Short answer. Count normal seedlings from a random sample, under controlled conditions, and report the percentage with the test date and sample size. Official labs test 400 seeds in replicates [2][3]; a small breeder should test at least 100 per batch (or 4 x 25). A seed that cracks but makes a deformed or weak seedling is not "germinated" for QC purposes.
What's happening
Germination (in seed-testing terms) is the percentage of seeds that produce normal seedlings under favourable conditions [1]. Viability is the percentage alive, which can be estimated quickly with a tetrazolium (TZ) stain that turns living tissue red, typically in 24-48 hours [1]. Germination ≤ viability, because some living seeds are dormant or weak.
Numbers that matter
| Item | Standard or target | Notes |
|---|---|---|
| Official test size | 400 seeds, in replicates | AOSA practice reported by state labs [2][3] |
| Breeder in-house test | ≥100 seeds (4 x 25) per batch | GenetixBay suggested minimum |
| Small batch (<200 seeds) | Test ≥10% or 20 seeds, state the n | Smaller tests are wide-error estimates |
| Official hemp test timing | 7-day prechill then 7-14 days warm; up to 3 weeks total | As described by Oregon State seed labbeing reviewed [1] |
| TZ test | 24-48 h | [1] |
| Storage | ~5 °C, dry, sealed | Recommended to maintain germination about 6 years in [4] (secondary report)being reviewed |
| Seed moisture | Low; dry seeds at room conditions for a week before sealing | Grower practice |
| Good fresh germination | ≥85-90% | Grower practice target for well-made seed |
Error bars for small tests
A germination percentage from a small sample is rough. Approximate 95% margin of error for an observed 90% rate [6]:
| Seeds tested | ±margin (approx.) |
|---|---|
| 20 | ±13 points |
| 50 | ±8 points |
| 100 | ±6 points |
| 400 | ±3 points |
(Normal approximation, 1.96 x √(p(1-p)/n); less accurate at very small n.)
Step by step: an in-house test
- Sample randomly. Mix the whole batch, then count out 4 x 25 seeds without picking the nicest.
- Set up. Two layers of moist paper towel (or germination paper) in a closed container, one replicate each. Use distilled water. Label each with batch ID and date.
- Incubate at ~24-26 °C (75-79 °F) in the dark.
- Count at day 3, day 5 and day 7 (and day 10-14 if many are slow). Remove counted seedlings.
- Classify each seed: normal seedling (healthy root, intact cotyledons emerging), abnormal (stunted, rotten root, no shoot), dead, or ungerminated firm (possibly dormant).
- Calculate germination = normal seedlings / seeds tested x 100. Report per replicate and total.
- Optional sex check for feminised batches: DNA-sex a sample of seedlings (e.g., 20-50) with validated markers [5]; one male in 50 suggests contamination worth investigating.
- Record the result on the GB-BAT with date, method, n and the person who ran it.
Quality control beyond germination
| Check | Why |
|---|---|
| Seed count per pack (count, not weigh) | Customer trust |
| Visual: cracked, immature (pale), insect damage | Removes poor seed before packing |
| Off-type rate in a grow-out | Shows whether the line is what you say |
| Sex ratio (regular) / male or intersex rate (feminised and regular) | Core product claim; a line that herms under normal conditions is not released (see diagnosing hermaphrodites) |
| Chemotype (where claimed) | Legal and customer claim |
| Re-test after storage (every 12 months) | Germination falls over time |
Common mistakes
- Testing the 10 best-looking seeds.
- Counting a cracked shell as germination.
- Storing seed in a warm drawer, or in a fridge without a sealed container (moisture cycles).
- Publishing a germination rate with no date or sample size.
