Seed Test Counts to Sowing Quantity

What a small germination test can and cannot tell you

A germination test is a count of outcomes in a sample. If you place 20 seeds on moist paper in a warm spot and 16 produce normal seedlings, your observed sample rate is 16 ÷ 20 = 0.8. That number describes those 20 seeds under those conditions.

The limits matter more than the number. Twenty seeds is a small sample, so the true rate for the packet could be meaningfully higher or lower. Your test conditions — temperature, moisture, light, and how long you waited — differ from a garden bed. And germination is not establishment: a seedling that emerges can still be lost to crusting, drying, damping off, cutworms, frost, or transplant shock. The University of Minnesota Extension describes damping off as the death of tiny seedlings due to pathogens at the surface of the potting mix, which is exactly the gap between "germinated" and "grew into a plant."

Treat the test as evidence about the seed lot, not as disease certification and not as a promise of field emergence. If you want a tighter estimate, test more seeds or run a second sample and compare. The same Extension page notes that the percentage of germination and seed purity is governed by law for commercial seed — that is a labeling framework, not a substitute for your own test, and your test is not a legal germination percentage either.

Turning the test into a seed quantity

The example table below follows the same steps. The example uses a hypothetical 20-seed test with 16 germinations, a target of 24 plants, and a separately stated 80% establishment assumption.

The calculation is a chain of divisions, not a single guess:

1. Observed sample rate = germinated ÷ tested. In the example, 16 ÷ 20 = 0.8.
2. Establishment assumption = the share of germinated seedlings you expect to become usable plants. This is your own labeled number, not something the test measured. The example uses 0.8.
3. Combined planning rate = sample rate × establishment assumption. In the example, 0.8 × 0.8 = 0.64.
4. Seeds needed = target plants ÷ combined planning rate. In the example, 24 ÷ 0.64 = 37.5.
5. Round up to a whole seed, because you cannot sow half a seed. The example rounds to 38 seeds.

The table shows the same steps so you can check the arithmetic. Note what the 38 does not mean: it is not a guarantee that 24 plants will stand, and it is not a claim about any particular variety, packet, or garden. It is a planning quantity built from two explicitly stated assumptions.

If you also want a deliberate reserve for thinning, culling weak seedlings, or a second sowing, add it as a separate, labeled step after the division — for example, multiply by 1.10 for a 10% reserve. Keep that number visible so you can see how much of the total is buffer rather than expectation.

Choosing the establishment assumption honestly

The establishment figure is a separate uncertain input; the sample rate is uncertain too. A few ways to keep it honest:

- Start from your own records. For this worksheet, divide usable plants at your chosen checkpoint by the number of seedlings that germinated in the corresponding group. Record both counts. Usable plants divided by all seeds sown is a combined rate; multiplying that rate by germination again would count germination losses twice. If you have no records, say so and pick a conservative figure.
- Label a guessed value and compare scenarios. Without records, this article cannot supply a measured establishment rate for your crop. The 80% value is hypothetical. For sensitivity only, substituting 60% while keeping 16/20 and a 24-plant target gives 24 ÷ (0.8 × 0.6) = 50 seeds rather than 38. Neither percentage is a recommended rate. Record the checkpoint and revise the assumption when you have actual counts.
- Revise downward after a failed stand. If a sowing produced far fewer plants than the assumption predicted, lower the figure for the next sowing rather than reusing the optimistic one. The assumption should track your results, not your hopes.
- Adjust for the crop and the season. Small-seeded crops sown into a crusting surface, or warm-season crops set out while nights are still cold, carry more risk than a robust transplant moved into a prepared bed. The University of Minnesota Extension advises checking packets for days until harvest so choices ripen before frost, and notes that many long-season vegetables must be started indoors in early spring — both of which affect how much time you have to recover from a poor stand.
- Adjust for the bed. A raised bed with consistent moisture behaves differently from a heavy clay row that ponds. The frost dates vs. hardiness zones guide explains why a Zone value cannot stand in for bed-level conditions.
- Do not reuse the germination rate as the establishment rate. They measure different things. Multiplying them is the point of the method; collapsing them into one number hides which assumption is doing the work.

