> 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.

Tests I have seen of this sort have some way of distributing the clamping force, such as a fillet or rubber in the clamp. Epoxying the ends of the test strip into tapered wooden pieces which are themselves clamped is one method.

For the most complete data you will want to measure the distance deflected as you increase the weight as well as the amount of weight at each interesting "failure" point.

> 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?

The application method for your test is part of what you are testing. As long as the application method is reported as part of your data, different application methods can be compared.