528 



that all the results of a particular day are likely to have been similarly affected by the weather conditions 

 prevailing on that day, and that it is fairer to proceed according to the method of Table II., and attach 

 equal weight to the mean for each day, whatever the number of observations from which it has been 

 derived. The difference in the actual results of the two methods is fortunately too small to be of importance. 



There is still another difficulty to be considered. The place of observation was not the same in the 

 winter months as during the rest of the year. From the 24th of May, 1902, till the beginning of August, 

 and again during the latter half of March and in April, 1903, the observations were taken on the ice 

 instead of at the usual place on shore. The height of the burning match above the ground or ice was the 

 same at both places. The apparent winter minimum might, then, be held to be spurious and due to the 

 change of the place of observation. Against this it is to be said that there is no obvious discontinuity in 

 the record of potential readings when the place of observation is changed, and that the values of the 

 potential for April, 1903, are greater than those of April, 1902, in spite of the fact that those of 1903, 

 unlike those of the previous year, were taken on the ice. There is, then, no evidence that the potential 

 above the ice was lower than that at the same height on shore. The change of place of observation, no 

 doubt, however, does introduce a serious uncertainty. 



Even, however, if we confine our attention to the observations taken under constant conditions from the 

 beginning of August, 1902, to the end of February, 1903, the evidence of an annual variation of potential 

 is very strong, the maximum occurring in December. The material for the rest of the year is much less 

 satisfactory, on account of the less frequent observations, as well as of the changes in the place of 

 observation. 



To sum up, we may say that the observations made at the " Discovery's " Winter Quarters give evidence 

 of the existence of an annual variation of the electric field, the maximum being in December (i.e., mid- 

 summer), there being probably a minimum in June or July, and a secondary minimum in March or April. 

 It has, of course, to be remembered that, these conclusions rest on the results of observations extending 

 over only one year. The midsummer maximum is very clearly indicated by the observations and is 

 remarkable when we remember that in lower latitudes the electric field has its maximum strength in 

 midwinter. Data for the study of any daily variation are practically wanting except for the month of 

 November. Here there is evidence of a minimum in the middle of the day (and this is in agreement with 

 the results of the December and January observations) and a maximum in the early morning hours. The 

 early morning maximum is remarkable, the one point in which there is agreement in the curves of daily 

 variation for different seasons and different localities in lower latitudes being the occurrence of an early 

 morning minimum. 



5. POTENTIAL AT 15 FEET ABOVE THE GROUND. 



During December, 1902, and January, 1903, in addition to the observations at the usual height, 

 measurements of the potential at 15 feet above the ground were also made. For this purpose a collector 

 match was attached to one end of a rubber-insulated wire, the other end being connected to the electro- 

 meter terminal. The match was lighted and then raised to a height of 15 feet with the help of a pole of 

 that length, to which the insulated wire was attached. 



The observations were taken about 3 minutes after those at the 5-feet level. 



In the absence of special tests the insulation in such an arrangement is by no means above suspicion. 

 Nor, even in the absence of error due to defective insulation, is it possible to deduce from the observations 

 the true undisturbed potential at 5 feet and 15 feet, on account of the deformation of the potential 

 surfaces by the observer and apparatus. But the results of the observations at the higher level are of 

 value as confirming the results obtained for the variation of potential at the lower level. The observa- 

 tions at the two levels not being simultaneous, we cannot expect a constant ratio between the high- and 

 low-level potentials for individual readings (even in the absence of volume electrification of the air in the 

 immediate neighbourhood of the place of observation), but there appears to be a fairly constant ratio 

 between the means at the two levels when we consider a reasonable number of observations. 



In December, 1902, and January, 1903, observations were taken fairly regularly at the hours 10 a.m., 

 4 p.m., and 10 p.m. If we confine our attention to days on which observations at each of these hours wore 



