8ULLETlN No. XXXCX.

ON THE DISPLACEMENTS OF THE SPECT'RUM LINE8 AT THE SUN'S LIMB

An investigation of the displ~cements of the spectrum lines at the sun's 11mb was made by W , 8. Adams in 1909, using the tower telescope and 30-foot, spectrograph a t Mount Wilson.* It w a s shown that, i f we except the lines chl~racteristic of the higher clir~~mosyhern which show little or no displacement, the grnst majority of the lines are d~splilced to the red ~elatlvely to ihe hnes a t the centre of the disk. T h ~ s relative slilft Increases with the wave-lengtb for all the elements ~nvestigated, and for iron tho displacements increase rather more rapldly than in direct proportion to wave-length.

Dr. Adams adopts the expla~~ation of this shift suggested by H alm, who believed it to be due mainly to pressure, the effective reg~on of abs~rption at the limb being supposed to be at a lower levol and there- fore h ~ g l ~ e r pressure than st tho ceniaa of the dirk, owlng to tho relatively longer yahh of tangential rays in the lower levels of the rovellslug layer compared with the rays passing normally through the solar atmosphere, as a t the centre uJ the disk.

As an alternative to this theory we wight suppcse the shift6 to be due to motion in the line of sight If the gases are ascending radially all over the sun w i t h tbe rocluired velocity, t,be relative shift to the red a t the llmb as compared wiLh the centre of the disk would result. This however is ruled out because we find a, descendmg, not an amending movament, when the positions of ihe lines at the centre are directly oornpartld with those from a, terrestrial source. If tbe shifts are the rewlt of movement then they can be explained only by a, motion pa,rall.llel to the solos surfi~ce, and directed away frorn the earth at, all points of the solar circumference. This snggests that the earth ltself controls the movement, exerting a

repelling aotlon on the solar gases. Obviously the pressure tlieory presents a much more rat~oual explanation of tho phenomenon than the

motion theory; yet there are difljcultiea in accoptmg tlhe forqar which have not been in any way lessened, but on the contrary hayo been largely increased, by further research. An initial d~fficulty which appeals to us is tho t&senoe of any evidence ol shading or indefinite edges on the red sides of tho lines near the sanjs limb. If the photospheric light coming from the limb passes t hough suocessive lagers of diminishing pressure, one would expect the absorption to begin gradually at tlho red edge of a line, especially as the absorption would be weakest in the lowest and Ilottcst layers where the pressure is greatest, and would increase as the rays cntered the cooler regions of less pressure. The red sides of the displaoed lines should therefore appear indefinitely bounded, yet no trace of any such effect is sppsrgnt. The lines O E the limb spectrum are broader tLan those of the centre spectrum, but they ere sharplyibounded on both red and violet edges. Perhaps the strongest argument in favour of the pressure theory which h h n s gave in his paper is based on the relatiyo shlfta of the iron lines which are most and least affected by pressure. B e found that thelmes most a,ffeoted by pressure gave the lilrgest limb -- centre shifts, and be also argmd that the large increase of shift with wave-length for the iron lines pointed to pressme as the main factor in the case,

We find it difficult to accept these results, because we consider that the relativo shifts of different lines at the llmb have no part~cular meaning when determined by reference to the lines at the centre of the diek, for these latter have shifts peculiar to themselves, and our measures show that the absolute shifts of the line8

* C____ __--.. - - -̂ "- . ,

* Aatrophysicfil Journal XXXI, 80, 1910.

a t the lirnb referred to a terrestrial standard sliow no relation to pressure shlfts, and further, the absolute shifts do not increase .wit11 the wave-length.

We have hscussed in Kodaikanltl Observatory Bnlletln Nos. XXXVI and XXXVIII the shifts of iho lines at the centre of the disk compared with the aro in air, and consider that our results clearly show tlint. pressure 1s not concerned i n the general displacement of the solar lines towards tlie red. A sinell pressure

effect is nevertheless traceable, but this indicates a pressure of less than one atmosphere in %he r egon of Iron absorption as observed at the centre of the disk.

This result aeems to argue against any large pressure effect a t the li~n'b, such as would be deduced fn3111

the shlfts hmb - centre.

Dctt~nmiry ation of limb shijls.

I n determliiing the abso l~ t~e shifts of the lines at the limb we have slrnply combined our rneasures o f the lumb - centre shifts with the oerrtre -- arc sl;lfts, the algebraical sum of the shifts representing the al,so- lute or 11inb - arc sh~fts. A number of direct measures of tlie limb - arc sllifts wa1.e also made by one of

us, aud these, so far as they go, confirin the indirect determinations. The spectrograph employed has already been described in Kodalkanal Obuervatotory Bullot~o No. XXXVI*

Fo r the limb - oeutre shifts photographs are obtained by means of a reflecting device pllaced in front of t l x e * spectrograph slit. Witli this apparatus silnultaueous expoeures are made with light from the centre of t l l r *

disk and frola poillts one-tliirtletil of the s~xn's radius inside the limb at the opposiie ends of a diarnuttar, The spectra form three contiguous strips on the plate each about 1.5 mm. in width. After tho espos~lrr~ on the sun, an exposureis made on the iron arc to impress the iron lines on the plate outsjde the solt~r- spectra : these serve to determine accurately the inc1ina:tioo of the lnicronleter thread t o the spectruxr~ I i i l t b b

In measuring the displxceilzents, biit are not used to deterinllle centre - arc or limb - aro shifts on thtt-t* plates. In this way the totttl fihlft west hmb -east lunb due to the solar rotatl~on may be acouritl,t*ly deteru~inecl. a s ~ve l l as the limb - centre s11ifts.

