I remember some years back Guns & Ammo magazine had an article testing revolver accuracy at various barrel lengths. With a .38 revolver in a machine rest, to eliminate human sighting errors, they fired identical loads strictly for group size. The barrel was 10 inches to start with, I believe, and after firing each group they cut one inch off and recrowned the muzzle, until it had been cut back to 2 inches. As I recall, there was virtually no difference in the group sizes.