service to Captain Hewett in his survey of the North Sea. This officer states, in a private letter accompanying his official report on it," that the daily rates derived from the comparison with his standard chronometer, perhaps never were excelled in chronometrical journal."

The obstacle which seemed to present itself to the use of glass in the chronometer was its extreme fragility; this, however, proved to be no obstacle, and having prepared a glass balance spring, we took a previously adjusted chronometer, having a hardened and tempered balance spring of steel, and having regisered its arc of vibration, which was 180°, from the point of rest, we removed this spring, and applied one of glass. The vibrations were immediately increased to an arc of 200°, thus proving that glass had greater elastic force than steel, for the weight of the balance was not disturbed by the application of the glass spring, and the arc of vibration being increased and the same time preserved, its strength was necessarily equal to the original one.

Satisfied thus far with our efforts, we took it to Captain Beaufort, the hydrographer to the Admiralty, who promised us his assistance in obtaining their lordships' order for any trial to which we might wish the experiment to be subjected in our progress. Thus encouraged, the necessary adjustments for temperature, &c. were commenced.

Two important questions suggested themselves at this early stage of the experiment. The first was to ascertain how far a low temperature would affect the fragility of the glass spring; and the second was to prove by experiment whether it would withstand the shock arising from the discharge of cannon.

To determine the first question, we placed the chronometer in a temperature of 12° of Fahrenheit. In speaking of this low temperature, it may not be considered as digressing too much to explain the manner in which it was obtained; and we may be allowed this opportunity of describing our method of procuring a low temperature even during the summer months. The chronometer submitted to the process is placed in a metal vessel, having a piece of glass in the top, to allow a thermometer in it to be read, and the chronometer to be compared. This vessel, which we may call the internal one, is placed in an outer one similarly constructed with a glass top, but having a space completely parted off within it at the distance of an inch and a half from its outer sides. The distance thus left being occupied by common air, prevents the exterior atmosphere from readily acting on the ice which is to be contained within the partition. The inner vessel containing the chronometer is then placed in the centre of this outer vessel, and kept in its place by blocks of wood, and the space between it and the partition above-mentioned is filled with ice and salt mixed together. The whole is then placed in an outer wooden vessel, which is subsequently incased in flannels. By this arrange

ment a low temperature may be preserved for many hours. The chronometer with the glass balance spring was placed in a vessel of this description, and the thermometer soon fell to 12° of Fahrenheit, at which it continued, with little variation, during the first twelve hours; at the end of twenty-four hours the thermometer was at 28°. The result of this experiment we considered satisfactory as to the power of the glass spring to resist the effect of a low temperature. To enable us to determine the second question, namely, the power of resisting the shock arising from the discharge of cannon, the lords commissioners of the Admiralty were pleased to order the experiment to be made on board H.M.S. Excellent at Portsmouth. The Rev. George Fisher superintended these experiments, and their results are shewn by the following.

Having by these experiments established the conclusion, that the fragility of glass was no obstacle to its application to the purpose of forming the balance spring of a chronometer, our next step was to ascertain at what degree of tension a glass spring would break. We made a glass spring for this purpose, and applied it to the balance of a chronometer. It was wound up to 360°, detached from the escapement, and suddenly released, being shortened one coil at each trial. The following table will shew the result:

As chronometers vary in their arcs of vibration from 180° to 220°, and as the glass spring was wound up 360°, and did not break until it was shortened to seven coils, we may fairly conclude that it would have performed the usual arc of 180°, even at the seven coils, without fracture. It will be remembered, that the glass spring was applied to a previously adjusted chronometer in our few first experiments; and in adjusting it for variation of temperature, we found that we had to deal with a substance, with the nature of which we were altogether unacquainted; for on applying the glass balance spring, an excess of compensation was found, and, after repeated alterations to reduce this excess, we applied the least compensation that could be afforded us by the usual balance. This was done by placing the whole of the compensating weight at the end of the arm of the balance, instead of attaching it in the usual way to the moveable end of the arc of compensation, and, as the arm expands in a direct line from the centre of the balance, we supposed that our object would have been accomplished.

