New CI estimation for RGN method: The original interval was the standard error of the bootstrap mean, so it narrowed as you added resamples, which is not a measure of uncertainty in Wm. The new percentile interval spans the actual spread of the bootstrap estimates, and it handles θ error through per-clast classification probabilities instead of a hard 3° margin. Measurement error still isn’t in the interval itself, so the p_min sweep is what shows how sensitive the result is to that error.