F. PENNY: 35.454 (0 = 16).

Six experiments on the conversion of silver into nitrate gave 100 Ag= 157.441 nitrate; extreme difference, 0.028. Twelve experiments by three different methods on the conversion of silver into chloride gave 100 Ag = 132.837 chloride. Four series of experiments on the interconversion of potassic chloride, chlorate and nitrate gave for the difference between the molecular weights of the chloride and the nitrate 26.56. Corresponding experiments with sodium salts gave the same difference 26.568. The mean combined with the data for the silver salts gives the molecular weight of argentic chloride at 143.424, and Cl = 35.454. For further details see Penny's determinations of potassium, sodium, nitrogen and silver. The weighings were calculated for vacuum. (Phil. Trans., 129, 1839, 32.*)

R. PHILLIPS: 35.688 (0 = 16).

In order to avoid the error possibly incurred by the melting of argentic chloride, etc., Phillips mixed known and nearly equivalent quantities of silver dissolved in nitric acid, or of crystallized argentic nitrate, with ammonium chloride; filtered, washed, and precipitated the comparatively minute amount of chlorine in the filtrate with silver solution. The fusion of this small quantity could cause no loss of importance. Phillips confesses that his ammonium chloride was acid and the only conclusions he draws are that Cl = 36, N=14, 08 and H1 may be taken without considerable error if silver is 108. [The method seems to have been original and is nearly that afterwards adopted by Pelouze. The acidity of the ammonium chloride would of course give Cl too high.] (Phil. Trans., 129, 1839, 35.)

C. MARIGNAC: 36.001 (0 = 16); 225.007 (O= 100). Determined by passing chlorhydric acid gas over hot cupric oxide and condensing the water formed. The mean of three experiments was Cl = 450.013; the extreme difference is 0.2 for O= 100. The gas was made from recrys

*This is one of the most elegant investigations of the kind to be found in chemical literature, though it scarcely receives a mention except from Stas, who accords to it the praise it deserves. Stas' wonderfully exhaustive researches were necessary to prove beyond question that chemistry has a mathematical basis, and that the atomic weights of the elements are incommensurate. Penny's investigation, taken in connection with Stas', shows that the highest degree of accuracy is not incompatible with the simplest means when they are applied with the care and acumen, without which exact results cannot, under any circumstances, be obtained.

tallized sea-salt and concentrated sulphuric acid and was dried by passing through nine tubes filled with sulphuric acid and pummice stone and with calcium chloride. The water was collected in a condenser to which drying tubes were appended. (Paris Comptes Rendus, 14, 1842, 570.)

=

100).

A. LAURENT: 35.468 (0 = 16); 221.672 (0 = Determined by three analyses of chloronaphthalintetrachloride, which he found to contain 58.22; 58.29; 58.28; per cent. Cl. The mean is 58.27 from which the value follows. (Paris Comptes Rendus, 14, 1842, 456.)

According to Maumené, Laurent confessed that his salt was impure, containing chlorose compounds, in Gerhardt's Comptes Rendus, 1845, 108. (Annal. de Chimie et de Physique, (3,) 18, 1846, 45.)

C. MARIGNAC: 35.37 (0=16); 221.07 (O = 100). One synthesis of argentic chloride showed that 100 silver equals 32.74 chlorine. Berzelius had found 32.75, which Marignac adopts. Marignac found by six experiments on the decomposition of potassic chlorate by heat, that the molecular weight of potassic chloride was 932.14. He tested the equivalence of potassic and argentic chlorides by precipitating the former with argentic nitrate, filtering without the use of paper through a funnel with a capillary neck. The precipitate was dried and weighed, then melted and reweighed, no loss being observable. 100 potassium chloride gave 192.33 and 192.34 argentic chloride in two experiments, or reduced to vacuum, 192.26. Hence the atomic weight is 442.13. The potassic chloride was prepared by heating chlorate which had been purified by repeated recrystallizations. (Liebig's Annal., 44, 1842, 23.)

C. MARIGNAC: 35.456 (016); 221.6 (0 = 100). In accordance with Pelouze's suggestion, Marignac repeated his determination of the composition of argentic chloride and of the equivalence of potassic and argentic chlorides, retaining the molecular weight of potassic chloride mentioned in the last paragraph. That value was obtained from the mean of six experiments on the decomposition of the chlorate which gave the percentage of oxygen at from 39.155 to 39.167; mean 39.161. Pelouze had got, as the mean of three experiments, 39.157. (Paris Comptes Rendus, 15, 1842, 959.) Marignac made eleven experiments on the equivalence of silver and potassium chloride by Pelouze's

method, a known weight of silver being dissolved in nitric acid and added to a known and nearly equivalent amount of potassic chloride in solution, after which the excess was titrated with decimal standard solution. 100 parts of silver were precipitated by from 69.049 to 69.067, in mean by 69.062 chloride. 100 parts of chloride were precipitated by from 192.33 to 192.37, in mean by 192.348 silver. Five experiments were made on the composition of argentic chloride by dissolving silver in nitric acid, with precautions against loss by spirting, precipitation with chlorhydric acid, washing, drying, melting and weighing in the same vessel. 100 parts of silver gave from 132.825 to 132.844 chloride, mean 132.84. Calculation from these data gives in vacuo Ag=1349.01; K = 488.94; Cl=443.20; for 0 = 100 [or Ag=107.921; K = 39.115; Cl = 35.456, for 016.] (Berzelius' Jahresbericht, 24, 1844, 58; Bibl. Univ., 46, 1843, 350.)

