The other method used to obtain the solid contents of some globes, consisted in determining the specific gravity of the glass of the globe from the part cut off in fusing on its stopcock, from which the solid contents of the globe minus the key of its stopcock could be computed; to which was added the solid contents of this key.

9. COMPRESSION OF GLOBES WHEN EXHAUSTED.

The volume of a hollow globe varies with the variations of pressure to which it is subjected. The amount of variation due to a given difference of pressure depends on the radius and on the mean thickness of the walls of the globe, and on the variations of the thickness on different parts of the surface. For the globes used in these experiments, the difference of volume due to the exhaustion of the globe was from one six-thousandth to one twenty-five-hundredth of the capacity.

A convenient

This change of volume must be determined with accuracy. method was employed. A copper cylinder, 77, Fig. 4, had a cover m, which could be soldered to the cylinder and easily removed again. In this the globe a was placed; sometimes it was nearly filled with water, sometimes it was held down by a sinker. The globe was connected by a thick rubber tube to the glass tube c leading to a syphon gauge and to an air-pump, and the cylinder was nearly filled with water.

The cover was then soldered in place, the cylinder was quite filled with water, and the rubber stopper n was inserted. Tof were connected the tubes g and h. On the tube was a mark: g was graduated in tenths of cubic centimetres. To the upper end of i was attached some simple device for varying the pressure of the air in it. These tubes were filled with water to the level of the mark h. The tube h is intended to act as a manometer showing the pressure on the water in the cylinder.

Let us suppose, for a moment, the temperature of the water in the cylinder to be constant; if we alter the volume of the globe, but keep the pressure on the water constant by watching the manometer, and introducing or removing water, we shall have a measure of the change of volume of the globe.

The manipulation is accordingly as follows: We bring the level of the water in the manometer tube h exactly to the mark, and note the reading on the graduated tube g. Then we exhaust the globe, and keep the pressure constant in the cylinder by introducing water into it from the graduated tube, which we read again. The difference of the two readings is the change of volume of the globe, provided the temperature of the water is constant. If now air is admitted, and the alternate. readings are repeated at nearly equal intervals, we can eliminate the effect of slow changes of temperature and determine the change of volume of the globe with

accuracy.

As to the convenience of the method it may be said that, when the globe was not filled with water, the whole operation of putting the globe in place, soldering the cover, making a sufficient number of determinations, and taking out the globe, could be finished in two hours. When the globe was filled with water, the intervals between readings were shorter, but the time required to fill the globe and to empty it again was considerable. Most of the determinations were made in this latter way; three determinations were the work of nearly a whole day. The method, as will be seen from the following example, gave concordant results; it also gave concordant results when applied to the same globe on different days.

DETERMINATION OF THE COMPRESSION OF GLOBE NO. 1.

The compression of each globe due to a difference of internal pressures meas

ured by seventy-six centimetres of mercury is given in the following table:

COMPRESSION OF GLOBES DUE TO 76 CENTIMETRES' DIFFERENCE OF PRESSURE.

10.-CORRECTION FOR COMPRESSION OF GLOBES WHEN EXHAUSTED.

The effect of the change of volume of a globe which is due to a change of internal pressure may be eliminated by suitable manipulation, or the amount of a correction to the weight of the globe may be computed from the amount of the change of volume, together with the density of the air at the time of weighing. There will doubtless be a difference of opinion as to which is the more convenient.

In the case of one of my globes, the change of volume was about eight cubic centimetres. In order, therefore, not to produce an error of one tenth of a milligramme in the value of the correction, it would be necessary to observe the pressure and temperature of the air at each weighing. But, with the balance which I procured for the present series of experiments, numerous weighings were required in order to determine a weight accurately enough, so that much labor would be involved in the observations and computations for the corrections. It was accordingly thought more convenient to make the needed correction a matter of manipulation consuming no time after the proper apparatus had been constructed. Each globe was therefore provided with a counterpoise equal in volume to the globe when exhausted. Then, for each globe, was made a pair of minute flasks whose volumes differed by the amount of the compression of the globe on exhaustion, and whose weights were exactly the same when they were weighed in a vacuum. For instance, the compression of globe No. 1, for the difference of pressure usual in my experiments, was 1.27 cubic centimetres. Two flasks were made whose volumes were 2.080 and 0.810 cubic centimetres, and whose difference of weight when weighed in air was equal to the weight of 1.27 cubic centimetres of air at the time, taking account of the true values of the weights employed.

Now, when the globe No. 1 was exhausted, it was weighed against its counterpoise, which had the same volume. When the globe was full of gas, the 0.81 cubic centimetres were placed with it on the balance, and the 2.08 cubic centimetres were likewise added to the counterpoise. The true weights of the globe and counterpoise suffered equal additions, for the true weights of the two additions were equal within a fiftieth of a milligramme. Therefore, the apparent difference

of weight between the globe and its counterpoise would be the true difference of weight as expressed by brass weights in air.

The method was found so convenient that it was developed and improved for wider use.

11. -AIR-PUMPS AND EXHAUSTION.

a

The exhaustions of globes and other apparatus required in this series of experiments were mostly effected with a mercurial air-pump of the form devised by Toepler. The tube by which it discharged itself was shaped as shown in Fig. 5. The recipient a could be itself exhausted with a piston pump when the discharge of the Toepler pump directly into the air began to be incomplete owing to the small volume to be discharged. The body of the pump had a capacity of 1.6 litres; during this series of experiments, it would exhaust a volume of twenty litres to two millionths of an atmosphere. But such a completeness of exhaustion was avoided. Since the tension of the vapor of mercury at ordinary temperatures is, if my experiments are sufficiently trustworthy, not much more than one millionth of an atmosphere, it was hoped that we could avoid the diffusion of mercury vapor into the globe which is exhausted, by opposing to the diffusion a rapid current of gas in the other direction. Now, if the tension of the residual gas in the globe is, let us say, ten millionths of an atmosphere, each stroke of the pump will draw a current of gas towards itself with considerable velocity.. Unless, therefore, some vapor can pass from the pump to the globe during the up stroke of the pump, it will not reach it at all while the pump is in action. But if the tension of the residual air is made equal to that of the mercurial vapor in the pump, air will enter the pump by diffusion mainly, and then mercurial vapor will enter the globe. It was hoped that if the vacuum produced still contained gas at a tension ten or twenty times that of mercury vapor, its entrance into the globe would not take place while the pump was working; by making the connecting tube some two metres long, it was hoped that, even after the stopping the pump to measure the vacuum, not much mercury would diffuse into the globe.

FIG. 5.-Discharge tube of Toepler pump for obtaining a better vacuum.

The exhaustion of globes, but not of other apparatus, was sometimes effected with a Geissler pump whose body has a capacity of 2.3 litres. When this is newly cleaned, it is possible to exhaust a volume of one litre to a ten-millionth of an atmosphere.