FAQ

The following criticisms were posted to the MAKE magazine blog.  They were directed at Dr. Osadchey’s first article in The Citizen Scientist e-zine, and therefore may not be applicable to our later work.  They are similar to responses we’ve received elsewhere, and so with heavy editing, I’ll use them as a start for an FAQ.   

 

The original comments are found at: http://blog.makezine.com/archive/2007/03/an_experiment_to_measure.html#comments   

 

I apologize in advance that some of the answers will read like self-righteous ranting.  This is no doubt due to neglect on my part.  With all the time I’ve been devoting to this project, I have barely even allowed myself the luxury of a self-published manifesto, much less surfing from blog to blog on the internet haranguing my critics and getting a reputation for myself, so now I have to catch up.  I can only offer the excuse that at least some of the answers are informative.  Plus, I rather enjoy the ranting and can’t freely indulge myself elsewhere.  I highly recommend that every kook consider the convenience of simply developing an FAQ to field snarky questions instead of taking time to smite all your foes personally.   

 

In the event your question is more reasonable, you are free to just skip ahead and start again whenever sanity returns. 

1.“Somebody didn't take undergrad physics.”   

“I think there is more to ‘warn’ people about; or rather, I think there is some responsibility to ensure there is something to it before posting.”   

“Anyone who went to the trouble to build the thing, and to mention it's contradiction of modern physics almost certainly knows it is hogwash. It is irresponsible to post such a thing with the title you have and not something with the word ‘hoax’ in it.”  Ad hominem arguments are generally neither true nor relevant, and these are not exceptions.  The claim that a measurement contradicts an accepted theory only describes the measurement, not the claimant, and the only responsibility of any party involved with publication was to ensure accuracy.  Any statement to the contrary is a falsehood.  Groundless accusations of a hoax serve only to stifle critical thought and analysis.  

2. “Barring some compelling evidence otherwise, the assumption on the face of it should be that this is hogwash.” 

There’s no need to assign malice where incompetence will do.  We are well aware that there is a good chance we just missed some perfectly banal causal agent – it’s been the motive behind much of our work to date.  We accept reasonable skepticism of our results as both inevitable and necessary.  In fact, it’s welcome.  Half the people who have replicated this experiment only started it to show Dr. Osadchey that he was just measuring his floor.    

3. “It is too easy for verified and unvetted science to become mixed [together] and presented to amateurs without the proper knowledge to either properly accept or reject with proper reasoning such an apparently bedrock shifting claim. Such an article should be only presented as process and craft in such a forum as this and not as science as it appears to have been highlighted.” 

The technical process and skills necessary to properly conduct an experiment are not somehow distinct from experimental science.  As for mixing verified and unverified claims, that’s a risk you take every time you read any scientific research that yields any result contrary to an accepted theory.  We’re not denying responsibility for accuracy and clarity on our part, but we’re not here to shield you from our experimental results, either. 

4. “There's a gigantic disclaimer on citizen science.” 

With good reason.  The usual social pressures keeping professional scientists in line are absent from our lives.  We are motivated by bright shiny objects.  This makes our claims somewhat less reliable than those of professional scientists.  However, being amateur scientists rather than professionals does give us one small advantage:  There are no social pressures to keep us from making discoveries, either.  We’ve no better opportunities down another research path.  Most touchy academics we deal with are only anonymous bloggers.  There’s no financial impact from wasting all our lab time chasing EMI for a year.  This won’t even get us on TV.  And at the end of the day, it’s only general relativity, not our job.  

5. “How likely is it that what is really being measured here is noise and heightened geek enthusiasm?” 

Random noise?  Less than 1/100 chance, or at least that’s what our statistical analysis says.  True, systematic artifacts can have statistical confidence, too, but we’ve done a lot of work to stamp those out, and so far our signal is still there.  Every new control experiment makes a systematic error just a little less likely.  Heightened geek enthusiasm?  About 50/50.  Our experiment is not yet complete.  It’s only been partially replicated by four independent researchers.  (Nobody’s done the whole thing yet, including us.)  We haven’t yet run all the tests we feel are necessary to confirm a regular variation with latitude (although all our data to date shows it).  We’re in the middle of two ongoing control experiments and start every week with the knowledge that last week’s data could prove we were just picking up WKRP all along.  And no one in the history of the world has ever successfully done this right.  So, our odds are better than average. 

6. “There was a test done that was several hundred thousand times more sensitive back in 1887. […]The test I'm referring to is called the Michelson-Morley experiment.”

“It looks like he's trying to replicate Michaelson-Morley […] but with less accuracy.“   No.  We are not replicating the Michaelson-Morley experiment.  We would need an interferometer to do that, and we’re not using one.  Our instrumentation is very different, and is not capable of measuring the same parameters as Michaelson & Morley’s apparatus.  The Michelson-Morley Experiment was indeed more sensitive than our own, though by less than a factor of 3.  And, like us, they reported an orientation-dependent variation.  Their result was superficially similar to what we are claiming, but with some crucial differences.  Their variation was interpreted as a change in the interference pattern (which we don’t think we’re seeing).  A Michaelson interferometer can only accurately detect components parallel to the light path that cause a change in the shape of the interference pattern – lateral deflection or aberration of the light creates drift and/or noise.  When your only tool is a hammer, all your problems become nails.  Likewise, when your only tool is a Michaelson interferometer, all your results are parallel to the light path.  Their result was significantly less than the predictions of the hypothesis they were testing (Fresnel’s classical luminiferous ether hypothesis of light propagation).  So – rather than claim to have discovered the special magic ether that only behaves like Michaelson & Morley says it should – they wisely accepted this variation as an error source and correctly declared a null result.  Later analyses of their data has borne this out, and their experiment was been successfully replicated with even greater accuracy by several other researchers.  

