From this rule then we obtain the most advantageous dimensions of such an apparatus, by knowing the limit in respect to thickness, which is determined by the nature of the steel. The thickness of the bar must not amount to much more than 1212 millimetres, as otherwise the steel cannot be properly hardened and magnetized throughout. We thence obtain the following dimensions of the deflecting bar, as those which combine the greatest advantages, namely, for its breadth and thickness 1212 millimetres, and for its length 100 millimetres. We have also the length of the magnetic needle 100 millimetres, and the smallest admissible distance between them, 300 millimetres.
By following these rules we obtain an apparatus, with which, in mean latitudes, the smallest deflections to be measured exceed 22°, as in the experiments related. At greater distances from the magnetic poles of the earth, this deflection becomes somewhat smaller; nearer to the magnetic poles it is much larger. Therefore, if these deflections can be accurately measured to within a tenth part of a degree, a final result can be obtained to within the 200th part of the force itself; since all other measurements required in the determination of the absolute intensity can be made with greater accuracy. This result, it is true, is far inferior to that which can be obtained with the magnetometer; but such results may still be of great utility in the absence of more accurate determinations.
V.
Explanations of the graphical representations, and of the table of results.
In Plates IV.–IX. are given the graphical representations of the changes of declination during six terms, amounting, in all, to forty-six curves, from fourteen stations, viz. Berlin, Breda, Breslau, Catania, Freiberg, Göttingen, the Hague, Leipzig, Milan, Marburg, Messina, Munich, Palermo, and Upsala. The graphical representations begin with the November term of 1835, when the Association was strengthened by the accession of several new and zealous cooperators. The representations of two terms of the year 1836 have been omitted, viz. those of March and May, as the changes they present are comparatively less interesting than those of the two terms of January and July, be-
