For almost a year now, I have been in the uncomfortable position of having an idea.

However, there is one nice thing about this idea. It makes a concrete, exact and relatively easy to test physical prediction.

The idea predicts that there is a 11,187 m/s (Earth’s escape velocity) aethereal wind directly into the surface of the Earth at its surface.

I believe it would be possible to test this by performing a vertical variation of modern versions of the Michelson-Morley experiment (MMX) with one arm pointed in the vertical direction. (Modern MMX)

In 2003, Müller et al. performed a normal (2 horizontal orthogonal arms) modern MMX using cryogenic optical resonators that found a “possible anisotropy of the speed of light c, (of) 2.6 +/- 1.7 parts in 10^15” ( arXiv )

In a brief conversation with Holger Müller a Professor at Berkeley and the lead author of that paper, he stated that to this knowledge no one had ever performed a variation of the experiment using a vertical arm. He also mentioned that such an experiment would be complicated by the fact that gravity would slightly compress the length of the vertical bar making two equal length bars no longer equal in length.

“They haven't been done as far as i know. The problem is that any interesting physics signal would be hard to tell from a large signal from stretching of the arms under their own weight.” - H. Müller

I am interested in attempting to run this experiment myself. To that end, I have the following questions:

  1. Given current Physics understanding, is there any reason to expect that such a vertical variation of the MMX wouldn’t return the exact same results as all other MMXs, namely that there is no anisotropy in the speed of light?
  2. Given the complications mentioned by Professor Müller, are there reasonable methods available to overcome them? Especially considering the size of the effect (c + 11,187m/s vs c) is substantially larger than the accuracy obtained in his and similar modern MMXs.
  3. What is the order of magnitude cost of such an experiment? If I am to fund this personally, would such a project cost \$10,000? \$100,000? \$1,000,000? more?

Any insight offered on this topic will be greatly appreciated.

EDIT: For what its worth, for people looking at this years later. As I originally alluded to in a comment to [WetSavannaAnimal aka Rod Vance]'s much appreciated and comprehensive answer. While, I've become increasingly comfortable with the idea that there is a flow of aether or spacetime or whatever you want to call it into the Earth at 11km/s; I strongly believe that a vertical MMX would return the same null result as the horizontal one.

And I believe it will do so because of length contraction; that length contraction exactly masks any anisotropy of the speed of light by ensuring that the roundtrip time of light in any direction is a constant; that length contraction ensures Lorentz co-variance.

  • 1
    $\begingroup$ Material compression isn't the only problem: the gravitational redshift is more than a little significant in this arrangement. $\endgroup$ Feb 17, 2014 at 0:44
  • 1
    $\begingroup$ @dmckee Is the redshift reversible? Will a photon sent from the surface of the earth and bounced off a mirror in space back to the surface have a net redshift or will it blueshift back to where it started? $\endgroup$
    – aepryus
    Feb 17, 2014 at 1:11
  • 4
    $\begingroup$ Certainly it reverse coming back in the sense that the wavelength of the light is the same when it gets back onto the shared optical path, but it introduces a phase shift which is exactly what an interferometer is sensitive to. What I haven't done is attempt to estimate if this is something you could subtract out of the data. $\endgroup$ Feb 17, 2014 at 1:16
  • $\begingroup$ @dmckee Ok, thanks a lot for the feedback. I'll work on the calculation. $\endgroup$
    – aepryus
    Feb 17, 2014 at 1:19
  • $\begingroup$ aepryus: "In 2003, Müller et al. performed a normal (2 horizontal orthogonal arms) modern MMX using cryogenic optical resonators [... arxiv.org/abs/physics/0305117 ]" -- There (p. 2) it is claimed "In our experiment (Fig. 1), we use two $L = 3 ~ \text{cm}$ long COREs (cryogenic optical resonators)". How did Müller et al. measure whether (or to which accuracy) this setup condition was and remained actually satisfied throughout the trial? (Surely that's not only a worry in case in a "vertical variation" of the setup?) $\endgroup$
    – user12262
    Mar 2, 2014 at 19:11

2 Answers 2


Firstly, you need to calculate how much your hypothesised effect will change the optical delay in each of the interferometer's arms and check that you expect to see any result with your proposed experiment. Otherwise put: what are the specifications of the interferometer (arm lengths, light source requirements etc, vibrational tolerances) that will let you see your effect if it is real and are they reasonable?