7. “Scores of physicists have done this experiment with better equipment and gotten the expected result.” 

No.  Scores of physicists have done variations of the Michaelson-Morley experiment or other interferometric tests of isotropic light propagation, and gotten the expected result.  In many cases, the accuracy of those results exceeds our own.  But we are measuring a different parameter – the lateral motion of our image central peak, not interference fringes.  That in itself doesn’t demonstrate an inherent superiority of our method – our equipment has errors, rattles and idiosyncrasies just like anyone else trying to push the limits of their instrumentation.  However, it does explain why we divorce our result from interferometric tests.  The two methodologies are different.  Most interferometers are not physically capable of accurately assessing a lateral motion unrelated to interference.  In fact, it’s a major source of error for them.  They dramatically improve their accuracy by actively editing out lateral motions, including our effect of interest.    It’s important to note that this works both ways.  Our instrument is not physically capable of accurately measuring the same parameters as an interferometer.  Therefore, we cannot credibly claim to have produced results contrary to any repeatable result in interferometry.  We are not contradicting Michaelson & Morley’s experiment because we can’t.  

8. “This is someone doing a crazy-crude reconstruction of a famous experiment, and getting different results.” 

Yes, but in this case the original experiment was the crazy-crude version.  Ernest Esclangon performed a conceptually similar experiment published in 1927, to test the classical luminiferous ether model of light propagation.  His hypothesis was a logical extension of the reasoning behind Michaelson & Morley’s hypothesis.  Just as the earth’s motion through a hypothetical ether was expected to produce a Doppler shift which would be detectable by an interferometer, so that same motion could be expected to create a flow field in the ether which would laterally deflect light in transit.  Can you imagine daring to use a micrometer and a piece of wire to measure deflections so small that they defy the limits of modern digital instrumentation?  Esclangon could.  Unfortunately, his imagination ran away with him.  He arrived at a result wildly in excess of the predicted value - a null result just like Michaelson & Morley’s, only in the other direction - but instead of accepting this, he decided that he rather liked the idea of a special magic ether that only does what Esclangon says it should, and he went with that.    Our two-dimensional measurements of deflection using the same light beam is a unique design variation that has enabled us to test the geometric consequences of Esclangon’s hypothesis.  This, along with our increased accuracy, is an important improvement on his version.    Our results to date contradict Esclangon’s hypothesis.  We are seeing a reproducible deflection that is not consistent with known causes of misalignment for this kind of instrument, but it does not match the predicted geometric consequences of light deflection in a classical luminiferous ether flow field, either.    The engineering behind Esclangon’s experiment is impressive, but on the analysis side it’s not much of a pedigree.    We take some comfort in the facts that our measurements are by far the more precise of the two, and that our experimental design was independently derived before we realized that M. Esclangon had attempted something similar 80 years earlier.  We’re trying to keep our evaluations more realistic, and we’re quite pleased to be getting a different result.    

9. “Shining a laser into a CCD imager causes blooming which makes the image pretty useless.”

We find no evidence of oscillating changes in the laser modes.  Thus, spreading of the central peak is an imager artifact.  It’s a function of intensity and optical characteristics, and can also be created by saturating the detector circuitry.  Once you’ve got the lens off of the photodiode to reduce lens aberration and internal reflection, this can be controlled quite easily by turning down the laser power and smoothing the laser power curve to reduce fluctuations.  We’ve got that under control.  

10. “The inherent noise in a CCD element would prevent this method from being even slightly useful. Yah, I suppose you could attenuate the light significantly but there is still that noise issue.” 

The image noise is atrocious – averaging around 1 micron error in every measurement, and up to double that for some instrumentation and settings - but it’s not so bad that it precludes useful measurements.  You can still squeeze a 3σ or better signal out of it.  This method has been used successfully by other researchers for measuring lateral deformations in mechanical systems, with similar accuracy.  The only difference here is the application. 

11. “Hmmm... don't I seem to recall that one of the starting principles of the special theory of relativity is that one *cannot* determine one's own velocity in a reference frame moving at constant velocity?” 

Yes.  We have retained the title “An Experiment to Measure the Absolute Motion of the Earth,” because that was the experimental goal of Dr. Osadchey’s original experimental design, and the various articles published to date are steps in that project.  However, by this point, it is only a title.  We have established that our results do not conclusively demonstrate an absolute reference, and therefore do not refute special relativity.  Our results to date still allow for an astronomical phenomenon as the cause of our observations, and this phenomenon is a useful orientation reference.  This presents a challenge to general relativity because it suggests the theory is incomplete.  There is no term in the gravity tensor equation of general relativity that would account for our observations without a cause specific to the instrument or otherwise restricted the to Earth’s surface.  Thus, it is increasingly fascinating that we can’t find one.  

