]\gt ; the Efficiency of Selenium as Detector of Light .
By E. E. FOURNIER D'ALBE , D.Sc .
( Birm .
) .
( Communicated by Prof. J. H. Pointing , F.R.S. Received bIay 7 June 19 , 1913 .
) The following investigation concerns the limits up to which minute quantities of light and minute variations of its intensity ma .
be discovered by means of seleninm .
There are two actions of on selenium which may be used\mdash ; ( l ) a change in the conductivity , first described by .
Smith in 1873 ; and ( 2 ) the generation of an E.M.F. in a yoltaic cell in which selenitun is an electrode ( Sabine , 1878 ) .
The of an apparatus for the detection of is the ratio of the amount of measurable effect it indicates to the of eceived by the apparatus .
Both the above actions static , the amount of light available is best stated in terms of the steady flux of , the unit the lumen , or the amount of falling upon 1 sq . .
distant 1 .
from a source of 1 candle-power .
Hitherto it has been customary to specify the sensitiveness\ldquo ; of a selenium preparation by stating the ratio in which its resistance is reduced by a iven illumination , reckoned in metre-candles lux This practice , however , ives no indication , the usefnl effect to be obtained from a restricted of , such as that out conically from a star , as this ) ends upon the extent of the sensitive surface to be coyered .
Since the obtainable with a given selenium preparation is directly proportional , as a rule , to its sensitive surface , the latter has to be taken into account in determining its efficiency , and this is done by specifying the in lumen rather than lux .
Voltaic elenium " " cells properly so called , are electrolytic cells in which illuminated selenium forms one of the electrodes and yenerates an E.M.F. Their efficiency should be defined as the E.M.F. in volts enerated per lume1l when the illumination is 1 lux .
The illumination must be specified , since the .
is proportional to the square root of the incident Thus , in the arrangement adopted for the selenium of stars , volt was obtained by .
a sensitive surface of sq .
cm .
means of a candle at a distance of .
The * G. M. Minchin , ' Bo Soc. Proc 189 vol. 58 , p. 142 ; and 1896 , vol. ) , 231 .
VOL. LXXXIX.\mdash ; A. Dr. E. E. Fournier d'Albe .
On the was lux .
With unit illumination , the E.M.F. would have been volt .
Then the flux of light intercepted by sq .
cm .
would have been microlumen , and the resulting voltaic efficiency is 1270 volts per lumen .
Th.e Galvanometric Effieiency .
Selenium " " cells\ldquo ; in which the action due to light is a in the resistance selenium are more properly termed selenium " " bridges\ldquo ; ( Minchin ) .
The problem here presented for solution is not so much to obtain the possible ratio of ' ' dark\ldquo ; to " " light\ldquo ; resistance , but the absolute increase of conductivity , i.e. the largest additional current on .
to light .
In this respect , the most " " sensitive\ldquo ; bridges may be comparatively useless unless they also have a reasonably low resistance , and this it has always been found difficult to secure .
The " " galvanometric efficiency\ldquo ; may be defined as the increase of conductivity produced in a selenium bridge under unit flux of light .
Conyenient nunlbers , applicable to most made , are obtained if the conductivity is reckoned in reciprocals of megohms micromhos and the flux of in lumen .
Thus , a selenium bridge 10 cm .
square , placed at 1 .
from a 100 candle-power source , would receive llumen of .
If , under those circumstances , its resistance changed just as it would if a megohm were connected in parallel with it , its " " galvanometric efficiency\ldquo ; uld be unity .
Observed Efficiencies .
The data hitherto published with to the results obtained by means of selenium preparations are often insufficient for their efficiencies .
M. Reinganum , * using photo-voltaic selenium cells with a sensitive surface of sq .
cm .
, obtained about volt with a Nernst lamp placed at 30 cm .
This gives a voltaic efficiency of about 6 volts per lumen at normal illumination if the square root law holds good for intense illuminations .
Low as this efficiency is , it is probably considerably above that obtained by .
Uljanin in 1888 , who observed volt in sunlight about .
As regards galvanometric efficiencies , the greatest interest , from the point of view of the present investigation , attaches to the figures given by Stebbins , recently studied the light-curve of Algol by means of a selenium bridge .
The light from the star was received by the 12-inch M. Reinganum , ' Phys. Zeitschr 1906 , vol. 7 , p. 786 ; and 1907 , vol. 8 , p. 293 . .
Uljanin , ' Wied .
Ann 1888 , vol. 36 , p. 241 .
J. Stebbins , ' Astrophys .
Journ 1910 , vol. 32 , p. 185 .
Efficiency of Selenium as Detector of Light .
objective of a refractor , and an extra-focal cm .
in diameter was formed on a Giltay selenium bridge having a " " dark\ldquo ; resistance of 3 megohms .
The illumination due to the star was thus increased 1,900 times .
Now , the illumination due to Arcturus is very nearly one microlux* .
The stellar magnitudes of Arcturus and Algol at maximum are given by the Nautical Almanac as and respectively .
Hence the unaided\ldquo ; illumin ation } ) Algol will be microlux , and the illumination of its extra-focal image 330 microlux .
This illumination , falling on a surface of area sq .
cm .
, gives a light flux of micro- lumen .
The deflection shown by the galvanometer was 8 diyisions .
The deflection represented a current of ampere .
The used was 6 .
Hence the additional conductivity was micromhos , and the galvanometric efficiency\ldquo ; was 20,200 micromhos per lumen .
That a alvanometric efficiency of 25,200 is unusually great will appear from a calculation of the efficiency of a cylindrical bridge described by Buhmer as the best out of a number of bridges tested .
The area exposed to an illumination of was sq .
cm .
, a flux of lumen .
resistance fell from 35,000 ohms to half that value , which means an additional conductivity of micromhos .
Hence the efficiency is micromhos per lumen .
Another Ruhmer , used by Korn for yraphy , had an efficiency of about 234 micromhos per lumen in " " diffused daylight\ldquo ; ( about .
Conditions In order to compare two selenium bridges under standard conditions , it is necessary to take into account a number of circumstances their indications .
These are the ( a ) Colour.\mdash ; For the ordinary sources of light , the maximum effect shown by selenium is that due to the visible red , and as the maximum shifts with the distribution of energy in the spectrum , this distribution must be specified .
In practice , it will be most convenient to adopt a standard istribution of energy in the spectrum , such as is offered by one of the } ) standards of rhtness .
I suggest , as the standard illumi]ation , that provided by the whole spectrum of Harcourt 's pentane standard , the distance from the .
M. Minchin ( C. Boys ) , 'Nature , ' , vol. 62 , p. 246 .
See also G. Muller , ' Die Photometrie .
Gestirne ' Engelmann , Leipzig ) . .
Ruhmer , ' Phys. Zeitschr 1902 , vol. 3 , p. 468 . .
Korn , ' Phys. Zeitschr 1904 , vol. 5 , p. 113 .
7 .
Dr. E. E. Fournier d'Albe .
On the source being such as to give one lux .
As an alternative , the Hefner lamp be adopted .
( b ) Temp found that the sensitiveness of a selenium bridge was considerably increased by its temperature permanently to C. found a reduction of 25 per cent. in the sensitiveness on plunging a bridge into liquid air .
I found a reduction of 10 per cent. under the same circumstances .
I propose to adopt C. as the most convenient standard temperature , freezing point being unsuitable for photo-Yoltaic selenium cells .
( c ) Voltage.\mdash ; The resistance of a selenium depends upon the , the percentage decrease of resistance being proportional to the logarithm of the applied I propose 1 volt as the standard voltage .
This has the advantage that the efficiency can be expressed in micro-amperes ( instead of micromhos ) per lumen .
