Page 1

THE

AMERICAN JOURNAL OF SCIENCE.

[THIRD SERIES.]

•••

ART. XXXVI.-On the Relative Motion of the Earth and the

Luminiferous Ethe?'; by ALBERT A. M.ICHELSON and

Enw ARD W. MORLEY. *

THE discovery. of the aberration of light was soon followed

by an explanation according to the emission theory. The effect

was attributed to a simple composition of the velocity of light

with the velocity of the earth in its orbit. The difficulties in

this apparently sufficient explanation were overlooked until

after an explanation on the undulatory theory of light was

proposed. This new explanation was at first almost as simple

as the former. But it failed to account for the fact proved by

experiment that the aberration was unchanged when observa-

tions were made with a telescope filled with water. For if the

tangent of the angle of aberration is the ratio of the velocity

of the earth to the velocity of light, then, since the latter

velocity in water is three-fourths its velocity ill a vacuum, the

aberration observed with a water telescope should be four-

thirds of its true value.t

* This research was carried out with the aid of the Bache Fund.

t It may be noticed that most writers admit the sufficiency ot the explanation

according to the emission theory of light; while in fact the difficulty is even

greater than according to the undulatory theory. For on the emission theory the

velocity of light !!lnst be greater in the water telescope, and therefore the angle

of aberration should be less; hence, in order to reduce it to its Lrue value, we

must make the absurd hypothesis that the motion of the water in the telescope

carries the ray of light in the opposite direction I

AM. JOUR. Sm.-THIRD SERIES, Vor .. XXXIV, No. 203.-Nov., 1881.

22

334

Michel80n and Morley-Relative Motion of the

. On the undulatory theory, according to Fresnel, first, tbe ether

IS su.pP?sed t.o be at rest e.x~ept in the interior o~ transparent

medIa, III whICh secondly, It IS supposed to move wIth a velociy

less than the velocity of the medium in the ratio n'-l where

n' ,

n is the index of refraction. These two hypotheses give a com-

plete and satisfactory explanation of aberration. The second

hyp~thesis, notwithstanding its seeming improbability, must be

conSIdered as fully proved, first, by the celebrated experiment of

Fizeau,* and secondly, by the ample confirmation of our own

work.t The experimental trial of the first hypothesis forms

the suhject of the present paper.

If tLe earth were a transparent body, it might perhaps be

conceded, in view of the experiments just cited, that the inter-

molecular ether was at rest in space, notwithstanding the mo-

tion of thc earth in its orbit; but we have no right to ex-

tend the conclusioll from these experiments to opaque bodies.

But there can hardly be question tilDt tbe ether can and does

pass through metals. Lorentz cites the illustration of a metallic

barometer tube. When the tube is inciined the ether in the

space above the mercury is certainly fmced out, for it is im-

compressible.:/: But again we have no right to assume that it

makes its escape with perfect freedom, and if there be any resist-

ance, however slight, we certainly could not assume an opaque

Lody such as the wbole earth to offer free passage through its

entire mass. But as Lorentz aptly remarks: "quoi qui'l en

soit, on fera bien, a mon avis, de ne pas se laisser guidel', dans

une question allssi im portante, pal' des considerations sur Ie

degre de probabilite ou de simplicite de l'une ou de I'autre

hypothese, mais de s'addresser a l'experience pour apprendre a

connaitre I'etat, de repos ou de mouvement, dans lequel se

trouve I'ether a la surface terrestl'e_"§

In April, 1881, a method was proposed and carried out for

testing the qnestion experimentally.11

In deducing the formula for the quantity to be measured,

the effect of the motion of the earth through the ether on the

path of the ray at right angles to this motion was overlooked.,

* Comptes Rendus, xxxiii, 349, 181;1; Pogg. Ann. Ergiinznngsband, Iii, 457,

)853; Ann. Chim. Phys., 111, lvii, 385, 1859.

t Influence of Motion of the Medi1l1l1 on the Velocity of Light. This Journal,

III, xxxi, 377, 1886.