12. “Of course, the Earth's velocity isn't constant, so perhaps that's the point here?” 

No.  We are not seeing a 1:1 linear correlation with any known velocity vector describing the lab frame.  We have a lot of speculations, few of them tested and many of which won’t be tested during this phase of our experiment.  We don’t want to introduce more than we can test for.  At this time, the only point is trying to isolate potential causes of the signal we are observing, then manipulate that signal to prove it.   One of those potential causes is a vector field of constant direction in heliocentric space, but it’s just one among a set right now.    

13. “If I understand his setup, he may be measuring the flex in his apparatus.” 

We thought so, too, so we checked it.    Measuring beam bending is one of the more conventional uses of this type of measuring system, so our instrumentation was suited to the purpose.  Our apparatus is light enough that we can flip it on its side or upside down to measure beam flexure directly.  So we ran entire trials upright then flipped for comparison.  Isolation of the support beam’s two horizontal axes of rotation was accomplished by suspending it so that it would always hang in the direction of gravity.  We rebuilt the mount with different designs and materials.  We varied the position of the apparatus, including relocation during the same trial using a wheeled platform, and at different locations over distances up to 300 miles.  We found nodes in the daily variation where the nominal variation on at least one axis was effectively zero, then went looking for beam flexure at those times.  Flexing of the support beam does occur, but is most commonly observed as a bump rather than a smooth curve.  On average we are keeping it at remarkably low levels – less than 2 microns mean deformation over the course of each trial, too small to separate it from the camera’s inherent image noise without control experiments, and too small to account for the effect of interest.  Trials with large geographic separations between them have shown regular changes in the magnitude of the observed deflection, and observations conducted at stations up to 1400 miles apart are roughly in phase.  Mechanical flexure doesn’t account for that, either.

14. “There are only 3 clear images of an actual setup.”  

If you’ll skim our subsequent articles, you’ll find several additional photos, schematics and even a parts list, along with discussions of the intricacies of the instruments’ construction on our web sites.  Between the articles and referenced web sites, we’ve published and posted enough information for readers to build an equivalent instrument themselves.  If you want to understand the layout, we think there is enough in the combined body of work to do so.  

15. “Assuming a 4x4 piece of wood is rigid enough (big assumption), the CCD (2 Mpx camera) isn't attached directly to that wood, but is on 3 hunks of 2x4 that look hand cut and screwed down, one on top of the other. And the camera looks to not have a remote trigger attached. This means he would have to be touching the camera for each event.”  

Yes, early versions of the apparatus had all these problems, right down to the experimenter’s thumb on the camera switch.  Needless to say, there have been several improvements over the past year.  We’ve both constructed sturdier setups with much less self-weight flexure, and a remote trigger was easy enough to rig.  We’ve addressed every one of these structural and procedural concerns along with several others.  And we are still getting the same result. 

16. “Been there, done that, to considerably more orders of magnitude.

“Back in grad school the lab down the hall had a rotating assembly with two atomic clocks and a microwave interferometer performing exquisitely precise measurements of any frame drag or the like.  Surprisingly, they found a statistically significant variation. The disturbing bit was that the asymmetry was locked to the frame of the earth and not to sidereal motion. One of my friends then waved his Jameco magnetized screwdriver over the mixing ferrites and the speed of light suddenly dropped. The instrument was acting as a giant compass!

“The moral is that there have been some incredibly precise measurements of GTR [General Theory of Relativity] (any secondary effects would be an express ticket to a Nobel) with no variations yet. But on the other hand systematic artifacts can be very easy to encounter.”  That’s not surprising.  Scientists have been seeing unaccountable oscillations like that one in high sensitivity tests of isotropic light propagation for more than a century, so it makes sense that one would show up in a similar test of relativistic frame dragging.  Waving a magnet at the equipment and seeing a change would indicate that magnetic field influence on the instrument circuitry is a potential cause of the deflection we’re observing.  We’ve tried something similar without result – it sounds like the artifact you’re describing was instrument specific – but it was important to check.  Similarly, we have investigated the effects of heating, refraction, flexure, electromagnetic interference, and other potential error sources.  We’re not done with it yet, but we’ve shortened the list of causal agents considerably over the past year.   We think that what we are observing is a common component of these artifacts.  Michaelson interferometers, along with most other types, can’t measure this lateral deflection accurately and function better when they edit it out.  It is only a systematic error in interferometry, so it’s often assumed to be nothing more for any other type of instrument. And it doesn’t fit anyone’s theory.  So, it’s been largely discounted as nothing more than an annoyance, or pressed into service to support pet hypotheses, with no one really testing it to determine what it is.  We don’t think that just finding a gap in the general theory of relativity is enough to justify the Nobel Prize all by itself.  To earn a Nobel Prize, we’d have to discover a new and potentially useful physical phenomenon, not just a discrepancy.  Needless to say, we’re working on it.  

17. “As a rule, the existing laws of physics hold, but each revolution refines the laws of the previous one.” 