The voltage does not sensibly affect the sensitiveness , unless it is so high that the Joulean heating produces a perceptible rise in the temperature of the , whereupon the sensitiveness decreases .
The wide divergence of results hitherto obtained with regard to the " " sensitiveness\ldquo ; of selenium is attributable to the differences in method of exposure and of interpretation .
The measurements are affected by the " " inertia\ldquo ; of selenium , shown in the slowness with which it acquires its final conductivity when illuminated , and the even greater slowness of recovery in the dark .
In selenium bridges with graphite electrodes , in which there is no chemical action between the electrodes and the selenium , the changes of conductivity during and after faint illumination closely approximate to what they would be if due to a progressive ionisation of the selenium under the action of light , the number of ions produced in unit of time being proportional to the flux , and the total number proportional to the total quantity of light , less whatever ions recombined .
The equation to the light-action curve is then where is proportional to the flux of incident energy and is the coefficient of recombination .
is the additional conductivity , and is proportional to the number of ions of both signs liberated .
In a steady state , or the final steady conductivity is proportional to the square root of the light intensity ( as actually found by Boss , Adams , Berndt , and Minchin ) . .
Stebbins , .
cit. Pocchettino , ' Rend .
R. Accad .
dei Lincei , ' 1902 , vol. 11 , p. 286 . .
E. Fournier d'Albe , ' Eoy .
Soc. Proc 1912 , , vol. 86 , p. 452 .
Efficiency of Selenium as a Detector of Light .
integration we get , or The -curve has an initial line portion inclined to the -axis , so that before recombination becomes considerable the change of conductivity on illumination is .to the incident , as is found to be the case for all feeble exposures .
For the recovery curve we have , by , the relation where is the additiorial conductivity at the moment of shutting off the light .
ma be most conveniently obtained from the recovery curve itself .
If the galvanometel deflection at the moment of obscuration is called zero , and are two taken at equal time-intervals after that moment , then This is the final value ained by the recovery after a very long time .
It is hereinafter called the " " total recovery and by .
It is , other conditions being equal , the net additional conductivity due to previous illumination , whatever may have been the method of exposure .
But since the rapidity of reaction varies greatly with the intensity of the tJht , it is best to adopt a method of exposure in which the light action has a definite relation to the recovery .
This is attained by alternately ) and obscuring for equal times until the curve oscillates between steady valuea Then the number of ions produced by illnmination is twice the number recombined in the same time , and if the number produced varies as the incident enelgy , the amplitude of ] ation will also vary as the incident energy .
This is shown below to be very approximately the case .
Thus the process proposed for ( the true increase of tlctivity for a given illumination is the following:\mdash ; Illuminate and darken for alternate periods of one minute each , and when the " " light\ldquo ; and " " dark\ldquo ; deflections have each become consistent , shut off the light finally and take readings after the first and second minute .
The conditions proposed for determining the normal galvanome efficiency of a selenium bridge may here be summarised:\mdash ; Dr. E. E. Fournier d'Albe .
On the Quality of : complete radiation from a standard pentane lamp .
( b ) Illumination : llux .
( c ) Voltage : 1 volt .
( d ) Temperature : C. ( e ) History : alternate and darkness for one minute each until steady state is attained .
( f ) Measurement : readings for successive minutes during recovery .
Evaluation of the ' normal alvanometric efficiency\ldquo ; in micromhos per lumen , in this case equivalent to micro-amperes per lumen , or micro-amperes per square centimetre of sensitive surface , divided by The Detection of Small Fluxes of Light .
A preliminary experiment to test the efficiency of selenium in detecting small fluxes of light was made as follows : thin rod of highly insulating porcelain was ground flat at one end , offering a circular surface sq .
cm .
in area .
Thick pencil lines were drawn along two opposite generators of the cylindrical surface , and a cap of selenium was thinly spread over one end .
The selenium was then sensitised in the usual manner by annealing , and the rod was inserted in a hole bored through a stick of ebonite , and fixed by means of binding screws .
