]\gt ; BAKERIAN \mdash ; The Pressure of Light the : the Recoil from Light .
By J. H. POINTING , Sc. D. , F.B.S. , and BARLOW , D.Sc .
( Received January 13 , \mdash ; Read March 17 , 1910 .
) All experiments on the pressure of radiation have hitherto been made on the force exerted by light or radiation on a receiving surface .
The experiment now to be described shows the pressure of radiation against the source from which it starts , and from with a gun we may term this the recoil from light .
It does not appear practicable to show this effect by using a source in which heat is developed intrinsically .
But if radiation falls on an absorbing body it heats the body and the heat so developed issues again as radiation , and it is possible to detect the effect of this issuing radiation .
Theory .
We may see the nature of the action to be looked for by an ideal case in which we allow a beam of light with energy per cubic centinletre to fall normally in a perfect vacuum in turn on each of four discs , the front and back surfaces of these discs being respectively as in fig. 1 , a 3 4 BACK where represents a fully or " " black\ldquo ; surface , and a fully reflecting or non-radiating surface .
When the radiation falls on an absorbing face , as in the case of either of the discs ( 1 ) and ( 2 ) , the temperature of the disc rises till a steady state is reached in which emission equals absorption .
We may suppose that the discs are so thin that the two faces are sensibly at the same temperature .
If we did not take into account the pressure due to issuing radiation , or if we only considered the initial effects before heating took place , the pressures on the first two discs would be in each case , due to the incident beam alone , and on the last two would be 2 , due to the sum of the incident and reflected beams .
We should have , therefore , ( 1 ) ( 2 ) ( 3 ) ( 4 ) Pressvre of Light the Source : the Recoil from Light .
535 But when a steady state is reached , the discs ( 1 ) and must be iving out as much radiant energy as they receive .
The first disc ives out equal amounts on the two sides , equal and opposite pressures .
All the radiation from the second disc is given out at the front side and is equal in to that of the incident beam .
Assuming this emitted radiation is distributed according to the cosine law , the pressure from it is easily shown to be , so that the total pressure on this disc is P. Since there is no absorption by discs ( 3 ) and ( 4 ) , we stillhave the pressures 2 ; hence we have now ( 1 ) ( 2 ) ( 3 ) ( 4 ) In a real case these results are modified in two ways : ( i ) By the possession of some small reflecting power by surface , and of some small absorbing and radiating power by surface S. ( ii ) By an inequality of temperature between front and back surfaces conditioned by the energy which is carried through from front to back to be radiated thence .
The vacuum is uot perfect , and there is radiometer action due to the residual gas , which , owing to the inequality of temperature , is not the same on the two sides .
This is probably the only way in which gas action is sensible , for the effects due to ordinary convection and conduction in the residual gas are .
The tenlperature diHerence , though sufficient to produce a differential radiometer action , is so small that in the radiation from the two sides of a disc we may take them as being at the same temperature .
In the experiment to be described the .
diffusion is so slight that we disregard it .
, then , only the reflection and absorption , let be the coefficient of reflection of the surface for the incident radiation , that of the coefficient of emission of for the emitted radiation , that S. It is then easy to show that the total radiation pressures on the four discs are respectively ( 1 ) ( 2 ) ( 3 ) ( 4 ) P. The emitted radiation is not of the same quality as the incident radiation ; and strictly we are not ified in assuming the elnissiye powers the two surfaces for the one quality of radiation in the same as the absorbing powers for the other .
But we shall for plicity suppose that the ratios are tho saule , a supposition which bless us to proceed , and which is probably not very far from the truth .
We have , erefore , 536 Prof Pointing and Dr. Barlow .
The of [ Jan. 13 , On this assumption Table Ihas been constructed , giving the four pressures for different values of and .
The pressure ratios in the last two columns are of connection with the experimental results , and will be referred to later .
Table I.\mdash ; Pressures on Discs , calculated for Different of the Reflection Coefficients .
