CHAPTER III

THERMO-ELECTRIC PYROMETERS

=General Principles.=—Seebeck, in 1822, made the discovery that when a junction of two dissimilar metals is heated an electromotive force is set up at the junction, which gives rise to a current of electricity when the heated junction forms part of a closed circuit. Becquerel, in 1826, attempted to apply this discovery to the measurement of high temperatures, it having been observed that in general the E.M.F. increased as the temperature of the junction was raised. No concordant results were obtained, and the same fate befell the investigations of others who subsequently attempted to produce pyrometers based on the Seebeck effect. These failures were due to several causes, but chiefly to the non-existence of reliable galvanometers, such as we now possess. It was not until 1886 that the problem was satisfactorily solved by Le Chatelier of Paris.

Although any heated junction of metals will give rise to an electromotive force, it does not follow that any pair, taken at random, will be suited to the purposes of a pyrometer. A junction of iron and copper, for example, gives rise to an E.M.F. which increases with the temperature up to a certain point, beyond which the E.M.F. falls off although the temperature rises, and finally reverses in direction—a phenomenon to which the name of “thermo-electric inversion” has been applied. Evidently, it would be impossible to measure temperatures in this case from observations of the electromotive force produced, and any couple chosen must be free from this deterrent property. Moreover, the metals used must not undergo deterioration, or alteration in thermo-electric properties, when subjected for a prolonged period to the temperature it is desired to measure. These and other considerations greatly restrict the choice of a suitable pair of metals, which, to give satisfaction, should conform to the following conditions:—

1. The E.M.F. developed by the junction should increase uniformly as the temperature rises.

2. The melting point of either component should be well above the highest temperature to be measured. An exception to this rule occurs when the E.M.F. of fused materials is employed.

3. The thermo-electric value of the couple should not be altered by prolonged heating.

4. The metals should be capable of being drawn into homogeneous wires, so that a junction, wherever formed, may always give rise to the same E.M.F. under given conditions.

It is a further advantage if the metals which fulfil the above conditions are cheap and durable.

The exacting character of these requirements delayed the production of a reliable thermo-electric pyrometer until 1886, when Le Chatelier discovered that a junction formed of platinum as one metal, and an alloy of 90 per cent. of platinum and 10 per cent. of rhodium as the other, gave concordant results. In measuring the E.M.F. produced, Le Chatelier took advantage of the moving-coil galvanometer introduced by d’Arsonval, which possessed the advantages of an evenly-divided scale and a dead-beat action. This happy combination of a suitable junction with a simple and satisfactory indicator immediately established the reliability of the thermo-electric method of measuring temperatures. As platinum melts at 1755° C., and the rhodium alloy at a still higher temperature, a means was thus provided of controlling most of the industrial operations carried out in furnaces.

So far, the effect of heating the junction has been considered without regard to the temperature of the remainder of the circuit, and it is necessary, before describing the construction of practical instruments, to consider the laws governing the thermo-electric circuit, the simplest form of which is represented in fig. 2. One of the wires is connected at both ends to separate pieces of the other wire, the free ends of which are taken to the galvanometer Two junctions, A and B, are thus formed, which evidently act in opposition; for if on heating A the direction of current be from A to B, then on heating B the direction will be from B to A. Hence if A and B were equally heated no current would flow in the circuit, the arrangement being equivalent to two cells of equal E.M.F. in opposition. Thermal junctions are formed at each of the galvanometer terminals, but the currents to which they give rise, when the temperature changes, are opposed and cancel each other. The law which holds for this circuit may be expressed thus:—

“If in a thermo-electric circuit there be two junctions, A and B, the electromotive force developed is proportional to the _difference_ in temperature between A and B.”

It is customary to refer to the two junctions as the “hot” and “cold” junctions; but it is important to remember that fluctuations in the temperature of either will alter the reading on the galvanometer or indicator.

A second law, which applies to all thermo-electric circuits, is that “the E.M.F. developed is independent of the thickness of the wire.” This does not mean that the deflection of the galvanometer is the same whether thin or thick wires are used to form the junction. The deflection depends upon the current flowing through the circuit, and this, according to Ohm’s law, varies inversely as the total resistance of the circuit. Consequently, the use of thin wires of a given kind will tend to give a less deflection than in the case of thick wires, as the resistance of the former will be greater, and unless the resistance of the galvanometer be great compared with that of the junction, the difference in deflection will be conspicuous. The E.M.F., however, is the same under given conditions, whatever thickness of wire be used.

Reference to fig. 2 will show that in order to realise this circuit in practice, one of the wires forming the couple must be used in the form of leads to the galvanometer. This can readily be done if the material of the wire is cheap; but if platinum or other expensive metal be used, and the galvanometer be some yards distant, the question of cost necessitates a compromise, and the circuit is then arranged as in fig. 3. The wires forming the hot junction are brought to brass terminals T T, from which copper wires lead to the galvanometer G. This arrangement results in three effective junctions, viz. the hot junction A to B; the junction A to brass, and the junction B to brass. It will be seen that the two junctions of copper to brass are in opposition, and cancel each other for equal heating; and the same applies to the galvanometer connections. A circuit thus composed of three separate junctions does not permit of a simple expression for the net E.M.F. under varying temperature conditions, and to avoid errors in readings care must be taken to prevent any notable change of temperature at the terminals T T in a practical instrument arranged as in the diagram.

A point of practical utility in thermo-electric work is the fact that if a wire be interrupted by a length of other metal, as indicated at C in fig. 3, no current will be set up in a circuit if both joints are equally heated, as the electromotive forces generated at each junction are in opposition. It is thus possible to interrupt a circuit by a plug-key or switch, without introducing an error; always provided that an even temperature prevails over the region containing the joints.

Another useful fact is that if two wires be brought into contact, they may be fastened over the joint by soldering or using a third metal, without alteration of thermo-electric value, except in rare cases. Thus a copper-constantan or iron-constantan junction may suitably be united by silver solder, using borax as a flux, thus avoiding the uncertainty of contact which must always occur when the wires are merely twisted together. Welding, however, is preferable to soldering.

=Metals used for Thermal Junctions.=—Until recent years it was customary to employ a platinum-rhodioplatinum or platinum-iridioplatinum junction for all temperatures beyond the scope of the mercury thermometer. The almost prohibitive price of these metals has caused investigations to be made with a view to discovering cheaper substitutes, with successful results up to 1000° C. or 1800° F., thus comprehending the range of temperatures employed in many industrial processes. Above this temperature the platinum series of metals are still used for accurate working, but it will be of great advantage if the range measurable by cheap or “base” metals can be further extended. Promise in this direction is afforded by the properties of fused metals when used in thermal junctions. An investigation by the author has shown that in general the E.M.F. developed by a junction does not undergo any sudden change when one or both metals melt, but continues as if fusion had not occurred. By making arrangements to maintain the continuity of the circuit after fusion, it may be possible to read temperatures approximating to the boiling points of metals such as copper and tin, both of which are over 2000° C. The base metals are not so durable as platinum and kindred metals, but as the cost of replacement is negligible, this drawback is of little importance. Moreover, base-metal junctions usually develop a much higher E.M.F. than the platinum metals, which enables stronger and cheaper galvanometers to be used as indicators.

THERMAL JUNCTIONS USED IN PYROMETERS. ─────────────────────────────────────────────────┬──────────────── │ Upper limit to │ which Junction │ may be used. Couple. ├────────┬─────── │ Deg. │ Deg. │ Cent. │ Fahr. ─────────────────────────────────────────────────┼────────┼─────── Platinum and rhodioplatinum (10 per cent. Rh) │ 1400 │ 2550 2 Rhodioplatinum alloys of different composition │ 1600 │ 2900 Platinum and iridioplatinum (10 per cent. Ir) │ 1100 │ 2000 Nickel and constantan │ 900 │ 1650 Nickel and copper │ 800 │ 1475 Nickel and carbon │ 1000 │ 1850 Nickel and iron │ 1000 │ 1850 Iron and constantan │ 900 │ 1650 Copper and constantan │ 800 │ 1475 Silver and constantan │ 800 │ 1475 2 Nickel chrome alloys of different composition │ │ (Hoskin’s alloys) │ 1100 │ 2010 Nickel-chrome alloy and nickel-aluminium alloy │ 1100 │ 2010 2 Iron-nickel alloys of different composition │ 1000 │ 1850 ─────────────────────────────────────────────────┴────────┴───────

The electromotive force developed by a junction of any given pair of metals when heated to a given temperature varies according to the origin of the metals. It is not unusual, for example, for two samples of 10 per cent. rhodioplatinum, obtained from different sources, to show a difference in this respect of 40 per cent. when coupled with the same piece of platinum. Equal or greater divergences may be noted with other metals; and hence the replacement of a junction can only be effected, with accuracy, by wires from the same lengths of which the junction formed a part. As showing how platinum itself is not uniform, it may be mentioned that almost any two pieces of platinum wire, if not from the same length, will cause a deflection on a sensitive galvanometer when made into a junction and heated. It is therefore customary for makers to obtain considerable quantities of wire of a given kind, homogeneous as far as possible, in order that a number of identical instruments may be made, and the junctions replaced, when necessary, without alteration of the scale of the indicator.

The alloy known as “constantan,” which figures largely in the foregoing table, is composed of nickel and copper, and is practically identical with the alloy sold as “Eureka” or “Advance.” It has a high specific resistance, and a very small temperature coefficient, and is much used for winding resistances. Couples formed of constantan and other metals furnish on heating an E.M.F. several times greater than that yielded by couples of the platinum series, and show an equally steady rise of E.M.F. with temperature. This alloy has proved of great service in connection with the thermo-electric method of measuring temperatures. Couples formed of nickel-chrome alloys, known as “Hoskin’s alloys,” have been introduced into Britain by the Foster Instrument Company, which may be used continuously to 1100° C., and for occasional readings up to 1300° C. Another couple, much used in America, consists of an alloy of 90 per cent. nickel and 10 per cent. chromium, and an alloy of 98 per cent. nickel and 2 per cent. aluminium, which may be used up to 1100° C. Other couples, formed of alloys of nickel, chromium, iron, aluminium, etc., have been introduced by different makers, but have not proved so satisfactory as those mentioned above.