Not yet it doesn’t.  We’ve got some good guesses.  However, you need to understand that we believe this effect has been reported and held up as a “revolution” several times before by other scientists, both amateur and professional, and none of that panned out.  We’ve been able to confirm that the displacements are real enough.  But the single unifying theme of all prior “revolutionary” claims has been unbridled speculation.   Most previous reports we’ve seen had either tied it in to their pet theory by now or dismissed it as an artifact without properly identifying it, and every one was judged based on those unjustified additions.   We’re having fun speculating, too, but we’ve tried to keep our descriptions strictly empirical, and postpone the revolution.  

18. “I'm aware that the experiment is what it is, but as Shawn noted it's a good to see someone trying this themselves -- that's part of science. I think I'd rather have people attempt to do things as opposed to just accept everything. That said, I wouldn't try this one :)” 

Rational skepticism is the correct initial response for you, and by far the best for us. So far, there have been three attempts at replicating aspects of this experiment by independent researchers in addition to Dr. Osadchey, all of them skeptical about one or more elements of Dr. Osadchey’s methods, design, or analysis.  What does any skeptic worth their salt do when making an honest attempt to replicate an experiment in which they think some banal artifact is being misinterpreted as a positive result?  They make changes to the experimental design in order to eliminate or control for that artifact.  They work at improving the instrumentation and driving down the error rate.  They run tests that the original researcher doesn’t have time or sense to run.  They don’t trust it when they happen to see the same results, but try again a different way just to be sure, all the while adding more data.  What does a true believer do under the same circumstances?  Build an exact replica of our instrument, see a fluctuation or two, applaud, declare victory, and go home satisfied.  There’s a problem with satisfaction, though – the data stops.  Complacency and acceptance don’t give us any more dots on our graph.    It doesn’t take much to realize which group made the greatest contributions here.    Send more skeptics!  

Reader Comments and Author Responses from The Citizen Scientist, 4-06-07

 

 The following comments are taken from The Citizen Scientist e-zine.  They are relatively informative, and present a clear picture of some actual problems with accurately interpreting results of our experiment.The following comments were made by George Hrabovsky.  The original comments appear here:  http://www.sas.org/tcs/weeklyIssues_2007/2007-04-06/project2/index.html   These comments were directed toward the same earlier article by Dr. Osadchey that was discussed on the MAKE magazine blog.  Note that Dr. Osadchey and I use different experimental plans, with a common experimental design for simultaneous control experiments.  Where our designs are dissimilar, I have cited my own work unless otherwise stated. 

1.  
   1. “Reversing the method of calculation as written, the drift measured was approximately 33.5 microns in 10 nanoseconds. This is a beam accuracy of 33.5 microns over 3 meters, or around 11 microns over 1 meter, or about 1 in 10,000. This is an astonishing level of accuracy. This is like shooting at a target a mile away and coming within six inches every time.”  33 microns is a typical amplitude of the oscillations seen by Dr Osadchey – not a wander or cumulative drift.  There is also a drift component, but it’s often less than a third of that (depending on the allowed warm-up time) and easily separable when the data spans multiple turns per trial.
      We could never achieve the kind of precision necessary to hit a target a mile away within six inches using our aiming system.  Our instrument is limited by beam width (the laser beam would spread to six dozen inches over a mile), and stability problems that are minor for our single, short semi-rigid platform become nearly insurmountable when trying to align multiple non-rigid platforms separated by a distance of a mile.  Fortunately, increased precision becomes easier at smaller scales.    The displacements being measured by Dr. Osadchey are about twice the thickness of a kitchen trash bag.  That measurement does require a very high degree of precision, but nothing beyond the reach of a good micrometer.  As for aiming, it is important to note that our camera photodiode chips average 6 mm wide.  They are clearly visible to the naked eye and an easy target for a screw adjusted laser mount.  Thus, the main concern for aiming accuracy is the ability to maintain alignment, not to attain it in the first place. 
   2. “I see no tests of the accuracy of the aiming system. This leads to the question of whether or not the data can be extracted from the mechanical noise of the experimental apparatus. I would like to see some sort of error analysis for the system.”  Yes, an error analysis is very important for evaluating this experiment.  The combined contributions of mechanical and electronic noise sources are relatively large compared to our signal.  We do not use a stabilized source. (Stabilization of lasers is often accomplished by editing out lateral drift using feedback from the detector, which edits out our effect of interest right along with it.)  That means we are subject to beam wander and jittering.  We’ve found that the beam wander (i.e., laser drift) can be approximated by a fitted polynomial function.  A second order polynomial has proven adequate for short duration trials (< 30 minutes).  Longer trials suggest that a fourth-order polynomial function may be a more accurate approximation on long timescales, but an accurate fourth-order fit isn’t practical for sample sizes smaller than n = 5.  There is a distinct daily variation of amplitude, and our average noise levels have varied over the course of the experiment as we refined our experimental designs, so we have inevitably produced some individual trials for which the signal does not rise above the computed experimental error for that trial.  However, these are a minority which tend to be more frequent in our earlier data sets and often correspond to daily minima.  Data sets with amplitudes of 3 times the experimental error or more are more common.  Dr. Osadchey’s data sets frequently achieve signal-to-noise ratios of 5 or more. 
   3. “I would also like to see some sort of data on the relaxation time of the elements in the CCD. Since we are talking about 10 nanosecond time intervals, I doubt that the CCD elements are able to clear within that interval. I would like to see some data about this phenomena, since the laser is exciting the elements that the beam actually strikes; this causes a lot of heating, too. There is no real estimation of these effects on the experiment.”  10 ns is not the applicable response time.  The transit time from source to detector is within an order of magnitude of that, but we’re only measuring the final lateral deflection at the detector.  The cameras used typically have very slow frame rates – 0.01s or more.  If they were attempting to photograph a subject with a fast relative motion or rapid fluctuation, the image would blur.  Circuit reset times are even longer – up to 5 seconds after each photo.  We’re incapable of measuring any fluctuation on the order of 10 ns (or even 1 ms), and fortunately it is not required.
       
