tubes, b, at each end, fitting over the rod and within the tube, snugly. Thus the cylindrical shell (air-space) is but ∙141 centim. thick and 50 centims. long; and the condenser is adapted to bring to bear a high-potential gradient (the shell of air being as thin as admissible in consideration of flexure) along a considerable length of air-column conveying phosphorus nuclei. The outside of the condenser (fig. 1) is permanently put to earth, while the rod inside is in metallic connexion with the electrometer-needle, as suggested in the figure at E. A suitable key is added in order to conveniently Notcharge the rod. Upwards of 300 volts were applied. withstanding its slender dimensions, no difficulty attends the use of the condenser.

The air from a gasometer-train, after passing the desiccator D, and the stopcock F, is charged with nuclei by the ionizer, P, containing pellets of phosphorus between strips of wire-gauze. It then traverses the condenser L K, and thence enters the colour-tube. The gasometer has already been described in the preceding papers of the present series.

3. The method of observation was as follows:-The two sets of data to be obtained, namely the readings at the volumeflask and at the electrometer, were taken in consecutive minutes, the stopcock F having been previously adjusted to show any desired colour in the steam-tube. From observations at 15 seconds apart at the gasometer, I obtained two sets of results for the discharge of air per second, the pressure of the jet (p), the temperature of the inflowing air (0, by the thermometer T in the figure), the pressure of the air in the gasometer, and its temperature.

Meanwhile the condenser had been charged to its full potential, and after breaking the charging-circuit the leakage of the electrometer was observed each 15 seconds. furnished two sets of data for the initial electric current.

This

4. The experiments with the train of apparatus described nsually progressed smoothly. The data of two out of four similar series are given in Table I., which is divided into three parts, referring respectively to the steam-jet, the gasometer (volume conveyed per minute, V being the essential datum), and the electrometer. It was thought superfluous to reduce the air-volumes to normal conditions, as the fluctuation of pressure (pressure-excess, p) and temperature (0) are relatively insignificant in view of the variability of the ionizer. For the steam-jet the pressure, p, and temperature, 0, are also practically constant. Relatively to the electrometer series, the two initial potential-differences, E, are given as obtained immediately after, and 15 seconds after charging. These are

needed in the reductions. The potential in these experiments usually begins with about 40 volts. The table also contains the value of E/E' for observations half a minute apart, two of these being available for each colour as specified. The current was computed as follows:-For a very short time after charging the fall of potential is appreciably logarithmic. Hence E/E'-10-(-) may be assumed, where E and E' are the potential-differences at the times t and t' in minutes, and a constant for a given colour in the tube. Hence

—C(dE/dt)=CEx. log 10=60i,

if C is the aggregate capacity of the electrometer and the condenser, and i the current flowing radially through it. Since

log (E/E' = x(t-t'), or -x=8 log E/dt,

the table furnishes these data with E, to compute i relatively; and it was thought superfluous to supply the factor Clog 10/60, sharply. Its estimated electrostatic value is about 230/60, as the combined capacity of the electrometer (60 cm.) and the condenser (39 cm.) is about 99 cm., roughly. This would make the capacity 1.1/1010 farads, and the factor is 4.2/1012. The table shows a few values of i in coulombs; but the quantity a, being more immediately derived from the observations, will be chiefly discussed below, and has therefore been constructed in terms of V, the air-volume per minute, in fig. 3. The colours seen in the steam-tube are added to each point, as explained in the table.

5. The effect of the presence or absence of this electrical current on the colour-tube is quite inappreciable, no change of tint or suggestion of change occurring on charging the condenser. I gave the question an additionally severe test (up to 300 volts), showing that only an insignificant part of the nuclei present take part in the convection of electricity radially across the condenser, as detailed in an earlier paper.

6. On the other hand, however, the current varies with the number of nuclei present, no matter how small the relative quantity actually promoting electric conduction. This is fairly well shown in the chart (fig. 3) already described, which graphically reproduces Table 1. Finally, in fig. 4 the radial electrical currents are given as ordinates, the corresponding volumes per minute (V) of the air passing longitutudinally through the condenser as abscissas. The relation of saturated litres per minute longitudinally and coulombs per second radially is thus apparent for the second and fourth series. The other series, if supplied, would add nothing new.

TABLE I.-Coincident Registry of the Gasometer,
Electrometer, and Colour-tube. Stminute.

Yl', YI" are yellows of the first and second orders respectively. Op. denotes opaque. Blue + and Blue

are dark and light blues. The letters Y.

O, V, B+, B, G, Y2, C are used in the chart (fig. 3).

In other words, it is noteworthy that the colours referred to volumes of charged air per minute (abscissas, V) are roughly in agreement; whereas the electrometer data referred to the same colour differ. Some variation of the constant of the electrometer, if in question, would have shown itself in the

30

Each series conforms pretty closely to a straight line, but these lines differ in inclination for the different series. The colours corresponding to the observations are added in a sug gestive way. Their real positions are not the same in all series, electrically, but they are more nearly so, volumetrically.