I can see a major problem with your setup which you will need to overcome. There are several effects which I can think of which will influence your experiment. The first is the change in optical length of a given interferometer arm that arises through the gravitational redshift of the light propagating between different gravitational potentials when the arm is vertical (as verified by the Pound-Rebka experiment - see wiki page of this name as opposed to the absence of this effect when the arm is horizontal. This effect is small, but it can be precisely calculated from the Schwarzschild Metric and it is repeatable. So this effect is not a problem for you. A second effect is the gravitational tidal effect which is owing to the second order variation of the Schwarzschild Metric). Indeed, this is an extremely mild version of spaghettification. The interferometer responds to this by assuming a strained state - it is equivalent (in the Newtonian limit) to the strain required to counteract a variation in gravitational acceleration given by $3\,g\,\Delta/R$ for a vertical distance of $\Delta$, where $R$ is the Earth radius when the interferometer is in freefall. That is, a few micro-g per meter of vertical distance. Again, this is a repeatable effect.

An effect which is going to be vastly bigger and IMO almost impossible to account for is the change in an optical arm's length through the weight-induced mechanical stress on the arm as the arm of the interfermeter is rotated from horizontal to vertical. And the interferometer must be rotated as in the Michelson-Morley experiment as there is no non-interferometric way to compare the optical lengths of the interferometer arms. The solution is to swap the arms' roles (by rotation) and check for the effect thus.

So this really means your experiment needs to be done under conditions of weightlessness. Does your proposed effect still exist in freefall according to your theory? Otherwise, you will need to develop a repeatable, independent-of-light means of measuring the weight induced strain in your interferometer arms to hundredths of wavelength accuracy.

Any custom built interferometer system of the precision you will be seeking will easily eat up tens of thousands of US dollars / euros at the time of writing (2014) in the optics production and the mechanical alignment system. You will need to become highly adept at mechanical design and production of engineering drawings to get what you need. Add \$10K for either the acquisition of software to help you do this (e.g. Solidworks) or professional engineering help (in the latter case, add \$20K to \$30K).

Now, add the cost to put your experiment into low Earth orbit. The Space Exploration Stack Exchange Question "What is the current cost-per-pound to send something into LEO?" may help you. It would be a fair bet, from the figures quoted there, that your looking at \$25K USD / kg. So a ten kilogram interferometer satellite system (don't forget data telemetry) is going to cost you a quarter of a million USD / euros to get it to where it will work.

So all up, I'd say a budget of the order of \$400K USD / \$400K euros is looking like a minimum figure.

  • $\begingroup$ Thanks immensely for this answer. Actually, this idea predicts that the effect will go away entirely in freefall (physics.stackexchange.com/questions/130323/a-clock-in-freefall). Beyond that, I have come to realize that even with the 11km/s wind this experiment would also return a null result. My current thinking is that length contraction is entirely a result of aethereal wind and always precisely masks the wind, resulting in Lorentz covariance in all cases. $\endgroup$
    – aepryus
    Oct 28, 2014 at 16:01
  • 1
    $\begingroup$ WetSavanna is wrong about gravitational redshift. Indeed it does NOT affect light itself but CLOCKS. As everybody knows gravitational redshift is due to light frequency measurements using clocks placed at different heights. It only affects clocks. If YOUR clock (not the photons' one, there's no photon's clock in relativity) is slower then you will measure a blueshift, on the contrary, if your clock is faster, a redshift. But in the formation of the interference pattern there's no frequency measured with clocks/frequency meters, and this phenomenon is out of question. $\endgroup$
    – mfc
    Mar 24, 2018 at 1:44
  • $\begingroup$ @mfc But the Scwarzschild metric will tell you that there is a shift. Imagine it like this. Imagine the interferometer built in a square or rectangular frame: displacements along its sides $\mathrm{d}x$, $ \mathrm{d}y$ and lets build it in deep space. It has a rectangular shape measured with respect to the local metric, and because this metric is flat, $\mathrm{d}x$ and \mathrm{d}y$ commute. Now bring the interferometer into the gravitational field. Our spacetime is now no longer flat and indeed there is spatial curvature too. So the same infinitessimal displacements .... $\endgroup$ Mar 24, 2018 at 3:04
  • $\begingroup$ .... $\mathrm{d}x$ and $\mathrm{d}y$ no longer commute. So what this tells you is that the interferometer cannot both have its original shape and the same sidelengths. So it will be in a state of strain - although it will be very small. Its response to the metric will depend on its elastic properties so it will end up with different path lengths. Moreover, the relative pathlengths will depend on orientation. I haven't calculated how much this is - maybe it's too small to be a problem. In any case it is repeatable, so even if significant its not a problem - it's a systematic error that .... $\endgroup$ Mar 24, 2018 at 3:07
  • $\begingroup$ @mfc can be corrected for. The much bigger and hard to correct for effect is the weight induced strain. To put it another way, the relativistic effect I speak of is an extremely mild version of spaghettification, which is an anisotropic effect. $\endgroup$ Mar 24, 2018 at 3:08

An example of a similar experiment is the famous measurement of a gravitationally induced phase shift in a neutron beam by Colella, Overhauser, and Werner (often called "the COW experiment"). It's interesting to note that while there was an unambiguous gravitational phase shift, its size was not as predicted.


Not the answer you're looking for? Browse other questions tagged or ask your own question.