      It is important that the apparatus be stationary during each measurement.  We strive for a kinematically determinate regime for the instrument mechanism, and experimental error is computed on that basis.  We typically allow from 5 to 40 seconds between measurements, in order for vibrations to damp out between photos.  We’ve varied that, including changes during individual trials.  Shorter damping intervals allow vibrations and pendular oscillations to persist, with correspondingly higher error rates.  With intervals in excess of 2 minutes, drift begins to overwhelm the effect of interest. 
   4. “How precise is the beam size for the laser source? Does it change in size? Could this change in size be within 34 microns over 10 nanoseconds? Assuming a 5 mm average diameter, 33.5 microns would be about 3 in 50, or 6% error in the beam size. For a laser pointer this does not seem unreasonable.”  The laser beam can indeed change in size over the course of each trial.  We are also seeing considerable beam wander.  However, these can both be reduced dramatically by allowing proper warm-up time.  Drift rates and beam shrinkage typically fall by 2/3 within 30 minutes of power-up.  Drift and shrinkage over the entire trial can be reduced to less than the effect of interest by extending warm-up to more than 2 hours, and keeping measurement intervals to less than 1 minute. 
   5. “What are these lenses that are being used and how accurate are they?” The only optics in our final design are the laser diode, the light path air and the camera CCD photodiode.  There are no other intervening optics, which saves us the trouble of characterizing them.  Early setups had attenuating filters, but we have found that turning down the laser power is equally effective at attenuating the beam.   
   6. “Unfortunately, without these factors being accounted for there is no reason to accept these results as anything other than experimental noise. I would like to see a careful analysis of the apparatus, including the mechanical stability of the system over the time intervals being discussed. Perhaps a photograph of the apparatus would also be good.”  We’ve looked at enough noise and systematic errors from our own instruments that we have no choice but to agree.  That’s why the current phase of our experiment has focused on accounting for instrument specific artifacts.  We’ve also tried to post enough information at our web sites to characterize the equipment and reproduce the experiment if desired – including photos. 
   7. “Let me be clear, I doubt that this experiment will yield positive results in the light of careful data analysis. Unfortunately, no such analysis has been performed. It is the responsibility of the experimenter to provide such data before their work can be considered as acceptable. After all, the point of an experiment is to control as many factors as possible and make accurate measurements. Accuracy is far more important than precision, in my opinion. After all, you can be precisely wrong.”  This was a legitimate criticism of earlier versions of this experiment.  However, we have subsequently performed basic data analysis and made it integral to our current experimental design.  (That is, after all, the responsibility of the experimenter.)  This analysis has increased our accuracy, and supports the claim that we have a non-random signal with a small but acceptable (and growing with improved instrumentation) signal-to-noise ratio that varies over time and position in a pattern consistent with an astronomical phenomenon.    Unfortunately, improved data analysis will not prove the claim by itself.    There is no shortage of meta-analyses concluding that something analogous to this effect has been observed before with varying degrees of statistical confidence in interferometer experiments using unstabilized sources.  These inspire curiosity and are consistent with our results, but they don’t have much value as evidence, because meta-analysis cannot differentiate between a real signal and a systematic error.  No result is acceptable without sufficient statistical confidence, but false signals can have statistical confidence, too.  Control experiments are necessary to isolate systematic errors and noise sources – a process just as vital as error analysis.  Without sufficient control experiments to eliminate alternate causes, our results won’t prove our hypothesis.  Therefore, we are encouraged by the initial results of our data analysis, but not placated by it.

 
The following comments were submitted to the Citizen Scientist by Aaron Kammerer in the same article as George Hrabovsky’s comments above. 

