12.--MEASUREMENT OF THE DEGREE OF EXHAUSTION EFFECTED.

The perfection of the vacuum produced in each experiment was measured with the gauge devised by McLeod. The capacity of the bulb of that one which was set up with the Toepler pump was about 300 cubic centimetres. On the tube were three marks, such that one millimetre of difference of level in the two tubes showed a vacuum of one millionth or five millionths of an atmosphere, or one measured by the two-hundreth of a millimetre of mercury in a syphon gauge.

The gauge used with the Geissler pump had a bulb whose capacity was 436 cubic centimetres. On the tube were six marks; the upper, third, and fifth indicate vacua of the ten-millionth, the millionth, or the hundred-thousandth of an atmosphere, by a difference of level of one millimetre; the second, fourth, and lower marks indicate in the same way, vacua expressed by one ten-thousandth, one thousandth, or one hundredth of a millimetre of mercury in a syphon gauge.

The tubes of the gauge belonging to the Toepler pump were of such size, and kept so clean, that difference of capillary action did not much interfere with the accuracy of its indications. This was often tested by exhausting the gauge to a tenmillionth of an atmosphere, when any irregularity in the level of the mercury in the part of the tube where the readings were commonly made could be detected. The error found has never been important in comparison with the quantities to be measured.

13.-MANIPULATION OF GLOBE WHILE FILLING WITH OXYGEN.

In the present series of experiments, an elaborate apparatus was used to measure the amount of impurity in the oxygen. The method required that while a globe was filling with oxygen or hydrogen to be weighed, a second globe should also be filled with a quantity of the gas absolutely identical in quality for subsequent analysis, and that this sample should be absolutely safe from leakage. Now, when measurements were made in this way of the amount of impurity in the oxygen used for weighing, it was found to be negligible. But some experiments which proved this in another way are easier to describe, and are conclusive; so that perhaps it is not worth while to detail more at length the apparatus really used at the time the present series of determinations were made. I therefore describe only the manipulation required in filling the globe.

The globe a, Fig. 6, which had been previously exhausted, was placed in the case mm, in which ƒƒ is a non-conducting envelope, and dd is a mass of water contained in a vessel consisting of two concentric cylinders of iron. At ee is a stopper for the mouth of the inner iron cylinder; it consists of a cylinder filled with water. In this stopper are three tubulatures for the two thermometers and for the tube

which is connected to the globe. The ground joint of the globe was placed in n, and made tight with melted wax or paraffin. The globe, tube, and stopper were put in place, and the tube o was connected by fusion to the tube J h leading to the air-pump, the mano-barometer, and the generator of oxygen. The stop.

FIG. 6.-Calorimeter case for measuring temperature of oxygen.

cock c being open, the whole connected apparatus was exhausted, and the degree of exhaustion measured. The stopcock of the globe, b, could now be opened, and the degree of its exhaustion verified, if required, as, for instance, if the globe had just been weighed while exhausted, and it was desired to know if leakage had occurred. The globe was then closed during the further preparation of the apparatus for producing pure oxygen.

14. PREPARATION OF OXYGEN.

In all the experiments of this series, oxygen was obtained from potassium chlorate. A tube of infusible glass, a a, Fig. 7, was filled with the required quantity of the salt, the end ad was drawn out and fused to a ground joint made of the

same glass. This was then cemented with wax to the joint d, made of soft glass, which was in turn fused to the rest of the apparatus.

It may be said that not a single connection of rubber was used, in any of my experiments, in contact with the gas on which I was working. It is of course true that with rubber joints the leakage may be so small as to be estimated and allowed for. But it was desirable to have no leakage at all, or, rather, to have no more than that which is inevitable between glass and mercury; and the manipulation which dispenses with rubber joints is by no means troublesome.

d

a

h

វ

FIG. 7.-Apparatus for preparing oxygen; e f g shows in plan the tubes seen in elevation to the right of c.

M

At c is a gauge and overflow tube dipping in mercury. The oxygen coming from d, in some experiments, passed through a tube containing finely divided silver heated to redness, in order to absorb chlorine. When this was used it was connected to the tube a by fusion, and the joint d was at the end of the tube containing silver. The gas next passed through a tube 1 metre long, and 2.5 centimetres in diameter, which was filled with glass beads moistened with a strong solution of potassium hydroxide; then through a similar tube with sulphuric acid. At this point was placed a stopcock, whose office was to keep the pressure of the oxygen filling the part of the apparatus so far described at about the pressure of the atmosphere, so as to allow sufficient time for the action of the reagents. The tubes lay horizontally, so that the reagents might be advantageously distributed. After the stopcock, the gas passed through a tube of the same dimensions, filled with phosphorus pentoxide between layers of glass wool. From this, it was led to the apparatus which was to be filled with it.

