DISRUPTIVE DISCHARGE.] ELECTRICITY 61 Of ive, distance of 00254 cm., p= 11 290, whereas for a distance 1524, p = 535. It appears, therefore, that the dielectric strength of a thin stratum of air is much greater than that of a thick one. It is very difficult to understand why this should be so. " Is it possible that the air very near to the surface of dense bodies is condensed, so as to become a better insulator ; or does tbe potential of an electrified conductor differ from that of the air in contact with it, by a quantity having a maximum value just before discharge, so that the observed difference of potential of the conduc tors is in every case greater than the difference of potentials on the two sides of the stratum of airby a constant quan tity equivalent to the addition of about 005 of an inch to the thickness of the stratum 1 ?" 1 It is remarkable that the limiting tension should be so small, somewhere about half a gramme per sq. cm., as compared with the atmospheric pressure, which is about 1032 gm. per jq. cm. A series of absolute measurements of the potential re quired to produce a spark between equal spheres at different distances has been made by Mascart. The method em ployed was very ingenious. 2 Effect of Pressure, Temperature, $c., on the Dielectric Strength of Gases. The dielectric strength of a given gas depends on its pressure, or at all events on its density. Harris, who experimented on this subject, inclosed two balls in a receiver which could be exhausted to any required degree, and connected them with the armatures of a battery of jars. He found that the charge which had to be given to the battery in order to produce a spark between the balls was proportional to the density of the air in the receiver, while it seemed to be independent of its temperature. This amounts to asserting that the difference of potentials re quired to produce a spark between the balls is proportional to the density of the gas and independent of its tempera ture. Since we keep the distance between the balls the same throughout, this statement is equivalent to saying that the dielectric strength of a gas varies directly as its density, and does not depend on the temperature. Masson, using the method which Faraday had employed in com paring the dielectric strength of gases (vide infra} arrived at the same conclusion as Harris. Knochenhauer, however, experimenting with pressures ranging from 3 to 27 4 inches of mercury, found that for a given interval the difference of potentials required to produce disruptive discharge was proportional to the pressure increased by a small constant quantity. Faraday, in the 1 2th and 1 3th series of his Experimental Re&arches, examines this subject ; and the reader who de sires to have a clear idea of what the issues involved really are will do well to begin by carefully studying Faraday s results, and still more his views on this matter. Faraday directs his attention to the specific behaviour of different gases. The gas to be examined was introduced into a receiver in which were arranged two balls s and I, of diameters 93 in. and 2 02 in. respectively, at a constant distance 62 in. apart. Two balls, S and L, of diameters 96 in. and 1 95 in., were placed on suitable insulating supports outside the receiver. S and s were connected with an electric machine, and I and L to earth. The distance u between S and L could be varied at will ; if it was greater than a certain value , the sparks always passed between s and I in the receiver ; if it was less than a certain value a, they always passed between S and L in the outer air. It might have been expected that a and & would be equal, or at least very nearly so, i.e. that there would be one definite value of u, for which the spark would hesitate between the alternative intervals. This is not so, however. Nor ag:iin is the value of u the same when s and I are negative as when they are positive. The following table will illustrate these points, as well as the relations of the different gases : 1 Maxwell, Electricity and Magnetism, vol. i. 57. a ElectririU, t. i. 481. 4 and I positive. s and Z negative. Gas. a
Mean. ! Mean. Air 0-60 0-41 0-55 0-30 0-56 64 0-37 0-89 079 0-60 0-68 0-44 0-72 0-86 0-61 1 32 0-69 0-50 0-61 0-37 0-64 0-75 0-49 1-10 0-59 0-50 0-59 0-25 O o8 0-69 0-47 0-67 0-68 0-52 0-70 0-30 0-60 0-77 0-58 075 0-63 0-51 0-64 0-27 0-59 0-73 52 072 Oxvgen Nitrogen Hydrogen Carbonic acid Olefiant gas . Coal gas Hydrochloric acid. It will be seen that the different gases present consider able variety, and cannot be classified in any way so as to connect the dielectric strength with any other physical pro perty. The numbers given cannot be regarded as measuring the dielectric strength, owing to the disturbing influences which cause the inequality of a and /3. This inequality is not by any means small ; e.g., for air the uncertainty amounts to about 32 per cent. These experiments show very clearly that the sign of electrification of the surface at which the discharge begins has a great effect on the limiting tension. The discharge passes much more readily from a small ball to a large one when the former is nega tive than when it is positive. Faraday made a variety of experiments to elucidate this point, and he was driven to the conclusion " that, when two equal small conducting surfaces equally placed in air are electrified, the one posi- Positive tively the other negatively, that which is negative can and ne- discharge to the air at a tension a little lower than that at ! v . e required for the positive surface, and that, when discharge does take place, much more passes at each time from the positive than from the negative surface." The inequality of a and /? may be due to various causes, among which may be mentioned the charging of the glass of the receiver, dust, &c., in the air, heating of the air, and the presence of finely divided metal dispersed by pre ceding sparks. The last of these causes would account to a considerable extent for the fact that the sparks show a tendency to persist in a path once opened, and that the interval /? a is less for the negative spark, which starts at a smaller limiting tension, and may therefore be sup posed to produce less mechanical effect. Wiedemann and Riihlmann have recently taken up this Wiede- subject in a research which has already been alluded to. 3 The gas and the spark terminals were inclosed in a cylindrical metal receiver with rounded ends. A small window allowed the light from the spark to fall on a rotating mirror fixed on the axis of a Hottz machine, which furnished the electricity. The images of the successive sparks were observed by means of a heliometer. One-half of the divided object-glass was moved until one of the images of^one discharge coincided with one of the images of the next ; then a similar coincidence was brought about by displacing the half-lens in the opposite direction. The difference (y) of the two readings on the micrometer of the heliometer measures the rotation of the disc of the Holtz machine between the two sparks. Preliminary experiments showed that the amount of electricity furnished by the machine while the disc moves through a given angle is independent of the angular velocity of the disc. It varies from day to day, however, according to the quantity of moisture in the air and the arrangement of the machine ; but, on the principle just laid down, correction can easily bo made by taking the reading each day of a galvanometer through which the current of the machine is sent. It follows, therefore, that y is proportional to the quantity of elec tricity which passes at each discharge through the gas, and by means of a galvanometer observations on different days can be compared. It was found that at the lowest pressures worked with (5 to 25 mm. of mercury) the discharge of the Holtz machine was still discontinuous ; and that in all the ex periments the tension at the electrodes was such that tho discharge was independent of the nature of the metal, in maiin 3 Abh. d. k. Sdchs. Gesellsch., 1871, or Wiedemann, Galv. ii. 2,
933, &c,