218 * The Spectrum of Magnesium and of the So-called Magnesium Hydride , as obtained by Spark Discharges under Reduced Pressure .
By E. E. Brooks , B.Sc. , A.M.I.EE .
( Communicated by Sir William Crookes , F.R.S. Received October 14 , \#151 ; Read December 12 , 1907 .
) In 1878 Living and Dewar* examined the spark spectrum of magnesium in ' an atmosphere of hydrogen and noticed a bright line of wave-length about 5210 , occasionally accompanied by other fine lines on the more refrangible side .
In later papersf they remark that this line is only visible in hydrogen or when the C and E lines are also visible , and describe more fully the " two sets of flutings and a pair of fainter bands " which accompany it , and which appear to have been independently discovered by Ciamician .
J Living and Dewar inferred that some compound of the metal and hydrogen was formed , and since the time of their research these lines and flutings have been known provisionally as the " hydride " spectrum .
The term will be retained , for convenience , but it must be understood that its correctness is not assumed .
Numerous researches have since been made to elucidate the relations between the arc and spark spectra of magnesium , partly in connection with the mechanism of electric discharges in general and partly on account of the importance of the line 4481 in astrophysics and the question as to whether its intensity in stellar spectra can be taken as a criterion of temperature .
Of especial interest are the observations of Crew , Basquin and others as to the effect of a hydrogen atmosphere on the arc spectrum of this and other metals .
The results obtained cannot be briefly summarised , but references are appended to the more important work dealing with the subject .
Very little attention , however , appears to have been given to the hydride spectrum , which is a highly complex system of bands and flutings extending over the whole range of the visible spectrum , from the hydrogen C line to the magnesium triplet at 3830 .
The writer 's experiments were suggested by phenomena met with in the course of a research on the spectra produced when compound bodies are used as electrodes in vacuo .
The discharge vessel was usually a lamp glass fitted * ' Roy .
Soc. Proc. , ' vol. 27 , p. 494 .
t ' Roy .
Soc. Proc. , ' vol. 30 , p. 94 ; vol. 32 , p. 197 .
J ' Sitzber .
Akad .
Wissenscli .
, ' Wien , 1880 , p. 437 Spectrum of Magnesium and So-called Magnesium Hydride .
219 .with rubber stoppers carrying the electrodes , and the pressures employed varied from atmospheric to 1 or 2 mm. of mercury .
A large number of observationsK extending over several years , were made in different gases and with different kinds of discharge , but in this paper only the results obtained in the two special cases of alternating currents of high and low frequency will be dealt with .
For high frequencies a spark coil was shunted with two large Leyden jars of total capacity about 0004 microfarad , and an adjustable spark gap was put in series with the discharge vessel .
The primary current was from 10 to 16 amperes as a rule , obtained from alternating mains at frequency 50 , no contact breaker being employed .
The electrodes usually consisted of two long spirals made from single strips of thin commercial magnesium ribbon , .enclosed in glass jacketing tubes with from to f inch projecting at the ends and about ^ inch apart ; this form being adopted because a simple .device enabled them to be fed forwards or backwards as required without impairing the vacuum .
Absolute gas-tightness was of course impossible , but the rate of leakage could be made quite small .
At a pressure of from t to jr inch of mercury and with an auxiliary spark gap of from 1/ 16 to 1/ 8 inch in length , an oscillatory discharge passes through the apparatus whose RM .
S. value ( as read on a hot wire ammeter ) is about 3 amperes , which implies a maximum value of the order of 100 or 200 amperes .
The discharge then assumes one of two possible forms which are essentially the same in all gases tried .
If the electrodes are too massive or too far apart , there is a luminous gas column between them and they are surrounded by green ragged haloes which do not tend to extend into the gas column .
This state may persist indefinitely without any marked waste of the electrodes or deposit on the walls of the vessel , and the spectrum is a mixture of gaseous and magnesium lines .
In the case of hydrogen , C and F are especially intense .
Increasing the pressure to 3 or 4 inches of mercury does not much effect the metal lines , but makes the four chief hydrogen lines very broad and diffuse ( the C line , however , often being much sharper than the others ) .
If the electrodes are not too massive , if they are the single spirals of ribbon already mentioned , this discharge quickly passes into a new and more interesting form .
The green glow increases in size and brilliancy , and becomes a green flame enveloping the electrodes and bridging across the whole distance between them , under favourable conditions completely replacing the gas column .
