reached after a long series of experiments carried on step by step. Mr. Kapp. This machine, as Mr. Crompton said, converted me to the use of plenty of iron; but at the same time it converted Mr. Crompton a little in the opposite direction, and I think he would not at the present day make a machine which contains so little copper and so much iron, because it would be too costly for its output.

If it be permitted to the author of a paper to compliment a member who has taken part in the discussion, I would congratulate Professor Forbes on his very careful and painstaking way of working out the losses by leakage, and of obtaining figures so closely in accord with actual practice. The example chosen was, as Professor Forbes said himself, a very simple one, and I am afraid the calculation would in cases of greater complication be beyond the power of practical engineers; but it is important to see that the thing can be done with a fair degree of approximation. As a general rule, when making a preliminary estimate of the cost of a dynamo to give a stated output, I do not find it necessary to even employ my own simple formula for determining the leakage; I only estimate it by comparison with other dynamos which I have already tested, and leave the exact calculation to a later time, when the details of the dynamo must be designed. The reason why it is not necessary to go very closely into the question of leakage, for the mere purpose of estimating the cost lies in this, that a few per cent. more or less leakage does not materially alter the output. Professor Forbes has shown us a diagram suggesting a new method of placing the exciting coils, by which he hopes to nearly do away with leakage. No doubt this object would, to a certain extent, be attained; but the greatest merit of his suggestion lies in the fact that wire will be saved provided the system is used on dynamos having long armatures of small diameter. This design labours, however, under the great practical difficulty of either having to bend the coil up at the ends of the armature or extend it so as to enclose the commutator and brushes, which would again increase, to a certain extent, the amount of wire. The latter difficulty could be overcome, without sacrificing any of the theoretical advantages, by adopting a design for a four-pole dynamo which

Mr. Kapp. I published in the Electrician about a year and a-half ago, and which is shown in the diagram:

There are only two coils employed which are of the usual construction and can be wound in any ordinary lathe. Each half of the machine to the right or left of the vertical centre line is virtually a Forbes dynamo, but each coil, instead of being wound in a vertical plane having to cross the armature, is wound in a horizontal plane. To make this design economical, the diameter of the armature should be large as compared to its radial depth, and the vertical cores of the field magnets should be as short as possible, the only consideration being to provide sufficient space for getting a reasonable length of coils. Curiously enough, this design has actually been adopted by Mr. Kennedy of Glasgow, who makes what he calls an iron-clad dynamo; also, today, Mr. Elwell, of Wolverhampton, called on me and asked whether this was really the arrangement described by me in the Electrician. He had made a machine in Wolverhampton exactly on the same principle and found that it works exceedingly well, giving a large electrical output for very little weight and expenditure in labour. You can see that the cost for labour and

material in this four-pole machine is very much less than in the Mr. Kapp. usual form, where you have to tie the magnets together with stays of gun metal. Another advantage is that the machine is absolutely rigid. Dr. Hopkinson has described his method of measuring maximum induction, or, as I call it, the maximum number of lines which can be got through a square inch of iron. Another method occurred to me, and if I had been with Mr. Crompton I would have employed it, and given the results in the paper. It is only applicable with double horse-shoe machines, and I hope Mr. Crompton will make the experiment. In the former Crompton machine the poles were made bulging out, but in the modern machines their external surfaces are straight, so that the thickness of metal in the centre of each is very much reduced. The experiment would be made in the following way: Remove the armature and alter the coupling of the exciting coils so as to produce continuous magnetisation without external poles. The magnetic circuit will then be completely through iron, and the lines will crowd through the reduced section of the pole-pieces, where saturation will be produced with a comparatively low exciting power. Wind a single turn of wire round the part thus saturated, and connect it with a ballistic galvanometer. The induction can then be found in the usual manner by observing the throw of the galvanometer, whilst the exciting current is made or broken.

