ments, made with especial care, on the reduction of purpureo-cobaltic chloride by hydrogen gave 59.09. (Reported by Gibbs. Berlin, Bericht der Chem. Ges., 4, 1871, 789.)

COPPER.

Regnault, Kopp, and others have determined the specific heat of copper. It corresponds to an atomic heat of about 6 if the atomic weight is taken at 63.3. (Gmelin-Kraut, l. c.)

R. CHENEVIX: F. H. WOLLASTON: 64 (0 = 16); 400 (0 = 100.)

Chenevix found 20 parts of oxygen equivalent to 100 parts of copper, whence Wollaston deduces the atomic weight. (Phil. Trans., 104, 1814, 21.)

J. J. BERZELIUS: 63.296 (016); 395.6 (0-100). Determined by two experiments on the reduction of cupric oxide with hydrogen, which gave 395.695 and 395.507. The water was not weighed. (Poggend. Annal., 8, 1826, 182; and Lehrbuch, 3, 1216.)

ERDMANN and MARCHAND: 63.456 (0 = 16); 396.6 (0 = 100.)

Determined by four experiments on the reduction of large quantities of cupric oxide in a current of hydrogen. The hydrogen was displaced by air after the completion of the reduction. The weight of the oxide and of the copper were reduced to vacuum, but not that of the weights employed. To obtain pure cupric oxide, pure vitrol was prepared and electrolytically decomposed. The copper thus obtained was dissolved in nitric acid, and the nitrate decomposed by heat. The value is the mean; the extreme difference is 0.056 for O= 8, or 0.112 for 016. (Erdm. Journ. für Prak. Chem., 31, 1844, 389.)

Berzelius points out that these analyses vary among themselves much more than his own. He makes the difference somewhat greater than it really is by neglecting the reduction to vacuum. (Ibid., 37, 1846, 72.)

Hampe shows that these analyses, correctly calculated, give Cu 63.46. (Zeitschr. für Berg Hütten-und-Sal- Wesen im Preus. St., 21, 1873, 261.)

=

J. DUMAS: 63.5 (016).

Dumas says that experiments on the reduction of cupric oxide and on the sulphidation of copper have shown him that the atomic weight of copper lies between 31.5 and 32, near 31.75, but that his experiments cannot be regarded as decisive. (Annal. de Chimie et de Physique, (3,) 55, 1859, 129.)

MILLON and COMMAILLE: 63.128 (0 = 16); 394.55 (0 = 100).

These (three) experiments were in most respects a repetition of Erdmann and Marchand's. The value is the mean; the extreme difference is 0.49 for O = 100, or 0.0784 for O 16. The sulphate was prepared free from iron or zinc by dissolving copper in ammoniacal sulphate or nitrate. The oxide was obtained by heating the nitrate. (Paris Comptes Rendus, 56, 1863, 1249; and 57, 1863, 145.)

=

Fresenius sees no reason for preferring this to Erdmann and Marchand's value. (Fresenius' Zeitschr. für Anal. Chem., 2, 1863, 474.)

W. HAMPE: 63.3296 (0 = 16).

=

In three experiments cupric oxide was reduced in a current of hydrogen with all possible precautions. The hydrogen was displaced by air before weighing, though it was shown by experiment that porous copper does not condense hydrogen. The metal was heated till incipient melting was observed. The reduction and melting were repeated without altering the weight. Hampe attempted to control his results by reconverting the metal into oxide, but was unable to effect complete oxidation. The water produced by the reduction was found to be perfectly pure. The mean result was Cu 31.6696, maximum, 31.6729, minimum, 31.6648. The oxide was prepared from metallic copper. To obtain pure metallic copper, sulphate free from bismuth was electrolytically decomposed, the finely divided metal well washed, then melted, first in a current of carbon di-oxide, afterwards in hydrogen, and then again in carbon di-oxide. From the metal, basic nitrate was formed and from this salt, by heating first in air and then in oxygen, oxide. In two experiments the atomic weight of copper was determined by decomposing cupric sulphate by electrolysis, and weighing the metal. The residual fluid was evaporated, and a minute amount of copper, which had escaped decomposition, was

=

recovered and determined as sulphide. For S = 16.037 and 08, these experiments gave Cu 31.6577 and 31.66. The value taken is the mean of the two series. All weighings were reduced to vacuum. (Zeitschr. für Berg Hüttenund Sal.- Wesen im Preus. St., 21, 1873, 260.)

DIDYMIUM.

W. F. Hillebrand found the specific heat of this metal 0.04563, which corresponds to an atomic heat of 6.60 for an atomic weight of 144.78. (Poggend. Annal., 158, 1876, 78.)

C. MARIGNAC: 148.8 (0 = 16); 930 (0 = 100).

Determined by decomposing disulphate with barium chloride. Assuming the lower oxide as a prot-oxide, he calculated the atomic weight at 620. As Marignac was not confident of the purity of his salt, and subsequently became certain that the method was untrustworthy, details are unnecessary. (Liebig's Annal., 71, 1849, 313.)