1.  
   1. “First, according to the work of Larry McNish at the RASC Calgary Centre, the earth’s total velocity through space is more like 600 m/s with respect to the "stationary" cosmic background radiation. He also has a table of the various velocities that states that the earth’s rotation contributes between 0 and .46 km/s, the earth’s orbit around the sun roughly 30 km/s, and the sun’s orbit through the galaxy at 200 km/s. “One can see that the dominant contributors dwarf the earth’s orbital and rotational velocities. Instead of the 6.7 km/s and 0.1mm CCD deflection measured by this experiment, it would seem that a much larger deflection (100 times as large) would be seen if one were actually measuring the earth’s total velocity.”  Not as large as 100X, but otherwise the point is valid.  We are not observing a lateral deflection nearly large enough to correspond to 28 km/s motion through a classical luminiferous ether.  The roughly 7 km/s value being computed by Dr. Osadchey does not support the classical luminiferous ether hypothesis, even after correcting for a possible sinusoidal variation with latitude.  For this reason, we have been forced to abandon Fresnel’s ether hypothesis.  Some of our other results also contradict Dayton Miller’s entrained ether hypothesis.  However, we are still observing the deflection, which is still behaving as we would expect an astronomical phenomenon to behave.  It’s still an orientation reference.  And we still want to know what it is even if it is not a pet hypothesis.   
   2. “As for the apparatus, it seems plausible that by rotating this rigid beam during the experiment, one could introduce a ~0.1 millimeter deflection between the two ends of the beam, which would than account for the difference in the positions of the incident light on the CCD.”  This is entirely plausible.  Sufficiently large weight shifts due to floor slope are theoretically capable of mimicking our signal.  Therefore, it was necessary to conduct control experiments to check for flexure of the support beam, mountings and other equipment.    The effect of interest is the same regardless of how the beam is supported (padded mount, suspended, left side up, right side up, upside down, etc.)  It has the same magnitude regardless of how much time is allowed for soaking and sagging.  The deflection maxima have “nodes” in the lab frame – periods at regular time intervals when the magnitude decreases to near zero – when flexure can be examined by other methods.  We do see flexure.  But it does not mimic our signal.  Either this is the special magic flexure that only behaves the way we want it to, or we are not observing mechanical flexure.  
   3. “One suggestion would be to leave the beam in one position and take measurements over a 24-hour period. In this way, the rigid beam is more stationary, but the rotation of the earth will provide measurement in two dimensions over that period of time. Adding a second beam perpendicular to the first and measuring through 24 hours would provide data in three dimensions. Given that the earth’s rotational contribution to the total is so small, it should not have much effect on the measurements.  “By eliminating the need to rotate the apparatus, one can eliminate the need for a rigid beam and instead just require a solidly anchored light source and CCD on opposite sides of the room.”  This is a logical variation, and has been attempted by three of the four independent experimenters who have worked on this experiment.  Unfortunately, the beam flexure that this is meant to compensate for is not the only source of error in our experiment or even the most important from a design perspective.  Thermally induced laser drift is the major source of error in our experiment.  The warm-up period necessary to maintain laser drift below the magnitude of the effect of interest is quite large even for a short duration trial on a rotating platform, and increased warm-up time is not effective for trials that last several day-night cooling cycles.  Without some means of temperature control, the laser central peak can drift right off of the beam profiler’s CCD in the course of a single day.  Daily (even hourly) temperature variations can be much larger than the effect of interest.  Without active temperature control, they can completely swamp it.  The effects of these variations are apparent even in stationary trials with good temperature control.  A roughly sinusoidal daily variation was observed in a month-long experiment conducted in a thermally stable environment by Hayden Brownell, as we expected.  However, that sinusoidal variation can only be reliably isolated from the first half of that data, and there is no means of determining from the data alone whether the sinusoidal variation was thermally driven.  Thermal noise washed out the last 14 days completely, even with relatively precise temperature control.  Stationary experimental setups soak up a lot of heat over the course of one day, even in thermally controlled environments.   Rotating trials are subject to the same heat transfer problems, but varying the rotation rate allows us to control the heat transfer rate for short periods, providing a basis to control for thermal effects.  Additionally, rotating trials have a recurring reference once per turn that can be used to determine thermal drift.  Stationary trials provide no inherent reference for separating thermally driven laser drift from the data, and there is no easy way to prevent it without editing out the signal along with the drift.   
   4. “Further, extending the length of the light path would also improve the signal to noise ratio.”  Yes.  We haven’t had a lot of experience with beam lengths in excess of 10 feet for rotating trials (that being the largest size of construction material that is convenient to haul home from the hardware store), but a set of experimental trials by Lance Osadchey using different beam lengths from 2 to 20 feet suggests that the magnitude of the deflection varies roughly linearly with the path length with comparatively little change in the experimental noise.    

Other questions have been posed by other sources during the course of this experiment.  Some of these are:   

