LECTURE.-On the Theoretical Explanation of an apparent new Polarity in Light .
By G. B. AIRY , Esq. M.A. F.R.S. , Astronomer Royal .
Received June 14 , --Read June 18 , 1840 .
IN the Report of the Seventh Meeting of the British Association* , and in that of the Eiglth Meetingt , Sir EDAVID BREWSTER has given a short account of experiments which have led him to inferl the existence of a polarity , in the rays of homogeneous light , having regard only to the sequence of colours in the spectrum .
I am aware that Sir DAVID BREWSTER has announced his intention of publishing a more elaborate paper upon these experiments .
But as the leading facts of the experiments have now been communicated to the public nearly two years , and by Sir DAVID BREWSTER hiimself , and as the accounts have been sufficiently explanatory to enable any other person to repeat the experi:ments , and as various persons have in consequence repeated them , there cannot , I conceive , be the most trifling impropriety in making the whole subject a matter of discussion , experimental and theoretical .
As the experimental facts observed by myself differ fiorom those of Sit DAVID BREWSTER in a degree which , as regards the experiment , is trifling ( and might easily pass unobserved ) , but which , as regards the theory , is important ; I think it necessary to give the following brief history .
My first repetition of the experiment was made in October , 1839 , in consequence of a friend requesting my opinion upon a proposed theoretical explanation .
To that friend I irmmediately cormmunicated ( by letter ) the experimental results at which I had artrived , with a request that he would verify them ; and he did accordingly verify them , as far as the apparatus in his hands allowed .
These results suggested to my mind the general train of the theoretical explanation ; and the numerical calculations necessary for following it into details were inlmmediately placed in the hands of a computer .
The incessant occupations incidental to my office prevented mroe from evien looking at the calculated numbers till the mnonth of May , 1840 , when the remarls of another friend recalled my attention to the subject .
mention these matters tnerely for the purpose of showing distinctly that the observation of the facts was with me long anterior to the explanation , and that therefore no leaning in favour of the theory can have warped my judgment in the perception of experimental facts not wholly coinciding with Sir DAVID BREWSTER'S .
Indeed , I should never have thought of opposing my evidence on a matter of observation to that of the Father of Modern Experimental Optics , unless the reasons for disagreement were not only , in my judgment , well-grounded , but also unmingled and independent .
The facts , then , as I have observed them , are as follows : 1 .
When a prismatic spectrum is viewed , out offocus , bands are formed by placing a piece of mica of a proper thickness ( indefinite within wide limits ) to cover that half of the pupil of the eye , which is on the same side as the violet end of the spectrum .
2 .
No bands are formed with any th'ickness of mica if it is placed on the same side as the red end of the spectrum .
3 .
When the eye is too far off to see the spectrum distinctly ; then , upon passing the mica before the eye from the violet end of the spectrum , bands are seen advancing in the same direction over the spectrum .
4 .
When the eye is too near to the spectraum to see it distinctly ; then , upon passing the mica from the violet end of the spectrum , bands are seen advancing in the opposite direction over the spectrum .
5 .
If the eye be so far off , and the spectrum in consequence seen so indistinctly , that the ruddy portions are nearly mingled with the blue ; upon applying the mica to cover half the pupil on the side corresponding to the violet end of the spectrum , bands are seen very well defined , but narrow .
6 .
If the eye and mica approach to the position of distinct vision of the spectrum , the bands become somewhat broader , and near the position of distinct vision sometimes disappear .
On still approaching to the spectrum , the bands reappear and becotne narrower , but are not , I think , seen so distinctly as when the eye is too far off for distinct vision .
I know not , however , whether this may depend upon the practical difficulty of that part of the experiment .
7 .
Bands which are visible when the mica is on the violet side and invisible when it is on the red side , never occur when the spectrum is pure .
8 .
Bands are frequently visible when the spectrum is pure ; but in that case they can be seen equally well , whether thie mica be advanced from the violet side or from the red side .
The bands menetioned in No. 8 .
have been long since observed and explained by Mr. TALBOT* , and of these therefore I shall make no further mention .
The general outline of the explanation of the bands seen when the spectruimi is out of focus , is as follows : By retarding the light which falls upon a portion of the pupil of the eye , bands are foirmed in the indistinct image produced by any one kind of homnogeneous light .
2 .
