« PreviousContinue »
out; the same at the centre of section as at the surface. The difference is measured probably by hundreds of a degree only. Professor Heinrich Streintz has calculated the temperature difference in certain cases; and it appears that for the present purposes it is certainly unnecessary to take it into consideration. It is allowable further to assume that a fine dust resting upon the surface of a flat conductor, such as is used in the meldometer, is very closely at the temperature of the conductor, especially when near the central line ; where, surrounded by neighbouring particles, it is sheltered from draught. It is, indeed, true that for the accuracy of the measurements made on the meldometer it is not necessary to assume that the agreement in temperature should be exact, as will be seen, but only that the difference should be about the same from one substance to another. The point here is that the difference being in any case very small there cannot be much diversity in the temperature difference between substance and platinum from one case to another. Assuming then that the average temperature throughout the cross section of the wire is that of its surface, and that the finer dust resting upon this surface is nearly at its temperature, it only remains to measure the temperature of the platinum ribbon in order to determine the temperature of the substance.
First attempts in this direction were directed at determining the temperature in terms of the electrical resistance, a well-known formula of Siemens's being used to express the relation between the two. Subsequently Mr. Callendar3 compared the readings of the air-thermometer and the platinum resistance thermometer up to 600° C. However, notwithstanding the help thus afforded, the cumbersome apparatus and tedious measurements required in the case of the meldometer in which the resistance has to be determined while a current is traversing the circuit, rendered the method inapplicable. It was accordingly abandoned after much labour had been spent upon it ; and at a later date the present form of meldometer devised which depends on the thermal expansion of the ribbon for the estimation of temperature. The general mode of procedure in the present method is as follows:Using a much longer ribbon than that previously described, we determine its increase of length (by a contrivance to be described)
1 Poggendorff's Annalen der Physik und Chemie., vol. clx., p. 409.
? If, however, the ribbon of platinum be raised nearly to its melting point, it will be found that signs of fusion appear first along the central line.
3 Phil. Trans. Roy. Soc., vol. clxxviii., p. 161.
when a salt of known high melting point is seen to melt upon its surface, and again when two or more other substances, differing from each other in melting point as much as possible, are melting. This is analogous to determining the fixed points on a thermometer, and assumes, as in the latter case, nothing as to the absolute co-efficient of expansion of the working substance. Thus, if we read the movement of the mercury in a thermometer when we raise it from the known temperature of melting ice to the known temperature of boiling water we may evidently, on the assumption of the uniformity of the expansion of the mercury between the two points and of the uniformity of the bore, estimate the temperature of any intermediate reading, using a system of numbering decided upon beforehand. Similarly the assumption is made in the case of the meldometer that there is uniformity of expansion over the limits of known temperature, and for some distance beyond them. In fact, in this method, we take advantage of the experimental work of past observers. Much of this, more especially that of Carnelley and Violle, is very careful. It is sufficient to determine five or six points at various intervals along the scale, and plot the extensions corresponding as ordinates along any uniformly divided line serving as a scale of temperatures. The points so determined are joined by a line which will be found very slightly curved over its lower length, in which we use the experiments of Carnelley. In the upper range-from 900° to 1500° C.-we have very careful data of Violle's, which plot as a line of slowly increasing curvature, convex to the axis of temperature, and meeting tangentially the range covered by Carnelley's experiments. To determine now the melting point of an unknown substance, we obtain the extension corresponding to its point of melting and scale from the chart the temperature proper to this extension. This is all that is required to determine the melting points of minerals to the degree of accuracy attained in the elaborate researches of Carnelley and Violle. Although the extensions, as determined directly, may be thus used to determine unknown temperatures, it is preferable to plot in each case the ratio of the extension to the original length of the ribbon, or the quantity 12-11/11, this value being independent of the length of ribbon used in making the experiments, a length which it is inconvenient to regulate to uniformity on all occasions. The value of this fraction is certainly applicable to all observations made with platinum of the one quality, or to the contents of the one reel if certain precautions be taken which will be described later. In this way a curve once plotted may be kept in use for many observations. The chart (Plate VI.) accompanying this paper is plotted in this manner.