.

t It may be objected that it may escape by the space between the mercury and

the walls; but this could be prevented by amalgamating the walls.

§\ Archives N eerlandaises, xxi, 2"'" Iivr.

~The relative motion of the earth and the luminiferons ether, by Albert A.

Michelson, this Jour., nT, xxii, 120.

,. It rna}' be mentioned here that the error was puin led ant to the author of the

former paper by M. A. Potier, of Paris, in the winter of 1881.

.Earth and the Luminiferous Ether.

335

The discusssion of this oversight and of the entire experiment

forms the subject of a very searching analysis by H. A. Lo-

rentz,* who finds that this effect can by no means be disregarded.

In consequence, the quantity to be measured had in fact but

one-half the value supposed, and as it was already barely be-

yond the limits of errors of experiment, the conclusion drawn

from the result of the experiment might well be questioned;

since, however, the main portion of the theory remains un-

questioned, it was decided to repeat the experiment with such

modifications as would insure a theoretical result much too

large to be masked by experimental errors. The theory of the

method may be briefly stated as follows:

Let sa, fig. 1, be a ray of light which is partly reflected

in ab, and partly transmitted in aa, beil)g returned by the mir-

rors band e, along ba and ea. ba is partly transmitted along ad,

b, b

I

I

I

I

I

I

1.

I

I

I

c

s

I

I

d

8

C

2.

and ca is partly reflected along ad. If then the paths ab and ae

are equa~, the two rays interfere along ad. Suppose now, the

ether bemg at rest, that the whole apparatus moves in the di-

rection se, with the velocity of the earth in its orbit, the direc-

* De l'Influence du Mouvewent de la Terre sur les Phan. Luw. Archives N eer-

landaises, xxi, 2m • livr., 1886.

336 Miohelson and Morley-Relative Motion if the

tions and dis~ances traversed by the rays will be altered thus:-

The ray sa 1S reflected along ab, fig. 2: the angle bab being

equal to the aberration =a, is returned along ball (~ba, =2~), and

goes to the focus of the telescope, whose direction is unaltered.

The transmitted ra.y goes along ac, is returned along ca" and is

reflected at ai' makmg cale equal 90-a, and therefore still coin-

ciding with the first ray. It may be remarked that the rays bal

and call do not now meet exactly in the same point a" though

the difference is of the second order; this does not aff'ect the

validity of the reasoning. Let it now be required to find the

difference in the two paths aba" and aca/"

Let Y = velocity of light.

v= velocity of the earth in its orbit.

D=distance ab or ac, fig. 1.

T=time light occupies to pass from a to c.

T/=time light occupies to return from c to ai' (fig. 2.)

Then T=yD ,T/=yD. The whole time of going and com·

-v,

+v

ing is T+T/=2D y: ., and the distance traveled in this time

-v

y'

(V')

is 2Dy._v'= 2D 1 + y' , neglecting terms of the fourth order.

The length of the other path is evidently 2Dll+~.' or to the

sam.e degree of accuracy, 2D( 1+ 2~')' The difference is there-

v'

fore Dy •. If now the whole apparatus be turned through 900 ,

the difference will be i~ the opposite direction, hence the dis-

placement of the interference fringes should be 2D;·.. Con-

sidering only the velocity of the earth in its orbit, this would

be 2DXlO-s • If, as was the case in the first .experiment,

D=2X10· waves of yellow light, the displacement to be

expected would be 0'04 of the distance between the interference

fringes.

In the first experiment one of the principal difficulties en-

countered was that of revolving the apparatus without produ-

cing distortion; and another was its extreme sensitiveness to

vibration. This was so great t.hat it was impossible to see the

interference fringes except at brief intervals when working in

the city, even at two o'clock in the morning. Finally, as be-

fore remarked, the quantity to be observed, namely, a displace-

ment of something less than a twentieth of the distance be-

tween the interferenc,e fringes may have been too small to be

detected when masked by experimental errors.

