PhysicsEssays volume 5, number 1, 1992

A New Michelson-Morley Experiment

E.W. Silvertooth and C.K. Whitney

Abstract
We have augmented a Michelson-Morley interferometer with additional instrumenta-
tion and secured additional measurements of a O/pe not previously reported by other
researchers. Where the original Michelson-aorley experiment looked only at the inter-
ference beaveen two beams that had traversed the two interferometer arms, we look
at the standing-wave interference within one of the arms as well. Assuming that the
interferometer is resident in an inertial coordinate frame, special relativity theory pre-
dicts nodes spaced X/2 apart. Our experiment shows a departure from this expected
nodal spacing. The discrepancy is here shown to be consistent with an anisotropy in
the speed of light having theform c(0) = c/[ 1+ (v o/c) cos (O) ], where c is the usual
3 x 10s m/s, Vo is about 378 km/s, and 0 = 0 lies in a unique direction relative to
fixed stars. The apparent Vo is consistent with both the known anisotropy of background
cosmic radiation and the believed orbital velocitv of the Sun in the Milky Way Galaxy.

Key words: special relativity, Sagnac effect, velocity detection

1. INTRODUCTION be detectable after all? An earlier experiment by Silvertooth(8~ attempted

The original Michelson-Morleyexperiment (1) currently plays a very impor- to answer in the affirmative. The experiment compared intensity signals
tant role in our understanding of twentieth-century physics. Their experiment from two detectors,(9~ one viewing the interference of cotraveling waves,
sought to detect the ether that MaxwelF2) had postulated as the medium and the other viewing the interference of countertraveling waves. The latter
for the propagation of light. The null result obtained by the experiment sensor was specially designed and fabricated for the purpose. (10) In the
was taken as evidence that no such ether exists. The ramifications go far experiment the optical paths controlling the two interferences were varied
beyond light propagation. The Maxwell equations present a more general together. Assuming velocity-insensitive speed of light, the intensity signals
problem of Galilean noninvariance which can be cured either by modifying ought to change together when the optical path lengths change together.
the equations, ,~ la Hertz, or by dispensing with space and time invariance, That is, if physical dimensions are such that both intensities start out at
a la Lorentz.(3) The latter having been the approach taken, it required a maximum, then equal path variations ought to reduce them both by the
justification. Without a physical ether to explain length contraction and time same amount. For continuing path variations, the two signals ought to
dilation, Einstein's mathematical theory of special relativity(4) became the oscillate together between maximum and zero. They did not in fact do this:
only apparently viable explanation. there was systematic evolution of the relationship between the two intensity
But within this same time period a conflicting line of thought has also signals.
been developing. Curiously the same Michelson is involved. The Michelson- The significance of the earlier experiment has since been debated in the
Gale experiment (s) confirmed a much smaller-scale experiment by Sagnac (6) literature.O1-15) Some of the debate has been based on a simple semantic
demonstrating that countertraveling waves accrue a phase shift that is linear confusion. Silvertooth suppressed noise in his experiment by mechanically
in rotation velocity.This means that velocity due to rotation can in principle dithering the path variation imposed and filtering the intensity signals for
be detected without external reference. But according to special relativity, content at the dither frequency. From the phase between the output dither
linear velocity cannot be so detected. Numerous experiments continuing signals he could tell if the intensities were varying together or contrary to
into modem times support both statements. (7) But there is an obvious each other. Thus the concept of "dither phase" is important to the description
conflict: with speed v equal to rotation radius r times rotation angular of the experiment. But the term "dither phase" was sometimes abbreviated
rate co, one can imagine increasing r and decreasing co indefinitely, while to just phase and was then understood by some to mean optical phase.
holding v constant and making it ever more nearly linear. So when does Optical phase would of course be undetectable, and that fact has been the
v cease to be, in principle, detectable? basis for misdirected criticism of the experiment.
Could it be possible that velocity, even linear velocity, would in principle Silvertooth/8~ held that the departure from expected behavior was indica-

