( 320 ) 



In the paper already quoted, Mr. Angot assumed that the deviations 

 do not show systematic differences in different months, and he sub- 

 jects the deviations taken conjointl}^ to the criterion of the law of errors. 



This assumption is not justified l\7 the results given in Table III, 

 from which it is evident that the values of h are subject to consider- 

 able and systematic variations and, if a satisfactory agreement 

 is still found between theory and observation, this can only be 

 accounted for by the fact that the probability of the occurrence of 

 deviations between fixed limits is expressed in a number of decimals 

 too restricted to indicate the differences which, as for Helder and 

 Batavia, must here exist between theory and practice. 



No more can it be affirmed that, if a satisfactory accordance exists 

 between the calculated and the observed number of deviations 

 between given limits, the average value will also be the most 

 probable one. Tn applying this criterion, as well as in calculating 

 li and SÏ, a possible (and probable) skewness of the frequency curve 

 is not taken into account because, by treating the deviations without 

 regard to their sign, symmetry with respect to the ordinate of the 

 centre of gravity of the figure is tacitly assumed. 



As the number of years over which the observations extend is 

 still far too small to allow frequency curves to be drawn for each 

 month separately, it is still worth while to consider the deviations 

 collectively, provided that at the same time the question be put, 

 what form the law of deviations will assume when they are com- 

 posed of groups which individually follow the normal law, the factor 

 of steadiness being different for different groups. Even then the 

 available data are insufficient to indicate with certainty a small 

 degree of skewness in the frequency curve, so that only the sym- 

 metrical form can be sought for. 



3. If, as in our case, the different groups occur with equal (sub) 

 frequency, it is not difficult to indicate in what respects such a 

 curve, the resultant of many elements, must differ from the normal 

 curve. The groups characterised by large factors of steadiness will 

 raise the number of small deviations above the number correspond- 

 ing with an average factor and contribute only in a small degree 

 to the number of large deviations, whereas, on the contrary, flat 

 curves with small factors will give rise to a greater number of large 

 deviations than is consistent with the normal law. Deviations of 

 average magnitude will then occur to a less degree than is required 

 by the common law ; conseciuently in drawing the two curves, they 

 will be seen to intersect at four points, as a mininium, 



