Fig. 7

The effects of sample size on reproducibility of w(α,β,t). For raw data a sample of size k = 0.20n ra (or near 2000 leaves) yields reasonable reproducibility (\( \widehat{\uprho} \)(k) = 0.9889). But, for quality controlled data similar reproducibility associates to only k = 0.10n rq (about 1000 leaves). Larger sample would not induce a significant changes in the values of \( \widehat{\uprho} \)(k)