1.  
   1. “Do you know what you are detecting?”  No.    What we are measuring is a change in laser central peak position over a fixed path length.  This change appears dependent on orientation in heliocentric space and varies regularly with time and (we suspect) latitude in a manner suggesting an interaction with an external vector field.  It might be a reliable orientation reference, and we are considering a continued set of experiments to confirm its global extent.  We are reasonably confident that it is not caused by interaction with a classical luminiferous ether.  There would be geometric consequences of an ether flow field that do not occur in our experiment.  But we can’t find a known cause for it, either, despite numerous control experiments.    Right now, we’re down to two remaining candidates: interference from the naturally occurring global electromagnetic field, or a failure of the modern construct of General Relativity.  Both could potentially yield discoveries new to science.   
   2. “Has anyone ever reported this before?”  Yes.  Mechanical misalignments and laser drift have long been acknowledged as error sources in interferometry.  At the level of precision of which some modern interferometers are capable, even microscopic deformation of a critical element can generate measurable systematic errors.  We see examples of these error sources in our own work.    However, we’re not using an interferometer.  There are only two optical elements in our system, which eliminates a lot of potential misalignments and makes the remaining error sources easier to control for.  Our instrument is not as sensitive as some interferometers, but because of this, it is also far less susceptible to certain types of noise.  Based on experience with our own simple optical system, we’ve also come to suspect that interferometers are not as susceptible to certain types of mechanical misalignment as is claimed in some references.  The computed misalignments in these cases would be misleadingly large because there is a contribution from our effect of interest that is not accounted for.  We’re aware of at least two other sets of published experiments conducted using a rotating beam profiler or interferometer setup that may have directly measured the same effect we are observing. Our results suggest that these researchers misinterpreted their data due to the lack of some key control experiment data, leading them to an incorrect description of the observed behavior.  Thus, we sadly cannot agree with either their result or conclusion.  The daily variation observed by Hayden Brownell using a stationary beam profiler confirms results previously reported by other researchers.   We are not the first to see this phenomenon, but, as far as we know, ours might be the first accurate description.   
   3. “How do you know this isn’t caused by electromagnetic interference?”  Because we tested for it.    EMI induced mode changes in the laser diode (and consequent shifts in the laser interference pattern at the detector) were a plausible mechanism for the observed deflection.  In control experiments, feedback from artificially induced electromagnetic interference was capable of mimicking the observed effect.  EMI is also theoretically capable of accounting for the fact that our signals are in phase at different stations.  We can eliminate most obvious sources.  For example, we have included filter circuits in instrument power supplies to filter a range of EMI induced frequencies.  Battery power supplies reduce constructive interference.  A grounded metal mesh Faraday Cage protects the instrument from electrostatic fields.  Controlling for characteristic variation can rule out the dominant 60Hz and 1 MHz sources measured in our lab.  (These EMF sources lack the necessary daily phase variation to account for solar time dependence, and attempts to artificially induce EMI at these frequencies indicate that the systematic error produced would have a 180 deg period rather than the observed 360 deg.  Artificially introduced magnetic fields have no effect, either.)  And use of 6” steel pipe in the support beam construction can provide partial shielding against the lab’s magnetic field (although we have since found wood to be a better construction material because of its greater availability, better thermal stability and lower self weight deflection).  These controls and precautions improve our signal-to-noise ratio, but do not increase the signal.  Unfortunately, no single one of these measures can eliminate all EMI at all frequencies.  Any EMI countermeasure can potentially leak.  So, it was still possible that our shielding simply is not effective at the frequencies that generate modal changes in the laser.  However, to date we have been able to eliminate artificial sources as well as the Earth’s magnetic field as causal agents.  That means that even if this was EMI, we were looking at the effect of a natural phenomenon.  We also controlled for EMI by substituting our equipment to vary equipment specific parameters.  This had no affect on the signal of interest, even when substituting lasers of different wavelength (and thus different modal response).    We can now discount EMI as a causative agent.  We’ll miss it.  Because of the application of beam profilers as measurement tools, EMI as a causal agent would have important implications to engineering.  The regularities inherent in our observed signal can be compensated for, improving measurement accuracy by up to an order of magnitude for instruments of this type.  That would allow beam profiler accuracies comparable to that of (much more expensive) interferometers.  The effect is also a useful orientation reference.  (Dr. Osadchey already has a patent.)  Confirming EMI as the cause would have opened up several potentially valuable research paths.  If it’s not EMI, then that opens up a world of research paths, too, because we’ve been able to eliminate all other known causes of misalignment for this system.  Conclusive demonstration that this effect is not EMI makes a failure of General Relativity a plausible explanation for our observations.  That might throw modern physics on its ear, but it had better get right back up again, because at that point the potential applications would expand into the realm of science fiction.  Needless to say, we are still satisfied with this outcome.   
   4. “Why do you say that your signal is behaving like an astronomical phenomenon?”  Because we are obtaining data on more than one axis, we can compute a three dimensional direction of a vector causal agent in various mathematical models.  The choice of model is arbitrary, but is still constrained by the results of our control experiments.  We can further constrain the model by insisting only on models for which the computed vector does not rotate during single turns of the instrument.  That allows us to say which direction our causal vector is pointing.  And a common result is a vector close to the astronomical position vector of the sun or at approximately a right angle to it.   We appear to be tracking the sun.    This produces a solar time dependence on one axis and what appears to be a sidereal time dependence on the orthogonal axis.  It’s important to note that this claim isn’t solely based on time dependence but on a computed vector.     
