FROM ORDINARY DAYS OF THE ELEVEN YEARS 1890 TO 1900. 
193 
thermogram measurements made on magnetic quiet days from 1895 to 1900, and 
these were taken as applying to the whole 11 years. This attributes to each 
January, for instance, the same diurnal inequality of temperature, while it was no 
doubt larger in some Januarys than others. This procedure no doubt introduced 
slight errors into the inequalities for individual months of individual years, but 
these would tend to disappear in results from groups of years or from the whole 
11 years. 
The temperature correction of the TI magnetograph is about Y7y per 1° F., and the 
range of the temperature diurnal inequality in most months was under 1° F, so a 
difference of temperature lag of even 1 or 2 hours between the magnet and 
thermograph would have had little effect. The fluctuations of temperature in the 
magnetograph room from one day to the next are sometimes much greater than the 
range of the regular diurnal variation, and the comparison of the readings from 
the thermograph and mercury thermometer on the one hand, and the corresponding 
fluctuation in the base values derived from individual absolute observations on the 
other, afforded grounds for confidence that uncorrected effects of temperature in the 
H inequalities must be trifling. The Y magnet has a much larger temperature 
coefficient—about 12'5y for 1° F.—and the range of the regular diurnal variation is 
considerably less in V than in H. Thus there is more reason to fear uncorrected 
temperature effect in the Y inequalities. It is, however, mainly in the absolute daily 
range— i.e., the difference between the highest and lowest values throughout the 
day—that temperature uncertainty comes in. Undoubtedly some individual daily 
ranges, especially those of Y, suffered considerably from this cause. In the case of an 
element like Y at Kew considerably affected by temperature, it is sometimes difficult 
to recognise the maximum or minimum. If the temperature change in the day has 
been large, the maximum force may come at quite a different hour from the maximum 
ordinate. It may be necessary to take half a dozen measurements of the force curve, 
with the corresponding thermogram measurements, before one can decide. This is 
especially true of quiet curves—and most Y curves are quiet—at seasons when the 
regular magnetic diurnal variation is small. No temperature corrections, of course, 
were possible until 1895, and mean results based on a number of days would have 
been practically useless in dealing with individual days. It was accordingly decided 
to attempt no temperature corrections to absolute daily ranges, but to derive them 
from the magnetic curves as if these required no temperature correction. 
The neglect of temperature could hardly prejudice one’s view of the character 
of the day as quiet or disturbed, but undoubtedly in a few cases it led to a quiet 
day being assigned a Y range more appropriate to a day of moderate disturbance. 
The neglect of temperature may even have exerted a slight effect on the estimate of 
the mean monthly value of the absolute range, as the occurrences of maximum and 
minimum are much more numerous at certain hours than others, and so the effect ot 
the regular diurnal variation of temperature would not be wholly eliminated. 
