ON OUR KNOWLEDGE OF SPECTRUM ANALYSIS. 283 



Taking the whole of these experiments together, they do, I believe, 

 indicate that the lines of sodium widen more easily at a comparatively 

 low temperature ; but, as they may also be seen very wide at high tempera.- 

 tnres and narrow at low ones, they leave the original cause of the widen- 

 ing unexplained. 



The fact that the sodium lines widen more easily at a comparatively 

 low temperature is in accordance with the theoretical speculations we 

 have given on the cause of the widening of lines. For in the passage 

 quoted Lippich has remarked that an increased widening of lines would 

 go on simultaneously with an increased deviation from Boyle's law ; and 

 that deviation will be greater when the vapour approaclies its temperature 

 of condensation. 



Moreover, the widening of the sodium lines seems to take place 

 chiefly at the temperature at which the line-spectrum changes into the 

 band-spectrum. It will be seen further on that, according to the opinion 

 held by most spectroscopists the band-specti'um is due to a molecule con- 

 taining a greater number of atoms than that giving the line spectrum. 

 If this opinion is true, sodium vapour ought to show a change of vapour- 

 density as one spectrum changes into the other, similar to that which has 

 recently been proved to exist in iodine by Victor Meyer.' That at the 

 moment when the atoms or molecules of sodium have a tendency to com- 

 bine with each other, the molecu^lar forces should be afEected and dis- 

 turbed in such a way as to produce a widening of lines seems perfectly 

 intelligible. At the same temperature at which the band spectrum of 

 sodium changes into the line spectrum, Mi*. Lockyer has observed some 

 very remarkable phenomena.^ In some parts of the tube in which the 

 sodium was volatilized, the Hues seemed to widen only toAvai-ds one side, 

 while in others they were widened towards the other side. It is, perhaps, 

 worth mentioning, in connection with a remark by Lord Rayleigh on dis- 

 turbed vibrations,^ that the two parts of the band-spectrum of sodium lie 

 on the two sides of the D lines. 



Referring again to the effect of pressure on the Avideuing of lines the 

 question arises, whether for a given temperature and pressure a line may 

 be of different width whether the molecule is placed in an atmosphere of 

 similar or dissimilar molecules. We shall have occasion to refer to this 

 point again, and to show that such a difference in all probability exists, 

 and tliat it is not due to a mere reduction or increase in the number of 

 luminous molecules in the line of sight. We may mention here for 

 instance that Mr. Lockyer ^ has observed that tlie lines of oxygen or 

 nitrogen may be obtained sharp at atmospheric pressure by mixing a 

 small quantity of one gas with the other. The gas which is present 

 in small quantities has its lines sharp. If, therefore, we observe that in 

 putting larger quantities of sodium into a flame we widen the lines, 

 we must take many questions into account, and not conclude merely 

 that an increased thickness of the radiating layer has produced the result. 



We finally refer to one cause which limits the sharpness of sjiectro- 

 scopic lines, and which was fix-st pointed out by Lippich* and later by 

 Lord Rayleigh.*^ The molecules of a gas are, in addition to their vibratory 

 motion, endowed with a ti-anslatory motion. Those molecules which are 

 moving towards us will, in accordance with Doppler's principle, send us 



' Chem. Ber. xiii. p. 394 (1880). " Pliil. Mar/, vi. p. 161 (1878). 



"^ Pror. Boy. Soc. xxii. p. ;S78 (1874). * Por/f/. Aiiii. cxxxix. p. 46.5 (1870). 



' P/iil. Miiff. xliii. p. 322 (1872). « Naivn; xvii. p. 148 (1877). 