   5. “Why don’t you bring this to the attention of an expert in optics?”  We have.  The general opinion was that we were probably only seeing something new to us, not new to science.  Thus, the majority of professional scientists we consulted were not interested.    There were notable exceptions.  Hayden Brownell, then at Dartmouth University, ran a trial for Dr. Osadchey with the expectation of showing him that we were only seeing a systematic error.  (His trial ultimately yielded results consistent with our hypothesis using a different methodology than our own, and he provided other helpful advice for our project.)  His level of involvement was welcome but unusual.  Not everyone was as polite in their responses to our requests.  Dr. Osadchey can now truthfully say that he has been threatened by a member of the establishment trying to quash his hypothesis. (The “threat” was a meaningless professional censure – it’s not like they told him some guys were coming – but it’s a great icebreaker with the other kooks.)  Other than these two unique and exciting episodes, seeking expert help was unproductive.    It was important to try to increase our available pool of expertise, and we were lucky enough to encounter a professional scientist genuinely motivated by scientific curiosity. As a rule, however, it’s unrealistic to expect a professional to do critical groundwork for you if you’re not paying them.  All scientific curiosity aside, this is as true for professional physicists as it is for professional carpenters.  We were not entitled to expert advice.  Reproducibility only becomes a factor after there is a reliable and accurate result to reproduce, and getting there is the job of the experimenter.  We had to take responsibility for our own work.   
   6. “What value are you measuring for the absolute motion of the earth?”  We have no proof that we are really measuring any such value, just computing an equivalent based on our results.  However, we’ve found that this is still a useful parameter because it provides a convenient comparison for our measurements and  those of other researchers who have conducted experiments to test various luminiferous ether hypotheses.  Dr. Osadchey is computing up to 6.5 km/s ± 25% at his station.  I am computing 1.5 km/s ± 50% maximum at mine, with increases to 2.2 km/s ± 40% for measurements made at 3deghigher latitude.  Our measurements have never agreed in magnitude, but have maintained the same approximate ratio between stations  Our current estimate of change rate with latitude is crude due to lack of data, and is nearly twice as steep as the sinusoidal change with a 23 deg offset that a linear flow field model predicts.  However, based on that model, we can still project a maximum of 8 to 14 km/s with the instrument laser perpendicular to the field.  None of which is very informative in itself, since our control experiments have already established that we’re not dealing with trajectory deflection in a flow field.  However, a number of other experiments testing the luminiferous ether hypothesis of light propagation have reported magnitudes in this range – enough that we are not ready to dismiss this as coincidence.   
   7. “How do your results compare to the Pioneer X Anomaly and other observed discrepancies in space probe positions?”  We can’t extend our result to the Pioneer X Anomaly at this time.  There are no measurements to confirm this effect in interplanetary space, so any direct comparison would be meaningless.  We’re already speculating as fast as we can.  However, the Pioneer X Anomaly does set limits on the long range behavior of our causative agent, with the largest magnitude case having Pioneer X exactly where General Relativity says it should be.   
   8. “How does this contradict General Relativity?  Why not Special Relativity?”  This experiment started out as a test of Special Relativity.  Contrary to popular conception, Special Relativity is mathematically simple.  Basically, the theory starts from two assumptions: #1) All motion is relative, and #2) The speed of light is the maximum velocity in any reference frame.  Everything else involved - down to the last equation - is taken straight from classical physics, and the theory is derived by applying correction factors to the equations of classical physics to make assumptions #1 and #2 happen.  So, to refute Special Relativity, you should ideally refute assumption #1 or #2.  (You could also try refuting practically the entirety of classical physics, but that is beyond the scope of our experiment.)  We have no means whatsoever of assessing the speed of light with this instrumentation, so that leaves the assumption that all motion is relative.  And that’s what Dr. Osadchey initially went after.  We have been able to conclusively demonstrate that the model of absolute motion that we were testing (based on Fresnel’s classical luminiferous ether hypothesis) does not hold true.  Further, the magnitude of the signal that we do observe is large enough that descriptive models with an absolute reference but not based on a classical luminiferous ether will predict observable astronomical consequences that do not exist.    We must report a null result for that portion of our experiment.  We’re seeing a reference, but it’s not absolute, so Special Relativity passed this test.  To understand how General Relativity could be called into doubt by our results, it is necessary to recall that “null result” does not mean “zero result”.  We still have a non-zero, statistically significant signal that remains unexplained.  Right now, establishing the cause of our signal is still an engineering problem, but we can already eliminate most of the known possibilities.  We’re down to causal agents that affect the light itself.  General Relativity allows for many agents that cause anisotropic (i.e., direction dependent) behavior of light in a non-inertial reference frame, including gravity and acceleration.  However, there is nothing in General Relativity that can explain the behavior we are seeing, which suggests that the current formulation of the theory is incomplete.  General Relativity might not be so general after all.

Our observation does not contradict any fundamental postulate of General Relativity, but is merely unaccounted for by the theory.  We speculate that the most likely shortfall in the General Theory of Relativity’s would be its treatment of angular momentum, since our measured deflections are by definition anisotropic changes in the angular momentum of the incident light, and would in fact be a local violation of the conservation of angular momentum if we force them to fit the current version of General Relativity without the assumption that they have an instrument specific driver (which we are not finding).  Einstein’s original formulation of General Relativity did not thoroughly define relative angular momentum (e.g., what rotating objects are supposed to be rotating relative to and why angular momentum should be a conserved quantity for relative rotation).  A detailed treatment of angular momentum in a relativistic framework was later added by other theoreticians, but this has not been tested with the same rigor as some other elements of General Relativity.  If no one can demonstrate an alternate cause for our observations, then this patch on General Relativity might have a hole in it.
 

It’s an interesting guess.  Perhaps it’s even correct.  Only continued experimentation will tell.