Even if there was no light, the speed of light, c, is a parameter with units of distance over time, which appears to be fundamental in any description about how two observers who are moving with respect to one another describe the relationships between the events they observe.
http://forum.physorg.com/index.php?showtop...ndpost&p=246365I reproduce this to spare you the pain of reading the nearby posts.
On Time Dilation (corrected and updated)Time Dilation, as was shown shortly after Einstein's 1905 papers, is a natural result which arises from testable intuitive statements of the nature of space and time and physics.
First of all, assuming God doesn't write a physics textbook, man will be forever ignorant of the actual mechanisms of the universe. Even if Professor Y comes up with the mechanistic ONE TRUE THEORY OF EVERYTHING, there will be no way to distinguish the universe of the mechanistic theory from another universe where everything conspires to act just like Y's mechanistic model. That's why physicists (as opposed to philosophers) use mathematical models to avoid talking about the mechanism and only the behavior. Our everyday experience of the universe has taught us a lot of everyday assumptions. The following 4 should be non-controversial when you neglect gravity. They are all statements about observed symmetries of the universe -- so all of them are falsifiable if you found a counter-example.
Four everyday assumptionsLet us
assume the laws of physics are translationally invariant in space. Then it follows a statement about a experiment happening in an arbitrary place will work the same if we center our coordinate basis with it. This also implies that we can calculate what's happening in an arbitrary place and apply a translation transform to it, and the physics is the same.
x' = x + T_x ; y' = y + T_y ; z' = z + T_z ;
Let us
assume the laws of physics are translationally invariant in time. Then it follows a statement about a experiment happening in an arbitrary time will work the same if we center our basis of "now" with it. This also implies that we can calculate what's happening in an arbitrary time and apply a translation transform to it, and the physics is the same.
t' = t + T_t ;
Let us
assume the laws of physics are rotationally invariant. Then it follows a statement about a experiment oriented in an arbitrary direction will work the same if rotate our coordinate basis to be aligned with it. This also implies that we can calculate what's happening in an arbitrary aligned experiment and apply a rotation transform to it, and the physics is the same.
x' = R_xx x + R_xy y + R_xz z ;
y' = R_yx x + R_yy y + R_yz z ;
z' = R_zx x + R_zy y + R_zz z ; Where R is a proper orthogonal matrix, which can be parameterized in various ways by 3 rotation angles.
Let us
assume the laws of physics are invariant with respect to inertial frame. Then it follows a statement about a experiment with a freely moving center of mass moving in an arbitrary direction will work the same if set up our coordinate basis to be co-moving with it with it. But clearly any corresponding change-of-frame transform must tie velocity, time and space together. Since we already assumed we are rotationally invariant and translationally invariant, let us work with v in the x direction and just coordinate differences.
Δx' = F_xx Δx + F_xy Δy + F_xz Δz + F_xt Δt + F_x1 ;
Δy' = F_yx Δx + F_yy Δy + F_yz Δz + F_yt Δt + F_y1 ;
Δz' = F_zx Δx + F_zy Δy + F_zz Δz + F_zt Δt + F_z1 ;
Δt' = F_tx Δx + F_ty Δy + F_tz Δz + F_tt Δt + F_t1 ;
where F is a function of v, which we have agreed to consider in the x direction.
Limiting the form of the velocity transformSince it makes no sense to talk about (Δx,Δy,Δz,Δt) = (0,0,0,0) which says that the two events happened in the same time and place in one frame transforming into other than (0,0,0,0) in the primed frame, it follows that (F_x1, F_y1, F_z1, F_t1) = (0,0,0,0). Thus F(v) represents a homogeneous transform.
If you think you know of a reason why a v in the x direction should involve displacements in the y or z direction, please let me know. I think that the rotational invariance we assumed earlier means that if v is in the +x direction, then it cannot have a reason to prefer +y or -y, and so the effect on y must be zero, and vice-versa, and the same for z.
Then F_yy = F_zz = 1 and F_xy = F_xz = F_yx = F_yz = F_yt = F_zx = F_zy = F_zt = F_ty = F_tz = 0. So we are over half done.
Δx' = F_xx Δx + F_xt Δt ;
Δy' = Δy ;
Δz' = Δz ;
Δt'= F_tx Δx + F_tt Δt ;
Since two events one frame which don't move at all have Δx = 0, but in the other frame Δx'/Δt' = v, then F_xt = v F_tt. Since if two events in one frame are connected by a particle moving at speed -v, and not moving in the other frame then Δx = -v Δt => Δx' = 0 = -F_xx v Δt + F_xt Δt => F_xt = v F_xx => F_xx = F_tt. Let's call that A(v).
Δx' = A(v) Δx + v A(v) Δt ;
Δy' = Δy ;
Δz' = Δz ;
Δt'= F_tx Δx + A(v) Δt ;
For the same reason that length-contraction must be in the direction of movement, we expect two observers to experience the same relative time dilation. Since there are no preferred directions, then nothing but convention distinguished -x from +x and so nothing distinguished -v from +v and so we expect that the time dilation to be the same for two observers in relative motion, if there is any time dilation.
