Uncertainties of a group of results

[Tangeton]

I got a table for a simple pendulum. I have 8 lengths, from 0.20m going up by 0.01 to 0.27. For each length, I have time for 10 oscillations (10T) that I've measured, and I have repeated the measurement twice for each length. Then I got the average time for 10T. I divided this average to give me an average for time period of one oscillation, T. I've plotted a graph of T² against L, and so I also worked out T² for each lengths' period time, T.

I am stuck on the uncertainty. I have tired tried to work it out but when I do it for T I get 6.77%, and I do it for the spread of my results for T, but when I rise it to the power of 2 to get T² its 45.89%. I can't have such a huge uncertainty. And also I am worried because it is uncertainty in measurement but I can firmly say that I didn't measure T but 10T, and so should I do the uncertainty for average 10T instead?

What uncertainty would be the best to put down on my write up?

I'm an A Level physics student.

Thank you for all your help!

 

[BvU]

No way one can untangle this. 46 is the square of 6.8, but that's not the way errors propagate!
Show your work.