388 Observations on the Fluorescence and Resonance Radiation of Sodium Vapour.\#151 ; I. By the Hon. R. J. Strutt , Sc. D. , F.R.S. , Professor of Physics , Imperial College , South Kensington .
( Received April 7 , 1915 .
) S 1 .
Experiments on Duration of the Resonance Radiation .
In a paper entitled " Luminous Vapours Distilled from the Arc , " * I have shown that sodium vapour , stimulated to give the line spectrum by electric discharge , and caused to distil away from the region of discharge , will carry its luminosity with it for a distance of many centimetres .
In other words , the luminosity is persistent , and goes on after the electrical stimulation has ceased .
In this case the spectrum emitted is the arc spectrum .
It includes not only the D lines , but also the lines of the subordinate series .
We have another method of exciting sodium vapour to emit the line spectrum\#151 ; I mean fluorescent excitation .
If the vapour is illuminated by a soda flame , it will emit the D lines by resonance .
It is true that in this case the subordinate series of lines are not excited , but still we should certainly expect at first sight that the D-line radiation would behave as it does with electrical excitation , and that the luminosity would move with the vapour if the latter was rapidly distilled away from the place of excitation .
An important experiment has been described by Dunoyerf which seems to show that this is not so .
He projected the distilled sodium vapour in a jet across a wide globe ; part of the jet was illuminated by a beam of sodium light ; and it was found , that when the vapour passed into the shadow it abruptly ceased to be luminous .
It may possibly be objected to Dunoyer 's experiment that , since the bulb traversed by sodium vapour was not kept hot , a cloud of condensed sodium may have been formed and that the observed luminosity may have been due to the diffusion of sodium light by this cloud , rather than to genuine resonance radiation .
For this and other reasons it has been thought desirable to repeat Dunoyer 's experiment in a modified form , under conditions more closely comparable with those of my own experiments on the electrically excited vapour .
* ' Roy .
Soc. Proc. , ' A , vol. 90 , p. 364 ( 1914 ) .
\#187 ; t ' Comptes Rendus , ' vol. 157 , p. 1068 ( 1913 ) .
Fluorescence and Resonance Radiation of Sodium Vapour .
389 Fig. 1 shows the arrangement .
The glass bulb , 4 cm .
in diameter , contains a few grammes of dry sodium .
It is inserted in b , the chimney of a small gas furnace , which supplies heat for distilling e sodium through the tube c ( 18 mm. inside diameter ) into the cold condensing bulb d , of 9 cm .
diameter .
It was necessary to prevent condensation in and this was achieved by warming it from time to time with a large Bunsen burner held in the hand .
The vessel was connected at eto a Gaede molecular pump .
A metal diaphragm , backed by a soda flame not too heavily salted , * was focussed on the side of c by means of a short-focus lens , indicated at / .
A lantern condenser of 4 inches diameter was generally used , but a cinematograph lens , giving less light but a better defined cone of rays , was sometimes substituted .
The most rapid distillation of the vapour was at the rate of about 15 grm. per hour .
In this case the vapour in c was so dense that the resonance radiation at g was limited to a superficial patch .
In other experiments , with less rapid distillation , a cone of luminosity converging to a point could be seen in the vapour.f In neither case was there the slightest indication of the luminosity being carried along with the vapour current .
The appearance down stream of the place of excitation was in no way different from the appearance on the up-stream side .
The experiment confirms Dunoyer 's conclusion , and emphasises the difference between electrical and fluorescent excitation .
In the former case the D luminosity is persistent .
In the latter case it is not so .
This will appear still more remarkable and anomalous if we compare the behaviour of sodium emitting the D line with mercury emitting the ultra-violet line at \2536 .
Mercury through which the electric discharge is passing emits this line along with the rest of the arc spectrum , and continues to do so for a time when distilled away from the place of excita-tion .
J As Wood has shown , mercury vapour can also emit the line \ 2536 * See Dunoyer , ' Journ. de Physique , ' January , 1914 .