82
 E.W. Silvertoothand C.K. Whitney

/'11 M1
I

Bs
1,12 ~@ BS
---2 M2

Figure l(a). Original Michelson-Morley experiment configuration. The ex-

i
periment measures the phase shift between the beam traveling out to mirror Figure l(b). Configuration augmented to provide a reference signal. Move-
M1 and back, and the beam traveling out to mirror M2 and back. merit of mirror M1 causes the signal at PO to oscillate between maximum
and zero.

tive of an underlying velocity that affected the speed of light. The velocity pending to it various additional features that facilitate an investigation of
magnitude could readily be inferred from the experiment data and was of the standing-wave pattern in one of the arms. Figures l(a) through l(d)
order c/lO00. Silvertooth held that this inferred velocity was probably dom- develop the experiment configuration.
inated by a velocity relative to an absolute frame of reference defined by Figure l(a) shows the standard Michelson-Morley arrangement of a light
the now well-known anisotmpy of cosmic background radiation at approxi- source, beam splitter BS, mirrors M1 and M2, and photodetector PD. The light
mately 3 K.(16) But if this were really the case, it would imply the existence source in Michelson-Morley's day was incandescent, but in any modem re-
of a preferred inertial coordinate frame, contrary to the most fundamental creation would be a laser in order to provide a strong, collimated, monochro-
assumption in special relativity theory. matic beam.
Prompted by the involvement of a Sagnac-type interfemmeter in the The return beams from the two arms of the interfemmeter interfere at
experiment, WhitneyO5) suggested that the velocity detected might have been PD and form a pattern of concentric rings, usually observed slightly offset
rotational and thus beyond the scope of special relativity theory. But the large so that one sees a handful of fringes. In the original Michelson-Morley
magnitude of rotation velocity to be accounted for requires a truly cosmic experiment the interfemmeter was rotated through various positions. The
interpretation. It excludes the Earth's diurnal rotation and its annual orbit observable fringes were found not be affected by any such rotation. This was
around the Sun as significant contributors, leaving only the orbital motion taken as evidence that the speed of light was a constant c, independent of
of the Sun within the Milky Way Galaxy as a viable explanation. This is direction.
certainly a very nearly linear motion. If detectable, it leads one to ask when, Figure l(b) shows the Michelson-Morley setup upgraded with a laser and
if ever, special relativity actually applies. augmented with devices to move one of the mirrors in order to vary the
The ultimate choice among competing interpretations will require further length of one of the arms. There is a micrometer for coarse motion and
data and hence further experimentation. The present experiment is a step a piezoelectric transducer for fine motion. For convenience there is also a
in this direction. Here we attempt to simplify the experiment configuration, turning flat. Given the result of the original Michelson-Morley expeirment,
expand the description of the data it produces, and better explain the analysis the turning flat has no effect on the signal at PD.
applied to the data. At PD the fringe pattern is centered, so that PD sees just the central spot
of the pattern. Then as mirror M1 is moved, the intensity at PD oscillates
between maximum and zero. Given laser wavelength ~, a displacement of
2. EXPERIMENTCONFIGURATION M1 by distance L/2 produces a full cycle of intensity variation at PD. This
The original Silvertooth experiment configuration was quite complex, provides a reference against which to compare additional signals that will
having two separate interferometers with two separate laser sources and three now also be described.
moving parts. In the present configuration the Sagnac-type interferometer Figure l(c) shows the Michdson-Morley setup further augmented with an
encloses zero area so that area is not a potential issue. Just one laser is used additional beam splitter and sensor. The beam splitter BS2 samples the two
so that correlation between two lasers is not an issue. And one moving part countertraveling waves in the second arm of the interferometer. The standing-
is eliminated so that possible sensitivity to positional variabilities is reduced. wave sensor SWD monitors the interference between the sample beams. The
The simplified experiment configuration can be explained systematically light paths from BS2 to SWD are contrived to form a figure eight, so that
by starting with a basic Michelson-Morley configuration and gradually ap- they will enclose zero total area and not constitute a conventional Sagnac
83
A New Michelson-MorleyExperiment

