Carpology studies fruits and seeds, especially their structure, variation and identification. It helps answer a deceptively simple question: what is this plant object, and what can its features reliably tell us? A small object called a “seed” may actually be an entire one-seeded fruit, a stone enclosing a seed, or a fragment whose original form is no longer visible.
The gold standard of carpology is a clear chain from specimen to observation, comparison and bounded conclusion. This guide explains that chain through fruit walls, seed coats, embryos, reference collections and original worked cases. All numbered specimens, measurements, candidate groups and sampling datasets in the exercises are invented for teaching; they are not identification rules for real species.
The Gold Standard collection provides the wider framework, and Education: learning that transfers into life explains why a learner should be able to apply a method when the specimen or question changes. Here, dependable work means making the evidence inspectable, including when the correct answer remains “not yet identified”.
THE GOLD STANDARD · CARPOLOGY
Read the evidence in fruits and seeds
Choose a route through structure, observation, identification and interpretation. All teaching chapters remain open below.
ROUTE 1 · CHAPTERS 1–4
Understand fruits and seeds
Distinguish objects, layers, counting units and fruit categories.
ROUTE 2 · CHAPTERS 5–8
Observe and measure
Read surfaces, compare views, calculate dimensions and inspect sections.
ROUTE 3 · CHAPTERS 9–12
Identify and sample
Use several characters, reference variation and honest denominators.
ROUTE 4 · CHAPTERS 13–16
Interpret and test
Connect dispersal, germination and archaeology; audit a classifier.
ROUTE 5 · CHAPTERS 17–20
Repair, transfer and continue
Diagnose errors and investigate a mixed museum teaching tray.
Full chapter index · Try the transfer task · Gold Standard subject hub
Chapter index
Understand fruits and seeds · Chapters 1–4
Observe and measure · Chapters 5–8
Identify and sample · Chapters 9–12
Interpret and test · Chapters 13–16
Repair, transfer and continue · Chapters 17–20
CHAPTER 1 OF 20 · Understand fruits and seeds
1. Begin with the object and the question
A carpologist may examine a fruit still attached to a plant, a dry seed from a reference collection, or a damaged object recovered from sediment. These situations offer different evidence. An attached specimen can retain relationships to flowers, leaves and other fruits. An isolated object may have lost precisely the structure needed to distinguish two possibilities.
Before naming it, write down what you want to establish. “Which family could this belong to?” differs from “Did people grow this crop here?” The first is an identification question. The second adds questions about human action, place, time and how the material reached its deposit. Correct identification can support that larger enquiry, but cannot complete it alone.
Kew describes carpology as the study of fruits. In practical fruit-and-seed work, associated seeds and dispersal structures are examined together. Its carpological collection also accommodates bulky plant material unsuitable for ordinary herbarium sheets. A collection’s storage category therefore need not be a strict definition of an organ.
Start your record with a specimen code and a neutral description: “Object C17, dry, approximately oval, one attachment scar visible.” Keep the proposed identification in a separate field. If the name later changes, the observations remain available. This simple separation prevents the original guess from silently becoming the evidence used to defend itself.
CHAPTER 2 OF 20 · Understand fruits and seeds
2. Follow the layers: fruit wall, seed coat and embryo
In flowering plants, the ovary wall develops into the fruit wall, called the pericarp. Inside, an ovule can develop into a seed with its own covering. The embryo is the young plant within the seed; nutritive reserves may occur in different tissues. The USDA introduction to fruits and seeds explains why recognising these nested structures matters for identification.
Imagine three envelopes, each carrying different information. The outer envelope may describe the fruit. The next may describe the seed’s covering. Inside lies the embryo and any other nutritive tissue present. Removing one envelope changes which features you can observe. A description of the outer surface must therefore state which surface it describes.
The analogy has limits. Some layers adhere closely, some fruits include additional floral tissues, and reserves are not always a separate block surrounding a tiny embryo. In many legumes, substantial reserves occur in the cotyledons, which are parts of the embryo. Do not label every pale interior region “endosperm” merely because it looks like food storage.
