"The influence of the crucial Michelson-Morley experiment on my own efforts has been rather indirect. I learned of it through H.A. Lorentz's decisive investigations of the electrodynamics of moving bodies (1895) with which I was acquainted before developing the special theory of relativity . . . What led me more or less directly to the special theory of relativity was the conviction that the electromotive force acting on a body moving in a magnetic field was nothing else than an electric field." -Albert Einstein

The very crux upon which relativity stands! Regardless of the repetitions which followed, this is the experiment upon which the second postulate of special relativity is based. While Einstein claimed at one point, many years later, that he was unaware of this experiment at the time of his development of Special Relativity, there is no question that the Lorentz equations so central to SR were developed by Hendrick Lorentz specifically because of this experiment. Those equations require light speed to be constant to have any meaning at all whatsoever. Without this experiment there was no reason to postulate light speed constancy in the first place.

There exists the claim that Einstein's idea of light speed constancy came from Maxwell's Equations and that they somehow required frameless constancy. This is utter nonsense since Maxwell formulated his equations based upon the assumption of an ether with a permittivity of permeability. (still in use today) It is tantamount to saying that Galilean Relativity was Einstein's motivation for postulating light speed constancy. However Galilean Relativity assumed carrying a medium along with the experiment. IE: While Galileo's example was in a ship, we'll take sound inside a jet as an example. It travels at the speed of sound in the jet's "frame" because the medium (air) is taken along with it. However, if a hole is cut in the front and back of the "ship" so that air can pass through, Galilean relativity no longer applies. At the time of the inception of SR ether was nearly one-hundred years of scientific fact and this medium would behave much like air. There is absolutely no inspiration for light speed constancy to be found in Maxwell's equations or in Galilean relativity at all. The only inspiration would have to come from pure imagination of a counter-intuitive process with no example or use in human experience or it had to be this experiment. Hopefully this makes the answer clear enough for you.

The "null" result of this experiment is what gives credence to time dilation and the host of other paradoxical properties of relativity. This experiment was the only reason SR was given any consideration by the scientific community of the time. Unfortunately, the experiment was very very far from null. It was smaller than expected but the ultimate answer, the speed of the ether wind, was only one-third what was expected. Because of the way relative velocities work combined with the construction of the interferometer, all that can be detected is something called a second order effect. This means the main effect is mainly cancelled out and a much smaller effect is left. If Michelson, expected a speed one-third of his original expectations he would expect a "reading" around one-tenth of his originally expected reading.

If some of this is already starting to sound confusing, do not despair, I will explain these concepts in great detail to help you understand all the ramifications of the experiment and subsequent experiments of the same kind. If you know nothing of the experiment, you should browse the Wiki to understand it as it is generally accepted by the masses. I recommend having a feel for the "accepted" (biased) view before moving on.

False Claims

The Experiment Was Null: The most common cutoff point used in various areas of scientific study is in the area of 5%. While I will call into question the original analysis of the data conducted by Michelson, his analysis showed not less than one-sixth to one-fourth the expected effect. This is 16-25% and well beyond the accepted norms for "null". Only the repeated heralding of Arthur Eddington in spoken and written media caused the faulty repeating of the MMX as "null".

The Readings Obtained were Within the Error of the Experiment: The first time Michelson attempted to perform this experiment, the device he used would only show a resolution within one-tenth of a wavelength. This older failed experiment is of little consequence and is never thought of as the "Michelson Morley Experiment". Only the 1887 experiment is of scientific significance and is widely accepted as the experiment that was null or within error range. However, the error range reported by Michelson in the subsequent paper was one-hundredth of a wavelength or "fringe". Even the highly averaged and dubiously compared reading he reported was over four times that error level. The raw data of the experiment showed readings that varied as much as forty times the error range. As I will show elsewhere, certain experimenter bias and the lack of proper temperature adjustments caused a published result well below the actual signal detected. Neither the raw data nor the properly adjusted and averaged data come close to being "within the error of the experiment".

See For Yourself

Unlike those who give you their opinions and never proffer the evidence, I provide for you here the original MMX paper as it was published in 1887. I will provide information and analysis of the experiment to help you understand and analyse it for yourself in the pages linked to from this one.