These bands are not symmetrical with regard to the centre of the indistinct image ; and the extent of asymmetry depends on the retardation .
3 .
Therefore if light issues from a number of luminous points of such character that the value of retardation changes gradually from one to another , the position of the bands , as measured from the centre of the indistinct image of eacht luminous point , will gradually change from one to another .
4 .
If then these lumtlinous points be spectrally separated in one certain direction , the centres of the indistinct images will be separated in a corresponding manner , and the bands produced by all these luminous points may be mrade to coincide , and thereby to produce strong bands , in the confused spectrum formed by the aggregate of all the indistinct iiages .
5 .
But if the luminous points be spectrally separated in the opposite direction , the bands will be remnoved further from coincidence than before , and all trace of them will be lost in the confused spectrum formed by the aggregate .
6 .
W thther the retina be thoo near to the pupil , or too far from it , for distinct vision of the luminous points , the spectral separation must be such as to carry the points from which issues the most retardable light towards the side on which the retarding plate is placed .
The remainder of this paper contains the mathematical development of this explanation .
Let the wave of light , issuing from any luminous point , have , after passing the lens of the eye , the form of a spherical surface convetging to the centre of the sphere .
Let the radius of the spherical surface be c , and let the distance of the retina from the lens be c+a ( this assumption corresponds to the supposition that the luminous point is too far off for distinct vision ) , and let it be required to investigate the intensity of light on a point of the retina , whose distance , from the point defined by drawing a line from the luminous origin through the centre of the sphere , is b. Let x be mreasured from the centre of the spherical surface along that line ; and let y be measured in the direction parallel to b , x and y being used to define a point on the wave surface .
Then x2 + y2 = C2 .
There is no necessity for introducing another co-ordinate of the wave-surface , as its effect would only be , to introduce a constant mnultiplier in the result .
The distance from the point x , y , to the point on the retina is V{X+a2 ?
y -b2 } = V{c2+2ax-2by+a2+b2 } .
y2 If for x we put its value c^ ( as far as the second power of y ) , this expression becomes V{c+ a2 + b2 2 by y2 ; and , expanding to the second power of 227 y , and putting e2 for c +2 a+ b2 , it is b ( b a 2 .
e y-2-e3 +~ / y With the small values of b , of which we shall have occasion to treat , b2 will be much less than .
Neglecting therefore the first multiplier of y2 , and putting e for the distance from the point x , y to the point on the retina , bae eY 2ce eea ~2ce eb2 ab cB\ 2f2 ce +a =f --~+ cbe c be wheref =e +e If now we suppose ( as in diffractions generally ) that every part of the great wave , after leaving the lens , is simultaneously the origin of a small wave diverging in all directions , at least through a large angular extent , we may represent the disturbance of ether on the point of the retina , produced by the snmall portion by of the large wave , by by X sin ( v t)x or Xsin v t-/ + 0 / + } or sin s ( v t -f ) .
*y cos -c c -( + +cos ( v t --f).ysine -+ 9 .
a The limits of y in the integral must be the limiting values which determine the extent as far as there is no additional cause of retardation of the wave .
But if , by the interposition of any refracting substance with parallel boundaries , a portion of the wave be dretarded by the phase R ( expressed as an angle ) , then the expression to be integrated will be s ; sin ( 2r { vt-f+ ce ( Y+ r b2}R ) 228 or sinA ( Vt-fV )fcos { e a(+ )2 R } + cos ( v t.-f ) sin Ac e ( y + a-R } or si-)(v t -c os FC-R in 7r ( vt f cos. f ( + c1 9 .
sinac be c+\ )+ cos ( v t -f ) os FL .
Rsin A--y + sin R. cos ( y +- ) } mC0S-V-jjPCO5 .
yXcea/ yea where the limits of y in the integral must be the values which determine the extent of that portion of the great wave which is affected by the interposition of the refracting substance .
Suppose now that a plate of mica or other refracting substance is so placed as to cover a portion of the pupil on that side on which y is considered positive and suppose that its limits , neasured in the same way , are from g to yea indefinitely great : suppose also that the limits of the pupil are from y=h to y=+ h ; then the integral which is independent of R must be taken from yh toy =+ g , and that which depends on R must be taken fiomy =+ g toy = h , and the integrals must be added to obtain the whole disturbance of ether at the point of the retina .
