., 66, p. 451; 1895. Beckmann. Black-body radiation, etc. Thesis, Tubingen, 1898. Lummer. On the gray glow and red glow. Wied. Ann., 62, p. 13; 1897. Radi- BIBLIOGRAPHY 473. ation from black bodies. Paris Cong. Rpts., 2, p. 56; 1900. Arch. Math. Phys. (3), 2, p. 157; 1901. 3, p. 261; 1902. Lummer and Jahnke. Ann. der Phys., 3, p. 283; 1900. Compau. Radiation laws. Ann. Chim. Phys. (7), 26, p. 488; 1902. Lummer and Pringsheim. The radiation from a black body between 100 and 1300 C. Wied. Ann., 63, p. 395; 1897. Distribution of energy in spectrum of black body. Verh. Deutsche Phys. Ges., 1, pp. 23 and 215; 1899. Infra- red radiation. Verh. Deutsche Phys. Ges., 2, p. 163; 1900. On black radi- ation. Ann. der Phys., 6, p. 192; 1901. The theoretical radiation scale, etc., at 2300 absolute. Verh. Deutsche Phys. Ges. (5), 1, p. 3; 1903. Lummer and Kurlbaum. Electrically treated black body and its temperature measurement. Verh. Phys. Ges. (Berlin), 17, p. 106; 1898. Ann. der Phys., 6, p. 829; 1901. On the change of photometric intensity with temperature. Verh. Deutsche Phys. Ges., 2, p. 89; 1900. Pringsheim. Deduction of Kirchhoff's law. Verh. Deutsche Phys. Ges., 3, p. 81 ; 1901. j Emission from gases. Paris Cong. Rpts., 2, 1900. Bottomley. Radiation from solids. Phil. Mag. (6), 4, p. 560; 1902. Planck. Entropy and temperature of radiant heat, etc. Ann. der Phys., 1, p. 719; 1900. Sitzber. Berl. Akad., 1, p. 440; 1899. On irreversible radia- tion. Ann. der Phys., 1, p. 69; 1900. On a bettering of Wien's spectral equation. Verh. Deutsche Phys. Ges., 2, p. 202; 1900. Ann. der Phys., 4, p. 553; 1901. Energy distribution in normal spectrum, Verh. D. Phys. Ges., 2, p. 237; 1900. Theory. Ann. der Phys., 31, p. 758; 1910. Rayleigh. Phil. Mag. (5), 49, p. 539; 1900. (6) 1, p. 98; 1901. Thiesen. Verh. Deutsche Phys. Ges., 2, p. 65; 1900. Rubens. Infra-red radiation. Wied. Ann., 69, p. 576; 1899. Rubens and E. Nichols. Wied. Ann., 60, p. 418; 1897. Rubens and Kurlbaum. Experimental verifications. Berl. Ber., p. 929; 1900. Ann. der Phys., 4, p. 649; 1901. Astrophys. JL, 14, p. 335; 1901. Ann. der Phys., 4, p. 649; 1904. Maxwell. Pressure of radiation. Electricity and Magnetism, chap, on Electro- magnetic theory. Bartoli. Wied. Ann., 22, p. 31; 1884. Galitzine. Wied. Ann., 47, p. 479; 1892. Pellat. JL de Phys., July, 1903. Lebedew. Ann. de Phys., 6, p. 433; 1901. Paris Cong. Rpts., 2, p. 133; 1900. E. F. Nichols and Hull. Phys. Rev., 901 and 1903; Astrophys. Jl., 17, p. 315; 1903. Rayleigh. Phil. Mag. (6), 3, p. 338; 1902. Day and Van Orstrand. The black body and the measurement of extreme tem- peratures. Astrophys. JL, 19, p. i; 1904. C. Mendenhall and Saunders. Astrophys. Jl., 13, p. 25; 1901. Rase h. On the photometric determination of temperatures, etc. Ann. der Phys., 13, p. 193; 1904. G. W. Stewart. Spectral-energy curves of black body at room temperature. Phys. Rev., 16, p. 306; 1902. 17, p. 476; 1903. Cotton. Kirchhoff's Law. Rev. Ge"n. des Sci., Feb. 15, 1899. Astrophys. JL, 9, p. 237; 1899. 474 HIGH TEMPERATURES Angstrom. Gaseous Absorption. Ann. der Phys., 6, p. 163; 1901. Bahr. Ibid., 29, p. 780; 1909. Valentiner. Stefan's law and gas scale to 1600 C. Ann. der Phys. (4), 31, p. 275; 1910. V. Wartenberg. Emissive powers vs. temperature. Verb. Phys. Ges., 12, p. 121; 1910. Holborn and Henning. Emissive powers. Berl. Akad. Ber., p. 311; 1905. Burgess. Radiation from Cu. Bull. Bureau Standards, 6, p. in; 1909. Poynting. Radiation in the Solar System. Phil. Trans., 202, p. 525; 1903. Fery. Ann. Chim. et Phys., 17, p. 267; 1909. Rev. Gen. d. Sci., Sept., 1903. Stefan's law. C. R., 148, pp. 915, 1150; 1909. Conical Receivers. C. R., 148, p. 777; 1909. Selective Receivers. C. R., 148, p. 1043; I 99- Nernst. From Gases. Phys. Zs., 6, p. 777; 1904. Holborn and Henning. Light Emission and Melting Points. Berlin Sitz. Ber., 12, p. 311; 1905. Lucas. Pt Radiation. Phys. Zs., 6, p. 418; 1905. Burgess. Pt Radiation. Bull. Bureau Standards, 1, p. 443; 1905. Hertzsprung. Radiation and Luminous Equivalent. Zs. wiss. Photographic, 4, p. 42; 1906. Jeans. Theory. Proc. Roy. Soc., 76, p. 545; 1905. Phil. Mag., 12, p. 57; 1906. 17, p. 229; 1909. Lummer and Pringsheim. Welsbach Mantle. Phys. Zs., 7, p. 89; 1906.- Lorentz- Jeans Formula. Phys. Zs., 9, p. 449; 1908. Swinburne. Proc. Phys. Soc., 20 f p. 33; 1906. Engineering (Lond.), 82, p. 217; 1906. Cantor. Radiation and Doppler's Principle. Ann. der Phys., 20, p. 333; 1906. Thomson (J. /.). Electrical Origin of Radiation. Phil. Mag., 14, p. 217; 1907. Theory. Phil. Mag., 20, p. 238; 1910. Lorentz. Theory. N. Cimento, 16, p. 5; 1908. Richardson. lonization and Temperature. Phys. Rev., 27, p. 183; 1908. Coblentz. Radiation Constants. Bull. Bureau Standards, 5, p. 339; 1909. Reflecting Power of Metals. Jl. Frank. Inst, Sept., 1910. Hagen and Rubens. Emissivity and Temperature. Sitz. Berlin Akad., 16, p. 478; 1909. 23, p. 467; 1910. H. A. Wilson. Theory. Proc. Roy. Soc., 82, p. 177; 1909. Phil. Mag., 20, p. 121; 1910. Pokrovzkij. Spectrophotometric Relations. Jl. Russ. Phys.-chem. Soc., 41, p. 73; 1909. Polara. Theory. Accad. Lincei, 18, p. 513; 1909. Bauer and Moulin. Stefan's Law. C. R., 149, p. 988; 1909. C. R., 160, p. 167; 1910. Einstein. Theory. Phys. Zs., 10, p. 185; 1909. Saurel. Theory. Phys. Rev., 30, p. 350; 1910. Fery and Drecq. Stefan's Law. Jl. de Phys. (5), 1, p. 551; 1911. Parmentier. Stefan's Law. Ann. Ch. et Phys., 22, p. 417; 1911. Warburg and Leithauser. Wien's law. Sitz. Berlin Akad., p. 925; 1910. P. T. R. Thatigkeit, 1910. BIBLIOGRAPHY 475 Humphreys. Summary. Astrophys. Jl, 31, p. 281, 1910. Violle. Report on Radiation. Ann. Ch. et Phys. (8), 2, p. 134; 1904. Drysdale. Luminous Efficiency of Black Body. Jl. de Phys. (4), 8, p. 197; 1909. RADIATION PYROMETER. Violle. Solar radiation. Ann. Chim. et Phys. (5), 20, p. 289; 1877. Jour, de Phys., p. 277; 1876. Rosetti. Ann. Chim. et Phys., 17, p. 177; 1879. Phil. Mag., 18, p. 324; 1879. Deprez and d'Arsonval. Socie"te de Physique, Feb. 5, 1886. . Boys. Radiomicrometer. Phil. Trans., 180, p. 159; 1887. Wilson and Gray. Temperature of the sun. Phil. Trans., 186, p. 361; 1894. Langley. Bolometer. Am. Jour. Sci., 21, p. 187; 1881. 31, p. i; 1886. 32, p. 90; 1886. (4), 6, p. 241; 1898. Jour, de Phys., 9, p. 59. Terreschin. Diss. St. Petersburg, 1898. Jour. Russ. Phys.-chem. Ges., 29, pp. 22, 169, 277; 1897. Petavel. Proc. Roy. Soc., 63, p. 403; 1898. Phil. Trans., 191, p. 501; 1898. Abbot. Bolometer. Astrophys. Jour., 8, p. 250; 1898. Belloc. Bolometer errors. L'e"clair. 61ec. (5), 15, p. 383; 1898. Scheiner. Radiation and temperature of sun (Leipzig), 1899. Warburg. Temperature of sun. Verh. D. Ges. (i), 2, p. 50; 1899. Fery. The measurement of high temperatures and Stefan's law. C. R., 134, p. 977; 1902. Jl. de Phys., Sept., 1904. Rubens. Sensitive Thermopile. Zs. Instr'k., 18, p. 65; 1898. Ftry. C. R., 134, p. 977; 1902. Spiral Pyrometer. Engineering (Lond.), 87, p. 663; 1909. Bull. Soc. Franc. Phys., p. 186; 1907. Coblentz. Radiation Instruments. Bull. Bureau Standards, 4, p. 391; 1908. Foster. Trans. Am. Electroch. Soc., 17, p. 223; 1910. Angstrom. Pyrheliometer. Astrophys. Jl., 9, p. 232. Solar Research Int. Union, 1, p. 178. Callendar. Absolute Bolometer. Proc. Roy. Soc., 77, p. 6; 1905. Phys. Soc. Lond., May 12, 1905; July 8, 1910. Kimball. Pyrheliometers. Bull. Mt. Weather Obs., 1, p. 83; 1908. Abbot and Fowle. Pyrheliometers and Bolometers. Annals Smithsonian Astro- phys. Obs., 2, 1908. Abbot. Silver-disc Pyrheliometer. Smithsonian Misc. Coll., 66, p. i; 1911. Thwing. JL Frank. Inst., May, 1908. OPTICAL PYROMETER. See also laws of radiation under Lummer, Pringsheim, Kurlbaum, Wanner, Wien, Rasch, Fe"ry, etc. Kirehhojf. Ann. Chim. et Phys., 59, p. 124; 1860. Ed. Becquerel. Optical measurement of temperatures. C. R., 66, p. 826; 1863. Ann. Chim. et Phys., 68, p. 49; 1863. Violle. Radiation from platinum. C. R., 88, p. 171; 1879. 91, PP- 866, 1204;. 1881. 