ferently? This point will be fully discussed in another Report which will be laid before your Committee, and in which it will be shown where the harddrawn wires become partially annealed, and annealed wires partially harddrawn, by age. It is a curious fact that a change in the molecular arrangement of the particles of wire of some metals which may be considered homogeneous has very little effect on its electric conducting power. Thus pure cadmium*, which when cold is exceedingly ductile, becomes quite brittle and crystalline at about 80°, and returns again to its ductile condition on cooling, shows no marked change in its conducting power at that temperature; in fact, it behaves as if no such change had taken place. Again, when iron wire is heated in a current of ammonia it becomes perfectly brittle and crystalline, without altering its conducting power to any marked extent. That a wire which changes its molecular condition in becoming crystalline does not necessarily materially alter in its conducting power, is an important as well as a very interesting point, and has also been proved in the case of German silver. 3. On the effect of annealing on the conducting power.-When hard-drawn wires of silver, copper, gold, &c., are heated to redness and cooled slowly, they become much softer, and on testing their conducting powers they will be found to have increased thus: Now there is a certain difficulty in drawing a wire which is hard-drawn ; and if annealed wires be used for the reproduction of standards, the molecular condition, or perhaps the process of annealing, has an influence on the increment of the conducting power. Thus, according to Siemens, the difference in the conducting power between hard-drawn and annealed silver varies between 12.6 and 8 per cent., and that of copper between 6 and -0.5 per cent. ; according to Matthiessen and von Bose, that of silver varies between 10 and 6 per cent., and that of copper between 2.6 and 2 per cent. Again, the annealed wires of pure metals are so soft that they would easily get damaged in covering them with silk or winding them on the bobbins, so that in using them the utmost care would have to be employed in order to prevent their getting injured. 4. On the influence of temperature on the electric conducting power.-It has been shown that the conducting power of most pure metals decreases, between 0° and 100°, 29.3 per cent.: pure iron has been found to form an exception to this law, its conducting power decreasing between those temperatures 38.2 per cent. If pure metals be therefore used as standards, very accurate thermometers are necessary, as an error of 0.1° in comparing two standards would cause an error in the resistance of about 0.04 per cent. Now there is great difficulty in obtaining normal thermometers; and we must *Phil. Trans. 1862, pt. 1. † Ann. de Chim. et de Phys. 1846, t. xvii. Matthiessen and Vogt's unpublished researches. Phil. Trans. 1862, pt. 1. || Phil. Mag. Jan. 1861. bear in mind that supposing the zero-point of the thermometer is correct today, we are not at all justified in assuming that it will be so in six months time; so that we ought to redetermine the zero-point of the thermometer before using it for the above purpose. Again, it has been proved that the influence of temperature on the conducting power of wires of the same metal is not always the same. Thus, for the conducting power of annealed copper wires the following values were found: showing therefore that if standards of pure metals be used, the influence of temperature on the conducting power of each would have to be ascertained. It must also be borne in mind that it is not at all easy to maintain a standard, even in a bath of oil or water at a given temperature, for any length of time. II. Those reproduced by a given length and section or weight of a pure metal in a liquid state. The only metal which has been proposed to be used in a liquid state for the reproduction of units of resistance is mercury. We shall only have to speak of its preparation in a state of purity, and on the influence of temperature on its conducting power. For a tube, carefully filled with mercury, will certainly form a homogeneous column, and its molecular condition will always be the same at ordinary temperatures. On its preparation in a pure state. Although this metal is one of the most easily purified, yet the use of it as a standard is open to the same objections, although in a less degree, as have been advanced against the use of pure metals in a solid state when speaking of their preparation. We there stated that metals prepared by different chemists conducted differently. Now although the same manipulator may obtain concordant results in purifying metals from different sources, yet that by no means proves that the results of different observers purifying the same metal would show the same concordance. Thus we find that the values obtained by one experimenter+ for the resistance of mercury, determined in six different tubes, varied 1.6 per cent. This difference, he says, is not greater than was to be expected. The resistances found were as follows:: Experiment... 1016-52 427-28 555.38 217-73 194-70 1142.3 Calculated.... 1025.54 427-28 555-87 216.01 193.56 1148.9 Again, the values found for the conducting power of different preparations of pure hard-drawn gold, by the same observer ‡, were found equal to Phil. Trans. 1862, part 1. + Phil. Mag. Jan. 1861. The same experimenter (Dr. Siemens) states, however, in a later paper (Pogg. Ann. cxiii. p. 95), that he is able to reproduce standards of resistance by means of mercury with an accuracy equal to 0.05 per cent., but does not indicate what other precautions he takes (see remarks on the above, Phil. Mag. Sept. 1861). Phil. Trans. 1862, p. 12. 