The General Concept

There are certain basic assumptions that the experimenters had that greatly affected the outcome and interpretation of the experiment. You must be aware of all of these and their implications to fully understand the experiment.

1) Light propagates in a medium called ether.(today called Aether) Light is not an entity itself nor does it have particles any more than sound has particles. It is a disturbance of this medium that occurs at a specific speed in relation to that medium. This was supported by nearly 100 years of science and was not in question at the time.

2) Ether is a universal reference frame that has specific coordinates. All of the bodies in the universe move in relation to this one singular coordinate system. This is equivalent to assuming the ocean has no currents and all the water in the ocean is always in the exact same place. We could then use latitude and longitude as the basic coordinate system and track the travel of a boat through the water by measuring the speed and direction at which the water passes by the boat.

3) The earth moves with respect to the sun at a speed proven by astronomy and that speed is ~30 km/s. The science behind this assumption has been proven in multiple ways by modern techniques as well.

4) The solar system as a whole is travelling toward some place in the galaxy

The purpose of the experiment was to try to determine the direction and speed at which the solar system as a whole is moving through the galaxy. It was the full expectation that a very large effect would be detected since the component motions of the earth and the solar system and the galaxy etc. could add up to a very great velocity with respect to the universal reference frame. This assumption of a stationary grid is quite presumptuous but it was the dominant view of the time.

In short, the idea is that if the medium is moving with respect to you, then you should be able to detect when you are propagating a wave upstream because it should take longer to go the same distance as cross stream from your perspective. These two scientists had no doubt of the existence of Aether but instead were determined to detect its motion using all the 19th century pre-conceived notions of Aether.

Though the test did not come back as truly null it was not what was expected. Many argue that the result was within the error of the mechanism in all cases but this is very certainly not true in Dayton Miller's experiments and likely not true in others when we see it in the light of all the evidence. Unfortunately, the problem lies in the expectations of the experiment. There were so many assumptions about a medium that we know almost nothing about. EM waves travel through matter; does this mean that ether travels through matter too? Is matter perfectly transparent or does it interact with ether to a small extent?

One of the proposals for why the ether wind was smaller than expected was that perhaps matter entrained ether to a great extent. Perfectly opaque to ether, it would drag it around. (You can visualize entrainment by thinking of a dirty tennis ball submerged in a river. The particles on the hairs further away from the surface will be washed away quickly but all those hairs don't allow the water to run as quickly across the surface of the ball deeper in. They entrain the water and slow down the flow.)

While an interesting theory it seems unlikely for a number of reasons such as how it would slow bodies moving through it and we do not believe there is evidence of this. Additionally there were some tests to try to entrain ether or detect its entrainment at this level. These all came back false.

Dayton Miller did not fall directly into any particular camp of thought when it came to the level of entrainment that occurred at an object level but he did believe that encasing an experiment in a metal housing and performing the experiment in a basement would almost surely reduce the efficacy of the experiment. This was his reasoning for the exceedingly small results of other experimenters.

The newer school of thought leans towards the idea that ether flow near an object is part of gravitation etc. This newest concept centers around the fact that the ether wind detected by both the MMX and Miller seemed to flow in a north/south direction which would indicate that "entrainment" is generated by the magnetosphere or is somehow closely involved in it. This is entirely contrary to the direction originally expected for the wind as well.

Wrongful Expectations

The nature of what should have been expected from the experiment is so complex that I must explain it a piece at a time elsewhere. If you understand the importance of this experiment and wish to make your own determination, it is imperative that you fully understand every part of the experiment and can predict what should have been expected yourself. I will provide as much understanding as a can as simply as I can on this website but I will only give a summary below. It is up to you to determine if you want to understand the "why" behind it and read my fuller explanation .

What should have been expected? There are a few different things they could have and should have expected but let's talk about what they did expect first.

First of all, if the wind is caused by the earth's motion, then looking at the solar system from a northern perspective, the Earth revolves around the sun in a counter-clockwise direction. This means that at noon, the wind would be coming from the west and at midnight the wind would be coming from east. Now depending on your decision on whether or not matter is transparent or opaque to the ether, the sixes would either show a north/south wind or no wind at all because the wind would be coming directly in from the top and not flowing along the path of either of the legs. Michelson expected the Earth to be opaque and expected there to be a north/south wind at the sixes. This is why his tests were performed at noon and six o'clock.