476 HIGH TEMPERATURES Kurlbaum and Schulze. Pyrometric examination of Nernst lamps. Verb. D. Phys. Ges. (5), 24, p. 425; 1903. D. Berthdot. On a new optical method, etc. Ann. Chim. et Phys. (7), 26, p. 58; 1902. Lummer. Photometric pyrometer. Verb. D. Phys. Ges., p. 131; 1901. Fery. Temperature of the arc. C. R., 134, p. 1201; 1902. Absorption Pyrom- eter. Jl. de Phys. (4), 3, p. 32; 1904. Le Chatdier Pyrometer. On the measurement of high temperatures. C. R., 114, pp. 214-216; 1892. Jl. de Phys. (3), 1, pp. 185-205; 1892. Industrie ^lectrique, April, 1892. On the temperature of the sun. C. R., 114, pp. 737-739; 1892. On the temperatures of industrial furnaces. C. R., 114, pp. 470-473; 1892. Introduction to Metallurgy (Roberts-Austen), 1903. i Discussion of Le Chatelier's method: (Violle) C. R., 114, p. 734; 1892. Jl. de. Phys. (3), 1, p. 298; 1892. (Becquerd) C. R., 114, pp. 225 and 390; 1892. (Le Chatdier) C. R., 114, p. 340; 1892. (Crova) C. R., 114, p. 941; 1892. Kayser's Spectroscopy, 1902. Crova' s Pyrometer. Spectrometric study of certain luminous sources. C. R., 87, pp. 322-325; 1878. On the spectrometric measurement of high tem- peratures. C. R., 87, pp. 979-981; 1878. Study of the energy of radiations emitted by calorific and luminous sources. Jour, de Phys., 7, pp. 357-363, 1878. Spectrometric measurement of high temperatures. Jour, de Phys.; 9, pp. 196-198; 1879. C- R., 90, pp. 252-254; 1880. Ann. Chim. et Phys. (5), 19, pp. 472-550; 1880. Photometric comparison of luminous sources of different lines. C. R., 93, pp. 512-513; 1881. Solar photometry. C. R., 96, pp. 1271-1273; 1882. End of Spectrum Method. A. Crova. Study of the energy of radiations emitted by calorific and luminous sources. Jour, de Phys., 7, pp. 357-363; 1878. W. Hempel. On the measurement of high temperatures by means of a spectral apparatus. Zs. f. Angewandte Chem., 14, pp. 237-242; 1901. Wanner Pyrometer. A. Konig. A new spectral photometer. Wied. Ann., 63, p. 785; 1894. Description of Wanner instrument. Iron Age, Feb. 18, p. 24; 1904. Phys. Zs., 3, pp. 112-114; 1902. Stahl und Eisen, 22, pp. 207- 211; 1902. Zs. Vereines Deut. Ing., 48, pp. 160, 161; 1904. Wanner. Pho- tometric measurement of the radiation of black bodies. Ann. der Phys., 6, pp. 141-155; 1900. Martens and Grinbaum. Improved form of Konig spectrophotometer. Ann. der Phys., 5, p. 954; 1903. Base. Measurements with Wanner pyrometer. Zs. Anorg. Chem., 15, p. 715; 1902. Nernst and Wartenberg. Verh. Phys. Ges., 8, p. 146; 1906. Low- temperature pyrometer. Jl. Gasbel., 60, p. 1005; 1907. HolbornandKurlbatimPyrometer. Sitzber. Berlin Akad., June 13, pp. 712-719; 1901. Ann. der Phys., 10, p. 225; 1902. See also laws of radiation; melting points. Morse Thermogage. American Machinist, 1903. Henning. Spectral pyrometer. Zs. Instr'kunde, 30, p. 61; 1910. Waidner and Burgess. Arc Temperature. Phys. Rev., 19, p. 241; 1904. Bull. Bureau Standards, 1, p. 109; 1904. Optical Pyrometry. Phys. Rev., 19, p. 422; 1904. Bull. Bureau Standards, 1, p. 189; 1905. Radiation from Pt and Pd. Bull. Bureau Standards, 3, p. 163; 1907. BIBLIOGRAPHY 477 Holborn. Engineering (Lond.), 84, p. 345; 1907. Brit. Assoc. Rpt, p. 440; 1907. Nernst. Photometric. Phys. Zs., 7, p. 380; 1906. Holborn and Valentiner. Optical vs. Gas to 1600 C. Ann. der Phys. (4), 22, p. i; 1907. Ffry. Vacuum-tube Temperatures. Soc. Franc. Phys., p. 305; 1907. Jl. de Phys., 6, p. 979; 1907. Kurlbaum and Schulze. Ibid. Berichte D. Phys. Ges., 1906. Pirani. Phys. Ges. Verh., 12, pp. 301, 1054; 1910. 13, p. 19; 1911. C. E. Mendenhall. Phys. Rev., 33, p. 74; 1911. H. W. Gillett.JL Phys. Chem., 15, p. 213; 1911. Thuermel. Ann. de Phys., 33, p. 1139; 1910. Wartenberg. Metal temperatures. Verh. D. Phys. Ges., 12, p. 121; 1910. EXPANSION AND CONTRACTION PYROMETERS. Wedgwood. Phil. Trans., 72, p. 305; 1782. 74, p. 358; 1784. Weinhold. Pogg. Ann., 149, p. 186; 1873. Boh. Die Pyrometer, Berlin, 1888. Guyton and Morveau. Ann. Chim. et Phys., ist Series, 46, p. 276; 1803. 73, p. 254; 1810. 74, pp. 18, 129; 1810. 90, pp. 113, 225; 1814. The Meldometer. /. Joly. Proc. Roy. Irish Academy (3), 2, p. 38; 1891. Ram- say and Eumorfopoulos. Phil. Mag., 41, p. 360; 1896. High-range Mercury Thermometers. Weber. Ber. Berlin Akad., Dec., 1903. Wiebe, ibid., July, 1884, Nov., 1885; Zs. Instr'kunde, 6, p. 167; 1886. 8, P- 3735 l888 - 10> P- 2 7; l8 9- Schott, ibid., 11, p. 330; 1891. Hovestadt. Jenaer Glas (trans, by J. D. and A. Everett). Marchis. Modifications per- manentes du verre, etc. Dickinson. Bull. Bureau of Standards, 2, p. 189; 1906. Guillaume. Treatise on Thermometry of Precision. (Gauthier Villars, 1889.) Fischer and Bobertag. Glass Therms, for High Temps. Zs. f. Elek. Ch., 14, p. 375; 1908. Bureau of Standards. Circ., 8, 2nd Ed., 1911. Quartz Thermometers. Dufour. C. R., 188, p. 775; 1900. Siebert. Zs. Elek- troch. (Halle), 10, p. 26. FUSIBLE-CONE PYROSCOPE. Lauth and Vogt. Pyrometric measurements. Bull. Soc. Chim., 46, p. 786; 1886. Seger. Tonindustrie Zeitung, p. 121; 1885. Pp. 135, 229; 1886. Tonindustrie Zs., No. 49, 1893; No. 52, 1895; No. 119, 1908. Stahl u. Eisen, p. 440; 1909. P. 1505; 1910. A. Beranger. Materiaux et Produits Refractaires (Paris, 1910). (Very complete account.) Zs. Angew. Chem., p. 49; 1905. Sprechsaal, pp. 118, 156, 391, 483; 1907. P. 1284; 1906. P. 561; 1908. Rothe. Tonindustrie Zs., 30, p. 1473; 1906. 31, p. 1365; 1907. Hoffmann. Tonindustrie Zs., 33, p. 1577; 1909. Cramer and Hecht. Tonindustrie Zs., No. 18, 1896. Seger's Schriften. 47$ HIGH TEMPERATURES Simonis. Tonindustrie Zs., 31, p. 146; 1907. 32, p. 1764; 1908. H. O. Hofmann. Trans. Am. Inst. Mining Eng., 24, p. 42. Thatigkeit P. T. Reichsanstalt for 1909, 1910. PYROMETERS BASED ON FLOW OF FLUIDS. A. Job. Viscosity pyrometer. C. R., 134, p. 39; 1902. U filing and Steinbart. Stahl u. Eisen, 1899. Carnelly and Burton. Water circulation. Jl. Chem. Soc. (Lond.), 45, p. 237; 1884. Also described in Sir Roberts-Austen's Metallurgy, 1902. Barus. Bull. 54, Geolog. Survey, 1889. E. A. Uehling. Proc. Cleveland Inst. Engrs., Jan., 1900. Electroch. and Met. Ind., 3, p. 160; 1905. RECORDING PYROMETERS. Le Chatelier. Study of clays. C. R., 104, p. 144; 1887. - Quenching of Steels. Rev. de M6tallurgie, 1, p. 134; 1904. Roberts- Austen. First Report to the Alloys Research Committee. Proc. Inst. Mech. Engrs., p. 543; 1891. Nature, 45, 1892; B. A. Report, 1891; Jour, of Soc. of Chem. Industry, 46, p. i; 1896. Proc. Inst. Mech. Eng., p. 269; 1895. Pp. 67, 243; 1897. Proc. Roy. Soc., 49, p. 347; 1891. Fifth Report Alloys Research Committee, 1899. Proc. Inst. Mech. Eng., p. 35; 1899. Metal- lographist, 2, p. 186; 1899. G. Charpy. Study of the hardening of steel. Bull, de la Soc. d'encouragement (4), 10, p. 666; 1895. Callendar. Platinum recording pyrometer. Engineering, May 26, p. 675, 1899. Phil. Mag., 19, p. 538; 1910. Stansfield. Phil. Mag. (5), 46, p. 59; 1898. Phys. Soc. London (2), 16, p. 103; 1898. Bristol. Air pyrometer. Eng. News, Dec. 13, 1900. Saladin. Iron and Steel Metall. and Metallog., 7, p. 237; 1904. Queen and Co. Electroch. and Met. Ind., 3, p. 162; 1905. Siemens and Halske. Zs. Instrumentenk., 24, p. 350; 1904. 25, p. 273; 1905. Einthoven. Arch. Neerland, 10, p. 414. Proc. Acad. Sci. Amsterdam (2), 6,, p. 707; 1904. Benedicks. Iron and Steel Inst., II, p. 153; 1908. Le Chatelier. Rev. de Metallurgie, 1, p. 134; 1904. Wologdine. Rev. d. Metallurgie, 4, p. 552; 1907. Kurnakow. Zs. Anorg. Ch., 42, p. 184; 1904. Schmidt. Chem. Eng., 6, p. 80; 1907. Harkhort. Metallurgie, 4, p. 639; 1907. Bristol. Trans. Am. Inst. Mech. Engrs., 22, No. 874, p. 143. Dejean. Rev. de M6tallurgie, 2, p. 701; 1905. 3, p. 149; 1906. Burgess. Methods of Obtaining Cooling Curves. Bull. Bureau Standards, 6, p. 199; 1908. Bruger. Phys. Zs., p. 775; 1906. BIBLIOGRAPHY 479 Brown. Electroch. and Met. Ind., 7, p. 329; 1909. Northrup. Proc. Am. Electroch. Soc., May, 1909. Rengade. Bull. Soc. Chim., 7, p. 934; 1909. Hayes. Proc. Am. Acad., 47, p. 3; 1911. VARIOUS PYROMETRIC METHODS. Quincke. Acoustic Thermometer. Ann. d. Phys., 63, p. 66; 1897. Krupp's Hot-blast Pyrometer. Von Bergen. Jl. Iron and Steel Inst., 1, p. 207; 1886. A. H. Sexton. Fuel and Refractory Materials. Chapter on Pyrometry; several industrial forms described. T. Hurter. Industrial Air Pyrometer. Jl. Soc. Chem. Ind., p. 634; 1886. WiborgWs Pyrometer. Jl. Iron and Steel Institute, 2, p. no; 1880. Fournier's Vapor Pressure Thermometers. Engineering (Lond.) 89, p. 447 ; 1910. MELTING POINTS. METALS. Prinsep. Ann. Chim. et Phys. (2), 41, p. 247; 1829. Lauth. Bull. Soc. Chim., Paris, 46, p. 786; 1886. E. Becquerel. Ann. Chim. et Phys. (3), 68, p. 497; 1863. Viotte. C. R., 86, p. 543; 1877. 87, p. 981; 1878. 89, p. 702; 1879. Holborn and Wien. Wied. Ann., 47, p. 107; 1892. 66, p. 360; 1895. Also Zs. fur Instr'k., p. 257; 1892. Holborn and Day. Ann. d. Phys. (i), 4, p. 99; 1901. Am. Jour. Sci., 11, p. 145; 1901. Wied. Ann., 68, p. 817; 1899. Am. Jour. Sci. (4), 8, p. 165; 1899. Ehrhardt and Schertel. Jahrb. fur das Berg- und Huttenw. im K. Sachsen, p. 154; 1879. Callendar. Phil. Mag. (5), 47, p. 191; 1899. 48, p. 519. Curie. Ann. de Chim. et de Phys. (5), 6, 1895. Earns. Bull. 54, U. S. Geological Survey, 1889. Behandlung u. Messung hoher Temp. Leipzig, 1892. Am. Jour. Sci. (3), 48, p. 332; 1894. Berthelot. C. R., 126, Feb., 1898. Le Chatelier. C. R., 114, p. 470; 1892. Moldenke. Zs. fur Instr'k., 19, p. 153; 1898. (Iron and steel.) Cusack. Proc. Roy. Irish Acad. (3), 4, p. 399; 1899. Landolt and Bornstein. Phys. Chem. Tabellen, Berlin. Carnelly. Melting and Boiling Point Tables, London, 1885. Smithsonian Physical Tables, 5th ed., 1910. . Holman, Lawrence, and Ban. Phil. Mag. (5), 42, p. 37; 1896. Proc. Am. Acad., 31, p. 218. Heycock and Neville. Phil. Trans., 189, p. 25. Jour. Chem. Soc., 71, p. 333; 1897. Nature, 66, p. 502; 1897. Chem. News, 76, p. 160; 1897. Heraus. Manganese. Zs. Elecktroch., 8, p. 185; 1902. Nernst. Zs. Elektrotech., 1903. 