In preparing oxygen, the apparatus was exhausted, and the degree of exhaustion measured; it was commonly the thirty-thousandth or fifty-thousandth of an atmosphere. Then the stopcock h was shut, and the chlorate was heated till it was thought safe to assume that any organic matter present had been oxidized. The chlorate was then cooled, the stopcock h was opened, and the apparatus was again

exhausted. Then the stopcock h was again closed, the tube containing the chlorate was heated till the pressure of the oxygen in it was nearly that of the atmosphere. The stopcock h was then opened so as to permit a slow so as to permit a slow passage of oxygen into the globe, the rate being determined according to the indication of the gauge c.

15.--PURITY OF THE OXYGEN PREPARED FROM POTASSIUM CHLORATE.

Oxygen prepared from chlorate might contain chlorine; it might contain nitrogen which had not been removed from the apparatus, or which entered it during the experiment; or carbon dioxide produced by the combustion of organic matter; or finely divided chlorate or chloride; or the vapor of water.

The vapor of water can be so of water can be so completely removed that the remainder is negligible. If the current is properly related to the dimensions of the drying tube, sulphuric acid does not leave more than one milligramme in four hundred litres of the * gas. The drying power of phosphorus pentoxide is yet greater, so that the amount of water vapor left unabsorbed is perhaps not more than one hundredth part as much as in the case of sulphuric acid. If, as seems to be proved, a current of three litres an hour is dried completely by a drying tube whose capacity is twentyfive cubic centimetres, it may be safely assumed that a current at five times this rate will be dried by a tube five times as large.

Cooke has observed that the desiccating power of phosphoric anhydride may be lessened by the formation of a sort of glaze over the surface. The difficulty will be overcome if the gas to be dried does not simply pass over the pentoxide, but passes through a long column of it. If the pentoxide be so filled into the drying tube that a channel can form, the difficulty would no doubt exist. But if the pentoxide is alternated with plugs or diaphragms of glass wool, between which it fills the tube completely, the difficulty is removed. The initial deliquescence which forms the glaze is long limited to the compartment first reached by the gas to be dried; then at each diaphragm the current of gas spreads itself throughout the whole area of the tube, and if the tube be properly filled, it also spreads itself throughout the whole area of the part filled with pentoxide except in the first compartment. But if no plugs of glass wool are interposed, the channel which has formed in this compartment gradually extends throughout the tube, and the gas is no longer properly exposed to the action of the anhydride.

Nitrogen was sought for by eudiometric analysis, as well as in other ways. The maximum amount found was one twelve-thousandth, the minimum was about five millionths; the mean was one thirty-thousandth. Since the densities of American Journal of Science, 34, 199.

*American Journal of Science, 30, 140.

oxygen and nitrogen differ but by an eighth part, even the maximum found was negligible in its effect on the density of oxygen.

In some experiments, about half of the oxygen contained in a globe was com bined with hydrogen, and chlorine sought in the water thus produced. Sometimes it could not be detected; it never amounted to the twentieth of a milligramme. It was in the same experiments that the presence of nitrogen was investigated; after twenty litres of oxygen had been combined with hydrogen, the residue was analyzed, and the amount of nitrogen measured, with the result given above.

If the absorption of chlorine was so complete, it may be hoped that the absorption of carbon dioxide was also complete. The experiments to detect chlorine also served to detect carbon dioxide, and the amount found was negligible.

Whether any vapor of mercury interfered with the trustworthiness of the determinations cannot be affirmed; the evidence is too indirect. But the evidence is perhaps sufficient that the error, if any, is negligible. If the density of oxygen is in error for this cause, the error in the determination of the density of hydrogen by the same method, in the same conditions, would be some sixteen times as much. But I have succeeded in determining the density of hydrogen, not only in exactly way used for oxygen, but also in a way in which the error from the presence vapor of mercury is entirely avoided; and this error is found to be not above some such quantity as one thousandth of the density of hydrogen; therefore in the case of oxygen it is not likely to be more than the ten or twenty-thousandth part.

the

of

16. MEASUREMENTS OF TEMPERATURE AND PRESSURE.

gas

The temperature of the gas in the globe a Fig. 6, while resting in the case mm, was observed by means of two thermometers made of Jena normal glass and divided into fiftieths. Their bulbs were diametrically opposite, one near the top and the other near the bottom of the globe. The pressure of the in a was measured with the mano-barometer shown in Fig. 8. The tubes a and b together constitute a syphon barometer; c is the tube which was used to measure the pressure of the oxygen in this set of experiments; a fourth being used for the experiments on hydrogen which were made at the same time. The tubes all stood in a box containing water, two of the faces of which were made of selected plate glass. In front of each tube and in contact with it was a glass scale graduated into millimetres. The lines of the scale are about three thousandths of a millimetre in width, so that the uncertainty in setting a division of the eye-piece micrometer for coincidence is negligible. The three scales were of the same kind of glass, graduated at the same time, and agreed well with each other. They were adjusted so that their three zeros were on the same level, as shown by