Consider a motionless clock. Two tick of the clock are separated by Δt, but Δx = 0.
so Δt' = A(v) Δt . Now let's move the clock at -v so it is motionless for Δx' = 0. So we want to solve Δt = A(v) Δt', Δx = - v Δt, and Δt'= F_tx Δx + A(v) Δt
So Δx = - v A(v) Δt', Δt' = F_tx Δx + A(v) A(v) Δt' and so
Δt' = - v F_tx A(v) Δt' + A(v) A(v) Δt' and so
F_tx = ( A(v)A(v) - 1 ) / ( v A(v) ) = (1/v) ( A(v) - 1/A(v))
Δx' = A(v) Δx + v A(v) Δt ;
Δy' = Δy ;
Δz' = Δz ;
Δt'= (1/v) ( A(v) - 1/A(v)) Δx + A(v) Δt ;
At this point both the Newtonian and the Relativist should be happy. The Newtonian
assumes that A(v) is a constant with value 1, while the Relativist sees that our four initial assumptions do not yet force that choice. A(v), based on our four assumptions, is just a number and may yet turn out to be a non-constant function of v.
http://www.courses.fas.harvard.edu/~phys16..._appendices.pdfWorking with the velocity transformationNow with translations or rotations, they form a group. (A group is a mathematical way of talking about symmetries.) So that if you apply T1 and then T2, you get T3 which is also in the form of a translations. (Same for rotations.) This should be the same for two transforms related to velocity.
Δx' = A(v1) Δx + v1 A(v1) Δt ;
Δy' = Δy ;
Δz' = Δz ;
Δt'= (1/v1) ( A(v1) - 1/A(v1)) Δx + A(v1) Δt ;
Δx'' = A(v2) Δx' + v2 A(v2) Δt' ;
Δy'' = Δy' ;
Δz'' = Δz' ;
Δt''= (1/v2) ( A(v2) - 1/A(v2)) Δx' + A(v2) Δt' ;
Δx'' = A(v3) Δx + v3 A(v3) Δt = A(v2) (A(v1) Δx + v1 A(v1) Δt ) + v2 A(v2) ( (1/v1) ( A(v1) - 1/A(v1)) Δx + A(v1) Δt ) ;
Δy'' = Δy ;
Δz'' = Δz ;
Δt''= (1/v3) ( A(v3) - 1/A(v3)) Δx + A(v3) Δt = (1/v2) ( A(v2) - 1/A(v2)) (A(v1) Δx + v1 A(v1) Δt ) + A(v2) ( (1/v1) ( A(v1) - 1/A(v1)) Δx + A(v1) Δt ) ;
When we equate our expressions of the double-primed coordinates in terms of the unprimed coordinates, we have the following relations in v and A(v):
- A(v3) = A(v2) A(v1) + (v2/v1) A(v2) A(v1) - (v2/v1) A(v2)/A(v1) ;
- v3 A(v3) = A(v2) v1 A(v1) + v2 A(v2) A(v1) = (v2 + v1) A(v2) A(v1) ;
- (1/v3) ( A(v3) - 1/A(v3)) = (1/v2) ( A(v2) - 1/A(v2)) A(v1) + A(v2) (1/v1) ( A(v1) - 1/A(v1)) ;
- A(v3) = (v1/v2) A(v1) A(v2) - (v1/v2) A(v1)/A(v2) + A(v2) A(v1) ;
From equations 1 and 4, we have the important equality:
A(v3) = A(v2) A(v1) + (v2/v1) A(v2) A(v1) - (v2/v1) A(v2)/A(v1) = A(v3) = (v1/v2) A(v1) A(v2) - (v1/v2) A(v1)/A(v2) + A(v2) A(v1) ;
or
(v2/v1) A(v2) A(v1) - (v2/v1) A(v2)/A(v1) = (v1/v2) A(v1) A(v2) - (v1/v2) A(v1)/A(v2) ;
or
(v2 v1/v1²) A(v2) A(v1) - (v2 v1/v1²) A(v2) A(v1) / A(v1)² = (v1 v2 /v2²) A(v1) A(v2) - (v1 v2/v2²) A(v1) A(v2)/ A(v2)² ;
or, for generic v1 and v2,
(1/v1²) (1 - 1 / A(v1)² ) = (1/v2²) ( 1 - 1 / A(v2)² ) ;
But since this is true for any v, then there is some constant K = (1/(v²)) (1 - 1 / ( A(v)² ) ) for all v. This means A can be written in the form A(v) = 1/√(1 - K v²) ; Using the binomial theorem, we can show that when K v² << 1, A(v) is approximately: 1 + 1/2 K v² + 3/8 (K v²)² + 5/16 (K v²)^3 + 35/128 (K v²)^4 + 63/256 (K v²)^5 + 231/1024 (K v²)^6 + ... ; so the Newtonian will always appear correct as long as |v| is "small," or K << 1/v².