480 HIGH TEMPERATURES Rase h. Ann. d. Phys., 1904. D. Berthdot. Ann. Chim. et Phys., 1902. Gold. C. R., 138, p. 1153; 1904. Richards. Application of phase rule to Cu, Ag, Au. Am. Jl. Sci. (4), 13, p. 377; 1902. Jaquerod and Perrot. Gold. C. R., 138, p. 1032; 1904. Pirani and Meyer. Ta. Verh. D. Phys. Ges., 13, p. 540; 1911. Fery and Cheneveau. Pt. C. R., 148, p. 401; 1909. Carpenter. Iron. Iron and Steel Inst. III., p. 290; 1908. Lewis. Cr. Chem. News, 86, p. 13; 1902. Holborn and Henning. Sn, Cd, Zn. Ann. d. Phys., 35, p. 761; 1911. Day and Allen. Phys. Rev., 19, p. 177; 1904. Day and Clement. Am. Jl. Soc., 26, p. 405; 1908. Day and Sosman. Zn to Pd. Am. Jl. Sci., 29, p. 93; 1910. Tammann and Associates in Zs. Anorg. Chem. Metals and Binary Alloys. Tammann. Pressure influence on Sn and Bi. Zs. Anorg. Ch., 40, p. 54; 1904. Johnston and Adams. Ibid. Amer. Jl. Sci., 31, p. 501; 1911. Arndt. Review. Verein. z. Beford. d. Gewerbefleisses, Verhandt., p. 265; 1904. Barker. Pt and Ni. Proc. Roy. Soc., 76, p. 235; 1905. Nernst and Wartenberg. Pd and Pt. Phys. Ges. Verh., 8, p. 48; 1906. Holborn and Valentiner. Pd and Pt. Sitzber. Berlin Akad., 44, p. 811; 1906. Waidner and Burgess. Pd and Pt. Bull. Bureau Standards, 3, p. 163; 1907. Pt. C. R., May 3, 1909. W and Ta. Jl. d. Phys., 6, p. 830; 1907. Reproducibility of Metal M.P.'s. Bull. Bureau Standards, 6, p. 149; 1909. Burgess. Iron Group. Bull. Bureau Standards, 3, p. 345; 1907. Chemical Elements. Jl. Wash. Acad. Sci., 1, p. 16; 1911. Dejean. Copper. Rev. d. Metallurgie, 3, p. 149; 1906. Mendenhall and Ingersoll. Pd, Rh, Ir on Nernst Glower. Phys. Rev., 26, p. i; 1907. Locke. A wire method. Zs. Elektroch., 13, p. 592; 1907. Wartenberg. W. Chem. Ber., 40, p. 3287; 1907. Refractory Metals. Phys. Ges. Verh., 12, p. 121; 1910. Ruff. Refractory Metals and Oxides. Chem. Ber., 43, p. 1564; 1910. Bolton. Niobium. Zs. Elek. Ch., 13, p. 145; 1907. SALTS AMD MISCELLANEOUS. Carnelly.J. C. S. Trans., p. 489; 1876. P. 365; 1877. P. 273; 1878. Le Chatelier. Bull. Soc. Chim., 47, p. 301. C. R., 118, pp. 350, 711, 802. V. Meyer, Riddle, and Lamb. Ch. Ber., 27, p. 3129; 1894. MacCrae. Ann. d. Phys., 66, p. 95; 1894. Hutner and Tammann. Zs. Anorg. Ch., 43, p. 215; 1905. Ruff and Plato. Ch. Ber., 36, p. 2357; 1903. Joly. Proc. Irish Acad., No. 2, 1891. Ramsay and Eurmorfopoulos. Phil. Mag., 41, p. 360; 1896. Day and Sosman. Am. Jl. Sci., 31, p. 341; 1911. Grenet. Rev. de Metallurgie, 7, p. 485; 1910. BIBLIOGRAPHY 481 Brearly and Morewood. Sentinel Pyrometers. Jl. Iron and Steel Inst., 73, p. 261; 1907. Liebknecht and Nilsen. Ch. Ber., 36, p. 3718; 1903. Watson. Phys. Rev., 26, p. 198; 1908. C. H. Burgess and A. Holt. Rapid for borates, etc. Proc. Roy. Sri. Lond., Nov. 24, 1904. W. C. Herceus. Ceramics. Zs. Angew. Chem., p. 49; 1905. Doelter. Silicates. Zs. Elektroch., 12, p. 617; 1906. Lampen. Porcelains, etc. Jl. Am. Ch. Soc., 28, p. 846; 1906. White. Thermoelectric manipulation. Am. Jl. Sci., 28, p. 453; 1909. Boudouard. Slags. Jl. Iron and Steel Inst. (i), 67, p. 350; 1905. BOILING POINTS. Earns. (See Melting Points.) Am. Jour. Sci. (5), 48, p. 332; 1894. Troost. C. R., 94, p. 788; 1882. 94, p. 1508; 1882. 96, p. 30; 1882. Le Chatelier. C. R., 121, p. 323; 1895. (See also under Thermoelectric Pyrom- eter.) Berthelot. Stances de la soc. de physique, Paris, Feb., 1898, and Bull, du Museum, No. 6, p. 301; 1898. Callendar and Griffiths. Proc. Roy. Soc. London, 49, p. 56; 1891. >Chappuis and Barker. Travaux et M6m. du Bureau Int. des Poids et Mesures, 12, 1900; Phil. Trans., 1900. Preyer and V. Meyer. Zeits. fur Anorg. Chem., 2, p. i; 1892. Berl. Ber., 25, p. 622; 1892. S. Young. Trans. Chem. Soc., p. 629; 1891. MacCrae. Wied. Ann., 66, p. 95; 1895. Callendar. Phil. Mag. (5), 48, p. 519; 1899. (Fusion also.) D. Berthelot. Ann. Chim. et Phys., 1902. Fery. Cu and Zn. Ann. Chim. et Phys. (7), 28, p. 428; 1903. R. Rothe. Sulphur. Zs. Instrumk., 23, p. 364; 1903. Greenwood. Of metals. Proc. Roy. Soc., A 82, p. 396; 1909. 83, p. 483; 1910. Chem. News, 104, pp. 31, 42; 1911. Kraft and Merz. S, Se, Te at reduced pressure. Chem. Ber., 36, p. 4344J 1903- Ruff and Johannsen. Alkali metals. Chem. Ber., 38, p. 3601; 1905. Harker and Sexton. Pressure change of S.B.P. Elect. Rev., 63, p. 416; 1908. Eumorfopoulos. Sulphur. Proc. Roy. Soc., 81, p. 339; 1908. Callendar and Moss. Sulphur. Proc. Roy. Soc., 83, p. 106; 1909. Holborn and Henning. Sulphur, etc. Ann. d. Phys. (4), 35, p. 761; 1911. Waidner and Burgess. Sulphur. Bull. Bureau Standards, 7, p. 127; 1911. Wartenberg. Metals. Zs. Anorg. Chem., 66, p. 320; 1908. Moissan. Metals. Ann. Ch. et Phys. (8), 8, p. 145; I 9 6 - Watts. Metals. Trans. Am. Electroch. Soc., 12, p. 141; 1907. Smith and Menzies. Hg. Jl. Am. Ch. Soc., 32, p. 1434; 1910. Jaquerod and Wassmer. Benzophenone and Naphthaline. Jl. d. Ch. et de Phys., 2, p. 52; 1904. Waidner and Burgess. Ibid. Bull. Bureau Standards, 7, p. 3; 1910. 482 HIGH TEMPERATURES PYROMETRIC MATERIALS. PORCELAIN: EXPANSION. Deville and Troost. C. R., 67, p. 867; 1863. Bedford. B. A. Report, 1899. Benefit. Trav. et Mem. du Bureau Int., 6, p. 190. Tutton. Phil. Mag. (6), 3, p. 631; 1902. Chappuis. Phil. Mag. (6), 3, p. 243; 1902. Holborn and Day. Ann. der Phys. (4), 2, p. 505; 1900. Holborn and Gruniesen. Ann. der Phys. (4), 6, p. 136; 1901. METALS: EXPANSION. Holborn and Day. Ann. der Phys., 4, p. 104; 1901. Am. Jl. Sci. (4), 11, p. 374; 1901. Le Chatelier. C. R., 128, p. 1444; 1899. 129, p. 331. 107, p. 862; 1888. 108, p. 1046; 1896. Ill, p. 123; 1890. Charpy and Grenet. C. R., 134, p. 540; 1902. Terneden. Thesis, Rotterdam, 1901 (Fortsch. der Phys., 1901). Dittenberger. Zs. Ver. Deutsche Ingen., 46, p. 1532; 1902. Day and Sosman. Am. Jl. Sci., 29, p. in; 1910. QUARTZ. Le Chatelier. C. R., 107, p. 862; 1888. 108, p. 1046; 111, p. 123; 130, p. 1703. Callendar. Chem. News, 83, p. 151; 1901. Holborn and Henning. Ann. der Phys., 4, p. 446; 1903. Scheel. Deutsch. Phys. Ges. (5), 5, p. 119; 1903. Verb. Phys. Tech. Reich- sanstalt, 1904. Shenstone. Properties of Amorphous Quartz. Nature, 64, pp. 65 and 126; 1901. Contains history to date. Dufour. Tin-quartz thermometer. C. R., 130, p. 775. Villard. Permeability for H at 1000 C. C. R., 130, p. 1752. Joly. Plasticity, etc. Nature, 64, p. 102; 1901. Moissan and Siemens. Action of water on. C. R., 138, p. 939; 1904. Solubility in Zn and Pb. C. R., 138, p. 86; 1904. Vapor pressure of. C. R., 138, p. 243; 1904. Heraus. Properties: a general summary. Zs. Elektroch., 9, p. 848; 1903. Brun. Fusion. Arch. Sc. Phys. Nat. (Geneva) (4), 13, p. 313; 1902. Hutton. Lamps, etc. Am. Electroch. Soc., Sept., 1903. Day and Shepherd. Science, 23, p. 670; 1906. GLASS: EXPANSION. Holborn and Gruniesen. Ann. der Phys. (4), 6, p. 136; 1901. Bottomley and Evans. Phil. Mag., 1, p. 125; 1901. Hovestadt. Jenaer Glas. BIBLIOGRAPHY 483 REFRACTORIES. (See also Furnaces.) E. K. Scott. Furnace linings. Faraday Soc., 1905. Electroch. and Met. Ind., 3, p. 140; 1905. A. Beranger. Materiaux et Produits Refractaires (Paris, 1910). Neumann. Stahl u. Eisen, 30, p. 1505; 1910. (Bibliography of alumina-lime- silica.) Fitzgerald. Furnace materials. Electroch. Ind., 2, p. 439, 490; 1904. Dunn. Soc. Ch. Ind. Jl., 23, p. 1132; 1904. Brauner. Bibliography of Clays and the Ceramic Arts. American Ceramic Society. Hutton and Beard. Conductivity. Faraday Soc. Trans., 1, p. 264; 1905. Wologdine. Conductivity, etc. Bull. Soc. d'Encour., 8, p. 879; 1909. Schoen. Use of Quartz Tubes. Metallurgie, 6, p. 635; 1908. Walden. Jl. Am. Ch. Soc., 30, p. 1351; 1908. Acheson. Siloxicon. Electroch. Ind., 6, p. 379; 1908. Arndt. Berlin Magnesia. Chem. Ztg., 30, p. 211; 1906. Goodwin and M alley. Phys. Rev., 23, p. 22; 1906. Aubrey. Bauxite. Electroch. and Met. Ind., 4, p. 52; 1906. Bleninger. Fire Clays. Proc. Eng. Soc. West. Penn., 25, p. 565; 1910. Bywater. Jl. of Gas Lighting, 102, p. 831; 1908. Boiling. Silundum. Electroch. Ind., 7, p. 24; 1909. Tucker. Ibid., 7, p. 512; 1909. Buchner. Fused Alumina. Zs. Anorg. Ch., 17, p. 985; 1904. Phalen. Alundum. Electroch. Ind., 7, p. 458; 1909. Day and Shepherd. Lime-silica. Am. Jl. Sci., 22, p. 265; 1906. .Shepherd and Rankine. Binary Systems. Am. Jl. Sci., 28, p. 293; 1909. Heinicke. Magnesia-alumina. Zs. Anorg. Ch., 21, p. 687; 1908. Iddings. Igneous Rocks. 1909. (Wiley & Sons.) Moissan. Le Four Electrique. (Also in English.) .Sexton. Fuel and Refractory Materials. VARIOUS SUBJECTS. Le Chatelier. Specific heat of carbon. C. R., 116, p. 1051; 1893. Soc. Franc. de Phys., No. 107, p. 3; 1898. Barus. Bull, of U. S. Geological Survey No. 54, 1889. (Pyrometry.) Report on the progress of pyrometry to the Paris Congress, 1900. Viscosity and temperature. Wied. Ann., 96, p. 358; 1899. Callendar. Nature, 49, p. 494; 1899. Long-range temperature and pressure variables in physics. Nature, 66, p. 528; 1897. Baly and Charley. Liquid-expansion pyrometer. Berl. Ber., 27, p. 470; 1894. Dufour. Tin in quartz pyrometer. C. R., 180, p. 775; 1900. Berlhelot. Interference method of high-temperature measurements. C. R., 120, p. 831; 1895. Jour, de Phys. (3), 4, p. 357; 1895. C. R., Jan., 1898; applications in C. R., Feb., 1898. 484 HIGH TEMPERATURES Moissan. Le Four Electrique, Paris, 1898. (Also in English.) Tijpler. Pressure-level apparatus. Wied. Ann., 56, p. 609; 1895. 67, p. 311; 1896. Quincke. An acoustic thermometer for high and low temperatures. Wied. Ann., 63, p. 66; 1897. K. Scheel. Ueber Fernthermometer. Verlag v. C. Marhold, Halle, 1898. 