Our expressionsthen depend entirely on the two integrals cos deys+ ea andJ sin Ace ( Y + - ) .
For estimnating these , we shall refer to FRESNEL'S invafound fom those y by the eqatio s =\ Oitting the costant m ?
ultiplier M/ ie s(remarking only that , as itecomes infinite when a= and impossible when a is negative , different investigations must then be used ) ; and putting C ( s ) forfcos -2 s2 and S ( s ) for sin-2 S2 ; and remarking that , as FRESNEL'S integrals vanish when s= 0 , and as the differential coefficients have the same values with the same sign for negative and positive values of s , the integrals must have equal values but with opposite signs for negative and positive values of s bei o sirl --A(vt-f ) { Ca(4 h._ cb)+1 C( +cb)}c * Mmoire sir la Dite eqraction , Mem. de nstitut , 1821 athed 1822 M6mBoire su.r la Diff~raction , M -em .
de l'Institut , 1821 and 1822 .
229 +COS ( Vt -f ) { S( 2a 6+ a + ?
sinA(Vt-).COsR { C ( \/ *.a+ .
) -h ( \/ .+- ) } +Csin ( vt-/ ).snR { s(.h+ ) -s ( .g ) } +cos27(vt-f).cosR{S(^x/ 'e.th+ a)-s(x/ A.c 6 a ) } si ( v t f).Cos R{C V\/ a gbiC ( / Qac The expression in this general form is rather troublesome .
If , however , we suppose a not to be exceedingly small , so that 2ah is equal to several units , and if we remark that the values of C ( s ) and S ( s ) approach rapidly to the limit , it will be seen that ( for all values of g which are not very nearly equal to h ) we shall commit no sen2II r2a cb 0b sible error n putting 2 for S ( / *h -+ ) S(/ .
hC(/ c.h__Y )C )2a , P ) C(V/ fW~.h+eaJ , S\a X\ce a ce a Making this substitution , our integral may be put into this form : sinV(vt-f ) X [ +C ( )COS R{C(eg+ )a + { sin R{-S ( V.2 e+ ) } ] X'ce Xce+ + cos '\ ( v Z-S( ) s~nR [ Ct ( v/ ~ acb+a)+ cos -f)X +s( ) sin Rc The expression in this general form is rather troublesome .
If , however , we suppose he intensity of light being estimated , as is usual in the theory of unduait s , and if we the sum of the squares of the coefficiend ( s ) approach rapy to the limt i will be seen sible error in putting kfor S ?
h+ s qS hfaind this iion or iteg yt it th isfooing eprein fopoint of the retina : sin { ( \/ at x a ) } +2 { s( &lt; a R gc+ )}2 +s RX [ l-2 { C(a Before entering into the detailed consideration of the numerical value of this expression , we will make a similar investigation for the supposition that the luminous point is too near for distinct vision .
Let the distance of the retina from the lens be c a ' .
The distance , of a point of the wave whose ordinate is y , from the point of the retina whose ordinate is b , is / == { x -a +-2 +y _ 2 } V/ { c2 2 ax 2 by + a'2 + b2 } =/ { c 12 + b2 _ 2by + !
; y2 } ; and putting e2= c a'2 + b2 , and expanding as before , this becomes ba =e -yy + ce e ( ' cb)2 =f +.Cce~ Y ^6 c bc where f/ = ee The integral which expresses the disturbance of ether at the point of the retina is now 2T , tsin / c^\2 2T/ cab2 si ( vt-f ) C cos-(vt-f sin -a/ )c e to be taken from y= -h to y= + g. And sin ( t-/ cossR.y , cos ( na -sinLR sin(ya 2 7r ( v/ fb iat -f2 -cos(vt/ ) cos R. ssin + sin cos c( -y- ) } to be taken from y=+g to y =+ h. Proceeding exactly as before , omitting the constant multiplier x/ -2ce supposing h to be large in the integral , and estimating the intensity of light on the point of the retina as before .
we obtain this expression : InR(/ '2a ga )t2+ { CS(/ 2dc 'k a Xce or { 2C ceV+2 -{ S(g/ *c_ ) } 2 +cosRX[l-2x T[(1V at eb b22+ con RX[2Xe --g 2 Xe g ) ) +sinRxL2C~/ 2a--cb2 -2 / ate 7b2 Xce a'2 XkV IV b)J This expression is exactly similar to that obtained for the case when the luminous point is too distant for distinct vision , in the following particulars , which we shall find to be very important .