1862. L These values agree together as well as might be expected, considering that 0.01 per cent. impurity would cause these differences. Now the values obtained by different observers vary between the numbers 59 and 78. If we now take the case of copper, the values found by the same experimenters for different preparations of the pure hard-drawn metal were— * They were drawn by themselves, and all, with one exception, electrotype copper. It is well known how differently the so-called pure copper conducts when prepared by different experimenters. In the following Table, in order to show these facts more clearly, we have given the conducting powers of the metals, taking that of silver equal 100 at 0°. Silver, copper, gold, and platinum were hard-drawn. All values given, except where the contrary is mentioned, have been reduced to 0°. If now mercury be taken as unit, we find the following values: A glance at the foregoing Tables will suffice to show how badly Lenz's series agrees with the rest when mercury is taken as unit; and, in fact, we obtain more concordant results if, in the above series, we take any other metal #Phil. Trans, 1862, p. 9. This and the following Table have been copied from a paper published in the Phil. Mag. for Sept. 1861. as unit. These facts therefore seem to indicate that mercury is not yet proved to be a safe means of reproducing standards of electric resistance. The influence of temperature on the conducting power of mercury, between 0° and 100°, is, comparatively speaking, small, being only 8.3 per cent., whereas that of the metals in a solid state decreases between those limits 29-3 per cent. This property would, of course, render the use of very accurate thermometers unnecessary; for 1° would only cause a difference in the conducting power of about 0·08 per cent., and therefore 0·1 only 0·008 per cent., so that an error of 1 or 2 tenths of a degree might almost be overlooked. A fact has just come to our knowledge through Mr. Jenkin. He informs us that, having to make a report on the electric apparatus in the International Exhibition, he tested, amongst other things, several resistance-coils. Now he found two sets of coils made by the same firm, the one exhibited in the Prussian, the other in the English department. Both were said to be multiples of the mercury unit proposed by Siemens*, and their resistances determined by comparing a coil in each set with that of a tube filled with mercury. Taking each set by itself and comparing the coils in it with one another in the proper combination, they were found to be perfect; in fact, the adjustment of them was perfectly accurate. When, however, Mr. Jenkin compared coils of the two sets with each other, instead of being equal, they were found to show a difference of 1.2 per cent.† III. On those reproduced by a given length and section or weight, at a given temperature, of an alloy. The alloy on which we have to speak is that composed of two parts by weight of gold and one of silver. The reason why this alloy was proposed is that the use of (say) 1 per cent. more or less gold does not materially alter its conducting power. 1. On its preparation.-It has been shown that the alloy may be made of commercially pure metals and have the same conducting power as that made from chemically pure ones; for the maximum differences in the conducting power between those made in different parts of the world are not greater than those of a pure metal, either in a solid or liquid state, prepared by the same experimenter. But it may be urged that part of the differences obtained by different observers is due to the different methods employed in determining their conducting powers, and therefore had the conducting power of these alloys being determined by different persons, much greater differences would have been found. In answer to this, we give, in the following Table, the determination of the conducting power of several alloys by Thomson and Matthiessen ‡, independently of one another. The alloys were made by Messrs. Johnson and Matthey. of the mercury standard. This discrepancy may perhaps be attributed to some inaccuracy in the reproduction Proceedings of the Royal Society, Feb. 1861. The differences here, with the exception of alloy 6 and copper 2, may be due to the temperature at which the observations were made not being in both cases the same; for 2 or 3 degrees' difference will account for them. The Table, however, shows that different observers do obtain the same values for the conducting power of the same wires. The values obtained for the conducting power of the gold-silver alloy, made by different persons, of different gold and silver, are given in the following Table which shows, therefore, that the alloy may be prepared in a commercial way, and still have a conducting power which varies less than that of a pure metal prepared at different times by the same experimenter. If we look at the hard-drawn series, we find five out of the seven wires tested agree together exceedingly well, the greatest difference being only 0.3 per cent. These five alloys were made, three in London, by scientific chemists, one in Frankfort-on-the-Maine, and one in Brussels. Those which agree least with the others were made in New York (No. 3) and by a well-known assayer in London (No. 6). 2. On its homogeneity and its molecular condition.-If the wires of the alloy made and drawn by different persons were not homogeneous, the values obtained for the conducting power could not have agreed so well together. It has been already mentioned that some of the alloys determined by Thomson, when redrawn, were found to have a different conducting power *. Of course, here again, some of these differences are due to the temperature in each case not being the same; but the differences found with the alloys 2, 4, and 6 were undoubtedly due to faulty wires. It was for this reason * Proceedings of the Royal Society, Feb. 1861. |