The modification of the values in the above table by radiometer action will be greater for the BB disc than for the others .
In that disc energy proportional to has to be carried through the disc , and therefore the temperature difference is the greatest .
It is only possible to guess at the relative magnitudes of the radiometer actions for the different discs .
We have therefore sought to make the vacuum so that the action nearly disappeared .
The Experiment .
In the final form of the experiment each disc of a pair of circular cover glasses , cm .
in diameter and about mm. thick , between which was squeezed a layer of asphaltum also about mm. thick , the temperature being first raised sufficiently to render the sphaltum molten .
It is difficult to make the discs of uniform thickness and free from bubbles of gas ; but a great number were made and the four best were selected for use .
Such a compound disc appears to be perfectly opaque , and its surface is the blackest black and the least diffusing that we have yet been able to obtain .
The reflecting surface was made by depositing silver on the outside of the compound disc by means of the discharge from a silver cathode in an exhausted receiver .
A similar deposit on clear olass just allowed an arc to be seen through it .
1910 .
] Light tloe : the Recoil Light .
Four holes the size of the discs were cut in a stout plate mica ABCD , the centres of the loles being at the corners of a 2-cm .
square ( ficro .
2 ) .
The discs were then fixed in these holes by a minute amount of celluloid varnish .
The snspension rod and the mirror holder were attached to the mica plate , at the nUddle points of its top and bottom edges respectively , by very small copper clips without any cement .
A platinised mirror was cut in half , and the two portions M2 were mounted back to back in a suitable clip of copper at the of the rod , the plane of the mirrors perpendicular to the nica plate .
This system was suspended by a quartz fibre , 9 cm .
long , the centre of a lass flask of diameter .
The upper end of the quartz fibre was fixed to a ) collar held by friction in the neck of the fiask .
Both ends of the fibre were silvered and coppered , so that they could be ) and to the support .
After suspension the mouth of the flask was sealed off , alate ] al tubs in the neck being still connection the } apparatus .
538 Prof Pointing and Dr. Barlow .
The Pressure of [ Jan. 13 , To carry out the exhaustion of the flask to a very high degree , the general arrangement shown in fig. 3 was adopted , and the successive stages in the process were as follows:\mdash ; .
( 1 ) Preliminary exhaustion by aid of a Gaede pump .
The experimental flask A was kept hot by gas burners below it , and the charcoal bulbs and were strongly heated electrically by enclosing them in asbestos tubes concoils of platinum wire .
From time to time the whole apparatus was washed out with dry oxygen generated from the manganese dioxide in the bulb E. A small discharge tube indicated the state of the vacuum , and examination of the spectrum gave useful information as to the gas given off from the bulbs and .
At the end of three days these bulbs appeared to have ceased to off any gas , and the Gaede pump would bring the down to a hard -ray stage in about 10 minutes after admitting oxygen .
The temperature of and was then somewhat reduced , and the apparatus was sealed off at and in the figure represent phosphorus pentoxide drying tubes , and is a manometer .
( 2 ) The charcoal bulb was put in liquid air , and the temperature of the second bulb and of the flask A was then slowly reduced to the room temperature .
After 30 hours the apparatus was finally sealed off at During both these stages the -tube was kept always immersed in liquid air .
This arrangement formed a perfect trap for mercury vapour , which otherwise diffused from the pump into the flask and attacked the silver mirrors .
A roll of silver foil was placed in one limb of the -tube , with the object of making the trap more effective , but this precaution was probably unnecessary .
( 5 ) In the final stage of the exhaustion , the charcoal bulb was surrounded by liquid air , which was boiled off continuously at the reduced pressure of about 2 cm .
of mercury several hours before and during the whole of the measurements .
Experience showed that the highest vacuum 1910 .
] the Source : the Recoil from Light .
was obtained probably two hours after the application of fresh liquid air , and that it was necessary to renew the liquid air after every four or five hours .
The source of light ( fig. 4 ) was an Ediswan 50-volt ' Focus lamp which was fed from accumulators .