48 pp. Heitmann. Ueber einen neuen Temperatur-Fernmessapparat von Hartmann und Braun. E. T. Z., 19, p. 355; 1898. Chree. Recent work in thermometry. Nature, 68, p. 304; 1898. Lemeray. On a relation between the dilation and the fusing points of simple metals. C. R., 131, p. 1291; 1900. Holborn and Austin. Disintegration of the platinum metals in different gases. Phil. Mag. (6), 7, p. 388; 1904. Stewart. (Same as preceding.) Phil. Mag. (5), 48, p. 481; 1899. Hagen and Rubens. On some relations between optical and electrical prop- erties of metals. Phil. Mag. (6), 7, p. 157; 1904. Kahlbaum. On the distillation of metals. Phys. Zs., p. 32; 1900. Kahlbaum, Roth, and Seidler. Ibid. Zs. Anorg. Ch., 29, p. 177; 1902. * Glaser. Latent and sp. hts. of metals. Me'tallurgie, 1, pp. 103, 121; 1904. Richards. Metallurgical calculations. Scudder. Liquid baths for melting-point determinations. Chem. News, 88 r p. 104; 1903. Guertler. Devitrification temperatures. Zs. Anorg. Ch., 40, p. 268; 1904. Kraft and Bergfeld. Lowest evaporation temperatures of metals in vacuo. Chem. Ber., 38, p. 254; 1905. Kraft. Boiling metals in vacuo. Chem. Ber., 38j p. 262; 1905. Wiebe. Melting Points vs. Expansion. Phys. Ges. Verh., 8, p. 91; 1906. /. W. Richards. Latent Heat of Vaporization, Metals, etc. Am. Electroch.. Soc. Trans., 13, p. 447; 1908. F. C. Mason. Magnetic Pyrometer. Am. Machinist, 33, p. 875. Pawlow. Melting vs. Surface Tension. Zs. Phys. Ch., 66, p. i; 1908. Hot-blast pyrometers. Roberts- Austen's Metallurgy. J. Iron and Steel Inst., pp. 195, 240; 1884. P. 235; 1885. P. 207; 1886. 2, p. no; 1888. Proc. Inst. M. E., p. 53; 1852. Jl. Soc. Chem. Ind., p. 40; 1885. P. 16; 1897. Wiborgh. Industrial Air Pyrometer. Jl. Ir. and St. Inst., 2, p. no; 1888. Callendar. Industrial Air Pyrometer. Proc. Roy. Soc., 50, p. 247; 1892. Measurement of extreme temperatures. Nature, 69, pp. 495 and 519 a review of various pyrometric methods. Siebert. Quartz Thermometers. Zs. Elektroch. (Halle), 10, p. 26. Mahlke. On a comparison apparatus for thermometers between 250 and 600 C. Zs. Instr'kunde, 14, p. 73; 1894. F. Kraft. Evaporation and boiling of metals in quartz in electric furnace. Ber. Deut. Ch. Ges., 36, p. 1690; 1903. BIBLIOGRAPHY 485 CHEMICAL DETERMINATIONS OF TEMPERATURES. Haber and Richardt. The water-gas equilibrium in the Bunsen flame, and the chemical determination of high temperatures. Zs. Anorg. Chem., 38, p. 5; 1904. Zenghelis. Chemical reactions at very high temperatures. Zs. Phys. Chem., 46, p. 287 ; 1903. Nernst. On the determination of high temperatures. Phys. Zs., 4, p. 733; 1903. THERM OCHEMICAL DATA. Haber. Technical Gas Reactions, 1908. (Tr. by A. B. Lamb.) Nernst. Q = RT 2 (d\ogK/dT). Phys. Zs., 4, p. 733; 1903. Haber and Associates. Gas Combustion Phenomena. Zs. Phys. Chem., 68, p. 726. 69, p. 337; 1909. 71, p. 29; 1910. Zenghelis. Zs. Phys. Chem., 46, p. 287; 1903. Wartenberg. Reviews. Fortschritte d. Chemie, 2, p. 205; 1910. V. Jiiptner. Energy relations. Zs. Anorg. Ch., 42, p. 235; 1904. Wartenberg. Phys. Ges. Verb., 8, p. 97; 1906. Nernst and Wartenberg. H 2 O and CO 2 dissociation. Zs. Phys. Chem., 66, pp. 5i3, 534, 548; 1906. Tucker and Lampen. Carborundum Formation Temperatures. Jl. Am. Chem. Soc., 28, p. 853; 1906. Button and Petavel. High-temperature Electrochemistry. (References.) Elec- trician, 50, pp. 308, 349; 1902. Phil. Trans., 207, p. 421; 1908. INCANDESCENT-LAMP TEMPERATURES. Le Chatelier. Jl. d. Phys. (6), 1, p. 203; 1892. Lummer and Pringsheim. Verb. D. Phys. Ges., 1, pp. 23, 215; 1899. Janet. C. R., 123, p. 690; 1896. 123, p. 734; 1898. Pirani. Verb. Phys. Ges., 12, p. 301; 1910. Fery and Cheneveau. C. R., 149, p. 777; 1909. Jl. de Phys., 9, p. 397; 1910. Waidner and Burgess. Bull. Bureau Standards, 2, p. 319; 1907. Elec. Wld. 48, p. 915; 1906. Henning. Zs. Instr'kunde, 30, p. 61; 1910. Grau. Elektrotech. u. Maschinenb., 25, p. 295; 1907. Coblentz. Bull. Bureau Standards, 5, p. 339; 1908. Morris, Stroud, and Ellis. Electrician, 59, pp. 584, 624; 1907. Jolley. Electrician, 63, pp. 700, 755; 1909. TEMPERATURE OF FLAMES. Kurlbaum. Phys. Zs., 3, p. 187; 1902. Lummer and Pringsheim. Phys. Zs., 3, p. 233; 1902. E W. Stewart. Phys. Rev., 1902, 1903. Fery. C. R., 137, p. 909; 1903. Haber and Richardt. Zs. Anorg. Chem., 38, p. 5; 1904. 486 HIGH TEMPERATURES Haber and Hodsman. Zs. Phys. Chem., 67, p. 343; 1909. Becker. Ann. der Phys., 28, p. 1017; 1909. Allner. Jl. fur Gasbelencht, 48, pp. 1035, 1057, 1081, 1107; Amerio. Accad. Sci. Torino, 41, p. 290; 1905. Ladenburg. Phys. Zs., 7, p. 697; 1906. Schmidt. Phys. Ges. Verh., 11, p. 87; 1909. Shea. Bunsen with Thermocouples. Phys. Rev., 30, p. 397; 1910. E. Bauer. Le Radium. Dec. 6, 1909. C. R., 147, p. 1397; 1908. C. R., 148, pp. 908, 1756; 1909- SOLAR AND STELLAR TEMPERATURES. (See also Radiation Pyrometer.) Abbot. The Sun, ign. Astro, phys. Jl. 34, p. 197; 1911. Pringsheim. Physik der Sonne, 1910. Goldhammer. Ann. d. Phys., 25, p. 905; 1908. Millochau. C. R., 148, p. 780; 1909. Jl. de Phys., 8, p. 347; 1909. W Using and Scheiner. Pub. Astrophys. Obs., Potsdam, No. 56, 19. Astrophys. JL, 32, 130; 1910. Abbot and Fowle. Pub. Astrophys. Obs. Smithsonian Institution, 2, 1908. Astro- phys. Jl., 29, p. 281; 1909. .Nordmann. C. R., 149, pp. 557, 662, 1038; 1909. 160, pp. 448, 669; 1910. Bull. Astron., Apr., 1909, p. 170. Kurlbaum. Sitzber. Berlin Akad., 25, p. 541; 1911. LABORATORY FURNACES. A. Kalahne. On electric resistance furnaces. Ann. d. Phys., 11, p. 257; 1903. E. Haagen. Platinum-foil furnaces. Zs. Elektroch., p. 509; 1902. W. C. HercBUs. Electrical laboratory furnace. Zs. Elektroch., p. 201; 1902. ^ Electrician, 49, p. 519. 50, p. 173; 1902. C. L. Norton. Laboratory electric furnaces. Elec. World and Eng., 36, p. 951; 1900. F. A. J. Fitzgerald. Principles of resistance furnaces. Trans. Am. Elec. Chem. Soc., 4, p. 9; Elec. chem. Indus., 2, p. 242; 1904. D. Berthelot. Ann. Phys. et Chim., 26, p. 58; 1902. Holborn and Day. Ann. der Phys., 2, p. 505; 1900. Doelter. Electric furnaces for melting points. Centralbl. f. Min., p. 426; 1902. Day and Allen. Phys. Rev., 19, p. 177; 1904. King. Carbon vacuum tube. Astrophys. Jl., 28, p. 300; 1908. Tucker. Tube furnace. Tr. Am. Electroch. Soc., 11, p. 303; 1907. Lampen.J\. Am. Ch. Soc., 28, p. 846. Stansfield. The Electric Furnace (McGraw-Hill Co.). H. Moisan. Le Four Electrique. (Also in English.) J. Wright. Electric Furnaces and Their Application. (Henley Pub. Co.) Hutton and Petavel. Inst. Elec. Eng. (Manchester Sec.), Nov. 25, 1902. JL of, 32, p. 227; 1903. BIBLIOGRAPHY 487 Button and Patterson. Electroch. and Met. Ind., p. 455; 1905. Trans. Faraday Soc., 1, p. 187. Button. Electrician, 68, p. 579; 1907. Howe. Electric muffle. Proc. Am. Soc. Test. Materials, 6, p. 202; 1906. Friedrich. Gas metallographic furnace. Metallurgie, 3, p. 206; 1906. Electric furnaces, Metallurgie, 4, p. 778; 1907. Jl. Am. Ch. Soc., 28, p. 921; 1906. Arsem. Vacuum Furnace. Jl. Am. Chem. Soc., 28, p. 921; 1906. Trans. Am. Electroch. Soc., 9, p. 153; 1906. Jour. Eng. Chem., Jan., 1910. Ruff. Vacuum Furnace. Ch. Ber., 43, p. 1564; 1910. V. Wartenberg. Tungsten furnace. Zs. Elek. Ch., 16, p. 876; 1909. Heraus. Iridium Furnace. Zs. Angew. Chem., 18, p. 49; 1905. Of Kriptol: Zs. Angew. Chem., 18, p. 239; 1905. Metallurgie, 4, pp. 617, 778; 1907. 6, pp. 186, 638; 1908. Mutter. Vacuum Furnace. Metallurgie, 6, p. 145; 1909. Oberho/er. Vacuum Furnace. Metallurgie, 4, p. 427; 1907. Sabersky-Adler. Electric hardening furnace. Trans. Faraday Soc., 6, p. 15; 1909. Barker. Solid Electrolyte Tube. Proc. Roy. Soc., 76, p. 235; 1905. THERMOSTATS AND FURNACE CONTROL. B. Darwin. Electric Thermostat. Electrician, 62, p. 256. Astrophysical JL, 20, p. 347; 1904- Morris, Ellis, and Stroud. Automatic Rheostat Control. Electrician (Lond.), 61, p. 400; 1908. Sodeau. Regulators. Soc. Chem. Ind. Jl., 23, p. 1134; 1904. Plato. Mechanical automatic rheostat. Zs. Phys. Ch., 66, p. 721; 1906. Portevin. Continuous water rheostat. Rev. de Metallurgie, 6, p. 295; 1908. Kolowrat. Electro-optical regulator. Jl. d. Phys., 8, p. 495; 1909. Bodenstein. Thermostats. Faraday Soc., May 23, 1911. METALLOGRAPHIC PRACTICE. Frankenheim. Cooling curves. Pogg. Ann., 39, p. 376; 1836. Plato. Cooling Curves. Zs. Phys. Ch., 66, p. 721; 1906. 68, p. 350; 1908. 63, p. 447. Osmond. Cooling Curves. C. R., 103, pp. 743, 1112; 1886. 104, p. 985; 1887. Annales des Mines, 14, p. i; 1888. Wiist. Cooling Curves. Metallurgie, 3, p. i ; 1906. Tammann. Thermal analysis. Zs. Anorg. Chem., 37, p. 303; 1903. 46, p. 24, 1905. 47, p. 289; 1905. Portevin. Thermal analysis. Rev. de M6tallurgie, 4, p. 979; 1907. 6, p. 295; 1908. Dejean. Cooling Curves. Rev. de Metallurgie, 2, p. 701; 1905. 3, p. 149; 1906. Le Chatelier. Microscopic Methods. Rev. de Metallurgie, 3, p. 359; 1906. Robin. Hardness of Steels at High Temperatures. Rev. de Metallurgie, 6, p. 893; 1908. 488 HIGH TEMPERATURES W. Rosenhain. Observations on Recalescence Curves. Phys. Soc. Lond., 21, p. 180; 1908. Proc. Inst. of Metals. G. K. Burgess. On Methods of Obtaining Cooling Curves. Electroch. and Met. Ind., 6, pp. 366, 403; 1908. Bull. Bureau Standards, '6, p. 199; 1908. Guertler. Treatise on Me"tallographie, 1909. Oberhojfer. Metallographic Examination in Vacuo at High Temperatures. Metal- lurgie, 6, p. 554; 1909. Heyn. Progress from 1906 to 1909. Rev. de Metallurgie, 7, p. 34; 1910. (With bibliography.) Ruer. Treatise on Metallography, 1907. Shepherd. Thermometric Analysis of Solid Phases. Jl. Phys. Chem., 8, p. 92; 1904. Desch. Metallography, 1910. (Longman.) Guillet. Traitements Thermiques, 1909. (Dunod.) Cavalier. Alliages Me"talliques, 1909. Carpenter and Keeling. Steels. Jl. Iron and Steel Inst., p. 224; 1904. SPECIFIC HEAT. Iron: P. Oberhoffer. Metallurgie, 4, pp. 447, 486; 1907. Weiss and Beck, Jl. d. Phys., 7, p. 255; 1908. Harker. Phil. Mag., 10, p. 430; 1905. And Nickel: Lecher. Phys. Ges. Verh., 9, p. 647; 1907. Carbon: Kunz. Ann. d. Phys., 14 p. 309; 1904. Gases: Pier. Zs. Elektroch., 16, p. 536; 1909. Holborn and Austin. Berlin Sitz. Ber., 6, p. 175; 1905. Holborn and Henning. Ann. d. Phys., 23, p. 809; 1907. '/ Metals: Stiicker. Wien. Sitz. Ber., 114, p. 657; 1905. Tilden. Phil. Trans. 