ox COMPARING AXD REDUCING MAGNETIC OBSERVATIONS. 85 



a comparatively short series of observations as those of Sabine's rougher 

 method when derived from a much longer series of observations. Such 

 a trial I have since been able to make upon the hourly tabulations 

 from the registers of the Colaba declination and horizontal force magneto- 

 graphs for the quarter November 1875 to January 1876 ; and the results 

 iire of so highly satisfactory a character, and bear so directly on the 

 inquiry upon which the Committee are engaged, that I deem it my duty 

 to place an account of them before the Committee. 

 The process may be described as follows : — 



1. Tabulations of hourly ordinates are entered upon monthly abstract 

 forms (Form A), which have the hours of the day marked at the head of 

 the columns, and the days of the month at the left-hand side of the lines,' 

 and upon these ruled forms the daily mean is taken and entered in a 

 column at the right-hand side of the twenty-four hourly entries of each 

 day, and the mean for the month of the entries in each hour column is 

 taken and entered at the foot of the column. Let us suppose this to have 

 been done for a given month, and for the two preceding and two following 

 months. 



2. Take the mean of the daily means for the first fifteen days of the 

 same month and the last fifteen days of the preceding month as the mean 

 ordinate for the beginning of the former month, and, for the present, let 

 the excess of the mean ordinate for the beginning of the next month over 

 this mean ordinate be taken to represent the progressive increase of the 

 ordinate for the given month, whether arising from instrumental change 

 or from secular or annual variation, and in allowing for such increase 

 treat it as growth at a uniform rate. 



3. On a blank strip of ruled paper cut out from one of the columns 

 enter the proportional corrections for progressive increase, to reduce the 

 tabulations to the standard of the middle of the month ; these corrections 

 will be zero for the middle of the month, and equal positive and negative 

 numbers at equal intervals before and after the middle of the month, and 

 the sum of these for the whole column will be zero. The strip is to be 

 placed close up to each column in succession for reference in the operation 

 of separating disturbed observations. 



4. Apply Sabine's method of separating disturbances to each hour 

 column in succession, taking account of the corrections for progressive 

 increase entered on the loose slip, and calculate final normals. The 

 separating values adopted are for declination '048 inch of tabulation or 

 •00150 of force, and for horizontal force "078 inch of tabulation or -00334 

 of force. 



5. Substitute for each disturbed tabulation the higher or lower limit 

 for that hour and day of an undisturbed tabulation according as the dis- 

 turbance is positive or negative. The deviations of the disturbed tabu- 

 lations above the higher or below the lower limits respectively are to be 

 called positive or negative ' disturbances without the limits,' and the laws 

 of their variations are to be determined by the method that Sabine applied 

 to his 'larger disturbances.' In what follows we are to confine our 

 attention to the original numbers entered upon Table A, except where 

 these have been replaced, in the case of disturbed tabulations, by the 

 higher or lower limits. 



' If the continuity of the record has been interrupted during the month, either 

 by accident or by instrumental adjustments, due allowances must be made to render 

 the whole month's tabulations comparable before proceeding further. 



