independence of the nature of the gas-demonstrated, in fact, by Emden-and its independence of temperature, which has not yet been investigated.

§ 13. The acoustical phenomenon accompanying the rush of gas through a slit has been studied experimentally by Kohlrausch (Wied. Ann. xiii. p. 545 (1881)) with respect to width of the slit and to pressure in the reservoir. The influ ence of the pressure outside, of temperature, and nature of gas, might be inferred by our method, too, from this research; but we do not enter into this matter, as the results, not fitting easily into analytical expressions, are represented by tables, which would be rather cumbersome for use.

§ 14. In the preceding investigations the effect of external forces has been entirely neglected-whilst there exist certain classes of phenomena where gravity plays a prominent part; for example, motions of the earth's atmosphere or convective currents produced by inequalities of temperature. By considerations analogous to those in § 5 we get three conditions for similarity to be fulfilled in such cases :

m

m2=n=ah; b=ß— he ;

n

(7)

Let us examine, in this respect, Lorenz's result concerning the amount of heat given off by 1 cm. of a vertical plane [height H, breadth infinite, temperature 9, above that T of the surrounding gas] which was evaluated approximately (Wied. Ann. xiii. p. 592, 1881):

The form of this expression looks rather peculiar; but we satisfy ourselves of its dimensions fulfilling the conditions of similarity-as far, however, only as the coefficient e is neglected, which points to a serious restriction of its validity.

§ 15. The presence of those convection-currents gives much trouble in the determination of thermic conductivity of gases. Their influence can be diminished by rarefying the gas; but rarefaction below a certain limit of pressure would imply another source of errors in certain molecular "discontinuities of temperature," as I have called these phenomena (Phil. Mag. xlvi. p. 192, 1898).

Now it may be noticed that relative measurements of conductivity can be strictly performed, notwithstanding the unknown convective currents, by using corresponding

pressures and corresponding dimensions of vessels for different gases (according to h=1; n=a; b=

В

Also the thermic variability of conductivity-not yet known with desirable precision-may be investigated in an analogous manner, by application of similar motions. If we make use, for the higher temperatures, of vessels with dimensions increased in proportion of the first, and of pressure increased in proportion of the (e-)th power of temperature, the quantity of heat transferred must be proportional to 0, whence e may be determined. The method of heating wires by electric currents may be easily adapted to this way of experimenting.

We confine ourselves to these few examples on this sort of similarity, since its range of applications is less extensive and since there is little experimental work hitherto done which could serve as a basis for further speculations.

LXXV. The Condensation Method of Demonstrating the Ionisation of Air under Normal Conditions. By C. T. R. WILSON, M.A., F.R.S., Fellow of Sidney Sussex College, Cambridge*.

SOME years ago with water I described experiments + which proved that when air saturated with water-vapour has been freed from dust particles, it will still give condensation in the form of drops on sudden expansion provided the expansion exceeds a definite limit. If v1 v2 be the volume of the air before and after the sudden expansion, then if v/v be less than 1.25 no drops are produced on expansion, but if this critical expansion be exceeded a rainlike condensation results. The drops remain comparatively few if v/v does not exceed a second limit about 1:38. It was found that exposure of the air to Röntgen or other ionising rays increased enormously the number of drops produced by expansions between these limits, the least expansion required to cause the formation of drops remaining, however, the same. It was concluded that the nuclei giving the clouds in air exposed to Röntgen rays are to be identified with the ions to which its conducting power under the action of the rays is attributed, and that the few drops always produced with expansions exceeding the critical value are due to ions of the same nature continually being produced even in the absence of the rays.

*Communicated by the Author.

+ Phil. Trans. vol. clxxxix. p. 265 (1897).

Further experiments showed that the number of drops produced by expansions between the above-mentioned limits in air exposed to Röntgen rays, is reduced in a very striking manner when a sufficiently strong electric field is maintained across the air before expansion, thus proving that the nuclei move in an electric field and are therefore electrically charged, and presumably identical with the ions to which the conducting power is due. On the other hand, similar experiments made in the absence of ionising agents failed to show any diminution of the number of drops by the action of even very strong fields. The absolute identity of the degree of supersaturation required to cause condensation upon ions and upon the nuclei to which the rainlike condensation is due, made it difficult to believe that the latter are not ions also, and to explain their non-removal by an electric field it was suggested that they might be ions produced in some way as a result of the expansion. When, however, subsequent experiments † on the leakage of electricity from conductors suspended within closed vessels, showed that a continual slight ionisation of the air is always going on in such vessels, it appeared more likely that the rainlike condensation really is due to this ionisation, and that the failure to detect any diminution in the number of drops under the action of an electric field is due to some defect in the conditions of the experiments. In the experiments thus far made the vessels used had been small, and to permit of a strong electric field being applied the air was enclosed between conducting surfaces generally only a centimetre or less apart; in many cases one of the conductors was a layer of water at the bottom of the vessel, the other being a horizontal metal plate coated with wet filter-paper. The drops were under these conditions very few whether an electric field was applied or not; it was thought that if a much larger volume of air were used there would be more chance of detecting the diminution in number when an electric field was applied. This expectation has been realized. With the large apparatus described below the effect of an electric field in removing the nuclei which gave rise to the rainlike condensation is very striking.

