THE EFFECT OF CERTAIN IMPURITIES ON THE SPECTRA OF SOME GASES.

By PERCIVAL LEWIS.

IT has frequently been observed that under certain conditions small traces of a foreign substance in a gas may affect the spectrum of the latter to an unexpected extent. Heretofore little systematic investigation of such phenomena has been made, and it was thought that it might be of interest to examine the interactions of some substances which are rarely or never absent from vacuum tubes, namely, mercury vapor, hydrogen, oxygen, and water vapor. The following cases have been studied:

I. The spectrum of hydrogen.—(a) Pure; (b) containing traces of mercury vapor; (c) containing traces of oxygen; (d) containing traces of water vapor.

2. The spectrum of oxygen.—(a) Pure; (b) containing hydrogen; (c) containing mercury vapor.

The method of investigation was first to observe the spectrum of the gas in as pure a condition as possible, particularly with respect to the substances whose influence was to be studied. It was impracticable to note all details, but the general appearance of the spectrum was observed, and photometric measure

ments of the luminosity of selected parts or lines made, at various pressures of the gas. Small quantities of the foreign substances were then introduced and the observations repeated.

In order to guard against disturbances due to the absorption or giving off of gases by internal metallic electrodes, external electrodes, such as those described by Salet,' were used. For this reason no measurements of current strength were practicable.

THE APPARATUS.

The general arrangement of the apparatus is shown in Fig. 1. Hydrogen is generated in the voltameter V from distilled water containing a small quantity of phosphoric acid. The traces of oxygen always present in hydrogen generated in this manner were removed by passing through a strong solution of pyrogallic acid in the vessel A. The gas is stored in the drying tubes C, containing respectively calcium chloride, solid potassium hydroxide, and phosphorus pentoxide. G is a sulphuric acid valve to exclude mercury vapor coming from the pump, and F is a tube containing solid potassium hydroxide to absorb the vapors coming from the sulphuric acid. The vacuum tube D is of the H-shaped end-on type. The electrodes are made of four pieces of brass tubing separated from the glass by mica. Without this precaution the tubes. were invariably sparked through, owing to the high potentials used. The capillary part of the tubes generally used was about 10 cm long and 4 mm internal diameter. E is a reservoir containing mercury, which, by opening a cock, may be admitted into the discharge tube at a pressure corresponding to the temperature of the reservoir, which was always lower than that of the vacuum tube. B is a glass bulb containing potassium permanganate from which oxygen may be generated by heat.

At first all parts of the apparatus were fused together. This gave rise to inconvenience on account of frequent change or renewal of parts of the system, and finally several sealingwax joints were used. No injurious effects were noticed unless. 'SALET, Ann. Chim. et Phys., 28, 20, 1873.

the discharge reached the joints, when carbon monoxide bands always appeared. The same was found to be true of stopcocks, which were avoided at first on account of the grease, the gas under investigation being introduced through a barometer tube. In studying the effects of mercury vapor this method was of course impracticable, and stopcocks were necessary. At low pressures the carbon monoxide bands always appeared after the discharge had passed for several minutes. In these experiments, however, the discharge tube was always connected with the pump, and observations were made on fresh gas which had not had time to become contaminated by vapor from the cocks. The diffusion of the vapor was also hindered by capillary contraction in the tubing between the discharge tube and the cocks.

A small induction coil was used which, under the usual conditions, gave a spark of some 5 cm length. It was fed by a current from the city leads, at a potential of 110 volts, through a Wehnelt interrupter. The tube with external electrodes is practically a condenser, and it was found that the interrupter would not work so easily and uniformly as with a closed secondary circuit. Furthermore, it was impossible to always maintain constant conditions, owing to frequent renewal of the interrupter, change in the concentration of its acid solution, etc. During a given series of experiments, however, the conditions usually varied little, and different series have been as nearly as possible reduced to the same scale by assuming that in a pure gas at a given pressure in a given tube, the luminosity is proportional to the current strength.'

Photometric measurements were made with a Glan spectrophotometer, with an ordinary glowlamp, behind a screen of oiled paper, as a standard. As relative values only were required, the luminous intensity was taken as equal to the square of the tangent of the angle read multiplied by a convenient constant factor.

RESULTS.

I. Hydrogen. The first experiments were made for the purpose of determining the relation between luminosity and pressure in the case of pure hydrogen. The luminosity is also a function. of the current strength, which in these experiments was unknown; but the source of the current supply was kept as constant as possible, so that the results are comparable.

Measurements were usually made on fresh hydrogen, so that in all probability the results are very little affected by the presence of carbon compounds or gases given off by the glass. In some cases observations were repeated on the same portion of hydrogen, with its pressure reduced by pumping out. Usually such readings (indicated by the sign [-] in the table of results) were a little smaller than those made on fresh hydrogen.

For the sake of more accurate photometric measurements, a slit of about 5 mm width was used. For this reason, the compound spectrum of hydrogen, which was invariably present, appeared as a group of wide bands in the red and orange, and as a continuous spectrum in the green. Photometric measurements were made on Ha (λ=6563), HB (λ=4861), and the compound spectrum in the neighborhood of the green mercury line (λ= 5460). Only a few observations were made on HB, as its luminosity seemed to follow the same law as Ha, and measurements in this part of the spectrum are difficult. The results are given on page 142.

These results for Ha and Hẞ are plotted in the curves (I. Ha) and (I. HB), and for the compound spectrum in the upper sinuous curve (I. H11) of Fig. 2. They show that, keeping the conditions of current supply constant, the luminosity of both the elementary and the compound spectrum of pure hydrogen reaches a maximum at about 3 mm pressure, and then diminishes very rapidly with the pressure. With increasing pressure the elementary spectrum diminishes in luminosity faster than the compound, and at pressures of more than 4 or 5 cm the latter alone remains visible, although too feeble to be measured. When these experiments were repeated with a tube having internal electrodes, the