When not exposed , the selenium was covered with a cap fitting over the aperture .
This miniature selenium is shown in fig. 1 .
Its resistance , when freshly prepared , was 50 megohms .
It was inserted into a circuit containing a battery of 20 volts and a Broca ( consequent pole ) galvanometer iving 0 division for ampere .
The " " dark\ldquo ; deflection was oompensated by suitably altering the netic field , Efficiency of Selenium as Detector of Light .
and the selenium was then exposed to various moderate illuminations .
The faintest illumination discoverable with certainty proved to be that due to a affin candle 5 .
from the selenium .
This illumination was lux , and the flux received on the sensitive surface was microlumen .
The deflection was 5 divs .
micro-ampere .
This means an additional conductivity of micromhos , and a galvanometric efficiency of omhos per lumen .
Both these efficiencies are low , and as the sensitive surface is of about the same area as the pupil of the eye , this method evidently falls far short of the ability of the human eye in detecting a candle at 5 Better results were obtained by a condenser through the selenium resistance and discharging it ballistically , but it was difficult to obtain consistent readings .
Electro ethods .
A Dolezalek quadrant electrometer giving 200 divisions per volt was used to measure the resistance of some resistance selenium brid , , either by Cardew 's method , or by inserting it in the galvanometer branch of a eatstone bridge .
A change of per cent. in the resistance was thus observed , due to an illumination of lux .
The galvanometric efficiency was of the order of 300 micromhos per lumen .
Results with : The best results , both as regards actual deflections and anometric efficiencies , were obtained with a Kelvin differential galvanometer coils of a total resistance of 6280 ohms .
In order to be able to vary the illumination measurably within wide limits , the arrangement shown in the diagram ( fig. 2 ) was adopted .
The Se No. 3 was fixed on the bottom of the box , which was optically blackened inside .
At a distance of 25 cm .
above the selenium was a circular ening 2 sq .
cm .
in area , closed by a plate of ground glass .
A metallic lamp was placed above the box at such a distance that the total candle-power of the ground-glass disc , as seen from the bottom of the box and determined photometrically , was .
The illumination at the bottom of the IJOX was thus when the ground glass was lighted to fiye candle .
By placing a small diaphragm on the surface , the ilhuuination could be reduced to any desired amount .
The faintest illumination so obtained was 10 microlux , or ten times that due to Arcturus unaided Since this was discovered by means of a selenium bridge , it means that it is possible to discover bright stars by means of selenium without the aid of any tical system whatever .
Dr. E. E. Fournier d'Albe .
On the As glass does not an illumination uniform over a wide angle , the precaution was adopted of picking out central , intermediate , and portions of the ground glass in turn .
Under the conditions here described no difference due to this shifting was discoverable , so that for the purposes of these measurements , as well as others described below , the rolmd may be considered as equivalent to a uniformly self-luminous surface .
One of the coils of the differential galvanometer was put in circuit with a battery of accumulators , giving 80 volts , and the selenium bridge .
The other coil , which had the same resistance , was tTaversed by a compehsating current of the order of 6 A set of was made by drilling ound holes of different diameters in a piece of sheet copper .
Complete series of measurements were made with four of these , called , and respectively .
Two smaller ones , called and somewhat uncertain results because the instrumental errors to be comparable with .
the quantities to be measured .
The table ives the dianoeters and areas of the apertures used , and the candle-power of the small of ground glass they exposed to the light , together with the illumination produced by that candle-power at the distance of the selenium bridge ( 25 cm Area .
BFED Illumina.tionmicrol .
The galvanometer having been made nearly aperiodic , readings were taken after half a minute 's exposure , then after half a minute 's recovery , and so on alternately until successive ' dark\ldquo ; and ' light\ldquo ; readings had become sensibly constant .
The exposures were made either by moving a cardboard shutter or by switching on the lamp , the room being otherwise in complete darkness .
Finally , the shutter having been closed , were taken for five or six half minutes in order to detel.mine points on the recovery\ldquo ; curve .