First , the general form is the same .
Secondly , the argument of the functions C and S has the quantity 6 with the positive sign in both expressions ; and therefore , in both expressions , the argument increases as b increases .
We shall also find the following remark to be not without importance ; that , in the arguments of the functions in the two expressions , the quantity g enters with different signs .
Our expressions now depend entirely upon the three functions , 1+2 ( C(s))2 +2 ( S())2 , 1 -2 ( C(S))22(S(s))2 , 2C ( s ) 2 S(s ) , which we shall call A(s ) , D(s ) , and E(s ) .
Adopting FRESNEL'S values of C(s ) and S(s ) , I have computed the following values of A(s ) , D(s ) , and E(s ) .
s. A(s ) .
D(s ) .
E(s). .
A(s ) .
D(s ) .
E(s ) .
0'0 0'1 0'2 0'3 0-4 0'5 0'6 0'7 I 0.7 0'S 09 1-0 1O1 1-2 1'3 1-4 ' 1.5 1'6 L 1.6 1'7 ' 1I8 1-9 2'0 2*1 2'2 2-3 2*4 2'5 2-6 2'7 2*8 +1*000 + 1.020 + 1080 +1l180 +1*318 +1-493 + 1-700 + 1-929 +2-169 + 2401 +2*601 +2'743 + 2802 +2*758 +2-609 +2'371 +2-084 +1-814 + 1630 + 1591 +l1713 +1'957 + 2-225 +2-398 r+2385 +2'186 d1'908 +1-719 +1-744 +1.000 +0*980 +0.920 + 0820 + 0682 +0'507 + 0300 +0-071 -0-169 -0-401 -0 .
601 -0-743 -0-802 0-758 -0609 -0-371 -0-084 +0-186 0-370 +0-409 +0287 + 0-043 -0'225 -0-398 -0-385 0-186 + 0'092 + 0281 + 0-256 0'000 +0'199 +0'391 + 0-571 +0-728 + 0856 + 0942 +0'976 + 0949 + 0'852 + 0-685 +0-457 +0-186 --0-093 0'339 -0'502 -0-545 -0-449 -0-233 +0-043 +0-291 +0-416 +0-363 +0-149 --0'128 -0-322 -0-321 -0-120 +0-153 2'9 3-0 3*1 3* ' 3*3 3*4 3-5 3'6 3-7 3'8 3'9 4'0 4'1 4'2 4-3 4-4 4-5 4-6 4'7 4*8 4'9 5*0 5'1 , 5 ?
J 5'2 5-3 5-4 1 5*5 +1-969 + 2227 +2-308 +2-139 + 1-869 +1-754 + 1-912 +2-176 +2-249 +2-042 + 1-809 + 1'850 +2-111 +2'221 + 2'018 + 1-812 + 1-930 +2-176 +2-126 +1-870 +1*879 +2'133 +2-132 +1'879 +1-903 +2-149 +2-070 +2'000 +0-031 -0-227 -0-308 -0-139 +0-131 + 0246 + 0'088 -0-176 -0-249 -0-042 +0'192 + 0-150 -0'111 -0'221 -0-018 +0-189 +0-070 -0-176 -0-126 + 0130 +0-122 -0-133 -0'132 +0-121 +0-097 -0-149 -0-070 0.000 +0-306 + 0.220 -0-039 -0-253 -0-226 +0-019 +0-236 4,0'193 -0-064 -0-234 -0'105 +0'157 +0-197 -0'042 -0-208 -0-047 +0'184 +0-103 -0-150 -0'125 +0-131 +0-130 -0-124 -0-115 +0-135 +0-087 -0-150 0'000 For negative values of s , the numerical values of A(s ) , D(s ) , and E(s ) are the same as for the equal positive values of s : the signs of A(s ) and D(s ) are the same for the positive and negative values of s , but the signs of E ( s ) are different .
1 .
IiIiI 232 IThe first remark which presents itself , in looking at this table , is , that the numbers D ( s ) and E ( s ) increase and diminish , and then change signs and increase and diminish , with increasing values of s ; that the maximum of one always coincides nearly with the vanisthing point of the other ; and that the taximumr values are nearly equal in both series ( regard being had to the general diminution of the numbers as s increases ) ; in fact , that these numbers may be represented as the cosine and sine of the same are , the radius being slowly variable .
