q4 ] On the Temperature Classification of Stars .
22/ " lid is determined by the nature of the atoms composing the yiral molecules , and is not a measure of the thermal work done in Cwper concludes with a discussion of the relations of specific " to atomic weight in the solid , liquid and gaseous states .
January 30\#151 ; Read February 18 , 1904 .
[ Plates 7\#151 ; 9 .
] Contents .
1 .
Historical Review ... ... ... ... ... ... ... ... 2 .
Aims and Conditions of the present Research 3 .
The Observational Conditions ... ... ... ... .
4 .
Description of the Instrument used ... ... .
5 .
Method of Work ... ... ... ... ... ... ... ... . .
6 .
Description of the Photographs ... ... ... . .
7 .
Discussion of the Photographs ... ... ... ... 8 .
Conclusions ... ... ... ... ... ... ... ... ... ... 9 .
Description of Plates ... ... ... ... ... ... . .
1 .
Historical Review .
PAGE In my first Bakerian Lecture in 1873 I dealt with the question of he spectra of stars , and pointed out that the facts accumulated up to hat time by Rutherford and others led to the view that in the eversing layers of the sun and stars various degrees of dissociation are it work .
_ I also suggested that the stellar evidence indicated that one of the r esults of dissociation temperatures could be to " prevent the coming together of atoms which at the temperature of the earth , and at a artificial temperatures yet obtained here , compose the metals , the metalloids , and compounds."* In a subsequent communication to the Paris Academy I wiote , t u II semble que plus un etoile est chaude plus son spectre est simple , et que les elements metalliques se font voir dans l orclre de leuis poids atomiques .
" This last generalisation rested upon the great preponderance of hydrogen in certain stars , which I classed as hottest on the gi ound that the blue end of the spectrum was open .
Of the spectrum of helium* * 4 Phil. Trans./ vol. 164 , p. 479 .
f ' Comptes Rendus/ vol. 77 , 1873 , p. 1357 .
Sir Norman Lockyer .
[ J 8,1 ) * a=r=.5=rk-'JxrjzsS raised , but it was not till 1892 that I was able to approach it by m of photography .
The interval was spent chiefly in researched he i^on solar and terrestrial changes in spectra when d\#174 ; " ^ and electric energies were obvious .
n In a paper on stellar spectra in relation to temperature , communicat , , 16 f0yal Societym 1902* I gave an account of an attempt at temperature classification of stars , utilising the fact that an extent \lt ; f spectra into the ultra-violet is produced by increased temperatur and further that a lower temperature in an atmosphere above a phot SvedTw TTthG f bs'rption in the blue end- The classified airived at was based on photographs obtained with instruments havin thd uWvilLTrlyT ^ * " 3 ^ absOTW"S ** \#171 ; The general results of the discussion was the conclusion that th .
stars so far considered might be divided into two series , one o !
ascending , the other of descending temperature .
Further , that th classification proposed was justified both by the relative extension , of the spectra into the ultra-violet and by the temperature sequence of .
the few typical lines then available for study .
By 1899 , laboratory work on the spectra of different substance i r 1 C11 er^ conditions , and the discovery of a terrestrial source of e mm by Kamsay , which enabled me to investigate the complete spectrum , had so far facilitated the study of the typical lines in the various stellar spectra , that I felt myself justified in attempting to c assi y the stars in relation to the chemical sequence revealed by the presence of gaseous and metallic lines , using especially the lines of helium and hydrogen and the " enhanced " and arc lines of the metals n his way I hoped to be able to test the classification of 1892 based on the relative lengths of spectra .
An account of this research was published in the Proceedings of the oya s ociety ( vol. 65 , pp. 186 191 , 1899 ) , and ultimately the complete results obtained were included in a " Catalogue of 470 .Brighter Stars Classified According to their Chemistry."f In this catalogue the stars were arranged in sixteen groups along a temperature curve with its apex in the central portion .