In table I: w e give a, list of all the iron lines of whlch we have measures of both limb -- contre sllilts :t r t $1 cenlre - aye shifts. The dgebraical sum of these given in column 6 represents the absoluto fih~ft of t l~u l i l i t + - -c

near the limb wlicn compared wit11 the iron arc in air at 580 ~ n m . pressure, tho normel prossurtb ; & a Kodazkeiial.

A (Rowland).

-___-_

3895 803 389R 860 31303'090 3906 628 3020.410

3923'064 3935 790 3028 075 3930 450 3B31269 3935*965 3837 479' 3948 925

Nutlibor of man,sr~ros. -----.-

aentre.

2 4 4 5 -L

4 3 2 5 i: 3 3 3 4 6 8 6 6 2 2 1 1 1 1

Inten- 'ltJ',

-__---- -

'7 7 10 10 10

12 5 8

h 11

4 b 1 4

4 1. 4 4 1 1 1 I. J. C 1. 4 3 1 l. 3 1 1 3

-

Sum

--

3950.102 1 5 3956 81'3 (3 3966.212 1 3

8 1 2 3 4

Bemarks

-

3969 413 3977 891 3986 321 3098'205 M05*408 400Q'864 4022018 ' 4045*976

a - arc,

A/lOOO I- 14.4 + 19 3- 1'7 4- 11 4- 15

-I- f 5 + 18

4- 6 $- 6 -I- 4 -k 4 3. '7 + 6 + 7 + 8 f 9 + 14 4- 8

Llrnk, -

Kodalkanal. --

AllOOO. - 3 - 4 - 2

2 :

+ 2

10 6 a 4 7 3 5 30

+ ' 6 + '7 ... +" 6 ... 4' 8

+' 9 ...

centre.

Mt TYllson.

A11000

<' 6

,.

+ 1 5

+ 5 + 4 + 4 .

f 2 f 8 f. -$ 3 3 + 8

4- 15 . -I- 15 -5- 17

+ 21 + 12 4- 20

+ 11 C 11 3-8

,..

3 plates give zero sh~f t and one plate - 6 (11nzb - centre)

+ 16 + 8 1- 13

0 + 6

4- 18 + 6

+ 17 -I- 3 C 1 4

4- 5 0

f 8 - 5 0 + -6

+ 11 + 7 + 16 f 4 3- 12 + 14

PC)

/ 3

TABLE I.-~~IFTs OF 1 1 1 0 ~ IJINES-CO~~~.

A (Ltowland).

1062 599 4063.759 4071'008 4076 792 41118 703 4127.767 $1 34'840 %140 089

41 1,4'088 4147*836

4J54 567 4175'806 4181'9 19 418712041 41191 595 CS~2*108 4220 509

429'7'G0(3 4233'1772 4236 112 4860~881 P ~ C O 640 42'7 1~ 427 l'!W & 4282'565 4308.081 4315 862 -A325 !)39 4337'81 M $362*130H 43639'041 437ti-107 4383'720 441)4*!127 4415+2U3 4487.4162 1

Inten- sity'

3

1 1

Sum Remarks.

4- 8 $ 6 + 7 + 6

-I- 1 -I- 8 + 4 -t- 14

4486.785 41442~610 4498'865 4447.898

A/lOoO. + 12 + 14 + 11 + 10 f 9 -1- I@ 4- 8 + 11

N~lmber of measnros.

Llulb con tre. -- 1 0 - arc- - arc..

Lnnb - --

Kodalkand.

-I- 9 -1- 13 + 13 4- 18

3 I - 8 0 -I- 4 3 4- 11 6 $ 4

A/lOOo. 4- 4 3. 6 + S $. 3 4 - 3 f 5 + 2 (+ l2)

-I- 7 '7

+lo

+ +I1 l a i- 10 4- 5 4- '7 + 8 f 8 4- 31 4 12 ... ... . a

... ...

. . ... a . 7

.vb

. . ,.,

-I-' 10 + + 8 $ 7 4- 9

centre, -- ----

Mt. Wilson.

A/1000. 5 -I- 8

2d 3. 8 15 + 3 4 $ 7 5 $ -8 4 3- 6 5 -1- li G - 1

15

2 4 3 1 2 2 2 ... 3 ... 2 4 I; 3 Ci 4

2 3 3 2 1 3 3 4 4 4 6 4 3 1 2 5 41 2 4

I

1 i : 1

1 i

4- @ -I- 7

$ I- 4 + 3 - 2 + lo -1- 5 4- 6

0 4- 1 4 4 + 8 4- 8 + 2 4- 5 - 1 4- 1 + l. + 9 -1- 8 -I- 1 - 1 f 2 4- 5 4- 1 + 1 4- 4 4- 6 -I- 2

441544552 446lS8 1 8 /4GO4727 4194738 $508'485 4515*508 '

4522 802 41528~798 4631'327 4548 024 4549'642 4556 063 4584.018 4&92+840

A/1000.

3-' 8

.+.

+ 8

Limb -- arc direct mea- sure.

].limb - arc dircct mea- anre.