TABLE TAT
IIM1
E MICROMETER IPZT~
-'L -\
i
SWD SWD
i
I

F1- F1 I

<-->

xt2 BS2
M2
BS1
,7 'I
- J - F2

BS2
M2

' I
-}~SClLLOSCOPE S ~ w S J
Figure l(c). Configuration augmented with additional sample sensor. The
sensor SWD probes the standing wave at variable position x.
Figure l(d). Experiment configuration completed with noise suppression.
The signals from PD and SWD appear as traces on the oscilloscope.

interferometer. Filters F1 and F2 are added to equalize the intensities of

beams interfered at PD and SWD. 3. EXPERIMENT RESULTS
The sensor SWD is forced to move along with M1 by having them both At 300 Hz the dither frequency of the M2 mirror is very low compared to
tied to a movable table that is driven in translation by the micrometer and the laser optical frequency v. So its only observable effects are dither in the
the PZT. The variable x measures the translational offset of SWD and BS2. time-average intensity levels observed at the two sensors, PD and SWD. The
The interference state at SWD is essentially equivalent to that at x. oscilloscope allows amplitude and phase comparison between the dithered
To compare the signals at PD and SWD as carefully as possible, any noice signals detected at PD and SWD.
should be suppressed. One way to do this is by imposing a low-frequency Figure 3(a) shows the first such comparison. The path lengths in the
dither input and then filtering the outputs for that frequency. Figure l(d) interferometer are such that the two signals are in phase with each other,
shows the experiment setup further modified in order to do this. A sine wave and of maximum possible amplitude. If x is now changed slightly by PZT1,
generator dithers PZT2 and also triggers the oscilloscope that displays the the two signals stay in phase, but decrease in amplitude. Changing x by a
sine wave outputs from PD and SWD. quarter wave makes both signals go to zero.
Figure 2(a) shows the actual experimental apparatus in its housing. Clearly, as x moves, intensity peaks and valleys traverse PD and are
During operation, the experiment is enclosed by a thin plastic sheet to traversed by SWD. Let us refer to these peaks and valleys as fringes. In Fig.
reduce air currents. The key standing-wave sensor SWD is held by the large 3(a) the fringes are exactly in phase with each other, so that dither produces
cylindrical container visible at the center of the photo. This container is the same output from both. The dither output is maximum amplitude when
attached to a horizontal plate, at each end of which there is a sheet of the nominal x centers on the midpoint of a fringe edge. Displacement by
tempered steel shim stock to allow parallel motion of the sensor. Two small a quarter wave then centers the dither on a fringe peak or fringe valley,
knobs near the right-hand end control micrometer motions that allow the whereupon the only output is at twice the input frequency, which is filtered
sensor and its two input beams to be aligned. The large circular disk to the out before display on the scope.
right is a variable filter used to equalize the intensities of the beams going Figure 3(b) shows another such comparison, this time after a macroscopic
to SWD. change in x imposed by the micrometer. The dither amplitudes are again
Figure 2(I)) shows the novel sensor SWD. In the large end is a substrate maximum, and displacement by a quarter wave again drives them both to
with a semitransparent photocathode surface. This assembly produces an zero. But now the signals are exactly opposite in phase.
output signal that is maximum when the two input beams are in phase at We believe that this behavior means the fringes at PD and SWD are now
the photocathode surface, zero when they are out of phase. At the other end exactly interleaved with each other, so that when dither excursion moves
is a five-stage photomultiplier tube that amplifies the signal generated. One toward a peak on one fringe pattern, it necessarily also moves toward a
of the vertical steel shims mentioned in the description of Figure 2(a) can valley on the other fringe pattern and vice versa. Displacement by a quarter
be seen above and behind the photomultiplier tube. wave then centers the dither on a peak of one fringe pattern and on a

84
 E.W. Silvertoothand C.K. Whitney

Figure 2(a). The experiment in its housing. The laboratory is located at Figure 2(b). The novel sensor SWD. Design and fabrication are discussed in
Silvertooth's address, in Olga, Washington, north of Seattle. Ref. 10.

