308 Royal Society, 



erbia lines were most distinctly seen when the slit of the spectroscope 

 took in the light from the extreme edge of the incandescent solid. 

 When the bright lines were best observed, the continuous spectrum 

 was relatively faint. Again, when the slit was made to cut the edge 

 of the ignited bead of the earth, the strong green line of erbia was 

 seen to extend to a very small but appreciable distance above or 

 below (as the case might be) the continuous spectrum. I could only 

 observe this for the strong line. I failed to get any trace of lines in 

 the spectrum of the flame beyond the incandescent erbia. 



" The erbia was next heated in the oxyhydrogen flame to the maxi- 

 mum temperature that the wire would bear for three and a half 

 hours ; but the green line was seen to be just as strongly marked at 

 the end as at the beginning of the experiment. The bulk of the 

 erbia was so much reduced by this treatment, that I have now scarcely 

 a trace left. 



" From the results of these experiments, I think we must con- 

 clude : — (1) that the light emitted by incandescent erbia is derived 

 chiefly from the ignited solid, but that the bright lines observed in 

 its spectrum have as their source a luminous vapour of extremely 

 low tension at even the highest temperature of the oxyhydrogen 

 flame ; (2) that this interrupted spectrum belongs either to erbium 

 or to its oxide. 



" If these conclusions are true, it follows that erbia is not an ex- 

 ception to the ordinary law. 



" It would appear that in these experiments three substances have 

 been employed, varying in their degree of volatility. At the tempe- 

 rature of the oxyhydrogen flame magnesia appears to be less volatile 

 than lime ; but I am in doubt what relative volatility to assign to 

 erbia, since its spectrum of bright lines can be seen when the earth 

 is heated in the plain hydrogen flame, and yet at the much higher 

 temperature of the oxyhydrogen jet the volume of luminous vapour 

 does not appear to materially increase. 



" Finally, we have yet to learn whether or not in all these cases 

 reduction of the oxide precedes volatilization ; if reduction takes 

 place, the luminous vapour must be that of the metal. The settle- 

 ment of this question would no doubt be very difficult. But I rather 

 incline to the view that the vapour whose spectrum is obtained on 

 igniting these earths is that of the metal ; for I find that the bright 

 lines are most easily observed when hydrogen is present in excess 

 in the oxyhydrogen flame. Moreover the actual amount of matter 

 volatilized on very prolonged heating is really very small ; and this 

 circumstance appears to favour the view that a slow surface-reduction 

 is in progress." 