194, p. 233; 1900. 201, p. 37; 1903. > A. fc K Steam: Holborn and Henning. Ann. d. Phys., 18, p. 739; 1905. Iron-carbon: Oberhoffer and Meuthen. Metallurgie, 6, p. 173; 1908. Bystrom. Fortschritte d. Phys., 16, p. 369; 1860. Silicates and Pt: White. Am. Jl. Sci., 28, p. 334; 1909. Platinum: Plato. Zs. Phys. Ch., 65, p. 736; 1906. Violle. C. R., 85, p. 543; 1877. Copper: Naceari. Atti. di Torino, 23, p. 107; 1887. Richards and Frazier. Chem. News, 68, 1893. Le Verrier. C. R., 114, p. 907; 1892. Ferromagnetic substances: Dumas. Arch. Sci. Phys. Nat., 29, pp. 352, 458; 1910. NH 3 and chemical equilibrium. Nernst. Zs. Elek. Ch., 16, p. 96; 1910. TABLES. PAGE I. TEMPERATURE CONVERSION TABLE 490 II. MELTING POINTS OF THE CHEMICAL ELEMENTS 491 III. BOILING POINT OF WATER 493 IV. BOILING POINT OF SULPHUR 493 V. RESISTANCE THERMOMETER SCALE (CENTIGRADE) 493 VI. RESISTANCE THERMOMETER SCALE (FAHRENHEIT) 494 VII. AUXILIARY TO TABLES V. AND VI.. 495 VIII. TEMPERATURE CORRECTIONS FOR PLATINUM OF DIFFERENT 5 495 IX. TRANSFORMATION TABLE FOR ABSORPTION COEFFICIENTS 496 X. ABSORBING POWERS FOR METALS, ETC 497 489 490 HIGH TEMPERATURES TABLE I. TEMPERATURE CONVERSION TABLE. (Dr. L. Waldo, in Metallurgical and Chemical Engineering, March, 1910.) c 10 20 30 40 50 60 70 80 90 -200 100 -328 -148 + 32 -346 -166 + 14 -364 -184 - 4 -382 202 22 -400 220 40 -418 -238 - 58 -436 -256 - 76 454 -274 94 -292 112 -310 -130 32 So 68 86 104 122 140 158 I 7 6 194 C F 100 200 300 212 392 572 230 410 590 248 428 608 266 446 626 284 464 644 302 482 662 320 500 680 338 518 698 356 536 716 374 554 734 i 2 3 1.8 3.6 5-4 400 500 600 752 932 III2 770 950 1130 788 968 1148 806 986 1166 824 1004 1184 8 4 2 1022 I2O2 860 1040 1220 878 1058 1238 896 1076 1256 914 1094 1274 4 i 7-2 9.0 10.8 700 800 900 1292 1472 I6 S 2 1310 1490 1670 1328 1508 1688 1346 1526 1706 1364 1544 1724 1382 1562 1742 1400 I58o 1760 1418 1598 1778 1436 1616 1796 1454 1634 1814 J 9 12.6 14.4 16.2 1000 1832 1850 1868 1886 1904 1922 1940 1958 1976 1994 10 18.0 1100 1200 1300 2012 2IQ2 2372 2030 22IO 2390 2048 2228 2408 2066 2246 2426 2084 2264 2444 2102 2282 2462 2120 2300 2480 2138 2318 2498 2156 2336 2516 2174 2354 2534 F o C 1400 1500 1600 2SS2 2732 2912 2570 2750 2930 2588 2768 2948 2606 2786 2966 2624 2804 2984 2642 2822 3002 2660 2840 3020 2678 2858 3038 2696 2876 3056 2714 2894 3074 I 2 3 0.56 i. ii 1.67 1700 1800 1900 3092 3272 3452 3IIO 3290 3470 3128 3308 3488 3146 3326 3506 3164 3344 3524 3182 3362 3542 3200 3380 3560 3218 3398 3578 3236 34i6 3596 3254 3434 3614 4 1 2.22 2.78 3-33 2000 3632 3650 3668 3686 3704 3722 3740 3758 3776 3794 7 3-89 2100 2200 2300 3812 3992 4172 3830 4OIO 4190 3848 4028 4208 3866 4046 4226 3884 4064 4244 3902 4082 4262 3920 4100 4280 3938 4118 4298 3956 4136 43i6 3974 4154 4334 8 9 10 4-44 S.oo 5.56 2400 2500 2600 4352 4532 4712 4370 4550 4730 4388 4568 4748 4406 4586 4766 4424 4604 4784 4442 4622 4 802 4460 4640 4820 4478 4658 4838 4496 4676 4856 4514 4694 4874 ii 12 13 6. ii 6.67 7.22 2700 2800 2900 4892 5072 5252 4910 5090 5270 4928 5108 5288 4946 5126 5306 4964 5144 5324 4982 5l62 5342 5000 5180 536o 5018 5198 5378 5036 5216 5396 5054 5234 S4I4 14 IS 16 7-78 8.33 8. 3000 5432 5450 5468 5486 5504 5522 5540 5558 5576 5594 17 9-4* 3100 3200 3300 5612 5792 5972 5630 5810 5990 5648 5828 6008 5666 5846 6026 5684 5864 6044 5702 5882 6062 5720 5900 6080 5738 5918 6098 5756 5936 6116 5774 5954 6i34 18 IO. 3400 3500 3600 6152 6332 6512 6170 6350 6530 6188 6368 6548 6206 6386 6566 6224 6404 6584 6242 6422 6602 6260 6440 6620 6278 6458 6638 6296 6476 6656 6314 6494 6674 3700 3800 3900 6692 6872 7052 6710 6890 7070 6728 6908 7088 6646 6926 7106 6764 6944 7124 6782 6962 7142 6800 6980 7160 6818 6998 7178 6836 7016 7196 6854 7034 7214 C 10 20 30 40 50 60 70 80 90 Examples. 1347 C. = 2444 F.+I2.6 F.=24S6.6 P.; 3367 F. = i8so C.+2.78 C. = i852.78 C. TABLES 491 TABLE II. MELTING POINTS (C.) OF THE CHEMICAL ELEMENTS.* (Standard Temperatures are in small capitals.) Element. Melting point. Remarks. Helium < 269? ( B. P. He= -268.5. Hydrogen 2Ca, Silicon Troost. Thorium >i7oo, i8oo Parsons. Ytterbium 1600 to 2OOO? Unknown. Titanium ( 22OO tO 24OO? Weiss-Kaiser. Rhodium \ 1800 to 1850 IQ2O? Hunter. Range 1907 to 1970. Ruthenium >i95o Joly. Niobium 2 2OO? v. Bolton=i95o. Boron 2200 to 2<;oo Weintraub. Iridium _ 2300? Range 2100 to 2350 Uranium near Mo Moissan. Molybdenum 2500? Range 2110 to >25oo. Osmium . .... 2700? Waidner-Burgess. Tantalum 2850 Waidner-Burgess = 2910. TUNGSTEN 3006 ==100 ( Range 2575 to 3250. Carbon ? I Waidner-Burgess = 3080. Unknown. TABLES 493 TABLE III. BOILING POINT OF WATER. Temperature Centigrade; Barometer in mm. of Mercury. mm. I 2 3 4 5 6 7 8 9 730 98.880 98.918 98.956 98.994 99.032 99 069 99-107 99-145 99-183 99.220 740 99-258 99.295 99-333 99-370 99.407 99-445 99.482 99 519 99-557 99-594 750 99-631 99-668 99.705 99-742 99-779 99.816 99.853 99.890 99.926 99.963 760 100. OOO 100.037 100.073 100. IIO 100.146 100.183 100.219 100.256 100.292 100.327 TABLE IV. BOILING POINT OF SULPHUR. Temperature Centigrade; Barometer in mm. of Mercury. mm. o I 2 3 4 5 6 7 8 9 730 442.00 442.09 442.18 442.27 442.36 442.45 442.53 442.62 442.71 442.80 740 750 442.89 443-79 442.98 443.88 443.07 443-97 443.16 444.06 443-25 444-15 443-34 444-24 443-43 444.34 443-52 444-43 443.61 444 52 443.70 444.61 760 444-70 444-79 444-88 444-97 445.06 445-15 445-25 445-34 445-43 445-52 I This table is based on the assumption that the normal boiling point of sulphur is "444.70. The other temperatures are computed by Holborn and Henning's formula. TABLE V. RESISTANCE THERMOMETER SCALE (CENTIGRADE). Values of Temperature Centigrade (t) in Terms of Platinum Temperatures (pt) for Thermometers with d = 1.500. pt t Differ- ence for i PL Pt / Differ- ence for 1 pt. Pt * Differ- ence for i pt. Pt t Differ- ence for i pt. o.ooo .985 250 255.99 .066 500 534-89 .170 750 844.26 313 IO 9.867 .988 260 266.67 .070 Sio 546.62 -175 760 857.42 319 20 19.762 .991 270 277-38 .073 520 558.40 .180 770 870.65 .326 30 29.687 .994 280 288.13 .077 530 570.22 .185 780 883.95 333 40 39.641 .997 290 298.92 .081 540 582.10 .190 790 897.32 340 So 49.625 .000 300 309.75 .084 550 594-03 195 800 910.76 . -347 60 59.639 .003 3io 320.61 .088 56o 606.00 .200 810 924.28 355 70 69.683 .006 320 33i 5i .092 570 618.03 .205 820 937.87 .363 80 79.758 .009 330 342.46 .096 58o 630.11 .210 830 951-54 370 90 89.863 .012 340 353-44 .100 590 642.24 .216 840 965-28 .378 100 IOO.OO .015 350 364.46 .104 000 654.43 .222 850 979-10 .386 IIO 110.17 .018 360 375-52 .108 610 666.67 .227 860 993 oi 394 120 120.37 .021 370 386.62 .112 620 678.97 .232 870 1007.00 .403 130 130.60 .024 38o 397-76 .116 630 691.32 .238 880 1021.07 .4" 140 140.86 .027 390 408.95 .120 640 703-73 244 890 1035-23 .420 150 151.16 .031 400 420.18 -125 650 716.20 .250 900 1049.47 .428 160 161.49 .034 410 431-45 .129 660 728.73 .256 910 1063.80 437 170 171.85 .038 420 442-77 .134 670 741.32 .261 920 1078.21 445 180 182.25 .041 430 454-13 .138 680 753-97 .267 930 1092.71 .455 190 192.68 -044 440 465.53 .142 690 766.67 .274 940 1107.31 .464 200 203.14 .048 450 476.97 .146 700 779-44 .280 950 1122.00 474 2IO 213.64 .052 460 488.46 .151 710 792.27 .286 960 1136.79 .484 220 224.18 055 470 500.00 .156 720 805.17 293 970 1151.69 494 230 234.75 .058 480 5ii-58 .160 730 818.13 299 980 1166.68 503 240 245.35 .062 490 523.21 .165 740 831.16 .306 990 Il8l.76 513 250 255.99 .066 500 534.89 .170 750 844-26 .313 1000 1196.95 .524 494 HIGH TEMPERATURES TABLE VI. RESISTANCE THERMOMETER SCALE (FAHR.). 5=1.50. Platinum Gas scale Platinum Gas scale Platinum Gas scale Platinum Gas scale tempera- tempera- tempera- tempera- tempera- tempera- tempera- tempera- tures. tures. tures. tures. tures. tures. tures. tures. o 0.56 510 522.7 1040 II22.8 1570 1805.5 10 10-35 520 533-4 1050 1134.8 1580 1819.4 20 20.19 530 544-2 1060 1146.9 1590 1833.4 30 30.03 540 554-9 1070 H58.9 l6oo 1847.4 32 32.0 550 565-7 1080 II7I.O 1610 1861.6 40 39-9 560 576.5 1090 1183.1 1620 1875-6 5 49-8 570 587-3 IIOO "95 3 1630 1889.9 60 59-7 580 598.2 i no 1207.5 1640 1904.1 70 69-5 590 609.1 1 1 20 1219.7 1650 1918.3 80 79-5 600 620.0 1130 1232.0 1660 1932.5 QO 89.4 610 630.9 1140 1244-3 1670 1946 . 8 100 99 4 620 641.8 1150 1256.6 1680 1961 . 2 no 109.3 630 652.8 1160 1270.0 1690 1975-7 1 20 "9-3 640 663.8 1170 1281.3 1700 1990.2 130 129.3 650 674.8 1180 1293.7 1710 2004.7 140 139-4 660 685.8 1190 1306 . i 1720 2019.3 150 149.4 670 696.9 1200 1318.7 1730 2034.0 160 159-4 680 707-9 I2IO 1331-1 1740 2048 . 7 170 169.5 690 719.0 I22O 1343 7 1750 2063.4 180 179.6 700 730.1 1230 1356-3 1760 2078.2 190 189.7 710 741-3 I24O 1368.9 1770 2093.1 200 1998 720 752.5 1250 1381.5 1780 2108.0 210 209.9 730 763-6 1260 1394-2 1790 2123.0 212 212. 740 774-8 1270 1406 . 