FROM ORDINARY DAYS OF THE ELEVEN YEARS 1890 TO 1900. 
180 
published in 1908. This paper was reprinted with the addition of an Appendix in 
the ‘Collected Researches’ of the National Physical Laboratory, vol. 5, 1909. The 
Appendix contains a list of the 209 disturbed days. 
In 1908-9 a further grant of £100 from the Government Grant Committee enabled 
the measurement to be commenced of all the H and V curves for the period 1890 to 
1900. A difficulty at once presented itself. While some of the 209 days, which had 
been classified as disturbed from consideration of the D curves alone, were quite 
ordinary so far as the II curves were concerned, other days which had been treated 
as ordinary from the point of view of the D curves were conspicuously disturbed 
from the point of view of the IT curves. V is a much less disturbed element at Kew 
than I) or H, and many of the V curves from the 209 days classified as disturbed 
could have been smoothed satisfactorily. The decision reached was to regard the 
209 days already selected as representing disturbed conditions for all three elements. 
Diurnal inequalities were derived from these, and these only, as representative of 
disturbed conditions. In their case the curves were read absolutely unsmoothed, at 
exact hours G.M.T. Notwithstanding the large irregularities in individual days, 
diurnal inequalities were obtained of a fairly regular character. # 
§ 2. Coming now to ordinary days, it was decided in the case of H to set aside the 
209 days already mentioned, and in addition all days when the TI curves were too 
disturbed to smooth, and to derive inequalities from the remainder. These were 
smoothed, when it seemed expedient, exactly in the same way as the D curves had 
been. In the case of V a different procedure was adopted. There was no single 
month in which a large majority of the curves could not be satisfactorily used 
without any smoothing. This being so, it seemed best to dispense with smoothing— 
which everyone admits is open to certain criticisms, while some dispute its necessity— 
though that entailed omitting a considerable number of days additional to the 209. 
The discussion of D results hardly comes under the present memoir, but D enters 
with H into such quantities as the north and west components (N and W), which of 
necessity are treated here. Thus particulars of the number of days’ traces actually 
used for the ordinary day D inequalities concern us as well as the corresponding 
data for D and H. It is simpler to enumerate the days not used than those used. 
This is done for individual years in Table I., and for the 12 months of the year 
in Table II. The total number of days in the 11 years, it should be remembered, 
was 4017. 
Natural disturbance was not the sole cause of omission of days. A feAv of the 
days—10 in the case of H, 19 in the case of D and V—were omitted owing to 
imperfections in the records. One cause of imperfection, stoppage of the clock, 
affected the three elements alike. Another cause, insufficient gas suppfy, affected all 
to some extent, but while one trace might be invisible another might be measurable. 
* ‘Phil. Trans.,’ A, vol. 210, p. 271, and ‘Collected Researches, National Physical Laboratory,’ vol. 7, 
p. 1. 
2 d 2 
