The Gold Standard of Seed Biology

Chapter index

The everyday question “Are these good seeds?” hides several different questions. Are they the intended species? Are the embryos alive? Will they germinate under the proposed conditions? Will they produce usable seedlings? Will the remaining stock survive storage? A useful answer names the outcome before selecting a test.

Imagine a packet labelled Lot A. The label connects a group of seeds with a source and handling history; it does not promise that every seed has an identical state. Some may have matured earlier, suffered damage or experienced a different position within a fruit. A test on a sample estimates something about the lot, with uncertainty. It does not inspect every seed automatically.

Define the unit as carefully as the outcome. A unit sold or handled as a “seed” may include a fruit wall or other covering. One numbered unit in a tray should retain that identity throughout the observations. If a cluster contains several seeds, counting clusters first and seedlings later can make an apparent germination percentage exceed one hundred without any biological miracle.

A precise question might be: “What fraction of randomly sampled units from this identified lot produces normal seedlings by the stated final count under this protocol?” Another might ask how rapidly radicles emerge. Both are sensible, but they are different experiments.

Before starting, write four things: the material, the conditions, the observation criterion and the decision the result will support. This small record prevents later confusion when an attractive photograph seems more persuasive than the actual measurements.

Contents · Next chapter

A seed brings together an embryo, protective structures and resources that support early development. In flowering plants, the embryo and endosperm arise through fertilisation, while the seed coat develops from maternal ovule tissues. Reserves may remain largely in endosperm or become concentrated in cotyledons. A bean diagram is therefore a useful example, not a universal map of every seed.

OpenStax’s account of seed development provides contrasting arrangements in flowering plants. The functional questions are straightforward: where is the embryo, what surrounds it, where are reserves stored, and what must change before the embryonic root can emerge? Seed biology then asks how those parts behave over time.

Think of an observational sequence. First examine the intact unit. Then compare a suitable prepared section or labelled teaching specimen. Finally relate the structures to germination observations from other, intact units. Cutting one seed open and then expecting it to provide an undisturbed germination result confuses two methods with different consequences.

An empty-looking section is not automatically proof of an empty seed. The section may miss a small embryo, or the observer may not recognise the tissues. Likewise, a full seed is not automatically alive. Structure and biological function require different evidence.

Use an annotated drawing to keep that distinction clear. Label “visible coat”, “suspected reserve tissue” and “embryo identified” only where supported. Then add a separate line for what remains unknown about viability. Good observation can be confident about anatomy while remaining appropriately uncertain about future growth.

Contents · Previous chapter · Next chapter

Seed quality has a history before a packet reaches your hands. Development, harvest timing and subsequent handling affect the material that will enter a germination or storage test. A mature-looking fruit and a seed ready for a particular conservation procedure are related observations, but they are not interchangeable measurements.

Kew’s post-harvest handling guidance emphasises maturity, moisture status and ambient conditions. Very immature material can be harmed by rapid drying; ripe material also needs appropriate handling. The general lesson is to determine the biological state before applying a storage routine, rather than treating all newly collected seeds as equivalent.

Consider two fictional samples taken from the same parent population. Sample A is collected earlier and kept for several days in damp fruit tissue. Sample B is collected later and processed promptly. If B later germinates better, the comparison does not isolate harvest maturity: handling changed too. A report should preserve both differences instead of naming whichever explanation sounds more convincing.

To investigate maturity separately, compare planned developmental stages with handling held as similar as the species allows. To investigate handling, allocate comparable material to different handling conditions. Include several source plants or batches where the question requires generalisation beyond one plant.

Keep collection date, maturity observations, processing date and storage transitions together. “Collected on Monday” is less useful than a record that also shows whether seeds remained inside moist fruits until Friday. This history becomes especially valuable when a later test gives an unexpected result and the team must locate the likely point of change.

Contents · Previous chapter · Next chapter

These terms describe related capabilities and events, not four names for the same percentage. Viability concerns the capacity for living seed tissues to produce growth under appropriate conditions. Germination is a developmental process with an operational endpoint. Vigour concerns performance properties, including how seed lots cope with conditions and storage. Establishment concerns the resulting seedling in its growing environment.