+ After an experiment sodium can be removed from the vessel by washing it out with alcohol .
If the glass has become brown with reduced silicon , it can readily be cleaned by washing with dilute hydrofluoric acid .
+ For references , see ' Roy .
Soc. Proc. , ' A , vol. 91 , p. 92 ( 1914 ) .
390 * Hon. K. J. Strutt .
Observations on the without the other mercury lines when light of precisely this wave-length falls upon it .
So far the behaviour of mercury and sodium run parallel .
Now comes the difference .
The luminous centres emitting mercury resonance radiation of wave-length 2536 can be .
distilled away from the place of excitation as well and easily as if they had been excited electrically.* The centres emitting the resonance radiation of sodium cannot , as we have seen , be distilled away from the place of excitation at all .
I cannot at present make any suggestion as to how these facts should be regarded .
It seems very strange that the analogy between the behaviour of sodium and mercury should go so far , and then suddenly break down .
S 2 .
Invisibility of the Resonance Radiation through Dilute Sodium Vajpour .
Dunoyer 's improved methods of observing the resonance radiation of sodium have already been referred to .
He noticed that when a soda flame a , fig. 2 , is focussed by a lens b , on to the wall of a bulb d containing sodium vapour at a density corresponding to a temperature of , say , 300 ' C. , I Fig. 2 .
the resonance radiation did not penetrate into the depth of the vapour but was confined to a very thin layer at c. The light from this layer of superficial resonance could be seen from any direction in front of the plane ee , but from any direction behind this plane ( looking through the bulb ) was invisible .
Dunoyer suggests that this may be explained simply by the opacity of the vapour for the particular radiation , though he apparently had some alternative * See F. S. Phillips , ' Roy .
Soc. Proc. , ' A , vol. 89 , p. 39 ( 1914 ) .
The rate of distillation of mercury used by Phillips was only about twice as great as the rate of my most rapid sodium distillation , measuring the rate in each case by the number of molecules passing per second.^ Fluorescence and Resonance Radiation of Sodium Vajpour .
391 explanation in mind .
The absorption is , however , a sufficient explanation , as is proved by the simple experiment illustrated in fig. 2 .
The spot of resonance radiation was viewed through a second exhausted bulb / containing sodium .
This bulb was heated , and as the temperature rose the resonance radiation appeared dimmer , finally becoming invisible .
The temperature at which it disappeared was 160 ' C. as nearly as could be determined .
The thickness of the layer of vapour in f was 4*5 cm .
, but this bulb was not very well exhausted , and it may be that with the sharper absorption band obtained in the pure vapour a less thickness or density of vapour would have sufficed .
The bulb f is , of course , at this low vapour density quite transparent to daylight or lamplight , and the inability to see the spot of resonance radiation through it produces so strange an impression that the observer almost feels as if he is being tricked .
air .
PUMP | 3 .
Resonance Radiation with the Vapour a Magnetic Field .
The conical poles of a Dubois half-ring electromagnet were adjusted to a distance of 8 mm. apart , and a glass tube of 4*5 mm. inside diameter was arranged between them as shown in fig. 3 .
The tube was kept in connection with a Gaede molecular pump , and , by warming it with a Bunsen burner held in the hand , sodium vapour could be distilled up into the region between the poles .
A salted flame was focussed on this part of the tube , as in previous experiments .
The effects depend upon how much salt is introduced into the flame .
Two kinds of flame have been used .
The first , referred to as the weak flame , was an ordinary Bunsen flame , with a dilute salt solution sprayed into the gas mixture .
This spray was obtained by electrolysis of the dilute salt solution.* The second , referred to as the strong flame , was from a Meker burner , with abundance of salt melted on to the nickel grid .
Illuminating the sodium vapour with the weak flame , the intensity of the resonance radiation was diminished by exciting the magnet .
The stronger the field , the weaker the resonance .
With a field of 19,000 units , which was the limit attainable , the resonance was reduced to a very small intensity .
