used. After the measurement had been made, they were to meet in a vacuous apparatus at two platinum jets and be mostly combined. The residue, when the pressure was too small for combustion to continue, was to be transferred by a selfacting leakage-proof Toepler air pump to a eudiometer, where eight hundred cubic centimetres of electrolytic gas have been safely exploded. After all but ten cubic centimetres had been exploded in this way, the residue was to be transferred from this eudiometer to a common Bunsen eudiometer, and analyzed. The apparatus is set up, ready for use, and it has been used, though no experiment has been finished, and there is no difficulty in its use, and since so much is self-acting it is not very laborious. But, as I have not had time uninterrupted enough, nor nerves steady enough, to finish, without assistance as I am, even a single experiment, the apparatus had to be broken up, unused, for lack of room.

The apparatus which is nearly ready is designed to measure in the same vessel three nearly equal volumes of hydrogen and oxygen, each of about 2.6 litres, and to combine them by combustion from platinum jets in a space containing a part of the measured volume, and afterwards to transfer the remainder to a endiometer for analysis. It is completed, except building a pier and putting together the apparatus; whether it will be used is doubtful.

The third method, which, with the apparatus already set up for other purposes, was easiest to carry out, has been successfully employed in a series of ten determinations. Unfortunately, it is that one of the three which, on account of the uncertainty in some of the constants needed in the reduction of observations, is the least helpful. The experiments by this method will now be described.

2.-METHOD EMPLOYED.

It will be recalled that Leduc made two experiments, in which he decomposed an alkaline hydroxide by electrolysis, and measured the density of the mixed gases. From this could be computed the ratio of the volumes of the components. No determination seems to have been made whether the electrolytic mixture would recombine without residue; which very much needs proof in any experiment claiming precision. If the numbers given in Comptes Rendus, August, 1892, page 313, are correctly printed, the reduction fails to take into account the physical conditions involved.

Having the apparatus with which the fifth series of determinations of the density of hydrogen was made, I resolved to make a determination by means of the process used by Leduc, but with material modifications. One consisted in avoiding the contamination of the gas with mercurial vapor, by weighing, not a

globe full of gas, but a voltameter from which the gas was delivered. Secondly, the composition of the gas, after weighing and measuring pressure, was determined by adequate eudiometric analysis. Lastly, in the computations of the composition of the mixture from the density, the deviations of the two gases from Boyle's law have been taken into account.

The observations attained a precision which it is hoped will be thought satisfactory. But the constants required in the computation may not be thought sufficiently well determined. The uncertainty, however, does not much affect the computed result. The remaining uncertainty will doubtless be removed in time, when the results here obtained will admit of re-computation, which will make them more useful. I therefore venture to give the details of the experiments.

3.-DENSITY OF ELECTROLYTIC GAS. VOLUME, TEMPERATURE, AND PRESSURE.

The apparatus for the measurement of

pressure, temperature, and volume was the same as that used in the last preceding determinations, with some trifling modifications.

72

-b

to regulate admission of gas to the eudiometer.

For the admission of the gas from the vol. FIG. 30.-Connection of globes to pump, serving tameter a stopcock had to be interposed, so as to keep the pressure in the voltameter equal to that of the atmosphere. In Fig. 29, h is the fusible metal valve by which hydrogen is to be admitted to the globes; to this is fused the stopcock g, which, during the operation of filling, was immersed in sulphuric acid.

The connection of the globes to the pump was modified so as to serve two purposes. In order to admit to an exhausted eudiometer an appropriate quantity of the gas, some stopcock was necessary; but one which could possibly leak was objectionable. The apparatus was therefore constructed as shown in Fig. 30. At a is the tube leading to the three globes in the ice. But the connection of this

tube with the Toepler pump was modified so as to contain the glass stopper b, ground so as to fit into its seat, and attached to a long glass rod which continued downwards about 90 centimetres within the glass tube c g. The end of this tube was prolonged in the rubber tube g, which was attached also to the rod, so that the space between them could be filled with mercury. The elasticity of the rubber tube permitted the stopper to be raised and lowered; the mercury prevented leakage. It is obvious that when this stopper is forced into its seat, and the vacuum below it is filled with mercury, gas cannot pass from the globes into the space below the plug, provided the pressure of the mercury upwards is kept greater than that of the gas downwards.

4.-DENSITY OF ELECTROLYTIC GAS. VOLTAMETER FOR PRODUCING AND WEIGHING.

A voltameter to produce electrolytic gas, and to make it perfectly dry, and to do it at the rate desired, without being too heavy or too bulky, took some pains in constructing. Using five voltameters combined in

one and connected electrically in series, it was deemed
best to use a current of four amperes. The plati-
num wires to conduct this current without undue
heating I found it somewhat difficult to introduce
into the glass of my apparatus without having them
afterwards crack. Several experiments were lost in

this
way; after the
passage of the current for four
hours, some minute crack around a wire would leak
at the rate of half a milligramme an hour. Gradu-
ally the faulty joints were made good, and the ap-
paratus worked without failure till a heedless
plumber knocked a brick through the wall, without
even looking to see what damage might be caused,
and crushed it into a hundred fragments.

