companying the isostatic adjustments may be a partial cause. Material is moved in this process, to zones of decidedly different temperatures, to hotter or to colder zones, and it seems to be logical to conclude that some chemical or physical changes may take place which would affect the density of the material transported. In the process of isostatic adjustment all the material of a column above the zone of flow is raised or lowered. The ordinary thermal expansion of the material of a column, as it changes its temperature, is capable of accounting for only a small part of the changes in the length of a column.

It should be clearly borne in mind that the theory of isostasy does not explain any vertical movements other than those necessary to maintain equilibrium. Some other theory is needed to explain elevation and subsidence and the writer feels that the theory of local expansion and contraction is in general in harmony with the geodetic data. He believes, however, that there are local vertical movements of small amounts which may not be due to these causes.

It is hoped that increased activity will take place in collecting geodetic data and in extending the investigations in isostasy. It is particularly desirable that we have the investigations include the ocean areas as soon as geodetic data may be available within them.

1 See page 90, "Investigations of Gravity and Isostasy," Spec. Pub. No. 40, U. S. Coast and Geodetic Survey, 1917.

2 "Strength of the Earth's Crust," by Jos. Barrell, J. Geol., January-February, 1914 (48); and "Discoidal Structure of the Lithosphere" by Bailey Willis, Bull. Geol. Soc. Amer., June 30, 1920.

3 "Investigations of Gravity and Isostasy," Spec. Pub. 40, U. S. Coast and Geodetic Survey, pp. 70-82.

4 "Investigations of Isostasy in Himalayan and Neighboring Regions," Professional Paper No. 17, by S. G. Burrard, former Surveyor-General of India, 1918.

EXPERIMENTS ON THE ELECTRICAL CONDUCTION OF A

HYDROGEN ALLOY

BY DONALD P. SMITH

CHEMICAL LABORATORIES, PRINCETON UNIVERSITY

Communicated by O. Veblen. Read before the Academy, November 17, 1920 In studies which were described some time since1 it was found that hydrogen, when discharged electrolytically upon any one of certain metals which occlude it, produces a temporary diminution in the electrical resistance of the metal. With metals, such as palladium, which occlude large amounts, and suffer, as has long been known, a lasting increase of resistance, the diminution of resistance, or supplementary conductance, is superimposed upon the opposite and more enduring effect; while in the

case of platinum, which has only a small occlusive capacity, and does not display the permanent increase, the supplementary conductance may be observed alone.

Regarding the nature of this peculiar electrical conduction, three explanations seem thus far to offer:

(1) The conduction is by a transient form of the occluded hydrogen, probably monatomic, and consists in a transport of electrical charges between points of different potential within the metal.

(2) The conduction is by unstable hydrides, which are produced only under conditions of electrolysis which are equivalent to a high pressure of hydrogen, and which rapidly decompose when these conditions cease to obtain. (Newbery.)2

(3) The conduction, while not the result of chemical combination, is due to a temporary effect which the occluded hydrogen exerts upon the distribution or mobility of the electrons within the metal.

Upon the second or third of these suppositions, there appears to be no reason to expect the character of the supplementary conduction to differ from that of metallic conduction in general; but the first-named hypothesis leads to a different expectation.

This first explanation, which has been suggested by the observation that the supplementary conduction and the volume of the alloy are directly related, may be stated somewhat more in detail as follows:

It is supposed that within a hydrogen alloy there exist (a) molecular hydrogen, (b) atomic hydrogen, and (c) hydrogen combined with the metal to form solid solution; and that these three forms are in mobile equilibrium in some such way as is indicated by the scheme

2H + 2xM2MxHH2+ 2xM.

(1)

Since the fact is well established that the formation of solid solutions between two metals brings about a great increase in the electrical resistance beyond the value which would result from simple mixture, it is natural to attribute the permanent increase of resistance in hydrogen alloys such as palladium-hydrogen to this combination; and if the supplementary conduction is attributed to the atomic hydrogen, it is evident that, upon the conception just outlined, this conduction must vary with the cathodic current density during electrolysis, as has been found to be the case, and must persist after the interruption of electrolysis until the equilibria of equation (1) have become established.*

If the foregoing explanation be essentially correct, the conduction in question is that of a gas confined within the cavities of the metal. It is accordingly not to be expected that it will conform to Ohm's law, but rather that the resistance observed will depend upon the intensity of the current employed in its measurement, or in other words, upon the applied e. m. f.

The experiments about to be described were, therefore, undertaken in order to test this prediction in the case of palladium-hydrogen since the earlier observations, having been made by null methods, gave only the resistance for "zero current."