The practical lesson is to establish relationships before collecting technical names. Draw an outer boundary, mark any separate inner covering, and record whether an embryo is actually visible. If the available view does not reveal it, write “not observed”. “Not observed” and “absent” are different statements with different consequences.
CHAPTER 3 OF 20 · Understand fruits and seeds
3. Recognise the unit that moves or gets counted
A dispersal unit is the structure that travels away from the parent plant. Depending on the plant and its condition, that unit can be a seed, an entire fruit, or a fruit segment. Additional structures may remain attached. What an analyst receives is therefore not always the organ that everyday language suggests.
The USDA fruit guide explains that dry, one-seeded fruits can be mistaken for seeds. A separate fruit wall enclosing a seed is useful evidence. Wings, hairs and persistent floral parts can also help, although none of these features alone provides a universal fruit-versus-seed test.
Consider a fictional tray with ten intact one-seeded fruits, eight loose seeds and six empty fruit-wall fragments. “Twenty-four seeds” would be an incorrect description. The tray contains twenty-four recorded objects, but only eighteen are currently known to contain or constitute seeds. The six fragments may represent several fruits or parts of fewer fruits; their original number is unresolved.
This distinction changes analysis. A count of objects, a count of complete fruits and an estimate of original seeds answer different questions. Preserve separate categories instead of adding unlike units into one attractive total. The same discipline applies when counting shell pieces, cereal grains or fruit stones in archaeological material.
CHAPTER 4 OF 20 · Understand fruits and seeds
4. Use fruit categories as structural descriptions
Fruit names organise observations about development, texture and opening. They are useful because they make comparisons precise. They become misleading when a learner treats one memorable supermarket example as the definition of an entire category.
| Structure | Useful distinction | What to inspect |
|---|---|---|
| Capsule | A dry fruit from a compound ovary that opens at maturity | Opening pattern, chambers and remaining seeds |
| Achene | A dry, one-seeded, non-opening fruit with the fruit wall separate from the seed coat | Relationship of outer wall to seed |
| Caryopsis | A grass fruit with fruit wall united to seed coat | Whole grain and any surrounding bracts |
| Drupe | A fruit with a hardened inner wall enclosing seed material | The stone and the seed inside it |
| Samara | A winged dry fruit | Which structure forms the wing |
The USDA overview distinguishes achenes from caryopses and discusses differences in terminology, including cypselae in the sunflower family. When references use different conventions, record the convention rather than pretending the disagreement is necessarily about the specimen.
A separate distinction concerns origin: a simple fruit comes from one ovary, an aggregate fruit from several separate carpels of one flower, and a multiple fruit from several flowers. Virginia Tech’s botany guide explains these categories. An isolated damaged fruit may not retain enough developmental evidence to settle every category. Use the classification as a question to investigate, not a label to force.
The seed surface offers several kinds of evidence: dimensions, shape, texture, colour, attachment marks and appendages. The hilum is the scar associated with attachment to the funiculus, or seed stalk. Its shape and position can be useful, particularly when combined with other characters. The USDA seed guide also cautions that colour is not always reliable and that size and shape can vary within a species.
Observe the whole object before selecting a striking feature. Turn it over, examine the edges, and distinguish a persistent pattern from a broken surface. A pale patch might be an attachment region, a scrape or material adhering to the specimen. Calling it a hilum without checking its position and structure can send the identification down the wrong route.
An efficient notebook entry records several independent-looking observations: dimensions measured in stated views; outline; surface relief; scar position; completeness; and condition. “Brown and bean-shaped” is a useful first impression, but it leaves most of the discriminating evidence unrecorded.
Keep uncertainty local. If the overall outline is clear but one end is broken, record the outline confidently and the damaged end cautiously. You do not need to downgrade every observation to “unknown”. Equally, confidence about colour should not spill over into confidence about the species name.
CHAPTER 6 OF 20 · Observe and measure
6. Make shape and surface descriptions repeatable
A single photograph converts a three-dimensional object into a two-dimensional outline. A flattened seed may look round from one direction and narrow from another. Record the view so a second observer knows whether “width” means the widest face or the thinner edge.
