was obtained from observations on short lengths of cable immersed in water at various temperatures. The gutta percha covering the long cables was used in alternate layers, with a varnish known as Chatterton's compound. The short lengths, tested in water of various temperatures, were three in number, and will be called numbers 1, 2, and 3. The length of each was one knot. The diameter of the gutta percha, in numbers 1 and 2, was 0.3 in., covering a single copper wire of 0.06 in. diameter. No. 3 was a knot of the Red Sea core: the diameter of the gutta percha was 0.34 in., covering a copper strand formed of seven wires, each 0.038 in. diameter. The coils were placed in a felted tub, and were covered with water of the desired temperature for several hours before the experiments were made. The loss or escape of electricity was measured on a delicate zinc galvanometer. Separate tests were made with the positive and negative poles, and at each test five readings were taken; the first one minute after the application of the current, the others at successive intervals of one minute. On coil No. 1, covered with Chatterton's compound and gutta percha, there was a marked difference between the tests made with positive and negative currents; whereas in coil No. 2, covered with pure gutta percha, there was no difference between those tests from 50° to 75° Fahrenheit. In both coils and at all temperatures the loss decreased rapidly during the first two minutes after application of the battery; this decrease continued till the fifth minute, when a minimum was nearly attained. This effect in coil No. 2 was regular, and between 50° and 80° Fahrenheit was not affected by a change in the size of the current. In coil No. 1 the decrease of loss was regular and nearly constant when the zinc pole was connected with the coil; whereas irregular results were obtained when the copper pole of the battery was so connected. The extra resistance or decrease of loss was still more marked in coil No. 3, where the gutta percha is of larger diameter. In this coil the loss decreased 30 per cent. in the interval separating the first and fifth minute. The same phenomenon (decrease of loss) was observed on the long cables. The insulation of a sound gutta percha covered wire is therefore improved by the application of either a positive or a negative current. It also appears that it is most necessary, in testing the insulation of a cable, to record the time separating the observation and the first application of the battery. The phenomenon of decreased loss or extra resistance is observed, whether the cable is dry or immersed in water. The annexed Tables I. and II., showing the relative loss from the two coils at different temperatures, give the result of the experiments as deduced from curves which result from the observations after all due corrections have been made for loss on connexions and varying electromotive force. The numbers in the Tables give no absolute measurement, but only the relative loss at the various temperatures. At 65° the two coils 1 and 2 test much alike. At high temperatures pure gutta percha rapidly deteriorates; at low temperatures it has the advantage. The irregularity of the copper tests of No. 1 lead to some suspicion of a chemical action between the various substances in contact. Table No. III. contains the results of similar experiments on coil 3. The absolute resistance of the insulating cover was obtained by comparison of the current flowing through the gutta percha, and that flowing through a coil of known resistance, or through the copper core itself. The actual resistance of the insulating cover being known, this formula, due to Professor Thomson, was used to determine the 2πί specific resistance, or the resistance of a cubic foot: R where R resist = ance of cylindrical coating, l= length of wire tested, ratio of the diameter of the copper core to that of the gutta percha covering, a specific resistance of the material, The specific resistance of gutta percha of the Red Sea core, at 6°Fahrenheit, calculated from daily tests of the Red Sea cables, was found to be 205 × 1018 absolute British units. Owing to the influence of the phenomenon of extra resistance described above, this number does not express a well-defined resistance such as that of a metal, but gives an approximate resistance at about thirty seconds after application of the battery. The corresponding specific resistance at 60° of No. 1 coil was found to be 195 × 1018. The specific resistance of No. 2 coil at the same temperature was 202 × 101. The resistance at other temperatures, or after longer application of the battery, is inversely proportional to the numbers given in the three Tables, and representing the relative loss. The employment of Chatterton's compound in different proportions from those used in the above coils, would necessitate fresh experiments in each case to determine the effect of temperature. Since writing the above paper a detailed account of the experiments, and the results of more extended calculations, have been communicated to the Royal