Figure 3(a). Initial in-phase condition of dither signals from PD and SWD. Figure 3(b). Subsequent out-of-phase condition. This is produced by mi-
The PZT1 has been adjusted until the signal amplitudes are maximum. crometer excursion and PZT1 adjustment.

valley on the other one, so the only outputs are again at the filtered double adjusting the micrometer/PZT1 combination until the in-phase condition is
frequency. observed, then reading the micrometer, then adjusting the micrometer/PZT1
Figure 3(c) shows the situation at a value of x halfway between those combination again until the out-of-phase condition is observed, then reading
for Figs. 3(a) and 3(b). Here, maximizing the amplitude from one sensor the micrometer again, and finally differencing the two micrometer readings.
results in zeroing the amplitude from the other sensor. Displacement by an The magnitude D of the displacement x required for phase reversal
eighth of a wave exchanges the traces: whichever one was small amplitude described varies with the diurnal rotation of the Earth. A minimum dis-
becomes large amplitude and vice versa. placement Do occurs at 12-hr intervals, with D becoming unmeasurably
We believe this behavior means that the fringe patterns at PD and SWD large at the halfway points in between. This behavior implies a dependence
are misaligned with each other such that the peaks and valleys of one on orientation with respect to fixed stars. In terms of an orientation angle
pattern coincide with the edge midpoints of the other pattern. O measured with respect to the constellation Leo, we have approximately
Evidently the relationship between oscilloscope traces from PD and SWD
changes systematically from dithered-in-phase to dithered-out-of-phase as
D (O) = Do for O = 0, m
x is varied. The experiment measures the value D of the displacement (la)
in x such that the relationship reverses. A measurement consists of first D (O) ---* oo for O = m2, 3 7t/2.

85
A New Michelson-MorleyExperiment

by Ax, one of the two paths to PD changes by 2 Ax, while each of the two
paths to SWD changes by 1 Ax. In terms of time-averaged intensities, and
hence observable dither amplitude and phase, these path changes ought to
change both output signals in exactly the same way.
But this is not what happens experimentally. Instead, the number of
fringes traversed by SWD generally differs from the number of fringes
traversing PO. This can only mean that the fringes traversed by SWD differ in
length from those traversing PD, and hence that wavelengths perpendicular
and parallel to x differ. Since the temporal frequency of optical oscillations
in the experiment is fixed by the laser, the propagation speeds perpendicular
and parallel to x must also differ.
One conventionally acceptable reason might be that coordinate x exists
in a frame that is not inertial. For example, there could be an underlying
rotation with a velocity component parallel to x. Then a propagation path
parallel to x that measures some length Lr in the rotating frame actually
has a different length Li in inertial space. This implies that in the rotating
frame the propagating speed is

Cr = CLr/Li -"#C. (2a)

Figure 3(c). In-between condition. Maximizing the amplitude from one
sensor results in zeroing the amplitude from the other sensor. The value of Li depends on both the normalized speed of rotation 13 = v / c
and the direction of the propagation in relation to the vector I~. Thus we
further expect that Cr bifurcates into direction-dependent propagation speeds
Is all this real? A measurement takes only about a minute to accomplish,
so we do not believe that any sort of electronic or thermal drift is causing the Cr = c+ and Cr = C-. (2b)
effect. Because the oscilloscope traces are so clean, we believe the difference
between the in-phase condition and the out-of-phase condition is not an Michelson and Morley expected the variation in the speed of light to be
artifact of noise. Because the measurement is inherently a difference, we consistent with the concept of vector addition. For propagation parallel and
believe it is not dominated by any kind of bias. antiparallel to [~,
How accurate are the measurements? At any 0 the D can, in principle, be
measured by counting the fringes elapsed at PD. The presence of signal noise c_ =c(1 -13) andc+ =c(1+ 13), (3a)
can, of course, make the detection of the exact phase condition difficult, but
the measurement is much facilitated by the filtering involved. Given perfect while for propagation perpendicular to I~,
noise suppression, the displacement can, in principle, be measured to within
a quarter of a wavelength ~.. co ~- c(1 - 132/2). (3b)
But with our 0.63 ~tm laser source, the observed Do corresponds to about
400 fringes. Without sophisticated instrumentation not yet available to us, Propagation forward and back to PD should then have produced a phase
there are too many fringes to count very reliably. Instead, we have used the shift between the two arms differing in second order from what it would
direct micrometer reading. So instead of being accurate to k/4 = O. 16 Bm, have been without the variation:
the measurements are really only good to 1/10 000 in., or 2.54 Bm.
In addition, there is some uncertainty associated with aligning the ap- 8r oc ( c / c _ + c/c+ - 2C/Cp)
paratus with respect to the constellation Leo on different days and with
judging that the dither traces at PD and SWD are exactly inphase or exactly ~[v(1-13) +v(1+13) -2/(1-132/2)] o~132. (3c)
out-of-phase. The latter especially limits the accuracy with which Do can be
stated. Considering the line widths as compared to the cycle lengths shown Michelson and Morley rotated their apparatus in order to reverse the
in Figs. 3(a) to (c), we believe we have at best second-order phase shift and display the reversal in the form of interfero-
metric fringe shifts. But they found nothing, and the physics community has
Do = O. 125 mm -4- 5%. (lb) generally concluded that there must therefore have been no variation in the
speed of light.
But later, Michelson and Gale demonstrated a phase shift between coun-
4. CONCEPTUAL ANALYSIS
tertraveling waves. This was a Sagnac effect linear in Earth rotation [3. In
Under conventional assumptions the difference between dithered outputs
the present experiment the sensor SWD looks at countertraveling waves, and
at PD and SWD ought to be insensitive to changes in x. &~suming the speed
the phase shift is
of light c governs the propagation of all optical signals in the system, the
dithered outputs at PD and SWD should differ by an amount depending
only on the path lengths of the various interfered beams. When x is varied 8r oc (c/c_ -c/c+) o,: 13. (4a)