C. MARIGNAC: 143.81 (0 = 16); 898.8 (0 = 100). Five experiments were made on the sulphate by decomposition with ammonium oxalate. The didymium oxalate was heated to redness, and the resulting oxide weighed. On the assumption that the oxide was protoxide, these determinations gave a mean of 598.2 for Di, with an extreme difference of 2.5. Three experiments were made on the chloride, the insoluble oxychloride, which is unavoidable in drying the salt, being separated. The chlorine was determined with silver, and the Di as in the previous experiments. These determinations gave Di at 600.2, with an extreme difference of 5.2 for Cl = 443.2 and S = 200. The salts were prepared from cerite. The cerium was extracted by treatment at first with dilute and afterwards with concentrated nitric acid. The sulphates of Di and La were separated by partial precipitation with oxalic acid and by partial recrystallization. (Annal. de Chimie et de Phys., (3,) 38, 1853, 148.)

=

16); 890.25 (0

=

100).

R. HERMANN: 142.44 (0 In one experiment sulphate which had been heated to a low red heat, was dissolved, decomposed with ammonium

=

=

oxalate, the precipitate incinerated and the oxide weighed. The result was Di 594.46, on the prot-oxide hypothesis, for S 200. In one experiment the chloride was decomposed with argentic nitrate, oxychloride being filtered off and allowed for, and the argentic chloride weighed. This experiment gave Di= 592.54 for Cl 443.2. For the preparation of the salt see Lanthanium. (Erdmann's Journ. für Prak. Chem., 82, 1861, 387.)

=

In five experiments the sulphate was exposed to a white heat until the weight became constant and the oxide on being tested showed no traces of sulphur. The results varied from Di = 46.585 to 48.08, probably, Zschiesche thinks, on account of the presence of La. S16. Di was separated from La by the partial precipitation of the nitrates with oxalic acid, the first portion falling being redissolved, and the partial precipitation repeated twenty times. (Erdmann's Journ. für Prak. Chem., 107, 1869, 74.

C. ERK: 142.695 (0 = 16).

The sulphate was decomposed with ammonium oxalate, the oxalate incinerated and the oxide weighed. The sulphuric acid was also precipitated as barium salt, and weighed. Three experiments gave a mean of Di = 95.13, on the prot-oxide hypothesis, with an extreme difference of 0.78. The Di salt was found to contain yttrium which was removed by repeated fractional precipitation with sodium sulphate. This re-agent precipitates a double salt of Di and sodium. The purification was continued until the atomic weight became constant. (Kopp's Jahresbericht, 1870, 319, Jena'sche Zeitschr, für Med. und Nat., 6, 299.)

Casselmann thinks that the salt may still have retained yttrium, and Fresenius objects to the barium sulphate determination on the well-known grounds. (Fresenius' Zeitschr, 10, 510.)

D. MENDELEJEFF: 138 (016).

From the analogy between Di and cerium and other elements, and from the fact that it forms two oxides, Mendelejeff believes that its lower oxide is a sesqui-oxide, and its atomic weight 138. Mendelejeff points out that an error is to be apprehended in the received values from the fact that we have no guarantee of the pureness of Di salts except recrystallization. (Liebig's Annal. Suppl. 8, 1871,

=

16).

P. T. CLEVE: 147.01 (O Determined by the conversion of didymium oxide into sulphate. The number is the mean of six experiments; extreme difference 0.58. The Di was separated from lanthanium by repeated precipitations of basic nitrate from nitric acid solution, conversion into formate and decomposition of this salt by heat. (Kopp's Jahresbericht, 1874, 259. Bulletin Soc. Chimique, (2,) 21, 246.)

W. F. HILLEBRAND: 144.78 (0 = 16).

Determined by one experiment on the conversion of metallic Di into nitrate, and then, by heat, into oxide. The impurities were determined. The metal was reduced electrolytically from the chloride. (Poggend. Annal., 158, 1876, 78.)

ERBIUM.

The physical and chemical analogies of the salts of this element have led Mendelejeff (Liebig's Annal., Suppl. 8, 1871, 195,) and P. T. Cleve (Kopp's Jahresbericht, 1874, 260; Bulletin Soc. Chimique, (2,) 21, 344,) to regard it as triatomic, and its atomic weight as about 170.

M. DELAFONTAINE: 113.04 (0 = 16).

M. Delafontaine investigated gadolinite by Mosander's method, and obtained besides yttrium, two substances which he regarded as erbium and terbium. From the sulphates, in which he supposed the metals to exist as protoxides, he determined erbium at 496 and terbium at 471 for O 100. Popp (Liebig's Annalen, 131, 189,) and Bunsen and Bahr (Ibid, 137, 1,) have shown that Mosander's method gives only mixtures. Delafontaine's terbium is thought to have been chiefly the erbium of other chemists. (Liebig's Annal., 134, 1865, 108.)

BAHR and BUNSEN: 168.9 (0 = 16).

A known weight of erbium oxide was treated with a very slightly excessive quantity of sulphuric acid; the solution evaporated and the excess of acid driven off at as low a temperature as possible. The increase of weight indicates 112.6 for S 32. The oxide was prepared from gadoli

=