Earth and the Ltl1niniferou8 Ether.

331

The first named difficulties were ent.irely overcome by mount·

ing the apparatus on a massive stone floating on mercury; and

the second by increasing, by repeated reflection, the path of the

light to about ten times its former value.

The apparatus is represented in perspective in fig. 3, in plan in

fig. 4, and in vertical section in fig. 5. The stone a (fig, 5)is about

1'5 meter square and 0'3 meter thick. It rests on an annular

wooden float bb, 1'5 meter outside diameter, 0'1 meter inside

diameter, and 0'25 meter thick. The Boat rests on mercury

contained in the cast-iron trough ee, 1'5 centimeter thick, and

of such dimensions as to leave a clearance of about one centi·

meter around the float. A pin d, guided by arms gggg, fits into

a socket e attached to the float. The pin may be pushed into

the socket or be withdrawn, by a lever pivoted at f This pin

keeps the float concentric with the trough, but does not bear

any part of the weight of the stone. The annular iron trough

rests on a bed of cement on a low brick pier built in the form

of a hollow octagon.

3.

At each corner of the stone were placed four mirrors d dee

fig. 4. Near the center of the stolle was a plane-parallel glaRs b.

These were so disposed that light from an argand burner a

passing through a lens, fell. on b so as to be in part reflected

to d,; the two pencils followed the paths indicated in the fiaure

bdedlif and bd,e,d,bf respecti vely, and were observed by tht: °tele~

scope f. Both f and a revolved with the stone. The mirrors

were of speculum metal carefully worked to optically plane

surfaces five cen ti meters in diameter, and the glasses band c

were plane·parallel and of the same thickness. 1'25 centimeter'

.

,

338

Michel80n and Morley-Relative Motion oftlw

their surfaces measured 5'0 by 'l'5 centimeters. The second of

these was placed in the path of one of the pencils to compen-

sate for the passage of the other through the same thickness of

glass. The whole of the optical portion of the apparatus was

kept covered with a wooden cover to prevent air currents and

rapid changes of temperature.

The adjustment was effected as follows: The mirrors hav-

ing been adjusted by screws III the castings which held the

4.

al

/

mirrors, against which they were pressed by springs, till light

from both pencils could be seen in the telescope, the lengths of

the two paths were measured by a light wooden rod reaching

diagonally from mirror to mirror, the distance being read from

a small steel scale to tenths of millimetel's. The difference in

the lengths of the two paths was then annulled by moving the

mirror e. This mirror had three adjustments; it had an adjust.

ment in' altitude and one in azimuth, like all the other mirrors,

Earth and the Luminiferou8 Ether.

339

but finer; it also had an adjustment in the directio~ of the

incident ray, sliding forward or backward, but keepmg very

accurately parallel to its former plane. The three a.dJust~~nts

of this mirror could be made with the wooden cover III posltlOn.

The paths being now approximately eq ual, t~e two imag~s

of the source of light or of some well·defined object placed m

front of the condensing lens, were made to coincide, the .teles-

cope was now adjusted for distinct vision of the expected l?-ter.

Ference bands, and sodium light was substituted for white hght,

when the interference bands appeared. These were now ~ade

as clear as possible by adjusting the mirror el; then white lIght

was restored, the screw alteri ng the length of path was very

slowly moved (one turn of a ~crew of one hundred threads to the

e

inch altering the path

5.

nearly 1000 wave.lengths)

a

till the colored interfer-

ence fringes reappeared

in white light. These

c

c were now given a con-

~

~

b Uc 9 al 9 ell b

g O-oQ

f

venient width and posi-

tion, and the apparatus

was ready for observa-

tion.

