1. The following researches originated from some experiments which were undertaken to discover an improved method for ascertaining the quantity of nitrate of potassa existing in crude saltpetre. After several unsuccessful attempts the action of hydrochloric acid was tried. The fact, that nitrates are decomposed by this acid, has been long known; but the nature of the resulting compound of potassium has not, so far as I am aware, been hitherto determined. I anticipated that the nitrate would be decomposed into chloride of potassium. To decide the question some pure nitrate of potassa was mixed with hydrochloric acid, and the mixture heated; while at common temperatures no perceptible action occurs, but immediately the acid becomes hot, decomposition commences. Chlorine and nitrous acid are evolved with copious effervescence, and the nitrate slowly disappears. The solution was gradually evaporated to dryness, and the dry salt treated with an additional quantity of acid until decomposition was no longer evident. The resulting salt was then carefully examined, and it was found to be pure chloride of potassium. This experiment was repeated several times, and all the results concurred in satisfactorily establishing the fact, that nitrate of potassa may be perfectly converted into chloride of potassium, provided a sufficient quantity of the acid be employed, and the temperature necessary to effect the decomposition be properly regulated. 2. So far the decomposition was admirably adapted for the object mentioned at the commencement. The usual impurities, such as chlorides, sulphates, silica, &c. which any sample of crude saltpetre might contain, would obviously remain unchanged, while the nitrate of potassa alone suffering decomposition, its quantity could easily be ascertained, by comparing the weight of the resulting salt with that obtained from a known quantity of absolutely pure nitrate. Several experiments were therefore performed to determine the exact quantity of chloride of potassium corresponding to a known weight of nitrate. The mean result of four experiments gave the ratio of 100 of nitrate to 73·730 chloride. I was then naturally led to compare this result with the equivalent numbers of these two compounds. In this country there are two series of equivalents in general use, one in which whole numbers are adopted, and the other in which fractional parts are admitted. For example, according to the former, nitrate of potassa will be 102, and to the latter 101·3. So chloride of potassium will be 76 and 74·6. Whence, according to the former, every 100 parts of nitrate should yield 74·51 of chloride: the latter gives the ratio of 100 to 73·613. But these results differ considerably from my experiments. In the one case, we have a difference of ·78, and in the other ·12. Whence therefore could they arise? Either the process must be defective, or the equivalent numbers, so generally considered as correct, must be erroneous.
Estimation of the mean normal measure from flat sections
