360 



used the diurnal variation on "quief days ; in consequence of the 

 use of the "international" quiet dajs the material for all the stations 

 refers to the same days and therefore is as homogeneous as possible. 

 By means of the calculated values of g and h the FouRiKR-coeffi- 

 cients for the diurnal variation were obtained for the places used 

 in the calculation, and for the 10 stations an average deviation 

 betvi^een calculated and observed amplitude (positive and negative 

 values taken with the same sign) was found, amounting to 



347o in the diurnal, 35"/; in the semi-diurnal period for LX 

 20 /(, ,, ,, ,, -to /j ,, ,, ,, ,, ,, Li I 



4. For a determination of the harmonic coefticients at other places 

 a graphic interpolation is the most suitable. The calculated points in 

 the diagram are joined by a smooth curve, by means of which the 

 coefficients for each intermediate latitude can be read off. Moreover 

 we can decide how far observations made at stations situated outside 

 the interval 60° N. — 14° S. follow the curve of the calculated points. 



A test in this manner gave satisfactory results and led to the 

 following conclusions : 



a. In the main the curves would show the same character if 

 more stations had been used for the calculation. 



b. The difference between ttj- and .t^ cannot be completely at- 

 tributed to insufficient observational material, i.o.w: The forces causing 

 the diurnal variation, taken as a whole, do not possess a potential, 

 although it remains always possible to dedut'e part of these forces 

 from a potential. 



\ 