The ISTA Vigour Committee connects vigour testing with practical outcomes such as emergence, uniformity and storage potential. That connection requires validation. An improvised stressful treatment does not become an internationally validated vigour test simply because it produces a numerical result.

Suppose ninety of one hundred units show a radicle, eighty form seedlings meeting a defined normality criterion, and sixty remain established at a later outdoor census. Report 90%, 80% and 60% with their respective endpoints. Selecting the largest number and calling it “success” would erase where losses occurred.

The denominators also matter. Sixty established seedlings are 60% of the original hundred, but 75% of the eighty normal seedlings. Both are correct calculations. The second describes a conditional transition; it must not be presented as the total outcome per seed sown.

Contents · Previous chapter · Next chapter

A viable seed can fail to germinate because its current environment is unsuitable, because dormancy prevents germination, or because the observation period and protocol are inadequate for that material. Dead and empty seeds can also remain unchanged. The shared appearance of “nothing happened” does not make the causes equivalent.

Physical dormancy involves a water-impermeable covering. Physiological dormancy involves internal regulation that prevents germination. Some seeds have embryos that need further development; some combine developmental and physiological constraints. Kew’s environmental factors and dormancy guidance explains these distinctions and the need for species-specific conditions.

Use the classifications to form testable questions, not to assign a diagnosis from appearance alone. A hard object may be a fruit wall rather than the relevant seed coat. A seed that absorbs water but does not produce a radicle does not thereby prove one particular hormonal explanation. Observations narrow possibilities; they rarely settle the entire mechanism at once.

For an unidentified lot, the first repair is often better identification and protocol research. Repeatedly adding water or applying stronger treatments can destroy the evidence while leaving the original question unanswered.

Imagine a test ending with sixty germinated seeds and forty unchanged units. The defensible first report describes those observations. It should not immediately call the remaining forty “dormant”, “dead” or “bad”. Further evaluation may distinguish empty units, damaged seeds and potentially viable ungerminated seeds. Keeping those categories separate is more useful than forcing every outcome into a binary success/failure label.

Contents · Previous chapter · Next chapter

When a dry seed takes up water, tissues rehydrate and the conditions for renewed metabolism change. Water uptake also has a physical component: non-living seed material can absorb water. Swelling is therefore an observation about hydration, not a complete viability test.

An original weighing exercise makes the distinction visible. Ten dry units have a combined mass of 2.00 g. After a defined hydration period and a consistent surface-blotting procedure, they weigh 3.10 g. Their mass gain is 1.10 g, or 55% of the starting mass. This calculation does not mean that they are now 55% water by total mass, because the starting material already contained an unknown amount of water.

It also does not establish that all ten absorbed equally. One damaged unit could behave differently from the others. If the biological question concerns variation among seeds, weigh or observe units separately where feasible. A pooled average can hide the difference between uniform moderate uptake and a mixture of strongly and weakly responding units.

Keep surface water under control when measuring uptake. A droplet clinging to the outside is not evidence of internal hydration. Describe the blotting method, timing and balance resolution; otherwise a small apparent treatment effect may reflect how the sample was handled.

The practical diagnostic is to connect hydration with subsequent events. Did tissues hydrate? Did a radicle emerge? Did development continue normally? Those successive observations provide a stronger account than one swelling photograph. For an introductory treatment of water, oxygen and growth, use How Seeds Germinate and Grow.

Contents · Previous chapter · Next chapter

Temperature, water availability, light and gases influence germination, but a seed does not respond to them as four independent on/off switches. The response depends on species, physiological state and the combination of conditions. A temperature favourable to one lot may inhibit another, and a condition that permits germination may be poor for later seedling survival.

One primary study makes the mechanism concrete. Toh and colleagues’ Arabidopsis research linked high-temperature germination inhibition with increased abscisic acid synthesis and reduced gibberellin synthesis and signalling. Genetic and physiological experiments supported the explanation. This is evidence for a studied system, not a recipe for every seed or a universal temperature threshold.

The important reasoning move is that an external condition can alter internal regulation. “Warmth speeds biological reactions” is too incomplete to predict germination. The same temperature change can affect competing processes and the regulation of growth. A warmer treatment need not produce more germination.

For a classroom comparison, record actual conditions rather than relying on labels such as “warm cupboard” and “cool windowsill”. Those locations may also differ in light, drying and daily fluctuations. If several variables change together, the experiment tests the combined locations, not temperature alone.

Do not turn this mechanism into an instruction to apply plant hormones. A mechanistic explanation and a practical propagation protocol have different evidence requirements. The useful next step is to locate a credible protocol for the identified material, then ask which measured condition could explain the observed response.

Contents · Previous chapter · Next chapter

The embryo must develop roots and shoots before a seedling can rely substantially on its own photosynthetic surface and resource uptake. Stored reserves support that transition. The emerging plant is therefore using an inherited resource supply while constructing the organs needed for its next stage.

This explains why the absence of visible green leaves does not imply the absence of activity. It also explains why a radicle is an early milestone rather than proof that the entire transition has succeeded. Damage to essential tissues, unsuitable growing conditions or failure of later development can still interrupt the process.

Consider a teaching sequence with thirty numbered units. Twenty-four produce a radicle. Of those, twenty develop the structures required by the exercise’s seedling criterion. Eighteen survive to the later census. The record shows losses at two transitions after radicle emergence. Treating all six non-established germinants as “failed germination” would conceal that sequence.

Use photographs taken at standard intervals to connect observations. A single end-of-test image may show a missing seedling without revealing whether it never germinated or germinated and died. A dated sequence can answer that question without pretending to identify every underlying cause.

For learning, draw a chain with separate boxes for the counted events, then attach each observation to its box. For practical planting, decide which stage matters to the intended decision. A nursery interested in transplantable seedlings needs evidence beyond the first appearance of roots. A study of dormancy release may legitimately focus on an earlier endpoint, provided it states that boundary.

Contents · Previous chapter · Next chapter

Begin with a defined seed lot and allocate material to treatments without selecting the largest or most attractive seeds for one group. Keep an untreated or standard-protocol comparison where it answers the question. Record the number allocated, the number lost during handling and the reason for any exclusions.

Replication must match how the treatment is applied. If all treated seeds occupy one container and all controls occupy another, a container difference can masquerade as a treatment effect. One might dry faster, receive different light or carry contamination. A hundred seeds within each container do not automatically provide a hundred independent treatment replicates.

A stronger small experiment might allocate comparable units among several containers per treatment and arrange them across the available space. Record container identity as well as seed identity. If conditions vary by shelf or position, account for that structure rather than pretending it does not exist.

Decide the final count and criterion before inspecting the outcomes. Ending one treatment as soon as it looks good but allowing the other more time makes the reported comparison difficult to interpret. If an unexpected problem requires a protocol change, document it and preserve the original record.

ISTA’s current Rules page describes standardised sampling, testing and reporting and identifies the 2026 edition. Formal testing should use the applicable current method. The experiments here teach reasoning; they are not certificates, official seed-grading procedures or substitutes for a validated species-specific test.

Contents · Previous chapter · Next chapter

Four fictional containers each begin with twenty-five randomly allocated seeds. At the predeclared final count, the numbers of normal seedlings are twenty, twenty-one, nineteen and twenty-two. The total is eighty-two out of one hundred, so the observed normal-seedling percentage is 82%.

Suppose five of the remaining eighteen have abnormal development, eight remain firm and ungerminated, and five are clearly dead at evaluation. The full record is more informative than 82% alone. If the eight ungerminated seeds need further assessment, say so. Do not silently add them to the normal-seedling count or remove them to improve the percentage.

Because all four containers start with equal numbers, averaging their percentages gives the same 82% result. With unequal numbers, an unweighted average of percentages can be misleading. For example, nine successes from ten and fifty successes from one hundred give 59/110, approximately 53.6%, overall. Averaging 90% and 50% would incorrectly give 70% for the combined sample.

Kew’s germination procedures and evaluation sheet explains why test records and evaluation of ungerminated material matter. Always specify which denominator your report uses, particularly where a protocol separately accounts for empty or infested material.