9 1800 2138.0 22O 22O.I 750 786.0 1280 1419.6 1810 2I53-I 230 230.3 760 797-3 1290 1432.4 1820 2168.3 240 240.5 770 808.6 1300 1445 2 1830 2183.5 250 250.7 780 819.9 1310 I458.I 1840 2198.7 260 260.9 790 831.2 1320 I47I.O 1850 2213.0 270 271.2 800 842.6 1330 1483.9 1860 2229.4 280 281.4 810 854.0 1340 1496.8 1870 2244-9 290 291.7 820 865.4 1350 1509.8 1880 2260.4 300 302.0 830 876.8 1360 1522.9 1890 2276.0 310 312.3 840 888.3 1370 1535-9 1900 2291.6 320 322.7 850 899.7 1380 I549-I 1910 2307-3 330 333-0 860 911.2 1390 1562.1 1920 2323.0 340 343-4 870 922.7 1400 1575 3 1930 2338.9 350 353-8 880 934-3 I4IO 1588.5 1940 2354.8 360 364.2 890 945-9 I42O 1601.8 1950 2370.8 370 374-6 900 957 5 1430 1615.1 1960 2386.8 380 385-1 910 969.1 1440 1628.4 1970 2402 . 9 390 395-6 920 980.8 1450 1641.8 1980 2419.0 400 406.1 930 992.5 1460 1655.2 1990 2435-3 410 416.6 940 1004.2 1470 1668.7 2000 2451.6 420 427.1 950 1015.9 1480 1682.2 2010 2468.0 430 437-6 960 1027.7 1490 1695-7 2O2O 2484.4 440 448.2 970 1039-5 1500 1709 3 2030 2500.8 450 458.8 980 1051-3 1510 1722.9 2O4O 2517-5 4 60 469.4 990 1063 . i 1520 1736.6 2050 2534-2 470 480.0 1000 1075.0 1530 1750.3 2O6O 2551.0 480 490.6 IOIO 1086.9 I54O 1764.0 49O 501 .3 IO2O 1098 . 8 I55O 1777.8 500 512.0 1030 ino.8 1560 1791.6 TABLES 495 TABLE VII. AUXILIARY TO TABLES V AND VI. Corrections to t for small changes in S. Centigrade scale. Fahrenheit scale. At tor A/ for Affor At for A5 = o.oi. A5 = o.oi. AS = o.oi. A6 = o.oi. 5 O.OO2 550 +0.247 IOO 0.003 IIOO +0-53 100 .OOO 600 .300 200 .OOO 1 200 .64 150 + .008 650 357 300 + .014 1300 .76 2OO .O2O 700 .420 4OO .038 1400 .90 250 037 750 .487 500 .07 1500 OS 300 .060 800 .560 600 .11 1600 2 3 350 .087 850 637 700 .18 1700 38 400 .I2O QOO .720 800 25 1800 56 450 157 950 .807 900 33 1900 73 500 . 2OO 1000 .900 IOOO .42 2000 95 Computations of / from pt are made by Table V, as if the thermometer had d = 1.50. The above corrections (A/) are then applied to the computed values of t for the value of 5 proper to the thermometer. Example. Let pt = 470.00, whence / = 500.00 C. by Table V. If 5 = 1.52, the corrected value of / is 500.40 C. by Table VII. TABLE VIII. TEMPERATURE CORRECTIONS FOR PLATINUM OF DIFFERENT 5. [Thermometer calibrated by Callendar method, ice, steam, and S. B. P.J Correction in C. for values of S given below. c. 1.525 1.550 1.575 i. 600 1.650 1.700 1.800 1.900 200 +O.O2 +0.05 + 0.08 + o. 10 + 0.14 + 0.16 + O.2O + O.2I 300 + .02 + -05 + .08 + .11 + -19 + -27 + -45 + -55 400 .OO .00 + .01 + -03 + .08 + -14 + .29 + -37 500 .02 -05 - .09 . ii - .18 - -24 - -39 - -57 600 - .09 - .18 - -30 - .40 - .62 - .88 - 1.42 - 1.96 700 - -33 - .70 - 1.03 - 1.32 - I. 7 8 2.2 - 2.9 - 3-5 800 - .90 -1.65 -2.24 2.7 ~ 3-6 - 4.4 - 5-8 - 7.1 900 1.90 -3-i 4.0 4-9 - 6.5 - 8.1 10.8 -13-5 IOOO -3-3 -5-2 - 6.8 - 8.2 -10.7 -13.1 17.1 -20.8 IIOO -5-5 -8.1 -10.3 12.2 -15-7 -18.7 -24-3 -29.1 The above table applies only when the value of 5 is that given by using the S. B. P. as third calibration point of a resistance thermometer. 496 s O E | W M 8 I o I ? t & O PM "O H I r .1 KH T3 t: S 1 A I HIG rf ^f H TEMPERATURES foo ON ON ON ON ON ONOO t^ r^ O *H CS co rf LOO t^-OO r? r? foN ON ONOO OO t^* O M O M W co rf to t^O^" r?' 4 "' r? M M t^ cs rf ONOO t^. to rf (N O O M ON ONOO 00 t^ O M .QO >.QO Cuprous oxide 7 7 65 .60 Iron oxide 65 t O .QO > oo > oo > oo 2"? t o so 07 t o . 13 IO IO 06 t O OQ .06 t o .09 Lime .10 t o .40 The absorbing power a = emissive power e = i - r, where r = reflecting power. INDEX ABBOT, 268, 274, 276, 277, 455. Absorption Coefficient, 254, 300, 335, 336. pyrometer, 311. ACHESON, 1 80. Actinometer, 263. Air, as thermometric gas, 18, 19, 23, 25, 3i, 35, 61. Alarm, temperature, 431. ALLEN, 452. Alloys (see also thermocouples). melting points of, 451. chemical changes in, 107. Aluminium, freezing or melting, 445, 449- in thermocouples, 170. AMAGAT, 34. ANGSTROM, 276. Aniline, boiling point, 450. Antimony, melting and freezing point, 62, 445, 447. Arc, carbon, temperature of, 273, 453. Argon, as thermometric gas, 26. ARNOLD, 191. ARSEM, 333, 342, 461. ARSONVAL (D'), n, 121, 135, 137, 212, 267, 271. AUSTIN, 76, 159. AVENARIUS, in. AYRTON, 122. BARR, 443. BARRETT, 169. BARUS, 9, n, 58, 76, 77, 103, 171, 378. boiling and freezing points, 6, 187, 188, 438, 443- gas thermometer, 65. thermoelectric pyrometer, 102, 105, 112, 118. BAUER, 247, 340. BECK, specific heats, 92, 93. BECKER, 340. BECKMANN, 251. BECQUEREL, iii, iv, 9, 10, n, 15 16, 51, 54, 57, 63, 109, 296. gas thermometer of, 62. melting and boiling points, 438, 439, 441. thermoelectric pyrometer, 101, 121. BEDFORD, 59. BELLOC, thermoelectricity, no, 169. BENEDICKS, 421. BENOIT, 55. Benzophenone, boiling point, 450. BERTHELOT, 99. calorimetry, 94. BERTHELOT, DANIEL, 180, 188. expansion of gases, 25. gas thermometer of, 85. melting and boiling points, 8, 88, 438, 439, 440, 441, 445- thermodynamic corrections, 31, 32, 34, 35- Bibliography, 465. Biju-DuvAL, calorimetry, 97. Bismuth, freezing point, 445. Black body, 239, 247, 293, 333. temperature, 242. Boiling points, 6, 88, 226. methods for, 187. standard, 463. Bolometer, 248, 268, 274, 425. v. BOLTON, 446. BOLTZMANN, 248, 251. radiation laws, 247. BOTTOMLY, 247. BOUDOUARD, v, vii, 305, 420. BOYLE, Law of, 13. 499 500 INDEX BOYS, 10, 267, 272. BREAJRLEY, 366, 413. Br. Association and platinum thermom- etry, 194, 227. BREGUET, 158. BRISTOL, 83, 156, 157, 170, 175, 408, 411. BROCA, 213. BROWN, 284, 285. BUCKINGHAM, corrections to gas scales, 32, 34, 35- BUDENBERG, 380. BUNSEN, 276. Bureau International des Poids et Mesures, 13, 20, 22, 24. gas thermometer of, 37. BURGESS, iii, v, vi, viii, 10, 79, 252, 330, 334, 414, 416. freezing and boiling points, 7, 199, 226, 435, 437, 438, 441, 442, 443, 444, 445, 451, 452, 454, 459. melting points, 9, 344, 446. optical and radiation pyrometry, 240, 243, 253, 286, 307, 320, 323, 332, 335, 337, 340, 344- resistance pyrometer, 199, 200, 205, 228, 231, 232, 233. thermoelectric pyrometer, 115, 185. BURTON, 377. Cadmium, boiling point, 66, 88. freezing point, 445. CAGNIARD-LATOUR, 380. CALLENDAR, 9, 10, 12, 59, 68, 74, 198, 199, 200, 211, 213, 225, 231, 234, 269, 274, 276, 378. gas expansion, 35. gas scale corrections, 21, 31, 32. gas thermometer of, 70. melting and boiling points, 6, 226, 434, 435, 438. recording pyrometers, 386, 388, 390, 425- resistance pyrometer, 194, 195, 196, 197, 201, 202, 217. Calorimeters, 94. Calorimetric pyrometer, 10, 89. Calorimetric pyrometer, precision of, 97, use of, 99. CAMBRIDGE SCIENTIFIC INSTRUMENT Co., 131, 143, 176, 194, 203, 213, 223, 237, 390, 408, 409, 413, 425,. 426. Carbonic acid, as thermometric gas, i8 r 19, 20, 22, 25. Carbon monoxide, as thermometric gas, 1 8. CARHART-CLARK, 136. CARNELLY, 377, 378. CARNOT, 26. Carnot's principle, 26. CARPENTER, 416. CARPENTIER, 124, 125, 145, 217, 221,, 393, 408. Cell, standard, 136. CELSIUS, scale of, 3. CHAPPUIS, 9, 21, 24, 30, 31, 34, 54, 59, 68, 70, 72, 198. expansion of gases, 21, 25, 35. gas scales, 20, 23, 434. standard gas thermometer, 42. standard mercury thermometers, 44, CHARPY, 191, 403, 404. CHAUVIN and ARNOUX, 132, 176, 408. Chromium, in thermocouples, 106, 170. CLAPEYRON, 380. CLARK, 136, 137, 428. Clays, heating curves, 399. CLEMENT, 68, 76, 79. gas thermometer, 75. melting points, 8, 440. Cobalt glass, 347. Cobalt, melting point, 446. Color scale, 346. COBLENTZ, 253, 268, 269, 341, 342. Cones (fusible or Seger), n, 368. scale of, 371, 373. Constantan, 167, 174. Contraction pyrometer, n, 357. Cooling curves, 383. derived differential, 384. differential, 382, 414. inverse rate, 384. methods for, 384, 396, 408. INDEX Cooling curves, rapid, 420. temperature-rate, 382. time- temperature, 381. with neutral, 414. Copper, emissivity, 286. eutectics, 441. in thermocouples, 167, 169. melting or freezing point, 65, 441, 447- specific heat, 93. CORNU, 296, 298. COUPEAUX, 58. CRAFTS, 57, 81, 84, 85, 380. boiling points, 6, 434. expansion of gases, 22. gas thermometry, 49, 51, 83. CRAMER, 369. CROMPTON, 159. CROOKES, 267. CROVA, 350, 352. Crucibles, 181, 188. furnaces, 180, 459. Curve tracer, 412. Curves, freezing point, 447. DARLING, 158. DARWIN, 431. DAY, vii, viii, 9, 54, 59, 67, 68, 72, 76, 77, 79, 81, 114, iSS, 160, 161, 171, 180, 253, 440, 456. expansion of metals, 55. gas thermometer, 75. melting points, 7, 8, 55, 438, 439, 441, 442, 445, 446, 451, 452. thermoelectric pyrometer, 112, 113. DEJEAN, recording pyrometer, 382, 399, 400. DEPREZ, n, 121, 137, 271. DICKINSON, 99, 206. DICKSON, resistance pyrometer, 200. DlESSELHORST, 145, 146. Displacement, Wien's law of, 251. DIXON, 1 80. DOERNICKEL, 171. DULONG, 245, 264, 265, 266. DUMAS, 85. EDELMANN, 421. EDWARDS, 230. EINTHOVEN, 421. Electromotive forces of thermocouples, 105, in, 116. Electrothermal recorder, 394. Emissive powers, 239, 244, 255, 291, 293- correction for, 255, 257. determination of, 301. Energy curves, 248. EUCHENE, specific heats, 92, 93. EUMORFOPOULOS, 59, 68, 74, 75, 200, 367, 434- Eutectics, 451, 452. Cu . Cu 2 O, 441, 445- Ag 8 .Cu 2 , 445, 451- FAHRENHEIT, scale of, 3. FARYTHER, 333. FERY, vii, 10, 247, 268, 283, 285, 286,. 333, 34i, 443, 454, 455- radiation pyrometers, 277, 280, 281, 284, 311, 314, 339, 354,. 