The timing was done by listening to the beats of a metronome marking seconds .
The accompanying diagram gives the so obtained in the of aperture A. On joining up the " " light\ldquo ; and ' dark\ldquo ; readings respectively , two curves Efficiency of Seleninm as a of Light .
are obtained whose vertical distance apart we may call the " " amplitude\ldquo ; of the alternating exposure culVe .
These readings , therefore , us two main results : the " " total recoyery\ldquo ; , calculated as explained , from ordinates in the recovery curve ; and the " " amplitude\ldquo ; P. Both these quantities should be plotted against the illumination I , in order to discover the laws of the action .
This is done in fig. 4 , which fives the total recoveries after ious illuminations , and , which gives the amplitudes FIG. 3 .
FIG. 4 .
The values for the apertures and were too small to be obtained in the way .
They were somewhat roughly estimated by waiting umtil the deflection in the light seemed to ) proach a final value , then , and until the ecovery seemed to do the same , and the mean of two deflections for the " " total recovery The conclusions can be drawn from these cllrves:\mdash ; ( n ) The amplitude is proportional to the illmnination .
The curve is a line , which , however , does not ) through the origin , and probably has a curved portion between and the origin ( lig .
5 ) .
( b ) The total recovery is proportional to the square root of the illumination .
This appears to hold idly down to the feeblest illuminations , as far as the Dr. E. E. Fournier d'Albe .
On the limits of accuracy allow one to judge .
These limits are indicated in the curve ( fig. 4 ) by rectangles including all admissible values of the illumination and of the total recovery .
( c ) The ratio of amplitude to total recovery is proportional to the square root of the illumination .
This follows from ( a ) and ( b ) , and is subject to the error specified under for very Saint illuminations .
One division ( 1 mm. at 1 of the galvanometer scale represented a current of micro-ampere .
This , with 80 volts , means that each millimetre of deflection indicates a gain or loss of conductivity amounting to micromhos .
Hence the " " galvanometric efficiency\ldquo ; of the selenium Se 3 is given by the following ures , which , of course show an efficiency inversely proportional to the square root of the illumination .
To discover still fainter illuminations , a minute hole , cm .
in diameter , was pricked in tinfoil , thus giving an illumination of 10 microlux ( about that due to Venus maided ) .
It was not discoverable with certainty by means of bridge Se No. 3 , but was easily discovered by means of a of resistance 20,000 ohms , which could be worked with 15 volts .
The " " amplitude\ldquo ; was divisions , derived from 20 alternate readings , during which , however , there was a strong drift owing to of temperature .
The efficiency is the same as in case of hole G. with Potassium Photo-electric Cell .
It is of interest to compare the above results with the best performance of a potassium photo-electric photometer described by Elster and Geitel .
* They placed a minute hole sq .
mm. in area in front of an amyl acetate lamp at .
The illumination thus received was 3 microlux .
The photoelectric current was then ampere .
With 10 microlux , in the Iast experiment , the current was -ampere .
Selenium , therefore , with the same illumination and extent of sensitive surface , is capable of , iving a J. Elster and H. Geitel , ' Pbys .
Zeitschr 1912 , vol. 13 , pp. 468 and 739 .
Efficiency of Selenium a Detector of measurable effect at least 100,000 times greater than that furnished by a potassium photo-electric cell .
Comparison with the The sixth magnitude is usually assumed as that of the faintest star cleal .
lie visible to the naked eye .
Zollner estimated the ratio of the htness of the sun and Capella as 55,700,000,000 .
The vertical illumination has been put at 50 , ( Exner ) .
This would make the illumination due to Capella ( magnitude ) microlux .
A star of would then give an illumination of microlux , and an illumination one-third of that would have to be considered the limit of human vision .
Let us call it 3 milli-microlux .
An attempt to test this limit experimentally was made as follows:\mdash ; A 10-volt 1 candle-power electric lamp was encased in a cylinder closed with a disc of ground lass .