The magnesium ions have become the carriers of the discharge , and the arrangement is analogous to a mercury lamp in which the mercury is replaced by magnesium .
This state is perhaps most readily 220 Mr. E. E. Brooks .
The Spectrum of Magnesium [ Oct. 14 , obtained in hydrogen , and in this gas the phenomena are least complicated by secondary actions .
At the instant it sets in , a fine black deposit of metallic magnesium begins to form on the walls of the vessel , and rapidly obscuring the view , although by various devices observations can be maintained as long as is necessary .
Meanwhile , the thin electrodes waste quite slowly away , neither fusing nor softening , and a slow absorption of gas goes on , so that a few bubbles must be admitted now and then in order to keep the vacuum from becoming too good and interfering with the oscillations .
This will be called the " high-frequency flame discharge .
" It gives a brilliant spectrum of perfectly definite character , which can be recognised at a glance , and which differs in many respects from the spark spectrum at ordinary pressures , and also from the normal arc spectrum , but bears some resemblance to the arc spectrum under reduced pressure , as described by other investigators .
The points of chief importance may be summarised as follows ( assuming at first hydrogen to be present):\#151 ; .
( 1 ) A good magnesium spectrum is obtained , in which the gas lines are not obtrusive , although the C and F lines are sharp and strong and the fine line hydrogen spectrum more or less visible at the red end .
The metal line 4481 is very intense , but still sharp and clear .
As regards the fainter magnesium lines , one at 3938 is always found on negatives taken with sufficient exposure , and there are two very faint lines near 3895 and 3900 , and another pair near 3848 and 3853 .
These appear to correspond to some of the fainter lines of the spark spectrum at ordinary pressures ( excepting the stronger line 3938 , which seems to be only mentioned by Fowler and Payn).* Two other faint lines at 3944 and 3961 may be due to aluminium as impurity .
( These results were obtained with glass prisms , and observations have not as yet been extended into the ultra violet .
) ( 2 ) In hydrogen the hydride spectrum is always present in great intensity .
This is , in fact , the most convenient method of obtaining it .
It is usually present , often with almost equal intensity , in the case of other gases , unless very special precautions are taken to exclude water vapour , but disappears in the complete absence of hydrogen and water vapour .
( 3 ) A very prominent feature is the presence of two pairs of sharp lines at wave-lengths approximately 4433*6 , 4428'2 , 4390 , 4385 , the more refrangible pair being slightly the stronger .
This is the only type of discharge which gives them readily .
Three of these lines appear to have been first observed , but not identified , by E. A. Porter.f * ' Roy .
Soc. Proc. , ' 1903 , vol. 72 , p. 253 .
t ' Astropliysical Journal , ' 1902 , vol. 15 , p , 274 .
1907 .
] cmcZ of the So-called Magnesium Hydride .
Fowler and Pain * observed both pairs in the spectrum of the magnesium arc in an air vacuum , and concluded that they were really enhanced lines of the metal .
No other reference to them has been found , although they are visible in a reproduction of the magnesium arc spectrum in , accom- panying a paper by J. Barnes.f As these lines will be frequently mentioned , it will be convenient to term them the " F and P " lines .
Their frequency difference is about 27 , a number bearing a simple relation to the frequency differences of 40*5 and 81 , which hold good respectively for two pairs and three pairs of magnesium lines in the ultra violet .
J The high-frequency flame discharge can be produced in a coal gas , air , nitrogen , or carbon dioxide vacuum ( these being the only gases tried ) , providing the terminals are not too massive .
A very great momentary current density appears to be an essential condition .
The type of spectrum just described remains the same , and , under ordinary conditions , the hydride fiutings and the C and F lines of hydrogen are invariably present , often with intensity little inferior to that obtained in hydrogen , so that photographs of the discharge in air or in hydrogen are practically identical .
As already stated , this is due to the presence of a trace of water-vapour , extremely difficult to get rid of .
Eventually it was eliminated by placing phosphoric anhydride within the discharge vessel itself for each experiment , , and then , in the complete absence of hydrogen or water-vapour , no hydride spectrum appears .
In a coal-gas vacuum the carbon fluting at 4315 comes out strongly .
In an air vacuum the first effect is a rapid fusion and scattering of the electrodes , due to the oxygen present , but the action quickly subsides , and the true flame discharge sets in with slow subsequent waste .
This effect is still more pronounced in carbon dioxide .