By means of an adjustable resistance in series with the lamp , the was maintained exactly at 60 volts , the lamp then taking a current of amperes .
The light was so steady for hours at a time that adjustment of the resistance was seldom necessary .
A photographic lens , of 15 cm .
focal length , provided with an iris , was arran ged to throw au image of the lamp filament on an achromatic lens of 19 cm .
focal length .
This second lens then formed a uniformly illuminated innage of the iris diaphragm on the disc to be forked with .
By adjustment of the diaphl.ag this image was made rather smaller than the disc , so that when the beam was centred on the disc unilluminated margin about lnm .
wide was left all round it .
The lamp and lenses were fixed to a board which could be lloved parallel to itself vertically and horizontal]y between guides , so that the could be sily directed on to the four discs in succession .
The centring of the on the disc was made by The flask was mounted on an ilon turntal ) with the ' accurately in the axis of rotation .
By rotation of the flask it was possible to experiment on the reverse sides of the four discs .
We shall refer to the observations made in two positions as ' direct\ldquo ; and " " reverse Thus the " " direct\ldquo ; BS disc becomes the " " reverse\ldquo ; SB disc .
The 540 Prof Pointing and Dr. Barlow .
The Pressure of [ Jan. 13 , flask was shielded from electrification and from extraneous radiation by enclosure in a cylindrical case of tinned iron , blackened inside and provided with windows to admit the beam of light and to allow the deflections to be obser For the deflections the image of an electric lamp on a millimetre scale , at a distance of 113 cm .
was used .
The definition of the was sufficiently good to allow deflections to be read accurately to mm. , the optical ularities of the glass flask rendered a telescope useless .
In finding the centre of .
from the deflections , a curious periodic motion was observed in its position , the centre } to and fro with simple harmonic motion .
The complete period of the torsional vibrations of the suspended system was seconds with no appreciable damping ; and the period of the motion of the centre of swing was found to be about seven minutes , that is nearly 11 torsional half periods .
The periodic motion was ultimately traced to pendulum motions of the system set up by external disturbances , for the amplitude of the displaced centre of swing was found to increase with such pendulum motion .
The pendulum period of the system was htly 1 for motion in the plane of the mica plate than for motion in a perpendicular plane .
Hence a pendulum motion once set up periodically from motion in a straight line to motion in an ellipse , and the complete cycle of these was actually gone through in seven minutes .
This meant a ) eriodic change in the angular momentum of the system about the axis of suspension , and to neutralise this the mica plate tended to turn with equal and opposite angular momentum about the vertical axis , and so to give a twist to the fibre .
Accordingly , we should expect the maximum twist to take place when the vibrations are linear , and this was observed to be the case .
To eliminate this effect , 12 consecutive turning points were always taken ( i.e. observations over a period of seven minutes ) , and the mean centre of swing calculated from these .
When the beam was allowed to fall on a disc , some initial effects were observed , of which we an below , and shall then a tentative explanation .
Results .
The observed deflections for the four discs are given in Table II .
These results are divided into two series , A and B. In Series A the beam of was kept on the disc under experiment until the deflection seemed nearly constant , the time of exposure being generally about 20 or 30 minutes before the deflection was read .
Attention was chiefly given to the BB and Deflections in scale divisions for the four discs .
tionR on the ( direct\ldquo ; and " " reverse ' ' sides ure denoted by and respectively .
* Rejected in taking as , owing to a breakdown of the for exh sting the liquid air , the vacuum had doub(less deteriorated .
In the Series made a month later , the ) was kept on each disc for one hour before reading the deflection , with the object of a still closer approach to a steady state .
One hour was also allowed before the zero after cutting olf the light .
The much greater time now required 542 Prof. Poynting and Dr. Barlow .
The Pressure of [ Jan. 13 , made it impracticable to observe the zero more often than twice while making observations on all four discs .
Experience showed that , provided the vacuum was well maintained , the zero seldom changed more than one or two tenth divisions during five hours .
