December 29, 1910] 



NATURE 



289 



ratio to the extent of riverless country. These statistical 



data are summed up in a fifth table. The Sahara region 



is most intensively characterised by lack of river-systems ; 



no rivers rise there, and those which flow through lose 



rather than gain during the transit. Eighty-eight per 



■nt. of this region is without off-flow. This huge desert 



rea Herr Banse regards as " the surest bulwark of Islam 



1 Africa." 



RECEXT PROGRESS IN 

 LIGHTING.^ 



ELECTRIC 



Incandescent Lamps. 



"T^HE most remarkable development within recent times 



is the production of an incandescent lamp with an 



ficienc}- approximating to i candle-power per watt. The 



st known of these lamps are of two kinds, one made 



ith tantalum as the material for the filament, and the 



Iier with tungsten. There are a great number of lamps 



ider a great variety of different names using tungsten, 



id the difference between the lamps is largely due to the 



■fferences in the processes of manufacture adopted. The 



ost recent development in the construction of tungsten 



mps is in the use of wire-drawn filaments (L. Gaster, 



intor Lecture, 1909). 



In order to test a statement that has been frequently 

 made as to the bad effect of switching off and on, a series 

 of tests is being carried out in the Electrical Engineering 

 Laboratories at Liverpool in which the lamps are switched 

 off for ten seconds and then switched on again, the process 

 being repeated at intervals of one minute. .Although these 

 tests are not complete, the results, so far as they go, have 

 been interesting. In all, twenty lamps, which have been 

 supplied by several makers, are being subjected to this 

 test. The lamps were divided into two groups, and each 

 one was adjusted to have an initial efficiency approxi- 

 mating to 14 watts per candle-power ; this adjustment 

 was effected by introducing resistance into the circuit of 

 each individual lamp. One set of ten lamps was con- 

 nected in circuit with an automatic switch driven by a 

 small fan motor at such a speed that a cam switched 

 on the light for fifty seconds and switched it off for 

 ten seconds. The reason for this choice of time was 

 that it appeared that ten seconds was enough to allow 

 the filament to become practically cold, and thus to give 

 the maximum contraction and expansion of the thread. 

 So far as the experiments have gone, the effect of switch- 

 ing does not appear to be serious ; the lamps that have 

 been burning continuously have given out to the same 

 extent as those that have been subjected to the continuous 

 switching off and on. 



The effective life of these lamps is found by them to be 

 roughly proportional to the 3-65th power of the initial 

 watts per candle-power, a law which corresponds with 

 that found for carbon filament lamps. 



We may sum up the position, so far as metallic fila- 

 ment lamps are concerned, by saying that at present there 

 is no difficulty in obtaining a 230-volt metal filament 

 lamp of about 25 candle-power which will give one hori- 

 zontal candle-power for 12 watts, and will burn near this 

 efficiency for more than 1000 hours, probably for a much 

 longer period under ordinary conditions. 



Before leaving the subject of incandescent lamps, it 



may be of interest to make some remarks on the character 



of the light that is emitted from them. The spectrum 



'^'itained corresponds with the spectrum given by an in- 



ndescent body, i.e. it is a simple band spectrum. So 



"as my own observations have gone, there is no evidence 



of selective emission, and the increase in efficiencv of 



metal filament lamps may be said to be entirely due to 



the higher temperature at which the lamp filament runs. 



The wave-length of maximum emission intensity corresponds 



fairly closely with Wien's law for the radiation emitted 



a black body. (It may be of interest to note here how 



arly the temperature of maximum emission intensitv for 



:■ yellow line, D, or centre of the visible spectrum about 



^ From a paper read before the lUutnlnatin? Eng-neirins Society on 

 I>ecemb€r 9 by Prof. E. W. Marchant. & s » j 



6200° C, corresponds with the estimated temperature of 

 the sun.) 



The nearer we can approach the temperature of the 

 sun with artificial sources of light, the more nearly will 

 an illumination be obtained which corresponds in all 

 regards with sunlight. 



Arc Lamps. 



The most notable advance in arc lighting within recent 

 3'ears is the flame arc, but these lamps have been used so 

 extensively for a number of years that the flame arc itself 

 is far from being a recent development. 



It is my intention in this paper to lay stress on only two 

 points : — (i) the improvements made in the distribution of 

 light so as to give more uniform illumination over a 

 large area ; (2) the actual efficiency or flux of light emitted 

 per watt consumed in a modern lamp. 



The attempt to obtain more uniform illumination on 

 the surface has been made in two ways, first by using 

 vertical carbons instead of V carbons, as in the earlier 

 form of flame lamps. As has been pointed out by many 

 people, the candle-power curve required to give uniform 

 horizontal illumination is much more closely approached 

 by the polar curve of light distribution given by a vertical 

 carbon lamp than by any other fornx 



The second method which has been adopted with the 

 view of improving the light radiation from V carbon lamps 

 is by the use of special globes. The most notable example 

 of this is the use of the dioptric globe. 



(2) Actual efficiency of modern arc lamps as measured 

 bv the influx of light per watt consumed (see subjoined 

 Table). 



Type cf lamp Volts Current M.^S.C.P. watt 



Enclosed flame lamp 



(clear globe) 58 ... 8-45 ... 22CO ... 4-5 



(opal globe) ... ... 57-8 ... 8-5 .„ 1430 ... 30 



Open flame arc (slightly 



obscured globe) ... 40 ... 70 ... 1040 ... 372 

 Singly enclosed arc 



ordinary carbon ... 86 ... 85 ... 1150 ... i'6 

 Midget singly enclosed 



arc lamp 77 ... 3*2 ... 245 ... I'D 



Vapour Lamps. 



The production of light from an incandescent vapour 

 is a method of lighting which has long been familiar, 

 though the only practical examples of it are the mercury 

 vapour lamp and the Moore tube. The two forms of 

 mercury vapour lamp which are being manufactured at 

 present are the quartzlite lamp and the silica lamp. 



The main characteristic of this lamp is that it produces 

 a large amount of ultra-violet light, to which quartz is 

 transparent, and which is screened off from the exterior 

 by a heavy lead glass clover. If the lamp is left burning 

 without this cover for a few minutes the smell of ozone 

 produced is very strong. It is a matter for discussion 

 whether mercury vapour lamps containing these very 

 strong lines only in the spectrum will not ultimately prove 

 injurious to the sight of those who are obliged to work 

 in it. A priori, it would seem to be bound to produce a 

 fatigue of those parts of the retina which respond to the 

 impulses given by the particular rays which the lamp 

 emits. 



A lamp of this type has recently undergone test in my 

 laboratory, with the result that the efficiency worked out 

 at 1-73 candle-power (mean hemispherical) per watt. The 

 lamp consumed 688 watts at 230 volts, and gave a mean 

 hemispherical candle-power of 1190. 



The mixing of the mercury vapour light with that of 

 the light from tungsten lamps has been tried at Liverpool 

 with quite satisfactory results, the ratio between the 

 amount of light required to produce complete mixing being 

 very easily found by the aid of the globe photometer and 

 two pieces of milk glass, one piece illuminated by a beam 

 of daylight from the outside and the other by the light 

 diffused on the surface of the globe by the two sources 

 of light inside the globe. This method is, of course, not 

 so exact as the colorimeter of Ives, but gives quite satis- 

 factory results. 



NO. 2148, VOL. 85] 