In nitrogen the discharge presents the usual features , but it is difficult to keep it steady for any length of time , and there is a strong tendency for a red nitrogen glow to appear at one of the electrodes .
In all gases absorption occurs during this kind of discharge .
The low-frequency discharge ( i.e. , without jars ) is quite different in character and in spectrum .
In this case it was desirable to have at command a stronger current than could be obtained from a spark coil , and hence a small 2000-volt transformer , fed from alternating mains at frequency 50 , and with an adjustable liquid resistance in the secondary circuit , was usually employed , also magnesium rods about 3/ 16 inch diameter were * ' Roy .
Soc. Proc. , ' 1903 , vol. 72 , p. 253 .
t ' Astrophysical Journal , ' 1905 , vol. 21 , p. 75 .
X Baly , ' Spectroscopy , ' p. 502 .
222 Mr. E. E. Brooks .
The Spectrum of Magnesium [ Oct. 14 , more convenient to manage than strip , for the electrodes fused with a current whose R.M.S. value was perhaps only one-tenth of that carried easily under the former conditions .
For a long time the results obtained were very puzzling , the " F and P " lines and the hydride spectrum appearing or disappearing under apparently identical conditions , but whilst the latter was more often present than not , the former were very seldom found on the negatives .
Eventually it became apparent that several distinct spectra were present , the facts observed being briefly as follows .
When a current of , say , 03 or 04 ampere is passed through rod electrodes in a hydrogen vacuum at about 2 or 3 cm .
pressure ( the vessel itself containing phosphoric anhydride ) , the first effect is a perfectly steady pale glow around each electrode , which gives a faint hydrogen spectrum , but there is no luminous gas column between them .
If no special care has been taken with the electrodes , i.e. , if the surfaces are oxidised , vivid green points may keep flashing out on them , most numerous at first and gradually diminishing as time goes on .
The spectrum of these points has been observed frequently , and also photographed .
It is characterised by the presence of the hydride flutings and the " F and P " lines , and the great strength of the line 4481 , but differs from the high-frequency flame spectrum in the comparative weakness of the minor magnesium lines .
If the electrodes be trued up in a lathe so that their whole surface is perfectly bright and clean , very few or no flickering green points are seen , and in any case they disappear more or less completely in a very short time , but when one does flash out , its characteristic spectrum flashes out with it .
The subsequent course of events is practically the same whether the rods are clean or not .
Gradually a pale green tint creeps into the perfectly steady halo surrounding each electrode .
This gives the b triplet and the line 5528 on a background of the two hydrogen spectra , but there is not a trace of the hydride flutings or of the " F and P " lines , 4481 being very faint and the other magnesium lines practically invisible .
This spectrum has been repeatedly photographed , but only the brighter lines are shown .
The green tint steadily increases in intensity until finally , with the suddenness of switching on a lamp , the discharge lights up brilliantly , the glow around the electrodes becomes intense , producing a flame effect not unlike that already described ( but really quite different in structure and properties ) , and accompanied by a rapid and copious evolution of brownish-black fumes.* With equal suddenness the gaseous lines disappear completely , * This stage marks the fusing point of the electrodes .
Later experiments with unidirectional current have shown that the green glow and subsequent phenomena occur only at the cathode .
1907 .
] and of the So-called Magnesium Hydride .
leaving an intensely bright and sharp magnesium spectrum , but without the hydride flutings and without the " F and P " lines .
The line 4481 is distinct and about equal in intensity to 4571 , and relatively much fainter than 4703 and 4352 .
It is difficult to eliminate completely the hydride spectrum , but when it appears it is in flashes , and in any case its intensity is very small compared with its great strength in air or nitrogen not specially dried .
Photographs show barely a trace of it .
Hence the mere presence of dry hydrogen is not sufficient to bring out the hydride flutings .
In one experiment , without disturbing the apparatus , the connections were then changed to the coil with jars and the high-frequency discharge passed .
As is usual with such thick electrodes , the flame state was not obtained , and although the " F and P " lines were well seen the hydride flutings were invisible , except that after long running a faint trace of the line 5210 appeared .
In another experiment of the same kind some approach to the true flame discharge was obtained for a few moments and then the hydride flutings appeared distinctly , but only during that time .
The general inference is ( since abundantly confirmed ) that the " F and P " lines are not in any way related to the hydride spectrum ; they are probably true magnesium lines , but require the jars to bring them out well , whereas either at high or low frequency the hydride flutings require something more than the mere presence of hydrogen .