425- Filaments, temperature of lamp, 340. temperature of Nernst, 342. wide comparison, 331. Fixed points, 5, 433. table of, 456. FISCHER, 155. FIZEAU, 43, 55, 61. Flames, temperature of, 338. Flicker photometer, 352. Fluorite, 278. FOSTER, 284, 285. FOURNIER, 12, 380. FOWLE, 276, 455. FRAZIER, specific heats, 94. Freezing points (see also Melting points). curves of, 447. reproducibility, 452. Furnaces, 179, 289, 458. control of, 430. temperatures, 191, 388. Fusible cones, n, 368. 502 INDEX Galvanometer, for radiation pyrometers, 283. ^ for resistance pyrometers, 209, 212, 213, 216, 220. for thermocouples, 118, 120, 131, J 33, 137. string, 421. Gases, coefficients of expansion, 17, 25, 35- variations of, 18, 20, 22. at high temperatures, 26. critical constants, 32. scale corrections for, 31. GASPARIN, 262. Gas pyrometer (see also Gas thermom- eter), 9, 37. Gas thermometer, 3, 14, 16. as standard, 13, 16, 37, 430. comparison of results with, 79. constant pressure, 15, 50, 64, 70. constant volume, 14, 44, 64, 67, 68, 75, 76. formulae and corrections, 44, 74. future experiments with, 80. indirect processes, 83. industrial, 82. method of D. Berthelot, 85. method of Crafts and Meier, 83. methods of Regnault, 84. methods of Sainte-Claire-Deville, 85. of variable pressure and mass, 15. recording, 385. substance of bulbs, 53. volumetric thermometer, 15, 51. GAY-LUSSAC, law of, 13, 27. GEIBEL, thermoelectric pyrometer, 171. Geophysical laboratory, vii, 8, 55, 56, 77, 181, 342, 460. Glass, absorbing, 300. as gas thermometer bulbs, 59, 68, 71. cobalt, 347. colored (monochromatic), 297, 334, 347- thermometric, 17, 363. Gold, in thermocouples, 172. melting or freezing point, 62, 65, 67, 88, 272, 439, 445. GOLDHAMMER, 456. Graphite, furnaces, 462. GRAU, 341. GRAY, 271, 272, 273, 305, 454. GREENWOOD, 342, 463. boiling points of metals, 9. GRENET, 366. GRIFFITHS, 74, 212, 214. melting and boiling points, 6, 226, 433, 434- resistance pyrometer, 194, 195, 197. GURNEISEN, 59. GUILLAUME, 244, 364. GWYER, 226. HADFIELD, v, 169. BARKER, 21, 24, 30, 68, 72, 437. melting and boiling points, 434, 443. resistance pyrometer, 198. specific heats, 92. thermoelectric pyrometer, 114, 143. HARRIS, 218. HARRISON, 169. HARTMANN, 323, 342. HARTMANN and BRAUN, 127, 158, 218, 408, 409. HAUSRATH, 146. Heat (see also Specific heat). total, of metals, 90. HECHT, 369. HENNING, 60, 68, 79, 159, 228, 256, 331, 336, 437- melting and boiling points, 8, 226, 434, 435- optical pyrometer, 330. HER^US, 60, 75, 152, 164, 171, 174, 180, 181, 205, 226, 331, 376, 458, 459. HEVESY, 174. HEYCOCK, 440, 452. melting points, 6, 199, 438, 439, 440, 44i, 443- resistance pyrometer, 194, 199, 202, 231. HOBSON, 12. HOFFMANN, 376, 418. HOLBORN, vii, 9, 10, 54, 56, 57, 59, 60, 68, 72, 76, 79, 159, 161, 171, 180, INDEX 503 200, 228, 247, 253, 328, 330, 337, 340, 342, 343, 345, 437, 44Q. boiling points, 8, 226, 434, 435. expansion of metals, 55. gas pyrometer, 67, 75. melting points, 7, 8, 67, 438, 439, 44i, 442, 443, 444, 445. optical pyrometer, 324, 325, 327, 329, 426. resistance pyrometer, 196, 199, 233, 234- thermoelectric pyrometer, 112, 113, 119, 126. HOLMAN, 378, 442, 443. thermoelectric pyrometer, 112, 113, 114, 139. HOSKINS, 170, 411. HOVESTADT, 364. HOWE, 191. HULETT, 137. HiJTTNER, 367. HUTTON, 462. Hydrogen, as thermometric gas, 18, 19, 20, 21, 22, 25, 31, 35, 68. Helium, as thermometric gas, 26, 31, 35. Ice point, 34. INGERSOLL, 342, 444, 445. International committee on weights and measures, scale of, 37. Iodine vapor, in gas thermometry, 62, 84. Iridium, expansion of, 56. furnace, 459. gas thermometer bulbs, 56, 75. in resistance pyrometers, 235. in thermocouples, 172, 174. melting point, 294, 444. specific heat, 91. Iron, emissivity, 256, 287. expansion of, 56. gas thermometer bulbs, 56, 62. in thermocouples, 101, 103, no, 168. melting point, 446. specific heat, 92. Isochromatic curves, 250. JAQUEROD, 60, 61, 68. boiling and melting points, 7, 437, 439, 440. expansion of gases, 25. gas thermometer, 70. JAGER, 211, 231, 451. JOB, 12, 378. JOLY, n, 220, 272, 341. meldometer, 361. JOULE, 21, 27, 30, 31, 34. KAHLBAUM, 453. KANOLT, 342, 375, 376. KEISER and SCHMIDT, 127. KELVIN (THOMSON), 21, 30, 31, 34, 109, 215, 274. Kelvin bridge, with resistance pyrom- eter, 215. KING, 247. KIRCHHOFF, 240, 244, 245, 333. radiation laws, 239, 243. KOLOWRAT, 430. KONIG, 315. KURLBAUM, Vii, 10, 247, 328, 330, 337, 342, 343, 346, 455- optical pyrometer, 324, 325, 327, 329, 426. radiation laws, 238, 240, 246, 274, 338, 340. KURNAKOW, 403, 404. Laboratorie d'Essais, 457. LADENBURG, 340. Lamps, pyrometer, 328, 329, 332. comparison, 332. electric, temperature of, 340. LANGLEY, 10, 248, 268, 274. bolometer, 267, 273, 425. LATIMER-CLARK, 136. LAUTH, 368. LAWRENCE, 443. Lead, freezing point, 445. LE CHATELIER, vi, vii, 10, n, 12, 51, 65, 67, 97, 103, 104, 171, 172, 182, 278, 306, 3", 3U, 3i8, 333, 336 > 340, 346, 369, 397, 421, 454. contraction pyrometer, 358. 504 INDEX LE CH ATELIER, expansion of porcelain, 58. optical pyrometry, 292, 296, 301, 302, 305- recording pyrometers, 382, 396, 400, 418, 420. specific heats, 93. thermoelectric pyrometer, 102, 105, 107, 108, in, 119, 122, 125, 128,. 191. LEEDS and NORTHRUP, 140, 145, 203, 205, 208, 209, 212, 216, 218, 224, 235, 236, 393, 408, 420. LEHRFELDT, 21. ice point, 34. LEITHAUSER, 253. LESLIE, 266. LE VERRIER, specific heats, 94. Logometer, 221. LOUGININE, 99. Luminous intensity, 293. LUMMER, vii, 338, 340, 454- radiation laws, 238, 239, 240, 246, 248, 250, 251, 252, 274, 352. McCRAE, 367. MALLARD, 51, 65. Manganin, 222. in galvanometers, 131. MARIOTTE, law of, 13, 27. Marquardt porcelain, 181. MARSH, 170. MARVIN, 234, 277. MASCART, 403. MATHIAS, 364. MEIER, 8 1, 83, 84, 85. MEISSNER, 376. Meldometer, n, 273, 361. MELLONI, 267. .Melting points, 6, 189. certified, 458. methods for, 179, 183, 188, 342, 343. of chemical elements, Table II, Ap- pendix. of salts, 190, 366. pyrometry based on, 365. standard, 433. Mercury thermometer, scale of, 16. standards, 44. MESURE, 348. Metals, certified, 458. MENDENHALL, 268, 328, 334, 342, 444, 445- MEUTHEN, 92. MEYER, 328, 446. MICHELSON, 276, 277. MlLLOCHAU, 455. MOISSAN, 60. MOREWOOD, 366. MORSE, vii, 10, 324, 325, 329, 330, 336, 346, 426. Moss, 59, 434. MOULIN, 247. MYLIUS, 60. Naphthaline, boiling point, 450. National Physical Laboratory, 457. NERNST, 253, 342, 459. melting points, 8, 442, 443, 444. optical pyrometry, 294, 319, 337. NEVILLE, 440, 452. melting points, 6, 199, 438, 439, 441, 442, 445- resistance pyrometer, 194, 199, 202, 231. NEWTON, 245, 265, 266. NICHOLS, 337, 338. Nickel, in resistance pyrometers, 234. in thermocouples, 103, 106, 167, ?68, 170, 174. melting point, 445, 446. specific heat, 93. Nitrogen, as thermometric gas, 18, 20, 21, 22, 23, 25, 31, 33, 35, 68, 80. NOBILI, 267. NORDMANN, 346, 353, 455- Normal scale of temperatures, 21. thermometer, 3, 22 NORTHRUP, ratiometer, 223. NOUEL, 348. NUTTING, 255. OBERHOFFER, 99. specific heats, 92. INDEX 505 OHM, 118. Ohmmeter, 218. Optical pyrometer, 10, 291, 296. applications, 338. calibration, 302, 306, 311, 316, 327, 33i. corrections, 257, 336. errors of, 306, 319, 322. extension of scale, 336. measurements with, 305, 326, 338. of D. Berthelot, 85. of Crova, 350. of F6ry, 311, 354. of Henning, 330. of Holborn and Kurlbaum, 324, 337. of Le Chatelier, 296. of Mesur6 and Nouel, 348. of Morse, 324, 329. of Shore, 311. of Wannerj 314, 324. range, 32^: scale of, 306, 307, 336, 434. stellar, 353. use of, 344. ORTON, 375, 376. OSMOND, 191, 384. OTTERHAUS, 155. Palladium, latent heat, 65. in resistance pyrometer, 233. in thermocouples, 101, 103. melting point, 64, 67, 185, 201, 272, 442. specific heat, 64. PALMER, 170. PARVILLE, 150. PASCHEN, vii, 268. radiation laws, 246, 248. PAUL, 129, 131, 158, 159, 218. PEAKE, 176, 426. PECHEUX, thermoelectric pyrometer, 167, 168. PELLIN, 128, 298, 406, 408, 420. PELTIER, 109, in, 131, 269, 270. PERROT, 61, 63, 68. expansion of gases, 25. gas thermometer, 70. PERROT, melting points, 439, 440. PERRY, 122. PETAVEL, 200, 443, 454, 462. PETIT, 246, 265, 266, 267. Phosphorus, effect on thermocouples, 107. Photometer, 296, 352. PIONCHON, specific heats, 92, 93. PlRANI, 259, 328, 341, 446. PLANCK, vii, 340, 353, 355, 455. radiation laws, 251. Platinum, alloys of, 54, in, 171. expansion of, 54. latent heat, 64. melting point, 64, 67, 115, 294, 334, 442. resistance pyrometer, 194, 203. specific heat, 63, 91. thermocouples, 67, 101, 105, in, 171. thermometer bulbs, 54, 61. Platinum-palladium, thermocouple, 101, i73- Platinum-iridium, expansion of, 43, 55. gas thermometer bulbs, 39, 43, 69, 75, 76. thermocouple, 103, 105, 173. Platinum-rhodium, gas thermometer bulbs, 54, 76. thermocouple, 67, 105, 108, 114, 116, 173- PLATO, 91, 367. Porcelain, expansion of, 58, 59, 67. gas thermometer bulbs, 44, 57, 62, 63, 65, 66, 68. insulators, 151. Potentiometer, 138. for radiation pyrometers, 280. for resistance pyrometers, 214. for thermocouples, 139, 143. precision requirements, 141. POUILLET, iii, 9, 14, 16, 54, 91, 137, 263, 264, 265, 441. gas pyrometer of, 61. pyrheliometer, 262. scale of, 3, 62, 346. thermoelectric pyrometer, 101, 121. INDEX PRIM, 342. PRINGSHEIM, vii, 338, 340, 454. radiation laws, 238, 246, 248, 250, 251, 252, 352. PRINSEP, 365. Purimachos, 151. Pyrheliometer, 262, 276. Pyrometers, 9. calorimetric, 10, 89. contraction, i, n, 357. dilution, 377. expansion, 360. fixed focus, 285. fusing