772

72

FIG. 31.-Voltameter for producing and weighing mixed gases.

In Fig. 31 are shown five tubes 2.5 centimetres in diameter, having platinum wires as seen in section at m n. To these was connected a tube filled with phosphorus pentoxide, and also a stopcock ending in a ground joint. Into the central tube was introduced a gauge to show the pressure of the gas in the voltameter; the gauge was filled with gas to the desired point (having being filled with liquid when constructed) by electrolysis at the wires ff, while the stopcock d was open to the atmosphere.

The voltameter was filled with sodium hydroxide which had been made from clean sodium; it was freed from carbon dioxide with barium hydroxide. The stopcock d was never turned except once to open and once to close for each experiment, and was lubricated often; it is thought that it did not leak, as the pressure on the two sides of the key was always nearly the same during the times when leakage might cause error. The regulation of the current required to maintain a pressure equal to that of the atmosphere was obtained, not from this stopcock on the volta meter, but from the one mentioned before, which was immersed in sulphuric acid.

For some experiments the stopcock of the voltameter was replaced by a fusible metal valve. The opening it by fusion is simple; but when it is to be closed, the metal has to be pushed into its place in the constricted part of the glass tube. When this has to be done without opening the tube to the atmosphere, the manipulation is accomplished by introducing into the tube an iron instrument which is moved by a current in a helix enclosing the tube.

5.-DENSITY OF ELECTROLYTIC GAS. FILLING GLOBES WITH THE MIXED GASES.

When a current had been passed through the voltameter for a few days, it was closed, wiped, and weighed against a counterpoise of equal volume and nearly equal weight. When the weights required had been learned, the two volumes on the opposite pans of the balance were made very nearly equal, exactly as described at page 48.

After the weighing, the voltameter was set in a tank of water and ice. A ground joint was fitted to that of the voltameter, shown in Fig. 31 at e, and then fused to the stopcock which was attached to the globes, as shown at e, Fig. 29. The tube f was now connected to the air-pump and a three-way stopcock, and the volume included between h, Fig. 29, and d, Fig. 31, was determined, exactly as the volume of the tubes connecting the globes to the barometer had been determined (page 67). This space was finally exhausted and sealed off. The stopcock d, Fig. 31, was then opened, and g, Fig. 29, was closed. The passage through h, Fig. 29, was now opened by fusion, a current was passed through the voltameter, and the passage of the gas into the globes was regulated by the stopcock till the globes were filled. The pressure in the voltameter was kept very nearly constant. The filling the globes took about three hours and three quarters.

6.-DENSITY OF ELECTROLYTIC GAS.

MEASUREMENT OF PRESSURE.

The measurement of pressure of the gas was made exactly as in the case of hydrogen, except that the tube for admitting the gas was not now closed by fusion, and that therefore a different method of being safe against leakage was needed,

7.-DENSITY OF ELECTROLYTIC GAS. CLOSING THE GLOBES DURING THE ANALYSIS OF THE GAS.

It was necessary to close the globes in such a way as to prevent the introduction of air, not only during the measurement of pressure, but also during the withdrawal of the gas used for analysis. During the measurement of pressure, the voltameter was left connected as during the filling. Both stopcocks were shut, and the pressure was the same in the globes and the voltameter. Leakage, therefore, could not take place.

In order to prevent leakage for a day or two, and against an inequality of pressure, the stopcock g, Fig. 29, could not be trusted. Of course it was impossible to fuse the tube. I therefore bent the tube by which the gas was admitted, as shown at rq. At the bottom of the short U tube was a branch through which mercury could be admitted to fill the bend. A mark at s defined the volume whose capacity was determined in the calibration. To this point the mercury was brought during the measurement of pressure. When the voltameter was to be disconnected, the mercury was made to fill the bend. If now slight leakage should take place through the key of the stopcock, it could be detected by the change of level of the mercury in time to admit of remedy before the gas in the was contaminated.

8.-DENSITY OF ELECTROLYTIC GAS. CLOSING THE VOLTAMETER.

apparatus

The stopcock of the voltameter was closed, or the fusible metal plug was replaced, and the apparatus was wiped and again weighed, as before. The volumes on the two sides were made equal, and the weighings were by reversal.

9.-DENSITY OF ELECTROLYTIC GAS.

PREPARING THE VOLTAMETER FOR A
FOLLOWING EXPERIMENT.

The weight of the gas delivered by the voltameter was about twenty-three grammes in each experiment. Instead of adding successive increments of twentythree grammes, to make up this loss of weight, the counterpoise was made lighter by this amount. The counterpoise consisted of a number of the tubes of the same glass as the voltameter, closed by fusion, one of which was loaded with mercury. A duplicate of this loaded tube was adjusted by removing mercury to the amount indicated by the weighing just made, so that when the weighing of the voltameter after an experiment was finished, the duplicate counterpoise was ready, with its weight about the same as that of the voltameter. A current was passed through it for a few minutes and its stopcock was opened to secure that the pressure in it should be that of the atmosphere, and the indications of the gauge were verified