The USDA shape guide separates outline descriptions from three-dimensional forms. Its surface-relief guide provides comparisons for features such as parallel striations, ridges and grooves. These resources help replace private impressions with descriptions another person can test.
For a teaching comparison, photograph each object in three agreed orientations: widest face, opposite face and edge. Include the same specimen code and a scale in each image. Use lighting that reveals relief without allowing a shadow to masquerade as a groove. Repeat an uncertain view with the illumination coming from another direction.
Do not improve a scientific drawing by making irregularities disappear. If a specimen has a broken margin, show the break. If a diagram reconstructs a missing part, label that reconstruction. The existing Botanical Illustration guide develops the relationship between observed morphology, accurate scale and an honest visual record.
CHAPTER 7 OF 20 · Observe and measure
7. Measure an image without trusting its display size
Suppose a photograph contains a 2 mm scale bar that spans 240 pixels. A seed’s measured length spans 630 pixels and its width spans 360 pixels. Assume the specimen and bar lie in the same measurement plane, the image has no appreciable distortion, and the chosen endpoints are appropriate.
Length = 630 ÷ 240 × 2 mm = 5.25 mm. Width = 360 ÷ 240 × 2 mm = 3.00 mm. The length-to-width ratio is 5.25 ÷ 3.00 = 1.75. This is a description of that specimen in that view, not a species diagnosis.
Now reduce the whole image to half its original pixel dimensions. The bar becomes 120 pixels and the seed length 315 pixels. The calculation remains 315 ÷ 120 × 2 = 5.25 mm. A properly preserved scale bar survives uniform resizing because both measurements change by the same factor.
A caption saying “10× magnification” may not survive changes in screen or print size in the same way. A scale bar that is separately pasted, stretched or cropped can also become unreliable. Preserve the original image and record any transformations. For precise work, use an appropriate calibrated imaging system rather than treating a casual photograph as a measuring instrument.
The ratio 1.75 is dimensionless, but it still depends on correct orientation and endpoints. Removing a wing from the length measurement while including it in the width would create a ratio of incompatible dimensions.
CHAPTER 8 OF 20 · Observe and measure
8. Use internal structure only when the specimen allows it
External features sometimes leave several candidate groups. Internal observation may reveal the embryo’s position and shape, nutritive tissues, or the boundary between fruit wall and seed coat. The USDA dissection guidance shows why a longitudinal view can add information unavailable from the exterior.
Decide whether opening the specimen is justified before doing it. A unique museum specimen, an archaeological object and a plentiful teaching seed have different constraints. Photograph and describe the intact object first. Keep all resulting pieces associated with its code, and record the treatment because preparation can alter what later researchers see.
For introductory learning, prepared sections or instructor-prepared material are often more useful than an improvised cut. The learning objective is to recognise tissue relationships, not to demonstrate cutting speed. A poor section may miss the embryo or crush a diagnostic feature while giving the impression of complete access to the interior.
Suppose one prepared section shows no embryo. There are several possible explanations: the cut missed it, the specimen was empty, the relevant tissue was damaged, or its structure was not recognised. Inspect the section plane and remaining material before concluding that the original seed lacked an embryo. A negative observation needs an opportunity for the feature to have been detected.
CHAPTER 9 OF 20 · Identify and sample
9. Build an identification from several characters
Identification is a comparison between an unknown specimen and a defined range of possibilities. A key narrows that range through character states. It does not create evidence that was absent from the specimen, and it cannot reliably name a taxon that lies outside its coverage.
The Fruit and Seed Family ID key tips explain that distantly related families can share visible features. Users should compare remaining candidates with fuller descriptions and images rather than assuming a thumbnail captures all variation. A family-level result is a family-level result; a species name requires additional support.
Try this fictional comparison. Candidate A has a short attachment scar, ridged surface and nearly circular widest-face outline. Candidate B has a long attachment scar, ridged surface and elongated outline. Candidate C has a long attachment scar, smooth surface and elongated outline. Your specimen has a long scar and elongated outline, but dirt obscures its surface.
A is poorly supported by the recorded scar and outline. B and C remain plausible. The sensible next observation concerns the surface, provided cleaning or imaging can reveal it without damaging the object. Guessing “ridged” because B looks familiar would manufacture the very evidence needed to choose B.
