18650's and their quirks

[BrianG]

How are you measuring IR? The best way IMO is to measure voltage unloaded, then measure again at X load, then take the voltage difference divided by the load current. Do that over the course of the discharge cycle to get a graph of IR. I think you'll find the IR for the poor performing cells is higher than the others.

When cells are in parallel, you are getting some conflicting results because a good cell is somewhat compensating for the not-so-good cells, and the not-so-good cells are surely affecting the performance of the relatively good cells. Basically, parallel cells masks each cell's true performance. Which incidentally, is one reason I don't like any kind of 2p+ setup.