Nitroglycerine Apparatus, etc." We have Mowbray's book before us but, although he gives an illustration of his apparatus opposite page 68, we do not find in this or any edition of Mowbray's work the cuts Eissler gives. These cuts occur in Hill's Notes on Explosives, and we have always understood that Hill devised this special form of apparatus though it was based on Mowbray's principle. Eissler repeats the statement that Sobrero discovered nitroglycerine in Paris, though Guttman has some time since announced that "It was expressly stated to me by the late Mr. Sobrero that he made his invention in Turin, where he was professor, and not in the laboratory of Pelouze,' and that a sample of the original nitroglycerine is still kept at Avigliana. Throughout the book there is a carelessness in the use of scientific and especially chemical terms, sulphate of potassa and potassium nitrate, for instance, occurring in the same paragraph, while in the description of the methods of analyzing acids such typographical errors as 180 BaSO, = 42.06 H,SO, are committed repeatedly.

The book contains much information and is useful, but it would have been a better book if the collated matter had been digested. CHARLES E. MUNROE.

BOOKS RECEIVED.

Text-book of Physical Chemistry. By Clarence L. Speyers, Associate Professor of Chemistry, Rutgers College. 1897. New York: D. Van Nostrand Co. vii+224 pp. Price $2.25.

A Short Handbook of Oil Analysis. By Augustus H. Gill, S.B., Ph.D. Philadelphia: J. B. Lippincott Co. 1898. Cloth. 139 pp. Price $1.50. Introduction to Electrochemical Experiments. By Dr. Felix Oettel. Translated by Edgar F. Smith. Philadelphia: P. Blakiston, Son & Co. 1897. 144 pp. Price 75 cents.

Practical Exercises in Electrochemistry. By Dr. Felix Oettel. Translated by Edgar F. Smith. Philadelphia: P. Blakiston, Son & Co. 1897. 92 pp. Price 75 cents.

Electric Treatment in Gout and the Uric-acid Diathesis. By Robert Newman. Reprinted from the Medical Record, Dec. 11, 1897. New York: The Publishers' Printing Co.

16 pp.

ERRATUM.

In the last (January) issue of the Journal, p. 29, line 11, the author's name read "H. D. Campbell" instead of “E. D. Campbell."

for

THE JOURNAL

OF THE

AMERICAN CHEMICAL SOCIETY.

FIFTH ANNUAL REPORT OF THE COMMITTEE ON ATOMIC RESULTS PUBLISHED DURING 1897.

WEIGHTS.

BY F. W. CLARKE.

Received January 27, 1898.

To the Members of the American Chemical Society:

During 1897 comparatively few new determinations of atomic weights have appeared. The data given are as follows:

Carbon.-Scott' has carefully studied the determinations of the atomic weight of carbon which depend upon the combustion of the element and the weight of the dioxide formed, and has discovered an important correction. This is due to the considerable change of volume in the potash solution produced by the absorption of carbon dioxide, which involves a notable change in the reduction of the weighings to a vacuum. Scott has measured the amount of this change, and has applied it to the work of his predecessors with the subjoined results. In each case the values given are the means of the several experiments recorded.

These values are based upon O= 16; with 0 15.879, the corrected mean result becomes C = 11.910.

Scott further criticises in detail the determinations by Stas which depend upon the combustion of carbon monoxide, and shows that they require possible corrections and involve various uncertainties. He concludes "that the atomic weight of carbon is by no means one of those most accurately known, and that a careful revision of it is imperative."

Nitrogen, Chlorine, and Silver.-Leduc', on the basis of his density determinations, assumes N 14.005 when 0 = 16. Applying this to the determinations by Stas of the ratios Ag: C1, AgC1: 0,, and AgCl : NH,C1, he finds a discrepancy of about one part in 320. That is, the work of Stas gives N = 14.044 instead of 14.005. This disagreement Leduc attributes to the presence of occluded oxygen, as pointed out by Dumas, in the silver used by Stas. For this he assumes a correction, and then, combining his value for nitrogen with the data given by Stas, he computes H = 1.0076, Cl = 35.470, and Ag 107.916. The correction applied by Leduc seems to be somewhat uncertain; and it is doubtful whether his determination of the density of nitrogen is entitled to greater weight than the other data which are involved in the calculations.

=

Aluminum. In the third and fourth reports of this committee Thomsen's data for the atomic weight of oxygen are given. In these determinations oxygen and hydrogen were compared with aluminum as an intermediary; but the aluminum used was not absolutely pure. Had it been pure, the atomic weight of the metal, with reference to the two gases, could have been computed.

Thomsen now gives the data necessary for the comparison, and deduces the atomic weight of aluminum. The impurities are determined, and corrections for them applied. There is also a correction for the change in volume of the potash solution in which the metal was dissolved. Applying the corrections, Thomsen finds that 0.99897 gram pure aluminum correspond to 0.11195 gram of hydrogen. Hence, with H = 1,

Al = 26.770.

1 Compt. rend., 125, 299, August 2, 1897.

2 Ztschr. anorg. Chem., 15, 447.

Again, 0.99897 gram of metal is equivalent to 0.88824 gram of oxygen. Hence, with O= 16,

These two values, referring to the two standard units, are determined independently of each other, and are independent of any measured ratio between O and H. They are notably lower than the values computed by myself from Mallet's data, 26.91 and 27.11, and possibly should supplant the latter. This question, however, needs further discussion, and probably some additional experiments.

Nickel.-Atomic weight determined by Richards and Cushman,' from analyses of the anhydrous, sublimed bromide, NiBr.. This substance is shown by the authors to be perfectly suited to the purpose of the investigation. The method of analysis was in all essential respects that used by Richards in other researches of the same kind, and was partly gravimetric and partly volumetric. On the one hand, the ratio 2AgBr : NiBr, was determined; on the other, the ratio 2Ag: NiBr,. All weighings were reduced to a vacuum. The results, in three series, were

From all these data the authors conclude that the atomic weight of nickel cannot be far from 58.69 when O = 16, or 58.25 if 0=15.88.

From series two and three 15.32086 grams of silver give 26.67078 grams of bromide. Hence the percentage of silver in silver bromide is 57.444, which agrees with the value 57.445 found by Stas. This is a good check upon the accuracy of the work.

Cobalt. The atomic weight determinations by Richards and Baxter' are in all essential respects like those of nickel which have just been cited. The three series of data are as follows: