special engine. C is the communication positive wire. D is auxiliary lamp main lead, and E is auxiliary battery main lead. It is evident that if D and E are joined, the auxiliary lamps will light up, provided the local switches, J J, are closed. F is the main battery of 52 volts. G, communication bell. H, auxiliary battery three-position switch. J J, auxiliary lamps local switches. I I, main lamps local switches. K, communication push. L L, auxiliary batteries (it is possible to do without the auxiliary battery in van). M, main lighting switch. N, bell battery. 00, main lamps with auxiliary filaments (the auxiliary lamps may be

FIG. 4.

FIG. 3.

separate from the main lamp). Before describing the action of the system, I may say that safety fuses are fitted in every carriage, but these have not been shown in the figure in order not to complicate it.

As will be seen from Fig. 1, when the train is coupled up the whole of the main lamps are lit from the battery, F, which is switched on to the circuit by a switch, M (in combination with a switch, R, and resistance, P, to equalise the discharge). There is

FIG. 5.

The communication system, with its push, K, in each carriage or compartment, its main lead, C, bell, G, and small battery, N, is self-evident and needs no explanation. It may be worked by the main battery if desired.

The action of the automatic coupler strips, DE, is as follows: Should the train be parted anywhere, the strips D E, in the couplers that are severed, come down to the piece J, Fig. 4, and the lead D, Fig. 1, is thus connected to lead E, and the current

Coupler.

thus flows from the auxiliary batteries to the auxiliary lamps. When the train is once more coupled up, the connection is broken at the piece J, Fig. 4, and the main lighting leads then come into action, and the auxiliary lamps go out.

The smaller diagram, Fig. 2, represents the working of the same systems by the use of two through leads instead of four. This we consider equally efficient and more economical.

The main lighting current from main + B passes round coil C, and, when switch D is closed, lights the large lamps or filaments, and in passing round C, it actuates armature, E, raising it and making contact, F, and holding it there during the whole time main lamps are lit. The auxiliary battery, G, is thus placed in communication with main by means of the switch H, which, when closed, charges the auxiliary battery through the resistance, J.

When the main lighting circuit is broken in any way, the current in C is also broken, and the armature, F, falls and makes contact, K, and the auxiliary switch, L, being closed the auxiliary lamps light up. When the main lamps are once more lit the auxiliary lamps are automatically put out by the action of E.

In the double-filament lamps, Fig. 3, the large + is connected to piece A, the small to piece B, while the brass casing, C, forms the common negative to both filaments, and this casing makes contact with the lampholder which is attached to the general negative main.

Fig. 4 shows cross-section of the coupler for the four-wire system; Fig. 5 the longitudinal section; and Fig. 6 the sectional plan. The five strips correspond, and are attached to the five leads, A, B, C, D, E, respectively, shown on main diagram, Fig. 1. The strips are all pivoted on the pin F, and are all insulated from it and from each other by the insulation shown in black. The pin G carries upon it the insulating piece, H, upon which the three strips A, B, C rest when the couplers are detached, and it also carries the brass piece, J, which is itself insulated from the pin G. The object of the piece, J, is to automatically connect the two strips D and E when the couplers are separated, and thus light up the auxiliary lamps, as shown on diagram of general arrangement. The pin, K, entering the hole, L, of the corresponding coupler, ensures the accurate meeting of the various strips and prevents short-circuiting. As the couplers are pressed together, the curved ends of the strips engage with each other and hold the couplers together. The springs, M, keep the strips in connection with their corresponding strips, or with the piece J when advisable. If the two-wire system is used, only two strips are necessary.

In conclusion, we would remark that we can only claim as absolutely new in principle (as apart from the details of couplers and double-filament lamps and other mechanical matters) the combination of a main lighting circuit with high E. M.F. and efficient economy, together with an auxiliary lighting circuit and lamps with small batteries and economy in weight, cost, and main

tenance.

DISCUSSION.