This exercise assumes these three candidates and their stated traits. A real specimen could belong to an unlisted group, and real traits may vary. Treat the miniature key as a model of reasoning, not a transferable botanical identification table.
CHAPTER 10 OF 20 · Identify and sample
10. Compare variation rather than ideal specimens
Reference material should reveal a range, not merely one beautiful example. Compare specimens from documented sources and developmental states, and look for variation in size, shape and preservation. An unknown does not have to match a single photograph pixel for pixel to belong to the same group.
The USDA Fabaceae family account illustrates why broad descriptions need care: this family contains substantial diversity in seed form, surface and internal structure. A familiar bean is not a complete model of every legume seed.
In a fictional reference set, group P has observed lengths from 4.2 to 5.8 mm, while group Q spans 5.0 to 6.4 mm. An unknown measuring 5.4 mm falls within both observed ranges. Size therefore does not separate P from Q here. Even a specimen outside one recorded range would require checking sample coverage and measurement conditions before exclusion.
The useful next step is not to report another decimal place. Seek a feature that discriminates between the remaining groups, such as a verified scar arrangement or internal character. More precise measurement of an uninformative character can produce a more precise ambiguity.
Keep the reference accession, source of identification and observation date. If the comparison specimen’s name changes, those records let you revisit the conclusion without reconstructing the entire investigation from memory.
CHAPTER 11 OF 20 · Identify and sample
11. Design a sample that can answer the question
A tray is rarely a transparent miniature of everything that existed before collection. Someone selected a location, collected material, separated objects and decided what to retain. Each step can alter the representation of large, small, fragile or easily recognised specimens.
For a school investigation using known commercial material, define the population before sampling: one packet, one batch, or several batches. Mix material appropriately and use a documented selection procedure. Choosing the twenty largest objects is a valid way to study the largest objects, but not a random sample of the packet.
Imagine two fictional batches. From batch A, a learner measures twenty intact seeds chosen at random from the accessible material. From batch B, the learner measures twenty seeds selected because they are easy to photograph. A difference in average size may reflect the selection procedure. The problem cannot be repaired simply by calculating both means accurately.
Record the number examined, number measured, number excluded and reason for exclusion. If broken objects cannot supply full length, keep their count visible. A report based only on intact material should say so. Completeness may itself relate to size or condition, making the measured subset systematically different from the rest.
The existing Plant Science guide extends this reasoning into experiments and environmental conditions. Carpology contributes an especially important preliminary check: make sure the compared units really are comparable organs and developmental stages.
CHAPTER 12 OF 20 · Identify and sample
12. Calculate a proportion with an honest denominator
Suppose a fictional sample contains 160 recorded objects. Thirty are confidently assigned to target type T, ninety are assigned to other types, and forty remain unresolved. The unresolved material is retained and recorded, not silently discarded.
Among all recorded objects, confidently identified T makes up 30 ÷ 160 × 100 = 18.75%. Among the 120 identified objects, T makes up 30 ÷ 120 × 100 = 25%. Both calculations are correct, but they answer different questions. Reporting “T is 25% of the sample” without naming the identified-only denominator would mislead the reader.
For a simple sensitivity exercise, assume every recorded object represents one comparable unit and that all forty unresolved objects could in principle be T or non-T. The possible proportion of T ranges from 30 ÷ 160 = 18.75% to 70 ÷ 160 = 43.75%. This is a logical range under stated assumptions, not a confidence interval and not an estimate of the most likely value.
The upper bound would be too broad if some unresolved objects could already be excluded from T. Conversely, the whole calculation would need reconsideration if fragments represented unequal portions of original fruits. State the counting unit before choosing the denominator.
A useful report gives the confidently identified proportion, the unresolved fraction and the evidence needed to narrow it. Uncertainty then becomes an actionable part of the result: which additional views, comparisons or specimens would materially improve the answer?
CHAPTER 13 OF 20 · Interpret and test
13. Connect structure to dispersal without inventing a journey
A wing, hook or fleshy covering can suggest how a plant structure interacts with air, animals or other surroundings. The structure supports a mechanism hypothesis. It does not reveal the actual route taken by every specimen that bears it.