The observations were

conducted as follows:

Around the cast-iron

trough were sixteen equidistant marks. The apparatus was

revolved very slowly (one turn in six minutes) and after a

few minutes the cross wire of the micrometer was set on the

clearest of the interference fringes at the instant of passing

one of tIre marks. The motion was so slow that this could be

done readily and accurately. The reading of the screw-head

on the micrometer was noted, and a very slight and gradual

impulse was given to keep up the motion of the stone; on

passing the second mark, the same 'pl"Ocess was repeated, and

this was continued till the apparatus had completed six revolu-

tions. It wtlS found that by keeping the apparatus in slow

uniform motion, the results were much more uniform and con-

sistent than when the stone was brought to rest for every ob-

servation; for the effects of strains could be noted for at least

half a minute after the storie came to rest, and during this time

effects of change of temperature came into action.

The following tables give the means of the six readin~s; the

first, for observations made near noon, the second, those near

six o'clock in the evening. The readings are divisions of the

screw·heads. The width of the fringes varied from 40 to 60

divisions, the mean value being near 50, so that one division

340

Michel80n and Morley-Relative Motion of the

means 0'02 wave·length. The rotation in the observations at

noon was contrary to, and in the evening observations, with,

that of the hands of a watch.

NOON OBSERVATIONS,

16,

1,

2,

3,

4,

5.

6.

7,

8.

9,

10.

11,

12, 13, 14. 15,

- - - -

- - - - -

- - -

-

- - -

JulyS ....... 44-'7 44'0 43'5 39'7 3,,'2 34'7 34'3 32'5 28'2 26'2 23'8 23'2 20'3 18'7 17'5 .16'8

July 9 ....... 57'4 57'3 58-2 59'2 58'7 60'2 60'8 62'0 61'0 C,s'3 65'8 67'3 69'7 70'7 73'0 70'2

July 11 .. , .. 27'3 23'5 22'0 19'3 19'2 19'3 18'7 18'S 16'2 14'3 13'3 12'8 13'3 12'3 10'2 7'3

Mean ........ 43'1 41'6 41'2 39'4 37'7 38'1 37'0 37'S 3,,'3 34'0 34'3 34'4 34'4 33'9 33'6 31'4

Mean in w.l, '862 '83'2 '824 '788 '754 '762 '758 '7M '706 '692 '686 '688 '688 '678 '672 '628

'706 '692 'C>86 '688 '688 '678 '672 '628 '616

Final mean. '784 '762 '700 '738 '721 '720 '715 '692 '661

P, M, OBSERVATIONS.

July·S ....... 61'2 63'3 68'3 68'2 67'7 69'3 70'3 6!I'S o!)"o 71'3 71'3 70'5 71'2 71'2 70'5 72'5

July 9 ....... 26'0 26'0 28'2 29'2 31'5 32'0 31'3 31'7 33'0 31,'8 36'5 37'3 38'8 ,n'o 42'7 43-7

July12 ...... 66'S 66'5 66'0 tW3 62'2 61'0 61'3 59'7 58'2 00'7 53'7 54'7 5.Ij'O 58'2 58'5 57'0

Mean ....... 51'3 51'9 52'5 53'9 53'8 54'1 M'3 53'7 5.3'4 54'3 53'8 54'2 00'0 56'8 57'2 57'7

Mean in w.l, 1'026 1'038 l'OSO 1'078 1'076 1'082 1'086 1'074 1'068 1'086 1'076 1'084 1'100 1'136 1'144- 1'154

1'068 1'086 1'076 1'084 1'100 1'136 1'144- l'lM 1'172

Final mean. 1'0471'0621'063 l'OSl 1'068 1'109 1'115 1'114 1'120

The results of the obset'vations are expressed graphically in

fig. 6. The upper is the curve for the observations at noon,

and the lower that for the evening observations. The dotted

curves represent one·eighth of the theoretical displacements. It

seems fair to conclude from the figure that if there is any dis·

,,,.,, .. ,,.-------...........

",

.........

,...

... ...

6.

- o'os2

"

..._--_ ...... '

,-

,",'"

....

