disturbed, namely, systems in which the satellites (negative electrons) are arranged in rings and those in which the satellites move independently. Under severe stresses a ring system might break down into a system having independent satellites. Considering the ring system as discussed by Thompson as the general type of atom, metallic as well as nonmetallic, we find nearly all the phenomena related to secondary spectra capable of simple interpretation. We should have primary spectra given off by the lateral oscillations of the rings when their steady motion was disturbed, secondary spectra by the independent satellites after the rings had been broken up by violent excitation. The ring systems of the atoms of acid-forming elements are not easily broken up, while those of the metals are already broken up before sufficiently excited to become luminous. The Zeeman effect would be shown only by secondary spectra and the wave lengths of the lines in the primary bands would fall under quite a different spectral series from those of the secondary. Lines of a secondary spectrum would not be related in position to those of the corresponding primary. A nitrogen atom would have as many rings as there are bands in its primary spectrum. Critical capacity would vary with the wave length and become infinite for short waves, because the larger, slower rings would be the first to break up, while the innermost rings could not perhaps be broken up at all. The abrupt and complete transition from primary to secondary spectrum shown by nitrogen, sulphur, iodine, and bromine would indicate that when the discharge exceeds a certain intensity the rings on all the atoms are broken up at once and remain so, while in hydrogen only a portion of the rings would appear to be disrupted at one time, probably on account of very rapid recombination.

EXPLANATION OF PLATES.

PLATE I. The upper series of spectra are of nitrogren at 1.5 mm pressure, showing the transition from primary to secondary by gradually adding capacity in steps of 0.03 m. f. The first spectrum of the series (c) is of the white cathode glow. The second spectrum was taken with the same capacity (0.06 m. f.) as the fifth, but taken in the bulb instead of capillary. The bulb spectrum is seen to be a nearly pure primary, while the capillary is a secondary in the green and yellow, illustrating very well the effect of current density on the production of a secondary. The third spectrum (p) is a pure (anode) primary, while those beneath were taken with the capacity indicated. Drawing a line roughly separating primary and secondary, this curve represents critical capacity as a function of wave length.

The lower series shows the same capacity effect in hydrogen. The capacity used is indicated, the bottom spectrum being a nearly pure primary.

PLATE II. The upper series shows the same capacity effect in sulphur. The first two spectra are both primary, but the upper of the two was taken when the vapor density was greater.

The lower series of spectra shows the current density effect in hydrogen. The spectrum at the bottom was taken with a very feeble current and long exposure, the second from the bottom with a greater current, the third with the largest current (0.06 amp.) the tube would carry. The two upper spectra were taken with the capacities indicated. He appears even with a feeble steady current, but Hy, H8, and He appear only with heavy current or capacity.

PLATE III. This plate shows the effect of inductance in hydrogen and nitrogen in bringing the secondary back to the primary. In each case the secondary at the bottom was taken with a large capacity (0.1 m. f.). The spectrum just above was taken with the same capacity with a small inductance (0.1 millihenry). The next spectrum was taken with the same capacity and a large inductance (0.8 millihenry). The topmost spectrum is a pure primary, in each case taken with reither capacity nor inductance.

94

i effect in hydrogen. The urrent and long exposure, the

e third with the largest enirent upper spectra were taken with the

a teeble steady current, bot Hy, Hs and pacity.

Teet of inductance in hydrogen and nitrogen in primary. In each case the condary at the bot ity 0.1 m. f.). The sper Pojist above was t. n

a small inductance : 1.

hry). The ne

*tance (0.8 m

...ken with reither e