Contents · Previous chapter · Next chapter

Two seed lots can reach the same final germination percentage at very different speeds. Timing may matter for coordinating nursery work or interpreting a treatment, but a timing statistic needs a clear definition and the full final outcome alongside it.

In a fictional observation series, twenty seeds newly meet the radicle criterion on day two, thirty on day four and ten on day six. Sixty germinate in total. The mean observed germination time among those sixty is (20 × 2 + 30 × 4 + 10 × 6)/60 = 220/60, approximately 3.67 days.

The word “newly” matters. Counts must represent additional germinations in each interval. Using cumulative counts of twenty, fifty and sixty in the same formula would count earlier germinants repeatedly. Keep both columns if useful, but label them clearly and use the one required by the calculation.

The statistic also excludes seeds that never meet the endpoint during the observation period. A treatment that kills slow germinants could appear faster among the survivors while producing fewer seedlings overall. Compare the final fraction, the timing distribution and relevant later outcomes together.

Observation frequency limits precision. A seed first recorded on day four may have germinated after the previous inspection. The calculated time reflects the recording schedule, not an exact biological timestamp. Do not report many decimal places when checks occurred only every two days.

When comparing experiments, align the starting event too. Time from sowing is not necessarily time from the beginning of a pretreatment. A method that produces quick emergence after a lengthy preparation may save nursery time at one stage while taking longer overall.

Contents · Previous chapter · Next chapter

Eight successes from ten and eighty successes from one hundred both equal 80%. They do not provide the same amount of information. In the smaller test, one seed changes the percentage by ten percentage points; in the larger, it changes it by one point. Sample size does not remove bias, but it changes how sensitive a result is to individual outcomes.

Suppose Lot A produces forty-one normal seedlings from fifty and Lot B produces thirty-nine from fifty. The observed difference is four percentage points. That arithmetic alone does not establish that A is consistently superior, especially if only one container was used for each lot or the samples were not representative.

Ask what conclusion is needed. A preliminary classroom comparison may justify a larger follow-up. A planting contract may require standardised laboratory testing and a decision rule. A rare conservation collection may require a carefully limited test because sampling consumes irreplaceable material. “Use more seeds” is not always an adequate answer.

An honest report separates the observed counts from the inference. Write “41/50 under this protocol” before writing “better”. Then explain whether the comparison was replicated, whether conditions were comparable and whether the observed difference matters to the practical outcome.

Avoid choosing a winner after trying many treatments and reporting only the most attractive result. An unusually high percentage can occur by chance among numerous comparisons. Preserve the full set, identify exploratory findings and check promising results with fresh material where feasible. Reproducibility begins with retaining the less exciting observations as well as the successful tray.

Contents · Previous chapter · Next chapter

Conventional dry, cold seed storage works well for many species whose mature seeds tolerate substantial drying. These are commonly called orthodox seeds. Recalcitrant seeds are sensitive to drying; intermediate behaviours complicate the simple two-category picture. Storage response must be established for the material rather than inferred from the fact that it is a seed.

Kew’s desiccation-sensitivity guide recommends combining existing information with appropriate testing. Morphology and origin can provide clues, but a clue is not a species-specific demonstration of tolerance. A tropical origin alone does not justify assuming that every seed is unsuitable for conventional banking.

Imagine a collection with no reliable storage history. Freezing all available seeds immediately would turn an uncertainty into a potentially irreversible whole-collection experiment. A better decision process first establishes identity, examines relevant evidence and designs an appropriately limited assessment. The amount available and conservation importance affect what testing is reasonable.

Keep three questions separate: can the seed tolerate drying, can it tolerate the proposed storage temperature at that moisture status, and how long does useful viability persist? Passing one question does not answer the others. A seed can survive an initial treatment yet deteriorate during subsequent storage.

For home learning, this distinction is enough to reject the slogan “all seeds belong in a freezer”. Follow reliable instructions for the identified material. Professional alternatives for difficult species may involve living collections, tissue culture or cryopreservation approaches, each with its own requirements. No single storage container is a complete conservation strategy.

Contents · Previous chapter · Next chapter

Moisture content needs a stated basis. Fresh-mass, or wet-basis, moisture content divides water mass by the total initial mass. Dry-basis moisture content divides water mass by dry matter. The same sample therefore gives different percentages without either measurement being contradictory.

In an invented example, a sample initially weighs 10.00 g and its measured dry matter is 8.00 g. Water mass is 2.00 g. Wet-basis moisture is 2/10 × 100 = 20%; dry-basis moisture is 2/8 × 100 = 25%. This is an arithmetic demonstration, not a laboratory drying protocol or a suitable storage target for a particular species.

Assume that dry matter remains exactly 8.00 g and that a hypothetical target is 5% water on a wet basis. Dry matter is then 95% of the final total mass. Final mass must be 8/0.95, approximately 8.421 g. Water remaining is approximately 0.421 g, so water removed from the initial sample is approximately 1.579 g. Subtracting fifteen percent of the original mass would give the wrong answer because the denominator changes during drying.

Equilibrium relative humidity is another quantity. It concerns air in equilibrium with the seeds in a closed measurement system, not the percentage of the seed’s mass that is water. Kew’s hygrometer guidance stresses equilibrium and temperature. A room humidity reading beside a packet is not automatically a measurement of seed moisture status.

Contents · Previous chapter · Next chapter

A storage label should connect the collection to its identity, handling conditions and testing history. The useful question is whether the material still retains the intended capability, not simply whether the container remains present in the freezer.

An ARS-reported study by Walters and colleagues analysed genebank monitoring data across many accessions and species. Storage longevity varied considerably; the study also found that its controlled-storage longevity comparisons did not correlate with soil persistence. A dry genebank and a soil seed bank expose seeds to different processes, so their lifespans should not be treated as interchangeable.

Consider a fictional stock of two thousand seeds. A test uses one hundred, leaving nineteen hundred. If the sample gives 80% normal seedlings under a suitable protocol, multiplying 1,900 by 0.80 gives an estimated 1,520 normal seedlings under comparable conditions. It does not leave two thousand seeds in the collection, and it does not guarantee 1,520 established plants outdoors.

If a later test is lower, examine both biology and measurement. Was the protocol unchanged? Did the sampled portion represent the lot? Did storage conditions deviate? Were observers using the same criterion? A changed percentage deserves investigation before a single cause is assigned.

Monitoring also consumes material. Plan the balance between learning about viability and retaining seeds for future use. Document withdrawals, not just test percentages. A collection can lose practical value through depletion even when the remaining seeds maintain excellent germination performance.

Contents · Previous chapter · Next chapter

Seed priming uses controlled hydration to advance early processes before radicle emergence, commonly followed by drying for handling. It can improve aspects of subsequent germination performance, but the result depends on material, treatment and later storage. It should not be confused with a universal dormancy-breaking method or unlimited soaking.

Bruggink and colleagues’ research investigated how faster germination after priming could be combined with maintained longevity in the species and conditions studied. Its practical lesson is to evaluate both immediate performance and subsequent persistence. A benefit measured soon after treatment does not automatically establish a benefit after storage.

Here is a fictional comparison. Immediately after preparation, Treatment P produces ninety normal seedlings per hundred seeds and Treatment U produces eighty. After a planned storage interval, corresponding fresh samples from the two treated stocks produce sixty-five and seventy-five. If the material will be sown immediately, the first comparison is relevant. If it will be stored before sowing, the second comparison is more relevant.

These figures are separate tests at different times. Do not multiply 0.90 by 0.65 unless the second number was explicitly defined as a conditional survival proportion; here it is another percentage of seeds tested. The incorrect multiplication would invent a transition that was never measured.

Keep a treatment record that includes preparation, storage interval and sowing conditions. Compare the entire intended sequence. Faster emergence is useful only in relation to the actual job, and preserving an untreated reference can help reveal whether a treatment benefit survives that job’s timing.

Contents · Previous chapter · Next chapter

The strongest repair is usually narrower than “try harder” or “add more water”. Locate where the chain diverged from expectation, then collect the observation that best separates plausible explanations.

Consider contamination carefully. Visible mould may accompany dead material, affect developing seedlings or reflect test conditions. Removing affected units from the denominator without documenting a rule can make a poor test look successful. Record what happened and whether the test remains interpretable.

Change one important uncertainty at a time where feasible. If you simultaneously alter substrate, light, temperature, hydration and counting date, a better result may be useful operationally but will not reveal which change mattered. Preserve the original result, describe the revised question and keep the explanation proportional to the evidence.

Contents · Previous chapter · Next chapter

Contents · Previous chapter · Next chapter

The Singapore Botanic Gardens Seed Banking page, updated 24 April 2026, describes its role in preserving Southeast Asian plant material, documenting collections and supporting research. This is a local example of seed biology becoming a sustained scientific practice: identity, processing, storage and testing remain connected over time.

Conservation asks more than whether a handful of seeds germinates. It also asks what populations and genetic variation the collection represents, what material remains available, and whether future plants can survive in suitable habitats. Seed storage supports future options; it does not recreate the ecological relationships of a habitat by itself.

The existing Seed Banks, Dormancy and Germination Evidence guide follows the conservation workflow in more detail. Pair it with this root when you want to connect the institutional process with the biological distinctions behind each measurement.

For a manageable learning project, use identified, lawfully obtained teaching material with a suitable protocol. First produce a dated observation record. Next calculate an endpoint percentage. Then distinguish new from cumulative counts and calculate a timing measure. Finally follow some seedlings to a later stage and explain why that result differs from germination.

Assess the explanation: can you name the denominator, locate the failed transition, distinguish observation from mechanism, and state what would need rechecking in a new environment? Those abilities are useful in horticulture, conservation, seed technology, experimental biology and data interpretation. The How X Works hub offers routes into the wider systems that connect these fields.

Contents · Previous chapter · Next chapter

Does no germination mean that every seed is dead?

No. It describes the result under a particular protocol and time window. Identity, environmental conditions, dormancy and remaining seed condition need evaluation before the result becomes a claim about death. A repeated unsuitable test does not resolve the original uncertainty simply by being repeated.

Can I compare two reported percentages directly?

Only after checking endpoints, denominators, timing and conditions. A percentage based on radicle emergence is different from a normal-seedling percentage. A proportion among full seeds may differ from a proportion among all units sown. Ask for the count table when the denominator is unclear.

Is the fastest lot always the best?

No. Speed is one property. The final fraction, uniformity, storage interval and later establishment can change the decision. The mean time among germinated seeds can also conceal a large non-germinating fraction. Keep the time curve beside the final outcome rather than replacing it.

Does more seed compensate for every problem?

No. A larger sowing can increase expected numbers under a stable probability model, but it does not repair misidentification, an unsuitable environment or a damaging treatment. Density may itself change performance. First identify the limiting step; then evaluate whether additional seed is an appropriate response.

Why can a freezer and a soil seed bank give different results?

They are different environments. Controlled storage aims to limit deterioration for suitable material. Soil exposes seeds to changing conditions and biological interactions while some germinate or are lost. Evidence from one environment does not supply a universal lifespan for the other.

What should a useful report contain?

State the lot identity, history, sampling and allocation, treatment, conditions, observation criterion, timing, raw counts, exclusions and uncertainty. Explain the decision supported and its limits. A second person should be able to reconstruct the percentage and understand what another experiment would need to check.

Sources and an independent next task

The linked Kew technical sheets support the concise explanations of handling, dormancy, testing, moisture and storage behaviour. ISTA provides the standards context and the purpose of validated vigour tests. OpenStax supplies the structural foundation; the cited Arabidopsis, priming and genebank studies illustrate specific experimental questions. Official Singapore Botanic Gardens information supplies the local conservation example. Sources were checked on 24 October 2026; the teaching datasets are original and are not findings from those studies.

Choose one published seed result and reconstruct its claim in one sentence: “These units achieved this endpoint, by this time, under these conditions.” Then change the destination or storage interval and identify which evidence no longer transfers. That is a practical test of the Gold Standard: knowledge that remains useful because its assumptions and limits are visible.

Contents · Previous chapter · Continue to How X Works