To prove this , let us make D ( s ) G ( s ) .
cos p ( s ) , E ( s ) =G ( s ) siln p ( s ) , and investigate the values of G ( s ) and ( ( s ) for every value of s in the Table .
They are easily found by the formula I(r)=2/ ( UCir~~Ts -ED( ) ( s ) G ( o ) o{t ( p + oE ( } ) , tan .
( s ) ( s ) s. In this manner the following Table is for med ; in which , fort more distinct appreciation of the progress of p ( s ) : the differences of p ( s ) are set down .
M CCCXL .
, i G ( s).0 : ( s ) .
Difference .
8G ( s ) .
GDifference .
0o0 1-000 o 68 29 0-307 7200+ 84 14 0l 0'1 1200 11 88 30 -316 135 43 190 1'0 1 '00 11 37 17 0.2 1-000 23 5 31 -311 187 10 54 1 '11 47 0-31 0-999 Z 34 52 3 ' 23 -288 241 11.8 04 '997 46 2 12 5 3-3 263 300 75 12 540 581 s.983 14 17 62 1 0'5 '993 165 I 0*5^993 : 59 21 : 29 3* 4 *2'49 10800+ 4 22 65 5 106.98892 151 5 35 '252a 6930 63 0 0.9 '943 115 155 398 235 259 52 24 I ?
oi ?
90'6i 16 0 71 24 1'0I '909 131 15 3-9 '218 331 16.1 7 18 27 72 59 1'1 '869 14826 171 4 ' '217 14400+ 46 16'2 590 1'2 '82 166 53 0 4-1 26 ; 119 15 1 20 9 ' 7'1 124 1]3 '762 187 202 4 ' 225 190 39 74 2 1 ?
5 ?
622 24 '4 80 51 1-4 '696 209 72 4-3 209265 83 1.5 *625 233 31 4-4 -195 345 59 83 5 16 -552 261 12 2 4-5 -198 1800 ?
+ 69 14 8 31 19 80 23 1-7 -490 292 31 4-6 -204 149 37 35 14 195680 23 1-8 '441 327 45 4-7 195 230 0 86 15 1-9 -413 360+ 63 18 4-8 -180 316 15 2'0 '424A.45 15 39 14 4I9 '180 2160 " + 47 13 90 58 21 '419 81 49 54 0 186 135 38 8825 2-2 428 121 47 3743 5'1 180 223 17 8739 2-3 -424 159 30 5-2 '169 316 32 95 2-4 -404 198 23 41 34 53 '169+ 95 825 '372 239 577 46::173 149 33 398 2-6 -335 9 28 59 55 *166 244 55 92 2-7 -309 336 53 O 2-8 0-300 720 ?
+ 30 52 3y -000 53 22 For negative values of s , the numerical values of G ( s ) and p ( s ) are the same as for equal positive values of s ; the sign of G ( s ) is always positive , but the sign of 0 ( s ) is the same as that of s. p i. n ir lImI 233 2 n The unifortmrity in the progress of the values , both of G ( s ) and of p ( s ) , is very remarkable ; and seems to render it probable that , although the functions D ( s ) andi E ( s ) have been formed in such different ways from C ( s ) and S { ( s ) , and although C ( s ) and S ( s ) are formed in a manner which scarcely permits a sitnple irelation between them , yet some simple expression for G ( s ) and s ( s ) must exist .
This presumption becomes stronger , when we examine the values of s which correspond to the quadrantal values of p ( s ) .
If we make p ( s ) successively equal to 0 , 90 ?
, 180 ?
, 2700 ?
&amp;c. , and take the corresponding values of s , and the squares of those values , we fo-IIrm the following series .
The values of s2 corresponding to the successive quadrants of p ( s ) increase pretty uniformly every time by 1 , except in the first instance , where the increase is 0 '5 .
This interruption of the law is not without analogues in Physical Optics .
I may mention the expression for the intensity of light produced by a grating before a lens , namely / sin n O\2\ sin-i ) ' of which the first and greatest maximum occurs when = 0 , and the snc , r5 ' ceeding maxima occur nearly when 0= - , n --2 &amp; co The small irregularities in the progress of these various numbers may probably arise , as I think , from irregularities in FRESNEL'S calculation of the original integrals .
