machine stopped turning. The spot was off the scale, and nine minutes elapsed before it appeared on the scale. The first reading on the curve was taken one minute afterwards, or ten minutes after the machine stopped turning (35.25 volts).

Curve 3.

March 12, 1894.-A Voss induction-machine was joined to the charging wire, and run by an electric motor

for 4 hours 19 minutes. A test was applied at the beginning of the run to make sure that it was charging negatively; and a similar test when it was disconnected from the charging wire in the vat showed it to be still charging negatively. The water-dropper was joined to the electrometer, and the spot appeared on the scale immediately. The first reading on the curve was taken half a minute after the machine was disconnected (30.65 volts)

Curve 4. April 23, 1894.—The friction-plate machine was turned positive for 30 seconds, with water-dropper running and joined to the electrometer. 20 seconds after the machine stopped the spot appeared on the scale, and the reading 14 minutes after the machine stopped turning is the first point on the curve (73 volts).

Curve 5. April 23, 1894.-The friction-plate machine was turned negative for 30 seconds, with the water-dropper running and joined to the electrometer. 10 seconds afterwards the spot appeared on the scale, and the reading 70 seconds after the machine stopped turning is the first point on the curve (7·6 volts).

The curves show, what we always found, that the air does not retain a negative electrification so long as it retains a positive. We also found, by giving equal numbers of turns to the machine that the immediately resulting difference of potential between the water-dropper and the vat was greater for the negative than for the positive electrification; though the quantity received from the machine was probably less in the case of the negative electrification, because the negative conductor was less well insulated than the positive.

§ 10. On the 21st of March, two U-tubes were put in below the edge of the vat, one on either side, so that it might be possible to blow dusty, or smoky, or dustless air into the vat. To one tube was fitted a blowpipe-bellows, and by placing it on the top of a box in which brown paper and rosin were burning, the vat was filled with smoky air. Again, several layers of cotton-wool were placed on the mouth of the bellows, so as to get dustless air into the vat. The bellows were worked for several hours on four successive days, and we found no appreciable difference (1) in the ease with which the air could be electrified by discharges from the wire connected to the electric machine, and (2) in the length of time the air retains its electrification.

But it was found that, as had been observed four years ago with the same apparatus*, with the water-dropper insulated and connected to the electrometer, and no electrification of any kind to begin with, a negative electrification amounting to four, five, or six volts gradually supervened if the waterdropper was kept running for 60 or 70 minutes, through air which was dusty, or natural, to begin with. It was also found, as in the observations of four years ago, that no electrification of this kind was produced by the dropping of the water through air purified of dust.

The circular bend of the tube of the water-dropper shown *Maclean and Goto, Philosophical Magazine,' August 1890.

in the drawing was made for the purpose of acting as a trap to prevent the natural dusty air of the locality from entering the vat when the water-dropper ran empty.

§ 11. The equilibrium of electrified air within a space enclosed by a fixed bounding surface of conducting material presents an interesting illustration of elementary hydrostatic principles. The condition to be fulfilled is simply that the surfaces of equal electric "volume-density" are surfaces of equal potential, if we assume that the material density of the air at given temperature and pressure is not altered by electrification. This assumption we temporarily make from want of knowledge; but it is quite possible that experiment may prove that it is not accurately true; and it is to be hoped that experimental investigation will be made for answering this very interesting question.

12. For stable equilibrium it is further necessary that the electric density, if not uniform throughout, diminishes from the bounding surface inwards. Hence if there is a portion of non-electrified air in the enclosure, it must be wholly surrounded by electrified air.

§ 13. We may form some idea of the absolute value of the electric density, and of the electrostatic force in different parts of the enclosure, in the electrifications found in our experiments, by considering instead of our vat a spherical enclosure of diameter intermediate between the diameter and depth of the vat which we used. Consider, for example, a spherical space enclosed in metal of 100 centim. diameter, and let the nozzle of the water-dropper be so placed that the stream breaks into drops at the centre of the space. The potential shown by the electrometer connected with it, being the difference between the potentials of the air at the boundary and at the centre, will be the difference of the potentials at the centre due respectively to the total quantity of electricity distributed through the air and the equal and opposite quantity on the inner boundary of the enclosing metal; and we therefore have the formula::

where V denotes the potential indicated by the water-dropper, a the radius of the spherical hollow, and p the electric density of the air at distance r from the centre. Supposing now, for example, p to be constant from the surface to the centre (which may be nearly the case after long electrification as performed in our experiments), we find V=πра2; whence p=3V/2πα2.

