ment screw, to coincide exactly with the zero-point of the graduation, the introduction of a 2 decimetre tube filled with a solution of 26.048 grammes of sugar in 100 cubic centimetres, should give a rotation of exactly 100 on the scale.

For this experiment it is necessary to use specially prepared pure sugar only. The best refined still contains from 0.1 to 0.2 per cent. of inorganic matter, and does not record more than 99.8 to 99.9. In sugar-candy, the frequent presence of invertsugar, recognizable by its reduction of Fehling's copper solution, causes it to record a rotation likewise too low. To obtain a pure material, sugar-candy should be repeatedly crystallized from alcohol of about 85 per cent. Or it may be prepared by putting one part sugar-candy in half its weight of water and heating till it dissolves, filtering the hot solution into a porcelain dish, adding two parts absolute alcohol, and stirring frequently till it cools. The sugar crystallizes out as a fine powder, which should then be brought upon a filter and washed, first with dilute, then with strong alcohol, and dried at a temperature of about 60° Cent. The product thus obtained does not yield more than about 0.005 per cent. of ash, and has no action on Fehling's solution.

If now a solution of 26.048 grammes of sugar so purified does not give a rotation of exactly 100 divisions of the scale, the reading being either too high or too low, it will be necessary, before using the instrument, to determine the particular normal sugar weight belonging to it. Suppose, for instance, that the above solution has given the reading 100-3° on the scale, the sugar-percentage answering to a reading of 100 must be determined from the proportion

100.3°: 26.048 grammes 100°: x

=

whence x = 25.970 grammes.

So that in all applications of the instrument the normal weight must be taken as 25.970 instead of 26.048 grammes, and 1° of the scale must be understood to indicate 0.2597 gramme sugar in 100 cubic centimetres solution. In this way accurate results can be obtained with such an instrument, but it presents the disadvantage that all tables calculated for the number 26.048 are useless, and must be reconstructed.

If the 100 point has been found in its proper place, still the scale, which is supposed to be equally graduated throughout its length, must be tested at a few other points as well.

Proceeding, according to Schmitz's table, given in § 82, solutions are prepared containing respectively 19.519, 13.003, and 6-496 grammes of pure sugar, which should give rotations of 75°, 50°, and 25° respectively. If, however, important discrepancies should occur in the readings, then it becomes necessary to prepare a whole series of solutions of known saccharine strengths, note the degrees of rotation indicated by each, and then draw up a special correctiontable for the instrument. This is best done by the graphic method.

Errors of this kind appear when the four faces of the two quartz wedges of the compensator have not been ground perfectly true, so that differences in the total thicknesses of the compensators, produced by the sliding of the wedges, are not perfectly proportional to the differences of reading on the scale. Scheibler1 has given a method by which such errors, which are of frequent occurrence, can be eliminated, at least when we are dealing with rotations exceeding 80°.

This, the so-called method of double observation, most generally used in the analysis of natural sugars, is as follows:-A sample of 26.048 grammes is made into a 100 cubic centimetre solution, or more commonly 13.024 grammes are taken and a 50 cubic centimetre solution prepared, and the rotation observed in a 2 decimetre tube in the ordinary way. If the degrees indicated be, let us say, 94.2° (which should therefore be the sugar-percentage), this result will be correct provided the quartz wedges at the point corresponding to this reading are of the proper thickness. To test this, we must calculate the concentration required to give a rotation of 100° by the proportion 94.2° 13.024 100°: x

=

whence x = 13.826.

A 50 cubic centimetre solution must then be prepared with 13.826 grammes of the sugar to be analyzed, and the rotation observed in a 2 decimetre tube. If it gives 100° on the scale, the first result, 94.2 per cent., is correct.

If, however, the second solution does not give exactly 100°, but some less number, as 99.6°, in that case the result of the first observation is incorrect. The correct sugar-percentage can, however, be easily ascertained, as it must stand in the proportion

13.826 99.6 13.024 : x

X = 93.8.

The true sugar-percentage is accordingly 93-8, and the number of degrees indicated on the scale in the first experiment (942) was 0.4 too high.

1 Scheibler Zeitsch. des Vereins für Rübenzuckerindustrie, 1870, 212; 1871, 318.

If, again, the second experiment gives, say, 100-2 instead of 100, the proportion will then stand:

13.826 100·2 = 13.024 : x,

and the correct sugar-percentage will be

x = 94·4 per cent.

