The thermometer is placed in a glass tube, T (fig. 102), through which the solution passes, running into the pipette. It is suspended by a cork having four channels cut in it to allow the free passage of the liquid. The scale is engraved on the tube itself, and is repeated on the opposite side, so as to fix the eye by this double scale to the height of the thermometric column. The tube is fused at its lower end to a narrower tube, which is fixed by means of a cork into the socket of the stopcock of the pipette. The upper part of this tube is cemented to a socket of copper, tapped inside, which in its turn is fastened by a cock B, with the extremity (also tapped) of the tube T', communicating with the reservoir of normal solution. The corks used as joints between the parts of the apparatus retain a certain amount of flexibility, and allow it being taken to pieces and put together again in a short space of time; but it is essential to pass them into a hollow tube of glass or metal, to prevent them giving way under the pressure they have to sustain. If care be taken to coat them with a little tallow to stop the pores, no escape need be apprehended.

Preservation of the Normal Solution of Salt in Metallic Vessels. This subject has already been discussed, and it may appear unnecessary to again refer to it; but as it is here a question of metallic vessels, some details seem

necessary.

The figure 103 represents a cylindrical copper vessel, C, holding about 110 litres. It is seen in section, Z, same figure. To its base is soldered a socket, D, to which is adapted a tube, with stopcock, T, through which the solution passes into the pipette; the upper part, which is slightly concave, having an opening closed by a screw stopper, B, the edges of which press on a washer. This stopper is traversed by the tube t, which passes nearly to the bottom of the vessel, and through which air enters the apparatus, without the power of again passing out, so that evaporation is effectually prevented. This tube can be closed by a stopper, m, when the apparatus is not in use.

The quantity of liquid contained in the vessel can be determined at any time by the aid of a wooden gauge, J,

graduated into litres. When used it is plunged vertically into the liquid, but is seldom needed.

Pure or tinned copper alters in contact with the solution of salt and air, and the solution continually decreases in strength. This inconvenience is remedied by coating the inside of the cylinder with a soft cement, such as described at page 108; or with that cement softened by the addition of one-third its weight of yellow wax. This operation may be performed

by removing the tubes T and t, perfectly cleansing the inside of the cylinder, and heating it. About four or five pounds of the cement, made very hot, are run in, and the cylinder so turned round and inverted that the cement may run over every part. The turning is continued until the cement is cold. All the parts just described are united in

the figure 103, forming a complete apparatus for the preservation of the normal solution of salt, for observing the temperature, and for measuring the volume.

Preparation of the Normal Solution of Salt, measuring by Volume.-The preparation of the normal solution of salt, measured by volume, is much the same as of the solution measured by weight; there is, consequently, very little to add to that already given at pages 501-505, and to which the reader is referred.

The cylinder, as already supposed, will contain about 110 kilogrammes of water: no more, however, than 105 are put in; so that sufficient space may remain in order to agitate the fluid without throwing any out. According to the condition imposed, that 100 cubic centimetres, or onetenth of a litre, of solution, should contain sufficient salt to completely precipitate 1 gramme of pure silver; and further, admitting 13.516 for the equivalent of silver, and 7-335 for that of salt, the quantity of pure salt to be dissolved in 105 litres of water, and which corresponds to 105 × 10=1050 grammes of silver, will be found by the following equation :

13.516 : 7.335 :: 1050 gram. : x = 569·83 gram.

And as the solution of commercial salt employed, page 501, contains approximatively 250 grammes per kilogramme, 2279.3 grammes of this solution will be required to furnish 569-83 grammes of salt. As the 2279.3 grammes of solution contain 569-83 grammes of salt, it will consequently contain 1709-5 grammes of water, which must be taken into account in measuring the 105 litres: that is no more than about 103.3 must be employed. The whole being well mixed, the tubes and pipette must be washed out several times, by allowing the solution to run through them. The solution so passed is again placed in the cylinder, and after each addition the contents are well agitated, and lastly, the standard of the solution is determined, the temperature being supposed to

remain constant.

To accomplish this more readily, two decime solutions are prepared; one of silver, and the other of salt.

The decime solution of silver, as already stated, is obtained by dissolving a gramme of silver in nitric acid, and diluting the solution with water until its volume is one litre.

The decime solution of salt can be obtained by dissolving 0.543 grammes of pure salt in water, so that the solution fills a measure of one litre; but it is best prepared with the normal solution itself, which is to be standardised, by mixing one measure of the latter with nine measures of water. It must, however, be understood, that this solution is not rigorously equivalent to that of the silver, and only becomes so when the normal solution employed in its preparation becomes fixed at its true standard. If the normal solution be correct to ten thousandths, or one hundredth, the decime solution may be correct to the same degree. If ten thousandths of the latter solution be employed, the error committed will be one-tenth of a thousandth; and only one hundredth when one thousandth is employed. Such errors may be entirely neglected; nevertheless, after having exactly standardised the normal solution, it is better to prepare a new decime solution.

After the preparation of the decime solutions, several bottles, as at fig. 89, must be prepared, each of which contains 1 gramme of pure silver dissolved in 8 or 10 grammes of nitric acid. To these will be given the name of check, or witness-assays.

To ascertain the standard of the normal solution pour a pipetteful into one of the check flasks, and agitate briskly until quite bright. After a few moments' repose, two thousandths of the decime solution of salt are added, which, by superposition, will produce a precipitate. The normal solution is consequently too weak, since the salt employed was not perfectly pure. It is again agitated, and two other thousandths are added, which produce a precipitate. The addition of successive two thousandths is thus continued until the last produce no precipitate. Suppose in all sixteen thousandths have been added: the two last which have been added are not reckoned, as they produce no precipitate the two preceding have only been in part

necessary; that is to say, that the acting thousandths added are above 12 and below 14, or, taking the mean, equal to 13.

Thus in the existing state of the normal solution 1013 parts are necessary to precipitate 1 gramme of silver, while only 1000 should be required. The quantity of concentrated solution of common salt to be added may be found by noting that the quantity of solution of common salt first employed that is to say, 2279-3 grammes-has only produced a standard of 1000-13=987 thousandths, and by the following equation :

987 2279-3 :: 13: x = 30.02 grammes.

This quantity of solution of common salt must, therefore, be mixed with the normal solution.

After having washed the tubes and pipette with a new solution, another check gramme of silver is operated on. It is found, for instance, by proceeding but by one thousandths at a time, that the first precipitates, but the second does not. The standard of the solution is therefore too weak, being comprised between 1000 and 1001; that is to say, it is equal to 1000: this, however, is not sufficiently near.

Pour into the assay flask two thousandths of the decime solution of silver: these will merely decompose the two thousandths of salt, and the operation will have retrograded by two thousandths; that is, it will be reduced to the point from which the thousandths were first employed. If, after brightening the liquor, half a thousandth of the decime solution is added, there will necessarily be a precipitate, as was before known; but a second half thousandth produces no cloudiness. The standard of the normal liquid is therefore between 1000 and 1000, or equal to 10001.

This for most purposes may be considered sufficiently near; but if it be desirable to correct it, it may be remembered that the two quantities of solution of common salt added,

have only produced 999-75 thousandths, and that it is necessary to add a fresh quantity corresponding to the