86
 E.W. Silvertoothand C.K. Whitney

This would imply some form of variation in c, even if not exactly that D ~ 1~4131 with 13(0) = [3o cos (0), (6b)
described by (3a) and (3b).
Replacing the right-hand side of (3c) with zero and allowing an undeter- which is indeed consistent with the direction variation tabulated in Eq. (la),
mined proportionality constant on the right-hand side of (4a), we have two given
relationships that the variation must satisfy. The simplest candidate solution
has the same factors as in (3a), but appearing as divisors rather than Do = k/413o. (6c)
multipliers:

c+ = c / ( 1 - 13) and c_ = c / ( 1 + ~ ) , (4b) We proceed to estimate a velocity magnitude defined by

with the corresponding Vo = C13o = c L/4Do. (7a)

cp - c , (4c) Michelson and Gale detected a Vo equal to Earth diurnal rotation speed,
about 350 m/s, while Michelson and Morley had expected to find a Vo equal
A possible interpretation for such a functional form is power series sum: to Earth annual orbit speed, about 30 km/s. The present experiment implies
c+ and c_ differ from c by a series of corrections in powers of 13. The 13 something much larger yet: inserting (lb) into (7a) gives
in question must then be a scalar quantity, the proiection of vector 13 in
the propagation direction. Vo = 378 km/s + 5%. (7b)
In effect, we are suggesting here an alternative explanation for the original
Michelson-Morley null result; namely, that the variation in the speed of light
6. PHYSICAL INTERPRETA_TIONS
could have had a functional form different from what they expected and
The velocity (7b) replicates the result found and reported earlier by
designed that experiment to detect.
Silvertooth.(8) The inferred velocity is very large indeed and so not easily
dismissed as a subtle experimental error. Furthermore, there are possible
5. NUMERICALANALYSIS physical interpretations. The magnitude of the velocity is consistent with
In the present experiment phase shifts are produced by variation of the the magnitude required to account for the known anisotropy of background
position parameter x. If x is changed by Ax, the phase shifts are witnessed cosmic radiation at approximately 3 K and similar to the speed generally
at both PD and SWD as the passage of fringes. With the divisor form of believed to correspond to the orbital motion of the Sun in the Milky Way
variation in the speed of light, the number of fringes traversing PD is exactly Galaxy.
The result supports a model that involves direction variation in the speed
NeD = (vkx/c+ + v k x / c _ ) - 2Axle.. (5a) of light. Combining the divisor form of variation in c from (4b) and (4c),
the angle variation in 13 from (6b), and the definition of Vo from (7a), we
At SWD the number of fringes traversed depends on the inertial lengths have