=Changes in Thermal Junctions when constantly used.=—No metal appears to be able to withstand a high temperature continuously without undergoing some physical alteration; and for this reason the E.M.F. developed by a given junction is liable to change after a period of constant use. At temperatures above 1100° C., platinum, for example, undergoes a notable change in a comparatively short period, but below 1000° C., the change is very slight, and if this range be not exceeded, a platinum-rhodioplatinum or iridioplatinum junction may be used for years without serious error arising from this cause. This liability to change is one of the factors which restricts the range of thermal junctions, which should never be used continuously beyond the temperature at which the alteration commences to become large. A second cause of discrepancy is the possible alteration in the composition of an alloy, due to one of the constituents leaving in the form of vapour, as is noted with iridioplatinum alloys, from which the iridium volatilises in tangible quantities above 1100° C., causing a fall of 10 per cent. or more in the thermo-electric value of the junction of these alloys with platinum. Constantan appears to be very stable in its thermo-electric properties, and the various junctions in which it plays a part show a high degree of stability if not overheated. Rhodioplatinum alloys are very stable, and for temperatures exceeding 1100° C. a junction of two of these alloys, of different composition, is more durable than one in which pure platinum is used. An extended series of tests on base-metal junctions made in America by Kowalke showed that continuous heating of couples as received from the makers altered the E.M.F. considerably, the change in some cases representing over 100° C. on the indicator. A stable condition, due to the relief of strains or other change, was finally reached, and the conclusion drawn that the materials should be thoroughly annealed before calibration. It is desirable in all cases periodically to test the junctions at some standard temperature, and if any conspicuous error be noted, to replace the old junction by a new one.

In addition to the errors due to slow physical changes, a junction may be altered considerably, if imperfectly protected, owing to the chemical action of furnace gases, or of solids with which the junction may come into contact. The vapours of metals such as lead or antimony are very injurious; and platinum in particular is seriously affected by vapours containing phosphorus, if in a reducing atmosphere. So searching is the corrosive action of furnace gases that adequate protection of the junction is essential if errors and damage are to be avoided. When a wire has once been corroded, a junction made with it will not develop the same E.M.F. as before.

=Electromotive Force developed by Typical Junctions.=—The following table exhibits the E.M.F. generated by several junctions for a range of 100° C., taken at the middle part of the working range in each case. These figures are subject to considerable variation, according to the origin of the metals.

──────────────────────────────────────────┬───────────────────────── │ E.M.F. in millivolts for Couple. │ a rise of 100° at middle │ of working range. ──────────────────────────────────────────┼───────────────────────── Platinum-rhodioplatinum (10 per cent. Rh) │ 1·1 Platinum-iridioplatinum (10 per cent. Ir) │ 1·2 Nickel-constantan │ 2·3 Copper-constantan │ 5·8 Nickel-copper │ 6·1 Iron-constantan │ 6·7 Hoskin’s alloys │ 7·4 ──────────────────────────────────────────┴─────────────────────────

It will be noted that the base-metal junctions give much higher values than the platinum series, and hence can be used with a less sensitive, and therefore cheaper, indicator. Base-metal junctions are also, in consequence of the greater E.M.F. furnished, capable of yielding more sensitive readings over a selected range of temperature.

=Practical Forms of Thermocouples.=—When expensive junctions are employed, wires of the minimum thickness consistent with strength and convenience of construction are used, a diameter of No. 25 standard wire gauge being suitable. A common arrangement is shown in fig. 4, in which J is the hot junction, the wires from which are passed through thin fireclay tubes which serve as insulators (or through twin-bore fireclay) to the reels R R, in the head of the pyrometer, upon which a quantity of spare wire is wound to enable new junctions to be made when required. Two brass strips, S, are screwed down on to the wires at one end, and are furnished with screw terminals at the other end, from which wires are taken to the galvanometer or indicator. A protecting-tube, T, surrounds the wires and hot junction. The head, H, may be constructed of wood, fibre, or porcelain, and should be an insulator for electricity and heat. There are various modifications in use, but the general method described is adopted by most makers. In order to guard against errors arising from alterations in the temperature of the cold junctions in the end of the pyrometer, some firms construct the head so as to leave a hollow space, through which cold water is constantly circulated (fig. 5), the arrangement being known as a “water-cooled head.” In some forms the supply of spare wires is made to take the form of two spiral springs in a hollow head, the upper ends of the springs being taken to terminals.

The choice of a protecting-tube is a matter of considerable importance. Obviously, such a tube should not soften at the highest temperature attained, and when expensive metals are used to form the junction the sheath should not be permeable to gases or vapours. It should also, if possible, be a good conductor of heat, so that the junction may respond quickly to a change of temperature in its surroundings, and should be mechanically strong. It is difficult to secure all these properties in any single material, and the choice of a sheath is decided by the conditions under which the couple is to be used. The substances employed, and their properties and special uses, may be enumerated as follows:—

1. _Iron or Mild Steel._—For temperatures not exceeding 1100° C. iron or mild steel covers are cheap and efficient from the standpoint of conductivity, although liable to deteriorate owing to oxidation. The tendency to oxidise is greatly diminished by “calorising” the exterior by Ruder’s process, in which the iron is heated in a mixture of metallic aluminium and oxide of aluminium, a surface alloy being formed which resists oxidation. A result nearly as good may be obtained by smearing the surface with fine aluminium powder, and bringing to a white heat. This treatment greatly prolongs the life of an iron sheath. Some makers employ an inner steel tube round the wires, and an outer tube which comes into contact with the furnace gases, corrosion of the latter being detected before the inner tube has given way and exposed the junction. Some makers do not consider it safe to expose heated platinum to an iron surface, with only air intervening, and hence use an inner cover of silica or porcelain, which the outer iron or steel tube protects from mechanical damage. For ordinary work seamless steam or hydraulic steel tubing, with a welded end, is satisfactory; but for dipping into molten lead or other metals the tube should be bored from the solid. The great advantage of an iron or steel sheath is its mechanical strength, which protects the couple from damage in case of rough usage.

2. _Nichrom._—Certain alloys of nickel and chromium, and especially that known as Nichrom II, may be kept at 1100° C. without oxidising to any appreciable extent; and hence sheaths of this material may be used up to the temperature named. In addition to being more durable than iron, nichrom possesses the same advantages of strength and good conductivity; on the other hand, it is more costly.

3. _Molybdenum._—This metal, which possesses a melting point of about 2500° C., may be dipped in molten brass, bronze, copper, etc., without being attacked, and has been used to form the tip of a protecting-tube designed to measure the temperature of molten alloys. A junction covered only by a thin tube of molybdenum quickly attains the temperature of its surroundings.

4. _Graphite and Graphite Compositions._—Carbon has the highest melting point of all known substances, and in the form of artificial or Acheson graphite may be easily machined to any desired shape. Graphite sheaths are sometimes used for immersion in molten metals, but at 1000° C. and higher Acheson graphite oxidises easily and becomes friable. It is a good conductor of heat, but is easily broken. Compositions of natural graphite and refractory earths, such as Morgan’s “Salamander,” are inferior to pure graphite in conductivity, but are stronger and not readily oxidised, and may be used to form sheaths for temperatures up to 1400° C. or possibly higher, when penetration of furnace gases to the junction is not of moment.

5. _Porcelain._—This material, in its best forms, may be used up to 1400° C., but must be efficiently glazed to prevent the ingress of furnace gases to the junction. It is easily broken by a blow, and when circumstances permit should be protected by an iron covering-sheath. The variety known as “Marquardt” has been found very satisfactory for high-reading thermal couples. Porcelain is not a good conductor of heat, and a junction encased in it does not respond quickly to external changes in temperature.

6. _Vitrified Silica._—This substance, which may be worked in the oxy-hydrogen blowpipe, is largely used as a protecting-tube. It is not advisable, however, to use it for continuous work above 1100° C., as beyond this temperature devitrification occurs, and the tube becomes porous. It is a fairly good conductor of heat, and withstands rapid changes in temperature without cracking. It is very brittle, and for this reason is generally encased in iron.

7. _Alundum._—This material is made from fused bauxite, and has a melting point of 2050° C. A special form of alundum, used for protecting-tubes, is non-porous up to 1300° C., and forms a satisfactory covering. Alundum is a moderately good conductor of heat, but is easily broken.

8. _Carborundum._—This is an electric furnace product, which may be heated above 2000° C. without damage. For making into pyrometer tubes, it is bonded with a suitable material, and baked after shaping. Carborundum, and the amorphous variety known as “silfrax,” have proved useful for protecting junctions at temperatures as high as 1600° C. The thermal conductivity is relatively good, but the tubes are easily broken.

9. _Magnesia._—Tubes of this material, which melts at a temperature considerably above 2000° C., have been used for special work. Magnesia is a poor conductor of heat, and has little mechanical strength.

10. _Zirconia._—This is a very refractory material, its melting point exceeding 2500° C. It may be made into a vitreous variety, which is non-porous and proof against sudden temperature changes. At present, only a moulded form of pyrometer tube, made from zirconia powder, is available, the material worked in this manner being termed “zirkite.” Although zirconia is a bad conductor of heat, its other qualities are such that it forms an excellent material for work at the highest temperatures possible for thermal junctions; and when the vitreous variety is available, may come into extended use.

It will be seen from the foregoing that the ideal protecting-tube has yet to be found, and the user must choose the one which comes nearest to his requirements. Special consideration must be given in cases when chemical fumes are present, and a sheath selected which is not attacked or penetrated by them.