If your establishment assumption is a guess, label it as a guess. A quantity built on a labeled guess is more useful than one built on an unlabeled one, because you can revise it next season.

Comparing two test outcomes

Compare a 16/20 result with an 18/20 result while holding the target at 24 usable plants and the establishment assumption at 80% in both cases. Run both through the same chain with the same target and the same establishment assumption:

- 16/20: sample rate 0.8, combined rate 0.64, 24 ÷ 0.64 = 37.5, rounded to 38 seeds.
- 18/20: sample rate 0.9, combined rate 0.72, 24 ÷ 0.72 = 33.3, rounded to 34 seeds.

The difference is four seeds for a 24-plant target. That is the honest scale of the effect at this sample size: two extra germinations in a 20-seed test moved the planning quantity by about 11%. That 11% figure holds the 80% establishment assumption fixed for both lots, so it is a comparison of the arithmetic under identical assumptions — not a claim that the second lot will perform better in the ground. If the establishment assumption differed between the two lots, the seed counts would move for that reason too.

This is also why a small test should not be over-read. With 20 seeds, a single seed is 5 percentage points of the observed rate. Two seeds flipping either way would change the answer again. This worksheet has not calculated a confidence interval or tested whether the lots differ. Avoid declaring one lot superior from a two-seed gap alone; consider storage history and additional observations as well.

The University of Minnesota Extension recommends not buying more seed than you will use in two or three years, and storing leftover packets in an air-tight container in a cool place, with the refrigerator described as ideal. Storage practice affects the next test more than this one.

Failure modes and when to stop trusting the number

Several things break the method:

- Moldy or shriveled seed in the test. A high count of abnormal seedlings is not the same as a high germination rate. Count only normal seedlings, and if most are abnormal, the lot is a poor bet regardless of the arithmetic.
- A test run too cold or too dry. You measured your test conditions, not the seed's potential. Re-run before concluding the lot is bad.
- Treating the result as certification. A home germination check says nothing about seed-borne disease, purity, or legal labeling. The University of Minnesota Extension notes that the percentage of germination and seed purity is governed by law for commercial seed — that is a labeling framework, not a substitute for your own test, and your test is not a legal germination percentage either.
- Ignoring the calendar. A quantity is only useful if there is time to grow it. Use the seed packet guide to confirm whether maturity is counted from sowing or from transplant before you commit bed space.
- Forgetting that germination and establishment are separate. If your stand fails, check which stage failed before blaming the seed.

Show how the answer changes under stated assumptions, rather than presenting one small-sample result as certain. A sensitivity range such as the 38-to-50-seed comparison above is not a statistical confidence interval and cannot guarantee the target stand.

A practical routine

1. Test a counted sample, record the number tested and the number of normal seedlings, and note the date and conditions.
2. Compute the observed sample rate and write it down with the sample size beside it.
3. Choose an establishment assumption from your own records or a labeled conservative figure.
4. Divide the target plant count by the product of the two rates, then round up.
5. Add any reserve as a separate, visible step.
6. Sow, then record what actually emerged and what survived. That record becomes next season's establishment assumption.

The loop is the point. A single test is a snapshot; a few seasons of notes turn the establishment figure from a guess into something closer to a local estimate. Until then, keep the assumptions labeled and the quantity cautious. A USDA hardiness zone is not part of this calculation: as the USDA explains, zones describe average annual extreme minimum winter temperature, not the lowest temperature ever recorded, and they should serve as general guides rather than guarantees. For annual sowing decisions, the frost dates vs. hardiness zones guide keeps that distinction clear.

Sources and method

• University of Minnesota Extension: starting seeds indoors — retrieved 2026-09-25
• USDA: using plant hardiness maps — retrieved 2026-09-25

AI-assisted writing with automated source, calculation and editorial checks. No hands-on testing or professional certification is claimed. Check the stated assumptions before applying this guide.