I was going to post this as a question in Q&A on an ATM subject, but I also want to discuss some of the consequences of this idea, and I am interested in seeing if it is viable, as it seems to be to me so far, so it might be best to attempt to defend it here the best I can to see where it goes.
Relativity postulates that physics is the same in any frame. Okay, so let's say we perform the M-M experiment again, but this time, instead of light, we use tennis balls with a speed of v'. Well, in the frame of the apparatus, the tennis balls will travel over the same distance in perpendicular directions and meet back in the middle in the same time, and that will occur regardless of which way we turn the apparatus. That much can be seen whether we use the theory of Relativity or ballistic theory, although ether theory might vary, so if we were to expect a different result depending upon our theory of an ether, we would be surprised in the same way we were with the original experiment. But then, what about an observer observing from another frame with the apparatus moving away at a speed of v? We would then have the same dilemma we had with the original experiment, and we would find that if the speeds are not ballistic and so do not add, and that the physics in every frame is the same, then it could be explained if there were a contraction in the line of motion of sqrt[1 - (v / v')^2], where again, v' is the speed of the tennis balls.
So, we could build our theory of Relativity around tennis balls, the Lorentz contraction and time dilation being sqrt[1 - (v / v')^2], Relativistic Doppler is based upon the rate that the tennis balls bounce off of or are emitted by objects and we receive them, and nothing can travel faster than a tennis ball. Granted, at first glance, all of that sounds ridiculous, and you might think I have lost it, but is there really any difference between that and the original experiment? Okay, well, obviously tennis balls don't travel at the ultimate speed, but what about light then? The ultimate speed that is to be used in Relativistic equations can apparently be almost any speed, not necessarily just that of light, just because it was light that was used in the experiment. What if we had used electrons, for instance? Could we then be saying that electrons travel at the ultimate speed, even though the speed of electrons is variable, only to find out later that light travels faster?
Okay, so, if not necessarily light, then what is the ultimate speed that Relativity uses? I am thinking it must just be some maximum universal speed, not directly associated to light or electrons or tennis balls at all. Light has been found to vary very little from c, however, so light must still travel very close to the universal speed, but perhaps not quite. Nothing in the universe can travel at or greater than the universal speed, including light. One consequence of this proposition is that light may have mass after all. Just a very, very small mass, so that any slightly significant amount of energy applied to an atom will cause a photon to be propelled away at very nearly the ultimate speed to an observer at the source. A mass for the photon with different energies applied will produce different speeds for different frequencies, as can already be observed when light travels through a material or medium. When using the Relativistic formula for energy, one finds that E = mc^2 = sqrt[(p c)^2 + m_0 c^4], where p = m_0 v / sqrt[1 - (v / c)^2] and m_0 is the rest mass, and if we also set E = hf, then we get
E = hf = sqrt[(m_0 v c)^2 / (1 - (v/c)^2) + m_0 c^4]
h f = m_0 c sqrt[v^2 / (1 - (v/c)^2) + c^2]
h f / (m_0 c) = sqrt[v^2 / (1 - (v/c)^2 + c^2]
(h f / (m_0 c))^2 = v^2 / (1 - (v/c)^2) + c^2
= [v^2 + c^2 (1 - (v/c)^2)] / (1 - (v/c)^2)
= [v^2 + (c^2 - v^2)) / (1 - (v/c)^2)
= c^2 / (1 - (v/c)^2)
Setting x = h f / (m_0 c), we find
x^2 = c^2 / (1 - (v/c)^2)
x^2 (1 - (v/c)^2) = c^2
x^2 - (x / c)^2 v^2 = c^2
(x / c)^2 v^2 = x^2 - c^2
v^2 = (x^2 - c^2) (c / x)^2
= (1 - (c / x)^2) c^2
= c^2 - c^4 / x^2
Since v < c, where v is the relative speed of the photon to the source and c is the ultimate universal speed, although very close to each other, then c^2 - c^4 / x^2 < c^2, and 1 - c^2 / x^2 < 1, so x > c, therefore h f / (m_0 c) > c, and finally m_0 < h f / c^2. That just gives an upper limit to the mass of a photon, though. For the frequency of the hydrogen atom, it must be many times less massive than the electron. The less massive the photon is, the closer its relative speed to an observer at the source will be to the universal speed.
Experiments would still show light to be non-ballistic and very close to the universal speed with a small mass for the photon. For instance, let's say the speed of a photon at a particular frequency travels relative to the source at .999 of the ultimate universal speed. Now let's say an observer at the source accelerated to .1 of the universal speed. Ballistic theory says that the observer would now say that the photon is travelling away at only .899 of the universal speed. But since the same formulas apply to this speed as c did before, but only the value of c has increase slightly, so the new speed of the photon to the observer travelling at .1 c would be (.999 - .1) / [1 - (.999) (.1)] = .998778 c, the relative speed dropping by only .00022 c, or .022%, instead of the full .1 c.