On the assumption that the chemical changes were due to temperature , me u mg in that term the possible results of electrical energy , the general arrangement of the stellar groups in the order both of ascending and descending temperatures was indicated , the group * ^ Phil. Trans. , ' A , vol. 184 , p. 688 .
t I ublications of the Committee on Solar Physics , ' London , 1902 .
14 j On the Temperature Classification of Stars .
\L fcajning the hottest stars , on the dissociation hypothesis , being placed he top of the curve .
It will be convenient to reproduce this table ' % 3 .
Crucian Taurian Rigelian Cygnian Polarian Aldebarian Antarian Argonian Alnitamian Acherman Algol ian Markabian Proto-metallic Stars J Metallic Stars Arcturian | Stars with fluted spectra Piseian Sirian Procyonian There was abundant chemical evidence to show that the mean mperature of the stars occupying the same height on both branches the curve was not very different , so that we have ten hoiizons , or ages , of mean temperature indicated in passing from the complete ited spectra of Antarian'Piscian stars to the simplified line spectra of xe y Argus type .
So far as we could judge from the photographs then available , the aemical changes gave a sequence identical with that desired for the tigth of ultra-violet spectra in 1892 , so that the latter classification as fully justified by the test which had been applied to it .
2 .
Aims and Conditions of the Present Research .
, As before mentioned in the work of 1892 , the relative lengths of the Itra-violet spectra were determined from photographs seemed with t nstruments having glass optical parts .
It was possible therefore o secure still another test by obtaining a new set of photographs ising calcite and quartz in place of glass , to enable the far ultra-violet o be obtained and studied .
Nor was this all , it seemed of the first mportance to utilise not only the length of spectrum in the blue , but he relative brightness of the different parts .
" W ' hat happens regarding he relative brilliancy of the different parts of the spectrum , as well as the extension into the ultra-violet by increased temperature , was very Sir Norman Lockyer .
clearly stated by Sir George Stokes* in 1876 , in the following won " When a solid body such as a platinum wire , traversed by a volt ?
7 current , is heated to incandescence , we know that as the temperati ?
increases , not only does the radiation of each particular refrangibili absolutely increase , but the proportion of the radiations of the differe refrangibilities is changed , the proportion of the higher to the low increasing with the temperature .
" This question was also investigated by Melloni and Crova ; and recent years exact determinations of the law of increase have bee made by Lummer , Paschen , and others .
Melloni showedf expen ?
mentally that the maximum radiation moved towards the moi 1 refrangible end of the spectrum as the temperature increased .
Cro\ made use of this fact in determining the temperatures of varioi incandescent light sources , and was one of the first to suggest !
that tf method was applicable to the determination of the temperatures of tli sun and stars .
3 .
The Observational Conditions .
The kind of spectroscope to be used and method of observation to b followed were indicated by the following considerations .
In order to utilise the effect of temperature changes to the full it wa necessary to record the red end of the spectrum as well as the ultrr , violet , as only in this way could the relative changes in intensity h recorded , hence it was desirable to employ only a small dispersion .
In addition to the natural differences photographed in the ultra violet , artificial differences due to the absorbing effect of oir atmosphere\#151 ; which , even when clearest , is more or less opaque to the ultra-violet radiations\#151 ; might be introduced , therefore it was considered advisable , in order to eliminate the effects of atmospheric absorption , to obtain the spectra of any two stars to be compared whilst they wen at approximately the same altitude .
Further , to avoid the many pit-fall ?
to which those who compare photographs taken on plates of unequal sensitiveness , and differently exposed and developed , are liable , it was obviously important that any spectra to be compared should be obtained on the same plate in order to secure identical plate sensitiveness and development .
Again , in order to secure similar optical treatment it was arranged to photograph both spectra near to the optical axis of the camera ; thus they are near together in the centre of the plate .
I am sorry to say this work has been considerably delayed by the long time taken in preparing a new camera and optical parts suitable for the research , as above defined , and latterly by a long spell of bad observing weather .
# ' Roy .
Soc. Proc. , ' vol. 24 , p. 353 , 1876 .
f Taylor 's 4 Scientific Memoirs , ' vol. 1 , p. 56 .
