192 
DR. C. CHREE: DISCUSSION OF IvEW MAGNETIC DATA 
improbably, differ slightly from what they would have been if derived from magneto¬ 
graphs recording N, W, T and I directly. The differences, however, cannot be 
serious, because all the largest disturbances naturally figured on each of the three 
lists. 
§ 3. The full publication of magnetic data normally includes tables of the hourly 
values of three elements in absolute measure. Besides hourly measurements of the 
curves, this entails for each element a knowledge of the scale value and of the base 
value for each day. When a temperature correction is necessary, and the tempera¬ 
ture alters sensibly throughout the day, it entails further a knowledge of the 
temperature coefficient and measurements of the temperature records at each hour. 
Two months’ hourly values of a single element in ordinary type fill a quarto page. 
Thus full publication of 11 years’ data would have filled 198 quarto pages simply with 
the hourly values of D, H and V. As there was no prospect of publication on this 
scale, and economy of effort was important, no more was done than was essential for 
the immediate object in view. So long as one can assume the base value constant 
for the whole of each day, or can adequately allow for its fluctuation by means of a 
non-cyclic correction, its value is immaterial, so far as the diurnal inequality is 
concerned. For simplicity, consider the case where no temperature correction is 
required. If the scale value can be treated as constant for the whole of a month, 
as was the case at Kew with rare exceptions, all that is necessary is to take the 
hourly measurements in millimetres, sum the hourly columns, divide the hourly sums 
by the number of days in the month, allow for non-cyclic change, find the algebraic 
excess of each hourly mean value over the corresponding mean for the 24 hours, and 
convert the inequality thus formed into C.G.S. units, through multiplication by the 
factor representing the equivalent in force of 1 mm. 
When a temperature correction is required, the inequality of temperature for each 
month can be derived from the hourly measurements of the temperature curves. 
This is converted into force from a knowledge of the temperature coefficient, and the 
result is applied with appropriate sign as a correction to the inequality already 
obtained. There would be a great increase of labour, and no gain in accuracy, so 
far as the inequality is concerned, if each hourly value of the magnetic curve were 
corrected for temperature. A second and conclusive reason for not correcting 
individual hourly values was the fact that continuous records of temperature were 
not taken in the magnetograph room until 1895. This being so, a course was 
followed which at least reduced labour to a minimum. During the 11 years no 
change had been made in the magnetograph room, or in the programme of work done 
in it or adjacent parts of the building. There was thus no reason to suspect any 
considerable change in the thermal phenomena in the room, and evidence pointing to 
the same conclusion was derivable from the 3 or 4 daily readings from mercury 
thermometers under the glass shades covering the H and Y magnets. It was thus 
decided to calculate mean diurnal inequalities of temperature, utilising the 
