Interesting thread.

Let's see if we can understand why people swing their arms in the first place.

According to Wikipedia [1]:

"Whether arm swing is a passive, natural motion caused by the rotation of torso or is an active motion that requires active muscle work has been a critical discussion on arm swing that could illuminate its benefit and function. A recent study concentrated on the energy consumption during walking showed that at low speeds arm swing is a passive motion dictated by the kinematics of torso, no different from a pair of pendula hung from the shoulders."

It's obviously more complicated than this, not least because the body flexes, but for what it's worth, here's my theory for arm swing changes in PD.

If the pendula model remains true for PwP, the main thing that we need to look at is the work done at the shoulder joint to allow the rotation of the arm. Any cogwheel rigidity in the shoulder joint could impose a large enough energy cost to reduce or, even, to stop the natural swing.

As the disease progresses the other side is affected and eventually the "friction" at the shoulder becomes enough to stop that swing too.

The partial restraint effect [PRE](the good arm restrained allows the bad arm to swing) can be explained as follows. The energy that forces the pendula (the arms) comes from the jerkiness of the walking gait. I hazard a guess that when this energy can only be used by one pendulum, it leads to an increase in the energy available to the other arm. If this is enough to overcome the cogwheel rigidity, then the previously unswinging arm comes back into action.

Finally, we need to look at the situation where the arm that swung no longer does, and the arm that did not swing now does. The simplest way for this to happen is if the disease progressed to the other side and got worse there than on the original side, and then the PRE takes over. But I've not seen such a severity shift in PD. I have, though, seen localized reversals of severity from one side to another. So, a reversal of the shoulder rigidity is more likely than a complete severity shift from one side to the other. A better theory, in my opinion, comes from taking into account the spread of the disease into the leg on the bad side. If this alters the gait, it is likely to impact more on the good side arm. Thereafter the PRE takes over to give the desired effect.

All this is speculation, but someone must have a lab or, failing that, a workshop to test the theory.

A quick look at the literature gives:

Ford et al. found [2] for healthy adults:

" A comparison between walking with one arm constrained and normal walking revealed decreased, transverse pelvic, thoracic, and trunk rotation, however there were slight increases in non-constrained arm movement amplitude."

My details follow.

At diagnosis:
arm swing: left=normal, right=very reduced, though I could will it to happen
hand tremor: left=none, right=noticeable, no problems

After 5 years and until the present 9 years:
arm swing: as before
hand tremor: left=bad, spilt water from glass, right as before

Reference

[1] http://en.wikipedia.org/wiki/Arm_swi...man_locomotion

[2] Gait Posture. 2007 Jun;26(1):135-41. Epub 2006 Sep 25.
"Arm constraint and walking in healthy adults."
Ford MP1, Wagenaar RC, Newell KM.
http://www.ncbi.nlm.nih.gov/pubmed/16997561

John