ON SECONDARY SPECTRA AND THE CONDITIONS UNDER

WHICH THEY MAY BE PRODUCED.

By P. G. NUTTING.

a

Since early in the development of spectroscopy it has been well known that some of the elementary gases electrically excited give two entirely different spectra. One of these, called by Plücker and Hittorf the primary spectrum, usually consists of a large number of broad lines well distributed over the whole visible and neighboring ultra violet spectrum. The secondary spectrum, on the other hand, consists usually of but few lines, and these very prominent. It is the spectrum obtained when capacity is connected in parallel with the tube. The two spectra are easily obtained with an ordinary Plücker tube having not too long nor too fine a capillary and containing nitrogen, sulphur, or iodine vapor at a pressure of from 2 to 10 millimeters. With a large condenser (at least one-twentieth microfarad) in parallel, such a tube shows a secondary spectrum; without the condenser it shows the primary spectrum if the current be not excessive. Some gases which exhibit secondary spectra also show anode and cathode glows having quite different spectra. The present investigation was undertaken to separate, in the Mendeleef system, those elements which show multiple spectra from those that do not, and to determine what conditions govern the production of secondary spectra.

Starting with the seventh group of the periodic system, it was found that chlorine, bromine and iodine, all of the group that could be worked with, gave multiple spectra. The effect is particularly striking with bromine and iodine. Likewise all of the sixth group available gave multiple spectra, namely oxygen, sulphur, selenium and tellurium. In the fifth group nitrogen, phosphorus and arsenic exhibit

a Plücker and Hittorf: "On the Spectra of Ignited Gases and Vapors, with Special Regard to the Different Spectra of the Same Elementary Gaseous Substance.” Phil. Trans., 155, pp. 1-29; 1865.

multiple spectra while bismuth does not. With arsenic, however, the primary (anode) and secondary spectra have two prominent green lines in common. In the fourth group only the more metallic elements-tin and lead-could be used in a Plücker tube. The spectra of these was found not to change on connecting a condenser in shunt. In the third group aluminum, indium and thallium; and in the second group magnesium, zinc, cadmium, and mercury give only line spectra and these are not essentially altered by capacity in parallel. Hydrogen excepted, the first group shows only single spectra. Without taking up the much-discussed" question of the rôle of water vapor in gas conduction, from the point of view of multiple spectra the evidence is very strong that the fine-line spectrum of hydrogen is a true primary spectrum while the "four-line" is a true secondary. The latter is relatively enhanced under the same conditions that bring out the secondary spectra of other multiple-spectra elements. The primary and secondary spectra of other elements, notably chlorine and bromine, may be made to coexist in the same way, and the presence of water vapor certainly favors the preponderance of the secondary spectra of other elements as well as of hydrogen. Oxygen and sulphur are striking examples.

In the helium group there are wide differences. Helium itself shows but a single spectrum. Anode and cathode glows give identical spectra and both remain practically unchanged on the passage of the condenser discharge. Argon, however, shows at least four widely different spectra. The red primary (anode) and blue secondary spectra have been described by several observers. But the cathode glow shows quite a different spectrum from the anode (capillary) glow. With low dispersion it appears nearly continuous in the yellow, green and blue, while the strong red line and violet group of the anode glow are lacking. When the blue condenser discharge is produced in the capillary there is a buff-colored glow in the bulbs at either end. The spectrum of this glow shows the blue group lacking, but the yellow group of the red primary anode spectrum present. This difference between bulb and capillary spectra can hardly be regarded as other than a current density effect, since this is the chief, if not the only, difference in the excitation. As we should expect, the bulb spectrum with condenser is intermediate between the capillary spectrum with

a Callendar: Nat. 56, p. 624; 1897. Trowbridge: Phil. Mag. [6], 2, p. 370; 1901. P. Lewis: Phil. Mag. [6], 3, p. 512; 1902. Parsons: Astroph. J., 18, p. 112; 1903. Trowbridge and Richards: Phil. Mag. [5], 43, p. 77; 1897. Kayser: Berlin Akad., 1896; Travers, Expl. Study of Gases, 1901, p. 312.

с Baly: Chem. News, 88, p. 26; 1903.

condenser and the red capillary spectrum without condenser. Of neon, krypton, and zenon only the spectra of the two latter are said to be seriously affected by the addition of a condenser. No differences between anode and cathode capillary and bulb spectra appear to have been recorded.

In the accompanying Mendeleef table the elements showing multiple spectra are indicated by heavy-faced type. Elements showing distinct anode and cathode spectra are bracketed.

say

Excepting then the anomalous helium group and hydrogen, we may

that in general, so far as observations go, the acid-forming elements have multiple spectra, while metallic elements do not.

