ON OUK KNOWLEDGE OF SrECTRUM ANALYSIS. 291 



of lines by means of laboratory experiments, we sball always first reverse 

 the longest lines, for the order in which the lines reverse will be the order 

 of intensities at the temperature of the reversing layer. This is also con- 

 firmed by experiment. Thus Coruu ' has reversed the two aluminium 

 lines between H and K. These are the longest lines, according to Lockyer, 

 and they are also the only two lines which are reversed in the sun, 

 although aluminium possesses some very strong lines in other parts of the 

 spectrum. Zinc and cadmium gave similar results. The order in which 

 the metallic lines reverse had been made the subject of a series of investi- 

 gations by Liveing and Dewar, and their results tend to confirm the law 

 given by Lockyer. Although some differences exist between the order of 

 reversal given by Liveing and Dewar ^ and the order of length given by 

 Lockyer, it must be left for further inquiry to see whether the differences 

 are real. The difficulty in estimating the relative length of two lines which 

 are not very near together, must be very great, and no doubt some of 

 the lengths as given by Lockyer may require some corrections. Thus, for 

 instance, Profs. Liveing and Dewar have reversed the lines 5085, 4799, 4677 

 of cadmium, but not 4416. This agrees with Lockyer's law, for the three 

 first are longer than the last ; but they have seen the line 6438 reversed 

 once, and this line is given by Lockyer as shorter than 4416. But from 

 the great intensity of 6438 in Lecoq's drawing, we should infer that it 

 was really a long line, and that the length given by Lockyer is not correct. 

 With the other metals, where a comparison is possible, the two lines of 

 investigation seem to lead to the same result. Lead first reversed 4058 

 and subsequently 4063 ; this is the order of their length, though 4063 is 

 faint at the temperature of the spark, and other much stronger lines have 

 not been reversed. 



The fact that a reduction of quantity, as for instance in an alloy, 

 destroys the shorter though perhaps stronger lines, and leaves the longest, 

 is more remarkable than might at first sight appear. Lecoq de Boisbau- 

 dran has studied the effect of diluting the liquids which he used as 

 electrodes, and his results confirm Lockyer's law of the longest lines. 

 Thus, for instance : lithium in the flame gives the red line very much 

 stronger than the orange line. The red line is the longest, but with a 

 concentrated solution and a spark Lecoq ^ found the orange line to be 

 stronger ; dilution with water, however, at ohce gave preponderance again 

 to the long red line. A similar remark applies to the red cadmium line, 

 6438, which is stronger in a concentrated solution than 4677, but is 

 •weakened by dilution, being in reality a shorter line. 



After having given the facts relating to the question of long and short 

 Imes, we have to see whether we can find a theoretical explanation of 

 these facts. 



The first explanation which naturally occurs to everyone would make 

 the appearance of the long and short lines depend on the greater thick- 

 ness of luminous matter surrounding the electrodes. As this thickness 

 decreases we should expect to see more and more lines disappear and only 

 the most persistent lines remain. These most persistent lines would be 

 the longest. But this explanation will not account for the facts ; for as 

 we have said the longest lines are not always the strongest, and we have 

 proved, in the first chapter of this report, that an increase of thickness 



' C. R. Ixxiii. p. .332 (1871). 



- See especiall}' Proc. Roy. Soc. xxix. p. 402 (1879). 



' a R. Ixxvi. p. 1263. 



U2 