Society in a paper read March 22, 1860. On the Retardation of Signals through long Submarine Cables. By FLEEMING JENKIN. This paper contains the result of experiments made on long submarine cables, at the establishment of Messrs R. S. Newall and Co. * Professor Thomson's theory is confirmed by these experiments, which indeed were only rendered possible by the use of Professor Thomson's Patent Marine Galvanometer. The deflections of the magnet in this galvanometer are read by means of a spot of light, reflected by a mirror attached to the magnet, and brought to a focus on a scale at about 22 inches from the mirror. The magnet and mirror together weigh only 1 grain, and a very small angular movement of the magnet causes the spot of light to move over many degrees of the scale. By this instrument a gradually and rapidly increasing or decreasing current is at each instant indicated at its true strength. When therefore this galvanometer is placed as a receiving instrument at the end of a long submarine cable, the following phenomena are seen. At the moment of completing the circuit through battery, cable, and earth at the sending end, no movement of the spot of light occurs. In a second or less the spot begins to traverse the scale, at first slowly, then rapidly, and again more slowly, until after perhaps a minute a maximum is attained. The interval of time which elapses between the first completion of the circuit and the arrival of the spot of light at various divisions of the scale was measured, and the observations being thrown into a curve, have given what may be termed the "curve of arrival" for various lengths. Instead of one continuous current, broken currents, such as form dots and dashes in the Morse alphabet, were also sent with regularity by means of a metronome, and the movements of the spot of light corresponding to the various signals observed and delineated by curves. The following are the results of the observations :- 1. The strength of the battery used does not affect the speed of transmission; to prove this, the curve of arrival was taken with 72 cells on a length of 2168 knots and with 36 cells. The two curves coincided when drawn to scales proportionate to the electromotive force of the two batteries. Thus is once more proved the fact that the speed of electricity is independent of the power of the battery, the current always reaching the same fraction of its maximum strength in the same time. 2. The same curve represents the gradual increase of intensity in a current when arriving, and the gradual decrease of intensity caused by putting the sending end of the cable to earth. 3. The curves of arrival, as obtained from lengths of 1000 to 2000 knots, agree in general character with those given by Professor Thomson's formula. Some discrepancies appear, due probably to electro-magnetic induction between the coils, and also in great measure to the varying resistance of the insulating cover, described in the author's paper "On Gutta Percha as an Insulator." 4. Whatever length of the cable was used for experiment, the amplitudes of oscillation representing dashes, or A's or other letters, were found to bear a constant proportion to the amplitudes representing simple dots sent at the same speed. This proportion is, however, different for each amplitude. Thus on a length of 2191 knots the speed of 15 dots per minute reproduced the same amplitude of oscillation as a speed of 30 dots on a length of 1500 knots; and the same relative speeds reproduced the same oscillations for dashes, A's, &c. in the two lengths. The amplitude is in this paper supposed always to be measured as a fraction of the maximum deflection obtained by keeping the circuit completed till the spot of light comes to rest. 5. The speed at which signals could be received on a relay, is easily perceived when the groups of oscillations are graphically delineated. A certain constant amplitude of dot corresponds to this speed (vide § 4). The speed at which a given amplitude of dot can be produced, varies inversely as the square of the length; and therefore the speed at which signals can be received by a relay varies also inversely as the square of the length of the cable. * Proceedings of the Royal Society, May 1855, republished in the Philosophical Magazine. 6. By the usual hand-signalling, it was found just possible that legible groups of dots and dashes should be received through 1800 knots at a speed of 20 dots per minute. 7. The amplitude of oscillation due to various relative speeds can be thrown into a curve which is the same for all lengths; and since the law of retardation does not depend on the nature or dimensions of the material forming the cable, we can by means of this curve determine from one single observation at any speed, the amplitude of oscillation which will be due to any other speed, or in other words, the possible speed of signalling. 