474 
PROFESSOR W. N. HARTLEY ON THE 
Method of determining the wave-lengths. 
It was usual to take a series of five or six photographs of each solution on one plate, 
and thus the spectra represent the rays transmitted by 1,2, 3, 4, 5, or more millimetres 
of the absorbent liquid. The wave-lengths referred to are derived from those given 
in the Phil. Trans., Yol. 175, p. 63, 1884, for cadmium, tin, and lead The scale 
numbers in hundredths of an inch are arbitrary, and quite different from those given 
beside the wave-lengths in the paper quoted; they were actually read off from photo¬ 
graphs of the electrodes by means of an ivory scale, the end of which coincided with 
the end of the glass plate. The scale originally used was a photograph of the spectrum 
of the electrodes with divisions etched on the glass, the gelatine film being scraped off 
just below the lines to admit of the glass being etched. The divisions were hundredths 
of an inch, and every fifth line was distinguished by the corresponding wave-length 
being written on the back of the plate opposite to the longer line. The figures, 
however, are too small, and, otherwise, from the transparency of the glass difficult to 
read when applied to the face of another photograph. It was found more satisfactory 
to apply an ivory scale with a bevelled edge, divided as aforesaid, so that- the scale 
numbers arbitrarily fixed first for the cadmium lines are read off on all the other lines. 
The scale is held in position by a pair of wooden spring clips. Supposing an absorp¬ 
tion hand occurs near the line 120 (X=3245\5), then 120 on the scale is brought over 
this line on the photograph; if it be seen to extend to 240 or thereabouts (X=2593), 
the reading on the scale is brought up to this line, and a more exact measurement is 
made. If the adjustment is not further from the correct number than xcToth °f an nick, 
it is sometimes not worth while to alter it ; but whenever it is necessary to make ac¬ 
curate measurements, the scale is capable of easy and perfect adjustment to any part of 
the spectrum. For the purpose of converting the scale measurements into wave-lengths 
and their reciprocals, two curves have been drawn on one sheet of paper. The scale 
numbers, wave-lengths, and oscillation frequencies of the fiducial lines employed for 
the interpolation curves are stated on p. 475. Fractions of a tenth-metre have been 
omitted as being unnecessary in dealing with absorption spectra. 
On the diagrams the scale of wave-lengths is reduced to millionths of a millimetre. 