point, 365. gas, 9, 37- optical, 10, 292. pneumatic, 378. radiation, 10, 262. recording, 12, 286, 381. resistance, 10, 194. spectral, 330. standardization of, 432, 457., stellar, 352. sentinel, 366. thermoelectric, n, 101. transpiration, 378. vapor pressure, 380. various, 357. viscosity, 378. Quartz, expansion of, 60. gas thermometer bulbs, 60, 70. insulators, 151. mercury thermometer bulbs, 365. QUEEN, 408. Radiation, laws of, 4, 238, 272. application to pyrometry, 253, 261, 266, 291. constants of, 247, 250, 251, 253. of Kirchhoff, 243. of Le Chatelier, 303. of Planck, 251. of Rasch, 294. of Stefan, 245. of Wien, 249, 251, 254. Radiation, intensity of, 238. monochromatic, 291, 295. Radiation pyrometer, 10, 257, 261, 277, 284. calibration, 288. computation, 290. errors of, 283, 287. focusing, 282, 288. of Brown, 284. of Fery, 277, 281, 284. of Foster, 284. of Thwing, 284. results with, 285. use of, 285. Radiobalance, 269. Radiomicrometer, 267, 271. Radiometer, 267, 275. RAMSAY, 367. RANDALL, 54. Range control, 426. RANKINE, 18. RASCH, 240, 295, 304, 444. Ratiometer, 221. REAUMUR, scale of, 3. Recording pyrometers, 12, 381. accessories, 426. autographic, 408, 417. curve tracer, 413. drum recorder, 410. gas, 385- methods of, 384, 390, 395, 396, 403. multiple, 428. of Benedicks, 421. of Callendar, 386, 390. of Carpentier, 394. of Charpy, 403. of Dejean, 399. of Le Chatelier, 396, 420. of Leeds and Northrup, 393. of Kurnakow, 403. of Roberts- Austen, 401, 415. of Saladin, 418. of Schmidt, 404. of Siemens and Halske, 392, 408. of Wologdine, 407. photographic, 396. radiation, 423. INDEX 507 Recording pyrometers, resistance, 386. semiautomatic, 413. thread recorder, 410. thermoelectric, 395. REGNAULT, iii, iv, v, 10, 14, 17, 18, 19, 20, 29, 51, 62, 85, 97. boiling points, 6, 434. experiments of, 16, 84. specific heats, 92. thermoelectric pyrometer, 101. Reichsanstalt, Physikalisch Tedmische, vii, 7, 54, 68, 75, 127, 139, 342, 372, 376, 452, 457. Resistance pyrometer, 10, 194. as standard pyrometer, 227. calibration, 224, 231, 234. t compensation for leads, 208, 230. constancy of, 232. -construction of, 202, 218. direct reading, 218. errors of, 228. formulae for, 197, 201, 225. heating by current, 228. insulation, 221. industrial installations, 234. industrial forms, 204, 216, 218. lag of, 229. -laboratory forms, 203, 208, 218. methods of measurement, 207, 218, 230, 386. nomenclature, 201. of impure platinum, 231. of palladium, 233. precautions with, 206, 228, 234. protection of, 235. recording, 386. reduction tables, 220, 232. results with, 226. scale of, 197, 226, 235. size of wire, 206, 231. sensitiveness, 216. use of, 234. Rhodium, in resistance pyrometers, 234- in thermocouples, 105, in, 116. melting point, 444. RICHARD, 403, 404, 408. RICHARDS, 99. specific heats, 94. ROBERTS-AUSTEN, 12, 151, 191, 407. recording pyrometers, 382, 401, 403, 404, 405, 415, 416. ROSE-lNNES, 21, 35. ROSENHAIN, 384. ROSETTI, 10, 245, 267, 273. radiation, 265, 266, 267. ROTHE, 418, 434. Roux, 402. RUBENS, 251, 269. RUDOLPHI, 171. RUER, 446. RUFF, 367, 443. RUTHENIUM, in thermocouples, 174. SAINTE-CLAIRE-DEVILLE, iii, 9, 16, 51, 54, 57, 62, 63, 83, 85, 438. SALADIN, 420. recording pyrometer, 407, 418. Salts, melting points of, 190, 365, 450. Scales, color, 346. control box, 427. electrical resistance, 197, 228. gas, 3, 21, 23, 432. gas scale corrections, 30. mercury-in-glass, 17, 364. normal, 21, 24, 37. of the Reichsanstalt, 7, 8. practical, 13, 17. standard, 13, 21, 37, 432. thermodynamic, 3, 4, 13, 26, 31, 35, 432- thermometric, 2, 432. SCHAFFER, 380. SCHEEL, 6l. SCHEIMER, 355, 455. SCHMIDT, 404, 405. SCHOTT and GENOSSEN, 298, 335, 337. SCHUKAREW, 99. SCHULZE, 340. SECCHI, 265. SEEKECK, 101. SEGER, n, 369, 370, 375, 376. fusible cones, 368, 372. Seger cones, n, 366. INDEX Selenium, in pyrometry, 355. SEXTON, 437. SHENSTONE, 60. SHORE, 312. SIEBERT and KUHN, 60, 365. SIEMENS, iv, 10, 210. calorimeter, 96. resistance pyrometer, 194, 195, 209. scale of, 3. SIEMENS and HALSKE, 12, 127, 131, 133, 145, 208, 210, 392, 408, 409, 417, 420, 425. Silicon, effect on thermocouples, 107. Silver, in thermocouples, 172, 174. melting or freezing point, 62, 65, 67, 88, 440, 445, 448. SMITH (F. W.), 230, 231. SOMERVILLE, 167. SOSMAN, viii, 68, 77, 79, 81, in, 114, 155, 160, 253, 440, 456. expansion of metals, 55.. gas thermometer, 75. melting points, 8, 438, 439, 441, 442, 445, 446, 451. thermoelectric pyrometer, 115, 116. Specific heat of metals, 91. pyrometer (see Calorimetric pyrom- eter). Standardization of pyrometers, 43 2, 45 7. laboratories for, 457. optical, 302, 331, 336. radiation, 288. resistance, f!6, 227, 236. thermoelectric, 178. Standards, Bureau of, 7, 212, 252, 325, 342, 364, 372, 375, 447, 457, 45, 459- Standard temperatures, 5, 9, 13, 433. Standard cells, 136. STANSFIELD, in. . melting points, 438, 442, 445. thermoelectric pyrometer, 113. STEFAN, 247, 250, 253, 254, 272, 273, 278, 280, 289, 455- radiation laws, 245, 246. STEIN WEHR, 211. STEWART, 338. String galvanometer, 421. Sulphur, boiling point, 75, 433. Sulphurous acid, as thermometric gas, 18. Sun, temperature of, 261, 273, 455. Tables, 489. TAIT, in. TAMMANN, 367. Tantalum, lamp temperature, 341. melting point, 446. Telescope, pyrometric, 277. Temperature, alarm, 431. black-body, 242. definition of, 2. equivalent, 243. of flames, 338. of glow lamps, 340. of industrial furnaces, 191, 346, 388. of stars, 352, 355. records, 381. within furnaces, 342. Temperature coefficient, of galvanom- eters, 131. of metals, 171, 173, 194. of gases, 17, 25,35. Thalpotassimeter, 380. Thermocouples (see also Thermoelectric pyrometer). annealing, 104, 149, 173. Becquerel effect, 109. calibration of, 178. chemical changes, 107. cold junction, 154, 155. compensating leads, 156, 176. compound, 275. constancy of, 159, 164. effect of reducing atmosphere, 107. inhomogeneity of, 106, 161. insulation and protection, 149. junction of wires, 148. neutral point, in. of base metals, 165, 193. of complex alloys, 170. of Noble metals, 171. reproducibility of, 164. resistance of, 118, 171. INDEX 509 Thermocouples, thermoelectric power, 161, 166. tensile strength, 173. Thomson effect, 109. types of mounting, 151, i$& parasite currents, 106. Peltier effect, 109. Thermodynamic scale, 3, 4, 13, 26, 31, 35, 432. ice point on, 34. Thermoelectric pyrometer (see also Thermocouple), n, 101. advantages of, 102. applications of, 191, 192. base metals, 116, 165. choice of couple, 105. experiments with, 102. extrapolation, 115, 174. formulae, 102, 108, HI. galvanometers for, 118, 120, 131. methods of measurement, 116, 118, i35, i79- potentiometers, 135. recorders, 393. Thermoelectric telescope, 277, 282. Thermogage, 324, 329. Thermometer, mechanical, 360. mercury for high temperatures, 362. standard gas, 38. standard mercury, 44. Thermophones, 376. Thermopile, 266, 268, 269. Thermostats, 430. THOMPSON, 342. THREW, 379. THURMEL, 352. THWING, 132, 157, 170, 284, 286, 408, 411,418. TlLDEN, 91. Tin, freezing point, 445. TOEPFER, 403. TORY, 200, 202. Total heat of metals, 91. TRAVERS, 60, 226. TROOST, 51, 62, 83, 85. TUCKER, 342. Tungsten furnace, 462. Tungsten furnace, lamp temperature, 341- melting point, 446. TUTTON, 59. TYNDALL, 246. UHLING-STEINBART, 12, 378, 379. Vacuum furnace, 461. VALENTINER, 56, 68, 247, 253. gas thermometer, 75. melting points, 7, 440, 442, 443, 444. VAN DER WAAL, 34. VAN DER WEYDE, 444. Vapors, metallic, 107. VERY, 454. VILLARD, 60. VlOLLE, iv, 10, 263, 265, 454. actinometer, 263. gas scales, 23. gas thermometer, 63. melting and boiling points, 5, 64, 65, 438, 439, 44i, 443, 444- specific heats, 63, 91. Viscosity (pyrometers), 376. Visibility curve, 256. VOGT, 368. WAIDNER, viii, 10, 79, 212, 253, 330, 334- freezing and boiling points, 7, 199, 226, 435, 437, 438, 44i, 442, 443, 444, 445, 446, 452, 454, 459. melting points, 9. optical and radiation pyrometry, 240, 243, 253, 307, 320, 323, 332, 335, 337, 346. resistance pyrometer, 199, 200, 205, 228, 231, 232, 233. thermoelectric pyrometer, 185. WANNER, vii, 10, 320, 323, 336, 342, 346, 454- optical pyrometer, 312, 315, 319, 322, 324. WARBURG, 253. INDEX WARTENBERG, 253, 255, 259, 319, 462. melting points, 8, 442, 443, 444, 445, 446. optical pyrometry, 337. WASSMER, 68. boiling points, 7, 437. Wedges, optical, 313, 333. WEDGWOOD, iii, i, n, 358, 359. pyrometer of, i, 357. scale of, 3. WEISS, specific heats, 92, 93. WENNER, 146. WESTON, 129, 131, 136, 137, 140, 212. Wheatstone bridge, for resistance py- rometer, 208, 211. WHIPPLE, 224, 286. WHITE, 99, 146, 161, 451. specific heats, 91, 94. WIBORGH, 83, 376. WIEN, vii, 57, 200, 250, 253, 254, 255, 294, 303, 306, 315, 317, 328, 334, 337, 339, 340, 342, 352, 354- WIEN, gas pyrometer, 67. melting points, 7, 67, 439, 441, 443, 455, 456. radiation laws, 239, 249, 251. resistance pyrometer, 196, 199, 233,. 234- thermoelectric pyrometer, 112, 119,. 126. WILSING, 355, 455. WILSON, 271, 272, 273, 306, 454. WOLFF, 137, 145, 174. WOLOGDINE, 407. WRIGHT, 185. YOUNG, 273. Zinc, boiling point, 62, 63, 65, 66, 88, 438. melting or freezing point, 62, 437, 445- UNIVERSITY OF CALIFORNIA LIBRARY BERKELEY Return to desk from which borrowed. This book is DUE on the last date stamped below. - ^3 1952 LU 150ct'55GB 05 LD 6sl6)476 YC 91170 THE UNIVERSITY OF CALIFORNIA