4! $" 7

4 + 6 5 f 6 5 -f- 7 6 f 2 6 + 3 8 Jr 3 3 + 4i Jr 4 1 6 + 7 8 - 7 8 - 3 10 + 7 6 + :,

15 - t i 5 + '7 6 -+ 10 4 j- 9 8 3- 2 5 + 8 4i + t; 41 + 12 0 + 4 15 -t- 1 20 - 1 8 -I" 3 5 3. 10

2 2

3

3 6

3 -I- '7 $t + ? 5 + 8 6 4- 6

...

... ,.. a , .

4- 5 -I- 8

4' 8 .,. -1: 4,

+' 6 + 7

+ ' t i -1 3 f a + tj

4' 5 ... .., '

(4- 2) + J "7 + '7 I::

+ 1% 4- 141 + 20 + 15 - 1 + 14, + 8 C 14 + 12 + 12 + 8 + 12 + 16 + 14 -1- 12 + 9 i- 0 3. 6 f 6 $. 6 + 16

9 i- 3- a f 8 3- 9 4- 11 + 9 4- 12 + 13 4- 11

4 3 8 8 5 3 2 8 4 4

+ 2 + 4 -t- 7 -5 2 $. 11

( 0 )

' + s - 1 + 4 + 6 -I- 7 "r r 9 d- '7 -4- 8 3. 9 -1- 10 -/- 6 +- 4 + 11 -1- 11 -b 9 + 9 + + "

Unsharp line in arc

Llmb - arc direct mea- sure.

I

p Fe m short ere p Ye in short are p 1'0 in short @TQ

p Fe ~n shortr arc p Pe in short arc p Fe 1x1 ahort arc

: I 2 2

..* J;." 4 4- 8 4-41 ... ...

+ a 4

$ 7 . + 4 3. '7 3. 41

f 6 + 7 f 6 + 9 3. s 3. 6 Jr 4 .,.

.,.

... ...

4803*12tj 460'7'831 1 H 4619.468

+ S - -I 11 + 9 + 5 + 15 + 8

+ 9 1- 7 - 3 + 3 + la -I- 8 $ 9 4- 14 + 18 4- 18 + 13 -+ 13 1- 10 + 23 4- 15 + JO 4- -k 15 + 13

I 1 2 3 4 l.

1 1 1 3 2, 3 8 2 4 7 Ci G 7 2 2 2 3 4 3

S 2 3 .,. ... .". 4 5 4 ., . . 4 3 2 2 2 3 2 1 1 1 2 2 1

. . . ..

. #.

1-A

4626 227 463'7'(585 4638.193 4647*(il'7 4654 6'72 4654'800 5667.626 4679.027 470'i'45? 9831779 4736'963 41787 003 478Qq8 49 4869 928

3 '& 41 1 1 1 3 2 2 1 1 1 2 2 1 1 1 1 1 2 I I I 1 1 1 1 1 1 a

5 Ei 4 4 4 5 4 6 E; 4 6 2 3 4

--- - --

I i Limb - contrc, I (Ro~rlnnd) rnten- -

I 0 -810 S1tY' Kodn~ksnal. Mt. W~laon.

Bum. llernerks ( Llznil- 1 centre. (3 - ""

p Fe in short alc p Fa ~n short arc

Faint and diffuse line In I &PC.

b 4 1 ' 4 1 1

p Ee in ~ h o r t arc 1 1 1 4

0 - short; tlrc = - 14 1 4 Q - short sra = - 11 2 111 a - R ~ O L ' ~ BTC = - 1. Z 4 a - shorl nro = 1 2 8

I 3

I a - sltort, am = - 9 I, 0 - sho~t; nro = - 7

I p Bo 1x1 short arc

I

Fain(; and dlll'neo rn arc

I n Porili~ng bhis t&lc we ]lave givon Dr. Adam$ values of tho 11111b shirk, zn column 3, iii~der t110 honc'l- ing cc Monnt Wilson" all2 these values have been used in Eormmg column 6 fo11 the lines wllich we havo not yet ~neas~~red. For lines ~easured . at 110th obso~va-tones the mean of tihe two clcterminations or limb - centre has been used. For most of the cornmoll liires the agreement between Mount Wilson end Kod:~,ilcounl is excellerrt, but there are two or thl?ee ~narked dlsorapanoies such as the lilies 3920v4?0 and 42'71~934, wllioh acoord~ni to our measures are sM%-x3 towards the violet instead of towards the red as in Dr. A d a s ' determmations. The vallues in column 6 for these lines must be subjeat to considersblo uncertainty.

AS regards the relative accuracy of the different determlnntions, we glve in the l a ~ t two columns of table 1 the number of measures on wl~ich each value depends. In most cases several plates taken at different dates and in different solar latitudes have been used for each determlllatio~i of hinb - centre or sun - n1.o shiCt. There appear to bo col~siderable systematic variations in the amount of the shiftsglven by different pl;~,tes, for both limb -- centre and centre - arc, so that vali~es obtained from one plate only are sr~bjoct tc t l ~ s vvoiation from the rneilll in addltlon to the greater accidental error of rnoasurement.

The gun -- arc measures are the same as those given in godaikaoal Observatory Bulletlll No. XXXVI, with addi$lons and improved values for many of the lines obtsined from later measures.

Lzmb eh i f t s in relation t o fhe inte?zsity of t h e l iraes.

