THE CONSTITUTION OF Till: I.I.KCTKIC SI'AKK. -jo; 



line 3663 (Hg, y) is parallel to the cadmium line 3613 (Cd, e), and the measurements 

 an- lii irly << insistent, but when the pole was zinc amalgam, the displacements of the 

 mercury lines seemed decidedly greater than that of the zinc lines. Tliis would 

 indicate a higher velocity tin- x.inc than for cadmium molecules, while our previous 

 results had given almost identical values fur tin- displacements of the zinc and 

 cadmium lines. There is here some contradiction which requires clearing up hy 

 further experiments. 



liimnuth (fig. 25, Plate 11) gave us must interesting results, as it possesses some 

 lines, such as y, which are curved so little that the velocities of the molecules giving 

 these lines is found to be larger than that observed in any other case except aluminium. 

 ( >n the other hand, the line at 3793 () indicates the smallest velocity measured. 

 There are some special difficulties standing in the way of the measurement of the 

 bismuth lines, which made us hesitate some time before definitely asserting the 

 different curvature of the lines, but we think that our best photographs, one of which 

 is reproduced in fig. 25, leave no doubt on the question. One of the difficulties lies in 

 the fact that bismuth and mercury are the only metals in which the lines are repeated, 

 owing to the oscillatory discharges. When six jars are in circuit, the appearance is 

 that of fig. 26, and the lines become mixed and difficult to measure. Another difficulty 

 lies in the great difference in the sharpness of the lines even on the stationary film, 

 the lines which show a small curvature l>eing much sharper. There was a possibility 

 of an illusion due to this cause, similar to that explained in the case of the double 

 zinc lines, but we cannot believe that the difference in curvature between the lines 

 marked (y) and () in fig. 5 can be due to this cause. 



[April 26.] We possess very few investigations on the spectrum of bismuth- 

 LECOQ DE BOISBAUDRAN, who is acknowledged to have purified his substances with 

 extreme care, gives the line 5209 (a) as one of the characteristic lines of bismuth, and 

 he also gives 4302 (8) and 4260 (e) as l>elonging to bismuth. The relative intensities 

 of the lines as given by him cannot be expected to coincide with our own, as he used 

 very different spark conditions. Our spectrum agrees, on the other hand, j>erfectly 

 with that given by HARTLEY and ADENEY,* who traced no coincidence between the 

 lines we made use of in this research with those of other metals. 



As the great difference in the molecular velocities suggested the possibility that 

 bismuth was a mixture of elements, we obtained, through the kindness of Messrs. 

 JOHNSON and MATTHEY, samples of bismuth prepared from three different sources. 

 The visible portion of the spectrum was examined with great care, but no difference 

 in the relative intensities of the lines could be detected. 



6. Discussion of Result*. 



When we compare together the results obtained for different metals, the first 

 question that arises refers to the connexion between the velocities and atomic weight, 



* 'Phil. Trans.,' vol. 175, p. 130, 1884. 