To secure an even illumination of the latter , the cylinder was lined with white cardboard .
An image of this disc was thrown by a lens on to screen G. The lens was stopped down by an iris diaphragm until the whole luminosity emitted by the screen approached the limit of accurate comparison with a pentane lamp .
At that point , the candle power of , in the direction of the original beam produced , was to be , as compared with the pentane standard .
The experiments were conducted in a dark and the eye was carefully protected from all stray light .
Further reductions were then made by inserting small measured stops into the iris .
When the area of the stop was sq .
cm .
, the computed candle power of the illuminated surface of sq .
cm .
in area , was micro-candle .
A faint " " star " " was then produced ) a small stop on the ground glass G. A stop of area sq .
cm .
was not visible at all .
A stop of area sq .
cm .
was visible after 10 minutes ' accommodation to complete darkness , and a stop of area sq .
cm .
clearly visible .
Averted or " " rod\ldquo ; vision was employed in every case as a final test .
The object was 20 cm .
from the eye .
The candle powers were respectively , and 0.395 milli-micro-candles .
The illuminations at 20 .
were , and milli-microlux respectively .
Another observer obtained the same results , except that it was very difficult to perceive the second stop .
On the whole , the results show a satisfactory agreement with the star estimates .
It was next attempted to find the limit of visibility of a faintly illullinated .
The surface examined was the ground ho.lassG , with the iris diaphragm stopped down by means of very minute holes pricked in tinfoil .
Dr. E. E. Fournier On the The results showed that a surface becomes invisible when its intrinsic brightness is less than candles per square centimetre ; 1 sq .
cm .
of the barely visible surface gives an illumination of 10 milli-microlux at 20 cm .
If its brightness were concentrated at a point , that point would be clearly visible .
The eye , therefore , suffers in sensitiveness when the light is diffused .
The selenium detector does not .
If the two were equally tive to a point source , the selenium detector would be mor .
sensitive than the eye for faintly illuminated surfaces .
The fluxes of light discoverable by the are very minute .
In the dark , the diameter of the pupil is about 6 mm. , and its area sq .
cm .
With an illumination of 3 milli-microlux , the flux received is lumen .
This is , therefore , the minimum flux of light perceptible to the eye .
The Theoretical Instrument liimit .
The limit reached with the experimental arrangements described above is still very far removed from the limit theoretically attainable with the most sensitiye instruments now available .
This is evident from the faGt that a potassium photo-electric cell closely approaches the efficiency of the eye , although the current obtainable from it is 100,000 times feebler than the CUlTenG furnished by the selenium bridge .
Elster and Geitel worked with currents of the order of ampere , which represents approximately the present-day limit of current measurement .
Now if a current of the order of ampere is obtained with an illumination of 10 microlux , a current of ampere should , if the square-root law holds throughout , be obtainable with an illumination of microlux , or the unaided illumination of a star of the magnitude , which is invisible in the most powerful telescope hitherto constructed .
If , on the other hand , the current amplitude varies simply as the illumination , an illumination of milli-microlux should be discoverable , which is at least 100 times less than that discoverable by the eye , so that stars of the 11th magnitude s. hold show an electric effect unaided .
This advantage over the eye would be kept whatever the optical system , so that a telescope would give no advantage to the eye without giving it to the selenium detector to the same extent .
Selenium , therefore , offers an extension of our field of perception far beyond present optical or even photographic limits .
Bearing on the Theory of Quanta .
The efficiency of a lamp which converts the entire energy supplied to it into luminous radiation of the most visually advantageous type is 55 candles Efficiency of ium , as a Detector of Light .
per watt .
This value , given by Buisson and Fabry , intermediate between the values found by Drysdale and by Hyde , who obtained 17 and 72 respectively .
That portion of the total flux of energy from a Hefner lamp lies within the visible spectrum is about 200,000 ergs per second .