Consider a winged fruit found beside a path. Several histories are possible: local fall, movement by wind, transport with soil or deliberate movement by a person. The wing helps explain potential aerodynamic behaviour, but its presence alone does not establish travel distance, parent-tree location or the route followed on that occasion.
A classroom investigation could compare the descent of paper models with different wing areas while holding their mass and release height as constant as practicable. Record repeated trials and landing positions. Such a model tests a simplified physical relationship; it is not a measurement of a real plant species’ dispersal distance. Outdoors, gusts, obstacles and release orientation introduce further variation.
The reproductive sequence also needs careful language. Pollination and fertilisation are distinct processes, and a visiting insect is not automatically an effective pollinator. The existing guide to pollinators, flowers and fruit explains that earlier part of the life cycle. Carpological observations concern the resulting structures and what remains visible in them.
Ask three separate questions: what feature is present, what mechanism could it support, and what evidence shows that mechanism operated in this case? This sequence keeps a plausible story from outrunning the observation.
CHAPTER 14 OF 20 · Interpret and test
14. Separate identity, maturity and ability to germinate
An object can be correctly identified yet fail to germinate. A seed can also remain ungerminated during a particular test without being dead. Identity, physical condition, dormancy and response under stated conditions are related questions, but they require different evidence.
Kew’s germination guidance explains that environmental conditions and dormancy must be considered. A visually complete seed is not a guarantee of successful germination, and one universal temperature or treatment is not suitable for every species. A morphological description should not be presented as a viability certificate.
In a fictional classroom test, 24 of 40 seeds meet a predefined germination criterion by day fourteen. The observed result is 24 ÷ 40 × 100 = 60% under that protocol and time window. It does not establish that the other sixteen are dead. They may need more time or different conditions; some may indeed be non-viable. Additional appropriate tests would be required to separate those explanations.
Keep the germination record separate from the identification record while linking both through the specimen or batch code. The seed germination guide follows the embryo into seedling development. The Seed Saving guide addresses the practical question of preparing planting material for a future crop.
For carpology, the key boundary is simple: recognising what a seed is does not by itself demonstrate what that individual seed can still do.
CHAPTER 15 OF 20 · Interpret and test
15. Read archaeological fruits and seeds in context
Archaeological specimens arrive through preservation and recovery processes. Historic England’s archaeobotany overview identifies charring, waterlogging, mineralisation and desiccation as preservation routes and describes comparative collections of seeds and fruits. The surviving object is evidence of a particular history, rather than an untouched sample of all plants once present.
Keep the carpological task focused: describe the retained structures, compare appropriate reference material and state the justified identification level. Then connect that result to the deposit, recovery method and other evidence. The existing Archaeobotany root develops the wider interpretation of people and plants.
Suppose a fictional deposit yields twelve identified fruit stones. The defensible first statement is that twelve stones of the reported type were recovered through the stated method. It does not follow that twelve fruits were eaten on that spot. Material could have arrived through disposal, movement or another activity; preservation and recovery may have removed part of the original assemblage.
Likewise, a securely identified crop seed is not automatic proof that the crop was cultivated beside the excavation. Transport and use are alternatives that require contextual evaluation. Strong interpretation comes from converging evidence, not from expanding the species label into an entire agricultural history.
For practice with sediment volumes, recovery and historical claims, continue to Teaching Archaeobotany. Its examples address the contextual step after the fruit-and-seed observations developed here.
CHAPTER 16 OF 20 · Interpret and test
16. Audit an identification method before trusting its accuracy
An identification method should be tested on specimens whose reference identities are independently supported. Avoid evaluating it only on the same images used to develop the rules. Several photographs of one seed do not provide the same independent evidence as several seeds from different sources.
