This the last, and, I believe, the best, result, is almost exactly equal to 9 for the solution. With the last apparatus and a solution of nitrate of copper, for which was measured and found 6·14, a perfect balance in both I should have liked to do a few more solutions, but something went wrong with the insulation of the bobbins, and I had no time to repair them. However, these two results appear to be enough to enable us that the equation to say is true for the junctions of the kind under examination, and that these thermo-electric phenomena are reversible. VII. " Experimental Determination of Poisson's Ratio." By C. E. STROMEYER. Communicated by LORD KELVIN, P.R.S. Received April 12, 1894. The experiments with which this paper deals were carried out between the years 1883 and 1886 by Professor Kennedy and the author, with an instrument which the latter had originally designed for measuring local strains in metal structures, but which proved itself to be so exceedingly sensitive that it seemed capable of being applied to the measuring of the cross contraction of test pieces while these were subjected to a longitudinal pull, thus providing the means for measuring Poisson's ratio direct. In its original form the instrument consisted of two small frames, which were secured to each other by means of two flat springs, in such a manner, that any relative motion was a perfectly parallel one. One of these frames carried a small piece of dark glass, and close to it, but on the other frame, a right-angled reflecting glass prism was secured. The two glass surfaces, which faced each other, were then carefully adjusted, so as to be nearly parallel, and, on throwing yellow sodium light into the prism, interference bands could be seen in the reflected light, and these would move either in one direction or the other, according as to whether the two glass surfaces, and with them their two frames, were either moving towards or away from each other. By counting the number of interference bands, which passed a mark which had been scratched on the dark glass, it was possible to estimate the amount of the relative motion of the two glass surfaces, each band representing a motion of half a wave-length of sodium light, or about 0.0000116 in. A centre point projected from the under-side of each frame, and these could be pressed against that part of the structure where it was intended to measure the variations of strains. Subsequently these centre points were replaced by two small brackets and set screws, and in this form the instrument has been 2 used in the following experiments. Fig. 1 shows a section through the instrument as altered, F, and F2 are the two frames, S1, S2 are the flat springs holding them together and keeping them parallel, G is the black glass, P is the right-angled reflecting prism, and L the ray of sodium light. B, and B2 are the two brackets, and T is the section of the test piece in position and ready for testing. It was soon found that the results which were obtained with this instrument differed materially from those which were obtained by less direct methods; it was therefore taken to South Kensington and calibrated in a Whitworth measuring machine in company with Mr. Boys, by carefully comparing the relative motion of the two screws a and b, fig. 1, with the number of interference bands which had passed the mark on the dark glass. It was found that each band represented 0.0000144 in. Evidently the spring of the brackets and difference, namely, 24 per Although the cause might of the frames must account for this large cent., over the true value of 0.0000116 in. be known, this large correction introduced an element of uncertainty, which the author hoped to eliminate by constructing a new instrument, B, fig. 2. In this sketch, T is the section of the test piece, which is pressed against the point on the frame F, by the screw S. G is the dark glass, which, as soon as T contracts, is pulled away from the glass prism P by means of the four helical springs Z, Z, which surround the columns C, C, and which are firmly secured to the frames F2, F3. The latter carry the adjustable glass prism P, which is so shaped that the ray of yellow sodium light L1 does not fall together with its reflected ray L2. The inclination of the rays of light in the narrow space between the prism P and the dark glass G was carefully measured, and found to be 19°, so that each interference band, as seen in the reflected yellow light, ought to represent a distance of 0.0000109 in., but careful measurements with the fine screw S showed it to represent 0·0000120 in., or 10 per cent. more. Both instruments A and B were used, and in the Table each experiment is marked with a distinguishing letter. In the earliest experiments (marked A1) a spirit lamp was used for illuminating purposes; it was enclosed in an asbestos-lined casing, but this soon got very hot, and must have affected the readings. Later on a Bunsen burner was used, and the test piece and instrument screened from its radiant heat. These experiments are marked A2, but even now the heat made itself felt, and the value 1/μ, last column, might in most of the experiments, as well as those marked B2, be reduced 5 per cent. In the case of those marked B3, the test piece was placed in position and the lamp lit from 30 to 60 minutes before commencing the readings. In most of the early experiments (compare Columns 3 and 4) five, ten, and even twenty bands were counted between each reading of the steelyard of the testing machine. This was not only very fatiguing to the eye, but it was subsequently impossible to determine whether any interference bands had been wrongly counted. In the later experiments, two, or at the utmost three, bands were counted for each steelyard reading. Judging by the results, the central position of each band can be estimated to within 10 per cent., and in many experiments the total number counted exceeded 20. Each test piece was strained to the maximum intended load before each experiment; but, in spite of this, the first experiments were always slightly unsatisfactory, and have generally been rejected. The author's original intention had been to use the instrument A both for measuring the longitudinal extension and the cross contrac tion, but as this instrument did not give reliable results as regards extensions, other strain indicators had to be used. I. Professor Kennedy's Lever Gear (C1). The short end of a little lever ended in a point, which was inserted into the centre punch mark at one end of a test piece. The fulcrum was connected to an arm, which was fixed to the other end of the test piece, and the long arm of the lever acted as a pointer. The leverage was 100 to 1. This instrument measured the elongation only on one side of the test piece, and would not give reliable results. In many of the experiments (those marked C2) the instrument was first fixed on one side of the test piece and then on the other. The same remarks apply to the following gear, D, and D. II. Mr. Stromeyer's Rolling-pin Gear D1. Two flat plates with projecting centre points at either end were attached to the test piece. The rolling pin, which was placed between the two plates, and held there by helical springs, was a fine piece of hardened steel wire, to which a large straw pointer was attached. In the first experiments the leverage was about 300 to 1; in the later ones it was nearly 1,000 to 1. |