Mr. Killingworth Hedges thought that the subject of train lighting was one of great importance. It was to be regretted that Mr. Timmis's paper did not come on earlier in the day, when more attention could have been given to it. With the main conditions laid down by Mr. Timmis, particularly the one which stated that every carriage ought to carry its own store of electricity, he cordially agreed; in fact, he had advocated the same plan three years ago at the British Association meeting at Manchester. He must say, however, that it had been rather difficult to convince the railway authorities that they must, in order to break up their trains, put a battery in each carriage. It appeared to him that the main circuit and the auxiliary circuit when connected must be a very complicated system, and, indeed, one that would be difficult to arrange unless rolling-stock were joined together. He could not see the good of the auxiliary circuit, because if they lighted with batteries connected up in a carriage, why not put in batteries of small capacity, perhaps, three or four hours' storage, and work them from the dynamo, which in ordinary times when the train is in motion lights the whole installation. The auxiliary limits of lower candle-power would offer grave objections, because people would sooner do without the electric light than have it suddenly lowered. In his (the speaker's) paper at Manchester, he mentioned batteries having zinc and lead electrodes. Since then little had been done with them, but the other day, when seeing Mr. Webb, of the North-Western, he was told that they were acting extremely well, experimentally. If such cells were used they might obviate the difficulty complained of by Mr. Timmis in ordinary secondary batteries. At any rate, he thought it would be quite practicable to put a set of batteries either of zinc and lead or the ordinary batteries in a carriage, and get sufficient capacity to attempt all the lighting during ordinary stoppages at stations. One great objection to heavy batteries was the liability to be broken. When travelling on the Midland the other day, where a good system of electric light was used, he found the electric light was not at work. Upon asking the reason, he found hat the train at one of the stations ran into the station buffers with sufficient force to knock the batteries out of the case and break it, and the orders were, that until it was put together again, not to use the dynamo. On the Midland they had tried successfully a very important experiment, and that was one advocated at the British Association by Mr. Preece; it was known as Mr. Barber-Starkey's plan, and consisted of filling up cells with plaster of Paris and making them quite solid, and he was told that it acted admirably. In conclusion, the speaker expressed confidence in the fact that electricity must very soon be the illuminant used for trains, and pointed out the

disastrous consequences of a collision between trains lighted by gas. Mr. Bennett referred to a system of lighting used on the North British Railway.

If

Mr. Smith, in replying for Mr. Timmis, said, in reference to Mr. Hedges's objection to heavy batteries, it was that very reason which impelled Mr. Timmis to put small batteries in each carriage. Mr. Hedges did not quite see that was that system-no dynamo was used. If there was a large battery, then there was no small one. It amounted to this, that if they had a charging station where they lighted the railway station, one used the same plant to charge the big battery in the guard's van. the big battery was employed to light the main lights, then only little batteries were required to light the auxiliary lights. These could be left in the carriages from one month to another, with only occasional supervision, because they were charged from the big battery. As to the diminution of light it was not so serious as Mr. Hedges supposed. The reduction was from 16 c.p. to 10 c. p., which was quite good enough to read by. This, however, only occurred in stations, and it was supposed that the diminution of light here would not be very noticeable-at least, it would form no serious inconvenience.

ON THE ELECTRIFICATION OF STEEL NEEDLEPOINTS IN AIR.*

BY A. P. CHATTOCK.

(Concluded from page 203.)

Although the silent discharge from a point does not appear to have the power of permanently clearing away surface-resistance, sparks do so readily enough-though under suitable circumstances they will also form it again. In making the measurements on needle B (Table I.), the discharge at the last recorded (+) reading took the form of a spark. /P was here 1:59. On repeating the reading it had risen to 195. A third discharge brought it back to 162, and after that it continued to oscillate between these two values for some 20 or 30 times with hardly a break in the regularity. Obviously one spark formed a resistance and the next blew it away again. The character of the sparks was quite different, too. The one which formed the resistance was thin and sharply defined, and spread out for a considerable distance over the plate like a splash. The other was much brighter, straight, and without any signs of a defined edge.

