Litz says that at just above the speed of sound, the drag on a .22LR bullet increases enough to cause it to slow down faster than than a round that is below the speed of sound. The result of this is that ES of sample of shots fired downrange will be less beginning at about 30 yards or so than closer to the muzzle.
I was not aware of this phenomenon -- mach trimming. It offers an explanation other than flatter trajectories for the desirability of MVs averaging around the speed of sound and just above.
Litz also confirms that, while group convergence doesn't occur with centerfire, sometimes it does occur with .22LR. In other words, sometimes (not very often) groups shot with the exact same ammo in the test tunnel can produce groups that are MOA-wise smaller at 100 meters than at 50. He says it helps to understand what happens by envisaging a corkscrew trajectory. He said modelling has helped to illustrate what happens.
In this regard, the difference between jacketed centerfire bullets and soft lead .22LR bullets may help explain why it happens only with the latter, not with the former. Unlike jacketed bullets, soft lead bullets (and airgun pellets) are much more difficult to produce with nearly perfect centers of gravity. Variations in center of gravity of bullets going downrange result in them having more unpredictable trajectories, perhaps even causing the corkscrewing that Litz thinks of when describing convergence.