Most metallic spectra are greatly altered by the addition of capacity, it is true, but the changes produced are by no means comparable with the change from primary to secondary, while metallic spectra, by every test known, are already secondary spectra before adding capacity. They are series spectra, show the Zeeman effect and the displacement effect due to pressure, and the new lines brought in by added capacity are of the same (secondary) character. That so many metallie vapors exhibit banded absorption spectra would seem to indicate that metals may have primary spectra as well as nonmetals, but if so they are to be sought in the electrodeless discharge and cathode luminescence where the excitation is very feeble, rather than in the condenser discharge. Nor would the oxyhydrogen flame, banded spectra of the metals studied by Hartley and Ramage be comparable with primary spectra. In these spectra the lines of the series spectra simply become the heads of the bands in the flame spectra, while the lines of a true secondary spectrum appear to have no relation to the bands of the corresponding primary. Another notable distinction between the spectra of electro-negative and electro-positive elements is that the former, both primary and secondary, are remarkably invariant, while it is difficult to obtain the spectrum of a metal twice alike.

Why should a condenser discharge produce a secondary spectrum? The oscillation frequency of a condenser is so much less than the frequency of collision of an electron in a gas at 1 mm pressure that an electron must collide a great number of times before the impelling force is reversed. Hence the oscillatory nature of the discharge per se could have little effect on the spectrum of a gas. On the other hand, the current density during discharge is thousands of times as great as during the steady flow of the same current. In testing other gases for the effect of excessive current density, specially designed Plücker tubes were used. One form had a very fine (thermometer) capillary and hollow cylindrical electrodes; another had a third bulb interposed in the middle of the capillary. In this way the current density could be increased a hundred or thousand times in parts of the tubes. Bromine, iodine, and hydrogen easily show the current density effect within the range of current that a common glass tube will carry. The primary spectrum changes continuously over into the secondary spectrum as the current is increased. Nitrogen and the sulphur group failed to show the secondary spectrum with the largest steady currents that glass tubes could carry. Argon is so sensitive to changes in current density that, in conjunction with a pocket spectroscope, it might be used as a milliammeter.

It was further undertaken to determine how much capacity was necessary to just produce the secondary spectrum in different gases, how this critical capacity varies with the spectral wave length, with the density of the gas, the amount of inductance and resistance in circuit, distance apart of electrodes, and sectional area of the dis charge. In this work photographic methods were employed throughout. Spectra obtained under varied conditions were photographed side by side on the same plate so that the minutest changes could be observed and followed. For this purpose a large model Fuess quartz spectrograph was used. This was provided with a large flint glass prism giving a spectrum about 15 cm long from 300 to 600μμ. Ten or more spectra could be recorded on the same plate. A large glass condenser was used, composed of 20 plates well separated and provided with mercury cups, so that the capacity could be varied by a plate at a time. Current was supplied by transformers of 1,000, 2,000 and 5,000 volts, and by a set of generators giving 5,000 volts continuous current. For inductance, a Seibt tuning solenoid of 120 turns 20 cm in diameter was used. The greater part of the work was done with ordinary short, stout Plücker tubes made by Boehm, of Chicago. These had electrodes about 4 cm apart, and capillary por

a Davis: Phys. Rev., 17, p. 501; Dec., 1903.

tions 2 mm in diameter and 12 mm long. Some of the work was done with tubes without any central constriction. To keep the pressure of the inclosed gas more nearly constant, a half liter bulb was sealed to each tube while in use.

With a tube of air at 13 mm pressure spectra were photographed with capacities of 0.12, 0.09, 0.06, 0.03 microfarad and with no capacity in parallel. A sudden change from secondary to primary spectrum was found to occur at a capacity of about 0.05 microfarad, equivalent to that of about 14 one-gallon Leyden jars. Adding capacity indefinitely above 0.06 microfarad produced little, if any, effect, nor do secondary lines usually appear in the primary spectrum until the capacity is nearly 0.03 microfarad.

Critical capacity and wave length.-Drawing a line separating primary and secondary spectra on the photographic plate (see Pl. I) the ordinates of the curve represent roughly critical capacity, abscissas wave length. The curve drops off very steeply on the short wave length end in spite of the greater dispersion, indicating that for waves perhaps not shorter than 300μμ the critical capacity becomes infinite. Critical capacity expressed as a function of wave length appears to be of the form

Approximate numerical results for air are given in the table:

Critical capacity and pressure.-The critical capacity increases slightly as the pressure decreases down to about 2 mm air pressure, when it suddenly becomes infinite, i. e., no amount of capacity (without an external spark gap in series) will cause the secondary spectrum to appear. With spark gap, secondary spectra may be obtained at pressures of but a few tenths of a millimeter, nearly to the pressure at which cathode-ray fluorescence appears. This critical minimum pressure at which the disruptive discharge becomes possible is considerably