ANALYSIS OF DISTURBANCES. 177 



less than those at Colaba, and apparently a little less than even those at Mauritius. When, however, we 

 consider the disturbances answering to those of the special type in the Antarctic, it is otherwise. 



96. Table LXVI, p. 183, gives results for the Antarctic corresponding to those given for the co-operating 

 stations. The movements include those treated in Table LXII, except the movement on May 8, 1902, 

 from 7.4 to 7.11 p.m., which was partly lost in the Antarctic, and they include, in addition, the second 

 phase of the commencing movement on April 5, 1903, which was essentially peculiar to the Antarctic. 

 The letter i denotes that the Antarctic records were not absolutely complete. On April 5, 1903, the 

 first phase was complete, but in the second phase the H curve went off the sheet. The amplitude shown, 

 300y, was probably not much exceeded. On June 19 and 28, and July 26, both phases suffered loss. On 

 June 19 the loss was confined to H, and on June 28 to D, but on July 26 both D and H went beyond the 

 limit of registration. In all these cases, so far as one could judge from the appearance of the curves, the 

 major part of the movement was recorded, though on July 26 the loss was probably greater than on the 

 other days. These losses, of course, introduce some uncertainty, and one would naturally have used other 

 examples of the special type of Antarctic disturbance but for the fact that a comparison was desirable 

 with other stations, and that days had thus to be chosen when records from the co-operating stations were 

 available. 



To reduce the uncertainty thus arising, a mean disturbance was derived from 51 examples of the special 

 type of disturbance for which the records were complete. The values of AD, AH, and AV were meaned 

 for the two phases separately, and from the mean AD and AH corresponding mean values for AN and AE 

 were found. From these and the mean AV there were calculated corresponding mean values of AX, AY, 

 and AZ, and the force vector deduced therefrom. Comparing the results thus found with those obtained 

 by meaning the disturbances of June 19, 28, 29, and July 26, 1903, in like fashion, we see that in the case 

 of the first phase there is a very close agreement in the values of the angles 6 and <. In the case of the 

 second phase the agreement is not quite so good. 



The sudden commencements were also grouped with a view to obtaining representative means. The 

 operation was limited to May 8 and August 20, 1902, and August 25, 1903, these being the only three 

 occasions on which both phases were completely recorded. In this case the phases in which the elements 

 increased were grouped together, irrespective of whether they occurred first or second. The- values 

 of and </> obtained for the two phases in this case correspond fairly closely with those obtained for the 

 two phases of the mean representative of the special type of disturbance. 



All the examples in Table LXVI can be fairly included in two classes, distinguished as before by the 

 letters A and B. Class B includes all the first phases of the special type of disturbance, and that phase of 

 the sudden commencements during which the element decreased in value. Class A, on the other hand, 

 includes all the second phases of the special type of disturbance, and that phase of the sudden commence- 

 ments during which the element increased in value. In choosing which letter to apply, the guiding 

 principle was, that at the co-operating stations A represented a type of disturbance in which the element 

 or elements of force chiefly affected during sudden commencements exhibited an increase. 



97. On July 24-25, 1902, there were recorded at Kew a number of oscillatory disturbances of no great 

 magnitude. On examining these I found that the turning-points in the three elements appeared identical 

 in a considerable number of cases, and also that corresponding movements could be traced at the other 

 co-operating stations. Table LX VII, p. 184, deals with the measured changes of force on ten of these 

 occasions, at Kew, Colaba, and Mauritius, and with a, minor number of them at Christchurch and the 

 Antarctic. As the afternoon of the 24th (G.M.T.) advanced, it became increasingly difficult to obtain 

 movements that appeared to correspond, the difficulty appearing first in the Antarctic data, and then in 

 the Christchurch ones. 



The first four movements consisted, at Kew, of a rise in force from 2.23 to 2.30 p.m., followed 

 by a fall from 2.30 to 2.38 p.m., then another rise from 2.38 to 2.43 p.m., and a second fall from 

 2.43 to 2.54 p.m. The two double movements appeared of the same type at all the stations ; thus, 

 instead of treating them separately I took a mean from the two, combining the two falls together, and the 

 two rises together. These are numbered (1) and (2) in the table, as if the means represented each a single 

 movement. The other eases, (3) to (8), represent actual single changes of force. If we examine the 



2 A 



