and the time of them may be compared with the phases of the aurora, which are contained in the record for that day. The excursion at g was so great, that it was found necessary to curtail it on the plate; but the extent will be readily seen from remarking that it reached to 71.4 on the scale, 47'.2 of which were traversed in 11 minutes of time. An aurora was seen on the same night at Philadelphia, New Haven, and at Toronto, U. C. A description of its appearance at New Haven may be seen in Silliman's Journal, No. 1, vol. xxxix.* Where facilities existed for making the observations, it was discovered to be accomplished with similar effects upon the magnetic declination as were felt at Cambridge. The magnetometer at Philadelphia experienced great derangements, although the limits were less, not exceeding 55'.8. The influence which an aurora exerts upon the earth's magnetism reaches as far and wide as the appearance itself; and probably the intensity of the effect is proportional to the brilliancy of the display. The greatest disturbance of the magnetometer at Philadelphia was, as at Cambridge, between 4 and 5h A. M., Gott. M. T. The deflection of the instrument at Cambridge amounted to about 57 minutes, and the extremes were separated by little more than 2 hours. Lieutenant Riddell informs us that at Toronto the arc traversed was 1° 59′, which was never equalled, and approached but once on a similar occasion. We also learn from him that an aurora was noticed at Greenwich, Great Britain, on the same day; but he adds, that the disturbances there and at Toronto were very different. Such full information is not possessed in regard to the aurora of August 28-9. It is evident, from the observations, that the magnetometer at Cambridge was more affected on that day than ever before, the whole change of declination amounting to 61'. At Toronto, where the aurora was also seen, the disturbances were equally surprising, and produced an oscillation of 1° 33′ in declination. The greatest amount of derangement at Cambridge was as follows:At 1' 23' A.M., Gott. M.T., the reading of the scale was 111.9 Range of 52'.4 East in 1h 10'. 164.3 Range of 52'.9 West in 1 hour. "2 35' 66 "3 35' 66 66 Again, at 5h 20′ A.M., Gott. M.T., the reading of the scale was 108.5 Range of 47'.6 East in 25 minutes. During the first of these periods, the aurora reached its culmination of splendour; between 5 and 7h it was faint and near the * See also the Journal of the Franklin Institute for June, 1840, which contains some observations made upon it at Southwick, Mass. H horizon. It does not appear, from an examination of the May or August term-day, that the maximum agitation of the magnetometer coincides in time with the greatest brilliancy of the heavens. In May, it had not accumulated its action when the aurora began to decline; and in August, although it accompanied the display, it continued with undiminished energy one or two hours after that had passed away. The most rapid motion of the bar was from 5h 20' to 5 45', being equal to 47'6. in 25 minutes. This is nothing strange; but might be expected from the time which all the forces of nature consume in communicating themselves to bodies, and penetrating large masses so as to overcome their inertia. More exact and frequent observations will doubtless conduct to a better knowledge of a connexion which is now so undeniable, and yet so imperfectly understood. If observers are careful to note the times at which the chief phases of the aurora are witnessed, and its position among the stars, and, where they have the opportunity, the simultaneous variations of the magnetometer, we may not despair of elucidating these two classes of intricate and interlaced facts-the aurora and the irregular perturbations of the magnetic meridian. It must not be inferred that other causes do not exist, in co-operation with the auroral phenomena, to derange violently the earth's magnetism. According to Ampere's theory of currents, a large fund of such derangements must be deposited under the crust of the earth. The equilibrium, although permanently stable, must be subject to constant fluctuations. Theory supplies the reason, and observation asserts the fact. Many of the small daily derangements have no apparent relation to the aurora; and in regard to the magnificent strides of the magnetometer, it cannot be told which is cause and which is effect. If further search shall prove that an aurora never fails to attend a great disturbance, we may conclude that the aurora itself is seldom displayed in the daytime; for the remarkable changes of declination almost always begin during the night, and seldom continue into the next day. If, however, an unseasonable aurora should occasionally arise, we may be able to perceive indications of its presence from the magnetic perturbations, although its light were eclipsed by the brightness of the sun. So far we have attended to relative only, and not to absolute declinations. The former are sufficient when the object is to find the times of maxima and minima, the daily range, and the diurnal curve. But it sometimes becomes necessary to know the absolute declination, so that the process will now be described of referring any reading of the scale to its absolute value. It is clear that if the absolute value of one reading can be ascertained, that of all the rest is known at once. It is convenient to have the absolute declination always referred to the same number of the scale; we will suppose this number to be, therefore, 100. To find, then, the absolute declination corresponding to 100 of the scale, we proceed thus :-The variation-transit, with which the Gauss magnetometer is