Ax+ = Axc/c+ and Ax_ = Axc/c_. (5b) c( O) = c/[1 + (vo/c ) cos (0)]. (8)

With the divisor form of variation in the speed of light these are The experiment itself is a variation on the Michelson-Morley experiment,
differing only by the addition of features that permit investigation of the
standing wave in one of the two arms. But the original Michelson-Morley
Ax+ = ~c ( 1 + 13) and Ax_ = Ax ( 1 - 13). (5c)
experiment is generally believed to have demonstrated constant light speed.
The smaller of these limits the number of fringes traversed by SWD. We see This belief now needs to be refined: the original experiment did not demon-
strate constant light speed; it failed to demonstrate variable light speed.
NSWD= 2Ax_/~. = 2&c(1 -13)/X,. (5d) Reinstatement of possibly variable light speed creates a paradox for Ein-
stein's special theory of relativity, insasmuch as constant light speed is one
From (5a) and (5d) we see that NswD is smaller than NpD, and we infer of the postulates required by the theory. What, then, does a demonstration
that the fringes traversed by SWD are longer than the ones traversing PD. of variable light speed mean? Are there ways in which Einstein's theory is
This explains why variation of x causes the dither output at SWD to evolve inadequate? For example, does it describe exactly a limit case, and yet fail
with respect to that at PD. to describe any physical case, however closely approaching the limit?
The accumulation of exactly half a fringe of discrepancy between SWD Several reviewers have suggested to us that the existence of this paradox
and PD would move SWD from one side of a fringe to the other and thus may involve the tacit assumption that photon rest mass is zero. The elim-
produce dither phase reversal in comparison to PD. With (5a) mad (5d) the ination of this assumption opens the door to a possibly more satisfactory
condition for phase reversal is physics/I7) in which gravitation and electromagnetics can be successfully
unified, divergences can be removed from quantum electrodynamies, intrinsic
NpD -Nswo = (2D/X)[1 - (1 - 13)] = 1/2. (6a) spin of elementary particles can be better explained, and macroscopic mass
circulation can lead to observable electromagnetic effects.
Thus we expect the measured D to satisfy Although not discussed in Ref. 17, light speed dependent on propaga-

87
A New Michelson-MorleyExperiment

tion direction within the rotating galactic plane would qualify as such an here that many of the experiments represent generic types that have been
electromagnetic effect. It is only because of the assumed zero photon rest critiqued elsewhere by Tyapkin3 2~
mass that anyone even expects the same propagation speed c to apply in At the moment, all the candidate interpretations of the present experiment
all directions, regardless of the underlying rotation velocity.(18) Without zero seem rather radical, and we do not wish to draw any final conclusion. But
photon rest mass, underlying velocity should be manifest in a variety of ways. we can comment on the situation. Considering the historical importance
Reference 18 cites another recent experiment(19) as an example: there, a attributed to the Michelson-Morley experiment, it is very disturbing indeed
nonzero photon rest mass would produce a frequency shift that would be that additional measurements remain to be wrung from it after so long a
second-order in 13o. Since our experiment implies a 13o at the level of 10-3, time. More disturbing yet is the possibility that those additional measurements
one would look for a frequency shift at the (10-3) 2 = 10 -6 level. The admit interpretations that so challenge long-accepted beliefs concerning the
data do indeed seem to show such a shift. Reference 19 holds that the shift possibility of velocity detection by interferometry, the constancy of light speed
is spurious, but Ref. 18 holds that it could, in fact, be real, and now we see in all directions, and/or the physical nature of photons.
that if real, it is apparently consistent with our own results.
To the best of our knowledge, our own experiment is the first to measure Acknowledgment
an effect that is potentially first order in underlying velocity. There have The authors are grateful for technical discussions with H. Aspden, W. Brown,
been many previous attempts, but all of them have produced null results. In H. Hayden, G. Shaw, J.P. Vigier, and H. Yilmaz. In addition, a thoughtful
each case explaining why the result was null or how the experiment could anonymous reviewer contributed significantly.
be altered to reveal what was previously hidden requires a full analysis at Publication of this work was supported by W.J. Schafer Associates, Inc.
the level of detail here committed to just one experiment. While such an
exercise is beyond the scope of may single article, we can at least point out Received 31 August 1990.