X ' Comptes Rendus , ' vol. 87 , p. 981 .
904 .
] On the Temperature Glassification of Stars .
4 .
Description of the Instrument used .
With regard to the new apparatus , I may state that an objective .4sm camera ( fig. 1 ) , was devised having a 2-inch 30 ' calcite prism sounted in front of a 2\#163 ; -inch quartz lens of 18 inches focal length .
Fig- .
1.\#151 ; Quarfcz-Caleite Prismatic Camera .
i The prism is so cut that its first face is perpendicular to the optic axis of the crystal , and it is arranged that the incident rays are normal to this face .
All the rays , therefore , pass through the prism parallel to the optic axis and in this way there is no double refraction .
The whole length of the spectrum is brought into focus in a plane inclined at 42 ' to the optical axis .
The apparatus is attached to the side of an equatorially mounted 6-inch Dallmeyer doublet camera with an angle in declination between the two optical axes equal to the angle of deviation of the calcite prism , so that the Dallmeyer is used as a I finder .
t Edward 's snap-shot isochromatic plates have alone been used , and as they are not sufficiently sensitive for wave-lengths between X 486 land X 550 ( approx. ) , there is a break in each spectrum about this i region followed by a further portion of the spectrum having its centre about " D. " The length of spectrum obtained is such that to ; H* is 0-355 inch ( 9-0 mm. ) and from He to Hy 0-165 inch ( 4*2 mm. ) .
5 .
Method of Work .
As shown in Table I , there are ten stages or steps of temperature to be investigated , and pairs of stars were selected in order to show the differences between adjacent stages or else the difference between stages widely apart .
Stages 1 and 10 are not available in consequence of the photographic faintness*of the Antarian and Piscian stars , and the fact that the only 232 Sir Norman Lockyer .
[ jan J star as yet known representing the Argonian group is situated in tfc Southern Hemisphere .
The stages already investigated are shown b brackets in the following table .
Table I. 10 .
9 .
8 .
Crucian .
7 .
Taurian .
6 .
Rigelian .
5 .
Cygnian .
Argonian .
Alnitamian .
3 .
Polarian .
2 .
Aldebarian .
1 .
Antarian .
Achernian .
Algolian .
Markabian .
Sirian .
Procyonian .
Arcturian .
-J Piscian .
Thus an Arcturian star ( second stage ) , was compared with a Sirian ( fourth stage ) , and so on .
In the case of some stars , e.g. , Capella , this was repeated several times , each photograph showing a different comparison .
In securing the photographs an attempt was always made to obtain the pair of spectra with the region between and Hy of the same intensity in each .
This condition is very difficult to fulfil in actual practice owing to the difference of magnitude of the two stars , their difference in declination , and hence in clock rate , and lastly the very important actual differences between the actinism of the two bodies in this region of the spectrum due to their different temperature conditions .
It was also arranged to obtain , whenever possible , the spectra of two stars near together , so that the chances of introducing atmospheric interference due to the different conditions possibly present in various parts of the sky might be reduced to a minimum .
In every case where the observer had reason to suspect any change in the atmospheric conditions obtaining during the two exposures , which sometimes extended over a period of 1^ hours , the result has not been included in the discussion .
Thus we have a series of comparison photographs from which all variable conditions except the natural variations in radiation have , as far as possible , been eliminated .
904 ] O/ i the Temperature Classification of Stars .
233 6 .
Description of the Photographs .
Many of the photographs taken are too faint to reproduce well aough to exhibit completely the striking differences shown between ie pairs of spectra .
I shall therefore content myself by giving a etailed description of each negative ( Table II ) , only reproducing hose photographs which are sufficiently dense to plainly show the henomena under discussion .
Table II .
No. of negative .
Stage of Type .
AltiDate .
Star .
* tem- tude .
perature .
0 From Stage 2 to Stage 4 .
Description .
i | i O f a Lyrae 4 Sir .
50 19 7.5.03 ! .
| [ a Bootis 2 Arct. 57 f 3 Ur see Maj. 4 Sir .