<.

'~ o·ooil

~

E ,,::'----------------.,~,,~s

u

W

to"

~, ... "

N

............

,,-

................. ---............ ~.,.

placement due to the relative motion of the earth and the

luminiferous ether, this cannot be much greater than 0'01 of

the distance between the fringes.

Considering the motion of the earth in its orbit only, this

1~'

-

13'')'

72'2

6'5

30'S

'616

75'')'

44'0

56'0

58'6

1'17

l!.arth and the Luminiferou8 Ethel"

341

v'

displacement should be 2DV,=2DX10-s. The distance D was

about eleven meters, or 2X107 wave-lengths of yellow light;

hence the displacement to he expected was 0'4 fringe. The

actual displacement was certainly less than the twentieth part

of this, and probably less than the fortieth part. But since the

displacement is proportional to the square of the velocity, the

relati ve velocity of the earth and the ether is probably less than

one f<ixth the earth's orbital velocity, and certainly less than

one-fourth.

In what precedes, only the orbital motion of the earth is con-

sidered. If this is comhined with the motion of the solar sys-

tem, concerning which but little is known with certainty, the

result would have to be modified; and it is just possible that

the resultant velocity at the time of the observations was small

though the chances are much against it. The experiment will

therefore be repeated at intervals of three months, and thus all

uncertainty will be avoided.

It appears, from all that precedes, reasollably certain that if

there be any relati ve motion between the earth and the luminif-

erous ether, it must he small; quite small enough entirely to

refute Fresnel's explanation of aberration. Stokes has given a

theory of aberration which assumes the ether at the earth's sur-

face to be at rest with regard t<Y the latter, and only requires

in addition that the relative velocity nave a potential; but

Lorentz shows that these conditions are incompatible. Lorentz

then proposes a modification which combines some ideas of

Stokes and Fresnel, and assumes the existence of a potential,

together with Fresnel's coefficient. If now it were legitimate

to conclude from the present work that the ether is at rest with

regard to the earth's surface, according to Lorentz there could

not be a velocity potential, and his own theory also fails.

Supplement.

It is obvious from what has gone before that it would be

hopeless to attempt to solve the question of the motion of the

solar system by observations of optical phenomena at the sU1face

of the earth. But it is not impossible that at even moderate dis-

tances above the level of the sea, at the top of an isolated moun-

tain peak, for instance, the relative motion might be percepti-

ble in an apparatus like that used in these experiments. Per-

haps if the experiment should ever be tried in these circum-

stances, the cover should be of glass, or should be removed.

It may be worth while to notice another method. for multi-

plying the sq uare of the aberration sufficiently to bring it within

the range of observation, which has presented itself during the

842 Michelson and MO'l'ley-Relatirve Motion of the

preparation of this paper. This is founded on the fact that re-

flection from surfaces in motion varies from the ordinary laws

of reflection.

Let ab (fig. 1) be a plane wave falling on the mirror mn at an

incidence of 45°. If the mirror is at rest, the wave front after

reflection will be ac.

Now suppose the mirror to move in a direction which makes

an angle a with its normal, with a velocity w. Let V be the

velocity of light in the ether supposed stationary, and let cd be

the increase in the distance the light has to travel to reach d.

cd

In this time the mirror will have moved a distance ------,-,;:----

.y2 cos a'

cd

w .y2 cos a

.

ac

We have ad = V

whICh put = r, and ad = 1- r.

In order to find the new wave front, draw the arcfg with bas

a center and ad as radius; the tangent to this arc from d will

be the new wave front, and the normal to the tangent from b

will be the new direction. This will differ from the direction

ba by the angle 0 which it is required to find. From the equal-

ity of the triangles adb and edb it follows that 0=2f/J, ab=ac,

o

1- tan-

tan adb= tan( 45° - ~) =

~ = :~ =1-r,

l+tan T

or neglecting terms of the order r",

O

r'

.y 2w cos a + w' •

=r+"2 = V

V.cos a.

