Left column, picture in the middle looks like the most correct angle. The higher numbers only reflect what you see in the middle third part of the thread. I wonder what would happen when you also measured this exact sample and the exact fiber looking at the sides. Does the curving to the sides affect the visual appearance of the twist angle or should you even out what you see and measure in a straight line from where a fiber becomes visible to where it turns under again?
New experiment: spin a yarn with a few fibers in a different color and have a group of idiots, sorry "Forum participant" go wild on measuring one colored fiber from every angle possible.
I'm not so sure anymore that there is a "correct" answer to this question, as the angle changes so much. It would be really interesting as well to take a close look at how often outside fibres distort the picture by lying on the outside of the thread at angles not really related to the amount of twist...