The first point to be noted in these measures is the relstlon to intensiry. If the lines are grouped zlcoording to their intons~t~y in the sun tlie following mean results are obtaitied :-

111 this tn,hle all the llnos of table 1 are used in forming the means excepting t v o . Tliese are the llnas at X 5162-449 and h 542/1!.290, wliich give very anoinalous shifts due to l3ecnliau condit~ons in the arc.

-- - - --

Many otlicr liner :~,lso are probably :~i'Tcctccl by the arc conditlone, os-pscldlp thoso whicl~ wlden unsymme- trically undor p r e s ~ u ~ o , I)al as xf; is not a l prosonti posslble to classify ell tlis 11iie.s of tablo I i t has seomed

-

Intansitij . I N~irnbar of l l n u ~ I Lnnb - ceotr o 1 gentre- aro I ~ l m b - arc. - ---- -- -- - _ _ _ _ _.-_-_-I ----

best to tako g e ~ l e ~ s l means, exclnding ouly those above-m~ntionecl lines which give enoztmons centre -- arc sLif ts which i1,re 1~1101ost ~~~'titbinly I I O ~ connect e(1 wit11 solar conditions.

Although indiv~tlaal lincfl for each ~ntonsilly give very diff crent shiEts for both 1in.h) -- cell tre and ct!n.lvc~

8 and over 7 ant1 B 5 4 3 and under

- arc, t110 relation tlo iutcinsily is wcll ~nitrlcr~d iii t h o means. '1'11e limb slllfts ~ncrease as tho intensity

-(- 0 0023 A -/- 0 0047 4- 0'0074 J, 0 0073 4- O*Oo(j6

24 35 2(i S 4 20

dirrlinxshes from the st,llongoet liues tlu in torltjliy 6, wllilst illlo cont~*s ~liiEt~s docroa~o avar tl10 same rsage. Below intensity 5 $110 slnltls two iloih~ly co~lst~,~,nt,. If the arc lii~ns tnsy be oons~dored i,o be in thoir normal posiiitions, tllail the roltltion t o mbsnrtity of l , l l~ liiub - contl-n shifts is only an z~ppwent one and is really clue to the varying sllilts of tklo lirirs altl tlle c~11t;ra of the dlsk, fox* the :~~dclcd duf t s given in ilm lest cok~ma 'C linlb - arc" show prac;tic:a,lly 110 rolatioir to ~ntonsjty.

+ OGl07 A -1- 0 0063 4- 0 0037 + 0 0051 + 0.0038

Itelalion bot~)o(!th limb s 7 ~ 4 f t ~ and prasszlrs slzifta,

-1- 0'0139 A 3. 00110 + 0 0111 + 0.01241 + OV0lci4

If we gronp fho botal ~Infta, li~lib - : I I ~ C i ~ t 580 imn, piTcsunrcr aocor(1ing to tho amount oE t1:e nrossnre sbifts, tho following ~esnl ls are ol,l,r~iuod :-

It is here Boon that the TtlQiUl limb nhif t~ (10 n o t incron~e i l s clo the pressliro Khifks in p n ~ s i n ~ from the more

C - - -* - - " -- ---- .. *- - ------ -- --- --- ----

refrltnglble to bhe less rehtmgiblo groups or linas, and thzllt the lines mosl affectecl by pressure are loash shiftod at the limb.

If t h e limb - arc sll ift~~ a1re cor1*ecirod SOY tho dof ect of pressure all Rodnlilc~nall horn llorrnal and the results for the two spect~al regions axe averaged, the limb -- t-llrc sl~ifts fox nor~ns l pressaxe beoome 4- 0.0065 A f o r the lines most itffoctecl by pressure, and -+ 0901 34 A for the lines least affected by p r ~ ~ s u ~ ~ r ? . If tho wave-

X,og$011,

lengths ol the arc linoa aro upp posed to bo analroctad by other oondrtions, this would moan a total pressurc at the limb o f 0.24 atl~aospbere only, bud i t is very pestxon&lo whether the a m under the conclitions of our experiments doe^ give '' normnl " wave-lsngths for many of tlie lmes. Dr. Royds has shown that lmes whioh are un~ymmetrical in the nrc (i.e., the majority ol lines with large prossure sh~Ctn) arc displacoll in the sbort arc compared with +be long arc." This shift is not dne tJo prossure diflerences or to motion, bui appears iobo

N U ~ ) O X of l i r l ~ ~ .

s density effect, and there ore reasons fox behoving that in the long arc (5 t o 7inm.J the conciitions, although approaching more nearly to solar condit,aons than m the short arc, are still iilr I~.om being the same as in thc reversing layer, where the gases alppear to be of tho last degree of tenuity. Many of the arc lines therefore

Troasclra ehLft por alniosphoro.

which we hayo compared with the sun, and especiallly those which give abnorlnal pressure shifts, may be

MGttrl &iffi limb - aye,

---- --..... ---------- ------ - -_l___l_~- _ - --- -----------..---_ -__-- _- -- * Kodailranal Observstory B~zllatin No, XXXVIII.

0

affectled by this density slilft which would in general tend to r~duce the sun - arc shifts for the lines most affected by pressure.

1 t woulii seem probable therefore that this density shift u ~ a y partly account tor the low values obtained for the lines most affected by pressure. Even i f we concede that the whole difference of shift between

the lines most and least affected by pressure (which would amount to about 0.007 A when the ai-c is at normal pressure) is due to the density effect, the figures would imply an absolute presswe near the limb of one atmosphere only, and the difference of prossure between limb and centre of disk about one-fourth of an atmosphere.

