from analytical data. The samples of cereals examined consisted of two samples of wheat, one of rye, one of unhulled oats, and one of hulled barley. The analytical data obtained by careful analyses follow:

The calculated calories for each of the samples given above

The calories found by direct combustion in oxygen were as

The direct comparison of the two numbers is seen in the

following table:

In the above data all the nitrogen has been computed as protein.

The agreement between the calculated calories and those actually determined in the bomb is satisfactory in the above cases with the exception of the sample of oats. In this case, as will be seen, the calories directly ascertained were 179 greater than those calculated from the analytical data. This fact suggests the possibility of the heat of combustion of the unseparated complex representing the unidentified carbohydrates of hulls and fodders being higher than for starch. Where such a difference exists the suggestion at once occurs that either the analytical data or the calories obtained by combustion are in error. One of the principal values of ascertaining the calories of combustion in analytical work is indicated by such a difference. The combustion will be a check on the analysis, and vice versa the analysis a check on the combustion. Where the differences are as great as noted, the indications are for a repetition of both the analysis and the combustion. The magnitude of the difference between the calculated and ascertained calories which can be allowed as fully within the ordinary errors of analysis and combustion can only be fixed by a long series of determinations, and perhaps after the factors employed for the calculations have been slightly changed to harmonize more closely with ascertained results. At the present time, we are inclined to the opinion that when the difference between the calculated calories and those ascertained on combustion does not exceed fifty to seventy-five calories, the check is sufficiently satisfactory.

We have applied the above process to the investigation of some of the cereal products; viz., flour, bread, breakfast and partially prepared foods and some miscellaneous products. The data which are given below include all of the determinations made on the products mentioned with the exception of six,

where the difference between the ascertained and calculated calories was so great as to indicate an error in one or the other. The following table contains all the data showing a comparison between the calculated and ascertained calorific power of the several substances mentioned:

BREAKFAST AND PARTIALLY PREPARED FOODS.-WHEAT PRODUCTS.

A study of the above data reveals the fact that while the variations in individual instances are considerable, a comparison

of the means shows that the factors which have been adopted must be very nearly correct, inasmuch as the mean calculated calories differ very little from those determined by actual combustion. In calculating the calories for wheat products both the factors 5.70 and 6.25, for converting nitrogen into proteid matter, have been used. Inasmuch as the calorific power of the protein is slightly greater than that of the carbohydrates, the total calories as calculated by the factor 6.25 are slightly greater than those calculated by the factor 5.70. In the case of breads and other baked products the differences are not so great as we had anticipated, on account of the difficulty of completely extracting the fat and oil from a bread. The individual differences, as in the case of a flour, are somewhat marked, but the means agree very closely.

While this paper was writing, in point of fact on August 2, we received Bulletin 35 of the Wyoming Station (June 1897) in which Prof. Slosson has called attention to work similar to ours which he has done at that station. The factors used by Prof. Slosson differ slightly from those which we have adopted, his factor for fat and oil being 9500, for protein 5700, and for carbohydrates 4200. In nineteen samples of wheat products the calculated calories from analytical data by our factors are 4472, and by Slosson's factors 4447. In six samples of miscellaneous cereal products, calculated by our factors, the number is 4833, and by his, 4810.

It is seen by the above that the mean results are not very greatly different from the two sets of factors, Slosson gaining in his higher factor for fat a portion of what he loses in his lower factor for protein as compared with our numbers.

In conclusion we beg to call attention to the fact that in the light of the data which have been presented in the above paper, we can with reason claim that the determination of the calorific power by combustion under pressure in oxygen is destined to be a valuable aid to the analyst in serving as a check upon the analytical data. We are further warranted in believing that whenever the calculated calories and the analytical data in hulled cereals and cereal products differ by as much as 100 from those obtained by combustion, the chemist will do well to repeat both the analysis and the combustion in order to discover the source

of error. The bomb calorimeter in this way becomes a valuable adjunct to the chemist in his work from a purely analytical point of view.

The data of heat determinations in bodies rich in cellulo-lignolignin matters such as bran, hay, straw, and fibers generally, are found to be regularly higher than when calculated from the data of analysis, and this leads us to the belief that these bodies exist in a state of greater molecular condensation than starch. If this be the case, the determination of the heat of combustion may prove helpful in approximately fixing the molecular weight of the polymers in any series insoluble in reagents without change. This subject we are now investigating, and it will form the theme of a subsequent communication.

PRELIMINARY NOTE ON SOME NEW DERIVATIVES OF

WHE

VANILLIN.

BY A. E. MENKE AND W. B. BENTLEY.

Received March 5, 1898.

HEN chlorine is allowed to act for some time on a solution of vanillin in chloroform, a white crystalline substance is formed, which after suitable purification and analysis was found to be chlorvanillin, melting-point 166°.

If chlorvanillin in small quantities be treated with sodium amalgam, chlorvanilloin can be obtained, melting-point 255°. We have not yet succeeded in obtaining chlorvanillin alcohol. Fusion of chlorvanillin with caustic potash seems to yield chlorprotocatechuic acid, melting-point about 235°.

The action of dilute nitric acid in suitable proportion upon. vanillin yields three products: a white substance containing nitrogen, almost insoluble in the usual solvents, melting-point about 300°; a yellow compound soluble in hot alcohol, almost insoluble in water, melting-point 178°-179°; and dinitroguiacol, melting-point 123°, soluble in water. The substance melting at 178° is either nitrovanillin or an addition compound of nitrovanillin and dinitroguiacol. By treatment with nitric acid it yields. dinitroguiacol, and by oxidation with permanganate forms nitrovanillic acid, melting-point 214°.

In an attempt to prepare chlorprotocatechuic acid, directly from protocatechuic acid, we found that passing chlorine in excess