Figure 3:

Setting viability test intervals at one-third of the time it takes for viability to decrease to 85% of the initial viability means that sampling intervals will vary. Predicted survival curves are shown for rice seeds stored at 7% moisture content and 3°C (estimated using the Seed Information Database; Royal Botanic Gardens Kew, 2016), with varying initial viability, as follows: 99% (red), 96% (green), 93% (blue) and 90% (cyan). Theoretical sampling intervals (red, green, blue and cyan numbers, respectively) are shown for each survival curve, calculated as one-third of the time predicted for viability to decrease to one-third of the initial viability (in years). The inset graph shows an example where it may be more appropriate to use 85% of initial viability for determining sampling intervals, when there is a proportion (in this case, 10%) of non-respondents in the population of ageing seeds. Hence, although the initial viability (as a proportion of seeds sown) is close to 90%, the sampling interval is almost double that of the seed lot in the main graph with 90% initial viability.