Now let the light fall on a parallel mirror facing the first, we

- V2w cos a

w'

should then have 0,=

V

+ V. cos"a, and the total de-

viation would be 0+8, = 2p' cos'a where p is the angle of

aberration, if only the orbital motion of the earth is considered.

The maximum displacement obtained by revolving the whole

apparatus through 90° would be LI=2p·=0·004/1. With fifty

such couples the displacement wOllld be 0'2/1. But astronomi-

cal observations in circumstances far less favorable than those

in which these may be taken have been made to hundredths of

a second; so that this new method bids fair to be at least as

sensitive as the former.

The arrangement of apparatus might be as in fig. 2; s in the

focus of the lens a, is a slit j bb cc are two glass mirrors opti-

cally plane and so silvered as to allow say one·twentieth of the

light to pass through, and reflecting say ninety per cent. The

intensity of the light falling on the observing telescope df

844 Michel80n and Morley-Motion qf the Earth, etc.

:vould . be about one.mi~lionth of the original intensity, so that

If sunlight or the electrlc arc were used it could still be readily

seen.. The mirrors bb, and ee, would differ from parallelism

sufficIently to separate the successive images. Finall V the

apparatus need not be mounted so as to revolve, as the e~rth's

rotation would be sufficient.

If it were possible to measure with sufficient accuracy the

velocity of light without returning the ray to its starting point,

the problem of measuring the first power of the relative velocity

of the earth with respect to the ether would be solved. This

may not be as hopeless as might appear at first sight, since the

difficulties are en tirely mechanical and may possibly be sur-

mounted in the course of time.

For example, suppose (fig. 3) m and m, two mirrors revolving

with equal velocity in opposite directions. It is evident that light

from 8 will form a stationary image at 8, and similarly light

from 8, will form a stationary image at 8.

If now the velocity

of the minora be increased sufficiently, their phases still being

exactly the same, both images will be deflected from 8 and 8,

in inverse proportion to the velocities of light in the two direc·

tions; or, if the two deflections are made equal, and the differ·

ence of phase of the mirrors be simultaneously measured, this

will evidently be proportional to the difference of velocity in

the two direetions. The only real difficulty lies in this measure·

ment. The following is perhaps a possible solution: gg, (fig.

4) are two gratings on which sunlight is concentrated. These

are placed so that after falling on the revolving mirrors m and

mil the light forms images of the gratings at 8 and 8" two very

sensiti ve selenium cells in circuit with a battery and a telephone.

If everything be symmetrical, the sound in the telephone will be

a maximum. If now one of the slits 8 be displaced through

half the distance between the image of the grating bars, there

will be silence. Suppose now that the two deflections having

been made exactly equal, the slit is adj usted for silence. Then

if the experiment be repeated when the earth's rotation has

turned the whole apparatus through 180°, and the deflections

are again made equal, there will no longer be silence, and the

angular distance through which 8 mnst be moved to restore

silence will measure the required difference in phase.

There remain three other methods, all astronomical, for at-

tacking the problem of the motion of the solar system through

space.

1. The telescopic ohservation of the proper motions of the

stars. This has given us a highly probably determination of

the direction of this motion, but only a guess as to its amount.

2. The spectroscopic observation of the motion of stars in

the line of sight. This could furnish data for the relative

345

motions only, though it seems likely that by the immense im-

provements in the photography of stellar spectra, the informa-

tion thus obtained will be far more accurate than any other.

3. Finally there remains the determination of the velocity of

light by observations of the eclipses of Jupiter's satellites. If

the improved photometric methods practiced at the Harvard

observatory make it possible to observe these with sufficient

accuracy, the difference in the results found for the velocity of

light when Jupiter is nearest to and farthest from the line of

motion will give, not merely the motion of the solar system

with reference to the stars, but with reference to the lumin-

iferous ether itself.