one known arm, and make up the other two by a slide-wire. Fig. 15 shows this arrangement, as commonly used by Ostwald and Kohlrausch, in conjunction with the induction-coil. A is a cell connected to the primary terminals of the induction-coil I. A key, K, allows the coil to be put out of action when not wanted. The secondary terminals of the coil are connected to ac; the telephone to b and d. Arm ab consists of the resistance-box R : ad of the electrolyte cell W, the resistance of which is to be measured. bc and cd between them consist of a wire, usually a metre long, stretched over a millimetre scale and provided with a sliding contact-maker c. If the wire be uniform, the resistance of the two segments bc cd will be proportional to their lengths. Hence, in accordance with the rule on p. 46, we shall have

The box R should be capable of considerable adjustment. If it is only to contain a few coils, Kohlrausch recommends that they should have the values 30, 70, 200, 700, 2000, 7000 ohms. It is better if a set of coils continuously adjustable from 1 to 10,000 ohms is available. The coils should be wound double to avoid self-induction, as is customary in resistance-boxes; but in coils of 1000 ohms or more this method of winding introduces so much electrostatic capacity as to make measurements of high electrolytic resistances uncertain. The larger coils should therefore be wound in the manner suggested by Chaperon, viz. in simple layers, but each layer in the reverse sense to the preceding.`

The slide-wire bd should be made of an unoxidisable metal, and must not be so soft that the sliding contact-maker damages it. Platinum-iridium alloy and constantan (CuNi alloy) are about the best materials. It should not be more than mm. thick, or the resistance will be inconveniently low. The wire is usually stretched straight over a scale, but is manipulate if wound on a drum.

easier to

The use of a slide-wire involves calibration from time to time, as it usually suffers a certain amount of injury from the working of the sliding contact. Measurements with it are

T. P. C.

E

more exact when made near the middle of the wire: it is therefore desirable to choose R as nearly equal to W as may be; then cd will be nearly equal to be. If a set of resistances adjustable by steps of 1 ohm be used, it is possible to confine the movement of the slider within quite small limits; eg., suppose W is 30'5 ohms; then if 30 ohms be taken out of the box R, the position of the slider may be calculated from 30 5: If bd (the entire length of the wire) is 1000 mm., this equation shows that cd must be 504 mm., bc 496; i.e. the slider need only be moved 4 mm. from the middle point. This suggests replacing the end parts of the wire by coils,

= 30 X

cd bc

A

W

FIG. 16.

which are more handy, and less exposed to change. A convenient arrangement is to take a wire 10 metres long, and coil up both ends of it, leaving only 26 cms. in the middle, stretched over a scale. If the sliding contact be moved to one

cd bc

end of this, the ratio = 105; at the other end

so that the resistance-box used must be capable of adjustment to within 5 per cent. of the resistance to be measured; the box should be of the dial-switch pattern, for facility of adjustment. The arrangement is shown in Fig. 16, where S is the slide-wire. The manipulation is then as follows: put the slider at the middle point of the wire, and keep it in contact with the wire; rotate the switches of the box (R) till the position is

found that gives the least sound in the telephone (T); then move the slider to and fro till the exact position for silence is found. With the short slide-wire just described errors of calibration are so small that they can usually be neglected in practice.

In Figs. 15, 16 the position of the induction-coil may be interchanged with that of the telephone. The arrangement shown in Fig. 15 is on the whole the better, as then the induction coil is not in action if the sliding contact is not pressed down; whilst if the telephone is connected to the sliding contact erratic noises are apt to be produced when the contact-maker is pressed or moved along.

If an electrodynamometer be used instead of the telephone the sliding contact may be dispensed with altogether. The procedure is then identical

b

The arrangements for using an electro-dynamometer with alternating current are shown in Fig. 17. One coil of the dynamometer G is put in series with the induction coil I (between a and c), while the other coil is connected to b and d.

The electrolyte to be measured is always contained in a glass vessel and provided with a pair of electrodes of platinum or palladium. The size and character of the electrodes depends essentially on the conductivity of the liquid to be measured. Smooth platinum electrodes only give completely satisfactory measurements when the conductance of the liquid vessel does not exceed o'0004 mho per square centimetre of electrode surface, but are available up to o'002 mho. They can be made about thirty times as effective by "platinising." This is done by immersing in a solution containing 3 per cent. commercial platinum chloride with per cent. lead acetate,

and passing a weak current for about ten minutes in each direction between the pair of electrodes. Three or four volts are required, and the current should be so regulated as to cause a moderate evolution of gas. The electrodes become coated with a firmly adherent layer of platinum black, which greatly increases their effective surface and power of absorbing gases. Electrodes prepared in this way can be used with solutions of o'oi mho per square centimetre perfectly, and o'05 with fair success. Even better than this is an electrolytic deposit of palladium black, which, according to Ostwald, gives perfect results up to o'04 mho per square centimetre, and moderately good up to o 2; but palladium is attacked by strongly oxidising liquids.

The shape of the electrolyte cell also depends on the conductivity of the liquid to be measured: if this is high, the electrodes may be separated by a long narrow tube; if low,

they should be put close side by side. In any case it is important that they should be solidly made and fixed, so that they shall not be distorted, otherwise the resistance of the cell will vary from one experiment to another. A very convenient shape is shown in

Fig. 18. The electrodes are of thick sheet platinum welded

1 There are cases in which platinum black cannot be used on account of its catalytic action on the solution.

or gold-soldered on to thick platinum wires sealed through the ends of glass tubes. The tubes should be about 1 mm internal diameter, and contain a few drops of mercury so as to make good contact with wires inserted into them. The lower tube passes through a hole in the upper plate. The tubes are firmly cemented into an ebonite plate which forms the lid of the vessel and has a hole to admit a thermometer. Other forms are shown in Fig. 19 (for bad conductors), Fig. 20 (for good conductors). It is also convenient to have immersion electrodes (Fig. 21) which can be inserted into a bottle containing the solution, and

pipette electrodes (Fig. 22) that can be surrounded by a liquid or vapour jacket for measurements at a high temperature. In all the forms so far described the electrodes are fixed so that when filled with a given liquid the conductance of the vessel is the same. It is necessary to determine the constant of the vessel, which is defined as

conductivity of electrolyte

conductance of electrolytic cell

E.g. if a cell contain two plates of 4 sq. cm. placed 16 cm. apart and be filled with an electrolyte of conductivity o'02, its conductance (according to the rule on p. 44) will be 0'02 X=0'05 mho. The constant of this vessel is 2.5. But as conductivity vessels are never perfectly regular in shape, the conductivity is determined in practice by a measurement