While, however, the above is evidence of the existence of surfaceresistance at used points, the fact that the values of ƒ at beginning and end of discharge are practically identical for a clean point for both positive and negative electricity, may be regarded as showing that such resistance is either very small or non-existent on points which have not been used before.

Atomic Charge

§ 6. Assuming, then, provisionally that cohesion in Grotthuss chains is the sole cause of the resistance offered to discharge at a clean point, and that gas atoms are consequently concerned in carrying the electricity when it does so,t there arises the interesting question as to the amount of charge carried by each atom; and one is tempted to see whether the above measurements throw any light on it. Without in any way pretending to settle the point, the following considerations seem to me to render it probable that the electro-chemical equivalent of gas atoms is of the same order of magnitude as that of the same atoms in electrolytes.

The essence of a Grotthuss chain is that it shall consist of molecules capable of being split into two parts, one of which is charged after the split with positive electricity and the other with negative. These charges may exist separately in the molecules to start with, or they may be induced by an electrostatic field-but in either case it is the field which subsequently arranges them in chains; and the breaking down of the chains occurs when the mechanical pull of the field on the charged parts is sufficient to overcome their mutual affinity. This affinity may be due to one of two causes or their combinations. Either it is (a) the electrical without any previously existing charges, or it is (c) both. One of attraction of two initial charges, or it is (b) a cohesive attraction these three it must be, if by cohesive attraction is understood the sum of all non-electrical forces between the two parts of the molecule.

Consider the first (a) of these cases. If the molecules hold together by internal charges, they may be represented thus in their free state:

Put in another way, the strength of the field which will do this is rather greater than that which would induce on each of the two end atoms of the chain half their assumed initial charges, supposing them to be connected by a fine conducting wire. It depends thus on the length of the chain being greater as the chain is shorter.

In the case (b), a constant cohesive affinity between the two parts of the molecule, it is necessary to suppose that electricity can pass from one to the other under the influence of induction when they are put in a field of force, in order that after separation they may be oppositely charged. The following figure shows

瓜

a chain of such molecules on the point of breaking down, the break occurring when the cohesive forces (~~) are just overcome by the electrical forces ( ) set up by the field. Here, again, the longer the chain the weaker the field required to break it; and as the cohesive forces are constant, and the electrical forces are dependent, for given molecular arrangement, only on the charges induced in the molecules, it follows that these charges will also be practically constant, no matter what the length of the chain may be.

In both these cases, therefore, for all lengths of chain, the atomic charges at the breaking point are constant; in case (b) being rather less than such as would be induced by the field on the end atoms, supposing the latter to be connected by a conductor; and in case (a) less than twice that amount (less because the chain is not a continuous conductor). Their values might thus be calculated in terms of ƒ if the geometrical conditions were known. This, however, is not the case under the conditions of actual experiment; but by arranging that the chains shall consist of single molecules only, it is possible to get an idea of the magnitude of the charge in question. Reduce in imagination the discharging point to molecular dimensions; and find, by what is rather violent extrapolation, the corresponding value of at discharge from the constant in Table II. The Grotthuss chains will have been reduced to single molecules by the reduction of the point, and by supposing them to be spheres the + and charges induced on their opposite sides may be calculated. These charges will then represent in the case of (a) half, and in the case of (b) the whole of the atomic charge.

This number is greater than the atomic charge (or than half the atomic charge) because it is calculated from measurements on chains of many molecules, which, as was pointed out above, are not conductors, but only lines of high S.I.C. Now the most probable value of the ionic charge of oxygen is 10-11, which, considering the extent of the extrapolation and the fact that the above number is too great, is in sufficiently strking agreement with it. It is at any rate satisfactory that the experimental number is the larger of the two.