Having now only the compensation rim composed of the usual laminæ of brass and steel, and the weight being also placed at the extremity of the arm, where no curvature could be produced, to bring it nearer the centre, we found that the laminæ themselves, without any weight, produced an excess of compensation; we therefore determined on finding the amount of error in time arising from the spring when subjected to a variation in the thermometer from 32° to 100°. To arrive at this, we made a solid disk of glass, and, having removed the former balance, we applied the disk in its stead, and brought the chronometer under this arrangement to mean time at a temperature of 32°. Assuming that no error would arise from the glass balance by its diameter being increased by heat, the variation shewn when the thermometer was raised to 100° would be attributable to the glass spring only.

The temperature was raised to 100°, when, to our astonishment, the chronometer lost only the small quantity of forty seconds in twenty-four hours: the experiments were repeated, and the same results obtained.

It next became an object of much interest to find the amount of error of a steel balance spring under similar circumstances. With this view, we removed a compensation balance from a chronometer, and replaced it with a glass disk. Having brought it to mean time at 32°, we raised the temperature to 100°, and the rate shewn was losing 385 seconds in twenty-four hours. We then followed up these experiments by trying a gold balance spring; also one of palladium; and their several results are shewn in the following table: the number of vibrations were 18,000 when the chronometer shewed mean time.

This table shews that metal balance springs vary in their results, when under different degrees of heat, in the order in which they stand in a table of expansions; and with this conclusion we might have remained satisfied, had there not been such a wide difference between the results with the glass and the gold springs. This at once led us to consider that such a difference could not entirely arise from an increase of length in the springs caused by direct expansion, but principally from a loss of elasticity occasioned by change of temperature.

Our attention was next directed to separate these two causes of error, and if possible to account for the anomaly between the glass, steel, and gold balance springs. We may naturally conclude, if these alterations in the rates of the chronometers arose entirely from an increase on the length of the balance springs, due to the change of temperature, that by shortening the spring by the same quantity by which it had been augmented by the increase of temperature, a very near correction would have been effected, (making an allowance for a very small change in the inertia of the balance). This was by no means the case; the following experiment will shew that it arose principally from a variation in the elastic force of the springs.

We first measured the length of the steel spring, and, on referring to a table of expansions, we found that being 11.04 inch in length. it would increase its length 0065, in a temperature from 32° to 100°. This quantity being so small, we therefore preferred 01 inch, being in excess of that given by the table. Having shortened the spring by this quantity, which we could determine by our gauge, and keeping the glass disk applied, we obtained the difference occasioned thereby in the rate of going at 32° and 100°, then reasoning from the table of expansion, that as the spring had lengthened 0065 inches, it was fair to conclude that we had reversed the order of going, and that the chronometer would then go at the same rate when at 100° that it had gone before 32°. Instead of this, we had only advanced towards it by less than one fourth part of the quantity, shewing that the variation in the rate of going arose chiefly from a diminution of the elastic force of the spring, as seen in the following TABLE

Having shortened the spring for the fourth experiment, by '01 inch, if the difference of rate had depended on its length, it would have shewn the original rate of +5° 74 at 100°; instead of this, the rate was 70, differing only 35.3 from the former at the same temperature. To the decrease in length of 01 inch, therefore, is due the loss of 3.3, while to the loss of elasticity is due the difference between 3s 3 and 16s 04, the whole difference of rate for 68°, i. e. +5 74, at 32° and - 10s 30 at 100°.

The fact having been proved, that glass does not lose its strength by heat in the same ratio as metals, and being now acquainted with the extent of its loss, which was 40s for 68° of Fahr. we next had to construct a balance suitable to correct so small an error, and our previous experiments having pointed out to us that a metallic balance was unfit for such a purpose, we continued to employ the glass disk b, as shewn in the annexed diagram

to which was attached the glass balance-spring a. To the disk we applied in an horizontal position, two lamine composed of platina, c, c. These pieces were in length 0.5 in., breadth 0.052 in. thickness 0. 004 in. and weight 2 grains. The lightness and thinness of these pieces only allowed of one method of compensating for temperature, which was to cut off with a pair of scissors a small portion from the top, and to draw out the timing screws d, d, to bring the chronometer to mean time again, in consequence of