There appeared , therefore , no objection to this course .
Examination of the table shows that in most cases rather different values are given by the " " direct\ldquo ; and " " reverse\ldquo ; observations .
This is particularly evident in the case of the BB disc , where the values , roughly 15 and 17 divisions respectively , differ by about 13 per cent. The explanation of this appears to be that the radiometer action was still sensible and acted differently in the direct and reverse positions of the disc .
The differen ce was probably due to inequality in the thickness of the two cover glasses enclosing the asphaltum layer .
Some observations were made without using liquid air , i.e. with the charcoal bulb at room temperature .
The deflections were then always very and the zero showed erratic behaviour .
A few deflections obtained under these conditions are given in Table III .
They merely serve to indicate the general effects of the gas action .
On the BB disc the want of symmetry referred to above is now greatly exaggerated , and points to the gas action being a suction on the direct\ldquo ; side and a pressure on the " " reverse It is also to be noted that the action is a suction on the BS disc .
Any tendency towards this action in tlJe high vacuum experiments would , therefore , tend to mask the recoil pressure for .
Table III .
Deflections for the four discs with charcoal bulb at room temperature .
( The minus sign indicates suction .
) On account of the uncertain values of the various corrections required , and on account of the existence of outstanding disturbances , it was felt that mere multiplication of observations would not lead to more exact results .
As the final values in scale divisions for the pressures on the four discs given by the experiment we therefore take BB .
BS .
: 1910 .
] Light against the : the Recotl from The values for BB and BS are the means given by both Series A and the values for SS and SB are from Series alone , since for these discs the steady state was not attained properly in the observations of Series A. Hence we have We select for comparison with these ratios those calculated for in the last columns of Table I , .
the values The latter ratio agrees better with the experiment than the former .
This .
is what we might expect , since the latter ratio does not involve the pressure on the BB disc , and that is the pressure which is most affected by the radiometer action .
As to the actual reflection coefficients for the surfaces of the we can do but little more than make a uess .
For the black surface we may take the glass sulface as reflecting 4 per cent. of the incident , and allowing another 1 cent. for the probal ) a reasonable estimate\mdash ; we have , finally , .
In the case of the silver surface the reflection was tested by means of a thermopile , and it was concluded that at least 96 per cent. of the beam was reflected .
The of the Beam .
A determination of the energy of the beam used was ma and afforded a means of calculating the absolut , values of the pressures to be expected .
As in the experiments of Nichols and Hul ] , and in other experiments on light pressure which we have made , the energy was measured by the beam to fall on a ackened disc of pure silver ( 2 cm .
diameter and cm .
thick ) , and by observing the initial rate of rise of tempel.ature by means of a constantan-silver thermo-electric junction soldcred to the disc .
A RubensPanzer ocalvano ) letter was used and was adjusted for a eriod of tbout two seconds and was then made d beat .
The transit of centimetre division of the scale across the field of view of the observing telescope was recorded on the drum of an electrically driven raph .
Just before and just after each series of transits , a set of 5-second intervals were also recorded in order to the peripheral speed of the drum .
number of micro- volts per scale division was then deternlined , tho thernno-electric power of the couple being known from a separate experiment , it was possible , by means of a graphical representation of the raph record , to calculate the rate of rise of temperature , and hence the of beam was 544 Prof Pointing and Dr. Barlow .
The Pressure of [ Jan. 13 , calculated to be erg per centimetre length .
This would be the force in dynes on a fully absorbing surface .
The moment of inertia of the suspended system was gramme-cm.2 , and its period was seconds .
The arm was cm .
Hence the beam should give a de , flection of divisions of the scale used when on a disc fully absorbing on both sides .
Assuming that the BB disc reflects 5 per cent. , the deflection should be divisions .
This is in close agreement with half the value obtained with SS and , and the excess over of the observed value obtained with the BB is probably to be ascribed to residual radiometer action .
The smallness of the excess shows that the radiometer action was reduced to a very small amount .
Initial Effects .
