sensible uniformity of temperature over the central six centimetres of the ribbon. I may observe that sometimes in the case of a ribbon very much cut away, and with a particular tension of the spring, keeping it stretched, a transverse vibratory motion of the ribbon takes place when it is heated. This is due apparently to aircurrents in the same manner as a flat body thrown into the air falls, with horizontal oscillations, from side to side. This vibratory motion, of course, renders vision indistinct. be stopped by bringing a fine platinum wire, fixed to a support, to touch the edge of the ribbon at its central point, but I have only occasionally noticed the phenomenon.

It may

FIG. 4.

Before using a particular ribbon for measurements all traces of kink or bend is got out of it by raising it to a white heat for half a minute. In this connexion it must be mentioned that the pull of the ribbon must be very slight, just sufficient to keep it stretched. A force equivalent to the weight of 1.5 grammes is sufficient. The ribbon I have found it best to use weighs 0.0073 grammes per centimetre run. The maker's number for this is 0.170. A less section is not advisable, as being too much affected by slight draughts. It is essential, of course, that the ribbon be of constant cross-section. There is some advantage in surrounding the ribbon by an open trough-shaped box of platinum-foil from end to end. But while this conduces to steadiness in the temperature of the strip, it has the disadvantage of introducing a fresh difference of conditions towards the ends, except troublesome arrangements are made. On the other hand, the amount of fluctuation obtaining in the case of observations made in a quiet room are small, and often not disadvantageous in observing the melting point. Thus we observe the condition of the substance changing with these small fluctuations; and we set the micrometer so that the galvanometer needle is free when the substance shows the first signs of melting, and is deflected on the instant that solidification takes place. In this way the melting points of non-viscous bodies can be caught with much accuracy. And I may observe here that, although some of Carnelley's observations' appeared to indicate a considerable difference between the melting point and the solidifying point of many of the salts he dealt with, this difference is not substantiated by observations on the meldometer. The salt may generally be observed, with

Journal of the Chemical Society, vol. i. (1876).

care, in such a state that the zone of liquefaction may be seen advancing from beneath, or retreating downwards at the least fluctuation of temperature. I think the method adopted by Carnelley for determining the point of the solidification was calculated to introduce an error due to radiation.

It is advisable not to assume that a curve plotted for one ribbon is suitable to another one which has replaced it, till we have verified a couple of points upon it, using such easily observed melting points as are given in the sequel. Then, if a speck of silver chloride, and of the black oxide of copper, CuO, be melted upon it, and finally a little palladium, we may assure ourselves upon this point. To avoid any chance of error due to the position of the substance upon the ribbon, no more than the central six centimetres should be availed of for determinations.

For the control of the temperature a higher resistance than that previously described is necessary. As the Rheostat I have in use is new in many particulars, and is very satisfactory, I think it well to describe it. It consists (fig. 5) of eight rods of carbon, having a total

FIG. 5.

resistance of about 4 ohms, fixed horizontally upon a board. Two of these are provided with a sliding cross-connexion; the others may be, one by one, thrown into circuit, or cut out by a few turns of the screws

seen at the front of the board in fig. 5. The raising of one of these screws, for example, permits a brass plate to spring out of contact with the underlying metal, and as the plate is cut out sufficiently to clear it of the shaft of the screw, connexion is broken when the nut is raised, till it is no longer followed by the plate. The current then must traverse the carbon c, returning to the point d by the dotted connexion. It is evidently possible to secure a perfect gradation of temperature by the use of this resistance, the travel of the slides commanding a greater resistance than that of a single carbon. Thus if, when a certain number of carbons are in circuit, and the bridge has been moved to cut out the entire length of the two bridged carbons, we need still a higher temperature, we restore the bridged carbons to circuit by moving back the slides, and then cut out one of the carbons, after which we begin again to advance the bridge so as gradually to cut out the bridged carbons. In this way we increase the temperature, nor is it necessary to take our attention from the microscope and the manipulations of the micrometer, while making these changes in the resistance.

The diagram (fig. 3, p. 52) shows the convenient disposition of the apparatus when effecting measurements. To observe the objects upon the ribbon a simple form of microscope, mounted on a bracket, projecting from a heavy base, is requisite. This can be moved along and directed to any point of the ribbon. One ribbon, of course, serves for many observations.

A principal advantage of this electrical method of reading the elongation of the ribbon is contained in the fact that so little of the observer's attention is called upon to effect this ordinarily difficult measurement. This is very important, for slight variations of temperature are continually occurring, due to draughts chiefly, and it is essential to be able to seize the reading at the moment the substance melts, especially in the case of bodies which flood out suddenly upon the platinum. The galvanometer needle may be observed simultaneously, with the object in the field of the microscope, if the head be raised a little above the eye-piece. The left hand is kept upon the micrometer screw, and at each small movement of the bridge (effected by the right hand) the elongation is followed by the micrometer, and the behaviour of the substance at the increased temperature observed.