Unless special precautions are taken to eliminate water-vapour , the low-frequency flame discharge in hydrogen always gives a strong hydride spectrum , but the " F and P " lines have never been observed in it , although carefully looked for .
For instance , rod electrodes which had been used to demonstrate the absence of the hydride spectrum in dry hydrogen were directly afterwards placed in another vessel , but without phosphoric anhydride .
This vessel was then well washed out with hydrogen ( passed through strong sulphuric acid and over phosphoric anhydride before entrance ) , but when the final state set in the hydride spectrum appeared steadily .
Hence this spectrum can be seen in the presence of a trace of water-vapour under conditions in which it would be totally absent without that vapour , a hydrogen atmosphere being present in both cases .
A larger quantity of water-vapour makes very little difference , merely bringing out the oxide fluting at 5007 .
Such results made it a possible hypothesis that the hydride bands were ; always connected with the presence of oxide or water-vapour , in which 224 Mr. E. E. Brooks .
The Spectrum of Magnesium [ Oct. 14 , event there would be no occasion to assume the existence of a " hydride " at all .
It would simply be the spectrum of the metal in a peculiar and nascent state .
Some important evidence in this direction had been obtained years previously in the case of other elements and compounds .
For instance , in a similar way , but with more difficulty , a jet-like glow can be obtained from point sources on an aluminium electrode , and this always gives the characteristic fluting spectrum of that metal , by some writers attributed to the oxide ; but Hemsalech has supported Aron 's view that it is really due to the element itself .
It was , however , found that the hydride spectrum appeared with great intensity in the .high-frequency flame discharge in hydrogen even when every effort was made to eliminate oxide and water-vapour .
It is always difficult to estimate how far such efforts have been successful , but apparently there are two distinct methods of producing the state or substance which gives this spectrum , one dependent on a chemical change accompanying ionisation and the other on ionisation by a current of very high maximum value .
The low-frequency discharge in dry air or nitrogen presents problems of more difficulty and is still under investigation .
It is , however , perfectly certain that there are at least two quite different spectra to be seen in one and the same discharge , having distinct and recognisable sources .
One of these is characterised by the intensity of the line 4481 ; 4703 is weak and 4571 invisible , and the hydride spectrum is absent .
In air the oxide fluting at 5007 is strong at first , disappearing later , and in nitrogen it is absent .
The other spectrum , which depends in some way on the presence of nitrogen , probably on the formation of nitride , is remarkable on account of the great intensity of the line 4571 , which , next to the b triplet , is by far the strongest line in it , 44S1 being usually invisible , 5711 faint , and 5528 and 4703 of only moderate strength .
When the air or nitrogen has not been specially dried the hydride spectrum appears with great brilliancy .
There is an initial stage in which the cathode gets hot and may fuse , even with small current , but if " nursed " through this , a current of from 1 to 2 amperes may be passed for an indefinite time without fusion .
The " F and P " lines are faint , but in all cases are seen when discharge is sufficiently intense .
It is hoped to deal more fully with this part of the work in a later paper .
As regards the long-debated question as to whether the intensity of the line 4481 is a criterion of temperature , it may be pointed out that in the high-frequency flame discharge the single spiral of thin magnesium ribbon never fuses , although it is carrying a considerable current , its temperature probably being kept down by the free ionisation going on .
Yet this is the discharge which gives the line 448.1 with the greatest intensity and sharp1907 .
] and of the So-called Magnesium Hydride .
ness .
In the low-frequency discharge the whole apparatus is much hotter and the electrodes readily fuse .
In fact , this latter spectrum has been photographed when the thick electrodes have been in a molten state and liquid globules dropping from them , without any marked increase in the intensity of this line , which is relatively weak in this kind of discharge , and , under these conditions , the spectrum as a whole is not so good as when the flame state occurs without fusion .
Observations were made in several different ways on the high-frequency discharge , in order to determine the relation of the various lines to the electrodes .
In some experiments a three-way globe was arranged so that one of the electrodes could be seen " end on , " the other electrode being a ring of iron wire .
An image of the former electrode was thrown on the slit of a spectroscope by means of a lens , and thus different parts of the glow examined .
The line 4481 was only conspicuous quite close to the electrode , a slight displacement suddenly cutting it off , as was also the case with the hydride bands .