It is difficult to say definitely whether it disappeared altogether , since there * The arrangement is described in ' A Treatise on Practical Light , ' by R. S. Clay , p. 507 .
Macmillan , 1911 .
UpSODIUM Fig. 3 .
Hon. R. J. Strutt .
Observations on the is inevitably some diffusion of soda light by the striae in the glass , and by specks of dust ; this diffused light cannot be easily distinguished from the resonance radiation , and remains when the latter is quenched by the magnetic field .
The explanation of this experiment presents no difficulty .
The D lines produced by the weak flame are narrow .
They were examined visually in the second order of a 10 feet concave grating , and the breadth of Dx was estimated at 0*2 Angstrom , that of D2 at 0*3 Angstrom .
Dx is split by the Zeeman effect into four components , and with the field employed the two inner ones would be separated by 0 41 Angstrom.* The component lines of the resonance radiation are so narrow that their breadth may be neglected .
Al ] the components of Dx in the resonating vapour would , therefore , be thrown off the exciting line , and the resonance of Dx would be destroyed .
D2 is split into six components .
In a field of 19,000 units the distances of the respective pairs would be about 02 , 0*6 , and 1 Angstrom .
The two outer pairs , but not the inner pair , would be thrown off the exciting line 0*3 Angstrom broad .
We see , therefore , that the field should extinguish the whole of the Dx resonance , and half of the D2 resonance.^ This is in good agreement with the observations .
Using the strong flame instead of the weak one , the effects are more complex .
When the full magnetising current is switched on , the resonance radiation is observed to become brighter , to pass through a maximum , and then to diminish again .
On switching off , the maximum is passed through again , and then the brightness sinks back to its initial value .
These changes each take perhaps two seconds .
The magnetic field takes time to reach its full value , and the maximum brightness corresponds to a particular value of the field .
This value was determined , though not , of course , very accurately , by diminishing the ( steady ) magnetising current until , on switching on , the light could not be observed to pass a maximum any longer .
This condition was first satisfied when the steady field was about 12,000 units .
To interpret the observations , it is necessary to consider the structure of the D lines in the strong flame .
As observed visually in the second order spectrum of the grating , they were seen to be very much broader than the lines emitted by the weak flame .
The breadth of Dx was estimated at 0*8 Angstrom , and of D2 as IT Angstrom .
Further , the lines showed extremely distinct reversals .
The breadth of the relatively dark centre of the line was perhaps in each * See Rung and Paschen , Kayser 's ' Handbuch , ' vol. 2 , p. 670 .
+ The polarisation phenomena indicate that the two outer pairs of D2 contain only half the total light .
Fluorescence and Resonance Radiation of Sodium Vapour .
393 case about a quarter of the total breadth of the line .
This would give 0 '2 and 0'3 Angstroms for Di and Da respectively .
In the absence of magnetic force the resonating lines fall on the comparatively dark centres of the exciting lines .
As the magnetic force increases a resonating line is split into components which fall on to the brighter parts of the exciting line , and the intensity of the resonance increases .
A further increase in the field pushes the components of a resonating line beyond the maxima of the exciting line , and the resonance begins to fall off again .
To predict the exact field strength at which the maximum should be attained it would be necessary to know the precise distribution of intensity in each exciting line , as well as the comparative intensity of each Zeeman component of each resonating line .
Such data are not available .
At the observed optimum field , 12,000 units , the components of Di would all be off the dark centre of the exciting line , though well within the limits of the bright part of this line .
In the case of Da the innermost pair would not be clear of the dark central strip of the exciting line , but the two outer pairs would be on the bright part of this line .
Thus the explanation of the maximum which has been given is in good general agreement with the observed breadth and structure of the exciting lines .
To see the resonance clearly pass the maximum it is necessary to use a tube only a few millimetres wide , as described , for the strong field required cannot be produced in a larger vessel .
But if it is desired merely to show the increase it is better to use a tube 2 or 3 cm .
wide between large flat pole pieces .
The light should be limited to a strip 1 cm .
wide , along that part of the tube which is in the field .
