PRELIMINARY VALUES OF THE ANNUAL CHANGES OF THE MAGNETIC 



ELEMENTS AS DETERMINED FROM THE GALILEE AND 



CARNEGIE RESULTS. 1905-1916. 



DATA AT INTERSECTIONS OF TRACKS OF VESSELS. 



Explanatory Remarks. 



The magnetic results of the Galilee and the Carnegie have been compared from tune to 

 tune with the values scaled from the magnetic charts in use. As has been seen from Figures 

 29-35, systematic errors were found which may be explained, at least for the magnetic- 

 declination charts, by a lack of accurate information of the secular changes over the ocean 

 areas. Constructors of magnetic charts have been obliged to rely principally upon the 

 secular changes shown by land observations along the coasts and on the islands, or upon 

 more or less uncertain sea observations, separated often by rather long intervals. Plates 

 20, 23, 24, and 25 show that during the various cruises of the Galilee and the Carnegie, a 

 veritable network has been formed by the crisscrossing of their tracks. 



Theoretically, every crossing should furnish data regarding secular variation in each 

 magnetic element, but there are practical difficulties in securing accurate values of the 

 annual amounts of the secular variation at sea, which can not be entirely avoided. If the 

 interval of time between the observations of two tracks that cross is very short, it is evident 

 that the secular variation may be masked by observational error. The comparison then 

 serves only as some measure of the accuracy of the results. On the other hand, when the 

 time-interval is sufficiently large the average annual change may be determined with some 

 precision for the time considered; as the annual change itself is a variable quantity, the 

 accuracy of its determination can not, however, be increased ad libitum by simply increasing 

 the time-interval, nor can it be assumed that the annual change at the end of the interval 

 is the same as the average value for the elapsed interval. 



Another difficulty, peculiar to ocean observations, is that the precise point of a former 

 observation can not usually be reoccupied, and the difficulty is increased when the decUna- 

 tion is not determined at the same time and place as the other magnetic elements. In 

 order, therefore, to compare the results of magnetic observations made at two different 

 dates for the purpose of determining the average annual changes, they must be referred 

 to practically the same geographic position, if, as most frequently happens, the values of 

 the magnetic elements are not independent of the geographic position within the area 

 considered. The accuracy of the determination of the annual change is therefore improved 

 by increasing the number of observations within the area considered. There are, however, 

 practical limits to the number of observation-results that may be utilized. If they extend 

 beyond the area over which their changes with geographic position may be considered 

 linear, it is necessary to include in the computation of the value of the magnetic element for 

 some common point, terms containing the latitude and longitude differences to the second 

 degree or more, and the number of unknowns is thus increased; in this case the time and 

 labor which would be required to make a solution are justified only when no more observa- 

 tions are likely to be available for some time to come. 



A preliminary investigation of the distribution of the magnetic-dechnation values over 

 limited areas around some of the intersections of the Galilee-Carnegie tracks, shows that 

 a single determination of the declination may occasionally differ many minutes from the 

 normal value as indicated by the remainder of the group. The cause of such abnormal 

 values is not always known; there might have been local disturbance or there might have 

 been abnormal conditions of the ship's motion. Whatever the cause, the inclusion of such 

 a station in a group that is utiUzed to determine the average annual change, may seriously 

 affect the result sought, especially, if the group is composed of but few units. The presence 



430 



