134 



might have been occasioned by an accidental distribution of tiie 

 errors of our observational results and in reality might not exist. 

 Let us take it that for the values found the real errors were the 

 following ones: 



Eastcomponent. 



Height 

 k.m. 



y> 



Value 

 calcul. 



real 

 error 



real 

 value 



value 

 calcul 



real 

 error 



real 

 value 



Phase 



6.5—7 

 7.5— S. 5 

 9—11 



—34 



— 12 

 50 



- 9 —25 



— 2 —10 

 52 — 2 



2 



26 



—23 



142° 

 124° 

 101° 



According to the above the real values for amplitude and phase 

 ought to be: 



This would do away completely with the inversion and would 

 carry on the original decrease of phase up to the highest heights 

 observed, which should be accompanied by a rapid diminution of the 

 amplitude. 



If we obsei've (he real eri-ors, it strikes us that, though the ones 

 for y.^ are one-sidedly negative, they are of about the same size as 

 the mean errors. 



But, as has been explained above and has been proved by the 

 graph, the mean error is much larger than the one to be expected. 



As regards ,v^ the real errors accepted for 6.5---7 and 7.5 — 8.5 

 km. are small, but the ones for 9 — 11 km. are much larger than 

 the mean ei-ror, already exaggerated, viz. 52 against 34. 



