Photographs of the Spark Spectra of Elementary Substances, 233 
altogether. When the condensing surface is too small the sparks are not sufficiently 
brilliant, a lengthened exposure is then necessary to obtain photographs, and these 
are frequently wanting in faint lines and minute details; moreover, an increased 
number of lines are discontinuous. 
When the battery power and coil are of the right proportion, but the condensing 
surface too large, the sparks are delivered slowly, but they possess too great a 
briliancy. The effect of this is to obliterate the distinction between short and 
long lines, lines usually discontinuous are made to extend in an unbroken line 
across the spectrum. The rapidity of the passage of the sparks should be at least 
at the rate of 500 or 600 per minute. The battery and coil previously described 
by me (Proc., R. D. Soc., Vol. 3, New Series), are suitable for such sparks. 
The distance between the electrodes is measured by a gauge consisting of a piece 
of plate glass about ¢ of an inch or 5™" in thickness. This is placed between the 
points, and they are made to touch it, when it is withdrawn without disturbing 
them. 
The slit aperture is measured by a micrometer screw, it may vary between x55 
and soo of an inch in width or be diminished in special cases to roo with advantage. 
Particular care is necessary to keep the slit free from dust and the condensed 
vapour of volatile metals. The use of a condensing lens of short focus will prevent 
the condensation of metallic vapours on the slit. I have at various times used 
a condensing lens of three inches focal length, but I prefer to place the spark close 
to the slit, the slit itself being covered with a thin plate of quartz. This not only 
keeps off metallic vapours but also particles of dust. It is, of course, cleaned 
simply by wiping with a leather. There is really no difference in definition between 
photographs taken with and without a condensing lens. 
When the metallic points wear away by reason of the excessive volatility of a 
metal, sometimes a portion of the spark does not play directly opposite the slit. 
In such a case the quartz plate is of the greatest utility, as by a glance one can 
tell from the reflexion from its surface whether or not the slit is completely covered 
by the spark. Good, thick, and broad electrodes are commonly productive of the 
most uniform results. The character of the lines varies with the conductivity and 
volatility of the electrodes; their intensity, as a rule, increasing with these pro- 
perties. The more volatile a metal the more continuous are its lines. The spectra 
of magnesium, zinc and cadmium are examples of this, while that of mercury is an 
extreme case. The length of spectrum obtained with the lens’of thirty-six inches 
focal length between the air line just visible, with a wave-length of 464-4, and the 
cadmium line 23, wave-length 231°8, is nearly 6:4 inch (160™"), with the fifteen 
inch lens it is 3°2 inches (or 80"). If the prism be placed at the minimum 
angle of deviation for any line less refrangible than No. 17 Cd. the spectrum will 
be longer, but the definition of the more refrangible lines will be defective. 
The highly refrangible nature of the extreme lines of cadmium, which throws 
2Q2 