It is interesting to see whether a similar result is to be got from measurements of sparks between plates. Taking Dr. Liebig's numbers above referred to, I find, following Prof. J. J. Thomson, that the formula f= +150 expresses the results for the smallest sparkเ

a

engths in air fairly well, considering that the numbers for that part of the curve are rather irregular. The values of a lie between 1.9 and 14 for the five smallest spark-lengths (l=0.0066 to 0·0245 centimetres. Extrapolating for a distance between the plates of × 10-8, the field between them comes to be 4 × 108. This would induce a still higher charge (8 × 10-9) on a spherical molecule; but it must be remembered that these data are much further removed from molecular dimensions than mine.

On the other hand, there is a case accessible of what may perhaps be called discharge between plates, which takes place actually within molecular dimensions: I mean the passage of electricity at the cathode of a voltameter. Here, if a step of potential of, say, one volt be assumed, and if ƒ stand for the corresponding field between the liquid and the metal of the cathode, 1

1

fx 10-8 = orf=2/3 10 E.S. units.

2

300'

This is capable of inducing on a spherical molecule charges of 15 x 10-1 É.S., a number which cannot be distinguished from the ionic charge. Moreover, no reduction is necessary here as the data of calculation are from measurements made direct on single molecules,

Taking, then, these three calculations together, and having regard to the fact that the nearer the conditions of experiment approach molecular arrangement the closer are the results to the value of the ionic charge, I cannot help thinking that they furnish strong grounds for supposing that electrified atoms in gases are associated with the same quantity of electricity as in electrolysis. As regards the third possibility of molecular cohesion je mentioned above, that it is due to a combination of (a) and (b), it is impossible, without knowing the relative values of the two forces at work, to get any idea of the atomic charge from the value of f All that can be said is that it will be less than the value calculated for atoms held together by electrical attraction only. Even in this case, therefore, there is nothing to negative the presence of ionic charges in gaseous conduction.

Effect of Pressure.

§ 7. If the conclusions arrived at above be correct, one may picture a metal point on the verge of discharging as a smooth curved conducting surface studded all over with Grotthuss chains standing up on it like bristles. The density of charge upon will thus be far from uniform. It will reach a maximum at the root of each chain, the quantity collected there being constant for a given gas, independent of the length of the chain, and equal perhaps to the ionic charge of the gas atoms. In between the chains the density will be much less.

Now the field, f, measured by the attraction of a plate on a needle-point is the average number of lines of force per square centimetre of the point-surface, and takes no account of the manner in which they are distributed over it. This is because the point is so small that the lines have room to become uniformly spread out before they reach the plate. Hence it follows that, for different dispositions of the chains, the measured values of ƒ may be very different, and yet the number of lines of force running through each chain be the constant number corresponding to ionic charge on its atoms; f, in fact, for a given amount of induction per chain, depends both on the length and on the closeness of the chains, Great length, or great closeness, or both, means that the greater part of the lines proceeding from the point have been absorbed by the chains, hardly any passing in between them. In this case fis practically proportional to the number of chains per square centimetre, and is independent of their lengths. On the other hand, very short chains, or very few to the square centimetre, or both, means that ƒ is sensibly independent of their closeness, but is now dependent on their length, being inversely proportional thereto so long as it is not very great compared with the radius of curvature of the point. In between these two extremes ƒ depends both on length and on closeness of the chains, varying in an inverse manner with the former, and in a direct manner with the latter.

well

Now, both length and closeness of the chains increase with increase of gas pressure; hence, there must be some pressure, A, above which is sensibly proportional to the closeness of the chains only, and some lower pressure, B, below which it is inversely proportional to the chain-lengths only. In the neighbourhood of A increase of pressure will affect ƒ chiefly by the resulting alteration of the closeness, and will therefore increase f. Near B it will have the opposite effect, as increase in the length of the chains means a decrease of f. Hence between A and B there must be some pres sure for which is a minimum. This point is, of course, known to exist, though I was unable to obtain a sufficiently good vacuum with my apparatus to reach it. If, however, ƒ be expressed in terms of some power (n) of the pressure, n will be 0 at the minimum point, and positive and increasing as the pressure rises from there. This increase is shown well in Table III. (calculated from Table II. for positive discharge only-the negative is too uncertain). Here n is calculated from the values of f, corre sponding to 76 and 40 centimetres of mercury, and n for 40 and 20 centimetres; ng is in every case less than n1.