8. The maximum speed of signalling by any given system corresponds, as has been observed, to a certain amplitude of oscillation produced by successive dots. The actual amplitude necessary for each system must be determined by experiment. For Morse-signals sent by hand, it can hardly be less than 15 to 20 per cent. of the maximum strength of current due to the battery used. Mechanical senders would greatly increase the speed at which signals can be transmitted. 9. A comparison was made between signals sent by alternate reverse currents and those sent by alternate contacts with one pole of the battery and earth. One diagram would serve for both sets of signals, by simply drawing a line parallel to the base-line of the curve at half the height of the maximum, this line being taken as the base- or zero-line for the signals sent by reverse currents, all deflections above this line being called positive, all those below negative. 10. The use of reverse currents is of advantage in the first signals sent after the line has been completely discharged; the nature of this advantage may be briefly indicated by pointing out that, when no signals are being sent, the spot of light rests on the base-line, which in the common system is at a remote part of the scale from that at which the dots and dashes appear, but in the system of reversals is in the very centre of that portion of the scale. The conclusions and experiments were, in the original paper, illustrated by diagrams. On some of the Methods adopted for ascertaining the Locality and Nature of Defects in Telegraphic Conductors. By CROMWELL F. VARLEY, Electrician of the Electric and International Telegraph Company, and of the Atlantic Telegraph Company, &c. The author said the plans adopted by him were various: viz.— Case 1.-When a conductor "makes dead earth," i. e. the connexion between the conductor and the earth offers no appreciable resistance, the operation is very simple, and consists solely in ascertaining how much resistance the conductor in question offers to the passage of electric currents. Modes of Measuring Resistance. He preferred using a standard of resistance and a differential galvanometer. A current from a battery, whose positive pole is connected to the earth, is made to divide and pass round the differential galvanometer in opposite directions. The one half of the current is made to enter the cable whose resistance is to be measured, and the other half to go through the resistance coils to the earth. So much resistance is then included in the latter circuit as shall make the divided currents equal in force, when the needle will stand at zero. The number of resistance coils required to make the needle stand at zero indicates the resistance of the conductor; and if the defect in the insulation be so large as to offer no appreciable resistance at the fault, the amount of resistance will indicate the locality of the fault. When no resistance coils are at hand, the following method may be adopted :1st. Having a galvanometer whose resistance is known, and a Daniel's battery, first ascertain that each cell is in good order, and offers no appreciable resistance compared with that of the galvanometer. 2nd. Connect one pole of the battery to the earth, and the other through the galvanometer to the cable. 3rd. Note the deflection. 4th. See how many cells will give the same deflection when only the galvanometer is in circuit. 5th. Repeat 2nd, 3rd, and 4th with various powers, and take the mean of many results. - The following formula will then give the resistance of the conductor:- Then Number of cells required in operation No. 3 ". n. n'. x. With moderate care this will indicate the resistance to within 2 or 3 per cent. Care must be taken that the earth-plate used, and also the battery-cells, offer no appreciable resistance. Resistance coils and a differential galvanometer are much more exact, and should always be used if possible. The author's standard, which has been adopted by the Electric and International Telegraph Company, the Atlantic and other Telegraph Companies, consists of the following units: 1, 2, 3, 5, 10, 20, 30, 50, 100, 200. These allow of the coils being checked by themselves; thus 1+2=3,2+3=5,2+3+5=10, &c., which is very useful in practice. Powerful currents must not be allowed to flow long through the coils, because they are thereby warmed and their resistance increased. Case 2.-When (as is almost always the case) the fault itself offers resistance, but the conductor is otherwise perfect, one of the two following methods will indicate with sufficient precision the amount of resistance due to the conductor between the operator and the fault, and also that of the fault, the former being the distance of the defect: : 1st. Have the conductor disconnected at the distant end (B), and This is the resistance of the conductor between the operator's 3rd. The resistance of the conductor alone when perfect. =R =r =S Calling the distance or resistance of the cable between the operator and the fault, y the resistance of the cable between the fault and B, and This operation should, if possible, be repeated at the end B, which will indicate the possible amount of error. Plan No. 2 requires that there be at each end galvanometers of known resistance, |