To particularize further, suppose the potential to have been 38 volts or 0.127 electrostatic C.G.S. (which is less than the greatest found in our experiments) and take a=50 centim. : we find p=2·4 × 10-5. The electrostatic force at distance r from the centre, being pr, is therefore equal to 10-*r. Hence a small body electrified with a quantity of electricity equal to that possessed by a cubic centimetre of the air, and placed midway (r=25) between the surface and centre of the enclosure, experiences a force equal to 24 × 10-9 25, or 6 x 10-8, or approximately 6 × 10-5 grammes weight. This is 4.8 per cent. of the force of gravity on a cubic centimetre of air of density 1/800.

§ 14. Hence we see that, on the supposition of electric density uniform throughout the spherical enclosure, each cubic centimetre of air experiences an electrostatic force towards the boundary in simple proportion to distance from the centre, and amounting at the boundary to nearly 10 per cent. of the force of gravity upon it; and electric forces of not very dissimilar magnitudes must have acted on the air electrified as it actually was in the non-spherical enclosure used in our experiments. If natural air or cloud, close to the ground or in the lower regions of the earth's atmosphere, is ever, as in all probability it often is, electrified to as great a degree of electric density as we have found it within our experimental vat, the natural electrostatic force in the atmosphere, due as it is, no doubt, to positive electricity in very high regions, must exercise an important ponderomotive force quite comparable in magnitude with that due to difference of temperatures in different positions.

It is interesting to remark that negatively electrified air over negatively electrified ground, and with non-electrified air above it, in an absolute calm, would be in unstable equilibrium; and the negatively electrified air would therefore rise, probably in large masses, through the non-electrified air up to the higher regions, where the positive electrification is supposed to reside. Even with no stronger electrification than that which we have had within our experimental vat, the moving forces would be sufficient to produce instability comparable with that of air warmed by the ground and rising through colder air above.

§ 15. During a thunderstorm the electrification of air, or of air and the watery spherules constituting cloud, need not be enormously stronger than that found in our experiments. This we see by considering that if a uniformly electrified globe of a metre diameter produces a difference of potential of 38 volts between its surface and centre, a globe of a kilometre

diameter, electrified to the same electric density, reckoned according to the total electricity in any small volume (electricity of air and of spherules of water, if there are any in it), would produce a difference of potential of 38 million volts between its surface and centre. In a thunderstorm, flashes of lightning show us differences of potential of millions of volts, but not perhaps of many times 38 million volts, between places of the atmosphere distant from one another by half a kilometre.

XXV. Preliminary Note on the Spectrum of the Electric Discharge in Liquid Oxygen, Air, and Nitrogen. By Professors LIVEING and DEWAR*.

N making the experiments here described we desired, if possible, to observe the emission-spectra of the several substances, stimulated by the electric discharge, while at temperatures of 180° to 200° below zero. It seemed probable that the characters of these spectra would give some indications of the physical state of the substances operated on.

In order to prevent the rapid heating of the electrodes by the discharge, they were made of considerable size. One was a disk of platinum about one centimetre in diameter, convex on one side, and having its convexity turned towards the other electrode, which was made of a piece of platinum wire about two millimetres thick. Even these electrodes were much heated, became red-hot when they were not in the liquid, but the spark passed through the gas immediately above the liquid. When actually immersed in the liquid they could hardly have been, except locally at the point of discharge, at any temperature sensibly different from that of the liquid. Experiments were made also with electrodes of aluminium, but with no different results as regards the spectrum except the introduction of the shaded bands due to alumina instead of the lines of platinum.

The liquids experimented on were contained in double testtubes of large dimensions, having the space between the two tubes highly exhausted. The electrodes, insulated, except at the extremities, by glass tubing and wax or gutta-percha, were passed through a rubber-stopper which closed the tube. Through this stopper was also passed a glass tube, which was left open while experiments were made at the atmospheric pressure, but was connected with a powerful rotary air-pump when it was desired to exhaust the gas in the tube.

* Communicated by the Authors.