Experiment 10.-Air in the glass cylinder saturated with moisture:

Another series of observations made, under slightly altered circumstances, gave the following results :

Experiment 13.-Air saturated with moisture:

Hence the deflection is always greater in dry air than in that which is saturated with moisture. Even when dry air was compared with air in the working-room, which was far from being saturated with moisture, the deflection was greater in dry air. In an experiment on this point a deflection of 8 divisions was obtained for the dry air, and for the undried air a deflection of 6.1 divisions.

It does not seem very easy clearly to make out the causes why the greater deflection was obtained with the spark in dry air. The following circumstance would have to be taken into account. It is necessary for the formation of the sparks in dry air that the electric density upon the pole-surfaces be greater than when the air is moist. Before the formation of sparks, the electrical density increases upon the polar surfaces until it is great enough to traverse the layer of air between them. Hence in moist air the spark appears sooner and with a smaller electrical density. The disintegration of the polar surfaces thereby becomes smaller, and consequently there is a diminution in the electromotive force of the disjunction. As the conductingpower of the spark doubtless depends on the quantity of detached metallic particles, this will also be smaller. Now, if a diminution of the electromotive force and of the conducting-power take place when the air is moist, a decrease in the magnitude of the deflection must necessarily follow. The deportment with rarefied gases seems to speak in favour of the same mode of explanation.

Experiment 16.-The glass cylinder was filled with carbonic acid, exhausted, and these operations repeated several times until it was certain that no air was left. The carbonic acid, which was prepared from marble and hydrochloric acid, and was ascertained by testing to contain scarcely perceptible traces of foreign gases, was dried before entering the apparatus previously mentioned. The polar cones in the glass cylinder had to be pushed very near together, because otherwise the spark would not pass when the cylinder was filled with carbonic acid. This is the reason why the deflections were relatively small.

With carbonic acid the following deflections were observed:

Experiment 17.-The glass cylinder was filled with air (un

dried):

A few other other observations gave the same result—that is, a considerably greater deflection for carbonic acid than for air. Experiment 19.-The glass cylinder was filled with hydrogen which had been dried before entering the cylinder. The following deflections were thereupon observed:

Experiment 20.-The cylinder was filled with air:—,

With hydrogen, then, the deflection was considerably greater than with air. When, on the contrary, the cylinder was filled with coal-gas, the deflection was but very little greater than with atmospheric air. With the former mixture the deflection 13.5 was observed, and with the latter 11.9.

3. In order to investigate the dependence of the disjunctioncurrent on the density of the gas in which the spark is formed, experiments were made with atmospheric air, carbonic acid, and

sphere. mm. mm.

coal-gas. The two former were dried, but the latter not. The means only may here be given :

Pressure in the 1 atmo- 140. 80 glass cylinder.

40

20

mm.

min.

4 mm.

When the pressure was reduced from one atmosphere to 140 millims., the deflection diminished from 44-4 to 6.6 divisions, when it again increased until for a pressure of 4 millims. it was greater than for one atmosphere.

Experiments with dried carbonic acid led to an analogous result; the deflection was least for a pressure of 140 millims., after which it again began to increase :

With coal-gas there was also a diminution in the deflection when the pressure was diminishad, although here the variations were not so great as in the two previous gases. For this mixture there was obtained the following result:

23.8

39.2

27.1

26.3 24.9

Deflections for 40.1 28.6

coal-gas Mean

21.8 41.9 25.0 49.3 39.7 27.9 25.1 24.9 23.4 45.6

That the deflections first decrease and afterwards again increase when the pressure is diminished, indicates that there are several causes for these variations. The magnitude of the deflection depends on the electromotive force, on the conductingpower of the spark, and on its duration. That the electromotive force decreases with the pressure follows from the fact that the disintegration of the polar surfaces becomes less as the gas is rarefied, because the electrical density of the polar surfaces which is requisite for the formation of sparks diminishes with the pressure*.

* In a previous paper I assumed, without any experimental proofs, that the electrical spark is subject to no perceptible change when a voltaic

If, now, the conducting-power of the gas, as is probable, increases, and the duration of the spark lengthens, when the gas is rarefied, the result obtained (that the deflections first decrease and then increase) contains nothing inexplicable. Further investigations, however, are needed to enable us to decide with certainty whether the mode of explanation which has been indicated is admissible.

In connexion with the above, experiments were also made with some Geissler's tubes, in order to ascertain whether the disjunction-current could be perceived in them or not. With three of them, one (according to the label) containing oxygen, another hydrogen, and the third chlorine, distinct deflections were exhibited; while another, which contained carbonic acid, as well as one without label, gave no distinct proof of the existence of a disjunction

current.

4. The current in the voltaic luminous arc is well known to occasion a greater disintegration of the positive than of the negative pole. When two equal polar surfaces, between which the discharge from an electrophorus machine has been for some time taking place, are closely examined, it is easy to discriminate the positive polar surface from the negative; for the former appears more altered than the latter. Hence positive electricity is most active in the disintegration. As positive electricity readily issues from a sharp point without producing there a more powerful disintegration, it must follow that, when one pole consists of a point and the other of a plane disk at right angles to the plane of discharge, the disintegration is greatest when the positive current goes from the disk to the point. When, therefore, the discharge goes through the spark from the disk to the point, it is to be expected that the disjunction-current will be stronger-partly because the electromotive force of the disjunction increases with the disintegration, and partly also because the quantity of particles detached from the poles is greater, and therefore the conducting-power of the spark is better.

To test the accuracy of what has here been said, a round brass disk 2.7 millims. in diameter was fastened on one of the two metal rods of the glass cylinder, and upon the end of the other rod a conically sharpened brass point was screwed.

current traverses it in either direction; and on this assumption I have based a method of directly measuring the electromotive force of disjunction. This assumption has been found to be incorrect. The spark undergoes a considerable change by the passage of the voltaic current, so that the determinations obtained can only be regarded as valid for the case in which a voltaic current traverses the spark.