59 4 29.12.02 \#166 ; [ _a Ursse Maj. 2 Arct. 58 1 i C a Geminorum .
4 Sir .
1 42 14 17.2.03 2 Arct. 39 0 f a Geminorum 4 Sir .
68 37 21.1.04 1 L a Aurigae 2 Arct. 68 From S tnge 4 to Stage 6 1 j f 3 Orionis 6 Rig .
21 11 28.1.03 ] i 1 1 i Canis Maj. 4 Sir .
i 22 Faintly red , fairly bright to Hy .
Intense red , scarcely extends beyond K. Red scarcely visible , maximum about A 400 .
Fairly bright red , maximum about A 460 .
Red scarcely visible , maximum about A 418 .
Fairly bright red , maximum about A 455 .
Weak red , maximum about \ 420 .
Yery dense red ; blue maximum commences at F , centre at A 460 , weak beyond X 388 .
Red not so intense , and ultra-violet much more extended than in Sirius .
Dense red , centre of maximum about A 422 .
VOL LXXIII .
S Sir Norman Lockyer .
Table II\#151 ; continued .
[ Jan. 3 No. of negative .
Date .
Star .
Stage of temperature .
Type .
Alti- tude .
Description .
From Stage 6 to Stage 9 .
f k Orionis 9 Alnit .
201 oo Orionis 6 Rig .
24 J f y Ursse Maj i I 10.12.03 ^ l^a Ursae Maj. f r ] Ursae Maj. 14.1.04 Various Intervals .
6 1 Mark .
Aurigse f e Orionis 14.11.03 H t^a Canis Min .
Arct. Cruc .
51 Arct. Alnit .
i 38 Proc. | 41 Fxtreme Interval .
f t Orionis Alnit .
10.12.03 l^a Tauri Aldeb .
Although spectrum of Rigel is gene rally much strong er , that of Alnitan .
extends as far ink the ultra-violet .
Faint red , maxi mum about A 416 well sustained up to H* .
Fairlv bright red , maximum at A 460 , rapidly falling off beyond A 430 .
Faint red , maximum at about A 418 , bright extension well beyond the end of the Capella spectrum .
Very bright red , maximum at about A 455 , faint beyond K. red , centre maximum at Faint of about Us .
Bright extension far beyond the hydrogen series .
Yery bright red , maximum about Hy , and falls quickly beyond H* .
Yery faint red , centre of maximum about Hs , i.e. , near the more refrangible end of the Aldebarian spectrum .
Yery strong red , centre of maximum about A 465 , end of spectrum about Hs .
c)04 .
] On the Temperature Glassification of Stars .
7 .
Discussion of Photographs .
The following is a detailed discussion of the photographs described Table II .
Stage 2 to Stage 4 .
\gt ; ^o 19 .
This photograph of Yega ( fourth stage ) , and Arcturus ( second age ) , is very striking in the relative intensities of the two spectra , he red portion of the Arcturian spectrum is considerably more intense id forms the one end of a maximum which\#151 ; except in the green \#166 ; o ; ion where the plate is not very sensitive\#151 ; extends from D to about 454 .
The part of the spectrum more refrangible than that rapidly ecomes less intense , until at and beyond K it is very faint .
Matters re very different in the spectrum of Yega , Here the maximum adiation occurs about a third of the distance from H5 to Hy ( A 422 ) ad the spectrum extends without any great falling off in intensity to [ vj beyond that it is weaker but continues without any great decrease ) twice the distance on the more refrangible side of H " than the inter is from K. From that point to the end the spectrum gradually eclines .
Whilst the maximum radiation in this spectrum has moved holesale towards the ultra-violet , the red is relatively only about half he density of the red in the Arcturian spectrum .
No. 4 .
The general appearance of these two spectra leads to the onclusion that that of f3Ursse Majoris ( fourth stage ) is much stronger han that of a Ursse Majoris ( second stage ) .
A more careful examination , however , shows that whilst the tetached red part of the former is only seen with difficulty , the same \gt ; ortion of the latter is comparatively prominent .