Mr. Walker said something-in fact, a great deal-about leakage, and he, like all others, has over-estimated the danger of leakage. I think that leakage is now in a fair way to be regarded by electricians as a great bugbear, and other things will be sacrificed in order to reduce leakage. I do not think that leakage is so very bad as some speakers would make it appear. If the leakage were 25 per cent., we need not waste 25 per cent.

of the exciting energy. The amount wasted depends on the cross section of the field magnet. Say that a magnet 4 inches by 10 inches would suffice if there were no leakage; if there is 25 per cent. leakage, the magnet would have to be 5 inches, and the perimeter, or length of one turn of exciting wire, would have to be increased by 2 inches, the total being about 30 inches.

Mr. Kapp. This increase is only about 7 per cent.; and the exciting energy required to produce an additional 25 per cent. of waste field will only be 7 per cent. greater than if there were absolutely no leakage.

Professor Thompson, in his introductory remarks, found fault with the scope of the paper, because it was not a paper on the characteristic of the dynamo at all, but on the magnetisation of a dead piece of iron. Well, that is true, but I did not mean to go to-night through and write a treatise on dynamo construction, showing all the geometrical problems which might be solved to deduct one curve out of another, because I thought that was a thing too elementary for an institution like the TelegraphEngineers.

It is very curious that most speakers have said something rather disparaging of the Frölich curve [the dotted line in Fig. 3], but, curiously enough, the severest condemnation came from Professor Thompson himself. He said that no man in his senses would work a dynamo at half-exciting power, and that the point (A) I have chosen in the curve was wrong. He forgets that, whether you like it or not, you are bound sometimes to work your machine at half-power. Suppose you want to make a machine to give you 130 volts at full output, and 100 on open circuit, so that it may properly regulate at 100 volts. at the end of a long circuit; or, take the case of a shunt machine driven from a shaft in the factory, which does not run at a steady speed, and requiring therefore an automatic regulator, so that it gives the maximum output with a minimum of speed, and vice versa, at the same pressure. It is quite possible that in such cases the machines would occasionally work with less than half the exciting power. Professor Thompson said that if I had chosen the point higher on the curve, Frölich's curve would have answered very well for saturation. Of course it would, because I should verify at the same point as that originally chosen. But if a curve is to be of any use it must represent the actual condition of the machine over a reasonable distance; and if Frölich's cannot fulfil this requirement it must be condemned.

Now I must say something about a subject which makes me

feel very uncomfortable, and that is the question of units. I need Mr. Kapp. hardly say that last Thursday my heart sank within me when I heard one speaker after another disapprove of them. Professor Thompson said that my system is an inextricable mixture of centimètres, grammes, minutes, and inches. Now, I protest against this; my system is no mixture, for I do not use centimètres, grammes, or seconds. I use minutes, because everybody is accustomed to counting revolutions per minute; and anybody quite unacquainted with French measure can work my formulæ, and would at the same time see what he is doing. The figures are of reasonable magnitude, and present to those who use them a definite meaning. They know what is meant by 17 lines to the square inch, but if we talk of 15,800 to the square centimètre a greater mental effort is required to grasp the meaning.

Last week, after Professor Thompson accused me of using a mixture, we had an example of how easily the C.G.S. system leads to misunderstanding, for one speaker called that a gauss which another called a milli-gauss-the proportion between the two being nothing less than one to a million. It is significant that practical dynamo makers have not raised any objection to my system of units. The translation from it to C.G.S. units is really not such a very difficult matter as we are told, and with a little training one could get to manage either system equally well; but we have to talk to our workmen, and then we could not use grammes, centimètres, and seconds. We must give them figures in the usual English measure, and therefore it saves labour if we make the calculations in a system where the figures are of reasonable magnitude and directly applicable to the various purposes of the workshop.

The following paper was then read :—

SOME EXPERIMENTS ON SECONDARY CELLS.

By JAMES SWINBURNE, Member.

This paper is a very condensed account of a large number of experiments made in the spring of 1883.

Abour four years ago Mr. Norman Cookson, of the well-known