To render it more convenient to observe the deflection of the galvanometer simultaneously with the melting of the substance, or whatever other phenomenon we are observing upon the ribbon, I use the following arrangement :-A small galvanometer coil, wound with fine

wire, and containing a freely pivoted magnetized needle is affixed to the eye-piece of the microscope. Projecting from the needle at right angles to its length, and entering a horizontal slot in the eye-piece is a fine index-hand. This moves across the margin of the field, being accurately in the plane of the image formed within the eye-piece by the lower optical train. Its motion is confined to a small arc, the magnetized needle without being stopped by pins at either side. This renders it fairly dead beat. When no current is passing the needle points to the centre of the field under the influence of a small control magnet affixed to the coil. Binding screws enable this galvanometer to be circuited with the larger galvanometer, which stands upon the table, and the Leclanchè battery. This is a very great addition to our power of effecting an accurate measurement.

The appearances which characterize melting vary according to the nature and properties of the substance. A viscous substance is seen to round its angles slowly, the smallest fragments going first. If the temperature is well over the softening point the process is rapid, and spreads from small to large fragments, the whole coalescing finally in pools upon the platinum. It is always possible in such cases to fix a temperature at which the substance decisively melts, and one at which prolonged heating seems hardly to effect it. Between those limits the melting point may be assumed to lie. In the case of substances which melt rapidly and recrystallize on solidification, it is easy to fix the melting point very closely. In fact the temperature may be so regulated that the small passing changes of temperature affecting the ribbon determine fusion or solidification; the result being that the salt appears as if vibrating between the solid and the liquid states, rays of crystallization darting across its surface, or again melting at the edges. These bodies are very easily dealt with the salts selected for calibrating the ribbon, i.e. for the construction of the curve, are of this nature-as the carbonate and nitrate of potash, as well as potassium bromide. Silver chloride is also a salt of this nature, but it possesses also the peculiarity of changing colour when melting. Solid, it is a greenish-white, crystalline substance; melted, it is an amber-coloured liquid, deepening in colour with rise of temperature. The passage from the one state to the other is a well marked point. Similar phenomena mark the melting point of silver sulphate. If the substance is a metal the melting point is looked for in either of two ways. We may scrape a little dust off the metal on to the platinum ribbon, and, raising the temperature continually, catch the reading when the little fragments suddenly run down. A second trial, in which we may more carefully approach the melting point,

will fix a correct reading. In the cases of silver and palladium, we may observe the behaviour of a pool of the previously melted metal; when such is again just melting a change of colour of the surface is visible. The appearance is as if a red or yellow flush (according to the temperature) passed across the surface; it is very characteristic, and seems due to the smoothing of the rugosities on the solid substances which keep the surface darker by their greater area of radiation. Points determined in this way, in the case of silver and palladium, agree with the points determined, by observation of the breakdown of very small scrapings of the metal. It is to be observed that metals should be determined only on the first or second melting, as there is risk of mutual solution or alloying with the platinum, which would effect the melting point. In general, it may be said that the breakdown of very small fragments, or a rocking motion apparent in the larger fragments, are the readiest, and probably the surest, signs of melting.

Coming now to the consideration of the range possessed by this apparatus for the comparison of melting points-for such it obviously is-it will be convenient to refer to the curve of melting points (Plate VI.) obtained by observations on the meldometer, such as I have described. The points fixed on the lower part of the curve are based, principally in the results of Carnelley (see his Melting and Boilingpoint Tables, or the résumé of his results given in the Physikalischchemische Tabellen of Landholt and Bornstein. Carnelley and Williams' observations were made by a calorimetric method, in which they heated a platinum vessel to the temperature at which the salt melts, and plunging the platinum in water determined the quantity of heat contained in the platinum at the unknown temperature.' The specific heat of platinum had been measured by Pouillet, by direct comparison with the air-thermometer up to 1200° C. The data so obtained are used by Carnelley. This physicist was so well acquainted with the whole subject that it appears hardly possible to say more in their support than that he was satisfied to accept them as the best basis available whereon to investigate and discuss the periodicity of the melting points of certain groups of compounds.

3

Violle, finding the variation of specific heat with temperature by reference to the porcelain air-thermometer, determined the melting points of silver, gold, copper, palladium, etc. His determinations therefore rest on a similar basis to those of Carnelley.

1 Carnelley and Williams, Jour. Chem. Soc., vol. i. (1876), p. 489.

2 Pouillet, Comptes rendus, vol. iii. (1876), p. 782.

3 Comptes rendus, vol. lxxxv. (1877); vol. lxxxvii. (1878); vol. lxxxix. (1879).