It is probable t ha t a better knowledge of the effect of presaure at the llmb inay be gained by a comparison of the shifts of o n l y those lines whloh widen symmetnoally in the arc under pressure. I n our

list there are only 4.7 line4 which are lrnown t o be of t i us character, and t h e pressure shifts of these lines do not vary very widely : tbey may nevertheless brt separated into two groups comprihing the more alld the less affected lines,

In table IT the shifts of t,he symmetrical lines are set out in detail with the mean values ln Angstrom units a t the foot of eaoh column. The pressure shifts in column 2 are from the tables of Gale and Adi~rns .~

TABLE IT.-MEAN SHIFTS OF SYMMETRICAL LINES IN RELATION TO PRESSURE E ~ ~ I F T S . ---- -- -

L ----- -- - ----

I A

I

8.-Lznes most aflected b y pressul e. 3903.090 39(i94413 4045.975 413llt 840 4144 038 4202.198 4271.n3.k 4337 216 4869'041 4383 720 4 b54 552 4,531'327 5227 362 5269.723 5328 236 5328'5'21 5405.989 542D19!1 54341 '740 544'7'1 30 5455 834

Me:~ns . .

+ 22 2.2 23 27 20 25 22 27 23 27 23 29 31 27 29 26 2'7 20 2'7 3 1 29

+ -0284 .L

-

+ 15 18 14

tl 22 15 9 13 23 10 13 12 G 8 18 15 18 15 18 15 27 . -------

+ #Ol49

- 2 I +15' + 4 14

G

8

* Astrophysical Journal XXXV, 8, 1912.

88950803 3899.850 3006 628 3920'4110 3923'054 3928 0'75 3930 4150 3956.819 307'1 891 4005 408 4OG3.769 4071'908 4282.565 4308.081 431 5.262 41325*93Q 4352.908 4376 107 4404.927 4415*293 ths27.182 4443,365 4461'818 4486.927

Means ...

I o 8 12 1 7 7 3

- 1 + 3 7 9 9 10 10 8 - - - -

+ 005s

11. 9 5 11 9 G 5 3 Q 15 8 9 6 8 5 19

+ .0091

B,-Ltnes Zeast affectetl by pl-essure. + 11

12 11 10 11 12 13 Ith 17 19 20 21 22 21 19 20 1'7

-I- 141 19 11 15 14 18 14 4 6 8 G a

8 7 8 9

10 4

- 3 - 4 + 4 2 7 2 2 '7 '7 8 8 3 '7 10 9 3 6

f 11 15 15 17 21 20 16 11 13 16 14 11 14 15 18 13 10 15 8 16 13 13 14 20

-.-- _-- + *OICf-i

I ---

18 21 18 17 19 15 1s

-I- .u1641

- 1 Y + 3 TO

9 '7 8 - - -

+ .0049

12 3 4 5'

12 -- -- + *009'7

The mean total shift limb -- arc is practically idei~tic~il for the two sets of lines, although the mean pres- sure ahlfts are in the ratio 1 : 1*6. This would of course imply that the total pressure a t the limb is about the same as that of the air at Rodailcanal, or three-fourths of an atmosphere. I t must be said however that some of the values of the limb - centre shifts are very uncertain and need remsion, especially those which give low values for the total shift. If we eliminate from the lists the five lines whioh yield total shifts less than 10, the mean shift for the r.x~ost aflected lmes would be + 0.0165 A, and lor the least affected lines + 0-0149 A, a difference which would imply a total pressure a t the limb of 1.27 atmospllero above the pressnre at Kodaikafial, or about one atmosphere above normal prossure.

If the total pressure at the limb were of the order of 6 atmospheres, as might be deduced from the mean total shift of syrnrnetrioal lines which is almost f 0.015 A, then there should be a difference of shift between the lines most and least aff@cted by pressure of 0.0075 A, a q r ~ a n t i t ~ which could not fail to appear in the mean results.

I11 this discussion we have *aLen no account of differences of level of tho effective regions of absorption for tlhe different lines. The m~ml is that the limb - arc sshlfts, as we have shown in table 11, are no^ appreci- ably afleoted by dil-re~ences of intensity. I the ~ntensiides of the linee near the llmb are related to level in the same way as St. John has found for sun spots on the iiisk, then in the region of iron absorption level appears to havo little or no e f l~c t on the cliuplacoments. For the symmetrical lines these are remarkably constant, not only for lines of grea.tly differing intensity but also for lines in very different regions of tho spec tirurn.

Another ruason for believing that tlm limb shifts are not due to 17ressure alone is furnished by the dls- placements of tholre iron l~nes which wit11 increased pressure are shifted to the violet (Mt. Wilson group e)

in contrast to the majority which are shifted to the red. For these lines we have only measures of limb - contre, the total shifts being unknown, but if the displacements at the limb of the majority of lines t o the red is due to incrcnaed pressure alone then the lines of group e ought to be shifted to the violet. W e have at present only two photographs of limb alnd centre containing lines of group e but their evidence is quite decisive, for each line is displaaad to the rod. These two plates include the regions X h 5365 to 5455 and XX 5585 to 5638, and the average displnoomont; limb - centre is given. in tho following table, together with the pressure hilt per atinosphere according to Gale thud Adams * :-

Pra~lsure shift Mean shlftj pcr ntmosphera, L~rnb- colitro.