observed, is placed firmly on the table, and then the line of collimation is adjusted in the meridian by means of polaris and the large transitinstrument; a section having been made in the roof for this purpose. The azimuthal circle is then read off. The circle is now turned until the line of collimation coincides with the direction of the magnetic meridian, as indicated by the needle that accompanies the variation-transit. At the same moment the circle is read off again, and the scale is noted through the telescope. The difference of readings from the azimuthal circle will indicate the angle which the magnet makes at that time with the astronomical meridian, and the reading from the scale shows to what number on it this absolute variation corresponds. Now since, according to their arrangement on the scale at Cambridge, an increase of numbers implies a decrease of declination, we readily find the absolute declination of 100 of the scale by adding or subtracting, as the case requires, the difference between 100 and the reading at the time. This will be easily understood from the following example : The azimuthal reading by Polaris, June 21, at 8 P.M., The azimuthal reading at the coincidence of the 310° 34' 40" 301° 15' 30" When the reading of the scale through the telescope was 100,835, the absolute declination.... The absolute declination at 100 of the scale As it may not always be possible to take an astronomical observation, on account of the state of the atmosphere, the azimuthal angle between some fixed mark and the true meridian is read off, and the position of the magnetic meridian determined by reference to this. Thus it appears, June 25th, that a certain mark on Gore Hall, which has been previously found to be 38° 11′ west of the north, reads on the circle Absolute variation for 102.966 of the scale = 66 for 100 ... 66 9° 19′ 40 The absolute variation corresponding to 100 of the scale being known, the real values of all the lower numbers are found from it by adding, and of all the higher numbers by substracting, the difference between them and 100. Here we suppose, of course, that all the readings on the scale are made in the same position of the tels escope as the one by which the original absolute variation was determined. To secure this condition, when the observations begin, the azimuthal circle must be firmly clamped at some place which is considered the fixed reading for this period; and the vernier should be occasionally examined, to see that the instrument has not been deranged. The absolute declination thus obtained cannot be relied on within so small a limit of error as that to which the changes of declination are subject. The chief uncertainty attaches to the coincidence of the needle with the line of collimation. Several readings, repeated in succession, are likely to vary three or four minutes, so that their mean is only an approximation to the truth. Hence the difficulty of ascertaining the yearly change of declination, which is so small as to be partially masked under accidental errors. If the feet of the variation-transit were firmly secured to some durable foundation, the yearly variation might be found at once from the scale, as we now find the daily ones. In this case, the fixed reading should not be altered from one set of observations to another. Hereafter, as the observations will be made with Lloyd's declination magnetometer and a fixed telescope, we shall not be subjected to this inconvenience. For greater accuracy, the absolute variation assigned to 100 of the scale for any period should not depend upon a single set of readings. But the process which has been described for finding the real value of any part of the scale, should be repeated as often as possible during the days of regular observations. Thus we have:— June 21 at 8h 0' P.M., Gott. M.T., the We now pass from the absolute value in declination of 100 of the scale to the absolute declination in this way. If the absolute variation for any single moment were desired, we should readily find it by adding or subtracting, as the case required, the difference between the reading on the scale and 100 to the absolute value of 100. But when speaking of the absolute variation, we generally intend some mean value which is the representative of that element for a whole day or month, or perhaps a longer period. We should certainly miss of this mean variation, if we adopted the regular maximum or minimum reading of the scale, or that of any extraordinary stride which may have been observed during the period. No single mark on the scale can lead to anything more than a momentary expression of this element. The mean of the daily maximum and minimum limits of the scale, or, more accurately, the mean of all the observations, furnishes a mark from which some durable value of absolute declination may be derived. Now, the mean of all the observations belonging to the 10 days of June is 103'.603 of the scale : Hence the variation sought is.. Again, the mean of the observations made during the 5 days of October is 98'.838; hence the variation is The mean absolute declination from June 9° 20′ 59′′-3′ 36′′ or 9° 17′ 23′′. 9° 17′ 29′′+1' 10" or 9° 18′ 39′′. to November, 1840, may be considered = 9° 18′ 01′′. The following table shews the variation of the needle at Cambridge and in the vicinity, from the period of the earliest observations : As there can be no great difference between the variation for Boston and Cambridge,* we infer, from the table, that from 1708 to 1793 the declination diminished at the mean annual rate of 1'.8, *The following values of the latitude and longitude are taken from the American Almanac for 1840: |