the fixed carbon was only 13 per cent. of the liquid absorbed, or 26 per cent. of the sugar in the solution. At this rate, every ounce weight added to the carbons would cost 14d., with sugar at its present cheap rate, besides the additional trouble of having the carbons to dry after each soaking. In using coal tar as a 'substitute, we first heated it to a temperature of 300°, so as to drive off the Naphtha Hydo-carbons, which left it in a semipitched condition; and, while still hot, the coke bars were soaked in it till saturated, and sank to the bottom. The percentage of fixed carbon left in the coke after the soaking, and heated to redness, was 32.5 per cent. of the tar absorbed; or 2 times the weight obtained from sugar, and secured at an infinitely less cost. Pure tar carbon is among the best that I have yet tried for Bunsen Batteries, and it was the success in some experiments that I was making with it that induced me to try it as a source of fixed carbon for the purpose now mentioned.

Three separate soakings in the tar, as described, and as many times heated to a high temperature, are necessary fully to close the pores of the coke. Before the last steeping, the carbons are ground upon a flat stone into the required shape. In grinding a little water is used, but only so much as may form the abraded powder from the carbons into a pasty state; and as the carbons are still absorbent at this stage, they imbibe the water from the paste, and leave the particles of carbon deposited in the pores on the surface, thereby leaving the close surface shown in the finished state. The carbons are again soaked in tar, and charred at a high temperature, when the process is completed by the final smoothing on a flat stone.

The manufacture of the carbons is conducted simultaneously with the process of gas-making; and thereby, with economy of heat, a considerable amount of carbon is deposited from the gas itself, in addition to the fixed carbon from the tar absorbed. For convenience of having them properly placed upon the coal in the retort, a long semicircular trough of iron is prepared (similar to a water run for eaves of houses) to contain from twelve to twenty at one time, and after the charge of coal is introduced, and before closed up, the long trough and its contents are thrust in over the coals, close up to the roof of the retort, and allowed to remain till the charge of coal is wrought of-a period of nearly three and a-half hours. On being with

drawn, the carbons are scattered about, to cool as rapidly as possible, and prevent their burning away by the action of the air; and, when cold, they receive the slight rub to smooth them, which finishes the process.

How far these carbons suit the purpose for which they are made, I would only refer to two occasions, from among a number, in which they have been used. The first, to that on which the electric light was exhibited from St. Anthony's Chapel, by Mr William Hart, Philosophical-Instrument Maker, Edinburgh, was on the arrival of her Majesty from the North in the autumn of last year; and the second, to their use in a lecture delivered by Professor George Wilson on "The Metals." On both of these occasions, the carbons worked satisfactorily.

I do not wish it to be understood that I consider these carbon elements have reached their perfect state. This notice. only shows the process in a comparatively infant stage; and I hope to be able to prosecute the experiments to greater maturity, so as to be able to place a still better article in the hands of the student of electricity.

On a New Method of Measuring Watch-Glasses. By ALEXANDER BRYSON, F.R.S.E., Her Majesty's Clockmaker for Scotland.*

The method hitherto employed for gauging watch-glasses was by a nonius scale, divided into centimetres, each of which were again divided into sixteen equal parts. By this method considerable accuracy was obtained in measuring the diameter, but as most watch-glasses have a certain amount of ellipticity, it gave no true indication of the extent of the circumference. To obtain the proper size of the glass, it required to be rotated on the scale, and the ellipticity estimated by the difference of the diameter.

This process was necessarily slow, and at least uncertain; and as the lines on the nonius scale are fine, considerable acuteness of vision was indispensable.

Some idea of the difficulty of measurement may be formed,

* Read before the Society, and Drawings and Instruments exhibited, 25th April 1859. Awarded the Society's Silver Medal.

when we find that the difference between size 12, the smallest, and 23, the largest glass, is only ninety-five hundredths of an inch (95), which has to be divided by the nonius into 176 equal parts. This difficulty is also increased by the ellipticity. in the larger glasses, amounting, in some instances, to three, or even four sizes.

To obviate these sources of error, the gauges to be described were constructed.

