DaleSpam said:
Personally, that is why I really dislike Mach's principle. It always seems to come down to completely useless and unverifiable speculation about the behavior of "otherwise empty" universes. Physics is about describing the behavior of this universe, not fantasy universes.
I agree that it's pointless to talk about "otherwise empty" universes, but that's what Einstein came up with himself when he first pointed out that GR is not completely compatible with Mach's Principle.
I became interested in Machian theories through thinking about the gravitational effect of the whole universe at the current location, which GR doesn't seem to handle very well, not even as an approximation. As Sciama demonstrated in his "Origin of Inertia" paper, if Newtonian gravity is extrapolated in a semi-relativistic way to the whole universe, it gives rise to inertia and rotational effects from relative motion. In contrast, we don't know how to extend GR to a whole universe (except for unrealistically hypothetical universes with special properties such as uniform density).
GR tells us precisely what the frame-dragging effect is due to the motion of a single object, and provided that speeds are non-relativistic and fields are weak, this can be extrapolated reasonably accurately up to almost any scale, if we define an "effective value" of m/r for a distant mass in terms of its contribution to the local potential. The more we do this, the more we see that a test object effectively feels acceleration and rotation caused by frame dragging of other bodies, and it would be logical that if we extend that scheme to all masses, the test object would perceive the overall frame of all the masses to be the rest frame.
With the known values for G and the distribution of the mass in the universe, the result could be (very roughly) around the right order of magnitude to support an exact match, giving the generalized form of the Whitrow-Randall relation:
sum(Gm/rc
2) = n
where the sum is for all masses in the universe as seen from any point and n is a simple constant which depends on the specific theory. For the simplest model of linear frame-dragging, n=1. Note that we can't really define the mass and distances for very distant objects, but we can however assume that the effective ratio for a given mass is a constant value (apart from perhaps systematic changes with time).
This expression obviously cannot be true when G is a constant as in GR, because even the variation due to location with respect to a local mass could apparently cause a detectable variation in G.
However, it is possible instead that the local variation in G due to a central mass is part of what we consider to be the gravitational potential, and the effective value of G is given by the above relation for all other masses, in which case it would effectively be "locally constant".
In fact, the Schwarzschild solution to the Einstein Equations can be expressed in terms of this form of "locally constant" G by a coordinate substitution, but the resulting expression can then be simplified so that it only contains the full "variable G" instead, so the solution treats all masses in the universe identically. (The resulting solution only remains a valid solution to the Einstein Field Equations when the "locally constant" G term due to all other masses is really constant).
Even if Machian theory doesn't require the effective value of G to vary in the central mass case, if the Whitrow-Randall relation holds then G would be expected to vary with time and location on a larger scale (although some of the effects might cancel one another out). At present, there are strong experimental constraints from solar system experiments, in particular Lunar Laser Ranging (LLR), which make it unlikely that G could be varying at the moment even as 1/T where T is the age of the universe.