234 
DR. C. CHREE: DISCUSSION OF KEW MAGNETIC DATA 
curves. They show one well marked maximum near 6 p.m., and an equally well 
marked minimum about 4 a.m. 
The I difference curve in fig. 12 is also markedly unlike the ordinary day I curve. 
It has a considerable resemblance to the W difference curve, but corresponding points 
occur some hours later in the day in the I curve. Some doubt must be entertained 
as to the reality of the depression shown about 4 p.m. 
The difference curves in fig. 12 indicate on the whole an increased amplitude in the 
ordinary day inequality as compared with that of the quiet day, but they also 
indicate a distinct difference in type. The nature of the difference is most 
conveniently considered in connection with the harmonic analysis of the inequalities. 
One outstanding feature has already been discussed.* 
Annual Inequality. 
§ 15. Let M represent the mean value of any element for the whole year, and AM 
the algebraic increment in the twelve months due to secular change, then if the 
secular change took place uniformly throughout the year the mean value of the 
element during the n th month should be 
M+ (n-%-6) AM/12, 
where n represents 1, 2, 3 up to 12. 
This neglects the difference in length between different months. 
If the actual mean value for the n th month proves to be p n , the values of 
p n — (M+ (n—%— 6) AM/12} 
for the 12 months constitute the annual inequality. 
If a magnetograph had a known and invariable base line value, the annual 
inequality of the corresponding element could be determined from curve readings 
alone. On the other hand, if an absolute instrument were invariably read when the 
corresponding element had its mean value for the day, the annual inequality could be 
got without any reference to the curves. But in practice we can tell the base line 
value of the curve only by reference to the absolute observations, and we cannot tell 
when the element had its mean value without recourse to the curves. 
When base values have been assigned to all the curves, a mean value can be found 
for each day, representing the mean of the hourly values. From these daily means a 
monthly mean can be derived, representing, according to the practice of the particular 
observatory, the mean of all the days of the month, or the mean of all but highly 
disturbed days, or the mean of a limited number of selected days, e.g., the 5 inter¬ 
national quiet days. Unless all days are used, the mean of the days employed may 
* ‘Roy. Soc. Proc.,’ A, vol. 91, p. 370. 