Several years ago I verified a portion of FRESNEL'S table ; and though the agreement of my numbers with FRESNEL'S was sufficiently close to show that the numbers might be used with perfect safety , yet there were some small discordances which seemed to indicate that a complete recalculation might be useful .
Adopting now for 12 ( C ( s))2 2 ( S ( s))2 and 2C ( s ) 2S ( s ) , the expressions G ( s ) cos p ( s ) and G ( s ) sin 9 ( s ) , and observing that 1+2 ( C ( s))2 +2 ( S ( s))2 =2G ( s ) cos p ( s ) , the expression for the intensity of light , on the point of the reQuadrant of ~ ( s ) .
Value of s. Value of s2 .
Quadrant of p ( s ) .
Value of s. Value of s2 .
0 0.00 0-00 16 3-94 15.5 1 0'73 0-53 17 4-06 16.5 2I 1-27 I 1.61 18 4-18 17*5 3 1-63 2 ' 56 19 , 4.31 18'6 4 1 ' 88 3-53 20 i 4-42 19'6 5 2'12 4-50 21 4-53 20'6 6 2-36 5'60 22 4-64 21-5 7 2-57 6-60 23 4-75 22'5 8 274 7-50 24 4-85 23-5 9 2'91 8'49 25 4'95 24-5 10 309 9-5 26 5'06 25'5 11 3'25 10-6 27 5'15 26'6 12 3-40 11-6 28 5-24 27-5 13 3-53 12-5 29 5-34 28'5 14 3-67 13-5 30 5-44 29'6 15 3-81 14'5 234 tilna whose ordinate is b , becomes , in the case in which the eye is too far from the luminous point to see it distinctly , 2/ GV/ + g ) .
cos P ( / Aac+ g ) a 'ce aa -G(A/ 2ag c. ) osi.8a c(/ ) +G Ac e* ac +-g ) 'sin R.sin V( Ae'aV + gc ___ cb ( A !
ba Ci +G ( ' +.cos pa+g - ) +G ( )(c e the weak light of another ; and thus the alter nations of light and dark will be visible only near the origins of s. Thirdly , that if the origins of s in the non-interfering streamns of light do coincide , but if the values of X are very different , the values of s for the same value of b ( expressed by / a a+ g ) may , when b is large , differ very much for the different values of X , and thus the strong light of one stream will be mingled with the weak light of another : and here also the alternation of light and dark will be visible only near the origins of s. I will now proceed with practical applications of our formulae .
1 .
Suppose that heterogeneous light issues from a point or a narrow line , and is viewed by an eye too distant to see it distinctly : a thin plate of mica , with its edge parallel to the line , is gradually brought across the pupil of the eye : to desclibe the appearance of the line .
The formula is 2--('a G)CA eAe Jr G / L XCe'a +g( 'c s.V , -'a +g 2U ( V^ ce ace a and the aggregate of all the values of this expression for the different values of X is to be taken .
Now the last term may be rejected at once .
For the values of R will vary very much , perhaps to the extent of many multiples of 360 ?
, for the differently coloured rays ; and therefore , in the aggregate , the expressions { M x ( c e a6 ) } which are added together , will have had all values , positive and negative .
The expression is therefore 2G ( G cb +g ) cos ( Ce +C )W hce a *\V ce a The alternations of light represented by this , for the reasons above mentioned , wil be sensible only where +g is small , that is , where b differs little fromg If give t the te bands of light on the right side of the object viewed : and therefore , upon moving the plate of mica , the bands will appear to move in the same direction .
Moreover , the band which is most conspicuous , because its coefficient of the variable part is largest , because it is unaffected by the variation of x , and because the other bands are ranged symmetrically on both sides , is that for which a+g is 0 , and for this the expression of intensity for every colour is 2G ( 0 ) cos ( 0 ) =21= 1 , which is its minimum value .
Hence there will be seen a central dark 236 band strongly marked , with bright and dark bands on each side faintly marked .
And the whole of these phenomena will be independent of the thickness of the Ilica ( within wide limits ) .
The reader may very easily verify these conclusions by experiment ; and the whole will be found strictly in agreement with observation .