Lilies dis1,lnccd slz'g71,iZy to tho red by increasecl premsnre , . . ... i- 0tO04 A. + 0.008 A. (?Lh. 5371, 5397, 5405, 5429, 5438, 5447).

Lines displaced yveatL!j to t 7 ~ e r ~ d by increesod pressure . .. .-• + 0,023 A. -+- 0°008 A. (hh 5393, 5565, 5509, 5573, 5576, 5586, 5603, 5615, 5624,

5688). Lines diuplacod groatly to the violet by increased pressure ... - G*018 A. + 0.0~5 A.

(XX 5365, 5383, 5411, 5415,54244, 5555, 5565,5598).

Tho lines shifted to tlie violet by prossurva havo, it is soon, smaller displacemanbs to the red than tho other lines. Assurning that t h ~ s relative shilt is due to prosaure, the deduced pressure at the sun's limb is, sccording to wliether we compare these lines with the first or tlie second group in the table, one-seventh or one-fourteenth of an atmosphere above that at the centre oi the disc ; either of these amounts is com- paratively insignificant. I t should bo mentionorl, llowever, that the relative shlft of these lines compared with tho other hnes is not necessarily due to pressure because we do not yet know the displacement a t the centre of the sun, for m the arc the wti,~~o-lengths are not normal.? It may possibly bo that the smaller

displnctomel~t s t the limb of the lines shifted to the violet by pressure will be compensated by a larger displacement at tho centre as was shown above t o be the case lor lines of different intensities.

Limb shqts ivr rslation to wave-length,

Our results do not confirm that large increase of shift with wave-length whioh Adams obtained with a much smaller number of lines. It; is true that the relative shiEts limb -- centre tend to increase to wards the red end of the spectrum, but this is counteracted by al decrease in the centre - arc shifts, so that the total

--- -- - - -- - -- --- ---.---

+ Astrophysical Jovrntll XXXVII, 891, 1913. $. See Kodaikanel Observatory Bulletin No XXXVIII.

sbift limb - arc remnins sensibly constant whether we take all the lines of table I or only the symmetrical lines. These results are shown in table TI, where the lines of table I are grouped in three differen5 regions of the spectrum, and the shifts for each grosp avereged. As in table I1 the lines 5162,849 and 5424*290 have been omitted,

TABLE V I -MEAN BHIBTS I N lGELATION TO WAVE-LENGTH.

Number cf Lirn b - ( lnll 1 l e g l o . ccllic, 1 C c ~ ~ t r e - m1o. 1 L~n~b-a ro , 1

I ( A ) A21 lznes emcegt t11uo. I

I ( B ) By mnzeti dcal le~aes only I

The symrnotrical lines llavc beon grollped in two regions only, owing to the small numbor of line8 available. Botb table A n a i l table R exhibit tlie sane characto~istic inverso relation between the limb shift ancl tho centre shift which resultls in sensibly constant values of the total s111ft, 01' limb-a,ro shift. Tho symmetrical lines glve higher values, probably Erorn the absenco o f the anomalous shifts caused by pec~~lirtr conditioils in the arc

If the total shifts at thc limb were dne to presslwe wo should cxpact to find a marked increase in t h o mean values for tho less TeCrtlllgiblc lines. 14or the symmetrical lines mean prassuro shifL of tho 23 more ~ ~ f r a , ~ g i h l ~ lines is On0025 A. per atmosphere, and oI the 22 less rcfrnngible lines it i s 0*0030 a pel4 atmosphere. Bnt the less refrangible lines represent lower levels in the reversing layer than. tlio more refrangible lilies, so that the mean limb sh~fts should increase in a greater ratio than 25 : 30. As they do not increase at all but tend to be srnder for tho less refrangible lmos, we conclude that pressure is not concerned in the general shift of the lines towards the rerl.

Tbe sh~ l t s oE the cyanogen btt~lds at thn limb i$nd at the centre of llle disk glve ~dd i t io l~a l evidence which is strongly ngaillst tho view that pressure 1s the cause of the 11mb shifts, for the bands or flutiligs are not appreciably affected by pressure, and tlierefore tba slllfts fozuld between limb a,nd centre and centru and aro can only be expla,iiled by rnotlon in the line 01 sight.

In the paper already cltad Dr. Adnms ~*eSeru to hw measures of the cyanogen flutings at X 3883 auri

X 4216, for mhich he h d s a. s~uall positive s l ~ I t limb - ceutre. 131s mean result for several platov and for 14 bands near 3883 is -1- 0.002 A. A few rnewures of these bands 111lve also been made at TCodaiiLallal with results whloh confirm Adams' mea,qures, showing relative sbift to rod at the hmb whicll i b notably smaller than that of the illon or titaninm lines 'L'he mean shift limb - centre of the best defined bands is +- 0.002 A from 5 plates.

The values vary considertzbly from pla,te to plate, and Adams considered this shift to be due to ascending movements of the cyanogel~ at tho centre of the disk, causing a violet shift at t l ~ e centre or a,

relattive shlft to red at tho limb. Obviously the detlermination of the absolute shifts of the ON bauds at the centre of the disk compared

with the bands in the carbon a,rc is of crucial rmportance in testing Adams' hypothesis and in oonnection wlth llmb shifts generally. Accordingly we have made very careful sets of measures of the bmds near 3883 in eight comparison spectra of tlle carbon arc and the centre of the disk, using carbon termii~als or one terminal carbon and the other iron.