Returning to the junction, it is advisable always to weld the wires, and not to rely upon the contact resulting from twisting them together. Platinum and the platinum alloys may be welded readily by placing the junction in a coal-gas blowpipe fed with oxygen instead of air. For work at lower temperatures the platinum metals may be soldered by means of a small quantity of gold, in the flame of a Bunsen burner.

When cheap metals are used for the junction the construction may be considerably modified, and often with advantage. In fig. 6, for example, which represents a thermocouple made by A. Gallenkamp & Co., the metals used are copper and constantan, and the hot junction, fastened by silver solder, is supplemented by a cold junction of the same metals located in the head. The copper wire from the hot junction passes directly to a copper terminal, from whence a copper wire lead is carried to the galvanometer; and the same procedure is carried out with the copper wire from the cold junction, thus realising the circuit shown in fig. 2. The cold junction is kept in oil, the temperature of which is registered by a short thermometer, thus enabling (as will be explained later) the correct temperature of the hot junction to be deduced under any circumstances. In this instrument twin-bore fireclay is used to insulate the wires, and the protecting-tube is of iron—which suffices for the upper limit (800° C.) to which the junction may be used. Iron and constantan could be used in this manner by employing iron leads to the galvanometer.

Another type of instrument, rendered practicable by the use of cheap metals, and which may be termed the “heavy type,” is constructed of thick pieces of the metals welded together instead of wires, thus ensuring greater strength and longer life. Messrs Crompton & Co. were the first to introduce thermocouples of this type, consisting of a heavy steel tube, to one end of which a nickel rod is welded, the other end being free, and the length of the rod suitably insulated from the steel tube; leads for the rod and tube being taken to the galvanometer. Fig. 7 shows a couple of this kind, made by Paul, consisting of an iron tube down the middle of which a constantan rod is passed, insulated from the tube by magnesia. At the tapered end the two metals are welded together, and at the free end a special cap, fitted over the tube and rod, the contact parts being insulated from one another, serves to enable leads to be taken to the galvanometer. Similar thermocouples are made by the Foster Instrument Company (fig. 8), and are simple, cheap, and reliable up to 900° C. with an iron-constantan couple, and to 1100° C. with nichrom couples. When worn out they may be replaced, at a trifling cost, by others made from the same batch of metal.

The drawback to the use of carbon as one of the materials for a junction is the difficulty experienced in securing a good contact with the metal with which it is coupled. In nickel-carbon junctions the contact is sometimes ensured by the aid of a spring, which presses the two substances together. Such an arrangement is evidently not so reliable as one in which the materials are welded, and a defective contact, arising from any cause, would lead to serious error. A preferable plan is to screw both the nickel and carbon rods into a cross-piece of either element.

When applying a thermal junction to the measurement of surface temperatures, such as steam-pipes or the exterior of furnaces, the wires may be passed through a thin disc of metal, about ¼ in. in diameter, and soldered at the back. Suitable materials are copper and constantan, soldered to a thin copper disc with silver solder, and brought to a cold junction in the head of the instrument as shown in fig. 6. The terminal piece of the insulation may be made of hard wood, with the holes countersunk so as to cover the solder and enable the wood to touch the disc, which, when pressed on the hot surface, will then rapidly acquire the temperature. The author has found, by trials under varying circumstances, that this method of measuring surface temperatures gives reliable and concordant results. For very high surface temperatures a platinum disc, with one of the usual platinum metal couples soldered to the disc with pure silver, and a piece of twin-bore fireclay brought to the back of the disc, will be found to suffice for most cases arising in practice. A small blowpipe flame is best for soldering the wires to the disc, borax being used as flux in the first case; but no flux is necessary in soldering the platinum metals with pure silver.

In deciding upon the length of a thermocouple it must be remembered that the temperature recorded is that prevailing in the region of the hot junction. When the temperature of a furnace is uniform it is sufficient to allow the end of the thermocouple to protrude about 12 inches into the interior, but when following the change of temperature undergone by objects in a furnace the end must be located near the objects. If the distance from the exterior of the furnace to the objects exceed 2 feet, the thermocouple should be inserted through the roof so as to hang vertically, as if placed through the side it would droop by its own weight at high temperatures. The distance between the exterior of the furnace and the cold junctions should be at least 15 inches in all cases in which the heating of the cold junction is not automatically compensated. After inserting the couple the opening through the furnace wall should be closed by means of suitable luting-clay.

In certain instances, such as flues, it is necessary to use a long instrument in a horizontal position. A rail may then be placed across the flue, at a suitable place, to serve as a support and so to prevent drooping.

=Liquid Element Thermocouples.=—An investigation by the author and A. W. Grace has shown that the continuity of the E.M.F. produced by a rising temperature is not interrupted by fusion, except in the cases of bismuth and antimony, which both show an abrupt change in thermo-electric properties at the melting point. It would therefore appear feasible to measure temperatures by constructing a thermocouple so as to retain the circuit after fusion, the advantage gained being that the range is restricted by the boiling point of the metals instead of the melting point and higher readings are rendered possible. The boiling points of some of the common metals are appended:—

───────────┬─────────────────── Metal. │ Boiling Point. ───────────┼─────────┬───────── │ Deg. C. │ Deg. F. ├─────────┼───────── Aluminium │ 1800 │ 3270 Silver │ 1955 │ 3550 Tin │ 2270 │ 4120 Copper │ 2310 │ 4190 Nickel │ 2330 │ 4225 Iron │ 2450 │ 4440 ───────────┴─────────┴─────────

From inspection of these figures, it will be seen that if a suitable couple could be obtained, common metals might be used to measure temperatures equalling or even exceeding the limit of the range covered by wire junctions of metals of the platinum series. Instead of using two metals, graphite might form one member of the couple, provided that no objection to its use existed on other grounds.

The form of thermocouple designed by the author to permit of the use of molten elements is shown in fig. 9. A rod of refractory material, R, is perforated longitudinally by two holes, down which are passed rods of the thermo-elements, A and B. The lower ends of A and B are inserted in a graphite block G, which is jointed on its upper face to R; the whole being surrounded by the refractory cover C. On either or both of the elements melting, the circuit is maintained through G, which serves also to prevent the mixing of A and B when molten, whilst not affecting the E.M.F. developed. In order to allow for the expansion of the metals on melting, A and B are made to fit loosely in R. When inserted in a furnace to a depth represented by EF, only the portion of the metals adjacent to the closed end will melt, the outer parts remaining solid. At present it has not been found possible to procure the refractory parts in a form suited to commercial use, but when this obstacle is overcome this type of thermocouple should prove of service for measuring temperatures beyond the scope of ordinary base-metal junctions.

=Indicators for Thermo-electric Pyrometers.=—As the electromotive force developed by a single junction when heated is small, a sensitive galvanometer is required to indicate the minute current flowing through the circuit. Delicate millivoltmeters, of the moving-coil type, are universally employed, as they possess the advantage of an evenly-divided scale combined with the requisite degree of sensitiveness. The original d’Arsonval galvanometer, consisting of a coil suspended by a metallic strip between the poles of a horse-shoe magnet, was used by Le Chatelier, who, by its aid, was enabled to lay the foundations of this branch of pyrometry. Three forms of this instrument are now in use, viz. (_a_) the suspended coil “mirror” type; (_b_) the suspended coil “pointer” type; and (_c_) the pivoted type. Examples of each will now be described.

Fig. 10 represents a mirror galvanometer working on the d’Arsonval principle, designed by Gen. Holden, F.R.S. The horse-shoe magnet is laminated, and an iron core, supported by a pillar, is placed between the poles. The coil, which moves in the space between the core and the poles of the magnet, is suspended by a thin, flat strip of phosphor-bronze, which carries a small circular mirror. A similar phosphor-bronze strip is fastened to the lower part of the coil, and is continued to an adjusting-screw in the base. The ends of the suspension strips communicate with the terminals of the galvanometer, and a current entering at one terminal passes through the metallic suspensions and the coil to the other. The effect of passing a current through the coil, which is located in a powerful magnetic field is to produce an axial movement tending to twist the suspension strips, which movement is greatly magnified by a spot of light reflected from the mirror on to a distant scale. When the current ceases, the untwisting of the strip restores the coil to its former position. Galvanometers of this type are remarkably “dead-beat” in action, that is, the movement and restoration of the coil are accomplished without vibration. A semi-transparent scale, placed at 1 metre distance, and 50 centimetres long, is suitable for use with this galvanometer. When used in workshops, it is necessary to protect a mirror galvanometer from the vibrations produced by machinery, which would cause the spot of light to become unsteady. The best method of effecting this is shown in fig. 11, which represents the mode of suspension devised by W. J. Lambert for use in the Royal Gun Factory, Woolwich Arsenal. The usual supports of the galvanometer are abolished, and the instrument suspended from the ring of a brass tripod, so as to keep three springs partly in compression. When suspended in this manner, a mirror galvanometer is quite suited to commercial use; in the quiet of the laboratory the ordinary supports may be employed. The advantage gained by using the mirror type is that a much longer scale is possible than with instruments furnished with a pointer, and hence greater accuracy in determining temperature readings may be secured.

In suspended coil instruments furnished with a pointer, the construction differs only in detail from the foregoing. In place of the mirror, a light pointer is attached to the suspension so as to rest on the coil and a scale is furnished over which the pointer moves. Fig. 12 is an example of this type, made by Messrs Siemens, the suspension being contained in the tube which rises from the body of the instrument. The maximum length of scale moved over by the extremity of the pointer is about 6 inches, as a longer and therefore heavier pointer would reduce the sensitiveness below the point requisite for thermo-electric work.

