[Goosegirl]

Quite agree Balders in this case anyway. Measuring statistical significance usually involves the Null hypothesis and
p-values which I won't go into as it's many years since I did them. Also important is the number of specimens/cases you use - the more the better - but I think the absolute minimum is 100 per each specimen trial (pot, grow-bag etc) where they all must be subjected to the exact same regime. Variables in this case are the plants themselves, as we all know how they can differ in germination times, growth and productivity but, more importantly, the growing medium must also be exactly the same for each trial. As was rightly said, soil conditions can vary even within a few feet which may or may not have been the case here, so that invalidates the soil test. The compost used in both the grow-bags and pots should have been thoroughly mixed before use then replaced in the pots or Grow-bags to eliminate any variations in content such as texture, feed distribution etc. The polytunnel should have had a fan to eliminate any mini-local variations in temperature. I doubt any of this was done so, in my view, I'd ignore the results.

[mumofstig]

it's just think it's an unreliable test.

Up to everyone to make up their own mind, perhaps do their own trial 

I didn't give the reasons why I said that .......................... but, I'm soo glad you agree with me 

[Goosegirl]

Thanks MOS - so glad you agree with me too! If I gave it more thought I could maybe think up more reasons but hey - it was a potty trial and not worth the effort!