The final numbers have to admit what the monitoring schedule couldn't see. The interval notation does it in three characters or fewer: within the anchor accession's anchor-storage series, find a = the latest year ≥ threshold and b = the first subsequent year < threshold. Series crosses below → "a–b" (the true interval is somewhere between). Never observed failing → "a+" (right-censored: at least a years, ceiling unknown). Never reached the threshold at all → "<b" (upper bound only). That's why the catalog says "18+" and "<1" instead of false precision.
The publication sheet then assembles one row per species: recommended storage temperature (from the storage that produced the interval), storage behavior mapped from it (−18 °C → Orthodox, 5 °C → Freeze-sensitive, otherwise Unknown), a desiccation protocol, the interval notation, the baseline percentage and whether it came from fresh seed (year 0), the most recent germination at the interval, the best value across supporting accessions at that year with its provider, years stored, supporting and potential accession counts, and the freeze flag with its reason.
Two output disciplines worth stealing: write the reason columns into the sheet itself (a recommendation that carries its own justification survives being emailed around), and sanitize typography for Excel — en-dashes and minus signs become plain hyphens, because a notation column that renders as "18â€"20" in someone's spreadsheet has failed at its one job.