If you know what speed the Earth circles the sun from astronomical observations, then it is a simple matter to determine the speed the earth circles the sun and that is ~30km/s. This is exactly the speed that is considered and expected by Michelson. However, one large consideration seems to be overlooked and it could be called the "helicopter effect." (This is a consideration that the entire solar system is moving.) When a helicopter flies forward, the blades experience very different wind speeds on each side. If, for instance, a helicopters blades circle at approx. fifty miles per hour, and the craft is flying forward at fifty miles per hour, then the wind speed at the tips of the blades will be 100 miles per hour on one side and zero on the other. Combined with the fact that this test was performed for only 3 days, it could have been very possible for it to show a null result even with an existing ether.

Another very important component of the test and determining magnitude is the behavior of what they were looking at. They were looking at vertical lines called fringes. Without getting deep into the principles of interference, let it suffice to say that what was expected was that as the device was rotated, what should have been observed was those lines moving back and forth. As the observed path was rotated into the wind, the cross beam would "get ahead" of the other and the lines would shift to one side. Then as the crossways beam shifted into the path of the wind, it would fall behind the observed one and the vertical lines would move to the other side. Obviously this means that there would not be some random movement, but a steady slide back and forth and it would go back and forth twice during one full rotation of the device. That is all the way left then back to the right, then left and back to the right again as the cross-shaped device was spun 360 degrees.

What was expected is that the entire path a fringe/line would traverse would be measured as .8 of the distance between one set of lines. That means that it moved .4 of a space from the centerline left and .4 to the right and hit each of those extremes twice during one full turn. Since all that is sought is the difference between the two paths, it might have been less confusing if the device was only turned one-eighth turn so that all that was seen is both affected equally and then one leg into the wind. This should show movement from the halfway point to the left side or from halfway point moving to the right. The distance it moved being only .4 of the width between the lines. Because they preferred more data to eliminate experimental errors, they turned the device round and round and averaged all of the equal components. This is important because these readings can be plotted out on a graph and should show very sinewave-like pattern that would not appear if all that was experienced was random error.

An unfortunate and little understood part of this experiment is that though the expectation was .4 of a fringe, a reading of .04 is not a reading of one-tenth magnitude or speed. This reading is something called a second order effect and because of the squared nature of the math, a very small difference in actual speed changes the size of the reading a very great deal. While a speed of ~30km/s might equal a reading of .4, a speed of ~10km/s would give a reading of only .04

Because of their lack of understanding, most people repeat the mantra that the result was one-tenth the size of the expectation and you should now see how this statement, while true in a way, is a gross mis-representation of the situation. Most significant though, is the fact that the readings changed in the way that was expected instead of in a random fashion. Even with many errors in experimental techniques, the sine-wave like pattern can still be seen.

Experimental Errors and the Experimenter Effect

The greatest errors in early interferometry were caused by temperature changes. The expanding and contracting of the device, while extraordinarily small, is enough to alter the results. Michelson, while the inventor of the interferometer, did not have a great deal of experience in operating his own device. Many later experiments, especially the tens of thousands carried out by Dayton Miller exposed the necessity of considering temperature fluctuation. It was shown that, while the fluctuations that occurred in metal and concrete devices like these typically happened in a slow and steady fashion, the effect was certainly evident. Miller found that during the period of time for a rotation of his device the temperature effect was linear and it could not change from expanding to contracting during any single rotation and in fact did not fluctuate in the rate of change except under very great and unnatural temperature conditions. Miller's interferometer rotated about once a minute while Michelson's rotated in approximately six minutes. Michelson's is still a quite rapid rate but perhaps not as reliable as Miller's.

What is very strange about Michelson's data is that while an extra reading of the beginning slot after each rotation was put in place, it was not used to adjust the results. For instance, if there are sixteen reading points and the reading is "10" for point one at the beginning of the reading but when it turns all the way around to reading point one again the reading is "15" then you can see that the temperature caused a drift of five points during the rotation time. Given that the change is constant and unidirectional, the temperature error can very easily be eliminated from the reading in a straightforward and intuitive manner. Michelson, while the error is very obvious in his data, did not make any attempt to correct it. Not only was the error left in but readings with opposing errors of different degrees (one going up the other going down) were averaged together. This lack of temperature consideration greatly reduced the accuracy of his analysis. Thankfully some of his original data is left for us to work with though they are averages of six different readings each that should have been corrected and are already adulterated with possibly different temperature errors.