Consider a fictional screening method for target type T. A validation set contains twenty genuine T specimens and eighty non-T specimens. The method calls eighteen of the twenty T specimens positive and calls eight of the eighty non-T specimens positive.
| Reference identity | Method calls T | Method calls non-T |
|---|---|---|
| T: 20 specimens | 18 | 2 |
| Non-T: 80 specimens | 8 | 72 |
Sensitivity is 18 ÷ 20 = 90%: the fraction of reference T specimens detected. Specificity is 72 ÷ 80 = 90%: the fraction of reference non-T specimens correctly excluded. Yet only 18 ÷ 26, approximately 69.2%, of positive calls are correct in this set. A claim that “a positive is 90% reliable” would confuse different quantities.
Now assume the same sensitivity and specificity held in a new collection where only ten of one thousand objects were T. Expected true positives would be nine, while expected false positives would be ninety-nine. The expected positive predictive value would be 9 ÷ 108, approximately 8.3%. These are hypothetical expectations, not a guarantee of counts in a real collection.
The second calculation shows why prevalence matters. It does not prove performance will remain stable across preservation states, cameras or populations. Both the specimen distribution and the method’s error rates need reassessment when the application changes.
CHAPTER 17 OF 20 · Repair, transfer and continue
17. Diagnose common failures and repair the observation
When a result feels uncertain, identify the failing step. Repeating the final name more confidently will not repair a missing scale, a damaged attachment scar or a reference set that excludes likely candidates.
| Symptom | Likely problem to investigate | Useful repair |
|---|---|---|
| Two photographs seem to show different sizes | Different image scales or views | Recalculate from a valid scale and match orientation |
| A key eliminates every candidate | Misread state, specimen damage or limited key coverage | Revisit uncertain observations and check scope |
| One observer sees ridges and another sees grooves | Lighting, terminology or viewing direction | Compare the same view under changed illumination |
| Every object is assigned to one familiar species | Anchoring on a favourite comparison | Test plausible alternatives and retain unresolved objects |
| A percentage changes between reports | Different denominators or counting units | Reconstruct the count table and label each denominator |
| No embryo appears in a section | Section plane, condition or recognition problem | Inspect remaining material before declaring absence |
Repair should preserve the original record. If an identification changes from species to genus, retain the earlier determination, date and reason for revision. If a measurement was taken from the wrong axis, mark it as superseded and record the corrected method. Quietly replacing old values destroys the evidence needed to understand why the conclusion changed.
An unresolved specimen is not wasted work when its observations, images and location are preserved. It can be re-examined when a better reference becomes available. A confident unsupported name is often harder to repair because later users may never realise that the original identification was provisional.
CHAPTER 18 OF 20 · Repair, transfer and continue
18. Transfer task: a mixed museum teaching tray
A museum lends a class photographs of a teaching tray. The tray contains sixty numbered objects: thirty complete one-seeded fruits, twenty loose seeds and ten fruit-wall fragments. All objects belong to the same loan, but their parent plants and collection locations are not documented in the exercise.
One photographed object has a removable outer wall enclosing a separate seed. The 1 mm scale bar spans 150 pixels; the entire object spans 720 pixels along its chosen long axis. The extracted seed spans 540 pixels in a second image whose 1 mm bar spans 180 pixels. A learner concludes that the seed is 75% as long as the fruit because 540 ÷ 720 = 0.75.
The class confidently assigns fifteen complete objects to type R and leaves the remaining forty-five tray objects unresolved. Another learner reports that one quarter of the original plants were R. A third says the unassigned objects must belong to other species because they failed to match the first reference photograph.
Work out the fruit length, seed length and correct length ratio. Then explain what can be said about the fifteen R identifications, why the plant-level claim fails, and which observations would be most useful next. Assume that the image bars are valid and both axial measurements are comparable.
Answer and reasoning
Fruit length = 720 ÷ 150 × 1 = 4.8 mm. Seed length = 540 ÷ 180 × 1 = 3.0 mm. The ratio is 3.0 ÷ 4.8 = 0.625, or 62.5%. Dividing raw pixel lengths across differently scaled images produced the wrong answer.
The removable outer wall and separate inner seed support describing the whole object as a fruit containing a seed. That observation does not establish a species. The photographs need additional external and, if appropriate, internal characters matched to a suitable reference.