It is necessary before proceeding further to describe the form of meldometer used in determining melting points in the manner just sketched out. The accompanying figure (fig. 2) from a photograph
shows the apparatus (made by Messrs. Yeates and Son) at present in use. It consists of a rectangular piece of slate, cut into the form shown, on which two forceps are affixed, one rigidly (the further one on the figure), the other so that it is free to rotate round a vertical axis at one extremity, beyond which it is prolonged by a fine flat spring of steel. When these forceps are parallel to one another the distance separating their jaws is closely ten centimetres. This is the length of platinum ribbon whose expansion is measured when making experiments. A small helical spring attached to the vertical axle of the moveable forceps, and to the fixed piece which receives the upper pivot of this axle (this piece conceals it in the figure), serves to confer a slight tension on the platinum ribbon, in that it is set to turn the forceps clockwise on its axis. Thus, when the ribbon expands the forceps rotates a little, clockwise. At the further extremity of the spring which prolongs this forceps, a small polished plate of gold is attached, and to meet the surface of this a gold wire, carried at the extremity of the micrometer screw seen on the figure, may be brought on advancing the screw. The point of contact of pin and plate is as nearly as possible at the same distance to the one side of the vertical axis of the forceps as the centre line of the ribbon is to the other, when the latter is in its proper position in the forceps. Thus, the expansion of the ribbon may be measured by the movement of the micrometer screw, if at each change of temperature we get the gold pin to touch the gold plate exactly. To effect this without troublesome observations matters are arranged so that the meeting of pin and plate closes a circuit in which a galvanometer and a Leclenchè cell are included. The bed of slate insulates the screw and forceps, and connexions are made as on the diagram (fig. 3). This is an extremely
sensitive method of reading a small movement. The screw pitch is about the 4th of a mm.; the micrometer head carries 100 parts. Half these last divisions may easily be read, that is to say, the both of a mm., or less than the godooth of an inch, a quantity approaching the limits of the highest powers of the microscope, and indeed little more than two wave-lengths of sodium light. This movement is equivalent to a change of temperature of about 1.5 degrees centigrade. The micrometer screw must have a uniform thread and be reliable in its readings, or error will of course be introduced, just as would be the case with a badly calibrated thermometer. To bring the current into the forceps two large binding screws (seen in fig. 2) are provided. These are
attached to copper straps passing beneath the slate bed, and communicating with the forceps. To insure good contact with the movable forceps an annular space around the axle is provided in the slate, into which a little mercury can be poured. A perforated wooden lid, loosely encircling the axle, keeps this mercury from accidentally being spilled out. The bottom of the mercury well is of brass, which receives the pivot of the axle on its upper surface, and at its under surface is screwed to the copper conductor. The forceps close by their elasticity, and are opened for the insertion of the ribbon by the small screws seen on the figure. A slip of polished brass, passing beneath the ribbon, serves to intercept heat from reaching the edge of the slate, and to receive chance fragments of substances falling off the platinum. The slate is cut back beneath this, for it is an object in the construction of this apparatus to preserve the slate from change of temperature. There seems but little heat communicated to the slate. I have not detected any sensible heating, even after prolonged experiments at high temperature. The coefficient of expansion of slate is so small that but little error can, I think, creep in this way.
The platinum ribbon used in this apparatus is similar to that used in the first described form of the meldometer. The resistance of ten centimetres of the ribbon, cold, is 0.43 ohm. A current of 6 ampères raises this length of ribbon to its melting point. The ribbon is cut away at each end where fixed in the forceps, so that its section is reduced considerably where meeting the cold metal of the forceps. This adjustment is very important in this form of the meldometer. Any serious loss of heat by conduction from the ends of the strip produces an effect increasing with the temperature, not in a simple ratio, but more as the square of the temperature, the electrical resistance being affected by the fall in temperature in such a way as to intensify the effect. Thus, the curve found for the extension will be affected; the curvature will tend generally to be diminished at high temperatures, and the position upon the ribbon we assign to the standard substances will also enter as a factor, for there will be inequality of temperature along the ribbon. A reduction of section to about one-fifth the full section will bring up the temperature at the ends sufficiently to reduce the error to a very small amount. It is easy to see, at a full red heat, how far the tint is uniform up to the forceps; but a more exact test may be made by reading melting points within about 12 millimetres from the ends and at the middle, which should afford a difference not exceeding 5° C. at a dull red heat. It will be found that if the ribbon be cut away with a scissors, and clamped as in figure 4, there will be
B.I.A. PROC., SER. III., VOL. II.