The line 4703 remained visible a little longer than 4481 , whilst the b triplet was brilliantly visible long after the latter disappeared , and , in fact , could always be seen more or less clearly in any part of the globe .
The line 5528 varied with the triplet\#151 ; a fact often noticed\#151 ; as if these lines have a common origin or are connected in -some way .
The discharge was also photographed by means of an ordinary camera with a diffraction grating in front of the lens , but without a slit .
The negatives showed the 4481 light locally strong at the electrodes , whilst an image of the whole lamp glass was formed by the light of the b triplet , and also , but more faintly , by the light of the violet triplet at 3830 .
Hence this latter radiation pervades the whole space as does that of the b group\#151 ; due mainly perhaps to reflection\#151 ; but partly to a real luminescence at considerable distances from the discharge proper .
Similar results were obtained by using an ordinary prism spectrograph without a slit , and the phenomena can be observed directly in a specially clear and interesting way by means of Dr. Marshall Watts ' " Binocular " spectroscope .
The 4481 light can be seen flickering brightly close to each electrode , and with a small gap and a strong discharge extending faintly across it .
The whole globe is seen in the light of the b group , and also in the light of the yellow line 5528 , these flickering in unison .
These observations are merely given as confirmatory of results obtained by other investigators .
It may , however , be remarked that the locus of the line 4481 is visible without any apparatus at all .
In the true high-frequency flame discharge the green glow envelopes the electrodes and YOL .
LXXX.\#151 ; A. Q 226 Mr. E. E. Brooks .
The Spectrum of Magnesium [ Oct. 14 , bridges over the space between them , but the root of the discharge is distinctly blue , and it is this blue region which gives the 4481 line most strongly .
( The low-frequency flame has no such blue core .
) The " F and P " lines were not examined in the same way , but in all probability it will be found that their source lies even closer to the electrodes .
Some useful information as to the relation between the various phenomena already described and the sign of the electrodes was obtained by separating the components of the current in a magnetic field .
For this purpose a narrow lamp glass was generally used , placed horizontally between the poles of an electromagnet so that the discharge was at right angles to the field , the magnesium electrodes being bifurcated so that on each the locus of origin of a positive and negative discharge could be readily seen .
Sometimes one of the electrodes was an iron ring .
With the means available the experiments were somewhat troublesome , and it has not been possible to obtain a direct current of sufficient voltage to confirm the results , which all pointed to the same conclusion .
The high-frequency discharge separated into two paths , equally pervaded by the green flame , and the faint bluish root which gives the 4481 line was found to be on the positive source only .
Blackening took place by diffusion uniformly all round the tube , no localisation being noticed .
When the low-frequency flame was used , each electrode showed a bright green glow , * also attached to the positive source , but without a blue core .
In this case the blackening was markedly local ; the glows behaved much more like jets , and could be driven in any predetermined direction by the field , instant blackening occurring in exactly that direction .
The experiments were decisive as to the positively charged character of the magnesium ions .
When one electrode was of iron , this also possessed a green positive glow , and by the same method it was found that positively charged magnesium ions were returning from it .
The pressure used was about 2 inches .
These experiments were made in hydrogen , but a remark occurs in a notebook , made nearly three years since , to the effect that " the green glow , due to incipient combustion in an air vacuum , clearly and definitely follows the field , always being on the anode , and accompanied by brilliant whitish points .
" ( These points are due to incandescent oxide .
) It was much more difficult to study the cathode by this method , and certain curious results obtained lack confirmation and are being investigated further .
In one experiment , also made long ago , a three-way globe was * This refers to a particular stage in the discharge .
The facts are somewhat complicated and are being examined further , but the gi'een jets always act like positively charged streams .
1907 .
] and of the So-called Magnesium Hydride .
used , fitted with a third idle terminal , a small metal disc which could be kept charged positively or negatively by a Wimshurst machine whilst the discharge was passing .
This was of the high-frequency flame type , and it was found that when the disc was kept positive , no black deposit formed on it , whereas when it was negative , such a deposit formed on it as well as on the walls of the globe .
If the third terminal was quite close to the discharge column , the black deposit formed upon it in either case .
Apart from the question of charge , there are certain well-marked peculiarities in the evolution of fumes and formation of deposits which seem worthy of brief mention .
It has already been stated that when the high-frequency flame discharge occurs in the presence of a trace of oxygen , there is an initial stage of fusion during which small particles of metal may be projected in various directions .
