( 218 ) 



As mean value of the difFerences between observation and com- 

 putation we here tind ±0".059J, an even less admissible result 

 from chance errors than the value, derived from Table 11. Here it 

 is the ditferences 0— C from 1907.0 uj) to about 1907.B, that 

 reach very abnormal values. From 1907.7 the agreement may be 

 considered satisfactory . 



This would seem to lead up to llie coucbisiou, that the change 

 in tlie second component of the polar motion must have taken place 

 rather rapidly, and some^iiere between 1907.3 or 1907.4 and 1907.7. 

 On closer examination, however, it seems to me, that the obsei'va- 

 tions do not suflicicnlly justify this conclusion. We may rather say, 

 that the real path of the pole during the year 1907 deviates more 

 and more from the foiiner orbit (1904 — 1907), to approach to that 

 deduced from the elemenis found for 1908 — 1911. 



Bettei' still than by the Tables II and III, this is shown by the 

 annexed figure, which rejn'esents the observed path of the pole, and 

 the two com|)uled ones from Tab'.e II and 111. The curve drawn 

 continuously shows the dis|>lacemeut of the pole according to the 

 observations; the computed curves have been rejn-esented l)y dotted 

 lines. The inner one results from the elements found for 1904— '07, 

 the ontei' one from those for 1908 — '11. 



It had alreaily appeared from Table I, and the mean errors deduced 

 from il, that the elements (4) represent the observations before 1907 and 

 after 1908 with sufficient accuracy, and the figure shows, that the 

 observations in 1907 indicate a gradual rather than a sudden ti-ansi- 

 tion from one orbit to another. Thus I simply sup[)Osed, that the 

 amplitude c, of the 14-monthly circle gradually increased in the 

 course of 1907 from 0".118 to 0".250, and the computation on this 

 basis corresponds so remai'kably well with the observed co-ordinates, 

 that I thought it unnecessary to extend the researches still in other 

 directions. 



For a closer investigation moreovei-, it would have been necessary 

 to go back to the original observations of the several stations. 

 AT>BRKt'HT's coordinates have been obtained by a process of adjust- 

 ment, and this turns even rather sudden changes into smooth transi- 

 tions. In the first place the time for such an investigation was 

 lacking, and secondly it remains to be questioned, whether the accuracy 

 of the separate results would admit of a decided conclusion. 



In foi-mula (4) I substituted therefore for the year 1907: 



6-., ~0".118 4- I ".132 '« - 1907.0) ■ (5) 



Combining the resulting values of .k^ and y^ with the yearly com- 



