in all of which a current at least as small as 10-8 amperes is pronounced audible, would indicate, not to mention the more striking results of Tait and Preece. They agree in order with the data of Lord Rayleigh for frequencies between 300 and 500. Perhaps, however, it would be more fitting to regard the extent of the silent interval as the sensitiveness of the telephone, in which case my results are of the order of those found by Lord Rayleigh for the 4-foot octave.

Throughout my work, it will be remembered, the telephone was excited by isolated taps, and pains were taken to assist the ears by aid of sounding-tubes. One therefore expects a difference of behaviour in the two cases.

In relation to the measurement of electromotive force, therefore, the telephone does not quite come up to the requirements. By using the shades of audibility at the margins of the silent interval for the determination, it is possible to define electromotive forces of the order of one volt with an accuracy of one in one thousand. For thermoelectric forces like those occurring in pyrometry, the telephonic method is available only in rough work. Thus a single iridio-platinum couple (20 per cent. iridium in the alloy), with its hot junction kept at low red heat, showed the following values of apparent electromotive force, e, at the margins of the silent field:

Noise. e=9800 microvolt. Mean Temperature =837° C.

9480

Mean

834°

795°

815° C.

792°

The temperatures corresponding to the two apparent values of thermoelectromotive force on each side of the silent field have been added. Hence the interval of silence corresponds to about 40° Centigrade; but the sensitiveness at the margins of the interval is such that the mean temperature deduced is in error only 2° or 3°. However, close measurement like this can be made only under conditions of exceptional silence in the surroundings. In a noisy room increments of less than 10° at 1000° will escape detection, though the results will be more favourable in proportion as temperature increases.

In endeavouring to account for the occurrence of the silent interval, one may note first that the thickness of the diaphragm has little influence within the limits of '015 centim. to 030 centim. employed. The range of silence does decrease, however, when the telephone is made more sensitive either by modifications of the key or, again, by moving the armatures and coils closer to the diaphragm when other things are left unchanged. In such cases the forces actuating the plate are

increased. To invoke the inertia of the plate seems inadmissible, seeing that telephones with diaphragms* fifteen (15) centimetres thick have been made to respond (Bréguet).

I have therefore hazarded a straightforward supposition †, that the interval of silence is to be referred to a molecular inertness of the plate of the telephone very similar in its nature to quiescent friction. The diaphragm at rest seems initially to resist further deformation, and this resistance must first be overcome before the telephone will respond with nicety to fine gradations of the actuating stress. In other words, as soon as the internal friction encountered in moving the diaphragm has passed from the quiescent to the kinetic stage, the instrument is ten or more times more sensitive than it was before the critical stress value had been exceeded.

Washington, D.C., U.S.A.

LXVI. Note on the Measurement of the Specific Inductive Capacities of Water, Alcohol, &c. By REGINALD A. FESSENDEN, Professor of Electrical Engineering, Western University of Pennsylvania ‡.

Τ

IT appears to have been hitherto accepted that the high values for the specific inductive capacities for water, alcohol, and some other similar fluids which have been obtained by various experimenters are correct. If this were so, Maxwell's rule for the relation between specific inductive capacity and the index of refraction would not hold in these cases for wave-lengths of visible light, and the phenomena of dispersion &c. have been called in to explain this anomaly. It may be well to point out that these high values are not correct, but that the true values are in every one of these substances very nearly equal to that called for by the theory.

This fact was first noticed by the writer in 1891. A sensitive electrometer had been constructed, with two fixed and one movable cylinder, mounted on jewels, and provided with a commutator, so as to act as an electrostatic wattmeter. With a pressure of 1000 volts on the fixed and 50 volts on the movable part, it made 300 revolutions per minute. At

*Cf. Du Moncel's 'Telephone,' American edition, p. 115, containing an account of experiments due to Bréguet and to Bell.

References to hysteresis, to inconstancy of the batteries, &c. seem inadmissible.

‡ Communicated by the Author.

low speeds, however, the friction threw the readings out a good deal. It then occurred to the writer to utilize the suggestion made by Messrs. Swinburne and Kelly for electrostatic voltmeters, i. e. to immerse the instrument in oil. This was done with good results. It was reasoned, then, that since pure water insulates as well as indiarubber, and has, according to the experimenters referred to above, a specific inductive capacity of over 70, or 35 times that of oil, all difficulties would be removed by its use. Water distilled in vacuo to remove foreign gases was then tried, but gave no better results than oil. The commutator was then taken off and the movable cylinder suspended by a bifilar suspension, with the result that in the case of oil and the alcohols the specific inductive capacity came out very nearly equal to the square of the refractive index, thus showing plainly that all the high determinations hitherto given were erroneous.

The reason of these errors was not far to seek. It lay in the fact, pointed out by Maxwell and others, that electrolysis gives a capacity effect. A number of determinations made by students in my laboratory show that, for 133 periods per second, and a current-density of 01 ampere per square centim., each square centim. of electrode-surface has an apparent capacity of 400 microfarads when the electrolyte is caustic soda and the plates nickel. It is for this reason that Kohlrausch's method almost always gives erroneous results, as what is measured is not the resistance of the electrolyte, but its impedance.

