I have not checked your calculations, but one thing stands out. You are subtracting two numbers of similar magnitude, with moderate uncertainties, and getting a significantly smaller number with relatively large uncertainty. You do this in 10.5 (subtracting two numbers with ca. 10% uncertainty to get an answer with ca. 40% uncertainty), then in 10.7 you take two such numbers and subtract them again, getting an answer with over 200% uncertainty. If you do this repeatedly in a multi-step calculation, the relative uncertainties will quickly increase. It is therefore important to make your initial measurements as accurate as possible - using a graduated cylinder is just not good enough.