These strike the glass at various angles , either rebounding or adhering as small metallic spheres flattened by the impact , and would not have attracted attention except for a peculiar structure invariably present in the splash .
The metal particle is surrounded by a dark ring , and outside this , at a greater distance , is another fainter ring , as if the flying particle was surrounded by an atmosphere much larger than itself .
According to the direction of impact , these rings may be nearly circular or portions of ellipses .
When a particle strikes at almost grazing incidence , it leaves a long trace broken up into a series of separate splashes , but on either side is always found the faint line which marks the boundary of this atmosphere .
The particles themselves may average a millimetre or so in diameter , the largest met with being 5 mm. and its " atmosphere " 2 cm .
in diameter .
The fumes from the low-frequency flame discharge are usually lighter coloured , more distinctly visible , and form in a quite different manner .
The projectile effect is very seldom met with , but instead one frequently obtains , especially in air or nitrogen , very large and remarkable spiral formations on the glass , which , under favourable conditions , may be seen to correspond to almost stationary whirls inside .
Single instances of this kind would not be noteworthy , but it is remarkable that when such formations occur at all they are always of exactly the same type , and their probable appearance or otherwise can be predicted in advance .
[ Note added January 2bth , 1908.\#151 ; Since this paper was written , Professor Fowler has shown that the hydride bands and flutings occur in the spectra of sun-spots ; a most interesting and important discovery ; and he remarks that " there need be no hesitation in attributing the flutings in question to the compound magnesium hydride."* It is , however , somewhat difficult to * 'Notices Roy .
Astron. Soc./ vol. 67 , No. 8 , p. 530 .
228 Spectrum of Magnesium and So-called Magnesium .
imagine a compound of the metallic hydride type existing as such in the sun , even at the relatively low temperature of a sun-spot ; but , assuming this to he the case , and also that it is really formed in the arc and spark , it is still more difficult to understand why it eludes isolation by chemical means .
One would naturally expect magnesium hydride to he an unstable compound , and apparently the only chemical evidence for its existence is to be found in a paper by Winkler , * who obtained a non-homogeneous substance ( by heating a mixture of Mg and MgO to redness for four hours in an atmosphere of hydrogen ) , which certainly appears to have contained hydrogen , but whose exact composition seems somewhat doubtful .
On the other hand , in the current number of ' Nature 'j* is a paragraph referring to an article in the ' Observatory , ' ] : by Father Cortie , who has many times recorded lines attributed to water-vapour in the spectra of sun-spots , and who suggests the possibility of the presence of superheated steam .
Again , it may be noted that the metallic banded spectra of which the hydride flutings are merely one instance , are especially readily obtained in the oxyhydrogen flame , and Hartley and Ramage attribute these to the metal itself .
Professor Fowler ( whom I find has also been working for some years at the hydride spectrum ) informs me that the most convenient way of producing such .spectra is in the " flame " of the electric arc , but from my own experience I have no difficulty in believing that amply sufficient water-vapour may be present in this case also , unless very special precautions are taken to ensure its absence .
On the whole , it seems best to regard the objective existence of the hydride as an open question , awaiting further experimental evidence .
] REFERENCES TO PREVIOUS WORK BEARING ON THE MAGNESIUM SPECTRUM .
Living and Dewar , ' Roy .
Soc. Proc. , ' vol. 27 , p. 494 .
" " " " vol. 30 , p. 94 .
" " " " vol. 32 , p. 197 .
Crew , ' Astrophysical Journal , ' 1900 , vol. 12 , p. 167 .
Basquin " " vol. 14 , p. 1 .
Schenck " " 1901 , vol. 14 , p. 116 .
Hartley and Ramage , 'Roy .
Soc. Dublin Trans. , ' 1901 , p. 339 .
R. A. Porter , 'Astrophysical Journal , ' 1902 , vol. 15 , p. 274 .
Fowler and Pain , 'Roy .
Soc. Proc. , ' 1903 , vol. 72 , p. 253 .
Hartmann , 'Astrophysical Journal , ' 1903 , vol. 17 , p. 271 .
Crew , " " 1904 , vol. 20 , p. 274 .
Barnes , " " 1905 , vol. 21 , p. 75 .
Loving , " " 1905 , vol. 22 , p. 290 .
* ' Ber .
Deutsch .
Chem. Gesell .
, ' 1891 , vol. 24 , p. 1973 .
t January 23 , 1908 , p. 281 .
J January , No. 392 , p. 51 .

