58 
PROFESSOR W. N. HARTLEY ON SPECTRUM PHOTOGRAPHY 
as the electrode is nearly immersed in water the intensity of the spark is increased. 
Hence, though the quantity of the element present in the solution might be insufficient 
to render the short lines visible with dry electrodes, the opposite effect introduced 
by the presence of water is sufficient to compensate for this. 
In order to ascertain whether the elements oxygen and sulphur in combination 
could yield spark spectra, and whether insoluble compounds could be made to yield 
metallic lines to the spark, ferric oxide and ferrous sulphide were examined. The 
substances were finely powdered and mixed with glycerine to prevent them being 
dispersed by the spark too rapidly. No iron lines were detected on the photographic 
plate exposed to the action of the spark for the normal period of two minutes, though 
there was a good photograph of the graphite electrodes and the air spectrum. 
To test the behaviour of insoluble but volatile substances, thallous chloride was 
treated in the same way, and the lines with wave-lengths 3778*4 and 3518’8 were 
rendered weakly. In no case did the non-metallic constituents cause any variations 
in the spectra. Hence we may conclude that:— Insoluble and non-volatile compounds 
do not yield spark spectra when diffused in liquids. 
Several attempts to obtain a spectrum from selenium, selenic acid, and sodium 
selenate have proved unsuccessful. 
On the spectra of ternary compounds. 
In an examination of ternary compounds the salts examined were sulphates, 
nitrates, and phosphates of magnesium, cadmium, zinc, aluminium and iron. The bases 
of these salts were, as in the former case, prepared from the metallic electrodes, the 
aluminium and iron compounds being prepared from pure aluminium hydroxide and 
ferric oxide. The ferric oxide was obtained by heating pure ferrous sulphate with 
sodic sulphate, lixiviating with hot water, and washing many times therewith by 
decantation. The oxide was then dissolved in the requisite acids. The observations 
recorded in the case of metallic chlorides apply equally well to sulphates and 
nitrates, while the difficulty of obtaining a spectrum from ferric phosphate diffused 
in glycerine was equal to that in the case of the sulphide. When dissolved in hydro¬ 
chloric acid the phosphate displayed the lines of iron only. A similar observation 
was made with cerous phosphate. Ammonium chloride was made into a saturated 
solution, and its spectrum photographed. There was no striking alteration in the 
spectrum of the graphite poles, but a careful examination showed that a group of fines 
which appear to be caused by the presence of nitrogen was greatly strengthened and 
made more prominent. These fines or flutings extend from about wave-lengths 3S81'8 
to 3829 '0. The following facts point to their origin. They are never seen when 
metallic electrodes are used, wet or dry ; they are therefore not air-bands. They are not 
seen when carbon electrodes are immersed in oxygen, and cannot therefore be due to 
carbon dioxide. They are always seen when solutions are used with carbon electrodes, 