Experiments.-The cathode investigated consisted of a wire of commercially pure palladium, 0.05 mm. in diameter, and approximately 110 mm. in length. The cell and thermostat were those which have been described elsewhere, as were also the manner of attaching and supporting the cathode wire, save that arrangements for observing changes of length

were dispensed with. The electrolyte was 2-n sulphuric acid, which was caused to circulate during the experiment, as heretofore. The temperature was maintained throughout at 25.00 0.02° C.

The electrical connections are represented diagrammatically in figure 1, where the electrolytic circuit is shown at the left (subscripts "1"); and the resistance-measuring circuit at the right (subscripts "2").

An experiment was conducted as follows: The switch S2 being in position 1, so that the battery B2 was short-circuited through R, and the resistance-measuring circuit was open, the electrolytic circuit was closed by means of the plug switch S1, and was regulated with the variable resistance R1 until the millammeter M1 showed the current corresponding to the desired cathodic current density. Electrolysis was then continued,

with occasional regulation of the current, until the cathode wire had attained saturation, the time required being calculable from the data previously obtained.

After the electrolytic current had been interrupted by withdrawing the plug S1 the resistance-measuring circuit was closed by throwing the switch S2 into position 2. The current, derived from the battery B2, which consisted of four large storage cells in parallel, was now regulated by means of the resistance box R2 until the millammeter M2 gave the reading sought. This was from 2 to 4 milliamperes, these values being selected upon the consideration that for the sake of constancy the current should be as small as the precision of the potentiometer would permit.

I-42

2

e

I= 4.1

I=2.1

2' 4' 6' 8' 10' 12' 14' 16' 18' 20' 22' 24' 26' 28'

Time in Minutes

FIG. 2

Dependence of resistance on current

By means of the Pohl commutator P the potentiometer was now connected alternately to the potential leads, p and p', coming from the ends of the cathode wire, and to those of the ten-ohm comparison resistance, readings of the fall of potential across the two resistances being taken several times in rapid succession, while the time was in each case noted, in order that the value for the ten-ohm might be calculated by interpolation for the exact moment at which that for the cathode wire was observed. This made it possible to eliminate the error which would otherwise have resulted from the gradual drift in the resistance of the cathode, and hence in the measuring current.

A complete experiment consisted of three such series of observations as that which has just been described, obtained at intervals as short as possible. The first and last were taken with a measuring current of ap

proximately 4 milliamperes, and the intermediate one with a current of half this intensity.

The potentiometer employed was a low-resistance "Type K" instrument by Leeds and Northrup, with which was used a suspended coil galvanometer by the same makers. The latter was adjusted by means of a shunt until its vibrations were aperiodic, and had under these conditions a sensibility more than adequate. Care was taken before the beginning of an experiment to bring the potentiometer battery to such constancy of e. m. f. that errors from this source were negligible, and this constancy was checked at least once in the course of each series of observations by comparisons with a set of standard cells.

The ten-ohm comparison resistance used was not a precision coil, so that the resistances found are only relative. Their relative accuracy may be estimated at 0.02% in the least favorable instances.

Results. The results of one experiment are recorded in table I, and displayed graphically in figure 2. Before this run the wire had been several times saturated with hydrogen, and immediately before the interruption of electrolysis the current had been maintained over night at a value of 2.3 milliamperes, corresponding to a cathode current density of about 1.4 amperes per square decimeter.

In the table below, column 1 shows the measuring current in milliamperes; column 2 the time elapsed from the first reading in seconds; and column 3 the interval between successive observations. The unenclosed numbers in the column headed E10 give the observed fall of potential across the ten-ohm resistance, and the numbers in parentheses are interpolated from the preceding for the time at which E, was read. E, is the observed fall of potential across the cathode wire; and R, is the resistance of the latter calculated from the relation R = 10 E1/(E10).

In figure 2 the values of t from table I are plotted as abscissae, and those of R, as ordinates. If the resistance were independent of the measuring current, the values for all three series of the table should lie upon an unbroken curve which is very nearly rectilinear; but it will be seen that the points for the intermediate series, taken with a measuring current only half as great as that used in the two end series, fall far out of line. A diminution of current of 50% has produced a diminution of resistance of some 30%.

This result was confirmed by that of an entirely similar experiment in which the changes of current and of resistance were, respectively, fifty and forty-two per cent; and finally by that of an experiment in which the electrolyte was withdrawn from the cell, before the resistance of the wire was observed, and in which the corresponding decrements were fifty and twenty-six per cent. It should be pointed out that only qualitative agreement between the several experiments was to be expected, since the magnitude of the effect is dependent both upon the quantity and