Micrometer VS Estimation:
Another problem with this experiment is the method for measurement Michelson used. To measure the amount the firnges (lines) shifted, a micrometer was placed in the field of view and adjusted to the center of a fringe at each reading point and the reading on the micrometer was written down. Miller, on the other hand, placed an immobile center mark in the field of view and the reader estimated the distance the center fringe was from the point in tenths of a fringe. At first examination it seems as though the former method is superior to the latter but this intuition is very far from the truth.

First is the problem of touching the device to turn the screw of the micrometer. The total difference being measured between the path lengths of the light is only a couple hundred nanometers. (A human hair can be as much as 100,000 nanometers thick) So touching the device even gently could slightly alter the experiment.

Secondly, the physical size of the fringes change with temperature as well as each time the device is reset, whereas the measurement taken does not change. A single fringe, regardless of its physical width, always represents the same differential between path lengths. If a fringe shifts one fringe width, the path length has changed the distance of one wavelength of the tested light. This means that measuring fifty thousandths(I'm assuming thousandths for their micrometer) on one reading is equal to one speed for the ether and fifty thousandsths on the next reading is entirely another speed for the ether wind. He reported that the fringe widths ranged from forty to sixty in width and just used the average of fifty as his determination. With the largest fringe width being 150% of the smallest fringe width, a large degree of inaccuracy is introduced. Averaging may subdue the error but it also subdues the magnitude of the reading to an extent. The scale of the measurements did not change but what they were measuring did (percentage of a fringe)and then they were averaged together. Additionally, this means that temperature drift within a single rotation of the device causes more than one error type simultaneously and averaging measurements with these dual errors could possibly cause a compounding adulteration of the data.

Miller placed a mark in the window of a device and used visual estimation in tenths of a fringe so that averaging was "apples-to-apples"; it also kept the device from being touched. Factoring out temperature drift on each individual reading combined with the sheer massive volume of readings Miller took to eliminate observer estimation error shows how much Miller advanced the ether drift experiment's technology.


To the left is a picture of the MMX readings after a simple, intuitive and straight-forward linear adjustment to reduce temperature error . This image is a graph produced by Dayton Miller himself to visually exhibit the easily recognizable sine-wave characteristic observed in the MMX.

The long wave of each set represents the average of the full 360 degree rotation of the interferometer on 3 different days with 6 rotations performed on each of those days. As done by Michelson, the second shorter wave is the average of the first and second half of the reading. This was done because each 180 degrees are equivalent in determining fringe displacement.

On the right, I have superimposed the actual readings onto the idealized variation of the readings that were expected of the MM experiment. What is so remarkable is that in this "null" experiment we do not have points arranged in some random fashion even after averaging 18 rotations , but instead we see the readings rising and falling exactly as expected of the wind. This, by itself, is proof positive of a non-null wind detection.

While Michelson's readings are a little sloppy because there were so few readings taken, keep in mind when you read about Dayton Miller that his work represents tens of thousands of readings performed by teams of different individuals in differing locations on differing devices over a great number of years with thousands of hours put into far more up-to-date equipment and far more stringent experimental controls and considerations. Miller's 1933 paper is so profound, and in-depth it takes many many readings to understand the reasoning and conclusions, but upon seeing the data no doubt can be left in the mind of one who has a rudimentary grasp of interferometry, wave mechanics, fluid dynamics and relativity: There is most certainly an ether wind.


By the calculations we've established here and on the expectations page, the Michelson Morely experiment cannot be considered null or within the error of the experiment. Instead we see readings indicating an ether wind of ~10 Km/s with a variation of the reading that coincides precisely with the turning of the device between into the wind and 90 degrees to it. This complex behavior is exactly what should be expected of the wind and not only has it not ever been addressed or explained but cannot be explained in any logical fashion that does not support ether.