owing to the disappearance of the birotation of the levulose. This latter statement, however, requires considerable qualification, since such a decrease in levorotation as that referred to was not observed in any of the liquid honeys examined. In the case of the twenty honeys of this list which showed no indications of granulation every sample showed a decided increase in polarization toward the left.

A neglect of the phenomenon of multirotation naturally introduces a serious error in the case of those analytical constants which are derived from polarization; such, for example, as the direct polarization and the sucrose by the Clerget method. Honeys Nos. 63 and 92, for example, showed an initial rotation of +3.7 and +0.8, which would exclude them as honeys according to our standards of purity. The constant reading of these same honeys, however, was -0.3 and -10.8, which brings them within the standards as being levorotatory. In the same way honey No. 92, using the immediate polarization (+0.8), indicates by the Clerget formula 11.77 per cent of sucrose, an amount considerably above that allowed by the standards. (8 per cent), whereas using the constant rotation (-10.8) a sucrose value of only 2.98 per cent is obtained.

DIRECT AND INVERT POLARIZATIONS AT 20° C.

In comparing the direct and invert polarizations of the various honeys there are several points which should be noted. The statement is sometimes made that a pure honey will rarely polarize more than 20° V. at 20° C., and that a higher reading than this, while not conclusive evidence of adulteration, may well be looked on with suspicion. This statement is hardly borne out by the analytical results given in the table, where sixteen of the 100 honeys examined polarized more than -20°, the results going as high -24.6° in a tupelo honey and -24.8° in a mangrove. A high levorotation in a honey may be due to several causes, such as imperfect sampling through sedimentation of dextrose, or to a natural preponderance of levulose in the floral nectar (the explanation in the case of the mangrove and tupelo honeys) or to the honeys being comparatively free from sucrose and dextrins of high positive rotation (the explanation in the case of the mesquit and algarroba honeys).

The invert polarizations at 20° C., owing to the removal of the disturbing influence of variable amounts of sucrose, show usually a closer uniformity than the direct polarization. The alfalfa honeys, for example, show a range of 9.4° to 20.5° V. for the direct readings, and a range of only -21.01° to -24.53° for the invert. The invert polarization is usually a good index of the purity of a honey as regards the amount of non-sugars; the alfalfa honeys, for example, having an invert polarization of -22.99° and 2.05 per cent organic non-sugars, the basswoods -12.25° and 6.52 per cent, the white oak

+5.17° and 15.47 per cent. There are, however, important exceptions to this rule, as in the tupelo honeys which had an invert polarization of — 28.38° at 20° C. and yet contained 7.34 per cent of non

sugars.

CALCULATION OF SUCROSE BY CLERGET'S FORMULA.

The error resulting from a neglect of the phenomenon of birotation in the calculation of sucrose in honeys by Clerget's formula has already been mentioned. Another very serious error is that due to the change in the specific rotation of the levulose of the honey after the inversion with hydrochloric acid. This error in sugar analysis has been so thoroughly discussed by Lippmann" and others that no detailed study of it need be undertaken here. Tolman," who has recently made an investigation of this error, shows by analyses that many honeys which when analyzed by the Clerget method appear to contain cane sugar contain in reality none at all, the error in many cases exceeding 1 per cent. From the general averages of the direct and invert polarizations given in the table (page 38) the average sucrose content of the 92 levorotatory honeys by Clerget's formula is 3.31 per cent, and the average percentage of sucrose as determined gravimetrically was 1.90 per cent, an average error of +1.41 by the Clerget formula. For the seven dextrorotatory honeys the average per cent of sucrose by Clerget's formula was 2.96 per cent and by the gravimetric method 3.01 per cent, an average, error of -0.05 per cent. In the latter class of honeys, however, there was present a considerable amount of optically active non-sugars (dextrins, etc.), and the influence of the hydrochloric acid upon the rotation of these bodies may have acted as a compensating error.

On account of the numerous errors just cited in connection with the use of the Clerget method and the difficulty of introducing any correction for these errors, the determination of sucrose in honey, when accuracy is required, should be made only by the gravimetric method. A neglect of this precaution might cause injustice to the producers. Honey No. 23, for example, had been found to contain 8.95 per cent of sucrose by the Clerget method, a larger amount than that allowed by the standard (8 per cent). The gravimetric method, however, gave only 7.09 per cent, which is well within the limits allowed.

DIRECT AND INVERT POLARIZATIONS AT 87° C.

These determinations, which have been recommended for honey analysis, are based upon the fact that a solution of pure invert

a Chemie der Zuckerarten, 3d ed., 1: 922.

U. S. Dept. Agr., Bureau of Chemistry, Bul. 73, p. 69.
Lippmann, Chemie der Zuckerarten, 1:919.

sugar at 87° C. becomes optically inactive. Hence any positive rotation observed at this temperature is due to the presence of other optically active bodies than invert sugar. The readings at 87° were made both before and after inversion, and the difference between these readings gives approximately the percentage sucrose as found gravimetrically. The invert readings at 87° were made in acid solutions, and the changes due to specific rotation of levulose probably explain the discrepancies. It is also most probable that a slight destruction of levulose took place at 87° in the acid solution. To determine the extent of this error two 26-gram samples of sucrose were inverted and polarized at 20° and 87°, the one in acid solution. and the other after neutralizing, with the following results:

Polarization of acid and neutral solutions of invert sugar at different tempera

The acid solution polarized higher to the left at 20° owing to the increase in rotation of levulose and higher to the right at 87° on account of a slight destruction of levulose by the acid at this high temperature. It is therefore always a safer plan to neutralize the acid after inverting, before making the polariscopic readings.

