^{page 669 note *} Phil. Trans. abridged, vol. xviii. pp. 29 and 355Google Scholar. The first of the conclusions is given by Mr Macdonald. He appears to have found, that the north end of the needle moved east from 7 A.M. till 5 P.M. The observations having been made from June 1794 till March 1795, the result from the observations for the first four months (June to September) should have been nearly the inverse of that from the observations for the remaining months. There can be little doubt that the instrument employed was incapable of showing the variations with much accuracy.

^{page 669 note †} Œuvres de F. Arago. Instructions, Rapports, etc., p. 152. Voyage de l'Uranie.

^{page 669 note ‡} Ibid. p. 196. Voyage de la Coquille.

^{page 669 note §} Ibid. pp. 25, 26.

^{page 670 note *} Instructions, Rapports, etc., p. 283. Voyage de la Venus. Report, dated 1840.

^{page 670 note †} Voyage autour du monde sur la Fregate la Venus. Physique, vol. v. par de Tessan, M., 1844, p. 417. See also p. 461.Google Scholar

^{page 670 note ‡} Œuvres, Rapports, etc., p. 288.

^{page 671 note *} Mr Caldecott gave the diurnal curve for the first half, and for the second half of October. In Notes to the Observations, dated 10th November 1841, which were forwarded to the Royal Society of London, in the same month, he deduced,—

He attributed the change of law at the time to the monsoon, which broke out during the month; and he caused the observations to be continued every ten minutes, in the month of November, in order to determine whether the change would continue. Mr Caldecott's remark was never published by the Royal Society of London, but it will be found among the MS. of Trevandrum Observations, in the archives of the Society, where I have myself seen it. It is necessary to add, that Mr Caldecott's observations, though affected by many errors (chiefly due to the imperfect construction of the magnetometer boxes), were sufficiently exact for the determination of this change of law.

^{page 672 note *} The details connected with the description of instruments and the precautions taken to prevent errors, will be found in the first volume of the Trevandrum Observations, which I hope will soon be published.

^{page 674 note *} This point will be examined afterwards when the lunar effect has been eliminated.

^{page 678 note *} As the moon takes more nearly 25 solar hours to return to the same meridian, it is assumed that there are 25 lunar hours in 360 degrees.

See “Proceedings of the Royal Society of London,” vol. x. pp. 482–3.Google Scholar

^{page 680 note *} I have been obliged here to enter upon laws which will be the subject of another paper, in order to explain certain apparent irregularities of solar action. I cannot leave the subject at present without suggesting that the variations of magnetic declination are probably due to currents in an electro-sphere, which, it appears to me, must surround each heavenly body. I have suggested elsewhere that the solar spots are due to disruptions of these currents within the solar atmosphere (below the photosphere, or forming part of the photosphere itself), produced by the planets, and depending for their number and magnitude on the position (latitude and distance) of the latter relatively to the plane of the solar equator. (See a letter addressed to Sir David Brewster, December 21st 1857, published in the “Philosophical Magazine,” July 1858.) If this idea has any basis, we may suppose it probable that the moon exercises some similar action upon the earth's electrosphere, an action depending for its amount on the electric tension of the spheres, these again depending upon that of the sun. Some such idea is necessary to explain the fact, that the lunar effect is very variable in its amount, varying from a movement of the free needle of about 5′, in some days, of one lunation, to a tenth part of this movement in succeeding lunations.

^{page 680 note †} See “Proceedings of the Royal Society of London,” vol. x. p. 1482, 861. I may note here that, as I showed in the “Makerstoun Observations for 1844” (Trans. Roy. Soc. Edin., vol. xviii. p. 354), the difference of the law of solar diurnal variation of declination in Europe for summer and winter was of the same kind as that betwixt the laws for the same periods near the equator, where the movement is inverted; so it followed from the inversion of the lunar law near the equator, with the sun in the northern and southern hemispheres, that in high latitudes, the lunar law should present a greater range in summer than in winter, and that the mean law in high north latitudes should be nearly the opposite of that for high south latitudes. The latter fact should have been evident from preceding observations in mean latitudes of the two hemispheres, and on the discovery of the inversal of the lunar variation with the sun's passage of the equator, I examined these observations for this end. Unfortunately the discussion of the Toronto Observations by General Sabine gave a result nearly the inverse of those derived from the observations at Makerstoun and Prague in the same hemisphere, as I pointed out in the note cited above. General Sabine then discovered that west had been substituted for east in the discussion of the Toronto Observations; and this correction made, the approximate opposition of the laws for the two hemispheres was at once evident. Of the former fact I satisfied myself by a rediscussion of the “Makerstoun Observations” rejecting the large disturbances; and it has been verified also by the late Mr Bache, from observations at Philadelphia, though only two years after the result deduced by me of the inversion of the law at the equator had made the conclusion, if not certain, at least extremely probable.

^{page 684 note *} The values in the table are taken from the reduced hourly observations, and not from the curves, which would give the ranges slightly greater in some cases.

^{page 685 note *} It should be remembered, that we are examining the ranges of the monthly mean diurnal variations, and that these are diminished near the equator, by the change of law occurring in months near the equinoxes.

^{page 686 note *} At all the stations there is a kind of double in the epoch of principal maximum, seen more or less markedly in May or June, and in August, in the northern stations; in October or November, and in February, in the southern stations; at Singapore the range for November is slightly greater than that for February. I should also notice, as a deviation from the law of change of minimum from one station to another, the occurrence of the minimum oscillation at Madras in December, instead of November, as its position betwixt Bombay and Trevandrum would indicate. I feel inclined to believe that this deviation would not have appeared had the ranges for Bombay and Madras been derived from observations during the same years.

^{page 687 note *} Observations were also made at a height of 6000 feet above Trevandrum, and twenty miles W.N.W. of it.

^{page 687 note †} Trans. Roy. Soc. Edin. vol. xviii. p. 354.Google Scholar

^{page 687 note ‡} It is not meant that the change was the same in amount in high latitudes and near the equator, though, when we consider the variations of the horizontal needle at Makerstoun and Trevandrum, the changes are nearly equal.

If we subtract the curve with maximum movement from that of minimum movement of the free needle at each station in the higher latitudes where the curves follow nearly the same law, or the two curves of maximum movement from each other where they have opposite forms (as the curve of August from that of December at Trevandrum, and that of August from the curve of February at St Helena), we shall find that the change of movement, or the range of the difference curve, is three or four times greater near the equator than in high latitudes, St Helena being the station of greatest change of form, the range of the difference curve being seven minutes and a half (7′·5).

If the variation of the movement of the free needle from summer to winter had been nearly as great at Makerstoun, Hobarton, or Toronto, as near the equator, the curve would have been completely inverted. It is simply because the variation of range is less in high latitudes than near the equator that the inversion does not take place there. This fact does not appear when we compare the movements of the horizontal needle.

^{page 688 note *} It has also been employed by the Rev. P. Secchi in the same way as by General Sabine. Monthly Notices Roy. Ast. Soc. vol. xv. p. 27.

^{page 690 note *} See Letter, dated Trevandrum, 21st December 1857. Philosophical Magazine, July 1858.

^{page 690 note †} See Letter just cited, where I have also noticed that Cassini and Mairan supposed a connection betwixt the intensity and extent of the zodiacal light and the solar spots.