The _carbonate_, Sc₂(CO₃)₃,12H₂O, is thrown down by addition of ammonium carbonate as a bulky white precipitate, easily soluble in a hot solution of the precipitant; the solubility in excess may be used in the separation of scandia from yttria. Addition of water to such solutions causes separation of a basic carbonate, but crystalline _double carbonates_ may be obtained by evaporation of concentrated solutions containing a large excess of alkali carbonate. The sodium compound, Sc₂(CO₃)₃,4Na₂CO₃,6H₂O, is very sparingly soluble, and has been used in the separation from thorium. The _oxalate_, Sc₂(C₂O₄)₃,5H₂O, differs from other oxalates of the group, which generally separate with 10 molecules of water of crystallisation, not only in its water content, and in its solubility in acids, but also in the ease with which it forms double oxalates soluble in excess of alkali oxalate; in this latter property it shows a further resemblance to zirconium and thorium. The _formate_ and _acetate_ have the formulæ Sc(OH)(HCOO)₂,H₂O and Sc(OH)(CH₃COO)₂,2H₂O, respectively. A large number of organic salts have been described by Sir William Crookes.[430] [430] _Loc. cit._; see also Meyer, _Zeitsch. anorg. Chem._ 1908, ~60~, 134; Meyer and Winter, _ibid._ 1910, ~67~, 398. ~Atomic Weight.~--The mean values obtained by Cleve[431] in 1879 were 44·96 and 45·20 by the analytical and synthetic sulphate methods respectively. In the following year Nilson,[432] using purer material, obtained the value 44·13 by the synthetic method. Meyer and others (_loc. cit._) have criticised Nilson’s estimation on the ground of his empirical method of obtaining the neutral anhydrous sulphate. Determinations made with material purified from thorium by the iodic acid method gave the values 44·11, 44·11, 44·20; material purified by the double ammonium tartrate method gave the atomic weight 43·90. Meyer has shown that small quantities of thoria in the oxide cannot be detected spectroscopically; the value of the magnetisation coefficient, however, showed the oxide obtained by the last method to be free from thoria, and he considers another determination of the atomic weight to be necessary. [431] _Loc. cit._ [432] _Loc. cit._ The value accepted by the International Committee is 44·1. ~Detection.~--Scandium gives no absorption spectrum in the visible region. The spark spectrum has been examined by Thalèn (_loc. cit._) and Nilson;[433] see also Exner and Haschek, Lockyer and Baxendall,[434] and Crookes (_loc. cit._). The arc spectrum has been examined by Fowler,[435] Eder and Valenta,[436] and Exner and Haschek. [433] _Compt. rend._ 1880, ~91~, 56, 118. [434] _Proc. Roy. Soc._ 1905, ~74~, 538. [435] _Phil. Trans._ 1908, A, ~209~, 47. [436] _Sitzungsber. kaiserl. Akad. Wiss. Wien_, 1910, ~119~, II_a_, 576. The most intense lines of the arc spectrum are the following: