Two Germination Tests: Decide Before You Sow

Record both tests before you compare them

Label the seed lot and each sample. For every sample, write down the number of seeds tested, the number of normal seedlings counted at your chosen endpoint, the date, and the test conditions that matter for your method. Apply the same counting rule to both samples so the numbers mean the same thing. Keep the raw counts even after you combine them, because a reader should be able to reproduce your arithmetic from the record alone.

Use separate seeds from the same lot for each sample. If moisture, temperature, light, or the counting endpoint changed between tests, flag the difference and investigate before treating the two counts as equivalent observations. Two similar counts do not prove that either sample represents the whole lot, and one unusual result does not prove the lot is bad.

The University of Minnesota Extension notes that the fresher the seed, the greater the chances that all the seeds will still be viable, and that fewer and fewer seeds from a packet will germinate over time. That is the reason to test stored seed rather than assume. It does not validate any threshold on this page, and it does not tell you what test temperature or waiting period to use. Follow the packet and crop-specific guidance for the method itself.

Keep the single-sample baseline visible

The supplied example is Sample A: 11 seedlings from 20 seeds. The observed sample rate is 11 ÷ 20 = 0.55. Keep a separate, explicitly assumed establishment fraction of 0.75 and a target of 30 plants at a named checkpoint. The combined planning fraction is 0.55 × 0.75 = 0.4125, and 30 ÷ 0.4125 is approximately 72.73, which rounds upward to 73 whole seeds.

The table inserted below records exactly this baseline. It does not contain a two-sample comparison and it does not establish which rate you should finally use. The 75% establishment figure was not measured by the seed test; it is a stated assumption. If your own records give usable plants divided by all seeds sown, that is already a combined rate, and multiplying it by germination again would count the germination loss twice.

Keeping this baseline visible matters because it makes the effect of a second test measurable. When you later pool counts, you can see how much the planning quantity moved and why. It also prevents a pooled rate from quietly replacing a raw count without leaving a trace.

Set the disagreement rule before you see the result

Write three things down in advance: how much disagreement between comparable samples you will tolerate for this planning exercise, what you will do outside that tolerance, and when you will stop testing. An illustrative rule might accept an absolute gap of at most 0.15 between equal-sized samples, investigate anything larger, and stop after one additional check.

That 0.15 is a chosen planning tolerance, not an Extension recommendation, not a confidence limit, and not proof of seed quality. State whether the boundary is included. For this example it is: a gap exactly equal to 0.15 takes the same branch as a smaller gap. If a poor stand would be costly for you, change the rule before you test rather than after.

Choose a concrete fallback and write it down. Options include postponing the sowing decision, checking whether the test method was consistent, buying a fresh packet, or sowing a larger area than planned. Do not keep running tests until one count supports the answer you were hoping for; that converts a decision rule into a search for confirmation.

Compare the counts, then choose which rate to use

Suppose Sample B produces 14 seedlings from 20 seeds. Its rate is 0.70, and the gap from Sample A is 0.15. Under the stated inclusive rule, these comparable tests qualify for pooling. Add the counts rather than averaging the percentages: 25 seedlings from 40 seeds gives 0.625. With the same 0.75 establishment assumption, 30 ÷ (0.625 × 0.75) equals 64 seeds. The change from the baseline 73 to 64 comes from the chosen pooled rate, not from a new guarantee about the lot.

For an inside case, 12 out of 20 gives 0.60 and a gap of 0.05, which pools under the same rule. For an outside case, 18 out of 20 gives 0.90 and a gap of 0.35, which triggers the fallback. Pooling that pair would still produce a calculable average, but it would ignore the decision rule and hide the disagreement you intended to investigate.

When sample sizes differ, combine seedlings and tested-seed counts rather than taking an unweighted mean of percentages. Keep that arithmetic separate from the decision about whether pooling is appropriate at all. A pooled rate from two comparable tests is still a small-sample estimate, and it still describes only the seeds you tested under your conditions.

Separate establishment uncertainty from test disagreement

Name the checkpoint for the 75% establishment assumption, such as usable plants at thinning. Record whether that fraction comes from comparable past records or is simply a scenario you invented for planning. Without relevant records, this worksheet cannot supply a measured establishment figure for your soil, your weather, or your crop.

To see how much that assumption matters, hold the pooled rate at 0.625 and change only establishment from 0.75 to 0.60. The planning quantity becomes 30 ÷ (0.625 × 0.60) = 80 seeds, compared with 64 before. That is a 25% increase, and it is not evidence that either assumption is correct. If the difference changes whether you buy another packet or whether the bed has room, resolve the assumption or keep both scenarios visible in your notes.

Any extra reserve is another labeled choice. Multiply after the division, for example by 1.10 for a 10% reserve, and keep that number visible so you can see how much of the total is buffer rather than expectation. Do not bury a reserve inside the sample rate, and do not count it twice. Record the unrounded calculation and the final upward rounding so the order of operations is reproducible.

Use the result alongside the packet and local conditions

A seed-count worksheet does not diagnose disease and does not forecast field emergence. Two samples can agree and still be unrepresentative of the lot. When results disappoint, check the method and the growing conditions rather than assuming the arithmetic identified the cause. Germination is not establishment: a seedling that emerges can still be lost to crusting, drying, damping off, cutworms, frost, or transplant shock.

The USDA map guide explains that hardiness zones reflect the average annual extreme minimum winter temperature in 10-degree Fahrenheit bands, based on 1991-2020 weather data. That is a perennial winter-survival reference, not a germination measurement and not a sowing calendar. For those separate questions, use the seed packet guide and the frost dates versus zones guide.

Save a short decision record: lot label, both raw counts, relevant test conditions, the preselected rule, whether you pooled, the establishment assumption, and the chosen fallback or sowing quantity. That record is the useful output. It shows why your action changed, without claiming that a small home test certifies the lot or guarantees the target plant count.

Sources and method

• University of Minnesota Extension: starting seeds indoors — retrieved 2026-09-26
• USDA: using plant hardiness maps — retrieved 2026-09-26

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.