It will be noted on comparing the polarization constants of the different honeys that while the invert readings at either 20° or 87° are subject to the widest variations, the difference between the polarizations at these two temperatures is a fairly constant quantity for nearly all the honeys. The range of this constant, which will be found of great value in honey analysis, is worth noting.

Range of difference in invert polarization of honeys between 20° and 87° C. Number of samples.

Degrees Ventzke.

In other words, 50 per cent of the honeys examined gave a constant ranging from 26 to 28; 80 per cent from 25 to 29, and nearly 95 per cent from 23 to 30. It is thus seen that we have a constant for honey which will be as valuable in analytical work as the Reichert-Meissl number is in the examination of edible fats. In a number of cases, as in honeys Nos. 40 and 89, the low polarization constant is due to an excess of water in the sample. It is therefore more accurate in doubtful cases to reduce the polarization difference to a uniform moisture content or to a constant basis of reducing sugars after inversion (see p. 60).

OPTICAL INACTIVITY.

An interesting fact in connection with the polarization of honeys is that nearly every honey under certain conditions is optically inactive. This may be produced by change of temperature, and, in fact, from the data given in the table, the temperature of optical inactivity for all the different honeys may be calculated. This inactivity to polarized light may also result in some cases from birotation. For example, honeys Nos. 63 and 92 directly after solution polarized to the right; in a short time the rotation had fallen exactly to zero, after which it became negative. Such honeys as these therefore have two points of inactivity—under birotation this point is only momentary, while under temperature influence it will continue constant as long as the temperature remains constant.

ANALYTICAL DATA.

MOISTURE.

The average moisture content of the honeys in the table was 17.59 per cent, with a range of from 12.42 to 26.88. Only two of the honeys, Nos. 40 and 89, which were unripe, exceeded 25 per cent of water, the limit allowed by the Committee on Food Standards and by the German "Vereinbarungen." The average water content of 138 pure honeys listed by König was 20.60 per cent, and of 17 honeys analyzed by Lehmann and Stadlinger, 19.30 per cent, results somewhat in excess of those of American honeys.

Shutt shows by numerous analyses that honey from uncapped or only partially capped combs is usually immature, containing a higher moisture content and having decidedly poorer keeping qualities than honey from fully capped combs. A practice among some bee keepers at the present day is to allow such unripe honey after

a Ripe and Unripe Honey. Experimental Farms Reports, 1902, p. 163. Ottawa, Canada.

a

straining to become reduced by evaporation in the sun or by low artificial heat to the consistency of ripe honey before bottling and marketing.

INVERT SUGAR.

The average percentage of invert sugar for the honeys examined was 74.44, with a range of from 62.23 to 83.36. König's average of 168 honeys was 73.13 per cent, and Lehmann and Stadlinger's average of 17 honeys, 73.45 per cent, results somewhat lower than the American honeys, as would be expected from the difference in water content.

LEVULOSE AND DEXTROSE.

The percentages of levulose and dextrose were not calculated for the individual honeys (except in a few particular cases) but only for the general averages. A considerable variation will be noted in the proportion of these two sugars, although the levulose both in floral and in honeydew honeys is always found in excess. Two exceptions to the latter rule are noted in the cases of the cotton honey No. 58 and the wild pennyroyal honey No. 55. In the case of the cotton honey, leakage of the liquid portion of the sample from the jar explains the deficiency in levulose; an irregularity in sampling may possibly explain the excessive abnormality of the pennyroyal honey. The non-crystallizable tupelo honeys were characterized by an abnormally high percentage of levulose, over 6 per cent greater than that observed for any other honey. The general average for ninety-two levorotatory honeys was 40.50 per cent of levulose and 34.02 per cent of dextrose. The average results of 83 honeys analyzed by Soxhlet and Sieben by the latter's method of destroying the levulose with sixth-normal hydrochloric acid and determining the residual dextrose, gave 38.65 per cent levulose and 34.48 per cent dextrose.

b

SUCROSE.

The average sucrose content of the honeys in the tables was 1.90 per cent, the samples varying from 0.00 to 10.01 per cent. Only two of the honeys exceeded 8 per cent sucrose, the limit fixed by the Standards Committee. The German "Vereinbarungen" allow 10 per cent of sucrose as the maximum for pure honey, and this figure is perhaps more liberal than that of the United States standards. The average sucrose content of the 138 honeys listed by König is 1.76 per cent with a range of 0.00 per cent to 12.91 per cent. In certain cases where bees have had access to molasses or sugar the

a Ripening Honey by Artificial Means. E. R. Root, A. B. C. of Bee Culture, 1903, p. 124.

Zts. Ver. d. Rübenzucker-Ind., 1884, p. 837.