This is a really great question, and there was some initial confusion regarding it.
I just got some clarification from Professor Thouless about the calculation of uncertainty for Kq:
1) You want to use the perturbation method mentioned in the slides and by Professor Oldham in Friday's lecture to determine the error for each individual Kq value.
2) You WILL use two other sections' data to get their Kq values (and uncertainties).
At the end of the day, you will have 3 Kq values and 3 Kq errors for each aluminum bar thickness. When you plot Kq vs. thickness, you'll obviously just average together the 3 Kq values to get one Kq at each thickness and get the total uncertainty for that value of fracture toughness.
Your lab is using sections 3 and 6 (as well as your own, of course).