We have already referred to the necessity for exposing the discs in general to the beam for a long time before taking the observations of the deflections .
On the BB disc , however , there was no marked initial effect , and after a few periods the deflection was nearly constant .
When the beam was first allowed to fall on one of the other discs an initial disturbing effect was evident , and in some cases even half an hour seemed insufficient to produce a steady state .
On the BS disc there were sometimes indications of a pressure for the first few seconds , but a strong suction always set in , and this suction , after reaching a maximum , rapidly subsided , giving place finally to a pressure to the limiting value to the steady state .
The SS disc showed want of symmetry .
On the direct side there was suction followed by pressure , as in the last case , but on the reverse side there was at first excessive pressure , reaching a maximum and then slowly falling to the value given in the steady state .
The SB disc showed an initial excess pressure very similar to that on the reverse side of the SS disc .
For both the SS and SB discs the duration of these initial effects was so drawn out that an appreciably steady state was not attained in much less than an hour .
In fig. 5 the effects on all four discs are represented by curves in which the time is taken as abscissa and the deflection as ordinate .
The , residual effects which follow the cutting off of the light are also indicated .
As an explanation we suggest that these effects are due to the heat of the beam driving off occluded gas from the silver films , the expulsion of the gas causing a back pressure on the film .
In the case of the SS disc it was known that the silver film on one side was decidedly thicker than that on the other , so that there is no difficulty in for the want of synumetry observed in that case .
1910 .
] Light against the Sonr.ce : the Recoil from Light .
This explanation is supported by two observations : first , that with a stronger beam the steady state is sooner reached ; secondly , that in the case of the BS disc it was noticed that if after the establishment of the steady Initial Effects.\mdash ; The represents the dtion , plotted against as abscissa .
The direction of } deflectiou for ) ) is shown by the .
For the SS disc curves and correspond to the " " direct\ldquo ; and " " lever sides respectively .
The light is put on at and cut off at state one cut off the light for a minute or two , and then pnt it on again , initial suction took place but was much lesR than Assuming that effects observed were really due to the of 546 Pressure of Light against the Source : the Recoil from Light .
occluded gas as suggested above , it is possible to form an estimate of the total mass of gas given off by calculating from the curves ( fig. 5 ) the total impulse given to the disc .
Thus , if we assume the gas to be oxygen and suppose , further , that the molecules leave the film normally with the ordinary molecular velocity , we find that for the BS disc the total was about gramme .
Taking the volume of the experimental flask as 2litres , this quantity of gas would give a partial pressure of about So that unless the gas is rapidly absorbed by the charcoal it would appear that the might be sensibly affected .
Unfortunately we had no means of forming even a rough idea of the actual pressure of the residual in the apparatus .
The Temperature of the Discs .
In the steady state all the energy of the beam absorbed by a disc must be radiated from thel faces .
By the fourth-power law of radiation , we may therefore estimate roughly the rise in temperature of the disc .
The results are C. and C. for the rise in temperature of the BB and BS discs respectively .
For the BB disc it was also estimated that the temperature difference between the two faces was probably less than C. The temperature differences for the other discs should be still less .
It should be noticed that the glass is black for the issuing radiation , hence the asphaltum layer alone counts in the case of the BB disc .
Early Exper.iments with Platinum Discs .
In some early experiments we used two discs , BB and BS , of platinum foil mm. thick , the black surfaces being formed by depositing platinum black .
Results were obtained somewhat like what we expected as to the ratios , but the deflections depended very greatly on the state of the vacuum , and under the best conditions were about 50 to 100 per cent. greater than the values calculated from the energy of the beam .
This disagreement was doubtless chiefly due to radiometer action .
The black surlace , being flocculent , is obviously badly conducting , the temperature slope is therefore increased , and we have in consequence a big differential radiometer action on the faces of the discs .
Moreover , the polished platinum is a poor reflector , so that such discs quite fail to approach the ideal conditions .
These considerations led to the use of the asphaltum discs .

