^{page 429 note *} See my paper on “Alkali-Proof Vessels,” Jour. Soc. Chem. Industry for 1884, page 303.

^{page 429 note †} Ibid.

^{page 430 note *} In the research referred to I took special care to eliminate from my preparation every trace of potassium, rubidium, and cæsium. Yet, when fused in platinum at a red heat in the presence of air, it attacked the metal badly, with formation of a dark-coloured platinite. This is contrary to a statement of Teoost's, , who says (Dictionnaire de Chimie, vol. ii. part i. page 228)Google Scholar, that “lithia” attacks platinum only if contaminated with rubidium or cæsium.

^{page 438 note *} According to this formula there should be a minimum of solubility at t = 5°·77, and at 11°·54 the solubility should be the same as at 0°. At 5°·77, by calculation, y=6·665. It would take very exact work to see whether this minimum has any existence outside the formula, which latter of course does not pretend to formulate the actual law in the relation between t and y.

^{page 440 note *} Liebig's, Annalen, vol. cxvi. page 129Google Scholar.

^{page 441 note *} ? Vide infra.

^{page 443 note *} Note lost.

^{page 444 note *} See my memoir on the “Oxides of Manganese” in the Proc. Roy. Soc. Edin., for 1864Google Scholar, also Chem. Soc. Jour., same year, p. 294; also Debray's, “Experiments on Carbonate of Lime,” Jahresb. for 1867, p. 85Google Scholar.

^{page 446 note *} On the occasion of these analyses I invented an improved form of the Liebig bulbs, which is described and figured in the Soc. Chem Ind. Jour. and also in the Chemiker Zeitung, for 1888.

^{page 450 note *} This analysis did not proceed quite regularly; yet I have no reason to suspect that the result is at all far out.

^{page 452 note *} I quote from Meyer, Lothar and Seubert, , Die Atomgewichte der Elemente, Leipzig, 1883Google Scholar. With Lothar Meyer and Seubert, “Mean” means not the arithmetical mean of the several determinations, hut the result as il comes out if all the analyses are calculated as one. Thus, for instance, in the case of four analyses of chloride of lithium hy nitrate of silver, the four quantities of chloride are added together, and compared with the sum of the four precipitates of chloride of silver.

^{page 452 note †} Lothar Meyer and Seubert refer to a memoir published in 1857. I find, in the Zeitschrift für Chemie for 1862, an abstract of a memoir of Troost's on the same subject, in which the numbers quoted as immediate data of the analyses agree with those given by Lothar Meyer and Seubert.

^{page 453 note *} See footnote, page 452.

^{page 455 note *} On the assumption that the substance is a mixture of 82·16 per cent, of carbonate of soda, and 17·84 per cent, of hydrate NaOH, equal (the latter) to 13·83 per cent, of Na₂O, which agrees fairly with the determination by chloride of barium.

^{page 458 note *} Only the blowpipe used in all cases.

^{page 458 note †} Mean of two closely-agreeing determinations. In this Experiment (9) for the first time carbonate of soda from oxalate was used.

^{page 463 note *} If K = 39·136, as found by Stas. If K = 39·00, the percentage becomes 31·884. I am far from taking it for granted that the close agreement of our number with Stas's is not a mere accident.

^{page 463 note †} Transactions of the Royal Society of Edinburgh for 1887, p. 618Google Scholar.

^{page 466 note *} R = 85·33, as found in Experiment II.