Fifteen of sixty recorded tray objects, or 25%, are confidently assigned to R. The number of original plants cannot be calculated: one plant may contribute several fruits, and fragments need not represent independent original units. Even a proportion of original fruits requires assumptions about the fragments and loose seeds that the exercise does not supply.
Failure to match one reference photograph leaves an identification unresolved; it does not demonstrate a different species. Useful next steps include matching views, documenting attachment scars and surface features, checking a wider reference range, and recovering provenance records if they exist. Keep the ten fragments separate throughout the analysis. The changed setting tests whether the method survives new scales, mixed counting units and missing context.
CHAPTER 19 OF 20 · Repair, transfer and continue
19. Build a practical learning route
Begin with a small set of known, lawfully obtained teaching specimens. Commercial food material can help introduce fruit walls and seed coats, but processing may remove useful structures. Keep packaging or source information and distinguish a supplier’s label from an independently checked botanical identification.
In the first session, draw each whole object and describe its parts without trying to name every taxon. In the second, produce comparable photographs and calculate dimensions from valid scales. In the third, use a suitable reference or key and record which candidates remain. In the fourth, exchange coded specimens with another learner and compare the observations before comparing the names.
Assess the explanation as well as the answer. Can the learner distinguish fruit from seed, identify the counting unit, show the measurement calculation, and explain why an alternative remains possible? These capabilities transfer better than memorising a row of specimen names.
For botanical work, continue into collections and taxonomy. For archaeology, practise contextual sampling and interpretation. For planting and conservation, add species-appropriate seed biology and testing. For visual documentation, develop drawing and calibrated photography. Each route uses carpological observations but adds evidence appropriate to its own decision.
The How X Works hub provides routes into the wider mechanisms behind living systems. A useful next investigation is to choose one unfamiliar fruit and explain the relationship between its visible structures, the seed it contains and the evidence still needed to identify it reliably.
CHAPTER 20 OF 20 · Repair, transfer and continue
20. Questions, sources and the next investigation
Is carpology the same as seed saving?
No. Carpology examines fruits and seeds as structures and evidence. Seed saving concerns obtaining and maintaining suitable planting material. They overlap when identification and maturity affect a practical decision, but a good description does not replace storage or germination knowledge.
Can one photograph identify a species?
Sometimes a photograph contains distinctive evidence, but one view can omit scale, attachment marks or internal structure. Report the level supported by the available information. A useful family-level identification is better than an unsupported species name.
Does a hard shell mean I am looking at a seed coat?
Not necessarily. The hard part could be a fruit wall or a hardened inner fruit layer enclosing a seed. Trace the layers and compare the appropriate reference. Everyday names such as “seed”, “nut” and “stone” do not settle the anatomical question.
Why keep fragments that cannot yet be named?
They document the limits of the recovered material and may retain useful characters. Excluding them without a record can distort proportions. Preserve their association with the sample and state how they enter, or do not enter, each calculation.
Does a reference collection prove that every label is correct?
No collection is beyond revision. Accession records, identification history, associated specimens and documented comparisons make a determination inspectable. Confidence comes from the evidence and its review, not from the appearance of an old label alone.
Which sources support this guide?
The main technical reference is Islam, Gottlieb, Maher, Scher and Redford’s 2024 Fruit and Seed Family ID, produced by USDA APHIS PPQ’s Identification Technology Program. Its scope statement, morphology pages, family descriptions and key guidance support the structural distinctions used here. The tool is intended to support family identification; further work may be needed for species determination. Its citation and copyright page identifies public-domain text and separate restrictions that may apply to images. No source images are reproduced in this article.
Kew’s collection account supports the discussion of carpological collections, and its 2022 Technical Information Sheet 13b supports the distinction between germination conditions and dormancy. Virginia Tech’s botany guide supports developmental fruit categories. Historic England’s archaeobotany overview, updated 23 October 2025, supports the short archaeological context section. These sources were consulted on 23 October 2026; their dates should not be confused with the dates of the invented exercises.
The numerical cases are original teaching models with explicit assumptions. None provides a commercial grading rule, species diagnostic threshold or prediction for an undocumented specimen. Your next task is to build a record another observer can check: object, view, scale, characters, alternatives and the strongest conclusion those observations support.