I hadn't considered the 'impact' testing that Nick brought up -- I had planned on repeatedly adding small weights to the center of a panel and recording the weights that caused the aforementioned failure modes. Given the variability of the wood, I was hoping to do 3 tests of each. Obviously, the number of panels I'd need starts to add up, especially if I try lay-ups such as 4mm cloth outside and 6mm inside (where we would expect tensile forces to be the greatest.) It's tempting to go with 6"x6" panels, but I'm a bit leery -- would panels this small be representative of the bulging of the hull of a kayak parked on a rock? The hull bends quite a bit to help distribute the load... I like the thin strip method, clamped at one end with a weight on the other end; the only problem I forsee is that the stresses are increased at the clamp, so most of the failures would occur there rather than in the unsupported span. I may combine all these ideas and do a 3-point test with 3"x12" or 6"x12" samples.

One more question: the issue of glass/epoxy application methods has been brought up many times before but: I would assume a rough-sanding with 80-grit, no fill-coat, a good squeege (or bubble roller) job to ensure that the cloth is as close to the wood as possible, and fill coats on what would be the outside surface before the epoxy completely sets. Any other considerations?