The Tube Muffle Furnace is practically the common laboratory analytical gas-fired combustion-furnace, with its glass tube replaced by the iron and copper tubes, G and M, and surrounded with brick-work to lessen radiation and to give better control over the temperature.
The burners, J, are those of the common combustion-furnace.
The bar, K, to be heated is laid on another, L, cut from the same rod, and inside the copper tube, G. L rests on little strips of iron, to raise it above the bottom of G, so that both bars may be heated alike by radiation rather than by conduction.
The copper tube, G, rests within the wrought-iron tube, M, and is held concentrically with M by packing a little ring of asbestos between them at either end of G.
The thermo-electric junction, N, formed by the junction of a wire of pure platinum with one of platinum alloyed with about 10 per cent, of rhodium, lies between the bars K and L, in a little notch filed in the latter. The leading-in wires pass to the galvanometer, which is at a convenient distance from the furnace, through the double-bored clay tube, O.
My special object in designing this furnace was, first, to have the thermo-junction at a temperature very near that of the bar, K, under observation ; to have this bar at a uniform temperature throughout; and to have this temperature completely under control.
(56) Precautions.—Unless special precautions are taken, the temperature of the thermo-junction may differ very considerably from that of the bar under treatment.
It does not suffice simply to lay the junction against the bar, unless the latter be at the same temperature as the surrounding atmosphere and walls. If it be not, then the junction will be at a temperature intermediate between that of the bar and that of the surrounding objects. It is difficult to have the bar and the surrounding objects at the same temperature, unless a muffle be used; and even with a muffle it is possible only when the temperature is stationary or at most moving very slowly. But by imbedding the junction between two like bars, which are under like conditions as regards heat, its temperature is but slightly affected by that of other objects. Nevertheless, even with the arrangement shown here, the influence of the walls of the inner muffle is not wholly effaced during rapid changes of temperature.
The galvanometer shows at any moment the temperature of the
thermo-junction itself, unless the leading-in wires be short-circuited ; if they are, the galvanometer shows their temperature at the point of short-circuiting. Now the leading-in wires, small though they be, are good conductors of heat, and with the very great differences of temperature which often exist between neighboring parts of apparatus for high-temperature experiments, they may conduct enough heat to the thermo-junction or away from it to affect its temperature seriously. For instance, it was actually found that, if the double-bored clay tube, O, was broken, so that the leading-in wires were exposed to the flame, they conducted so much heat to the thermo-junction that the readings of the galvanometer were much too high. On cutting off the gas the temperature of the couple would drop 4° or even 6° instantaneously, clearly because the wires, superheated by the naked flame, had been conducting heat to the thermo-junction, and raising its temperature above that of the bars which surrounded it, and so above the temperature which was under examination. After such a drop of 4° or so, the galvanometer took up the regular gradual movement due to the cooling of the bars. But when the wires were protected by the clay tube, no drop of this kind could be induced.
Apart from the influence of the exterior part of the leading-in wires, as the temperature of the apparatus rises or falls, the thermo-junction tends to outrun the bar of steel under treatment and observation, to be slightly hotter than the steel while the temperature is rising, and slightly cooler when it is falling. This, of course, because the steel must always lag behind the walls of the muffle, and lag behind them more than the leading-in wires do; as these outrun the steel, so, owing to the heat which they conduct to the thermo-junction, will it outrun the steel. Moreover, the outside of the steel bars will always slightly outrun their interior; and as the wires are very apt to short-circuit through the skin of the bars where they pass from between them, our readings on this account, too, tend to outrun the average temperature of the steel.
But even the thin walls of the clay tube, O, which surround the wires probably conduct heat rather more slowly than the double metallic walls of the muffle. This difference tends to make the wires lag where they pass through the tube, O, and thus tends to counteract their tendency to outrun the steel under treatment. As one error tends to counteract the other, the net error is probably very slight. I doubt if it can amount to more than a degree, at least when the temperature of the whole is changing at a moderate rate only. It was observed repeatedly that, when the upper bar was withdrawn, leaving the thermo-junction exposed to direct radiation from the walls of the muffle, the reading of the galvanometer would vary very slightly; indeed, not at all, if the temperature had been changing only very slowly. If the more powerful radiation from the walls of the muffle thus substituted influenced the temperature of the thermo-junction so slightly, the much feebler influence of the leading-in wires, when the junction is protected by the bar under treatment, can be neglected in such a furnace as this, during slow changes of temperature.
Of course, when any such apparatus cools rapidly, the leading-in wires must outrun the bars operated on, and so must the thermo-junction, and through it the readings of the galvanometer. The faster the apparatus cools, the more will the readings of the galvanometer outrun that of the bars under treatment, and the more will their true temperature at any moment be above their temperature as reported by the galvanometer. The same thing applies, mutatis mutandis, to variations in the speed of heating.
I have dwelt on this at length, because it casts some doubt on a series of results obtained by M. Osmond. He found that increasing the rapidity of cooling lowered V, the recalescence point.
This would be a striking and important fact. But his apparatus was so arranged that the reported lowering of V may be due in part, if not wholly, to the source of error which I have just described. The apparent temperature of V would be that of the thermo-junction at the time when the bar itself was really at V. This apparent temperature would be below V, the more so the faster the cooling. Hastening the cooling would certainly cause an apparent depression of V, possibly even sufficient to explain the depression recorded by Osmond.