The construction of the apparatus (shown in the figure) is the same in principle as in the experiments on condensation nuclei which I have described in previous papers.

Phil. Trans. vol. cxcii. p. 403 (1899).

On

+ Geitel, Physikalische Zeitschrift, vol. ii. p. 116; C. T. R. Wilson, Roy. Soc. Proc. vol. lxviii. p. 151.

The apparatus was made by Messrs. W. G. Pye & Co., Cambridge. To Mr. Pye I am indebted for many suggestions as to the mechanical details.

account of the much larger size of the new apparatus, the mechanism by which the sudden expansion is produced was

an

constructed, however, of brass, not, as in the older experiments, of glass. The cloud chamber A, in which the drops formed by expansion are viewed, is a glass cylinder 18.5 cms. in internal diameter and 5.9 cms. high. Its roof consists of a thick brass disk cemented to it by means of sealing-wax. The cylinder rests on an indiarubber ring lying on annular brass plate F, which forms a flange at the top of the expansion cylinder B. The glass cylinder is squeezed down on the indiarubber by means of an upper annular brass plate R resting on the roof of the cloud chamber, from which it is separated by a second indiarubber ring; the upper and lower annular plates are connected by six bolts, by means of which the necessary pressure can be applied. The external diameter of the annular plates is 26 cms. ; about one cm. from the edge of the lower one on its upper surface a thin ring L of brass 12 cm. high is soldered. This serves to contain water, all risk of air leaking in below the edge of the glass cylinder being thus removed. Through three symmetrically placed tubes penetrating the lower plate of the cloud chamber are

sealed three insulated brass rods supporting a horizontal brass disk D, 15.3 cms. in diameter. Between this disk and the roof of the cloud chamber, 47 cms. above it, any desired difference of potential could be maintained by means of a battery of storage-cells. Both this disk and that forming the roof were covered on the surfaces facing one another with wet filter-paper. In addition to the three tubes through which pass the supports of the brass disk, the floor of the vessel is pierced by a fourth smaller tube T, by means of which air can be removed from or admitted into the apparatus.

Below the cloud chamber and supporting it is a vertical brass expansion cylinder B, 10 cms. in internal diameter and 30 cms. long. Sliding freely in this and serving as a piston is a thin-walled brass cylinder open below and with a hemispherical top, the length of the cylindrical part being 18.75 cms., the thickness of the walls being less than one millimetre. The expansion cylinder is bolted by means of a flange at its lower end against a thick brass disk, an indiarubber ring, of which the internal diameter is considerably less and the external diameter greater than that of the cylinder, being inserted between them. Rising up from the centre of the disk is a brass tube 18 cms. long and 1.3 cms. in internal diameter. The cylinder is filled with water to within a few cms. from the top of this tube. By means of the mechanism to be presently described, the central tube can be put into sudden communication with a vacuum chamber V, thus causing the piston to fly sharply down against the indiarubber at the bottom of the cylinder, and to remain pressed tightly against this so that no air or water can escape. It is in this way that the sudden expansion is produced on putting the central tube in communication with the atmosphere instead of the vacuum chamber, the piston rises to its original position.

The thick brass disk to which the expansion cylinder is attached rests upon an iron tripod (not shown in the figure) to the top of which it is firmly fixed by three screws. The feet of the tripod are screwed down to a board. The tubes for making connexion with the vacuum chamber are shown below the expansion cylinder. For convenience the connexions are made with screw-joints, indiarubber washers being inserted to prevent leakage. The vacuum chamber was a brass cylinder 22 cms. long and 14 cms. in diameter, with rounded ends; it was maintained at low pressure by a waterjet pump. A gauge was connected to avoid the risk of making an expansion while the vacuum was not sufficiently good. The construction of the mechanism for making sudden