myself of this by the identity of the results in several repetitions of the experiment. This also showed me that the electromotive power of the magnet, at least after having already undergone several removals, did not become weaker; proofs of this will also be furnished in some experiments hereafter to be mentioned. In the above-described arrangement of the apparatus, I could now with the right hand perform the removal of the armature from the magnet, which was fixed to a table, while at the same time my eye observed in the telescope the consequent deviation of the index of the multiplier. This index was a thin lath, which was fixed by means of some wax to the wire which served as a common axis for the two needles of the multiplier, and formed a diameter of the graduated circle. Being thus able to observe the deviation for every result which was to be deduced therefrom, first on the one and then on the other end of the index, I freed this result from the influence of the eccentricity of the axis of the needles, and turning first the end A and then the end B of the spirals towards the north arm of the magnet, and allowing the needles of the multiplier to deviate first on the one side then on the other, I made the result independent of a second error which arises if the cocoon threads to which the needles of the multiplier are suspended possess a rotatory motion. Further, I carefully avoided every disturbance of the multiplier during a series of combined experiments, because it is impossible that every coil of the multiplier could act in the same manner as another (this would presuppose that they were all in the same plane, and parallel to one another), and because even if this might be presupposed, the action would still vary according as the needle when stationary might be exactly parallel to the coils, or form a greater or less angle. The positions of the needles when at rest seldom differ more than 0°·3 from one another. According to the above statement, a complete experiment always demanded four observations, namely, two (at both ends of the hand) for the position where the end A of the spiral was turned to the north pole, and two where B was directed to the north pole. Besides this I have repeated almost every experiment twice over in order to convince myself that no accidental fault had crept into the reading off; if the two observations differed much from one another, I again repeated each of them. The first preparatory experiments were made on the influence of the combinations of the conducting wires with the electromotive spirals and with the wires of the multiplier, in order to see whether I should content myself with winding the ends of the wires, which had been freed from their silk and were clean, very closely round one another, or should be obliged to produce a closer connection, for instance by immersing them in quicksilver. I proceeded on the supposition that if the connection effected by winding them many times closely round one another was not sufficient, an increase of convolutions
