520 



NATURE 



[October 19, 191 1 



the actual improvements in detail which enable the 

 electrical engineer to substantiate his claim that great 

 progress has been made. 



Taking the case of electric lighting first. The 

 efficiency of the incandescent lamp depends entirely 

 upon the temperature of the filament, if we neglect 

 for our present purpose the hypothesis of selective 

 emissivity of certain substances, which, although 

 advanced from time to time, has never been com- 

 pletely established, and in any case would only slightly 

 affect the accuracy of the above general statement. 

 Tungsten, being more refractory than carbon, can 

 be kept continuously at a higher temperature without 

 volatilising, and a tungsten filament would give three 

 times the light of a carbon filament lor the same 

 consumption of power and the same life. But as the 

 specific resistance of tungsten is lower than that of 

 carbon it was necessary to produce a finer filament 

 than the carbon one, and to find a means of supporting 

 a greater length of this fine filament in the lamp. 

 These difficulties were at first only partly overcome. 

 The filament of a 25-c.p. 220-volt carbon lamp has a 

 diameter of about o'i6 mm., and a length of about 

 350 mm., while the filament of a tungsten lamp of the 

 same candle-power and voltage is about o'02 mm. in 

 diameter and 830 mm. long. 



To produce pure tungsten in the form of a rod with 

 a diameter in the order of a few hundredths of a mm. 

 it was necessary to mix it with other materials, to 

 press or "squirt" a filament of the required 

 diameter, and to eliminate the foreign substances by 

 various processes, at the same time " forming " and 

 strengthening the pure tungsten filament which was 

 left. This resultant filament was exceedingly brittle, 

 especially when cold, and it could only be made in 

 short lengths, so that several separate loops had to be 

 independently supported and joined in series in each 

 lamp ; obviously a delicate operation in view of the 

 fragility of the filament and the difficulty of welding 

 tungsten to other metals. 



At first, therefore, the filament could only be made 

 of small enough diameter and high enough resistance 

 for comparatively high candle-powers, and for voltages 

 not greater than no volts; and, in fact, when the 

 first tungsten lamps were put on the market as 

 articles of commerce only one diameter of filament was 

 used, and the length varied according to the voltage, 

 so that the buyer had to be satisfied with one par- 

 ticular candle-power of lamp suited to his voltage. 

 Month by month and year by year, however, the 

 manufacturing processes were improved, the filaments 

 were made stronger, finer filaments were pressed and 

 formed, and the methods of mounting them were 

 bettered. The range of sizes and voltages was gradu- 

 ally increased, and during the first half of iqoo, 16-c.p. 

 100-volt lamps and 32-c.p. 200-volt lamps with 

 tungsten filaments were placed on the market, fol- 

 lowed in the autumn of that year by a 25-c.p. 200-volt 

 32-watt lamp. 



Willi this, the limiting size of a squirted filament 

 seemed to have been reached, but in the meantime 

 painstaking research work had been carried on in the 

 laboratories of the chief lamp manufacturers, with the 

 object of finding a process for actually drawing 

 tungsten in the form of a wire. This has now been 

 finally accomplished on a commercial scale, and during 

 the last two months three of the leading incandescent 

 lamp manufacturers in this country have already prac- 

 tically "scrapped" their comparatively new filament 

 presses used for the squirting process, and are making 

 their lamps with filaments of tungsten wires. The 

 'ire lilnment lias an enormously greater strength, 

 and as in addition it can be used in the lamps in one 

 continuous length without joints, the robustness of 



NO. 2190, VOL. 87] 



the lamp has been so improved that it is as great as 

 that of one with an ordinary carbon filament. Finally, 

 it has been already found possible to produce finer 

 filaments of drawn wire than by the squirting pi 1 1 

 and, as has already been announced in our columns 

 (September 28, p. 420), at the opening of the Elec- 

 trical Exhibition at Olympia on September 23, a 16-c.p. 

 220-volt 20-watt Osram lamp was shown, as an article 

 of commerce, with a filament only about 0015 mm. 

 in diameter. 



During the various stages of the introduction of 

 the metal filament lamp, the electrical engineer set 

 himself the task of making the utmost use of the 

 economy which it occasioned in the consumption of 

 electrical energy. For equal light, one-third of the 

 energy was being consumed as with the carbon fila- 

 ment lamp, and the consequent cheapening of electric 

 light opened a much wider field, and brought it within 

 reach of the smaller householder. There were, how- 

 ever, two obstacles to overcome : first, the fact that 

 the metal filament lamps for the usual town-lighting, 

 voltages were only to be obtained in comparativt 1\ 

 large sizes, so that although the consumer obtained 

 three times the light at the same cost, he could not 

 light a small room at one-third of the cost; and, 

 secondly, the expense of wiring a house for electric 

 light acted as a deterrent to the small householder. 



The first of these difficulties was overcome, in alter- 

 nating-current systems, by the interposition of a smalt 

 transformer or "auto-transformer" between the con- 

 sumer's main switch and his lamps. A voltage of 

 25 or 50 could thus be obtained, for which pressures 

 metal filament lamps down to the smallest candle- 

 powers have been obtainable during the past four 

 years. A wastage of electrical energy is, however, 

 liable to occur when no lamps are on, unless the 

 consumer then turns off the supply at his main switch. 

 for otherwise magnetising current is passing through 

 the primarv of the transformer all day long, whethi 1 

 lamps are switched on or not. To prevent this wasti 

 automatic switches have sometimes been emploj 1 1 

 disconnect the transformer when all lamps are extin- 

 guished, and to reconnect it as soon as a lamp is 

 turned on. 



One of the latest types of these (made by Messrs. 

 Muirhead and Co.) is shown in Figs. 1 and 2. In 

 this case an auto-transformer is employed ; that is to 

 say, a coil connected across the mains with a tapping 

 at a point corresponding to the pressure required for 

 the secondary circuit. TThe construction of the switch 

 is clear from the general view and diagram. When 

 no lights are on, the main circuit of the auto-trans- 

 former is open at the mercury cup 1. As soon as one 

 lamp is switched on, a circuit is established through 

 the coil marked starting coil, and the "auxiliary" 

 coil and "main" coil in series, and the core seen in 

 the two illustrations is sucked into the coil, making 

 contacts 1 and 2. This connects the auto-transformer 

 to the mains at contact 1, and also short-circuits the 

 starting coil at contact 2. When a few more 

 lamps are in circuit, bringing the current up to 2'2 

 amperes, the core has been sucked in further, and 

 contact 3 is made, short-circuiting the "auxiliary" 

 coi! and leaving the main or retaining' coil only in 

 circuit. The obiect of successively short-circuiting tin- 

 first two coils is to diminish the voltage drop. On thi 

 lamps all being switched off, the arm falls by 

 gravity, and the auto-transformer is out of circuit. 



One of the largest items in the wiring of houses is 

 the cost of fixing the steel tubing or wooden casing 

 in which the wires are placed. In those houses whi re 

 surface work is not objected to. use is now frequently 

 made of metril-encn^ed wires, which are simply fixed 

 on the walls by metal <=ndclles. The metal casing is 



