123 



F'rom the above it will appear that it is impossible to state 

 this dependency quantitatively to any amount. 



The mean errors have been calculated for the groups 4 km. 

 (6.5_7) km. and (9—1 J) km. as follows: 



The equality of the mean errors of a\, .r, and //, and the greater 

 value of the m. e. of y, are striking; this being a consequence of 

 the distribution of the ascensions over the day. 



The mean errors of .v„ y„ x^ and y, are mainly in inverse 

 proportion to respectively \sin t sin t\, \cos t cos t\, [sin '2i sin, 2t\ 

 and [cos 2t cos 2t\, and the greater part of the couples consisted 

 of one ascension between the hours of 6—12 or 18—24 together 

 with one between the hours of 0—6 or 12—18. For such a com- 

 bination the values of [sin t sin t], [sin 2t sin 2t\ and [cos 2t 

 cos 2t\ are indeed about the same, but [cos t cos t] is much smaller, 

 which may be easily ascertained when making up the limits between 

 which the values for sin t, cos t, sin 2t and cos It are fluctuating. 



sin t cos t. sin 2t. cos 2t. 



6li_12^ 1.0 to 0.0 0.0 to — 1.0 O.Oover— 1.0 to 0.0 —1.0 to 1.0 



12''— 6^1 0-0 ^o 



-1.0 —1.0 to 0.0 O.Oover 1.0 to 0.0 1.0 to— 1.0 



(difference) 

 absolute value 



to 2.0 



0.0 to 1.0 



0.0 to 2.0 



0.0 to 2.0 



As regards the values determined for //,, these show indeed 

 — in accordance with the mean errors, which are nearly twice as 

 large — a greater spreading than those for x„ ,i', and ?/,. The 

 values determined are shown in the following table and graphic. 



The figures in the table (p. 124) show, first -. that the influence of land 

 and sea breezes is distinctly visible in the curves for the diurnal variation 

 of the N.-component ; this influence seems to make itself felt up to 

 4 k.m. Together with this influence is mixed up the one, which 