A fact which lends plausibility to this suspicion is the following: When he used small square bars he found that V was lowered by only some 29° by accelerating the cooling; but when he used still smaller round bars, V was lowered by some 90°. Clearly, the influence of the leading-in wires would be likely to be much greater in case of round than in that of square bars, for the latter hug the thermo-junction much better than the former do.
(57) Evidences of Accuracy.—An interesting phenomenon testifies that the temperature during very slow heatings and coolings was extremely uniform within the copper tube. When, during such a heating or cooling, the inner tube was opened, its contents were wholly invisible for a few seconds, and then the nearer end of the two bars gradually appeared. This was true even at low redness, as well as at all higher temperatures, and evidently was in no way due to dazzling. As the contents of the tube were all at very closely the same temperature, so they all glowed with precisely the same intensity and tint, and the eye, which sees things solely by difference of tint or shade or brightness, simply saw uniform red or yellow light, and nothing more. After a few seconds the cool outer air, entering the tube, cooled the ends of the bars so far that their tint was slightly darker than that of the rest of their length and of the tube itself, and these ends became visible, while the rest of the bars and the interior of the muffle remained invisible.
Some of the tensile-tests of bars which had been treated agree very closely, and testify to the uniformity of heating. Such are the tests of bars 42 and 52 of Table 18; of bars 38 and 39, and 36 and 37 of Table 23.
Where bending-tests were made on bars which had been treated alike, or which had not been treated at all, the results agreed. See, for instance, the tests of bars 96 and 28 of table 26, both of which bent 37° ; and bars 50 and 51 of the same table, both of which bent 45°.
The sharpness with which the retardations, and especially the re-calescences, often begin, shows that the temperature, not only of the two bars as a whole, but also of the different parts of each bar, is approximately uniform; for any considerable irregularity in the temperature of the metal would make the recalescence begin gradually instead of abruptly. For instance, suppose that one end of the two bars was considerably hotter than the other, and that, at any given instant, there was a regular increase of temperature, as we passed from the cool towards the hot end. If, in this condition, the two bars are cooling towards V, different parts of them will successively reach V, and will successively recalesce ; that is to say, the recalescence, instead of beginning simultaneously throughout the bars, will travel from the cool end towards the hot end, as one point after another cools as far as V. Under these conditions, the descent of the temperature of the thermo-electric junction, instead of continuing at full speed till V is quite reached and then abruptly changing to a rise of temperature, would necessarily slacken much and progressively before reaching V, because the recalescing of the adjoining parts, which reach V earlier, would gradually retard the cooling of this junction more and more as the recalescence, travelling along the bars, drew nearer and nearer to the junction. Yet actually the recalescence begins very suddenly—often with hardly any warning. I give here the intervals (seconds per 1.4° C.) in the cooling near and at V, in a few cases :
2.5; 2.5; 2, 2; 4; 3; 5; 3; 2; 2; 2; 95; 5; 5; 8.
2; 2; 3; 3; 4; 4; 4; 2; 3; 4.5; 4.5; 107; 4; 3; 4; 4.
2; 2; 2; 2; 4; 118; 5; 4; 4; 4.
But the converse does not follow. A gradual instead of an abrupt beginning of the recalescence would not argue that the temperature of the bars was not uniform, because the gradualness might be due to the very nature of the recalescence itself.
By like reasoning, an abrupt beginning of any retardation argues that the temperature is uniform throughout the bar treated.
The constancy of the apparent position of W and Wa, noted in (16), is further evidence that the conditions were uniform, and evidence, indeed, of the accuracy attainable under these conditions. Even if the true position of W and Wa be absolutely fixed, we should expect that errors of observation would cause the apparent position to vary. And it was to be feared that these errors would be considerable; for variations in the conditions of heating and cooling were introduced intentionally; and the considerably varying temperature of the room should affect the readings of the galvanometer, and together with varying draughts, and varying conduction of heat through the furnace walls, might alter the difference between the temperatures of the metal under observation and of the thermoelectric junction by which its temperature was observed. Indeed, the fact that in the eleven normal cases in which it was observed in Table 7, Wa varied by only 1.4° C., while in eight of those cases W varied by only 1.4° C., and in ten of them by only 2.8° C., argues that the net effect of these sources of error need not be serious.
The air-space between the iron and copper pipes tends to equalize the temperature of the latter; and copper was used for the inner muffle because it conducts heat so rapidly that the temperature of its different parts was very uniform.
The double sheet-iron diaphragms shown were designed to cut off radiation to and from the clay plugs at the outer ends of the iron tube.
It is not difficult to control matters so that the temperature of a bar, at the moment of its withdrawal from the furnace, shall not vary more than 1.5° C. from the temperature aimed at; this demands only that the temperature shall vary slowly, and to ensure this requires no great care in design. But it is a far different and very difficult thing to hold the temperature stationary at a predetermined point. Yet, so good was the control over the furnace that its temperature has appeared absolutely constant for six minutes, during which it certainly did not vary more than 1.4° C. In another case the apparent variation was only 1.4° C. during a period of seven minutes.