Since Messrs. Cohn and Arons, for example, used an induction-coil to charge the plates of the electrometer immersed in water, it follows that the voltage on the waterimmersed quadrants might easily have been 100 times that on the other quadrant, and that the method is inapplicable. All capacity measurements made by discharge are also incorrect; and the only correct method consists in purifying the fluid till it no longer conducts appreciably, and then measuring the attractive force between the plates when these are charged from a continuous current source of high voltage. Either a torsion or chemical balance may be used. A certain amount of leakage will always take place, but this must be provided for by using a powerful source of current.

It might also be mentioned that all the determinations of specific capacity of substances such as sulphur &c. are incorrect. For the reasons of this, those writers who have treated of the capacity of laminated dielectrics may be consulted.

LXVII. The Influence of the Relative Volumes of Liquid and Vapour on the Vapour-Pressure of a Liquid at Constant Temperature. By SYDNEY YOUNG, D.Sc., F.R.S., University College, Bristol*.

THE

HE question whether the vapour-pressure of a liquid at a given temperature depends on the relative volumes of liquid and vapour has been frequently discussed, and has been the subject of many experimental investigations. Within recent years extended researches on the relations between the temperatures, pressures, and volumes of several liquids have been carried out by Prof. Battelli, and he arrives at the conclusion that when, in a tube containing a perfectly pure liquid and its vapour, the volume is diminished and the vapour caused to condense, the vapour-pressure rises: in other words, that the smaller the relative volume of vapour the higher is the vapour-pressure.

These results are entirely opposed to those obtained by Dr. Ramsay and myself and, while referring to this question in a letter to the 'Philosophical Magazine' last February, we ventured to characterize the conclusions of Prof. Battelli as incorrect, and to suggest that the error was due to the presence of small quantities of air or other impurity in the liquid examined and to the employment of insufficiently purified substances for heating-purposes.

In a letter published in the August number of the 'Philosophical Magazine' Prof. Battelli adheres to the conclusion previously stated, and does not admit the existence of either of the sources of error suggested. "I would rather observe, he writes, "that, in order to observe such a phenomenon, an apparatus is necessary which enables us-as in my caseslowly to compress the vapour, and to maintain it for a time under constant pressure.

There can be no doubt, I think, that such an apparatus is required in order to decide the question whether the phenomenon exists, but it seems hardly necessary to point out that the conditions described are fulfilled in the apparatus employed by Dr. Ramsay and myself.

During the present year I have been engaged in an investigation of the thermal properties of isopentane--a liquid, boiling at 28°, which can, by suitable treatment, be obtained in a pure state †.

* Communicated by the Physical Society: read November 9, 1894. † Full details of the method of purification adopted will be described later when the research is completed.

Phil. Mag. S. 5. Vol. 38. No. 235. Dec. 1894. 2 Q

A great number of determinations of the vapour-pressure of the liquid were made at various temperatures, and in many cases the volumes of liquid and vapour were read. The results obtained prove conclusively that the vapour-pressures of isopentane are independent of the relative volumes of liquid and vapour; and it may be of interest to give all the observed vapour-pressures at two temperatures, with the corresponding volumes of vapour and liquid.

The same apparatus was employed as in my previous work on benzene and its derivatives, the esters, &c.: it is similar in principle though it differs somewhat in detail from that made use of by Ramsay and myself (Phil. Trans. 1887 A, p. 59).

The tube containing the isopentane was heated by the vapour from pure liquids (Trans. Chem. Soc. 1885, p. 640; 1889, p. 483) boiling under reduced pressure.

The pressures are corrected for (1) the difference in height of the columns of mercury in the tube containing the isopentane and in the air-gauge; (2) the expansion of the heated column of mercury; (3) the pressure of the column of isopentane; (4) the deviation of air (in the air-gauge) from Boyle's law, as determined by Amagat.

Several series of determinations were made with different quantities of liquid in the tube. As a rule four readings of pressure were taken in each series at each temperature.

It will be seen that although the relative volumes of vapour and liquid vary within very wide limits, there is no such corresponding variation in the vapour-pressures, the greatest difference from the mean value at 140° being slightly less than 0.1 per cent., and at 90° slightly greater-in no case outside the limit of experimental error.

It has happened occasionally in the course of this or previous investigations that a trace of air has entered the tube or has been left in the liquid, and in a very few cases (ethyl formate, propyl formate) a small quantity of permanent gas has been formed by partial decomposition of the liquid by prolonged heating at high temperatures. The presence of permanent gas (or of very volatile impurity) is clearly indicated by the increase of pressure required for the complete condensation of the vapour, and the following points have been noticed in such cases:- -(1) The pressure does rise as the volume of vapour diminishes; (2) the readings taken with diminishing volumes are higher than with increasing volumes. This may be readily explained; for when the whole of the substance is in the state of vapour the air is diffused uniformly through it, and when condensation is brought about