138 ABSOLUTE RANGES. 



suffered in this way. The information as to the smallest ranges does not suffer from any such uncertainty, 

 but subsequent to September, 1903, the number of days of observation was too small to give results of 

 much value. The smallest recorded ranges on days when the record was complete were 



forD, ll'-Oin July, 1902; 

 H, 14y June, 1903; 

 V, 15y July, 1902. 



Perhaps the most natural way of comparing the different months or seasons of the year as to their 

 relative liability to disturbance is to consider the size of the ratio 



(mean absolute range)/(range of diurnal inequality). 



This criterion cannot be very readily applied in the present instance, except to the Vertical Force. For 

 it the above ratio has the following values : Midwinter 4'1, Equinox 2'5, Midsummer 2'6. The ratio is 

 thus decidedly largest at Midwinter. The same phenomenon has been observed at Kew in the case of the 

 Declination. 



The practical equality of the ratios for Equinox and Midsummer suggests that in the Antarctic the 

 former is not a season of specially large or frequent disturbances. If instead of taking for our criterion 

 of disturbance the ratio of the mean absolute to the inequality range we were to take the amplitude o 

 the mean absolute range, or the frequency of occurrence of specially large ranges, we should come to the 

 conclusion that the equinoctial months were much less disturbed than those at Midsummer. For instance, 

 in the case of D, taking all days, complete and incomplete, the absolute range was in excess of 3 on only 

 1 1 per cent, of equinoctial days as against 40 per cent, of Midsummer days. 



44. Tables LIV to LIX show the frequency of occurrence of the daily maxima and minima in D, H 

 and V at different hours of the day. For instance, from Table LIV we see that, taking the 24 days of 

 April, 1902, for which data exist, the maximum in D occurred eight times between 8 and 9 a.m., 

 six times between 9 and 10 a.m., three times in each of the hours 10-11 a.m. and 11 a.m. to noon, 

 and once in each of the hours 5-6 a.m., 7-8 a.m., noon to 1 p.m., and 11-12 p,m. Maxima and minima 

 were assigned for D and H even on days of incomplete record. In some cases the imperfection was of 

 such a nature e.g. trace beyond a limit of registration during part of one hour as to introduce no uncer- 

 tainty into the hour of occurrence of either maximum or minimum ; in a greater number of cases there was 

 uncertainty as to one only of the two quantities. The number of days in which uncertainty affected the 

 maximum and the minimum often differed widely. Thus in October, 1902, the hour of minimum in H 

 could be assigned on 26 days, but the hour of maximum only on 12, In the case of V, only days of 

 complete record were included, so that maxima and minima were equal in number. The occurrence of 

 will be noticed in some of the entries for D and H, especially the latter. This may mean equality in the 

 extreme ordinates during two different hours, but it usually signifies that the trace was beyond a limit of 

 registration during part of each of two hours. Days were omitted in which either limit of registration 

 was exceeded during parts of more than two hours. 



If irregular disturbances did not exist, the daily maxima and minima would synchronise with the maxima 

 and minima in the diurnal inequalities. But in the highly disturbed conditions of the Antarctic one would 

 not have been entitled to assume a priori that the absolute maxima, for instance, would have shown even a 

 preference for the hour of the inequality maximum. Comparing, however, Tables LIV and XIII, we see 

 that the daily maxima for D cluster thickly round the hour 9 a.m., when the maximum presents itself in 

 the mean diurnal inequality for the year. This is true of all the seasons, especially Midsummer. It is 

 truly remarkable that at this season, when disturbances were so large, out of the 79 days for which data 

 exist not one gave an absolute maximum during the 12 hours ending with 2 a.m. 



If we define the " concentration " of the frequency as the percentage which the occurrences during the 

 three consecutive hours which, combined, give the greatest number of occurrences bear to the total number, 

 we find in the case of the D maxima for the concentration 41 in Midwinter, 60 in Equinox, and 62 in 

 Midsummer. The time of greatest frequency seems a trifle later in Equinox than in the other two seasons, 

 but the difference is minute. 



