86 



EEPOET — 1886. 



6. Construct now a new table (B), eact entry in which is the 29-day 

 mean of the numbers for the same hour in Table A, viz., of the numbers 

 for the day of the entry and of the fourteen preceding and fourteen 

 following days. The numbers of Table B for all the hours of a given day 

 we may take to represent very approximately the mean solar-diurnal 

 variation — plus a constant — for that day, the average extending over the 

 lunation of which that day is the middle day. They will be affected by 

 progressive change of the values of the tabulations and by disturbance 

 within the limits. 



7. The excesses of the numbers of Table A over the corresponding 

 numbers in Table B, plus a constant round number, are now entered on 

 a third table (C). The numbers in this table will be affected only by 

 that part of the solar-diurnal variation which goes through a cycle of 

 change in a lunation, and by disturbance within the limits. On the left- 

 hand margin of table C mark the days of the moon's age, the number 1 

 being placed opposite the first day of which at least a full half follows 

 the time of new moon, and the other numbers, in order, up to 29 or 30. 

 If table C were calculated for each month of a long series of years it 

 would be practicable to re-arrange the days in tables, of which there 

 would be one for each day of the moon's age in each month, with a 

 probability of obtaining characteristic diurnal variations from the numbers 

 of each table ; but as the trial calculations extend over three months only, 

 these (November to January) were grouped together, and the days of 

 the moon's age were arranged in eight groups as follows : — 



Thus eight tables were formed corresponding to the times of the four 

 quarters of the moon, and to the four intermediate phases, and tb& 

 numbers of Table C were duly distributed amongst them.' The hourly 

 means were taken of all the numbers on each of these tables, and then 

 the excesses of those means above the general mean for the twenty-foui- 

 hours of the same table, and finally these excesses were converted from 

 inches (of tabulations) into m.g.s. units of force. In this way were ob- 

 tained the excess solar-diurnal variations for each of the eight phases of 

 the moon. From these were calculated the luni- solar-diurnal variations 



of the formula 



fosW and f,.^(h) 



f,.,(h) cos 2(^^ ) +f,.,ih) sin 2(^py 



' By inadvertence the last three days of Januaiy, which formed part of a fourth 

 lunation, were left undistributed ; so that the results will be for the three lunar 

 periods from November 1, 1875, to January 28, 1876. 



