ABSORPTION SPECTRA OF THE ALKALOIDS. 
473 
Several modifications were introduced with the object of gaining accuracy and 
reducing the amount of labour entailed by a large series of observations. 
1st. The solutions were as far as possible made up to the same strength, and cells 
of definite thickness were employed in series as 5 millims., 4 millims., 3 millims., 
2 millims., and 1 millim. This obviated the necessity for repeatedly diluting the 
solutions, and so greatly facilitated the work; while it diminished the possibility of 
errors, especially in the descriptions of the spectra. 
2nd. The measurements were all reduced to wave-lengths, and the absorption 
curves were plotted accordingly. As the curves have been made continuous the 
necessity for shading has been avoided. 
3rd. While the absorption curves serve for the comparison of one alkaloid with 
another, as a ready means of distinguishing them and of estimating the quantity 
present in a solution, a very careful description of the spectra from which the curves 
were drawn has been furnished, so that no essential details have been omitted such 
as were found to be imperfectly represented by the drawings. 
4th. The rays intended to give a suitable spectrum were obtained from metallic 
points, which, for the purpose, are superior to all others. In previous researches, 
electrodes of cadmium or of nickel alloyed with a small quantity of copper, in certain 
cases metallic indium and iron have been used. None of these are quite satisfactory. 
The cadmium and indium lines are not sufficiently close together or numerous, the 
wave-lengths of the iron and nickel lines had not been determined when these experi¬ 
ments were in progress; moreover, they were deemed unsuitable because there is too 
great a difference between the intensity of the lines in different parts of the spectrum, 
so that deceptive results are obtained when feeble absorptions occur. 
The electrodes I prefer to all others are cadmium, tin, and lead; the lines of tin and 
lead are numerous and well distributed throughout the whole spectrum ; they are 
distinctive in character, and therefore easily recognisable, which is not the case with 
the iron and nickel lines. The cadmium lines preponderate in intensity, but the 
emissive power of the cadmium is purposely reduced by diminishing the quantity of 
that metal in the spark. 
Thus I take for one electrode an alloy of tin with 25 per cent, of cadmium, for the 
other an alloy of lead with the same proportion of cadmium. When the two are used 
as opposite electrodes a spark spectrum is obtained every line in which can be easily 
recognised; the tin lines are on one side, and stretch half-way across ; the lead are on 
the other side ; while the cadmium contained in both electrodes causes the lines of that 
metal to stretch across from point to point. A somewhat longer exposure than usual 
fills the intervals between the lines with a weaker continuous spectrum. The period 
was three minutes. The extent of the continuous rays is always noted in the de¬ 
scriptions of the absorption spectra. To secure well-defined spectra the photographs 
were taken with the solutions placed in front of the slit, and the rays were con¬ 
centrated by a condensing lens of two inches diameter and three inches focal length. 
