IN RELATION TO QUANTITATIVE CHEMICAL ANALYSIS. 
53 
a lump of the material, it was found that the points showed more or less unmistakable 
indications of the presence of iron. 
From this it appears that the same piece of the mineral may be of unequal purity. 
Alterations in spectra caused by moistening the electrodes. 
When electrodes of graphite are moistened with water, the only alteration remark¬ 
able in the spectrum is a lengthening of the short carbon lines, with wave-lengths 
3590*0, 3583*5, 2836*8, 2836*0, 2511 *6, 2508 , 7, and 2478*3, so that they extend from 
pole to pole. The other lines of carbon do not appear to be more than slightly 
lengthened, and in addition the line 4266*3 is much weakened. This change would 
i*eceive a natural explanation by assuming the formation of some gaseous or at least 
volatile carbon compound. If it be due to carbon dioxide, a similar change should 
occur when the electrodes are surrounded by pure oxygen. In order to put this 
matter to the test of experiment, pure dry oxygen was passed for some time through 
a glass tube in which were fastened electrodes of graphite, the end of the tube being 
closed by a plate of quartz. The change seen in the spectrum under these circum¬ 
stances is the following :— 
Wave-lengths. 
4266-3 This line does not appear altered, 
3919'5 This line remains short. 
3875'9 This line l’emains short. 
3590'7 This line remains short. 
•0J 
6 ) 
*0J 
2836-8 
2836 
2746-6 
2640 
2511-6 
2507-8 
2478-3 
2297-6 
These lines are so lengthened as to be rendered continuous from pole to pole. 
These lines are somewhat lengthened. 
>These lines are lengthened so as to be rendered continuous. 
A map to the scale of wave-lengths is appended (Plate 3). 
In addition to the above, the following lines are seen with difficulty, a pair with 
wave-lengths about 3168*0 and 3166*3 is much lengthened, and a very nebulous ray, 
extending from about 2995 to 2990 (\=2993) is rendered much larger and more 
distinct. It will be seen that the two spectra differ considerably ; it does not, 
therefore, seem at all probable that the change is due to the cause suggested. Indeed, 
it is not, as may be learnt by a study of the action of the spark on metallic electrodes. 
When photographs are taken of dry electrodes of copper the lines in the spectrum 
with lesser wave-lengths than 2766*2 and extending to 2243*5 are all short lines. 
When the electrodes are moistened with a solution of a metallic chloride, with 
hydrochloric acid, or even with pure distilled water, these lines are greatly lengthened, 
some of them so much so as to become continuous. When platinum, iridium, and 
gold electrodes are partially immersed in water, the same lengthening of the short 