Only at high speed would there be evidence that K is not zero. (If K is zero, then A(v) = 1, just like the Newtonian assumed.)
http://www.courses.fas.harvard.edu/~phys16/Textbook/ch10.pdfhttp://arxiv.org/abs/physics/0302045The generic velocity addition lawFrom equation 2 we see something that with a little algebraic reworking can become our velocity addition law from our four assumptions.
v3 A(v3) = (v2 + v1) A(v2) A(v1) ;
or
v3 / √(1 - K v3²) = (v2 + v1)/( √(1 - K v2²) √(1 - K v1²) ) ;
or
v3² /(1 - K v3²) = (v2 + v1)²/( (1 - K v2²) (1 - K v1²) ) ;
or
v3² = [ (v2 + v1)²/( (1 - K v2²) (1 - K v1²) ) ] [ ( (1 - K v2²) (1 - K v1²) ) / ( K(v2 + v1)² + (1 - K v2²) (1 - K v1²) ) ]
or
v3² = (v2 + v1)² / ( K (v2 + v1)² + (1 - K v2²) (1 - K v1²) ) ;
or
v3² = (v1 + v2)² / ( K v2 ² + 2 K v1 v2 + K v1² + 1 - K v2² - K v1² + K² v1² v2² ) ;
or
v3² = (v1 + v2)² / ( 1 + 2 K v1 v2 + K² v1² v2² ) ;
or
v3 = (v1 + v2) / ( 1 + K v1 v2 ) ;
Clearly if K is measured to be zero, then A(v) = 1 and there is no time dilation. However the results of an 1859 experiment (among thousands of others) are inconsistent with K = 0.
Fresnel and Fizeau's measured value of KIn the discredited dragged-ether theory of Fresnel, light is "slowed" and "dragged" by a transparent dielectric with dielectric constant n. It is slowed to V=c/n, but if the medium is moving at speed v, it is dragged and the measured speed is about U = c/n + v(1 - 1/n²). The amount of "ether dragging" by a moving dielectric is measured as the unexplained Frensel drag coefficient, 1 - 1/n². The result eventually help caused the downfall of the dragged-ether model, for where a physical medium can support a wide variety of waves, the phenomenon of dispersion shows that n is a function of wavelength, and so the Fresnel drag coefficient must also be a function of wavelength, and so there must be a different ether to drag for every wavelength of light.
But let's just take V=c/n the experimental velocity of light in stationary medium, and apply our generic velocity addition law to it and see how it predicts an observer moving relative to the medium measure its speed at.
v3 = (v1 + v2) / ( 1 + K v1 v2 ) = (v + V) / ( 1 + K v V ) = (v + c/n) / ( 1 + K v c/n )
if we assume K v c/n << 1, then we can use the binomial theorem to approximate v3 as (v + c/n) - (K v c/n) (v + c/n) + (K v c/n)² (v + c/n) + ...
= c/n + v - K v² c/n - K v c²/n² + K² v^3 c²/n² + K² v² c^3/n^3 ... or, if you drop the terms which aren't linear in v, v3 = c/n + v (1 - K c²/n²)
If v3 is close to the observed value U, then K c² = 1, or K = 1/c²
http://en.wikipedia.org/wiki/Aether_drag_hypothesishttp://en.wikipedia.org/wiki/Fizeau_experimentEpilogueThis was written for Bryn, who hasn't explained why he doesn't believe in time dilation. It was written to show that time dilation, by which I mean that observers who differ in velocity must have A(v) different than 1, is the only physical result if you accept the four assumptions. The other consequences of this idea have been well-developed. K is very close to zero in ordinary units, but we have thousands of experimental results which suggest that it is much closer to 1/c² than to zero.
For simonrach, this should show that if the four assumptions are good, then even experiments not involving light will show that c is an physically important speed in our universe. Further E = m_0 A(v) c² gives us E = m_0 c² + 1/2 m_0 v² + ... which has been used to relate Newton's approximate formula for kinetic energy to Einstein's Relativity.
http://relativity.livingreviews.org/Articl...05-5/index.htmlAccording to
http://arxiv.org/abs/0708.0929 people have shown this to be true with various degrees of rigor since at least 1910.
W. A. von Ignatowsky, “Einige allgemeine Bemerkungen zum Relativitatsprinzip,”
Phys. Z. 11, 972-976 (1910).
Following Einstein, Minkowski (1908) showed that algebraically this was the same as saying space and time were not separate things, which is how all physicists work today. Both Length contraction and time dilation arise from treating space and time as separate things with separate meanings, which is the Newtonian and intuitionist view. In Minkowski space-time, these are trivial (boring!) effects.