C. GERHARDT: 36 (0=16).

By heating potassic chlorate in a current of oxygen Gerhardt got, when he took precautions against loss by spirting, a mean of 60.949 chloride, from which he deduces 36 for chlorine without giving further data. (Paris Comptes Rendus, 21, 1845, 1280.) Marignac shows that no data have ever been published which, in connection with Gerhardt's experiments, would give this value for chlorine. He adds further experiments of his own which, without aiming to establish more exactly the true atomic weight, prove it less than 36 (Liebig's Annal., 59, 1846, 284; Bibl. Univ., Arch. des Sciences, 1, 1846.)

E. J. MAUMENÉ: 35.462 (O=16).

Maumené made seven analyses of argentic chloride by reduction in a current of pure hydrogen. Five of these experiments were made with quantities less than 10 grammes, and gave a mean of 100 silver = 32.736 Cl. Two experiments were made with about 30 grammes each, and gave 100 silver equal to 32.86 and 32.853 chlorine. Maumené prefers the latter, and deduces from them for chlorine the value 443.67 or 35.494 taking silver according to his own experiments at 1350.32. [If silver is taken at 107.93 (Stas) the same analyses give 35.462.] (Annal. de Chimie et de Physique. (3,) 18, 1846, 41.)

A. LAURENT: 35.5 (O=16); 221.88 (0 = 100).

A single experiment was made as follows: pure silver was weighed off and placed in a matrass, nitric and chlorhydric acids were added, the liquid was evaporated and the chloride melted. An empty test was carried on at the same time to act as tare. Silver was taken at 1350. (Paris Comptes Rendus, 20, 1849, 5.)

J. DUMAS: 35.5 (0 = 16).

Determined by chloridizing different weights of pure silver by heating the metal in a current of chlorine. Experiments on 10 grammes and 20 grammes gave a mean of 35.5055, the difference being 0.013, for chlorine, if silver is 108. (Annal. de Chimie et de Physique, (3,) 55, 1859, 135.)

J. S. STAS: 35.457 (0 = 16).

Stas found the atomic weight of chlorine by three independent methods:

(1.) From analysis of argentic chlorate and synthesis of argentic chloride. A known weight of the chlorate was dissolved in water, precipitated with sulphuric acid to secure advantageous division of the salt, and reduced while in suspension by a slow stream of sulphurous anhydride. The chloride was washed, dried, and weighed in the flask in which it was produced. The minute amount of chloride present in the chlorate was collected and taken into consideration, and the wash-water was carefully examined for silver. Two analyses (of about 140 and 260 grammes) gave for the molecular weight of the chloride 143.383 and 143.407, mean 143.395. A variety of syntheses of argentic chloride in the wet and in the dry way showed that 100 parts silver combined with nearly 32.850 parts chlorine. Stas assumes that none of his syntheses can possibly have given too much chloride and accepts the relation stated. These data give = 35.458.

(2.) From the mutual relations of potassic chlorate and chloride and argentic chloride, combined with the composition of the last. The chlorate was decomposed either by gentle heat or in the wet way with chlorhydric acid. 100 parts of chlorate were found to contain 60.846 parts chloride as the mean of eight experiments; extreme difference, 0.012, which gives the molecular weight of potassic chloride at 74.59. The relation between potassic and argentic chloride was ascertained by Pelouze's method, (see Marignac's

determination above.) Twenty experiments on quantities of 32 grammes, and less, of silver gave 100 parts Ag= · 69.103 parts KCl; extreme difference, 0.008. These data combined with the composition of argentic chloride given above, indicate for chlorine 35.460.

(3.) The composition of argentic nitrate was determined, and the difference between the atomic weights of nitrogen and chlorine. In two experiments silver was dissolved in nitric acid, the solution evaporated to dryness, and the nitrate kept melted until there was no further loss of weight. The result obtained was that 100 silver = 157.484 nitrate; difference, 0.008. From series of experiments on the relation of the chlorides of potassium, sodium, lithium and silver to the nitrates, Stas found the difference between a chloride and a nitrate from 26.586 to 26.591; mean 26.588. These data show that the atomic weight of chlorine lies. between 35.455 and 35.460, and confirm the mean of all the determinations of Penny, Marignac, and Stas, 35.457. The silver for this investigation was either distilled or compared with distilled silver; it was found impossible to reduce the amount of silica in the alkaline salts below 0.002 of one per cent., it was therefore determined and allowed for; every possible method of purification by recrystallization and otherwise was resorted to to ensure purity. The weighings are all reduced to vacuum. (Stas, Unters. über Chem. Proport., Leipzig, 1867.)

CHROMIUM.

The specific heat of chromium, as determined from that of the oxide by Kopp, Regnault, and Neumann, corresponds to an atomic heat of from 5.4 to 5.98, if the atomic weight is taken at 52.4. (Gmelin-Kraut, l. c.)

=

= 100).

J. J. BERZELIUS: 56.29 (0 = 16); 351.819 (O 100 parts of plumbic nitrate, on precipitation with potassic chromate, gave 98.772 parts plumbic chromate. The value follows for Pb 1294.498, and N = 88.518. (Poggend. Annal., 8, 1826, 22.)

T. THOMSON: 64 (016); 400 (O

=

100).

3.14 grains of metallic chromium, converted into chromic acid by heating with potash and nitre, gave a precipitate of 16.23 grains plumbic chromate. (Phil. Trans., 117, 1827,