compared with the Motions of Waves of Light. 135 



revealed to us by the fixity of the rays in the spectrum of each 

 gas, and by their number within the limits of the visible spec- 

 trum and the doubtless greater number that exist beyond. That 

 the motions are, however, not altogether free from perturbation 

 is made clear by such cases as that of the lines corresponding to 

 1) in the sodium spectrum, which are not linear except at low 

 temperatures, but of a breadth increasing with the temperature, 

 and ill defined at the edges. This suggests a perturbation of 

 motions that would otherwise be regular by the disturbing influ- 

 ence exerted by other molecules when they are near enough. If 

 during some of the appulses the period of a vibration be some- 

 what accelerated, by others somewhat retarded, and if in all cases 

 the vibration tend to settle down from the constrained towards its 

 normal period during the undisturbed intervals between con- 

 tacts, then will the effect be such as we see. If the irregu- 

 larity fade away so readily as to last but a trifling part of the in- 

 terval between the collisions, the line in the spectrum will be 

 sharply defined — as in the sodium-line in a spirit-lamp, or the 

 thallium-line in the flame of a Bunsen's burner. If it continue, 

 though with a marked decrease of intensity, through a consider- 

 able part of the time during which the molecule is pursuing a 

 straight course, the spectral line will have a sensible breadth and 

 hazy edges, as in the sodium-line of a Bunsen's burner. And if 

 the commotion be so violent that but little restoration can be 

 effected before a fresh contact and a fresh perturbation, the spec- 

 trum of a gas may even become continuous. All the stages up 

 to that of a continuous and nearly uniform spectrum can, for 

 example, be produced in hydrogen. It has been observed, too, 

 that rays of high refrangibility are more easily dilated into 

 spectral bands of the class with which we are here dealing than 

 rays of low refrangibility*. Perhaps this may arise from the 

 circumstance that a ray of high refrangibility is due to an atom 

 of the molecule the periodic time of whose motion is shorter 

 than those of other motions which give rise to rays of low 

 refrangibility; this atom, therefore, goes through many more 

 of its little evolutions in the instant while a passing mole- 

 cule is near enough to affect it, and is likely to suffer a propor- 

 tionately more serious disturbance. If we have been right in 

 assigning the cause of this class of dilated spectral lines, it is im- 

 possible to resist the impression that cases will sometimes arise in 

 which the periodic time of an atom of a molecule will yield to a 

 force tending to disturb it more freely in one direction than in 

 the other, and will thus produce a ray expanding on one side 



* This, however, is far from being uniformly the case ; and there are 

 even instances of gases, as, for example, oxygen, in which the dilatation 

 first takes place at the red end of the spectrum. 