In the double-pivoted type, the suspension is eliminated, and pivots are fastened to each end of the moving coil which rest in bearings. The turning of the coil is made to compress a hair spring, made of phosphor-bronze; and when the current ceases the unwinding of this spring restores the coil to its former position. The coil carries a pointer which moves over a scale. These instruments are not so sensitive as those in which the coil is suspended, but can be made sufficiently sensitive to work with any kind of junction in practical use. The pivoted form is cheaper and stronger than the suspended type, and is used whenever sufficiently sensitive.

The “Uni-pivot” galvanometer, made by R. W. Paul, is shown in figs. 13 and 17. The coil, which carries the pointer, is circular, and moves round a spherical core of iron placed between the poles of the magnet. A hole is drilled in the iron core, and the coil rests on a single bearing at the bottom of this hole. A phosphor-bronze control-spring serves to restore the coil to the zero position. The lessened friction due to the use of a single pivot enables this instrument to be made very sensitive when needed, so that a relatively small rise in the temperature of a junction may cause the pointer to traverse the whole length of the scale.

=Special Features of Indicators.=—All moving-coil instruments, whether suspended or pivoted, are liable to alteration of the zero point owing to what is termed “creep.” The suspension strip, when first fixed in position, generally possesses a certain amount of initial torsion, which comes into operation gradually and causes a slight movement of the coil. Similarly, in a pivoted instrument, the strength or shape of the control-spring undergoes a gradual alteration at first, causing the pointer to move away from the zero position. For this reason adjusting arrangements are fitted by means of which the spot of light or pointer may be brought back to the zero. This creeping ceases after a time—often requiring twelve months—and if not subjected to any strain, error from this cause does not recur to any notable extent. With a mirror galvanometer it is better to move the scale, or turn the galvanometer round on its axis to restore the correct zero, rather than to twist the coil back; but with a fixed scale and pointer the only remedy is to turn the coil bodily round. In a single-pivot indicator constantly used in the author’s laboratory, the creep amounted to a movement of the end of the pointer through an angle of 2 degrees in the first few months, since when, after the lapse of several years, no further alteration has occurred. It is advisable to test the zero point of an indicator from time to time by breaking the circuit, and if an error be discovered the pointer should be re-set, or an allowance made in taking a reading.

The resistance of an indicator should be so high that the readings should not be perceptibly altered by any fluctuations in the resistance of the circuit which may arise in practice. If leads of considerable length were used to connect the pyrometer with the indicator, and were subject to fairly large alterations of temperature, the consequent changes in the resistance of such leads would be noticeable on a low-resistance indicator; and similarly, if a pyrometer were inserted at different depths in a furnace at separate times, thus heating up varying lengths of the junction wires, a discrepancy would arise for the same reason. The resistance of an indicator, however, cannot be raised beyond a certain point without reducing the sensitiveness below the required limit. A mirror galvanometer of the type described may have a resistance—partly in the coil and partly in an added series resistance—of 1000 ohms or more, and still be sufficiently sensitive; and in the latest types of instruments provided with pointers the resistance may be made as high as 1000 ohms, although it is more usually 400 to 500 ohms. Many indicators are in use, however, in which the resistance is 100 ohms or less. As, from Ohm’s law, the current varies inversely as the total resistance in the circuit, any alteration in resistance should be small relatively to the total to render the error negligible. This point is made clear in the following example:—

_Example._—A thermocouple and leads have a resistance of 5 ohms and are subject to alterations amounting to 1 ohm. To find the errors resulting when indicators of resistances 800, 400, and 50 ohms respectively are used.

From Ohm’s law, C = E/R, the variation in C, with E constant, will be 1 in 805, 1 in 405, and 1 in 55 respectively. As the indications are proportional to the current, the alterations caused will be approximately ⅛ per cent., ¼ per cent., and 2 per cent. The first two may be ignored; the last may be quite serious and lead to the failure of an operation.

It will be seen from the foregoing that low-resistance indicators should only be used for fixed thermocouples and short leads not subject to temperature changes, or, in other words, in a circuit of fixed resistance.

The resistance of an indicator, when unknown, may be found by the following method, suggested by the author:—A resistance box is joined at one end to one terminal of the indicator. To the other terminal a fairly stout iron wire, 18 inches long, is connected, and a similar length of constantan wire is coupled to the other end of the resistance box. The free ends of the wires are twisted into a junction which is dipped into boiling water. The deflection obtained with no resistance in the box (D_{1}) is noted, and resistances (R) are then unplugged until the deflection (D_{2}) is approximately one-half of D_{1}. The resistance (G) of the indicator, ignoring that of the wires, is then given by the formula

D_{2}R G = ─────────────── (D_{1} - D_{2})

as may readily be proved from Ohm’s law, E being constant. This method is extremely simple and reasonably accurate.

Reliable indicators are now procurable from many instrument-makers at a comparatively small cost, progress in this direction having been most marked in recent years, particularly in the case of pivoted instruments. The most convenient form for workshop use is made with an edgewise scale (fig. 14) and may be placed in a suitable position fixed to a bracket. The flat-scale pattern is preferable for use on a laboratory table, or for a portable pyrometer. The sector pattern is also good for workshop use, the dial being visible from a distance.

=Standardizing of Indicators to read Temperatures directly.=—The temperature scale of an indicator, for use with a given thermal couple, is always marked by the maker in the case of instruments furnished with a pointer, and, generally speaking, is correct within reasonable limits. It is customary and necessary to send with the instrument a statement of the cold-junction temperature for which the markings are correct; say 20° C. or 60° F. The user should then endeavour to maintain the cold junction at this specified temperature when taking a reading, or otherwise a considerable error may be introduced. It is highly desirable, however, that the user should be able to perform the standardizing himself, if only for checking purposes; and when using a mirror galvanometer as indicator it is necessary to standardize on the spot at which the instrument is fixed. Ability to prepare a temperature scale is further useful, inasmuch as any good millivoltmeter, of range 0 to 20 millivolts, may be used for thermo-electric work of all kinds, and may be calibrated for different junctions, a suitable series resistance being added to enable E.M.F.’s higher than 20 millivolts to be measured. Such an instrument may thus be made extremely useful, both in the workshop and laboratory.

Standardization may be effected either by subjecting the hot junction to several known temperatures, and noticing the deflections corresponding thereto; or by measuring the electromotive force developed by the junction, and calculating the corresponding temperature from a formula which is known to hold for the range comprehended by the instrument. The former method is simpler; and if carefully conducted is quite accurate. The latter method possesses the advantage that readings in millivolts may be translated directly into temperatures when the constants of a given thermal couple are known. It is now usual to mark indicators with a double scale, one reading millivolts and the other temperatures.

=Standardization by Fixed Points.=—Taking any millivoltmeter which, with a maximum of 20 millivolts at the terminals, will give a full scale deflection, the first step is to arrange that the pointer (or spot of light) shall just remain on the scale at the highest temperature to be attained by the junction. This may be done by placing the hot junction in boiling water and noting the deflection obtained, either in millivolts or equal arbitrary divisions, and also the temperature of the cold junction. The deflection observed is due to a difference of temperature (100-_t_) deg. C, where _t_ is the temperature of the cold junction. If the highest temperature to be measured is 10 times (100-_t_), the deflection should be rather less than 1/10 of the scale, and similarly for any other required temperature limit. If the observed deflection exceed this proportion, a series resistance should be added until the correct value is obtained. This resistance is then permanently installed in the circuit for use with the junction under trial.

Before proceeding further it is necessary to consider whether the pyrometer is to possess a single cold junction of ascertainable temperature (as in fig. 6), or whether it will be arranged with two cold junctions in the head, as in fig. 4. In the former case it is simpler to prepare a “difference” scale; that is, one which reads differences of temperature between the hot and cold junctions, from which the temperature of the hot end may be obtained by adding to the difference that of the cold junction. In the latter case the cold end should be kept by artificial means at the temperature likely to be attained in practice—say 25° C.—a water-bath being suitable for this purpose. It is advisable to remove the shield of the pyrometer when standardizing, so as to expose the hot junction, as closer readings can then be taken.

A number of materials—preferably cheap—of known boiling points or melting points are then selected from a table of fixed points (page 16) so as to give about six points, distributed fairly evenly over the scale. As an example, if it were desired to prepare a temperature scale from 0° to 1000° C., the following might be chosen:—

════════════════════════════════════════════════╤═══════════════════ Substance and Condition. │ Temperature. ────────────────────────────────────────────────┼─────────┬───────── Water at boiling point │ 100° C. │ 212° F. Tin at melting point │ 232 │ 449 Zinc at melting point │ 419 │ 786 Antimony at melting point │ 631 │ 1167 Common salt at melting point │ 800 │ 1472 Copper at melting point (covered with graphite) │1084 │ 1983 ════════════════════════════════════════════════╧═════════╧═════════

The hot junction is allowed to attain these temperatures successively, and the corresponding deflection in each case is noted. It is then possible to divide up the whole of the scale to read temperatures directly.

The first reading is taken by placing the junction in a vessel of boiling water, and for a locality near sea level it is not necessary in ordinary work to take account of fluctuations in the boiling point due to alterations of atmospheric pressure. To ensure that the other readings are taken when the substances are exactly at the melting point, the procedure is as follows: about 2-3 lb. of the substance are melted in a salamander crucible, and a small fireclay tube, closed at one end, is inserted in the molten mass. The hot junction is placed in the fireclay tube, and the intervening space filled with asbestos fibre. Great care must be taken not to let the junction touch the fused substance. The crucible is now allowed to cool, and a reading of the deflection taken every half-minute. When the substance is exactly at its solidifying point—identical in general with the melting point—the deflection remains stationary for several consecutive readings, owing to the liberation of latent heat of fusion in sufficient quantity to balance the loss by radiation. This stationary reading is noted for each substance, and represents the deflection given when the hot junction is at the temperature corresponding to the melting point, and the cold junction or junctions at the temperature existing when the observation is made. For melting the materials, a Davies furnace with a large Teclu or Meker burner is convenient up to 850° C.; but to melt the copper a blast lamp is requisite. The molten mass may be allowed to cool in the furnace.

