2 AB si?i A 

 we lind from (6) and (9) 



( 56 



y = ^- (12) 



12 cos (f (5 4- y 



M= ^— = _J:^J_. .... (13) 



9 cos^ y + 4 1 4- ^ y 



111 these formulae the ^'allles of A — B and cos A do not appear, 

 and (12) olFers the advantage that it enables us to calculate the friction 

 eoefiicient without tlie intervenlion of a quadratic equation, as when 

 (lOj is used. 



This advantage however is only apparent as, for A = B and 

 A = 270°, i?== — 1, and the quantity 7, disappears fi-om the 

 formula; in fact, for a latitude of 52° .1/ approaches very nearly to 

 unity (0.986), as required by theory. 



It is therefore necessary to have recourse to another method and 

 as, without doubt, angular values can be determined with a greater 

 degree of accuracy than amplitudes aud tlie relation 



Hj = 7 3 f^i (^os (f 

 can be only approximatively true, we assume that: 



)., = c, ;., = (;+ 90^ 



or, i. o. w. the problem is thus formulated: what is the value to be 

 given to k in order to ensure agreement between the angular values 

 of the windvariation (6\ and C.,) on the one hand, and of the 

 barometer variation (6') on the other, for the semidiurnal variation. 

 As in this way not one but actually two relations are derived IVom 

 theory, obviously two values of k may be deduced from (5); taking 

 these together, the relation 



B sin (f., — A COS ffj 



A sin d^ -\- B cos 6^ 



<i\ = C-C,, d.,= c-c, 



is easily found. 



If k is known, then, with the help of form. (10), three different 

 values for the angle of deviation m may be derived, namely 

 m^ {q = 2) for semidiurnal periodic winds 

 7)1^ {q = 1) „ monodiurnal ,, „ 



m^ {q = 0) „ constant non 

 For the last named quantity form. (10) gives : 



a 

 tang ^n^ = — , 



A' 



the same value as, according to Oberbeck's theory, obtains for the 