A suitable diaphragm should be placed in front of the flame and focussed upon the tube .
S 4 .
Exciting Flame in the Magnetic Field .
The flames used were as before , and either of them could be placed between large flat pole , 'pieces , allowing of a maximum field of 9000 units .
The flame in each case was focussed on the wall of a bulb of 300 c.c. capacity containing sodium .
The bulb was kept hot over a gas burner , and was continuously exhausted by the Gaede molecular pump .
With the weak flame the resonance was diminished by the field .
The experiment serves as a striking demonstration of the Zeeman effect , and of the various experiments described it is perhaps the best for this purpose .
Since the vessel containing sodium vapour has not to get into the space between the magnet poles , it can be made large , and a large patch of resonance radiation produced upon it .
The effect of the field on this could quite well be shown to 10 or 20 persons at once .
Hon. R. J. Strutt .
Observations on the The explanation is , of course , analogous to that already given for the case when the same flame is used , applying the field to the resonating vapour , In the present instance the exciting line is split up and displaced so that the resonating line no longer falls upon it .
The case of the strong flame in the field is less simple .
The observed result is that the resonance radiation is increased , though this is not quite so striking as the diminution with the weak flame in the field .
In interpreting this result we have first to consider what effect the field will produce on the broad bright line .
Assuming that the Zeeman components are each as broad as the original bright line , the field is insufficient to separate them to any important extent .
We have , therefore , still a continuous bright background .
On the other hand , the comparatively narrow lines of the reversing layer are definitely separated .
Thus they may have uncovered the centre of the bright line , which is the part of it effective in producing resonance.* As a matter of fact when the source was examined with the concave grating , it was observed that the dark reversed centre of Di became brighter , on exciting the magnet , so that it was now difficult to see that the line was reversed .
In the case of Da the reversal became broader , though I could not be sure that it was less dark .
But the increase in the central intensity of Di accounts for the observed increase of resonance radiation .
Both the experiments of this section can readily be made with the volume resonance as well as with the superficial resonance .
It is merely necessary to work at a lower temperature . !
The effect on the volume resonance is the same as that on the superficial resonance\#151 ; decrease by magnetising weak flame , increase by magnetising strong flame .
S 5 .
Summary .
1 .
The centres emitting resonance radiation of sodium vapour excited by the D lines are not persistent enough to be carried along when the vapour is distilled away from the place of excitation .
This result is extraordinary , because it contrasts absolutely with the behaviour of sodium vapour excited electrically .
It also contrasts absolutely with the behaviour of mercury vapour , whether excited optically ( 2536 resonance radiation ) or electrically .
2 .
The resonance radiation of sodium cannot be seen through even a very dilute layer of sodium vapour placed in front of it\#151 ; a layer quite transparent to white light .
This explains why the spot of superficial resonance produced * Of course this discussion of the subject makes no pretensions to completeness , t See Dunoyer , loc. cit. \#166 ; Fluorescence and Resonance Radiation of Sodium Vapour .
395 on the wall of a glass bulb can only be seen from in front , when the light passes to the eye without traversing sodium vapour .
From the back it cannot be seen , as Dunoyer has observed .
3 .
The resonance radiation of sodium vapour is changed in intensity when the vapour is placed in a magnetic field .
If the exciting flame is weakly salted , the radiation diminishes with increasing field strength .
If the exciting flame is strongly salted , the radiation increases to a maximum and then diminishes again .
4 .
A change in intensity of resonance radiation can also be observed when the exciting flame is placed in the magnetic field .
In this case a weak flame gives diminished radiation in the field , while a strong flame gives increased radiation in the field .
5 .
All the facts summarised under 3 and 4 can be explained qualitatively and quantitatively , so far as the available data will go , by taking into account the known Zheman resolution of the D lines , and the observed width and structure of these lines as emitted by the flames used .
The latter data were obtained by observation with a concave grating of high resolution .
I have much pleasure in thanking Prof. A. Fowler , F.R.S. , and Mr. F. S. Phillips for help in making these observations with the grating .