The portion intercepted by 1 sq .
cm .
at 1 .
is 16 erg per second .
This is the equivalent of 100 micro-lumen , and the illumination is llux .
The eye , as we have seen , can an incident luminous energy of 8 .
lumen , which is equivalent to ergs per second .
Planck'sf energy quantum for a frequency is .
For a frequency , which is that of maximum visibility , the quantum is erg .
The number of quanta received by the eye when receiving the minimum visible light is , therefore , about 360 per second .
Any instrument , therefore , which has a sensitiveness 1000 or more times greater than that of the eye may be effective in discovering discontinuities in the light emission at feeble illuminations .
As selenium is by far the most efficient detector known , the efforts in this direction made by N. Campbell with Na-K alloy cells should be repeated selenium of In.visible Radiatio In comparing the efficieucy of selenium with that of the eye or of other detectors , it is necessary to deduct the effect of the invisible spectrum .
The maximum energy of most terrestrial sources of is in the infra-red , and if selenium were , like the bolometer or the thermopile , simply a of radiant energy , its performances in sources of terlestrial radiation could not be arded as evidence of superiority in visible ions .
In order to study this question experimentally a new form of seleniurn bridge was constructed , consisting of a porcelain rod .
in diameter and 5 cm .
long .
Two thick parallel pencil lines were drawn thwise the rod , a clear white line cm .
wide and 4 cm .
between them .
The white line was then over with selenium , was then sensitised .
The result was a ' line \ldquo ; suitable for scopic vork .
After constructing such elements the most sensitive chosen .
It had a normal galyanometric efficiency of micromhos per lnmen .
This line was exposed to the spectrum of a Nernst lamp produced by flint glass prism .
The isible spectrum was cm .
* Buisson and Fabry , ' Comptes Pendus , ' 1911 , vol. .
Planck , ' Ann. der Phys 1901 , vol. 4 , .
Campbell , ' Camb .
Phil. Soc. Proc 1910 , vol. 1 ) Dr. E. E. Fournier d'Albe .
On the portion which affected the selenium was not longer than 8 cm .
It was divided into half centimetres , and the lin bridge was successively placed at eacf ] division .
Half minute " " amplitudes\ldquo ; were taken for each position , moving the bridge first from ultra-violet to infra-red and then in the opposite direction .
The mean amplitudes so obtained for each position were divided into two groups , those for the invisible spectrum and those for the visible spectrum respectively .
Expressed in percentages of the total radiation the effectivebnergies were\mdash ; Ultra-violet Visible Infra-red Although it is evident that the invisible radiations are not in the regate as effective as the visible rays , the above figures do not necessarily represent the quantitative distribution of the effective energies .
The latter cannot be accurately obtained without ascertaining the law of light action for each radiation separately .
This was done by Pfund* for short exposures and for two diff'erent intensities .
He found a somewhat abrupt transition at a wave-length of 6500 .
Below that wave-length the deflection was proportional to the square root of the energy , while from .
to the end of the spectrum it was simply proportional to the energy .
I have already mentioned that the deflections so obtained are not necessarily proportional to the total effect , and a more detailed investigation of the law of action will be necessary for each part of the spectrum .
Meanwhile , the following qualitative experiment shows that the dissymmetry found by Pfund does not apply to the final deflection .
The Nernst lamp spectrum was allowed to fall on a ground-glass screen closing the front end of a tube 43 cm .
long , lined with white cardboard .
Any portions of the spectrum could thus be recombined separately , and the selenium at the other end received a definite portion of the aggregate energy of whatever parts of the spectrum were transmitted through the ground glass .
The spectrum was divided into two portions which exerted equal actions .
The division proved to be slightly on one side of the line towards the blue .
The separate effects were always in the greater than the total effect , showing that the effect varies as a power of the energy below the first power .
The slit of the spectroscopic arrangement was then * A. H. Pfund , ' Phys. Rev 1912 , vol. 34 , Effi , ciency of Selenium as Detector of Light .
arrowed or widened , and the selenium exposed alternately to the two porvions , as before .