R6sum6
Nous avons augmentd l'interf&om~tre de Michelson-Morley avec un instrument
suppldmentaire et fair des mesures compldmentaires de ~pe qui n'dtaient pas
prdMdemment prdsent&s par d'autre chercheurs. Tandis que l'expPrience originale de
Mic~lson-Morley btudiait seulement l'interf#ence entre les deux faisceaux qui avaient
d~'a traversds les deux bras de l'interfdrombtre, nous btudions, en plus, l'interfdrence
de l'onde stationnaire ?t l'int&ieur d'un bras. En supposant que l'interfdrombtre reside
dans un systbme de coordonne& inertielles, la thdorie de la relativitb sp~cialeprddit
des noeuds sdpar& d'une distance de s Notre exp&ience montre une d&iation de
cette sdparation nodal. Ce d&accord apparait d'btre consistant avec l'anisotropie de la
vitesse de la lumiPre et il prend la forme suivante: c(O) = c[ l+(v0/c) cos (0) ], o3
c est bgal d 3 x 10s m/s, Vo est d peu pros 378 km/s, et 0 =0 est orientb dans une
direction unique et relative aux btoiles stationnaires. La vitesse apparente Vo est ft la
fois en accord avec l'anisotropie connue de la radiation cosmique de fond et avec la
vdlocitb orbital du so~'l dam la galaxie de la Voie lact&

88
 E.W. Silvertoothand C.K. Whitney

References
1. A.A. Michelson and E.W. Morley, Am. J. Sci. 34, 333 (1887).
2. J.C. Maxwell, A Treatise on Electric#y and Magnetism, two volumes,
3rd rev. 1891 (reprinted by Dover, 1954).
3. T.E. Phipps, Jr., Heretical Verities:Mathematical Themes in Physical
Description (Classic Non-Fiction Library, Urbana, IL, 1986), Chap. 4.
4. A. Einstein, Ann. Phys. 17, 891 (1905).
5. A.A. Michelson, hstrophys. J. 61, 137 (1925); A.A. Michelson and H.G.
Gale, Astmphys. J. 61, 140 (1925).
6. M.G. Sagnac and M.E. Bouty, C.R. Acad. Sci. 157, 708, 1410 (1913).
7. H.C. Hayden and C.K. Whitney, Galilean Electrodyn. 1, 71 (1990).
8. E.W. Silvertooth, Nature 322, 590 (1986).
9. ldem, Spec. Sci. Tech. IO, 3 (1987).
10. E.W. Silvertooth and S.F. Jacobs, AppI. Opt. 22, 1274 (1983).
11. T.E. Phipps, Jr., Phys. Essays 1, 150 (1988).
12. B.A. Manning, Phys. Essays 1, 272 (1988).
13. C.W. Sherwin, Phys. Essays 2, 125 (1989).
14. H. Aspden, Phys. Essays 3, 39 (1990).
15. C.K. Whitney, Phys. Essays 3, 161 (1990).
16. 3.J. Fixsen, E.S. Cheng, and D.T. Wilkinson, Phys. Rev. Lett. 50, 620
(1983).
17. R.I. Gray, Unified Physics (Naval Surface Warfare Center, Dahlgren VA,
1988).
18. J.P. Vigier, IEEE Trans. Plasma Sci. 18, 64 (1990).
19. E. Riis, L.U.A. Andersen, N. Bjerre, O. Poulsen, S.A. Lee, and J.L. Hall,
Phys. Rev. Lett. 60, 81 (1988).
20. A.A.Tyapkin, Lett. nuovo Cimento 7, 760 (1973).

E.W. Silvertooth
Star Route, Box 166
Olga, Washington 98279 U.S.A.

O.K. Whitney
Tufts University Electro-Optics Center
Medford, Massachusetts 02155 U.S.A.

Also with:
WJ. Schafer Associates, Inc.
Chelmsford, Massachusetts 01824-4191 U.S.A.

89