Method of conducting the experiment.

Take as much distilled water as may be required and boil it thoroughly, keeping it boiling for some time, say 15 or 20 minutes. It may then be cooled to the temperature of the room by surrounding the containing vessel with cold water. The glass cylinder in which the piece of glass is to be weighed is then filled with the water to the desired height. Some of the water should remain over, and may be used if required to supply the place of the water lost through the steam arising from the glass cylinder during the heating process.

Supposing the copper vessel supplied with the proper amount of water and the stirrer, thermometer, &c. in their proper places the experiment may then begin.

For some time before each observation the stirrer is to be employed, care being taken not to make the water in the copper vessel splash over into the glass cylinder. When the thermometer indicates the temperature at which an observation is to be made the supply of gas may be regulated so as to keep the temperature fairly constant. The piece of glass is then examined for bubbles; these being removed and the half lid at once replaced an accurate weighing is made. This determines the excess of the weight of the piece of glass over that of the water displaced. Since the variation in this excess is very small it is only a question of finding how many milligrammes or centigrammes, as the case may be, must be added to an already known weight. Thus the observation may be made rapidly, which is of importance, especially at high temperatures, when there is more risk of bubbles attaching themselves to the glass. It also diminishes any error that might arise from a variation in the temperature.

The expansion of glass may for our present purpose be determined with sufficient accuracy by the following method. A rod of glass of about 50 centimetres long is placed inside a glasstube of somewhat greater length, and of 4 or 5 cm. bore. There are two scratches on the glass rod, perpendicular to its length near the ends. The rod rests on two pieces of cork, so that the scratches are on its upper surface. The ends of the glass-tube

are closed by corks, through each of which is passed a small piece of glass-tubing and also a thermometer. One of these pieces of glass-tubing is connected with a boiler, from which steam can be passed into the tube, the other piece communicates with a condenser. By means of the steam the glass rod may be heated to a determinate temperature. The expansion of the rod between the initial temperature and the final is found by means of two reading microscopes, one for observing the position of each of the two scratches on the rod. By means of these we can find the increase of the distance between the two scratches, and by means of a centimetre scale determine the initial distance. The linear expansion for one degree of temperature may then be calculated, since the whole rise in temperature is known; and the cubical expansion may be taken as three times the linear.

For a full account of the method see Glazebrook and Shaw's Practical Physics, § 36.

The following is a statement of some of the results obtained:

Weight of piece of glass in air 18-648 grammes.
Coefficient of cubical expansion of glass taken as '0000237.

The lowest temperature of observation being 10o, the density there has been in the above calculation assumed to be 99974.

No. 7. DETERMINATION OF THE POINT OF MAXIMUM

DENSITY OF WATER. JOULE'S METHOD.

WATER and liquids the chief constituent of which is water differ from all other liquids in having a point of maximum density, that is, a temperature such that both cooling and heating therefrom produce change of volume in the same direction, namely dilatation. The point of maximum density of water having been chosen as the standard for comparison of densities, the exact determination of the temperature at which this point is reached becomes of great practical importance. Many different experiments have been made for this purpose, and the results obtained by different methods agree fairly well amongst themselves. The different methods employed can be divided into two main classes.

1. An accurate knowledge of the coefficient of expansion of some solid is necessary.

a. The indications of a water thermometer are compared with those of a mercury thermometer (Delie and Dalton's method).

b. A solid is weighed in water and the temperature determined at which the loss in weight is greatest (Hallström's method).

c. The variation in weight of a quantity of water filling the same vessel at different temperatures is determined (Blagden and Gilpin's method).

2. The change in density is determined by the convection currents produced.