III .
" On the Diurnal and Annual Inequalities of Terrestrial Magnetism , as deduced from observations made at the Royal Observatory , Greenwich , from 1858 to 1863 being a continuation of a communication on the Diurnal Inequalities from 1841 to 1857 , printed in the Philosophical Transactions , 1863 .
With a Note on the Luno-diurnal and other Lunar Inequalities , as deduced from observations extending from 1848 to 1863 .
" By GEORGE BIDDELL AIRY , Astronomer Royal .
Received July 27 , 1868 .
( Abstract .
) The author states that the instruments employed are precisely the same which were used in the second part of the former investigation , from 1848 to 1857 , mounted in the same place , and treated in the same manner .
In describing the treatment of the photographic curves , he first gives the number of days which have been omitted in different years ; because the character of the observations or curves was too disturbed to permit the usual treatment of the observations , or the drawing by hand of a pencil curve that would fairly represent the general course of the curve .
The greatest numbers of omitted days occur in the years 1846 , 1847 , 1848 ; 1851 , 1852 , 1853 , 1854 ; 1859 , 1860 .
As the estimate of the amount of irregularity has been made throughout by the same person , he considers that these years may be accepted as those in which the disturbances were the greatest .
If they point to any cycle at all , it is one of 6 or 6years .
These days being omitted , the ordinates of the pencilled curves on the other days were used as basis of ' all the following investigations .
For the solar inequalities , they were treated by collecting the measures for every complete solar day , or for every solar hour bearing the same ordinal number , according as the annual or diurnal inequalities were the subject of inquiry ; but in all cases these quantities were next grouped by months , and the monthly means were taken .
In the further treatment , the means of the monthly means of every complete day for all the months of the same name in the different years were taken and corrected for secular change ; the corrected numbers do not appear to indicate any sensible annual equation .
Then the means of the monthly means of every solar hour for all the months of the same name in the different years were taken , giving the diurnal inequalities on the mean of years for the twelve separate months ; and these present , for the declination ( north to west ) and horizontal force , for the period 1858 to 1863 , sensibly the same differences between the summer months and the winter months as those for the period 1848 to 1857 .
For the vertical-force -163 curves also , the nodal passage in both periods is earliest in the summer months ; but it is not quite certain whether the curves in autumn , in the period 1858 to 1863 , are quite so bold as those in 1848 to 1857 ; the difference , however , if any , is inconsiderable .
After this , the monthly means of every solar hour are taken through each year , giving the mean diurnal inequality of each year ; and here a very remarkable change is observable .
To explain this , it is necessary to refer to the former paper , where it is shown that the curves for diurnal inequality of the horizontal forces had very slightly increased from 1841 to 1847 , but had rapidly diminished from 1848 to 1857 , giving the smallest and most winter-like curves in 1856 and 1857 .
Now it is found that from 1858 to 1863 the curves have increased , with a little irregularity in 1861 , till they are sensibly as large as they were at first .
Thus1858 nearly resembles 1856 1859 , , , 1851 1860 , , , 1850 1861 , , , , 1851 1862 , , , , 1847 1863 , , , 1841 With regard to the diurnal inequality of vertical force , it appears that the curves gradually increased in boldness to 1855 , and have gradually diminished to 1862 .
The nodal passages , it was remarked in the former paper , had been much accelerated in the hour of the day , from 1842 to 1857 .
Now , from 1858 to 1863 , the hours of nodal passages have been retarded , till in 1863 they are again nearly the same as in 1848 .
In all these remarkable changes there is no appearance of cycle .
The author then proceeds to the treatment of lunar inequalities from 1848 to 1863 .
The bases of their treatment were thus obtained : the exact time of moon 's transit was laid down on the time-scales of the photographic sheet , and the intervals were divided into lunar hours , and a new system of ordinates , corresponding to the lunar hours , was measured to the pencil curves .
The system of grouping was precisely similar , mutatis mutandis , to that for the solar inequalities .
First , for the menstrual inequalities .
The declination seems to exhibit a distinct lunar menstrual inequality , with + maximum about the fifth day of lunation ; the horizontal force seems to show a lunar semimenstrual equation with maximum about the second day ; the vertical force shows nothing certain , proving only that , if there is anything , it is very small .
Secondly , for the lunodiurnal inequalities .
The luno-diurnal inequalities in declination and horizontal force on the mean of 1858 to 1863 agree so closely with those on the mean of 1848 to 1857 , as to leave no doubt of their existence and law as luno-semidiurnal inequalities , with no trace of luno-diurnal or other inequality .
Remarking the singular difference for different years which has presented 164 [ Dec. 10 , itself in the discussion of the solar inequalities , it appeared to the author very desirable to examine whether there is any discoverable difference in the lunar inequalities for the same years .
The years were accordingly thus divided : Large solar curves. .
1848 to 1852 , 1859 , 1860 , 1862 , 1863 .
Small solar curves. .
1853 to 1858 , 1861 .
On discussing these , it was found that in all cases the lunar horary epoch for the inequality was sensibly the same for years of large solar curves and for years of small solar curves ; but the coefficient was different .
The value of the fraction lunar semidiurnal inequality in years of large solar curves lunar semidiurnal inequality in years of small solar curves is For declination ... ... ... .
1 35 For horizontal force ... . .
125 The author remarks that it would seem possible to suggest two conjectural reasons for this remarkable association in the time-law of changes of solar effect and lunar effect .
One is , that the moon 's magnetic action is really produced by the sun 's magnetic action ; and a failure in the sun 's magnetic power will make itself sensible , both in its direct effect on our magnets and in its indirect effect through the intermediation of the moon 's excited magnetism .
The other is that , assuming both actions ( solar and lunar ) to act on our magnets indirectly by exciting magnetic powers in the earth , which alone or principally are felt by the magnets , the earth itself may have gone through different stages of magnetic excitability , increasing or diminishing its competency to receive both the solar and the lunar action .
The epochs of lunar inequality in western declination from north and in horizontal force to magnetic north are sensibly the same ; and the coeffi .
cients expressed in terms of horizontal force on the mean of all the years are sensibly the same , and equal to 0'000061 .
The direction of the composite disturbing force is therefore sensibly N.W. and S.E. magnetic , or ( roughly ) in the direction of a line from the Red Sea to the south of Hudson 's Bay .
It may be remarked in opposition to this that the solar diurnal action is mainly in the S.W. direction .
The luno-diurnal inequality of vertical force on the mean of all the years appears to consist of a luno-diurnal and a luno-semidiurnal term .