The needle marked X was in reality a steel ball used for bicycle bearings. This was put opposite a tin disc of 6 centimetres diameter at a distance of 3.6 centimetres. n was calculated from measurements of the difference of potential between ball and plate, as ƒ could not be measured. The values of n fit in well with the rest. (n, was calculated in this case from ƒ at 40 and 60 centimetres mercury.)

Now pressure alters the length and closeness of the chains at the same time; but, for a given field strength at a point-surface, alteration in curvature gives rise to alteration of chain-length only, the length being less as the point gets sharper. Hence at a sharp point is more dependent on the length of the chains than at a blunt one; it is, in fact, nearer its minimum value, and n is consequently less. This, too, is shown very clearly in

Table III.

Röntgen, Weidemann's Electricitat, vol. iv., § 582,

A pretty illustration of the influence of point curvature on n was accidentally met with in the case of needle C. In getting it into the apparatus for repetition of the curves obtained with it, its point came against the metal box, and was flattened slightly to a width of about 36 × 10-3 centimetres. Tho values of 1 and n obtained from it after this were respectively 0:42 and 0:30, instead of 0.31 and 0.2 for the finer point.

Grotthuss chains, coupled with constant atomic charge, are thus well able to explain most of the phenomena described in this paper. There still remains one which is perhaps the most important of all-the difference in the behaviour of positive and negative discharge. This I hope to discuss in connection with experiments now in progress, but it may perhaps be well to place on record the results so far obtained.

These are given in Table IV., where k is the ratio at a pressure of 20 centimetres of ƒ for positive discharge to ƒ for the negative. The ratio shows a distinct tendency to decrease as r increases. It is almost useless to give values at higher pressures on account of the uncertainty as to the condition of the point, and the great effect this may have on the negative discharge (Curves III.); but the ratio is distinctly less at higher pressures, for points which seemed quite clean; the decrease of k at 70 centimetres mercury varying from 3 to 8 per cent. as compared with k at 20 centimetres. For dirty points the decrease may reach 50 per cent.

FIG. 1.-T, brass tube forming body of instrument; length 10 centimetres, diameter 6 centimetres.

H, tube for suspension wire, G. F, ebonite plug.
M, mica disc suspended from G, carrying light metal
clip B, holding needle N.

S, tin screen soldered into T.

K, metal cover. E, ebonite cover.

A A, ground air-tight joints, greased.

CD, spherical metal cup to be electrified; C D = 3.35 centimetre, radius of curvature = 1.76 centimetre.

P, short brass tube on S; length = 0.5 centimetre, diameter 12 centimetre.

=

Point of N at centre of curvature of CD. Length of needle projecting beyond P = 2 centimetres.

absolute measurement of much smaller quantities by Prof. Minchin. The attracted disc is suspended, and the force of attraction measured by tilting the whole instrument until the disc falls back by its weight to a fixed point. The disc is a sheet of mica, covered on one side with tinfoil and metallically connected with the case of the instrument, which is a tin cylindrical box 30 centimetres long by 30 centimetres diameter. The disc hangs just outside a 3in. hole in a tin screen at one end of the case, and opposite the hole inside the case is an adjustable disc of tin which forms the attracting plate. The sensitiveness of the instrument may be varied very greatly by altering the distance between this plate and the suspended disc. This arrangement does not of course permit of absolute measurement, but it was thought that calibra

tion in terms of spark-lengths was sufficiently accurate for the work in hand.

