A very interesting experimental result was emailed to me. It was a classic demonstration of defects of the bullet on accuracy. The gent didn't care to publish the results for various reasons. I learned the hard way teaching physics lab to freshman and sophomore university students. It was the first year to let the student solve problems experimentally of their own design. What a waste of time. Torque to open a door knob, how do you do that?
The sum total of the tiny masses vectorially of the bullet starting from any reference point it will always get to the same point in the object, the cg, here a bullet from your pick of a reference point. Mine may be a different point say the nose. One could simply hang one like a plumb, but that's only one axis. Hang from a different hook or point the intersection of the string line will go the the CG of the object. The more points you hang the object the better the accuracy in finding the CG experimentally.
Real marksman, reloaders, gun nuts, have figured this out a long time ago, that is you shoot uniform bullets, weight dimension etc. Bullets have what are know as static and dynamic stability. A bullet with a CG off it axis is out of balance statically, some revelation right! Missiles and rockets are drag stabilized like the arrow with feathers at the rear. Bullets we shoot are spin stabilized.
So now with defects we have a little off center line problem. The amount off center will determine how far the bullet will be thrown from the bore's axis. Bullets can be made to have small groups around the points of the clock if the defect is made to leave the barrel at the same orientation each time. This has been demonstrated in laboratory settings and the agreement with the digital models is excellent.
It's like a little man inside the gun with a sling shot releasing the CG in the direct of rotation at the muzzle. The length of the string is how far off axis the CG is. I just had to put my hat on from a different life.
Charlie