If the eye had been supposed too near to see the line of light distinctly , the investigation would have been precisely the same , but the place where the bands are cb sensible would have been determined by making g small , or b nearly equal to + -g ; from which it will appear that , upon moving the plate of mica , the bands will appear to move in the opposite direction ; which agrees with observation .
2 .
Suppose that the linear origins of the various kinds of homogeneous light are separated , either by prism-refraction , or by the diffraction of a grating , or in any other way which arranges the colours in an order corresponding to the order of the values of X ; suppose that the eye is too distant to see the lines of colour distinctly ; a thin plate of mica , with its edge parallel to the lines , is gradually brought across the pupil of the eye : to describe the appearance of the spectrum .
First , suppose the red end of the spectrumr to be on the same side as the plate of mica , or on that side on which b and g are considered positive .
Let k be the ordinate measured from a fixed point on the retina to the centre of the confused image of any one colour ( k therefore is a function of k ) ; and let I be the ordinate , measured from the same fixed point , of any point of which the intensity of light is to be ascertained .
Then k+b= 1 , or b=I k. Substituting this in the general expression for intensity , it becomes 2--G Qa ~- ; I~ c1Ick ( / a ccl ck\ 2G -Ce ' a os .
c4+g -c +G -i +g-a ) *.cos a +g --R Now the second term of this expression may at once be neglected , without consideration of the position of the spectrum .
For as k is a function of x , and as the spectral separation is considerable , A/ a * will vary rapidly with A , and therefore the angle is least for the most refrangible rays .
Moreover , R is greatest , or R is least , for tte most refrangible rays .
Hence the effect of the addition of the term R is to make the variation of the argument of the cosine still more rapid for the variation of % , and the positive and negative values of the cosine will therefore destroy each other , or the third term may be neglected .
The expression for intensity is , therefore , reduced to its first term 2 , or there are no visible bands in the spectrum .
Secondaly , suppose the blue end of the spectrum to be on the same side as the plate of rnica .
The second term of the expression may be neglected , as before .
But with regard to the third term , the circumstances are entirely different .
For k is now least for the most refrangible rays ( the blue end of the spectrum forimed on the retina being on 2a +l ckg the side opposite to the mica ) , and therefore p(x - ' a -+ g ) is greatest for the most refrangible rays ; and therefore the chromatic variations of the different parts of the argument p / 2a .
a+g -R have a tendency to destroy each other .
And by proper selection of the thickness of the piece of mica , the chrorinatic variations of R ( for the colours which fall upon the same point of the retina ) may be the same as the chromatic variations of ( A+ g , for those values of the ftunction which make the bands most brilliant .
In this case , then , the bands produced by all the neigllbouring colours will be aggregate in intensity , and therefore strong bands will be seen on the spectrum .
With regard to the place at which any bright or dark band is seen , as depending on the place of the edge of the mica ; that is , with regard to the value of I for one of these bands , as depending on g ; it will be evident that the intensity ( whether strongest cl or weakest ) will be preserved the same by keeping a+g the same ; that is , there will be a band of the same character so long as 1 varies in the same degree as ' , but in the opposite direction ; that is , the band upon the retina will shift in the direction opposite to the shift of the mica , or will appear to the mind to shift in the same direction as the mica .
But this shift will be small when a is small .
3 .
Suppose that the eye is too near to see the lines of colour distinctly , and that other circumstances are the same as in the second problem .
The expression for intensity in this case is ( , at cl cc os ( / k el c a ' g\ 2-G ( / cg_ ) cos { ' g/ b ( r ( c / / " cla ' ckek\at cIck + ; 3:U\ Gce a : cos e - .
238 The investigation of the effect of the different terms may be conducted in the very same words as in the second problem , and the result is precisely the same ; namely .
that when the red end of the spectrum is on the same side as the mnica , no bands are produced ; but that when the blue end of the spectrum is on the same side as the mica , and the thickness of the mica is properly adjusted , the bands produced by the neighbouring colours will correspond , or nearly so , and strong bands will therefole be produced in the aggregate effect on the eye .
With regard to the place at which any bright or dark band is seen , as depending on the place of the edge of the mica , the intensity will now be preserved the same cl by keeping are g the same ; that is , a band of the same character will be preserved by making the variations of I equal to those of g , and in the same direction ; or the bands upon the retina will shift in the same direction as the shift of the mica , anl the mind will therefore perceive bands to shift in the opposite direction to the shift of the mica .
