20 Land Magnetic Observations, 1914-20 



in ocean work the error may be five or more units in this decimal place. For these 

 reasons it appears inadvisable for field results to adopt so small a unit as a small 

 gamma, 7=10- 5 C.G.S. unit; it would be necessary otherwise at times to round 

 out the observed value by one or more zeros. If the conditions under which an 

 intensity result was obtained were such as not to warrant publishing the fourth or 

 fifth decimal, this is shown by stopping with the decimal deemed certain. In 

 general, however, as will be seen, the value to the fifth decimal is given, but it 

 should be understood that no claim is made as to the correctness of the last figure; 

 it has been retained here primarily in order that when all reductions to common 

 epoch have been applied on account of the magnetic variations, an error of a 

 unit in the fourth decimal, due purely to computation, will not enter. 



The first column in the table is headed "Station"; this gives the name of 

 place at which the magnetic elements were observed, the spelling adopted being 

 in accordance with the most reliable information at hand and conforming as far 

 as possible to local usage. The next column gives the geographical position, lati- 

 tudes, and longitudes, as derived in most cases from the observers' local astro- 

 nomical observations following the methods already described in Volumes I, II, 

 and III (see also pages 23-29). When the latitudes are the results of fairly 

 complete circummeridian observations of the Sun, or the means of several reoccu- 

 pations of the same station, or are derived from reliable large-scale maps, then 

 they are given to the nearest O'.l, though it should be distinctly understood that 

 this accuracy is not guaranteed, as even for these cases the error may be as much 

 as 0'.5, and even in some instances a whole minute of arc. When the latitudes 

 are given only to the nearest minute, there were either no astronomical determina- 

 tions, or they may have been incomplete or defective; these values are usually 

 taken from standard atlases and for some regions may be in error by several min- 

 utes. Owing to the numerous sources of error of a longitude determination, and 

 especially because of the uncertainty in more or less unexplored countries of the 

 adopted chronometer-correction on standard time, the longitude in no instance is 

 tabulated closer than to the nearest minute of arc. Usually it is derived from the 

 observers' astronomical observations. Considerable use was also made of reli- 

 able large-scale maps, whenever available, and of standard atlases; the values in 

 regions but slightly surveyed may be out sometimes by several minutes (see pages 

 23-25). 



The date on which the magnetic observations were made will be found in the 

 fourth column. The following abbreviations have been adopted for the months 

 of the year: Jan, Feb, Mar, Apr, May, Jun, Jul, Aug, Sep, Oct, Nov, Dec. The 

 values of the magnetic elements will be found in the next columns as observed at 

 the local mean time, expressed to nearest 0.1 of an hour, opposite each value. 

 Occasionally it has appeared desirable, where diurnal variation in declination was 

 observed or where numerous observations were made during a limited interval, to 

 give the local mean times of the beginning and of the end of the series and to 

 indicate the number of determinations from which the mean value is derived by 

 a number inclosed in parentheses: thus 9*1 to ll h 3(7) is to be read "the mean 



