and Vegetable Colouring -matters by the Spectrum Microscope. 157 



Ammonia does not produce this effect when the colour is dissolved 

 in alcohol, the solution remaining fluorescent and still giving the 

 narrow band ; and, as a general rule, that solvent greatly impedes or 

 entirely prevents such changes, and on this account almost invari- 

 ably shows absorption-bands to the greatest advantage. 



We therefore see from the above examples that the absorption 

 follows the general rule, and is raised by acids, and depressed by alka- 

 lies ; but this only applies to the absorption when viewed as a whole, 

 and not to the separate bands ; for those reagents change their 

 intensity, but not their position. In some cases, indeed, their posi- 

 tion is slightly altered ; so that perhaps it would be more correct to 

 say that acids and alkalies may raise and depress the general ab- 

 sorption to an extent equal to a considerable fraction of its own 

 great breadth ; whereas they either do not change the position of 

 narrow absorption-bands, or merely raise and depress them by a frac- 

 tion of their own narrow width. It is their very definite position 

 that makes them so useful in this method of analysis. 



Unfortunately I have not hitherto been able to find a sufficient 

 number of colouring-matters giving rise to three or more well-marked 

 absorption-bands to warrant a general conclusion ; and therefore it 

 is perhaps premature to conclude that their centres always occur 

 at equal intervals. At the same time it is certainly a very common 

 fact. "When the maximum point of transparency occurs between the 

 different bands, there may be, as it were, a double interval ; but 

 then, sometimes, even this missing band may be seen under favour- 

 able conditions. A difference in the general absorption may also 

 somewhat alter the apparent position of a band, if it is strongly shaded 

 on one side and not on the other ; and the presence of impurities 

 may also modify the results ; so that absolute accuracy cannot be 

 expected in all cases ; and occasionally very narrow bands occur 

 which appear to belong to a second system. It must be borne in 

 mind that the bands are equidistant only when measured by means 

 of the interference-spectrum. Thus, in the case of Alkanet-root, 

 when measured with a micrometer, instead of the intervals/? to b and c 

 tod being equal, they are related to one another about as 1|- and 2 ; 

 and thus the general law is entirely obscured. If subsequent research 

 should prove that the bands are normally at equal intervals, it will 

 be a fact of great value in deciding whether certain spectra are or 

 are not due to a mixture of colours, since, if a band occurred at a 

 perfectly unequal interval, it would show that there must be at least 

 two substances. Even in the present state of our knowledge, any 

 inequality should make us carefully search for some satisfactory rea- 

 son for such a divergence from a common rule. My meaning will 

 be better understood from the following examples : — 



If a little of the colour of Brazil-wood be added to the solution of 

 Alkanet-root, the bands are not altered, and are seen at 4^, 5|, 

 and 7 \ ; but a little bicarbonate of ammonia developes a well-marked 

 band whose centre is at 5|, and therefore at an interval of 1 instead 

 of 1|. The same is also well seen in the case of a mixed solution 

 of Brazil-wood and blue Lobelia. I therefore argue that if an un- 



