was never so clearly marked at 87° as at 20° on account of slight striations produced in the heated liquid, hence readings could not be made with the same degree of accuracy as at the lower temperaNevertheless, consecutive readings could usually be obtained. agreeing within one or two-tenths of a degree.

INVERT POLARIZATION.

Fifty cubic centimeters of the solution for the direct polarization were made up to 55 cc in a small sugar flask with concentrated hydrochloric acid at 20° C. and allowed to remain at the laboratory temperature over night (18 to 20 hours). The solution was then polarized in a 200 mm silver-lined jacketed polarization tube, first at 20° and then at 87°, as in the case of direct polarization. The hydrochloric acid of the inverted solution had no perceptible action upon the silver lining of the tubes. In case silver-lined tubes are not available, the acid of the inverted solution should be neutralized to avoid its attacking the metal walls. The invert saccharimeter readings at 20° and 87° were increased by one-tenth to correct for dilution, and the values thus obtained were recorded.

CALCULATION OF LEVULOSE.

Inasmuch as the variations in polarization at different temperatures are due almost entirely to the change in specific rotation of levulose, it is possible to calculate with a fair degree of approximation the levulose content of any saccharine solution. Wiley's " optical method for estimating levulose is based upon this principle. In the method as described by Wiley it is shown that 1 gram of levulose in 100 cc shows a variation of 0.0357° V. for each degree centigrade; the variation for 67° would, therefore, be 2.3919. The difference in the direct polarization of a honey at 20° and 87° C. divided by 2.3919 will give, therefore, the grams of levulose in a normal weight of honey; from this the percentage of levulose may be obtained easily. The above method of calculation, however, only holds for solutions which have been made up at 20° and 87°. The polarizations at 87° in the table (page 23) were all made upon solutions prepared at 20°, so that a correction must be introduced for the dilution due to the expansion of the liquid. One hundred cubic centimeters of water at 20° will expand to 103.15 cc at 87°; the polarization at 87° must, therefore, be multiplied by 1.0315 to obtain the reading corrected for this expansion. The uncorrected readings at 87° are

a " It is always much the better plan to neutralize before polarizing the inverted solution at 87° C., on account of the danger of destroying part of the levulose. (See p. 44.)

Principles and Practice of Agricultural Analysis, 1897, 3 : 267.

given in the table and these are sufficient for purposes of comparison; if it is desired to calculate the percentage of, levulose, however, it is important that this correction be made.

CHEMICAL METHODS.

MOISTURE.

For the determination of moisture in honeys 2 grams of the sample were weighed out in a flat-bottom aluminum dish (2 inches in diameter) containing from 10 to 15 grams of fine quartz sand which had been thoroughly washed and ignited. A small glass stirring rod was weighed out with the dish and sand, and after the addition of the honey the latter was dissolved in 5 to 10 cc of distilled water and thoroughly incorporated with the sand by stirring with the rod. The dishes were then placed in a vacuum oven and dried at 65° to 70° C., under 20 to 24 inches vacuum, to constancy in weight. The average length of time required for drying the samples was about eighteen hours; with honeys of high purity, such as those of the alfalfa and clover type, twelve hours' drying or even less was sufficient, while with low-grade honeys of the honeydew class, which were high in dextrins and gums, thirty-six hours or longer were required to secure constancy in weight.

REDUCING SUGARS.

As dextrose.-Ten cubic centimeters of the solution used for the direct polarization (26 grams to 100 cc) were made up to 250 cc and the reducing sugars as dextrose in 25 cc of this solution determined by Allihn's method.

As invert sugar.-The dextrose by Allihn's table was converted into invert sugar by multiplying by the factor 1.044."

CALCULATION OF DEXTROSE.

The method of calculating the levulose content of honeys from the difference in polarization between 20° and 87° has already been given. The percentage of levulose subtracted from the total reducing sugar as invert will give very closely the percentage of dextrose if these two sugars are present in nearly equal amounts. If the two sugars differ widely in the percentages, an error is introduced into the calculation of invert sugar and hence into the percentage of dextrose. It is therefore more accurate to reduce the levulose to its

a Browne, Analysis of Sugar Mixtures, J. Amer. Chem. Soc., 1906, 28: 446. For the Dextrose Invert Sugar same weight of reduced copper the ratio = 0.958 and Invert Sugar 1.044.

Lippmann, Chemie der Zuckerarten, 1:898.

Dextrose

dextrose equivalent in copper-reducing power by multiplying by the factor 0.915. This subtracted from the total reducing sugars as dextrose will give the true percentage of dextrose.

SUCROSE.

On account of the various errors involved in the estimation of sucrose in honeys by the Clerget method of double polarization (such as the difference in specific rotation of levulose in neutral and acid solution), the sucrose was determined in the honeys by the gravimetric method. Ten cubic centimeters of the solution used for invert polarization were made up to 250 cc after neutralizing the free acid with sodium carbonate, and the invert sugar determined as before. The difference between the percentage of invert sugar as found before and after inversion multiplied by 0.95 gives the percentage of sucrose.

ASH.

