SECT. 1] THE MAGNETIC FIELD OVER THE OCEANS 191 



on either side. The rapid changes of field are confined to the dayhght hours. 

 Within an area affected by the electrojet, short-period fluctuations in times 

 of the order of one hour and with amphtudes of up to 100 y may occur at 

 any time, but particularly during the hours of daylight. There is no way of 

 distinguishing these time variations from spatial variations, other than by a 

 resurvey or by having a second magnetometer in a moored buoy. Similar 

 complications arise in the auroral zones. Experience suggests that, except in 

 these special areas, the error incurred by neglecting the normal daily variation 

 may be acceptable. 



Magnetic storms present a more serious problem. It is doubtful if a useful 

 correction can be made using data from a land station and the only way of 

 eliminating the uncertainties is to reject observations taken when the level of 

 disturbance at a land station exceeds some arbitrary limit, or to provide a 

 magnetometer in a moored buoy to determine the correction. Electric currents 

 induced in the ocean will modify the storm-time fluctuations, particularly the 

 high-frequency components, in much the same way as they modify the daily 

 variation, but the extent of this modification is unknown. A number of magnetic 

 storms occurred during the survey of the north-east Pacific described in a later 

 section, but without any significant effect on the contours (which were plotted 

 at an interval of 50 y). The effects of magnetic storms and daily variation would 

 be more serious in more detailed surveys. 



A correction for secular variation may be necessary if the survey of an area 

 is spread over more than a few months. The rate of change of the field can be 

 taken from neighbouring land stations or from magnetic charts. A single value 

 can usually be used for an area up to 1000 km across. 



In order to plot anomalies in an intelligible form it is necessary to remove 

 the regional trend. Since the removal of a complicated or irregular trend may 

 introduce spurious features into the pattern of anomalies, it is desirable that a 

 simple and easily understood method should be used. A linear variation with 

 distance on a Mercator chart is often satisfactory over areas a few hundred 

 miles across. In the magnetic survey of the north-east Pacific a second-order 

 polynomial in latitude and longitude was fitted to the observed field, averaged 

 within 20' rectangles ; a third-order polynomial has been successfully used in 

 an area 40° x 40° in the north-western Indian Ocean. A linear variation plus 

 a two-dimensional Fourier series was satisfactory in a 20° x 20° area in the 

 Atlantic (Bullard et al., 1962) but cannot be generally recommended since the 

 Fourier series necessarily has the same values on two opposite edges of the area, 

 which often prevents a good fit being obtained with a few terms. The deter- 

 mination of trend from observations within an area is in a large degree arbitrary 

 and it is to be hoped that Project Magnet will so improve our knowledge of the 

 field that a spherical harmonic expansion of the whole earth can be used to give 

 the trend in the areas covered by detailed surveys. 



When the ship is in range of accurate navigational aids the measurements 

 at the crossing points of tracks usually agree to within 20 y, but after a few 

 hours' steaming in the open ocean, discrepancies of 50 y are not uncommon. 



