EFFECT OF PRESSURE UPON ARC SPECTRA. GOLD. 139 



wave-length than the first upon which it was previously believed to depend. The 

 results of the present investigation favour a dependence upon the 3rd power of 

 the wave-length, and this agrees with some theoretical deductions made by 

 Dr. O. W. RICHARDSON.* 



(5) Resolution into Groups of Lines. Diagram 1 aftbrded very little encouragement 

 to any attempt to classify the lines according to the amounts of their displacements, 

 but Diagram 2 provided the clue by which this might be accomplished, and indicated 

 that/2 and r2 were to be associated with the group of lines o2, pi, ql, </2, falling 

 between them in Diagram I, and that the remaining lines that had been investigated 

 over a wide range of pressures, gl,jl, and ll, belonged to another group or groups. 



Diagram 4, which was drawn on the assumption that the displacement varies as 

 the cube of the wave-length, further justified the separation of these three lines from 

 the others and supplied good reason for believing that they belong to three groups 

 with different rates of displacement ; this diagram makes the scheme so much more 

 orderly and coherent that there can be little doubt but that this method of treatment 

 is correct. 



In the iron arc it had been found possible to divide the lines into three groups 

 according to the amounts of their displacements, and, following the nomenclature 

 there adopted, the three groups in the gold arc are named Groups I., II., and III., in 

 order of increasing displacement. 



For gold the mean value of the displacement of Group 111., divided by the cube of 

 the wave-length, is 653 x 10~ 13 . 



The only representative of Group II. that has been measured over the whole range 

 of wave-lengths is ll, and this line was particularly difficult to measure on account of 

 its great breadth, and because its intensity curve under pressure is very flat-topped 

 (Plate 3). Its mean value divided by the cube of its wave-length is 384xlO~". 

 Two other lines, however, si and t2, which have only been observed at two or 

 three pressures, are seen from Diagram 2 to possess displacements of the same order 

 of magnitude as ll. If these lines be included in Group II., the mean value for the 

 group is 356 x 10~". 



Save in their mean values there is little agreement between the measurements for 

 the lines gl and jl, but they certainly are displaced less than any of the other lines, 

 and it is consequently justifiable to place them in a separate group, of which the mean 

 value divided by the cube of the wave-length is 195x 10~ 13 . 



The ratios of the displacements divided by the cube of the wave-lengths are : 



Group I. : Group II. : Group III. = 195 : 356 : 653. Unfortunately, only two lines 

 belonging to Group I. and only three belonging to Group II. have been measured, so 

 that sufficient data have not been forthcoming to decide whether this ratio may be 

 more simply expressed as 1 : 2 : 3 or 1 : 2 : 4. The latter ratio was found to expreas 

 the relationship between similar groups in the spectrum of the iron arc, and thus 

 * RICHARDSON, 'Philosophical Magazine,' vol. 14, p 557, 1907. 