I n table VI I the mean rasults are giveo for two sets of measures of 4 plates, each by different measurers, the initials N, N, R, and E at the heai of oolumns 2 and 3 indicate the measurers Nagamjja Ayyar, Narayana Ayyar, Royds and Evershed reslieatively.

A (Rowland). N.N.R. E.

(4, plates). (4 plates).

3863 533 . . . . . . . . . ... ... I I a ... 3- 10 ... 3864438 ... ... ... ... . . . . . . ... + 10 + 7 3876'448 ... . . ... ... ..a . . . . . . - 6 + 1.

-331 ... ... . . . . . . . . . . . . . . . . . -796 ... 3879 {.SS1 . ." . . # . I . .me ... 46.5

... . . I -* . ... . . I .* I . .

... 3980 { 1:;; ... ... ... ... ... + 5

3ESl {'E: ... * * a ... .. . . . ,.. + 8

,-- --- ... Means . . . . . . . . ... . , . ... + 0.0052 A + 0 0045 A

Elach set of 4, plates was pllotographed independently at different dates and required differelit corrections, to be appliecl for the ethrth's movements ralntivo to the sun. The lines or bands chosen for measurement were thoss most clearly defined in the sun nnd free from evidence of any lnterferenoe by lines due to other substsnces. All of thorn are nssigneil by Rowland to 0 oiily. As the two sets oE measures were made quite iudependcnfly rather a diSEerent seXeotioii was made, but the agreement oE the valuos for the cornmoil lines 1s as good as conld be expected coii~idering the breadth of many of the bands of which the component linos nue only partially resolvod in the photographs. There is a goueral agreement also in the differences of shift Tor tho di1Teron.l lines or bands, showiug that these differences are not due t o errors of measure- ment, but are possibly duo to tllc pocnliar condit,ions in the arc, which as we have shown tend t o produce srna,ll, or eve11 11eg;lt,iv(3, SUZI - arc shifts ia tho case of some of tho iron liizes. The two lines 3876*$48 end 3877.481 wllieh give uegative shiCbs in $110 firsi set of rnoasurof; and smell positlvo shifts i r i the seco~..d set would ~aetn to bu al'tectad in this way, and $he rliflerei~ce i t ) the maasu~es may be due t o a sho~ter arc having bean useil ln t l ~ o first set of plrlotugraplns t11an in tlie second.

130th sssts of motksurss agrce ili showit~g a ganoral shsft of t h e ON bnnds t o the roil at the centre of

tho disk amounting to $ 0gC105 A, a veluo whi& is almost certainly too small, as the general eiEoct of the arc conditions 1s to ~ * e d ~ c e I ~ o S U I ~ - BTC shirts. If the lines gzving nogstive shifts n,re excluded the general mean becolnos 4- O*QUG4 A, ccluivalent to a, motion of descew* on tho suil of 0.50 kmlseo. Thls is the same ordar oS volooily as we have fount1 for 1x011 in the reversing layer, being intermediate between the velocities obtained from thd high level find low love1 iron, lines. According to St. John tlze CN bands rcprescnt a rathey low level in tllo solsr n,tmosphere, or about EL mid-lsvsl 3n the revorsing layer, so that the velooi~y found above is in strict lla,rmony with oar results Cor Iron.

As the opallogen gas in the s ~ m is descending st Lho centre of tho dlsk and not ascending, the limb - centre shifts to tho rod cannot be explained as due to a violet sliift at the centre. If the limb -- centre an2 centre - arc shifts arc added, the total or limb - arc shift amounts to about + WOO8 A, indicatiiiig a, reoession at t l ~ e llmb parallel to the solar ssvrface oE 0.62 km/seo. But this movement of recession at tlm l i sb suggests s simllar ~iiove~nent for iron and other elements. For iron the total shift at the limb is as we have shown + 0*015 A for the symmetriaal lines, and almost constant for dl@~rent spectral regions. This would imply velooities varying from 1 4 1 km/sec. at 4000 to 0°.9 km/sec st X 5000.

The velocity interpretation of the lirub s h i h of the iron liues seernb forced upon us by the C N skifts, and involves very rernarkkble consequences, for we have to suppose the iron and other gaseous subst~iucos receding from tho earth at opposxte points on the sun's hmb, at the centre, and by inference r~ll over the disk and all rollnd the circumference, the velocity inoreaslng from the centre of the disk to\vurds the lllnb where, unimpeded by the denser gases d t the base of the reversing layer, it attmlns a veloclty of about 1 kin/sec.

But a movement colzstant in direction relative to the ea?*th and inalntail~edl at all seasons of the year means an earth eEect-wn alahu~;l repulsion of the solar gases by the eautlz, aud not apparently by the other planets.

While fully appreciating the absurdity of this idea we feel that tliere may be some jostification for i t in the appareut influence of the earth on the distr~bution of s~xnspots on the vl5ible disk alid of the prominencetr onthe east and west limbs. That the earth exerts some sort of iuflaence on solar phenomena which is not shared hy the othe;,planets is s, fitartlhg and perhaps rncredible nupposition, but the f;iots wh~ch have recently beell disclosed in this connection have not yet been explaluecl otherwise.