The maximum intensity in cc is situated at about A 450 and does tot vary a great deal between there and K. After the latter point s reached , however , the fall is rather sudden , and the spectrum soon lies out .
In ( 3 the maximum intensity is between H and H " ; and the iltra-violet up to HK is fairly strong .
Beyond H* the intensity of he spectrum drops rather suddenly , and then continues with a gradual ieerease for some distance .
No. 14 .
There is no great difference in the lengths of these two pectra , the one of a Geminorum ( fourth stage ) , the other a Aurigse second stage ) , but the red portion of the second stage spectrum is cost decidedly more intense than that of the fourth^stage , the latter , n fact , being scarcely visible at all .
In a Aurigse the maximum ntensity is at about A 454 , but in a Geminorum it must be placed at \gt ; r near to H$ ( A 410 ) .
No. 37 .
In the taking of this negative the spectrum of Capella second stage ) received the advantage of exposure and is consequently fnuch stronger in the Hy\#151 ; region than that of Castor ( fourth 236 Sir Norman Lockyer .
[ Jan. 3i stage ) , and yet the latter extends as far into the ultra-violet as tl former .
Furthermore , the intensity of the blue part of the spectrum Capella rises to its maximum immediately at and commences decline towards the violet at Hv , whereas the maximum regie of Castor does not commence at once after leaving the green gap , ai attains its centre at about A 422 .
It will be observed that whether we take Arcturus or Capella i represent Stage 2 , the spectra of stars of higher stages have relative !
longer ultra-violet and reduced red radiation .
It has to be note ' however , that there are indications that Stage 2 will , as a result t further work , have to be divided , for Capella is certainly hotter tha Arcturus as determined in the manner now under discussion .
Stage4 to Stage 6 .
No. 11 .
On examination of this negative it is seen that the detache red portion of the spectrum of Sirius ( fourth stage ) is decidedly mor intense than the same portion of the Kigelian spectrum ( sixth stage : In the ultra-violet , however , we find that although both stars ar fairly high on the temperature curve , and , therefore , both spectr extend far into the ultra-violet , the extension of the spectrum of Rig\lt ; is more intense and greater than that of Sirius .
Stage 6 to Stage 9 .
No. 35 .
By reason of its greater exposure the spectrum of Rigt ( sixth stage ) is generally much stronger than that of k Orionis ( nint !
stage ) , and especially is this so in the red portions of the two spectra This inequality notwithstanding , the spectrum of extends practical !
as far into the ultra-violet as that of Rigel .
Various Intervals .
No. 34 .
In a Ursse ( second stage ) the red part of the spectrum i comparatively very bright , nearly as bright as the region betweei G and F. In y Ursae ( sixth stage ) this red portion is barely visible Again , in the a Ursae spectrum the maximum occurs at A 460 , and thei the intensity gradually declines to K , beyond which it is very faint The maximum intensity of the y Ursae spectrum is situated at abou1 A 422 , and it extends without becoming greatly impaired to He .
No. 36 .
In this comparison we have a very striking case .
Tht spectrum of Capella ( second stage ) is compared with that of Ursa Majoris ( eighth stage ) ; Capella has been over-exposed , so that tin red portion is abnormally intense , Ursae Majoris received tin correct exposure , and the red part of the spectrum is rather faint ' 904 .
] On the Temperature Classification of Stars .
237 [ ( ^withstanding this difference the spectrum of Ursse extends irther into the ultra-violet than does that of Capella , and not only oes it extend further , but the maximum intensity is extended much irther into the ultra-violet than is that of Capella , which drops off ipidly beyond K. No. 23 .
In this comparison the red part of the Procyon ( third age ) spectrum is much brighter than that of Alnitam ( ninth stage ) , he intensity in the longer portion of the spectrum of Procyon rises t once to its maximum at H^ , has its centre of maximum at about 460 , and at commences to diminish towards the violet .
In the pectrum of Alnitam , however , the maximum is delayed until the ggion about X 426 is reached , and is then sustained up to H ?
