1905.] The Arc Spectrum of Scandium, etc. 539 



amongst the rarer elements it apparently stands by itself from this 

 point of view. This prominence of scandium lines in some stellar 

 spectra and particularly in the chromospheric spectrum, makes it 

 desirable to give as complete a record of the lines as possible and also 

 to analyse them in relation to their appearance or non-appearance in 

 extra-terrestrial spectra. 



Some time ago Sir William Crookes was good enough to send a 

 sample of scandium oxalate, and very good photographs of the arc 

 spectrum have been obtained with a larger Rowland concave grating 

 having a ruled surface of 5| x 2 inches (14| x 5 cm.), and a radius of 

 21 feet 6 inches. The scale of the photographs is such that the 

 distance between K and D is 30J inches, or 77 cm. This is equivalent 

 to 2-6 tenth-metres per millimetre. The scandium oxalate was 

 admittedly impure, and for the purpose of eliminating lines due to 

 impurities, the spectrum has been directly compared with the spectra 

 of all the chemical elements available at Kensington, which were 

 photographed under identical instrumental conditions. The chief 

 sources of impurity were found to be cerium, thorium and ytterbium. 

 The lines of the first two elements were easily eliminated by com- 

 parison with the Kensington photographs of their respective arc- 

 spectra. In the case of ytterbium it was a more difficult matter, as 

 this is one of the elements not investigated at Kensington. The 

 elimination of its lines has, however, been accomplished as far as 

 possible, by ascertaining whether there were lines in the scandium 

 photograph in the position of the stronger lines of ytterbium as 

 recorded by Thalen,* and Exner and Haschek.f If such were found 

 to be the case, and the intensity in the scandium photographs such 

 that the line was thought to be due to ytterbium, it was discarded 

 from the list of scandium lines. 



The fiducial lines used for the reduction of wave-lengths were the 

 H and K lines of calcium (which occur as impurity lines in the 

 scandium spectrum), and Rowland's solar-scandium lines 3907*62, 

 4020-55, 4082-59, 4247*00, 4314-25, 4400-56, 4670-59, 5672-05. The 

 coincidence of these lines was first confirmed by a direct comparison of 

 the Kensington photographs of the solar and scandium spectra. In 

 addition to the foregoing, well-marked scandium lines were found to be 

 exactly coincident with the isolated solar lines 5031*20, 5239'99, 

 and 5527*03, for which Rowland had given no origin. The solar 

 wave-lengths of these were adopted and used in the reduction of the 

 wave-lengths of the remaining lines. 



The table at the end of the paper gives the residuum of lines after 



* ' Oilversigt k. Vetensk. Akad. Forhandl.' (1881). 



f Wellenlangen Xabellen fur Spektralaualytische Untersuchungeii auf G-rund 

 der Ultravioletteu Bogenspektren der Elemente,' Leipzig und Wien, Franz- Deuticke, 

 1904. 



