The first of these conditions is fulfilled by making the points L and L', of hardened steel, and sharpened to a knife edge, like the point C, as, in this case, the motion of the beam will not sensibly change the points of support, and, consequently, the distance between C and L will be accurately preserved.

The second condition is to make the balance sufficiently sensible, which is accomplished by attention to the centre C, diminishing its friction as much as possible. For this purpose, the planes that support the knife-edge are highly polished; and, in order that they may be preserved in their original state, the beam is not suffered to rest upon its centre but when in actual use. To sustain it when not in use, the two forks, F F are employed, which raise it from its support, and preserve it in a horizontal position. These forks are movable by means of the handle N. When the balance is to be used, the forks are lowered, and the beam set at liberty; and, as soon as the observation is completed, the forks are raised, and the beam elevated from its supports till again required. To preserve the balance from the motions that would result from currents of air, it is sometimes inclosed in a glass case, having apertures in it large enough to admit the substance to be weighed to be put into the pans.

When the instrument is not in use, it is recommended to place within the case a small saucer filled with muriate of lime, or some other substance of strong hygrometric power, to absorb the moisture that would otherwise settle on the instrument, and destroy its polish by oxidation.

In order to ascertain the value of a balance, the scales may be removed from the beam, to see whether the beam balances without them. They may then be put on again in opposite sides, and tried. Equal weights should then be placed in each scale, and afterwards changed to the opposite one; and if the beam maintains its horizontal position during all these trials, it may be considered as accurate. The utility of good balances for weighing different substances, is not limited to the accurate performance of delicate experiments, but applies also to the saving of much time in weighing, when a smaller degree of accuracy is required. If a pair of scales, loaded with a certain weight, be barely sensible to one-tenth of a grain, it will require a considerable time to ascertain the weight to that degree of accuracy, because the turn is small, and must be observed several times over; but, if a balance were used that would turn with the hundredth part of a grain, and the weight was not required to any greater accuracy than the tenths of grains, a single tenth of a grain, more or less, would make so great a difference in the position of the beam, that it would be seen immediately.

If a balance be found to turn with a certain additional weight, and is not moved by any smaller weight, a greater sensibility may be given to it by producing a vibratory, or tremulous motion, in its parts. If the edge of a blunt saw, file, or similar instrument, be drawn along any part of the case, or support, of a balance, it will produce a jarring which will diminish the friction on the moving parts so much, that the balance will turn with a third or fourth of the addition that would otherwise be required. In this way, a beam that would barely turn with one-tenth of a grain, may be made to turn with a thirtieth or fortieth of that quantity. The improvement in the balance has progressed with the general advance of the mechanic art, to such an extent, that it would seem impossible to attain a higher degree of perfection than that which has been attained in the construction of some modern balances. Mr. Read's balance, described in the sixty-sixth volume of the Philosophical Transactions, readily turned with one pennyweight, when loaded with fifty-five pounds, but distinctly turned with four grains, when tried more patiently. This is about 1/96000 part of the weight.

In the same volume, a balance, by Mr. Whitehurst, is described, which weighed one pennyweight, and turned with 1/2000 of a grain, or 1/48000 of its weight. Ramsden's balance, turning on points instead of edges, is described in the seventy-fifth volume of the Philosophical Transactions. With a load of four or five ounces, a difference of one division in the index was made by 1/160 of a grain This is 1/384000 of the weight, and, consequently, this beam will ascertain the weight correctly to five decimal places.

The Royal Society's balance, which was recently constructed by Ramsden, turns on steel edges upon planes of polished crystal. Dr. Ure states, " l was assured that it ascertained a weight to the seven millionth part. l was not present at this trial, which must have required great care and patience, as the point of suspension could not have moved over much more than the two hundredth of an inch in the first half minute; but from some trial, which l saw, l think it probable that it may be used in general practice, to determine weights to five places and better. The assay balance is of a similar kind to that which is here described in detail, but small, and extremely delicate. It is used in docimastical operations, to determine exactly the weight of minute bodies. The beam should be made of the best steel, and of the hardest kind, as this metal is less apt to be spoiled with rust than iron, and it more easily takes a perfect polish, which, at the same time, prevents the rust. The longer the beam is, of course, the more exactly may the weight be determined; but, in general, ten or twelve inches is considered a sufficient length.

The thickness of it should be so small, that two dracbm3 might hardly be hung at either end without its bending; for the largest weight put upon it seldom exceeds one drachm. The whole surface of the beam should be without ornament, as these only collect dust, and render the balance inaccurate. The whole apparatus is, when used for nice experiments, enclosed in a case with glass faces, and which are opened only so far as may be necessary to introduce the weights and the body to be weighed.