ON OUR KNOWLEDGE OF SPECTRUM ANALYSIS, 139 



a group of three yellow rays, four green bands, and a line in the yellowisli- 

 green (only visible in tbe spark). In the case of rubidium these are 

 replaced by a group of four red rays and four double lines in the yellow 

 and green. The spectrum of potassium ends with a violet line, since 

 proved to be double, and the spectrum of rubidium, as is known, also 

 ends with two violet lines. If we form the ratio of the wave-length of 

 the centre of each of the five intermediate groups in rubidium to those 

 of the corresponding group in the potassium spectrum, we obtain the 

 numbers 1-0G4, 1-065, 1-056, 1-058, 1-063, which numbers are approxi- 

 mately constant. The ratio of the middle of the two red rubidium lines 

 to the middle of the two red potassium lines is 1-022, and the correspond- 

 ing ratio of the violet lines is 1-04. If we accept the fact of corre- 

 spondence of these lines we see that the atom with the higher atomic 

 weight vibrates more slowly. The weak point of the comparison consists 

 at present in the uncertainty as to which of all these lines and bands 

 belong to the metal and which to the oxide. We could not of course 

 attempt any correspondence between lines of the oxide of one metal and 

 the lines of the other metal itself. Similar relations exist, according to 

 Lecoq, between the two spectra we have discussed and that of calcium. 

 The two blue calcium lines concluding the spectrum are again less 

 refrangible than the two corresponding rubidium lines, and at the same 

 time they are wider apart. There is, therefore, a progressive change, as 

 pointed out by Lecoq, in the behaviour of these blue and violet lines, 

 which in every case are truly metallic. In potassium they are very close 

 together and in the violet ; in rubidium they are wider apart and less 

 refrangible, though still in the violet ; in calcium they are still wider 

 apart, and in the blue. If we connect this fact with the similar change 

 which, as we have pointed out, sometimes occurs in one and the same 

 spectrum, where double and triple lines go closer and closer together as 

 they approach the violet, we seem certainly to have a suggestive analogy 

 which may serve as the basis for further inquiries. 



We shall only follow Lecoq into his comparison between the spectra 

 of the chlorides of the alkaline earths, as his comparison of the metallic 

 and oxide lines seems to be uncertain. 



Both the spectrum of the chloride of calcium and that of strontium 

 consists of five bands. The differences between the bands of the chloride 

 of calcium gradually decrease ; they become more refrangible ; and the 

 same holds for the four least refrangible bands of the spectrum of 

 strontium chloride. Forming the proportion between the wave-lengths of 

 corresponding bands we find that the ratios of the bands of strontium 

 chloride to those of the calcium salt are 1-063, 1-065, 1-066, 1-070, 1-071, 

 and these ratios are seen to increase gradually with the decrease of wave- 

 lengths. The salt with the higher atomic weight vibrates, again, more 

 slowly. The spectrum of baz-ium chloride resembles in its general 

 arrangement that of the strontium salt, but it has six bands instead of 

 five, the central band being apparently broken up into two. Other dis- 

 crepancies are noticed on further inspection, and the barium bands, con- 

 trary to the rule given by Lecoq, are more refrangible than those of the 

 lighter metals. Lecoq gets out of the difiiculty by supposing that the 

 barium chloride spectrum which we observe is not the one corresponding 

 to that of the sti-ontium and calcium chloride, but that it is its higher 

 harmonic, and that we are to look in the ultra-red for the true corre- 

 spondence in the barium spectrum. This explanation may be correct, and 