There appears in fact to be no dteruatlve to illis earth-effect hypothesis, at any rate w ~ t h regard to the cymogen sh~fts at the limb, u~lless w e assume some ca,use for llne slnfts olher than il~utioti or lwassure. According to the '' TheoTy of Relahvity " of Emstein, tlie sun's gravitstlo~lal field should diminlsll the fre- quency of the light emitted, and the meal1 sl~ift to the red found by us a t the oentre oE the disk agrees very olosely with the theorehical gravitational shiCt calculated by E F. Freuncllich.* Rut the large varrnbiolls of shift from line to line at the centre of the disk whlcl~ depend ina~nly on intensity, and the constancy of the limb shifts for different wave-Jengtlls are facts which would al3parently offer serious dificnlties to tlns explanation. One of 11s has found t that the displacement at, the centre of the sun'^ disk decroases rapidly wit11 depth; i b would p~*esumably vanish at a depth a little lower tl~arl tliet of tlie f n m t iron linw wh~lstl tlis gra,vita,tionnl force call only be slightly sillnller than at higher levels sucl may, ioclced, be lnrgor.

Su~nmn.ry.--T1lo maill results of t h ~ s investigation and the conclwions reacherl may be briefly recapitulated in the f 0110 wing paragraphs :-

(I) In stndyilzg the limb shifts zt is cous~dered essential to dotermine bbe total slllfb l ~ m b -- s ~ o

instea& of the relative shifts betwoeu limb and centre of disk as has Iseen done hitherto, (2) The total sh~fts of 130 Iron lines at the sun's limb have been determined by nddmg the limb -

centre shifts to the centre- arc shifts, and also directly in some cases by measuring $lie limb - arc shifts. (3) The limb - celltre shr f t s and the oentre -- arc shifts are Eouiid to be relatod to tlio inhcnsity of tlie

solar llnes in ax1 opposlta sense, tho formel* decreasing as the intensity moreases and tire latter ~ucreasing at about tlze same reto. 'Phe total or 11mb - arc shifts a~le therefore sppr oximn~tely constant for all intensit~es.

(4) The relatiou between limb - tlrc shifts aud pressuye shlfts 1s discnssed for all the lxlles witll. known pressure shifts. Grouping the lines into those more or less affected by pressure and into different spoctl~ol regions, i t is found that the more aEected liuos are much less shifted than thc less afkctecl. This result is believed to be partly due to certain yeculiax conditioris ill the arc whlch bend to reduce the sun -- arc shifts, and espeoxaily the slzif ls of those lilies most affected by pressure.

(5) Taking symmetrical lines only, vhioh aye presumably free from the disturbing effects of the ~ I Y - J

conditions, there is found to be no difference of shift berween groups of lines whose average prossure shifts diff el1 in the ratlo 1.6 to 1. This implies a, pressure in the effective reglon of absorptloii at the lirllb equal to t l ~ a t of the air at lCodailralla1 or 3 atmosphere, a result which may however be contiidorably modi- fied by further research.

(6) The large shilts of the symtnetrlenl lines at the hmb, amoontlng to 0.015 A towards the red, cannot be due to pressure because, if so, the difforent~al shlfts of the lines more and less affected by pros-

sore would be of an order of rnaguitude which could no t fail to appear in the measures. Moreover the pressure hypothesis requires that certain iron lines should be displaced at the limb to the violet whereas they are actually shifted to the red of thoir positions a t the oentre of the disk.

- * Phyaikslisohe Zeltsohrlft XV, 2, 1914. f Ko6aikaa81 Observatory Balletln No. XXIVI,

(7) I t is shown that clltliough the relative shifts, ltmb -- centre, of tlie iron lines tend to lllcrease toward the lotlger mave-lengths, tlze total shifts, limb - arc, remain remarkably constant when means are taken In three spectral regions betweeu h 3895 alld h 561.5. This colistaucy 01 slilft is also shown atrilringly by the s$rmmetrical l i n cs alone, and tells l~eavily agamst the pressure theory.

(8) The coustimcy of the limb - arc shifts for symmetrical lines of greatly d~ffering intensity shows that level is not au important factor in determining the limb shifts, if 1t m a y be ilssa~ned that intensity is related. to lave1 for lines near the llmb as ~t appears to be for lines in sunspots on the dlsk.

(9) Tho cyanogcn bel~ds are shown to bo displaced towards the red at the limb relatively to t h e centre of the dial;, and a t the oentye eomp~~recl wit11 the arc. As these bands are not slnfted appreciably by pressure the sliifts can only l ~ e oxplained by assuming a movement of recession from the earhh-lb descending lnotiolz on the disk and a movemen"l;pslrallel to the solar surfaca a t the limb. The total shift is : ~ b w t OaOOS A, indicntwg recession near tlle l i ~ n b of 0 62 km. per see,

(10) Tlie shlfts of the iron line#, amounting to a mean value of 0 015 A for the syrnrrzetrical lines, is ~nterp~otocl sirnllarly as due to a ~ocession of the iron vapour at the limb of about 1 lrm. per reo. A move- ment of recoesion froin $110 earth at the centre and over tlie entire disk also follows.

(11) The view rhnt tho solar go,ses are actua,lly repelled by the ea,rth receives some snpport fro111 ot1ier lli~es of eviilenoe, but :m alterrintive hypotliesis is conotdered, namely, that ihe suo's yi-avitationsl field ~llif~cts tlm wave-length of tlio light culltted, in slcco~dance with Ernstein's Theory of Relatlviijr.