In this way errors arising from imperfect construction of the quartz wedges can be eliminated, and as the second observation invariably lies close to the 100 point, the position of which has already been accurately fixed, it is found that the results given by different instruments correspond within ±0.1 per cent., while by the ordinary method the differences between them may be much greater. As to the effect of the imperfect proportion between rotation and concentration upon this method, we have already said (§ 82) that when we are dealing with sugar-percentages recording more than 84°, such as we find in crude sugars, it may be neglected. When, however, the solutions indicate deviations ranging from 30° to 76°, this is no longer the case, and the method of double observation fails to afford correct results.

It will be seen from the foregoing remarks that the quartz-compensation principle in polariscope instruments involves considerable difficulties. Errors and corrections, like those just mentioned, do not occur in instruments with rotating Nicols (Wild's and Laurent's); moreover, these give generally more accurate results.1 For the method of verifying the length of solution-tubes see § 66. For the influence of glass end-plates see § 64.

§ 85. Influence of Temperature on Determinations by the Saccharimeter. The normal temperature at which the determination of the 100 point is made being 17.5° Cent. (63-5 F.), the experiments will ordinarily be made at some other temperature; for the experimental tubes not being usually provided with water-jackets, but exposed to the air, are subject to any variations in the temperature of the apartment. It is true, indeed, as Tuchschmid's researches have shown, that the specific rotatory power of sugar is not in itself affected by heat, but the directly observed deviation is influenced thereby. Thus, when the temperature rises, the length of the tube,

2

1 Schmidt and Haensch, opticians, Berlin, have lately brought out a half-shade instrument, with quartz-wedge compensators and Ventzke's scale. This instrument admits of much more accurate adjustment than the usual bi-quartz colour instrument. The variations do not exceed 0.1 division and they can be used by colour-blind persons. 2 Tuchschmid: Journ. für prakt. Chem., New Ser. 2, 235.

on the one hand, is increased, while, on the other, the density of the contained solution is reduced by its increase of volume. The former tends to increase the deviation, but this tendency is overpowered by the larger decrease due to the latter. Mategczek,1 taking as his basis the dilatation-coefficient of glass, § 66, along with Gerlach's researches on the density of saccharine solutions at different temperatures, has calculated the variations arising from this source. He gives a table for Ventzke's scale, of which the following is an abridgment :

:

1 Mategczek: Zeitsch. des Vereins für Rübenzuckerindustrie, 1875. 77.

2 Gerlach: Idem., 1862, 283.

Hence, for example, if a thermometer plunged in the contents of the tube, after the rotation has been observed, indicates, say 20°, then, in calculating the sugar-percentage, the constant 0-26064 must be used instead of 0.26048.

(b.) The Soleil-Duboscq Saccharimeter.

§ 86. The original Soleil instrument, as used in France, and manufactured at the optical instrument works of J. Duboscq, Paris, agrees essentially in principle with that just described, except as regards the scale. This is so devised that the deviation produced by introducing a plate of dextro-rotatory quartz 1 millimetre thick is taken as a fixed point, and the space between this and the zero-point is divided into 100 equal parts. Now, according to Clerget,1 the same amount of deviation is produced by a 2 decimetre column of a sugar solution containing 16:471 grammes in 100 cubic centimetres. This concentration was subsequently corrected by Duboscq2 to 16·350 grammes, and the scales for this instrument are now so constructed that the 100 point is recorded by a solution containing 16.350 grammes of pure dry sugar-candy in 100 cubic centimetres. Consequently each degree of the scale represents 0·1635 gramme in 100 cubic centimetres.

In construction, the French-made instruments differ from the German (Ventzke-Scheibler) in having both quartz wedges movable. Besides this, the regulator for producing the transition-tint is placed in the eye-piece of the instrument, with its Nicol immediately behind the eye-glass of the telescope, and capable of partial rotation by means of a rim projecting through the telescope tube.

In using this form of instrument, 16.35 grammes of the substance to be examined are weighed, made into a 100 cubic centimetre solution, and the solution examined in a 2 decimetre tube. The degrees of rotation on the scale indicate directly the number of grammes of sugar in 100 cubic centimetres. In other respects, the method of observation is precisely the same as in using the SoleilVentzke-Scheibler saccharimeter. To avoid errors from imperfect

1 Clerget: Ann. Chim. Phys., [3] 26, 175.

2 According to the still later researches of Schmitz and Tollens, a still more correct val for this constant is 16.302 grammes. (From the equation for c, given on p. 180chschm.