(1) As regards the slow oxidation of the V₂O₃ (Berzelius’s suboxide); a sample made on November 13, 1866, weighing 0·7507, weighed on June 12, 1867, 0·8733. This corresponds to an oxide higher than V₂O₄ (V = 51·4). Now I want this oxide and tube drying under the air-pump and weighing carefully and keeping for further examination (the weight of tube and oxide was 3·6066 on June 12th last). (2) When a neutral solution of V₂O₂ in SO₄H₂, got by zinc reduction, and neutralized by excess of zinc, is exposed to a current of air, it goes brown, and this brown colour does not alter. Analysed by permanganate it would seem to contain V₂O₄. I want this confirming. (3) Try to get sight of a copy of Greg and Lettsom’s “British Mineralogy” (you can go and call on Mr. Robt. P. Greg, Greg Bros., Chancery Lane, and ask him to lend you the book for me). Under “Vanadinite” you will see some mention of Heddle and his observations. I find nothing about him in Rammelsberg’s paper. (4) Has V₂O₄ been prepared by heating the suboxide V₂O₃ in the air at low temperatures? Is it green? The thing above all others necessary for us now is to get the _metal_. We must set about this at the beginning of the session. If you have time before you go away to clear up these four preliminary points I shall be glad. * * * * * RODDAM, _September 26, 1867_. The first thing to do when we begin will be to try to get the metal V = 51·3 by forming the ammonio-chloride and reducing the nitride in ammonia.… If we only can get 0·5 of metal the question of atomic weight is settled beyond dispute. I have now collated all the experiments: (1) On the oxides of vanadium; (2) on the determination of atomic weight by reduction of V₂O₅ to V₂O₃; (3) on the oxychlorides and determination of atomic weight from the chlorine determinations. These latter require re-calculation as I do not know how much should be allowed for as impurity in the silver. Still, I see already that this alteration can be but very slight, and the numbers agree very well, viz. (1) mean at. wt. from reduction expts. 51·362 (probable error 0·068); (2) mean of 9 volumetric chlorine determinations of the oxytrichloride 61·28; (3) mean of 8 weight determinations of chlorine 61·21, so that as far as I yet see the true atomic weight is 51·30. I am obliged by your extract from Greg and Lettsom. When in St. Andrews I saw Professor Heddle, and he promised to send me some crystallized specimens of vanadinite and pyromorphite existing in one crystal.