[414] See also Meyer and Wuorinen (_loc. cit._). Egan and Balke[415] have recently found the ratio Yt₂O₃ : 2YtCl₃ to be very suitable as a basis for atomic weight determinations; the oxide is converted into the anhydrous chloride in a quartz flask. In a preliminary experiment, they obtain as a mean of three consistent determinations the provisional value 90·12; the yttria employed was considered to contain not more than one-half per cent. of erbia. [415] _J. Amer. Chem. Soc._ 1913, ~35~, 365. Recent work by Meyer and his co-workers[416] indicates that the accepted value is too high. Preliminary work with the synthetic sulphate method gave the values (corrected) 88·71 and 88·73; the mean value of six analytical sulphate determinations, made on material carefully purified by the iodate method, was 88·75, the extreme values being 88·71 and 88·76. They consider that the true atomic weight is 88·7, the value of the second decimal figure being a little uncertain. [416] Meyer and Wuorinen; Meyer and Weinheber, _loc. cit._ ~Detection.~--The spark spectrum of yttrium has been examined by many authors, and the ultraviolet as well as the visible regions have been mapped; _vide_ Exner and Haschek; Eder and Valenta, also Becquerel.[417] [417] _Compt. rend._ 1908, ~146~, 683. The arc spectrum has been examined by Kayser, Eberhard,[418] and Eder and Valenta;[419] Exner and Haschek give the following as the most intense lines: [418] _Zeitsch. wiss. Photochem._ 1909, ~7~, 245. [419] _Sitzungsber. kaiserl. Akad. Wiss. Wien_, 1910, ~119~, IIa, 1. 3216·83 3242·42 3328·02 3600·92 3611·20 3621·10 3633·28 3664·78 3710·47 3774·52 3788·88 3950·52 3982·79 4077·54 4102·57 4128·50 4143·03 4177·74 4302·45 4309·79 4348·93 4375·12 4883·89 6191·91 6435·27 Pure yttrium compounds should be colourless, show no absorption in the visible region, and yield a perfectly white oxide. ~Scandium~, Sc = 44·1 The scandia obtained by Nilson in 1879 was isolated from the minerals gadolinite and euxenite; it consisted very largely of ytterbia, as shown by spectrum examination[420] and by atomic weight determinations, which gave the value 90. In the same year[421] Cleve prepared the oxide in a much purer state, using as his source the minerals gadolinite and keilhauite; he described several salts, carried out atomic weight determinations by the analytical and synthetic sulphate methods, and showed that scandium corresponds with the Eka-boron of which the existence was predicted by Mendelejeff in 1871.[422] Starting from a large quantity of euxenite, Nilson[423] in the following year prepared several grams of approximately pure scandia, which contained only traces of ytterbium. [420] Thalén, _Compt. rend._ 1879, ~88~, 642; 1880, ~91~, 45. [421] _Compt. rend._ 1879, ~88~, 419. [422] See also Mendelejeff, _Ber._ 1881, ~14~, 2821. [423] _Ber._ 1880, ~13~, 1439.