From these observations a calibration curve may be drawn either for differences between hot and cold junctions, or for a steady temperature of the cold junctions. Two sets of data are appended to illustrate the procedure.

════════════╤═════════════════════════════════╤═════════════════════ Temperature │ Pyrometer 1. Iron-constantan. │ Pyrometer 2. of Hot │ (Series resistance in │ Platinum- Junction. │ galvanometer circuit.) │ iridioplatinum. ├───────────┬─────────┬───────────┼───────────┬───────── │Deflection.│ Cold │Difference.│Deflection.│ Cold │ │Junction.│ │ │Junction. ────────────┼───────────┼─────────┼───────────┼───────────┼───────── 100° C. │ 8·9 │ 15° C.│ 85° C. │ 5·5 │} 232 │ 21·8 │ 17 │ 215 │ 15·6 │}Constant 419 │ 40·6 │ 19 │ 400 │ 29·4 │} at 631 │ 63·8 │ 19 │ 612 │ 45·5 │} 25° C. 800 │ 83·0 │ 20 │ 780 │ 59·0 │} 1084 │ ... │ ... │ ... │ 82·0 │} ════════════╧═══════════╧═════════╧═══════════╧═══════════╧═════════

Fig. 15, A, is a calibration curve for thermocouple 1, connecting deflections with corresponding differences between the temperatures of the hot and cold junctions. In order to read from this curve the temperature of the hot end, the reading corresponding to the observed deflection is added to the existing temperature of the cold junction. Thus if a deflection of 56 divisions were obtained with the cold junction at 25°, the temperature of the hot junction would be (540 + 25) = 565° C. The advantage of this method of calibration is that it is unnecessary to take precautions to keep the cold junction at a steady temperature; and when a single cold junction is used, as in fig. 6, this plan should always be followed. It will be noted that this curve passes through zero, as no deflection represents no difference of temperature.

Fig. 15, B, represents the calibration curve for pyrometer 2, and is such that direct readings may be obtained corresponding to any given deflection, for a cold junction temperature of 25°. This curve, therefore, cuts the axis of zero deflection at 25°, as no deflection corresponds to the condition when both hot and cold junctions are at 25°. This method of calibration may be used with advantage for couples of the type shown in fig. 4, where two cold junctions exist in the head, and the simple rule of adding the cold junction temperature does not apply. Many suggestions have been made for correcting for alterations in the temperature of the cold end of such a couple, but none are accurate, and it is necessary to keep this part at the temperature of standardization to secure correct readings. In both of the above calibrations the galvanometer used possessed a scale divided into 100 equal arbitrary divisions.

In making permanent temperature scales from these curves to attach to the existing galvanometer scale, intervals of 100° may be taken and marked opposite to the corresponding divisions on the existing scale. Each 100° may then be equally subdivided into as many parts as the length of scale permits, and numbered at suitable intervals. If the junction used yield a calibration curve departing greatly from a straight line, every 50° interval should be taken, or, if necessary, every 25°. In the examples given both curves are nearly straight lines in the working region, viz. 400° to 800° for the iron-constantan junction, and 500° to 1100° for the platinum-iridioplatinum.

One precaution necessary in standardizing an indicator by this method is to ensure that the metals used are pure, as impurities lower the melting points. If ordered as “pure” from any dealer of repute, the metals will generally be found satisfactory. The common salt used should be the ordinary salt sold in blocks, and not a prepared table salt. A second precaution, when observing melting points, is to guard against a possible error due to the substance becoming “surfused” or “overcooled”; in which case the temperature falls below the ordinary freezing point before solidification commences. When freezing occurs, however, the temperature rises to and remains at the true melting point, and an increase of deflection following a gradual fall always indicates overcooling. The higher deflection then attained is the true freezing point. Antimony frequently overcools to 600° before freezing, but on setting rises to the correct figure—631°. All metals and salts are liable to overcooling occasionally.

=Standardization by Measurement of E.M.F.=—It has been found, as the result of experiments, that the relation between the E.M.F. developed by a junction and its temperature—under constant conditions of the cold junction—may be expressed approximately by a formula as under:—

log E = A log _t_ + B (Holman’s formula),

where E = electromotive force in microvolts, _t_ = temperature in Centigrade degrees, and A and B are constants depending upon the junction. With certain junctions this formula may be applied over the working part of the scale with an error not exceeding 2° C., but with others the discrepancy is greater. In order to determine the constants A and B, it is necessary to measure the E.M.F. at two known temperatures, which should be chosen as far apart as possible in the working region. When these constants are known, a measurement of E enables the temperature _t_ to be found by calculation.

_Example._—Le Chatelier found that a junction at the temperature of melted aluminium (657° C.) gave 6200 microvolts; at the melting point of copper in air (1062° C.) the figure was 10580. Applying in the above formula log 6200 = A log 657 + B and log 10580 = A log 1062 + B, the value of A is 1·2196 and of B 0·302, as may be found by taking logarithms and solving for A and B.

The values of the constants A and B vary for different junctions, and also for different melts of what are reputed to be the same materials. When once determined for a quantity of homogeneous wires, to which the formula applies with sufficient accuracy, it is evident that an indicator with a millivolt scale may be made to read temperatures directly without any necessity for further experiment, although it is always advisable to take one check reading at a fixed point in the working range.

In order to determine the E.M.F. of a junction at different temperatures, the potentiometer method is used, in which the E.M.F. of the test-couple is balanced against the known E.M.F. furnished by a constant cell. The circuit is shown in fig. 16, in which B is an accumulator which sends a current through the resistances R_{1}, R_{2}, and the calibrated wire DE. The cold ends of the couple are attached at P so as to be in opposition to B, and in this branch of the circuit are included a sensitive galvanometer G and a portion of the wire DE. A standard cadmium cell, S, is connected between R_{1} and R_{2} at one end, and may be put in circuit with the galvanometer through the switch A. In commencing, S is connected to the galvanometer and R_{1} adjusted until no deflection is obtained on G. The switch A is now moved over to the circuit of the couple, and the terminal F moved along the wire until zero deflection is again obtained. The E.M.F. of the couple is determined from the relation

E of junction Resistance of DF ────────────────── = ───────────────── . E of standard cell R_{2}

By exposing the hot end of the junction to successive standard temperatures, and maintaining the cold ends at a known constant temperature, the necessary data for inclusion in a formula may be obtained.

In fixing a permanent temperature scale, calculated from the formula, to a millivoltmeter, it must be remembered that the values given by the experiment are absolute, and independent of the resistance of the circuit composed of the thermo-element and galvanometer. On the other hand, a millivoltmeter is marked to read difference of potential at its terminals; and if in series with a junction and leads of notable resistance, its indications will not be the E.M.F. of the junction. An example will make this point clear.

_Example._—A millivoltmeter has a resistance of 100 ohms, and is marked to read P.D. at its terminals. A thermocouple and leads connected to the millivoltmeter have a resistance of 5 ohms. To find the relation between the true E.M.F. of the junction and the readings of the indicator.

If E = the E.M.F. developed by the junction, and V, the reading of the millivoltmeter, = P.D. at its terminals, then the current in the circuit = E/105 = V/100; and V = (100/105)E. That is, the readings are lower by 5 per cent. than the true E.M.F. of the junction. In the same way a low resistance voltmeter, if applied to a cell of high resistance, shows a lower reading than the E.M.F. of the cell.

This example indicates how a table connecting true E.M.F.’s with reading in millivolts may be calculated when the resistances concerned are known. It is presumed, in preparing a scale in this manner, that the resistance of the couple will not be subject to such alterations as to affect the reading.

The advantages of this method of calibration are manifest when a number of junctions are being made from a given batch of wires, as it is only necessary to divide the scale of the indicator so as to represent millivolts—a simple operation—and then to attach a temperature scale. This procedure is much more expeditious than standardizing each indicator at several fixed points when a number are concerned, but for a single junction the fixed point method is easier. The potentiometer method of measuring E.M.F. may also be used to determine temperatures in place of an indicator, and is of great service in cases where very accurate readings are specially required, being far more delicate in detecting small differences of temperature than an indicator. Special potentiometers for thermo-electric work are made by the Cambridge and Paul Instrument Company, Siemens, and others, and are useful in conducting accurate research, but are too elaborate for workshop or ordinary laboratory practice.

=Cold Junction Compensators.=—The necessity for paying attention to the cold junction has led to various attempts to compensate automatically for changes of temperature at this part of the pyrometer. A thermometer located near the cold junction, as in fig. 6, is all that is needed to correct a two-junction circuit; but when a three-junction circuit is used a correct reading is not secured by adding the excess temperature of the thermometer over the calibration temperature to the reading on the indicator. In Bristol’s arrangement a mercury thermometer, with a large bulb and wide stem, is stationed at the cold junction, and participates in any temperature change. In the stem is placed a loop of thin platinum wire, which forms part of the pyrometer circuit. When the mercury is heated it expands up the stem and short-circuits a portion of the loop, thereby diminishing the resistance of the pyrometer circuit, and tending to increase the deflection on the indicator. Simultaneously the cold junction will be heated, tending to diminish the current, and so to cause a less deflection. By adjustment these two tendencies may be counterbalanced, so that the reading is unaffected, but such adjustment will only apply to a given E.M.F., and therefore to one temperature of the hot junction. Hence this method fails in general application.

Peake’s compensated leads are intended to remedy cold-junction errors by transferring this junction, in effect, to the galvanometer. They are used for pyrometers in which the platinum metals are employed, and consist of wires of two different alloys of copper and nickel, which connect the cold end to the indicator. These alloys are such that the electromotive forces set up at the junctions in the head—Pt and Cu-Ni 1, and Pt-Ir with Cu-Ni 2—are equal and opposite at all working temperatures, and hence changes at the cold junctions do not affect the reading. At the indicator, however, temperature changes would cause an alteration in deflection; but as the indicator is generally placed well away from the furnace, and is not liable to notable heating or cooling, the possible errors are greatly reduced by the use of these leads. They are obviously of no value for use with base-metal pyrometers, as the wires used in such may be prolonged to the indicator, with an identical result.