1 finally xtending to the ultra-violet with a marked superiority , comparatively , ver that of the Procyon spectrum .
Extreme Interval .
No. 33 .
This comparison of two type stars respectively situated lear the extremities of the temperature curve is naturally one of the most striking pieces of evidence in support of this method of tempera-ure classification .
The spectrum of Alnitam ( ninth stage ) is nowhere io intense as the red and blue parts of the Aldebaran ( second stage ) ipectrum , and yet it extends , more than twice as far towaids the iltra-violet , from Hp , as the hydrogen series , whilst the mo^re refrangible limit of the Aldebaran spectrum only extends to K. The maximum intensity of the blue portion of the Alnitam spectrum occurs much nearer the ultra-violet than that of the Aldebaran spectrum , the lattei attaining its maximum at about A. 465 .
8 .
Conclusions .
It may be pointed out that the temperature classification , confirmed by this research , does not agree with that published by Sir W illiam and Lady Huggins who , in their " Atlas of Representative Stellar Spectra " containing " a Discussion of the Evolutional Order of the Stars , " place the solar stars on a higher temperature level than the white stars .
A reduction of intensity in the corrtirruous spectrum beyond the hydrogen series to which attention has been called by more than one observer , Schumann* among others , does not affect the results which I have stated .
Another paper dealing with this and similar points is in course of preparation .
The result of the research may be stated as follows :\#151 ; Taking the stars assumed to be hottest in the chemical classification , we find that in all cases the relative length of the spectrum is reduced , * 1 Smithsonian Contributions to Knowledge , ' No. 1418 , 1903 , p. 23 .
238 On the Temperature Classification [ Jan. 3(1 and the relative intensity of the red is increased , as a lower temperature is reached .
rl hat is to say that where two spectra having their inter sities about the region Hp \#151 ; Hy equal are compared , we find that in th cooler stars , according to the chemical classification , the emissions i the red preponderate , whilst in the hotter star the ultra-violet is mor extended and intense .
My best thanks are due to Messrs. Rolston and Goodson , who tool the various photographs to which I have referred , the former als\lt ; helping me in the preparation of this paper , and to Mr. Wilkie fo : preparing the enlargements of the negatives .
9 .
Description of Plates .
No. No. of negative .
Stars .
Stage .
Type* j Plate 7 .
1 1 f V ega 4 Sir .
Arct. 1 19 \ Arcturus 2 o f Castor 4 Sir .
Arct. Sir .
Arct. 14 \ Capella 2 \ i O *7 f Castor 4 3 3/ \ Capella 2 Plate 8 .
J Eigel 6 Rig .
Sir .
Alnit .
4 11 \ Sirius 4 5 35 f k Orionis 9 !
1 R'gel 6 Rig .
6 34 f y Ursa ?
Maj. . .
\ ci Ursa ?
Maj. . .
6 2 Mark .
Arct. Plate 9 .
7 3fj J rj Ursa ?
Maj. . .
8 Cruc .
\ Capella 2 Arcfc .
8 A In it am 9 Alnit .
23 1 Procyon 3 Proc. Q qq J Alnitam 9 Alnit .
oo \ Aldebaran ... .
2 Aldeb .
In producing these plates the original negatives have been enlarged about 31 times .
Stage4 VEGA ARCTURUS Stage jt Stage 4 CASTOR CAPELLA Stage 2 Stage 4 CASTOR CAPELLA Stage I1 \#163 ; c* $ S ' vol. 73 , Plat Stage 6 RIGEL SIRIUS Stage 4 Stage 9 K ORIONIS RIGEL Stage 6 Stage 6 BURSAE MAJ ; OC URSAE MAJ : Stage 2 2$$j | f S " 00 vol. 73 , Pla\ Stage \amp ; 1 7 ] URSAEMAJ : ?
CAPELLA Stage 2 .
Stage 9 ALNITAM PROCYON Stage 3 Stage 9 ALNITAM ALDEBARAN Stage Z -^T 02 It JO

