decomposed by muriatic acid, the slight precipitate of hydrosulphuret produced was added to the first portion, and the whole was well washed, dried, and heated to redness in a glass tube; the sulphuret of bismuth thus obtained, fused into one mass, weighed 44.7 grains. I had previously ascertained the proportion of metal in this sulphuret, and found it to be 81.8 per cent. 44-7 grains of sulphuret, or 55 grains of the butter, must therefore contain 36.5 grains of bismuth; and hence, 100 of bismuth appear to consist of

33.6 chlorine

66.4 bismuth

100.0

The butter of bismuth may be called bismuthane. Among the preceding combinations of the metals and chlorine, there is a surprising difference in respect to volatility and fusibility. Iron and manganese, two difficultly fusible metals, form with chlorine readily fusible compounds, and a combination of the former metal and chlorine is even volatile; the compounds of tin and chlorine, and of chlorine and antimony, are very volatile substances, though the metals themselves are fixed at very high temperatures; on the contrary, the combinations of chlorine with bismuth, zinc, and lead, do not exceed in fusibility; indeed are not quite so fusible as the metals themselves. I can offer no explanation of these phe

nomena,

Another singularity attending the liquid fuming compounds of chlorine, such as the liquor of Libavius, the fuming liquor of arsenic, and the oxymuriats of sulphur and phosphorus, is, that they do not become solid at low temperatures. I have

reduced, by means of a mixture of snow and muriat of lime, the temperature of all these substances 20 degrees below the zero of FAHRENHEIT's thermometer, but without affecting their liquidity.

The influence of atmospheric air on the compounds of the metals and chlorine at high temperatures is curious, and worthy of particular attention. The combinations of chlorine with lead, zinc, copper, and bismuth, appear to be volatile in open vessels, and fixed in closed ones. How moist air operates in these instances, it is difficult to say. In other cases, where it evidently acts chemically, the changes explain themselves; thus, when the compounds of iron and chlorine and of manganese and chlorine are heated in the open air, hygrometrical water of the atmosphere seems to be decomposed, as muriatic acid fumes are produced, and oxides of the metals formed. Probably the volatility of the other compounds is connected with similar circumstances. This action of moist air has hitherto been much neglected; it is certainly worthy of being more fully inquired into, both in a theoretical and practical point of view. Its importance in practice is exemplified in the reduction of horn silver, and in the formation of several of the compounds of chlorine and the metals; if moist air be admitted in these operations, the silver will be lost, and the compounds not formed.

Guided by analogy, I have been led to try whether the muriat of magnesia, which is readily decomposed by heat in the open air, would not, when the air was excluded, by introducing it into a glass tube with a very small orifice, afford a permanent compound. The result was agreeable to my expectations; I obtained, by strongly heating the muriat for a

quarter of an hour, a substance like enamel in appearance, being semi-fused, and which appeared to be a mixture of magnesia and the true compound of magnesium and chlorine, for heated with water magnesia was separated, and a muriat of magnesia formed.

5. On the Relation between the Proportion of Oxygene and Chlorine in Combination with several Metals.

Errors being very common in chemical analyses, even in those conducted most skilfully and carefully, all possible means should be taken to discover them; and no means, I think, promise to be more effectual for this purpose, than the general analogy of definite proportions. From a great variety of facts, it appears that oxygene and chlorine combine with bodies in the ratio of 7.5 to 33.6. With 1 part by weight of hydrogene, for example, 7.5 of oxygene unite to form water, and 33.6 of chlorine unite with the same proportion to produce muriatic acid gas. To judge therefore of the accuracy of the analyses of the preceding combinations of the metals and chlorine, it is only necessary to compare them with the analyses of the oxides of the same metals. If the two agree, there will be reason to consider them both correct, but should they disagree, there is equal reason for supposing one or both of them to be wrong.

Thus, as the orange oxide of copper is analogous to cuprane and the brown oxide to cupranea, the oxygene and chlorine should be to each other in these compounds as 7.5 to 33.6. And from comparison of my analysis, with those of Mr. CHENEVIX and M. PROUST, it appears, that in the two first, copper being as 60, the oxygene is to the chlorine as 7.79, instead of 7.5 to

33.77, instead of 33.6; and in the two last as 7.5 to 33.6, or as 15 to 67.2. Coincidences as near as might be reasonably expected.

There is not the same agreement between M. PROUST'S analyses of the oxides of tin and the preceding ones of the combinations of this metal and chlorine. This discordance induced me to repeat my analyses, and obtaining the same result as at first, I directed my attention to the oxides of tin, and made the following experiments to ascertain the proportion of their constituent parts.

42.5 grains of tin, which had been precipitated from the muriat of this metal by zinc, were heated with nitric acid in a platina crucible, and slowly converted into peroxide; the acid and water were driven off by gentle evaporation at first, and afterwards by a strong red heat continued for a quarter of an hour. The peroxide thus produced was of a light yellow colour, and being very gradually dried, it was semi-transparent, and hard enough to scratch glass; it weighed 54.25 grains. Hence, as 42.5 grains of tin acquire, on conversion into peroxide, 11.75 grains of oxygene, this oxide appears to contain 21.66 per cent., of oxygene, just the quantity found in the native oxide by KLAPROTH, instead of 28, the proportion stated by PROUST.

M. BERTHOLLET, jun. has shewn that M. PROUST's estimate of 20 per cent. of oxygene in the protoxide is incorrect. To ascertain the true proportion, 20 grains of tin were dissolved in strong muriatic acid in a retort connected with a pneumatic apparatus, and without the assistance of heat; 16 cubic inches of hydrogene gas were produced. (Barom. 30, thermom. 60) as the production of this quantity of hydrogene indicates an

absorption of oxygene by the tin equivalent to 8 cubic inches, or (as 100 cubic inches weigh 34.2 grains) to 2.736 grains, the protoxide of tin appears to contain 11.99 per cent. of oxygene.

These analyses of the oxides, compared with those of the combinations of tin and chlorine, are found very nearly to agree. The ratio of oxygene to chlorine in the two first similar compounds, the tin being as 55, is as 7.5 to 33.4; and in the two last, viz. the peroxide and the liquor of Libavius, as 7.6 to 33.5, or as 15.2 to 67.

As the black oxide of iron is formed by the decomposition of ferrane by a solution of potash, and the red oxide by that of ferranea, it is evident that these oxides and combinations of iron and chlorine should coincide in the proportions of their constituent parts. This appears from the analyses* of Dr. THOMPSON to be nearly the case, for iron being as 29.5, the oxygene is to the chlorine in the black oxide and ferrane as 8 instead of 7.5 to 33.6; and in the two others as 8 to 33.6, or as 13.2 to 55.5. Here the agreement is less than in other instances; but this is not surprising considering the different estimates of the proportions of oxygene in the oxides of iron, and the difficulty of ascertaining them correctly.,

The yellow oxide of lead and the white oxides of antimony, bismuth, zinc, and arsenic are formed, when the combinations of these metals and chlorine are decomposed by a solution of potash. But on comparison with the best analyses of the oxides, there is not, excepting in the case of zinc and arsenic, that coincidence of proportions which might be expected. Zinc being as 34.5, the oxygene in the oxide from the analysis NICHOLSON's Journal, Vol. XXVII. p. 375.