An automatic compensator for use with base-metal pyrometers has been devised by the author, and is illustrated in figs. 17 and 18. A spiral made of a compound strip of two metals is attached to the needle of the indicator, and coils or uncoils when cooled or heated, thereby moving the pointer over the scale. The length of the spiral is such that an alteration of a given number of degrees in its temperature moves the pointer by the same number of degrees on the scale—or, in other words, the temperature scale of the pyrometer is identical with that of the spiral. The metals forming the junction are continued, in the form of wires, to the interior of the galvanometer, where a cold junction is formed, which will always possess the same temperature as the spiral. The scale is constructed to represent differences of temperature between the hot and cold junctions, and before coupling up the pyrometer the pointer indicates the temperature of the spiral; that is, of the cold junction. On connecting the thermocouple the pointer is moved by the coil of the indicator through an amount represented by the difference in temperature between the two junctions, and therefore finally indicates the temperature of the hot junction.

_Example._—If the cold junction were at 20°, the pointer, before connecting the couple, would indicate 20° on the scale. If the hot junction were 580° hotter than the cold, then on completing the circuit the pointer would move 580 additional degrees along the scale, so that the figure indicated would be (20 + 580) = 600°, the temperature of the hot junction. If now the indicator were heated by 10°, the spiral would tend to augment the deflection by 10°, but simultaneously the deflection due to the junctions would fall off by 10°, and the reading would still be 600°.

This method of compensation renders the readings independent of the cold junction, and, in addition to its use for high temperatures, enables ordinary and low temperatures to be read simply and correctly, as will be shown later. The spiral is located in the tower rising from the top of the indicator in fig. 18.

In Paul’s method of compensation the thermocouple and indicator are placed across a Wheatstone bridge, two arms of which contain resistances of copper, whilst the resistances in the other two arms are of manganin. Any change in temperature at the cold junction is shared by these four resistances, and, whilst affecting the resistance of the copper parts, no change is caused in the manganin parts, as this alloy has a negligible temperature coefficient. If, therefore, the bridge were initially balanced at 20° C., and the temperature rose to 30°, the increased resistance of the copper would destroy the balance, and permit of a small current passing through the indicator. A fall to 10°, by diminishing the resistance of the copper, would cause an equal current to pass through the indicator in the opposite direction. The amount of this current is arranged so as to add the rise in temperature of the cold junction to the reading of the indicator in the one case, and to subtract the fall in the other, thus retaining true readings for the cold-junction temperature at which the couple was standardized.

=Constant Temperature Cold Junctions.=—If the cold junction can be kept at a steady temperature, compensators are unnecessary, but no good practical means of achieving this end has yet been devised. Water-cooled heads have already been referred to; but in many situations the connecting-pipes entailed would be objectionable, and hence this arrangement is not greatly used. An alternative method, suggested by Prof. A. Zeleny, is to bury the cold junction in the ground. Recent experiments, conducted at Cambridge by R. S. Whipple, showed that a junction buried 10 feet deep did not vary in temperature by more than 2° C. over a period of three years. This has led to the adoption of buried junctions in special cases; but it is probable that much greater variations would be experienced in the ground beneath large furnaces, in which case the advantages of this procedure would be lost. A common workshop method is to locate the cold junction in a thermos flask filled with oil, when a temperature constant to 2° C. may be secured, although the changes in the temperature of the surrounding atmosphere may be as great as 150 C. For special work, ice may be used in the thermos flask, thus securing absolute constancy; but this procedure is not feasible in ordinary works practice.

=Special-Range Indicators.=—When the working range of a pyrometer is from 600° C. upwards, it is evident that the part of the scale occupied by the first 600° is useless, and that it would be an advantage if the whole scale could be utilised for the special working range, so as to secure more exact readings. This may be accomplished by a “set-up” against the movement of the pointer caused by the thermocouple, so as to prevent any motion over the scale until an assigned temperature is reached. For example, a junction developing 12 millivolts at 1000° C. may be coupled to an indicator in which the full-scale deflection of the pointer is produced by 6 millivolts. If an E.M.F. of 6 millivolts be opposed to the junction, no deflection will occur until the temperature at which the couple develops 6 millivolts is reached—when the opposing E.M.F. will be overcome. This temperature may be 500° C., so that the whole scale may be divided up between 500° and 1000°. The length of the indicator scale is thus effectively doubled; and by using different values for the set-up, it is evident that any desired range may be obtained within the limits of sensitivity of the indicator. The method of procuring the opposing E.M.F. varies with different makers. The Cambridge and Paul Instrument Company employ a dry cell and a series resistance, connected so as to oppose the thermocouple; and by adjusting the resistance any desired set-up may be obtained, the value of which, in degrees, may be read off by connecting the cell and resistance to the indicator, the couple having been switched out of the circuit. Thus, to adjust for a range of 500°-1000° on an indicator giving full-scale deflection for 500°, the resistance is regulated so that the cell alone causes the pointer to move to the end of the scale. The method adopted by Paul consists of suitable resistances inserted in a Wheatstone bridge, which may be thrown off the balance, and thus cause an opposing E.M.F. of the correct amount at the terminals of the indicator.

A mechanical set-up has been introduced by the Cambridge and Paul Instrument Company, the indicator in this case having a suspended coil. By turning a milled-head a twist may be given to the suspending strip, and by the turning of a second head the pointer may be brought back to zero, retaining the initial twist, which is opposed to that produced by the current due to the couple. Thus, if the imposed twist were such as to move the pointer to the 400° mark on the scale, the temperature indicated by the junction would be the observed reading plus 400. By this method it is possible to obtain any desired range within the limits of the indicator. The danger of producing errors due to “creeping” is said to be negligible.

=Potentiometer Indicators.=—The advantage of measuring E.M.F. by the potentiometer method is that the result is independent of the resistance of the circuit under test, whereas an indicator is affected by changes in the resistance of the circuit in which it is inserted. When long leads are used to connect a couple to its indicator, notable errors may be caused by the varying resistance of the leads, due to changing temperature; and, in addition, the resistance of the couple-wires varies according to temperature and depth of insertion in the furnace. Attempts have therefore been made to produce indicators based on the potentiometer principle, suitable for workshop use, and one form, known as Northrup’s “Pyrovolter,” is arranged as shown in fig. 19, A. A cell D sends a current through a rheostat R, a copper coil C, and a manganin coil S. The copper coil has the same resistance as the copper winding of the indicator G. The couple is connected, with G in circuit, across the manganin coil S, the resistance of this material being unaffected by temperature. By adjusting R until no deflection is shown on G, the drop of volts across S is made equal to the E.M.F. of the couple. To measure this drop, a key is pressed, altering the circuit as shown in B, the indicator being now in series with S and the couple detached. The value of the current passing through S is unchanged, as the indicator coil has the same resistance as the copper coil C, which it now replaces. The deflection on G indicates the value of this current, and, as the drop of volts across S is proportional to the current, G may be marked off to read E.M.F. and the corresponding temperature of the junction. The advantages claimed are that the indicator may be used with any type of junction, and is unaffected by temperature changes in the circuit. A similar instrument is made by the Brown Company of Philadelphia. Up to the present potentiometer indicators have not been adopted to any extent in Britain, and the adjustments necessary to obtain a reading must be accounted a distinct drawback from a workshop standpoint.

=Recorders for Thermo-electric Pyrometers.=—It is frequently of importance to know not only the existing temperature of a furnace, but also the fluctuations to which it is subject. Continuous observation of a pyrometer would involve too much labour, and it is therefore evident that an automatic recorder would possess many advantages in such cases. A continuous record shows whether the attendant has maintained the temperature between the prescribed limits, and furnishes a permanent history of a given operation, which often serves as a guide to future procedure.

The first successful recorder, suggested by Sir W. Roberts-Austen and designed by Gen. Holden, F.R.S., was used in conjunction with a mirror galvanometer. In its original form, the spot of light from the mirror was made to fall on a sensitized plate, to which a gradual vertical motion was conveyed by connecting the dark slide to a water-float by means of a chain and pulley. The float was placed in a tank of water, which was gradually emptied through a tap, causing the float to sink and the plate to rise. If the deflection of the spot of light remained steady, a vertical straight line was traced on the plate, fluctuations producing a sinuous line. Trials at known temperatures enabled a standard plate to be obtained, divided into degrees, which could be superposed on a trial plate, and the temperatures thus determined. Much valuable work was accomplished with this recorder by Roberts-Austen for the Alloys Research Committee of the Institution of Mechanical Engineers.

In its modern form (fig. 20) the photographic plate is replaced by a sheet of sensitized paper wound round a drum which rotates at a known rate—say, once in 12 hours—by means of internal clockwork, shown to the left of the figure. The galvanometer is placed at the opposite end, and the mirror is illuminated by means of an electric lamp placed externally, the rays from which are reflected from a prism in the interior on to the mirror. The ray of light leaving the mirror is broken into two portions, one of which passes through a narrow slit on to the sensitized paper, whilst the other portion is reflected on to a ground-glass scale on the lid, divided so as to read temperatures. In this manner the arrangement serves not only as a recorder, but also indicates the existing temperature without necessitating the examination of the sensitized paper. The whole arrangement is made impervious to light, so that it may be used in daylight. A dark room is necessary for fixing the records. When desired, records of two or more pyrometers may be taken on the same sheet, a clockwork device being used to switch each instrument in turn on to the galvanometer for a given period, an external dial indicating which pyrometer is for the time being in circuit.

Whilst it is a drawback to the use of this recorder that the record is not visible, the use of a mirror galvanometer confers a high degree of sensitiveness to the instrument, not possessed by the recorders to be described subsequently.

