I agree that machining has a built in error factor. The question is whether it could be large enough to make a 'marginally stable' bullet stable.
I have no doubt that it is difficult to get an exact twist on barrel rifling, but I doubt if a 1:9 twist turning is going to be off by another 1/2 inch or more out of 9 inches.
I wouldn’t argue that the error could be 2.5% (or less than ¼ inch out of 9 inches) but I would think that more than that would be a major screw up in setting up the rifling lathe.
Maybe Dave or some qualified machinist could give us some idea of how much error would be the norm.
Just to see if the premise that the barrel twist was cut faster than specified, I did the twist test with a cleaning rod and the barrel was very, very close to 1 twist in 9 inches. It wasn't off by anything close to 1/2 inch.

So I went back to the Berger calculator that I had used before but abandoned because all my best shooting bullets were being classified as 'marginally stable'. Somehow I just didn't believe that.

I entered the temperature, velocity and altitude that I was shooting at, and the BC and bullet length of the 77 gr SMK and 77 gr TMK bullets gave the following results:
77 SMK -- bullet stability rating = 1.36 – twist to stabilize = 8.50. Rated as 'Marginally Stable'.
No way a “little faster twist” is going to be ½ inch faster (5%). That lathe would have to really be humming to be that far off.
77 TMK – bullet stability rating = 1.12 – twist to stabilize = 7.75. Rated as 'Marginally Stable'.
That ‘little faster twist’ would have to be a 'whole lot faster' by 1.25 inches or almost 14 %. A 1:8 twist would only be 11.1% faster than a 1:9 twist.


To test the sensitivity of the stability calculator to altitude, temperature and velocity for the 77 gr TMK bullets:
I changed altitude from 250 feet to 2500 ft (900% change) and stability rating changed from 1.12 to 1.22. (0.010 or 8.9%) That is 1% for each 250 feet in altitude.
I changed temperature from 70 to 100 degrees (42.5%) and stability rating changed from 1.12 to 1.18. (0.06 or 5.3%). That is 1% for every 7.5 degrees of temperature change.
I changed velocity from 2540 fps to 2700 fps close to Pmax (6.2%) and stability rating changed from 1.12 to 1.14 (0.02 or 1.7% change). That’s 1% for every 100 fps.
None of those rates of change are incredible and could account for the accuracy that these bullets are showing.

I then tried to determine the effect of bullet length or BC changes on the stability rating:
Then I adjusted only the length of the bullet from 1.069 for the 77 gr TMK to the same length as the 77 gr SMK (1.000) and the stability rating as the 77 SMK changed to the stability rating of the 77 SMK 1.36.
I then adjusted only the BC of the bullet from the 0.420 of the 77 gr TMK to the same BC as the 77 gr SMK (0.372) and the stability rating didn’t change at all.
The Berger Calculator makes Length the only really significant driver in the stability calculation - but contrary to general belief that long thin high BC bullets are better – the calculator says that shorter is better and according to the calculator, BC makes no difference.

Somehow, to me, that is counterintuitive, especially since all my long, high BC bullets shoot more accurately than the stubby low BC bullets, regardless of caliber.
How can the more accurate bullets be less stable?