To exhibit more distinctly to the eye the cause of the annihilation of the bands when the red end of the external spectrum or the blue end of the spectrum on the retina is on the same side as the mica , and the cause of the formation of the bands when the blue end of the external spectrumn or the red end of the spectrum on the retina is on the same side as the mica , I have constructed diagrams ( figs. 3 , 4,5 , 6 , 7 , ) founded on the following calculations .
The Table below contains computed values of 2G ( s ) .
cos p ( s ) +G ( s ) .
cos { f ( s ) R } for all the values of s which are likely to produce sensible effects in the result ; and for the values of R , 0 , 30 ?
, 60 ?
, 900 , 1200 , 150 ?
, 180 ?
, 2100 , 240 ? ?
270 ?
, 300 , 330 ? .
These of course apply also to the values of R , 2 n ; , 2nr+ 300 , 2nr+ 60 ?
, &amp;c. 239 struction of bands is here , as to sense , complete , although the shift of s is the same in amount as that which ( when taken the opposite way ) produces strong bands .
Fig. 5 .
contains a representation of the effect / of shifting the bands in the direction of the shift produced by placing the blue end of the spectrum next to the mica , so that the shift 1'2 in the value of s corresponds to 360 ?
of R. Fig. 6 .
has the same with the shift 18 in the value of s corresponding to 3600 of R ; and fig. 7 .
has the same with the shift 2*4 in the value of s corresponding to 360 ?
of R. Each of these figures has at the bottom a curve showing the effect of the aggregation , supposing twelve curves taken ( the sum of the ordinates being divided by twelve ) , and another showing the effect supposing twenty-four curves taken ( the sumn of the ordinates being divided by twenty-four ) , the value of R for the twenty-fourth being double of that for the twelfth .
The intensity of the bands in the result is here seen well .
It must be remarked that no sensible error is produced here by making no variation in the value of 1 .
For the different kinds of light which are mingled are not those from all parts of the spectrum , but those from parts of the spectrum of a very limited extent ; and the properties of the difference of quality of the different rays enter , not immediately in consequence of the variation of k , but in consequence of the variation of R. It is also to be remarked that strong bands are produced as well in fig. 5 .
as in fig. 7 , although in the former there is a shift of 1P2 in the value of s corresponding to 360 ?
of R , and in the latter a shift of 2*4 in the value of s corresponding to 360 ?
of R. It appears therefore that , with the same spectrum , considerable latitude in the thickness of the mica is allowable , or considerable latitude in the degree to which the spectrum is viewed out of focus .
For the same reason , the same plate of mica , which exhibits bands in the spectrum formed by a prism , may also exhibit bands in the spectrum formed by a grating upon a lens or by reflection from a strliated surface , though the proportionate degree of separation of the colours , in different parts of the spectrum , is exceedingly different in these cases .
The intervals of the bands will , however , always be approximately determined by certain numerical changes in the value of R , and there will , therefore , always be nearly the same number of bands on the spectrum , and this number will always be nearly the same as the number of bands remarked by Mr. TALBOT when the spectrum is seen distinctly .
The bands , therefore , will generally become broader as the spectrum becomes broader , that is , as the eye ( supposed at first too far off for distinct vision ) approaches to the position of distinct vision .
It may , however , happen that , as a becomes small , the changes in the value of s will not nearly correspond to those of R , and therefore , in a position intermediate to that at which these bands and that at which Mr. TALBOT 's bands are clearly seen , no bands whatever may be visible .
In the whole of the special conclusions deduced from the theory , the agreement with observation is complete .
I wish it to be clearly understood that I confine this statement to the general appearance of the phenomena , for measures are yet 242 wanting in every part ; and soIne of them , depending , as they must , on the variable state of the eye as to focal adjustment during the observation , cannot be obtained with ease or certainty .
The train of phenomena and their related theory may be considered remarkkable in the first place for the apparent obscurity of the explanation , which was sufficient to induce an experienced philosopher to ascribe the appearances to a new property of light ; in the next place for the unexpected simplicity of the relation between the nutberus which occur in the investigation , which relation contributes materially to facilitate the understanding of the results deduced from them ; and lastly , I may perhaps add , for the completeness of the explanation which the phenomenoi recetiv.es from the undulatory theory .
Royal Observatory , Greenwich , May 30 , 1840 .
2m2 243