=The Thread Recorder.=—In this instrument an intermittent record is secured in ink, possessing the advantages of visibility during the period over which readings are taken, and of permanence without subsequent treatment of the chart. The principle is shown in fig. 21, where A is a boom terminating in a V-shaped piece of ivory, and attached to the galvanometer suspension B. By means of a cam E, rotated by clockwork, a bar D is made to descend at stated intervals, pressing the end of A on to an inked thread G, and causing the thread to touch a paper wound round the drum C. This drum rotates on its axis once in 25 hours by the action of internal clockwork. The continued rotation of the cam E alternately raises and depresses the boom A, leaving it free for a sufficient time to enable it to attain the position it would occupy if the mechanism were absent. The thread G is passed over pulleys, and is wound round through an ink-well, so that the portion opposite A is always moist. With the bar D descending every two minutes, the successive dots form a nearly continuous line. The paper on C is divided horizontally into temperatures, and vertically into time units, so that the temperature existing at any given time may readily be ascertained. The front of the bar D, or a separate strip parallel to it, is divided so as to enable temperatures to be read without reference to the chart. The actual instrument is shown in fig. 22. When several simultaneous records are required, the drum C is extended, and other galvanometers introduced, to which the separate pyrometers are connected. Several records can be taken on one chart by introducing a clockwork mechanism to couple each pyrometer in turn to the one galvanometer.

=The Siemens Recorder.=—In this instrument (fig. 23) the boom from the galvanometer terminates in a knife-edge, and moves over a thin horizontal rail, the top of which is rounded. Between the rail and the boom are placed an inking ribbon and a paper chart, which is moved forward by clockwork. A chopper-bar, also actuated by clockwork, descends at about half-minute intervals, and depresses the end of the galvanometer boom, thus producing a small dot on the chart. The paper is 12 cms. wide and 40 yards long; it is divided into time and temperature units, and moves forward at the rate of 2 cms. per hour. Levelling screws are fixed to the base of the recorder.

=Foster’s Recorder.=—Foster’s recorder (fig. 24) is designed for use with base-metal couples of the nickel-chromium type, known as Hoskin’s alloys, which yield an E.M.F. about five times as large as a platinum-rhodioplatinum couple. The force available in this case enables the coil of the galvanometer to be pivoted in a horizontal position, the pointer being vertical, and yet to be sufficiently sensitive. The chart is mounted on a vertical plate which rotates on its axis, the time ordinates taking the form of concentric circles, which are cut at an angle by the temperature ordinates. At the terminus of the pointer is placed a small capillary tube, fitted with an inked wick, which, when pressed upon the chart, makes a mark. The presser-bar is curved to the same radius as the pointer, and carries a pad wetted with ink, so that at each depression the supply of ink to the wick is replenished by an amount equal to that imparted to the chart. This recorder is sometimes fitted with special contacts, so that when the correct temperature exists an electric lamp with a white bulb remains lighted; whereas when too low or too high a green or red lamp is lit up, and an alarm thus given. Such an addition involves the use of a relay circuit, but is advisable in cases where expensive articles might suffer if overheated. It can be modified to permit of several simultaneous records being taken, and possesses the advantage that the whole chart is visible at any time. On the other hand, the circular coordinates may be accounted a drawback by some, as not being quite so familiar to read as charts in which the lines are straight. Robust construction is a feature of this recorder.

=Paul’s Recorder.=—In the recorders previously described, the motive power is furnished by clockwork. R. W. Paul has introduced an instrument in which all the moving parts are actuated by a motor driven with power from the mains. This recorder is shown in fig. 25. The motor is furnished with a special type of governor to ensure constant speed, and is connected by suitable gearing to the mechanisms moving the chart, presser-bar, and inking ribbon, provision being made to vary the speeds of these movements by changing the gear. The galvanometer is of uni-pivot pattern, and the pointer is pressed at intervals on to a typewriter ribbon which lies above the chart. Immediately beneath the ribbon is placed a thin metal rod over which the paper passes, and the result of the contact is to produce a small dot. As in the thread recorder, the chart is divided into rectilinear coordinates, the ribbon in this case serving the same purpose as the thread in the former instrument. The lower part of the recorder is prolonged so as to display a considerable length of the chart, which is in the form of a roll, and is drawn forward by the mechanism. When two records are taken simultaneously the ribbon consists of two strips, one moistened with black ink, and the other with red; and it is arranged that each strip in turn is over the thin rod on to which the pointer is pressed, so that the records appear in separate colours. This recorder can also be arranged for multiple records, or fitted with a scale-control. With a view to workshop use, all the covers are fitted with faced metal joints, which are much better for keeping out dust than wooden ones. A further useful feature is that the various units in the recorder—galvanometer, motor, feed and record mechanism, and reducing gear—are all separate and interchangeable. By introducing a suitably divided chart this recorder will also serve for a radiation pyrometer, or, as will be shown later, for a resistance pyrometer.

=The Leeds-Northrup Recorder.=—The Leeds and Northrup Company, of Philadelphia, manufacture a recorder which is largely used in the United States. As in Paul’s recorder, all the mechanism is motor driven; but the other arrangements are entirely distinct. Instead of measuring the deflection of the pointer, a zero deflection method is used. The pyrometer forms part of a potentiometer circuit, and the function of the mechanism is to oppose an E.M.F. equal to that of the pyrometer, from which the temperature is known. This has the advantage that the measurement is independent of the resistance of the leads, and is capable of great accuracy. The manner in which the adjustment of the opposing E.M.F. is controlled may be understood from fig. 26, in which the galvanometer coil is shown at the top of the figure. The shaft from the motor carries four cams, B, C, D, D, and at each revolution the cam B raises the bar (5) so as to press it against an arm attached to the galvanometer coil. At the same moment the cam C pushes against the bar (3), and thereby releases a clutch (2) from the disc beneath. As shown, the boom from the coil is to the right of the central position, and is gripped between a bar (5) and the lever (4) when the former rises, causing an angular movement of the clutch-arm (2). As the rotation continues the cam C leaves the bar (3), which then springs back and engages the clutch on the disc. The cam D then descends and presses on the projection of the clutch-arm to the left, causing the disc to rotate. The movement of the disc is conveyed to an arm which moves over the slide wire of the potentiometer; and this movement continues until the galvanometer boom is in the central or zero position, when neither of the levers 4, 4 is gripped, and consequently the disc is not fed in either direction. If the boom swing to the left, the movement of the disc will evidently be in the converse direction to that described.

In this recorder considerable power is available to drive printing or other mechanisms. The arm moving over the potentiometer wire carries a pen which marks the moving chart, or, when several records are taken simultaneously, a stamping machine is used which impresses the number of the pyrometer on the chart. The same galvanometer mechanism serves also for use with resistance pyrometers, as will be explained later.

=Control of Furnace Temperatures.=—Many attempts have been made to secure the automatic regulation of furnace temperatures by means of mechanisms controlled by an indicator or recorder. In the arrangement employed by the Brown Company of Philadelphia, movable stops are provided, which may be brought to any part of the scale, the mark between the stops representing the temperature it is desired to maintain. The indicator (or recorder) is provided with a presser-bar which descends periodically; and if the temperature be too low the depressed pointer completes a circuit through the inner stop, whilst if too high the circuit is through the outer stop. Both circuits contain a relay which brings a mechanism into operation, the result being to increase the supply of electricity or gas if the temperature be too low, or to diminish the supply when too high. When correct, the depression of the pointer fails to complete either circuit, and in this manner control between small limits may be ensured. In the case of large furnaces the relay circuits are employed to light lamps of different colours, the adjustment then being made by the man in charge of the furnace. Arrangements of this kind effect a considerable saving in fuel by preventing unnecessary heating, and are particularly valuable in cases where overheating would be deleterious to the articles in the furnace. The future will probably witness considerable developments along these lines.

=Contact-Pen Recorders.=—The force with which the pointer of an indicator is urged over the scale is relatively small, particularly in the case of pyrometers in which the platinum series of metals are used, as these furnish only a low E.M.F. If, therefore, the pointer terminate in a pen which is in continuous contact with the record-paper, the friction thus occasioned interferes considerably with the free movement of the pointer. When cheap-metal pyrometers are used, which yield a much higher E.M.F., the use of the pointer as contact-pen becomes more feasible, and if uniform friction at all parts of the paper can be ensured, records may be taken in this manner; and a recorder so constructed is simpler and cheaper than those of the intermittent type. Contact-pen recorders are used in America to some extent, being made by Bristol, Brown, and others; but so far British makers have not developed the manufacture of these instruments. At present, contact-pen recorders must be considered less accurate and reliable than those in which the contact is intermittent.

=Installations of Thermo-electric Pyrometers.=—When a number of furnaces in the same establishment are to be controlled, considerable economy may be effected by making one indicator serve for all the couples, which in this case must necessarily be made up of wires identical in thermo-electric value. Such an arrangement is shown in fig. 27, in which H^1 and H^2 represent two couples, one wire from each being connected to one of the terminals of the galvanometer G. The other terminal of the galvanometer is connected to the arm D of a switch, and the remaining thermocouple leads are connected to the points 1 and 2 respectively on the circumference. As shown, H^1 is connected to the galvanometer, and by turning the arm D to the point 2 the other couple would then be connected. Any number of junctions may thus be arranged with a single indicator. When this arrangement is adopted in a workshop, it is advisable to construct a small wooden building at a spot convenient for most of the furnaces, in which the indicator and switchboard are kept, and which could also contain a recorder if necessary; a spot as free as possible from vibration being preferable. Separate indicators are only necessary when a furnace is used for special work.

