102 DIURNAL INEQUALITII.s. 



gave a smaller range for the quieter days than for all days, the difference being usually conspicuous, 

 especially at Midwinter. It should also be remembered that the days of the largest disturbances did not 

 contribute to Table XII, as the traces did not remain on the sheet, while the quieter days included about 

 half those that were tabulated, and so cannot bo regarded as representing any very high standard of 

 Antarctic quietness. Thus the differences between the amplitudes of the regular diurnal inequalities on 

 the very quietest and the most disturbed days must presumably be considerably larger even than those 

 presented by Tables XII and XIV. 



27. In the case of the Horizontal Force prior to April, 1902, and subsequent to September, 1903, there 

 were practically no days during which the registration was complete, so that diurnal inequalities are 

 limited to the 18 months April, 1902, to September, 1903. Even between these dates there were various 

 months, especially January and February, 1903, when only about one day in three gave a complete record. 

 Considering this fact, the inequality curves, especially the seasonal ones, are much smoother than might 

 have been expected. As with the Declination, the type of the diurnal inequality seems to vary but little 

 throughout the year ; the tendency is towards a single maximum about 2 or 3 p.m., and a single minimum 

 in the early morning. The annual variation in the amplitude of the diurnal inequality is well marked. 

 But even in June, 1902 which was not far from sunspot minimum i.e. in the depth of the Antarctic 

 Winter, the amplitude exceeded 20y. 



As to the relative sizes of the diurnal changes in Declination and Horizontal Force, a change of 1' in D 

 answered roughly to a force of l'91y, so that taking the mean results for the year we have for the 

 amplitudes 45' 5 or 87y in D as against 35y in H. It must be remembered, however, that a good many 

 days contributed to the D inequalities which were too disturbed to contribute to the H, and if these 

 days had been excluded from the D inequalities the amplitudes of these would probably have been 

 reduced. Still, even if we confined ourselves to the quieter clays of Table XV, we should obtain for D, from 

 the year as a whole, an average daily range of 34' or 65y. This is nearly double the corresponding value 

 for H. It is thus certain that the forces to which the diurnal inequality is due are less potent in the 

 plane of the magnetic Meridian than in the perpendicular plane. 



28. The Vertical-Force trace itself seldom went off the sheet, but loss of temperature trace was less 

 uncommon. Also in some months a good many days' traces which were complete were omitted because 

 there was reason to doubt whether the instrument was working. In some cases the magnet was evidently 

 stuck. The space in which it moved gave very little clearance, and presumably the magnet whose 

 azimuth was approximately east and west magnetic tended to shift in azimuth, and so came in contact 

 with the adjacent metal. In other cases it was impossible to feel certain from mere inspection of the trace 

 that the magnet was stuck, while the extreme quietness of the trace suggested that it was. In such cases 

 one was generally able to settle the question by reference to the temperature trace. If temperature 

 changed sensibly and it usually did and the Vertical-Force trace showed no sympathetic movement, 

 then one inferred that the magnet was stuck, and rejected the day's record. 



Table XVIII gives inequalities for all the individual months from March, 1902, to December, 1903. The 

 inequalities, however, for November and December, 1903, are each based on only 4 days' records, and the 

 temperature-correction data for these were more than usually uncertain. Thus little weight has been allowed 

 these as compared to the corresponding months in 1902 when forming the inequalities in Table XIX. 



There is an apparent difference, clearly visible in fig. 11, between the types of the diurnal inequality of 

 Vertical Force at different seasons. At Midsummer the force, instead of falling continuously from a 

 maximum near midnight to a minimum near noon, shows a tendency to a slight increase about 6 to 8 a.m., 

 and there is at least a suspicion of the same phenomenon in the equinoctial months. The phenomenon 

 may be a real one, but it ought to be mentioned that disturbances in V were more prevalent from 4 to 

 9 a.m. than at other hours, and their satisfactory elimination might require several years' data. 



29. The diurnal inequalities of Inclination in Table XX were calculated from the corresponding V and 

 H inequalities by means of the formula 



51 = SVx 00441 -SHx -0477. 



Here 81, SV, 8H denote the departures at any, the same, hour from the mean values of I, V and H for the 

 day. The numerical multipliers "00441 and '0477 answer to the mean values of V and H during the stay 



