774
ATMOSPHERIC
issues passes through a hole in a window. From the tube, or the tank, an insulated wire passes to a quadrant electrometer. Readings can be taken by eye, or a continuous photographic record can be obtained. In the latter case, light reflected from the mirror attached to the electrometer needle falls on prepared paper, which is wound round a drum driven by clockwork. Installations of this kind are, or have been, in operation at a good many observatories, including Kew, Greenwich, Lyons, Perpignan, Lisbon, and Batavia ; of late years a water-dropper has been in operation during summer near the top of the Eiffel Tower. At Kew Observatory the water-dropper tube is connected to the needle of the electrometer, whose quadrants are connected, the one pair to the positive pole the other pair to the negative pole of a battery of cells, whose centre is earthed. The arrangement is shown diagrammatically in Fig. 1. The most obvious weak point in a water-dropper Water Tank on Insulators
is the tendency to freeze in cold weather. To remedy this, less easily frozen liquids have sometimes been employed, or the water has been heated or the tube protected. By such means water-droppers were kept in continual operation at two of the stations (Sodankyla and Cape Thorsden) of the international polar year 1882-83 during very severe cold. § 4. Before passing to the observational data it is expedient to refer to various sources of uncertainty. As already stated, we may in this inquiry regard the earth as at a uniform zero potential. Above a limited area of a level plain of absolutely smooth surface, devoid of houses, trees, or grass, the equipotential surfaces under normal conditions would be horizontal planes, and if we could determine the potential at a point 1 metre above the ground we should have a definite measure of the potential gradient at the earth s surface. The presence, however, of apparatus or observers upsets the conditions, while above uneven ground, or near a tree or a building, the equipotential surfaces cease to be horizontal. In an ordinarily moist climate a building seems to be practically at the earth’s potential ; near its walls the equipotential surfaces are far from being horizontal, and near a ridge they may lie very close together. The height of the walls in the various observatories possessing water-droppers, the height of the water-dropper tube above the ground, and the distance it projects from the wall, all vary largely. Again, the tube may project from the centre of a long wall or from a corner, and there may be external buildings or trees sufficiently near to influence the potential. An interesting
ELECTRICITY example of the influence of the situation is presented by Batavia Observatory. During the years 1887-90, with the water-dropper at a height of 2 metres above the ground, the mean of the potentials recorded was 79 volts ; but during 1890-95, with the water-dropper at a height of 7 ’8 metres, the mean voltage was 967. This will show the futility of comparing the absolute voltages met with at two stations, unless means exist of allowing for the difference in the environments of the apparatus. This is the principal reason why Tables I. and II., p. 775, give the ratios borne by individual hourly or monthly values to the mean value for the day or year. § 5. There are other sources of uncertainty. If the shape of the equipotential surfaces near the jet is influenced by trees, shrubs, or grass, the influence will vary throughout the year. Again, in winter, the varying depth of snow or the formation of icicles may exert an appreciable influence. There are further sources of trouble in the apparatus itself. Unless the insulation is perfect, the potential recorded is not that of the atmosphere at the spot where the jet breaks. The action of the jet is opposed by the leakage through imperfect insulation, and if this is greater at one hour of the day, or at one season of the year than at another, there may be a fictitious element in the diurnal or annual variation. A similar result may arise from variability in the pressure under which the jet issues, especially if the insulation is indifferent. The potentials that have to be dealt with are often hundreds, and sometimes thousands, of volts, and insulation troubles are more serious than is generally appreciated. As a check on the records from waterdroppers, it is desirable that scale-value determinations should be regularly made, and that the curve readings should from time to time be compared with eye observations, taken with a portable electrometer or electroscope in a fixed position, at a sufficient distance from buildings or shrubs. In interpreting the records from a water-dropper, allowance must be made for the existence of purely local phenomena. The necessity for this was shown as long ago as 1860 at Glasgow (Kelvin, Papers on Electrostatics and Magnetism, § 392). Two water-droppers were kept running in the university buildings, at no great distance apart but at different levels, and it was found that they often did not show pronounced changes of potential simultaneously. From this Lord Kelvin drew the conclusion—which more recent observations have only tended to confirm—that there are often electrified portions of the atmosphere in motion close to the earth. What may be regarded as accidental temporary disturbances have probably little influence on the results found for mean diurnal or annual inequalities, but this is not the case with influences such as the site of the station, whether in a valley or on a hill, near a river or the sea. § 6. When rain is falling the potential frequently changes rapidly. These changes are often too sudden to be satisfactorily dealt with by an ordinary electrometer, and they sometimes leave hardly a trace on the photographic paper. Again, the aviations rain dripping from all parts of the efflux tube may ° , ... materially affect the situation. For these and other P° ea ,a ’ reasons, it is customary at some observatories to take no account of days on which there is an appreciable amount of rainfall, or else to form separate tables for “dry” or “fine” days, and for “all” days. In other cases negative potentials are excluded from certain of the tables. Speaking generally, the exclusion of days of rain and of negative potential comes pretty much to the same thing (see, however, § 15). Of the accompanying tables (p. 775) the first gives the mean diurnal inequality at various stations for the whole year, while the second shows diurnal inequalities for “winter” and “summer.” These each include six months, except at Sodankyla and Perpignan, when three months at mid-winter and three at midsummer are meant. For reasons already stated in § 4, the hourly values are expressed as percentages of the mean value for the twenty-four hours. The height h of the efflux tube above the ground, and the distance l which it projects from the building, are in metres. The hours are in most cases mean local time. In the case of Florence, however, the entry under an hour is really the mean for the previous sixty minutes. At Sodankyla apparently a correction of about twenty-seven minutes would be necessary to give local time. The data employed in calculating the tables are derived from the following sources: Cape Thorsden {Observations faites au Gap Thorsden Spitzberg, par VExpedition Suedoise, tome ii. 2, par S. A. Andree, Stockholm, 1887); Sodankyla {Expedition Polaire Finlandaise, tome iff., Helsingfors, 1898) ; Kew Observatory (Everett, Phil. Trans, for 1868, p. 347, and Whipple, Brit. Assoc. Report for 1881, p. 443) ; Greenwich (annual volumes of Greenwich Magnetical and Meteorological Observations) ; Florence {Met. Zeit. for 1891, p. 357); Perpignan {Met. Zeit. for 1891, p. 113, and 1890, p. 319) ; Lisbon (annual volumes of Annaes do Obs. do Infante D. Luiz); Batavia {Observations made at the Mag. and Met. Obs. at Batavia, vol. xviii., 1895) ; Cape Horn (Hann, Met. Zeit. for 1889, p. 95). § 7. Some of the results in Table II. are shown graphically in Fig. 2, along with some corresponding