In some instances a second indicator is kept in the shop office, to which all the pyrometers are wired, and which serves as a standard. The scale of the office indicator is checked daily at one point; and by connecting a given couple first with the shop indicator, and immediately afterwards with the office standard, any errors can be detected. It is also possible to ascertain the temperature of any given furnace in the office at any time, and so to control the whole. In fixing up such an arrangement it is necessary that each couple and its leads, up to the indicator, should possess the same resistance, or should not differ by an amount sufficient to affect the readings. The general experience of a properly managed installation is that the cost is saved in a few months in fuel alone; and, in addition, the work is carried out to much better advantage owing to complete control from the office.

=Management of Thermo-electric Pyrometers.=—Generally speaking, thermo-electric pyrometers give little trouble in practice, but the management should always be placed in skilled hands. It is advisable to test each instrument periodically at a fixed point near the working temperature, by the method explained on page 57; and if two or three pounds of material be used, the protecting shield need not be removed. A useful material for checking pyrometers near the critical range of steel is an alloy of 60 per cent. of copper and 40 per cent. of tin, which gives a well-defined freezing point at 738° C., and which may be used indefinitely in a reducing atmosphere Any serious error is easily detected by observing that the indications differ widely from those generally obtained under the same working conditions. If an error of 20° C. or more is noted, it is advisable to form a new junction, as the discrepancy will probably become greater, being due to a change at the hot junction. A small error, of the nature of 5 or 10° C., may be due to “creep” in the indicator, which may be adjusted accordingly, or a numerical correction may be made when taking a reading. An iron protecting sheath may be saved from rapid oxidation by black-leading once per week, which greatly prolongs its useful life, but should be replaced immediately it becomes dangerously thin in any part. Coating with aluminium powder also greatly prolongs the life of an iron sheath. When used in lead baths, the immersed part, if of iron or steel, should be bored from the solid, and left thick at the portion opposite the surface of the lead, where most corrosion occurs. A graphite tube, or one made of a composition containing graphite, is often useful in cases where iron is readily corroded, and can be used to much higher temperatures.

When a number of instruments are in use, it is advisable to keep a standard pyrometer for checking purposes, preferably one which has been certified by the National Physical Laboratory. In conducting a test, the couples, with protecting-tubes removed, may be placed in the tube of an electric furnace of the type shown in fig. 29, in close proximity with the standard junction. On raising the temperature gradually, the readings of each working instrument may be compared with the standard, and the necessary corrections discovered. Care must be taken to prevent contact with the furnace tube, and this may be accomplished by passing the wires through an asbestos stopper fitted into the end of the tube.

When recorders are used the attendant should make himself thoroughly conversant with the details of the mechanism, so as to be able to remedy any minor ailments, which are, as a rule, easily cured. On no account should an unskilled workman be trusted with recorders; it is better and safer to keep these in the office, where they will not be likely to be damaged or tampered with. All records should be kept for future reference, properly dated, and labelled according to the operations represented.

=Laboratory Uses of Thermo-electric Pyrometers.=—Numerous operations carried out in muffle furnaces at prescribed temperatures require no special precautions beyond those previously given. In determining the melting points of metals or alloys, however, a porcelain or silica sheath is inadvisable, as they are easily corroded. An iron sheath is proof against some metals, but not against others, and it is always safer to fix a thin fireclay sleeve, closed at the end, over the part immersed. A sheath of graphite or graphite composition may be used for temperatures above 1100° C.; and occasionally a sheath bored from a thick arc-lamp carbon, coupled to an iron tube beyond the heated part, will be found useful at high temperatures. Alundum is useful up to 1600° C, and for temperatures of this order the higher refractories such as silfrax and zirkite may also be used to advantage.

The determination of the “critical” points of steel call for special mention. In cooling down a mass of steel the fall of temperature is arrested at one or more points, observations of which are frequently of service in deciding the subsequent treatment of the steel. A method commonly employed is known as the “differential method,” and is indicated in fig. 28. The sample of steel, A, is placed side by side with a piece of nickel, B, of equal dimensions, in the tube of an electric furnace. A naked junction, C, is placed in a hole drilled in A, and is connected to the galvanometer G, which is calibrated to read temperatures. A two-junction circuit, formed of a junction D placed in the hole in A, and another junction E located in the hole in B, are connected to a delicate galvanometer H. The furnace is heated until the galvanometer G indicates 900° C., when the arrangement is allowed to cool. As A and B, under normal circumstances, cool at an equal rate, the junctions D and E will be at the same temperature, and no deflection will be observed on H. When, owing to recalescence, the cooling of A is arrested, B, not being thus affected, will continue to cool, thus producing a difference between the temperatures of D and E, and consequently a deflection on H. The temperature of A at the time this occurs is read off on G.

The furnace illustrated in fig. 29 is suited to this determination. It consists of a silica tube 1 foot long, wound with a special resistance wire and efficiently lagged, and may be heated in safety to 1000° C. for long periods, and to 1200° for a short time. It may be placed across the electric mains directly, and reaches 900° C. in less than half an hour. It consumes 600 to 700 watts at the highest temperatures, and the cost of re-winding is small. This furnace is useful as a general laboratory appliance, and may be kept at a given steady temperature by the use of an external resistance.

The wires in this experiment should be platinum and iridioplatinum or rhodioplatinum, or a good pair of base metals, and the junctions in A should be separated from each other and from the specimen by asbestos; the same precaution being taken to prevent the junction E from touching B. A thin layer of mica should be used beneath A and B, to avoid contact with the furnace tube, which, when hot, allows of leakage of current from the heating coil. Both A and B may be 1½ in. long, ¾ in. diameter, with a hole ¼ in. diameter drilled to a depth of ¾ inch.

An alternative method is to insert a junction in a hole in the specimen, and to take direct readings as the temperature slowly rises or falls, when an arrest in the movement of the pointer of the indicator shows that the change-point has been reached. Special sets are made for this purpose.

=Measurement of Lower Temperatures by the Thermo-electric Method.=—Many cases arise in practice in which a thermal junction and a sensitive galvanometer are preferable to a mercury thermometer; and below -39° C., at which temperature mercury freezes, a thermal junction is frequently better to employ than an alcohol or pentane thermometer. A number of practical examples of the use of thermal junctions for ordinary and low temperatures will now be considered.

=Measurement of Surface Temperatures.=—A mercury thermometer, when laid on a hot surface, only touches along a line, and does not show the true surface temperature. The construction of a thermal junction suitable for this purpose is described on page 41, and for steam-pipe surfaces, hot plates, and the exterior of furnaces, a specially calibrated millivoltmeter, giving a full-scale deflection with 20 millivolts, may be used. In making the temperature scale, boiling water (100°C.), boiling aniline (184° C.), and melting tin (232° C.) are convenient standards. If the surface temperature be less than 100° C. a mirror galvanometer should be used, and the junction standardized in paraffin wax (freezing point usually about 50° C., but should previously be determined with an accurate thermometer), absolute alcohol at boiling point (79° C.) and boiling water. The author has found that this method yields excellent results in the case of steam-pipes, the exterior of rotary cement kilns, and hot surfaces generally.

=Measurement of Low Temperatures.=—Junctions of iron and constantan, Hoskin’s alloys, copper and German silver, or copper and constantan, are suited to these measurements. In a laboratory the cold junction may be kept in ice in a Dewar vessel, the mechanically protected form known as the “Thermos” flask being very useful for this purpose. With a good mirror galvanometer precise readings may be secured, 1/10 of a degree C. being easily detected. Calibration between -40° and +40°C. may be effected by comparison with a standard mercury thermometer, a water-bath being used above 0°, and alcohol surrounded by a freezing mixture of ice and calcium chloride crystals below zero. For very low temperatures (-200°C. or less) the junction maybe calibrated in solid carbon dioxide (-78°C.) and liquid air (-184°C.). Dewar has found that copper and German silver form a reliable junction for very low temperatures, and the author has successfully used a couple of Hoskin’s alloys for special work down to -200°C., a pivoted indicator being employed. No couples tested show a linear relation between E.M.F. and temperature at these low ranges.

Owing to the magnitude of the error caused by changes in the cold junction, the thermo-electric method is not suited to the measurement of atmospheric temperatures, or for explosive magazines or cold stores. In such cases instruments of the resistance type, to be described later, are used.

=Temperature of Steam, Exhaust Gases, etc.=—For measuring the temperature of ordinary or superheated steam, the exhaust gases from internal combustion engines, etc., iron-constantan junctions, with suitable indicators, are satisfactory. When placed in a pipe the junction should be as nearly as possible in the centre, so as to avoid the cooling effect of the walls. Several junctions, situated in different parts of the pipe, may be used with a single indicator and suitable switchboard. The above remarks also apply to the hot-blast for blast furnaces, and similar instances where the temperature does not exceed 900° C.

=Measurement of Differences of Temperature.=—Cases frequently arise in practice in which the difference in temperature between two points is required, and if this difference be subject to rapid changes, a mercury thermometer, from its large mass, would not respond with sufficient rapidity to indicate these changes. In such cases a circuit is made after the manner of fig. 2, one junction being located at each point; thin wires of iron and constantan being used. For small differences—1° C. or less—a mirror galvanometer should be used. Calibration may be performed by placing one junction in hot water and the other in cold, the water temperatures being read with an accurate thermometer.

=Advantages of the Thermo-electric Method of Measuring Temperatures.=—Compared with other methods, the thermo-electric possesses the following points of superiority:—(1) Simplicity, no special experiment being necessary to obtain a reading; (2) cheapness of outfit; (3) adaptability to a variety of purposes; (4) ease of repair in case of damage; (5) robustness, not being liable to get out of order under workshop conditions; and (6) suitability to the purpose of a centrally controlled installation. The drawbacks are:—(1) Liability to error owing to fluctuations in the cold junction (which may be avoided with care); and (2) lack of sensitiveness at very high temperatures compared with the resistance method—a point seldom of great practical importance, as the limit of accuracy is usually within the amount by which an ordinary furnace fluctuates in temperature under working conditions.