Plate II., fig. 1, represents the gauge used for ascertaining the circumference of the glasses, and is drawn to size. It consists of two plates of brass, one fifteenth of an inch in thickness (0.066), supported on six pillars two-tenths of an inch in height (0-20). The positions of the pillars are indicated by the screw-heads in the drawing. Numbers 1, 2 and 3 are hardened steel rollers, forty-one hundredths of an inch in diameter (0.41), they have each two pivots, working in steel sliders, which fit accurately into the three slots into the front or upper plate, and also into three corresponding slots in the plate below. The pivots in the upper plate protrude above it two-tenths of an inch (0 20), and embrace the glass at three points of its circumference. The breadth of the slots in which the slides or pivot carriages move is one-tenth of an inch (0·10), and six and a-half tenths in length (0·65), giving a range sufficient to measure glasses from size 12 to 23. C is a fine watch chain, fourteen inches in length, attached to a hook fixed to the front plate at h. This chain works in a grooved gun-metal roller, placed on the pivot of roller No. 1; it is then conducted to a similar grooved roller on No. 2, and then to roller No. 3, which also carries a gun-metal grooved roller. The chain is now warped round a grooved steel roller on No. 1; but above the gun-metal roller formerly described, it is then carried round a hardened steel cylinder, No. 4, which is of the same exact diameter as the other three-viz. 0·41 of an inch. The chain makes two and a-half revolutions round this cylinder, and is attached to it in the same way as a watchchain is to the mainspring barrel, by means of an oblique hole drilled into it for the reception of the hook. On the axis of the cylinder No. 4 is placed the index or pointer I, indicating on the dial A the size of the glass to be measured. The pivot of this cylinder proceeds through the lower plate 0·30 of an

inch, and has attached to it a watch mainspring, which tends to turn it in the direction of the arrow; by this means the chain is wound round the cylinder, and the three rollers are pulled towards the centre.

If a glass be now placed within the three pivots, it will be embraced at the corresponding points of its circumference, and the value indicated by the pointer I.

If any depression or ellipticity exists at points 1, 2, or 3, the exact amount must be told by the index, as each roller is independent of the others, and thus the extent of the periphery of the watch-glass ascertained with precision.

D is a cam-wheel with three epicycloidal curves, for the purpose of extending the rollers. The axis of this wheel is pivoted in both plates; on the upper it is level with the plate, but extends through the lower 0·50 of an inch; to this another watch mainspring is attached, tending to throw the epicycloidal arms outwards, and is of such a strength as nearly to counterbalance the spring on the axis of the index. Beyond the box which contains the mainspring, the axis is squared, and a milled head adapted, by means of which the hand is enabled to throw out the rollers for the admission of the glass to be measured.

A is the dial divided into ninety-six equal parts, each division being the equivalent of the sixteenth part of a centimetre. We are thus enabled to measure watch-glasses with an instrument which corresponds with the gauges used at Geneva, and by the principal manufacturers in England, with an amount of accuracy hitherto unattained. The ease of reading the new gauge, as compared with the old one, is as 155 to 1, and the increased accuracy in the same ratio.

Figure 2 in Plate is a side view of the instrument also drawn to size, and shows the disposition of the chain round the rollers and index cylinder. It exhibits also the mainspring boxes at A and B, and the milled head D, by which the rollers are extended.

Figure 3 in Plate is an enlarged representation of roller No. 1, with its five pieces separated, to show their positions; the sliding pivot-carriage above and below. Also the position of the gun-metal grooved roller, and the steel one of double thick

VOL. V.

U

ness, on which the chain is warped one whole turn before being attached to the cylinder or index axis.

By thus warping the chain round the roller No. 1, the relative angles which the three rollers make in all positions with the cylinder axis are the same, and hence equal increments of size in the watch-glasses are indicated by equal spaces on the dial.

The annexed table exhibits an analysis of the measurements of 9778 glasses made by the new instrument to test its accuracy as against the standard nonius Geneva gauge. It is necessary to remark, that all these glasses had been twice sized and marked-first by the makers in Geneva and London, and secondly by myself-before they were passed as correct, and placed in their cells, where they are kept according to their sizes. These glasses were all, therefore, as accurately sized as the former method was capable of attaining, and gave me an opportunity of testing completely the value of the new instrument.

The upper portion of the table shows the measurements of the Geneva glasses and the lower of the English. The first column contains the sizes, where 12 is the smallest glass in % ordinary use for Geneva watches up to 23%, the largest; each number, therefore, from 12 to 23, contains sixteen sizes of glasses.

The second column contains the number of glasses which were found rightly sized by both methods, and the third column those that were found wrongly sized, as detected by the new instrument.

The fourth column shows the difference of the right and wrong-sized glasses, and in the fifth is exhibited the proportional difference per cent.

From this analysis it will be seen that the old method, although very faulty in all the sizes, measured the smaller glasses more accurately than the larger; but this arises from the fact that a small disc has less chance of being elliptically ground than a larger one.

This fact is shown in the third column, where 163 glasses of the largest sizes had been passed as correct by the nonius gauge, but were all found wrong by the new instrument.