204 Original Articles. [April, 



thick entirely cuts off the red end of the spectrum to D, so that, on 

 placing in front of the tube one containing a strong solution of the 

 perchlortde, no light whatever passes. Mixing the two solutions pro- 

 duces no change, and the opacity is so complete that direct sun-light 

 does not penetrate through it. We may thus readily explain why the 

 protoxide combined with peroxide gives a black oxide ; for one cuts off 

 all the rays which the other transmits. Taking 12 as the width of 

 the entire spectrum, a foot thick of protochloride also cuts off the blue 

 end up to 2 below D, and only transmits bright green rays. On 

 reducing the thickness to about 4 inches, one-half of the spectrum from 

 the red end to D is cut off, but the blues and violet are slightly 

 transmitted ; and when the thickness is an inch or so, the whole of 

 the blue end readily passes, though the extreme dull red is cut off 

 for about ith of the distance to D. The crystallized salt acts in the 

 same manner. So far as I have been able to ascertain, similar prin- 

 ciples apply to the oxides of iron when combined with silica. Dif- 

 ferent kinds of bottle-glass show this very well. Since the intensity 

 of the colouring due to a given amount of oxide may vary, it is impos- 

 sible to do more than form an approximate conclusion as to the 

 relative action of the two oxides in producing the colour ; and it is in 

 this sense that I shall speak of one or other preponderating. How- 

 ever, I think we may deduce the following conclusions : — If a mineral 

 cuts off the blue end of the spectrum nearly to D, but transmits the 

 red up to the extreme limit, there can be little or no water or prot- 

 oxide present ; whilst, on the contrary, if it contain little or no com- 

 bined water, and be sufficiently coloured to cut off the red end halfway 

 to D, and yet transmits all the blue, there must be very little peroxide 

 present. If it cuts off all the red end to D, and yet transmits the green 

 part of the spectrum, there must be relatively more protoxide than 

 peroxide. If the blue end is cut off to near D, and also a considerable 

 part of the red, so that only a narrow band of light is transmitted 

 above and just below D, unless much water be present, there must be 

 a good deal of both peroxide and protoxide ; whereas, if there is only 

 a slight action at both ends of the spectrum, there cannot be much of 

 either. For example (assuming that the colour is chiefly due to the 

 oxides of iron), the red mica in a lava from the Lower Eifel transmits 

 the extreme red, but cuts off half the spectrum from the blue end, and 

 therefore nearly all the iron must exist as peroxide. Augite in a lava 

 of Vesuvius, which has been melted in situ, whilst the surrounding 

 rock remained nearly solid, cuts off a little of the extreme red and 

 also half the entire spectrum from the blue end, and hence contains 

 more per- than prot- oxide. Bright green chlorite in slate from Kes- 

 wick cuts off about J of the distance to D from the red end, and also 

 £ of the whole spectrum at the blue end, and therefore contains rela- 

 tively less per- than prot- oxide. Such facts might often be good 

 checks on tho results obtained by analysis, even if material enough for 

 that purpose could be procured. 