The intensity was thus varied in the ratio of at least , but the total deflections were sensibly equal for both portions .
shows that there is no great difference in the exponent of the for the two branches of the resonance curve .
The general result for the spectral distribution of the effective is that about three-quarbers of the total effect are due to visible radiations , eyen when the energy maximum is in the infra-red , as it is in most terrestrial sources .
This proportion is maintained , or exceeded , with faint illuminations .
If Pfund 's relation held , selenium would be sensible to none but those radiations which to the visible spectrum , if the were faint enough .
In any case , only some 30 per cent. at most need be deducted the obseryed effect in selenium in order to reduce it to visible light .
The Detect of Small Ch anges of Brightness .
The limit of accuracy in photometric estimates is enerally rnised to per cent. , and as there is no optical means of increasing the contrast of extended surfaces , a limitation is thus imposed upon the estimation of bri , htness such as physicists are not content to accept in other measurelnents .
Contrast can be increased by almost indefinitely , * it is a rocess , and gives no measure of the original contrast .
It is therefore of interest to inquire whether selenium , besides efficient as a detector of feeble , is superio to the eye as the appreciation of ulinlte differences of illnmination .
To take a concrete case .
A selenium of efficiency 1000 will 1000 micro-amperes with one volt if its surface is 1 sq . .
and the illnmination llux .
With an illumination of ( feeble the current will be 100,000 micro-amperes ( assuming the square-root law ) , or 10 microif the sensitive surface is only 1 sq .
cm .
A diffel.ence in the illumination lnting to 1 per cent. will make a difference of cent. in the current , a difference of micro-ampere ( say , 100 divisions on the scale of a sitive talvanometer ) .
There is , then , no reason why a difference of cent. should not be electrically discovel.ed , which , of course , is eyond the power of the eye .
This conclusion was confirmed by a nulnber of experiments , one of which the following:\mdash ; A disc , 21- ) .
cm .
in area , ated by a lamp from itbove .
The disc then transmitted a eqnivalent to 1/ 8 candle .
selenium bridge placed 25 cm .
below the disc was exposed to this , and *E .
E. Fournier d'Albe , ' Roy .
Dub .
Soc. Proc 1909 , 12 , vol. 11 , p. 97 .
Efficiency of Selenium as Detector of Light .
was connected up to a P.O. resistance box .
A battery of 8 volts and a Broca galvanometer adjusted to give 100 divisions per micro-ampere were also connected up .
Adjustment being made for equilibrium , a black thread mm. thick was drawn across in contact with the ground glass surface .
When passing across the centre of the disc it shaded off per cent. of its area .
A deflection of 8 divisions was obtained on the galvanometer .
With 20 volts a deflection of 20 divisions was obtained .
It was possible , then , to discover a variation of per cent. with this comparatively insensitive arrangement .
It is evident , therefore , that for such purposes as photometry and half-shadow polarimetry the selenium bridge must become a valuable accessory , capable of bringing these measurements up to the level of accuracy of other standard physical determinations .
1 .
The efficiency of seleninm preparations used for detecting is strictly defined , and standard conditions are chosen for determining it .
2 .
The efficiencies of some selenium preparations are evaluated from data yiven in previous publications .
3 .
Experiments are described which were made in order to detect minute quantities of light by means of selenium .
4 .
The minimum illumination reached is 10 microlux .
5 .
The law of light action is investigated , and shown to be , in the main , square-root law , down to the feeblest illuminations .
6 .
The theoretical limit of light action discoverable by means of presentday methods and apparatus is calculated , and shown to be , with selenium , very far beyond the power of the eye .
7 .
It is shown that this offers a means of the question of the discrete structure of radiant energy ( theory of quanta ) .
8 .
It is shown , theoretically and experimentally , that selenium apparatus is capable of differences and variations of luminosity quite inappreciable to the eye .