Five grams of honey were carefully heated in a platinum dish until intumescence ceased and then ignited at low redness until a white ash was obtained.

With impure honeys of the honeydew class, which are usually high in ash content, it is sometimes necessary first to carbonize the honey and wash out the soluble salts with hot distilled water; this solution is then added to the ash from the leached char, and the whole evaporated and ignited at low redness as before.

DEXTRIN.

The determination of the dextrin bodies in honey was attempted by a number of different methods, but none of these was found to be reliable or altogether accurate. The methods based upon the removal of the sugars by fermentation and determination of the residual dextrins were found to be inaccurate on account of a partial fermentation of the honey dextrins. Methods based upon the increase in copper-reducing power after heating with hydrochloric acid gave too low results on account of a partial destruction of levulose. Methods based upon a precipitation of the dextrins by means of absolute alcohol and weighing of the same after drying were also found inaccurate on account of the occlusion of sugars in the dextrin precipitate. The method finally adopted in the present work was a modification of the alcohol precipitation method and was carried out as follows:

Eight grams of honey were transferred to a 100 cc flask with 4 cc of water and sufficient absolute alcohol to complete to the mark. A little care is required to effect the complete removal of the honey from

Browne, Analysis of Sugar Mixtures, J. Amer. Chem. Soc., 1906, 28: 446,

the weighing dish without using more than 4 cc of water. The transference is best made by decanting as much as possible of the liquefied honey into the flask, then adding 2 cc of water to the dish to take up any adhering honey and again decanting. By using 1 cc more of the water in two successive washings and adding a few cubic centimeters of the absolute alcohol each time before decanting, the honey can be completely transferred without the necessity of using more water than the 4 cc. Absolute alcohol is used finally to rinse out the dish and is then added to the flask with continual agitation until the volume is completed to 100 cc. After shaking thoroughly the flask is allowed to stand until the dextrin has settled out upon the sides and bottom of the flask and the supernatant liquid has become perfectly clear (usually in twenty-four hours). The clear solution is then decanted through a filter and the precipitated residue washed with 10 cc of cold 95 per cent alcohol to remove adhering liquid, the washings being also poured through the filter. The residue adhering to the flask and the particles which may have been caught upon the filter are dissolved in a little boiling distilled water and washed into a weighed platinum dish. The contents of the latter are then evaporated and dried in a water oven to constancy in weight. Should the amount of precipitate be considerable, it is necessary to dry upon sand in vacuo at 70°.

After determining the weight of the dried alcohol precipitate the latter is redissolved in water and made to a definite volume. The following dilutions were employed in making up the solutions: Dilutions of dried alcohol precipitate.

The sugars were then determined in aliquots from the filtered solution of alcohol precipitate both before and after inversion. The total precipitate less invert sugar and sucrose gives the per cent of dextrin.a

While this method of estimating dextrin in honeys gives much more accurate results than the direct weighing of the alcohol precipitate, it can not be said in any way to give the true dextrin content of the honey, although it is believed that the figures obtained are a close approximation. A small amount of dextrin always escapes

With honeydew honey, which gives a large amount of alcohol precipitate, it was found best to take only 4 grams of honey for analysis; in other respects the method of procedure is the same.

precipitation with alcohol; furthermore, no account is taken of the other ingredients which may be occluded in the alcohol precipitate other than the sugars, and no correction is made for the copperreducing power of the honey dextrin itself. This latter factor, though apparently very small, might prove to be of some importance if much dextrin were present. Notwithstanding these limitations, however, the percentage of dextrin as determined by the method described has been found to have a decided value, especially when it is wished to compare honeys of different origins.

UNDETERMINED MATTER.

The undetermined matter of the honey (wax particles, pollen grains, higher alcohols, tannin, essential oils, combined acid, etc.) was estimated by difference.

FREE ACID.

The free acid in the honey, expressed as formic acid, was determined by titrating a weighed portion of the sample, after solution in distilled water, with tenth-normal sodium hydroxid, using phenolphthalein as indicator. It is customary to express the free acid in honey as formic, although other acids may, no doubt, be present. In soured or fermented honeys acetic acid is always formed.

TABULATION OF THE ANALYTICAL RESULTS.

In the tabulation of the analytical results the honeys were classified according to the floral origin. Ninety of the samples examined were labeled with the name of one definite flower, and these have been grouped together according to their botanical classification. Such a grouping has in some ways only a nominal value, since none of the honeys, as was shown by the microscopic examination of the pollen grains, was derived exclusively from the nectar of one kind of flower. In several cases the admixture of other kinds of nectars was indicated by the aroma. The botanical classification may be said, therefore, to represent simply the predominating kind of nectar gathered by the bees. Even with this limitation it will be noted that the chemical and physical constants of each class of honeys show a certain conformity to type.

The remaining ten honeys were labeled as being derived from mixed nectars, although several of these samples might be placed in the previous class with as much correctness as many of those marked unmixed. For general convenience it seemed more practicable to retain the nomenclature indicated by the bee keepers rather than to attempt a reclassification of the honeys based upon microscopic evidence as to the predominating kind of pollen.