The same principle of tilting was used in measuring the attractions on the needle-points; the needle in this case taking the place of the suspended disc. The zero position was determined by a hair in the eye-piece of a microscope through which the point of the needle was observed; the latter being at the centre of curvature of CD, Fig. 1, when it coincided with the hair. The needle was illuminated by small windows in T (not shown). L was connected to the Wimshurst; T to earth; and G-i.e., N-through a high-resistance galvanometer to earth, the indications of the galvanometer being therefore due only to current discharged from Ň. A much larger instrument, constructed on the same lines, was used for needle A and other measurements. The body of this instrument is a tin cylinder, 43 centimetres long (horizontally) and 23 centimetres in diameter. The needle is suspended in a small metal box at the centre, and the electrified plate is supported like CD, Fig. 1, from one end.

The tilting is effected in each of the above instruments by fixing them to a brass base, provided with pivots at one end and a vertical micrometer-screw with large divided head at the other. The method works admirably.

In conclusion, I wish to express my thanks to my friend Mr. F. B. Fawcett, a former student of University College, Bristol, and to my assistant Mr. J. Quick, for much careful help in carrying out the measurements described in this paper. To Prof. Lodge my thanks are so numerous that I cannot express them. His kindness, both by word and by deed, has been unceasing. Indeed but for him this paper would probably never have been written. Results.

1. Point-discharge depends only on the strength of field close to the point between pressures of 76 and 10 centimetres.

2. Field here measured absolutely by its pull on the point. 3. Seat of resistance to discharge shown to be in gas only, for a clean point (Grotthuss chain).

4. Atomic charge, probably of same order of magnitude as ionic charge.

5. Effects of pressure and of point-surface curvature in accord. ance with Grotthuss chains.

PORTSMOUTH.

Mr. Shoolbred suggests in his report to the Portsmouth Town Council the low-tension system for private lighting with storage batteries, and the three-wire system for distribution, using armour-protected cables, laid underground, and embedded without any other covering whatever. The capacity of the electrical installation which he recommends to be laid down at first would be of a maximum output of 5,000 lights each of 16 c.p., burning at one time. This figure of 5,000 lights would mean a total of about 8,000 lights connected with the system of distribution. The buildings at the generating station should, however, be constructed large enough to contain without addition generating plant for about double the electrical output above mentioned, so as to allow for reasonable extension of the first supply. The boilers preferred are of the Lancashire type, steel and double-flued, working at 180lb. pressure per square inch. He recommends that steam engines of the inverted vertical type, running at moderate speeds, driving the dynamo direct, and placed in the same bed-plate therewith, be adopted. The sizes of steam engines which he recommends are those capable of developing 150 i.h.p. and 75 i.h.p. respectively. Should sea water for condensing purposes be not readily available (and it is not at present in central Portsmouth) the engines should be one of the compound non-condensing type.

The capital expenditure for this part of the scheme is put at £45,000, and the revenue at £10,390, with a profit calculated to reach £1,451. We do not agree with Mr. Shoolbred's estimates-£10,390, with a maximum even of 8,000 lights wired, is rather more than we should expect, but as a rule the maximum number of lights estimated for are not wired for some years. For the public lighting Mr. Shoolbred suggests are lights, the figures for which are: Cost of 25 lights on the Clarence Esplanade, 25 on the South Parade, with engines, dynamos, boilers, cables, lamps, and posts, etc., £5,500; 40 arc lights at refuges and other prominent points in streets named in the electric lighting order, with engines, dynamos, etc., £5,000; contingencies, 10 per cent., £1,050; making a total of £11,550. The committee, having considered the general question of the electric lighting of Portsmouth, and inspected the systems in vogue elsewhere, stated they were of opinion that the scheme furnished by Mr. Shoolbred was the best that could be adopted. They had come to the conclusion, after exhaustive enquiry, that the low-pressure system was the most suitable, by reason of its absolute safety and economy in working. A further advantage of this system was that it could be utilised for motive power. In Mr. Shoolbred's report the total cost of the installations for public and private lighting was estimated at £56,550; but to this should be added a sum for acquiring a site and meeting other expenses. The committee recommended that Mr. Shoolbred's report be adopted, and that application be made to the Local Government Board for sanction to borrow £60,000 for the purposes of the electric lighting of the borough.