Proceedings of the British Association. 261 



intensity, Col. Sabine appears to agree with Prof. Kreil, and to find 

 that the number of observed disturbances of that element is greater 

 in the winter than in the summer months. 



Having thus stated the conclusions which have been hitherto 

 drawn, in connection with this subject, Dr. Lloyd proceeded to lay 

 before the Section the results to which he had himself arrived, by a 

 different mode of investigation, as applied to the observations made 

 in the magnetical observatory of Dublin. 



The problem which he proposed to himself had for its object 

 to determine the law of probability of disturbances, as dependent 

 upon the hour of the day, and upon the season of the year — a 

 question, the solution of which will be seen to be of very great im- 

 portance with reference to any physical theory of the phenomenon. 

 The methods hitherto applied, although they indicate in a general 

 manner the times of greater and less disturbance, do not solve this 

 question. In the investigations of Prof. Kreil and Col. Sabine, 

 no account is taken except of disturbances exceeding a certain 

 arbitrary limit ; and, with respect to these, the results are not com- 

 bined in such a manner as to give the law in question. The de- 

 duction of this law, although somewhat laborious, is nevertheless 

 simple in principle. We have only to take differences between each 

 partial result and the monthly mean corresponding to the same hour, 

 and to combine these in the same manner as the errors of observations 

 (to which they are analogous) are combined in the calculus of 

 probabilities. Thus, the square root of the mean of the sum of the 

 squares of these differences is a quantity analogous to the mean error, 

 in the partial observations of a constant quantity ; and the probable 

 disturbance at any hour is inferred from this, by multiplying it by a 

 constant factor. The values of this function (which Dr. Lloyd 

 proposed to call the mean disturbance) have been deduced for the 

 several hours of observation in each month. The corresponding 

 values for the entire year are deduced from those of the separate 

 months, by a repetition of the same process ; they are given, reduced 

 to minutes of arc, in the following table : — 



1 



2'-16 



3 



2'-09 



5 7 

 l'-09j2'-45 



9 

 3'-46 



11 



4'-10 



13 



2'-81 



15 



2\ r )2 



17 



2'- 16 



19 

 l'-93 



21 | 23 

 l'-87r-94 



