Commingled remains can preserve clear lesions while obscuring which bones belonged to the same person. That creates a measurement problem: a count of affected fragments is not automatically a count of affected individuals.
The Gold Standard begins by naming the unit of observation, building a defensible minimum number of elements and individuals, and reporting a range when person-level attribution cannot be recovered.
This approach does not make commingled assemblages useless. It makes their evidence more precise by separating what the bones show from what the present association can no longer prove.
The central question is: how many distinct people must have had the lesion, how many could have had it, and which assumptions connect those bounds?
eduKateSG · The Gold Standard · Archaeology
Find your next learning step
Move from fragments to elements to people without silently changing the denominator.
Full chapter index · Worked example · Transfer check · Archaeology subject root
Continue through The Gold Standard of Education, the Gold Standard collection and the How X Works hub.
Chapter contents
Define the units · Chapters 1–3
Build the bounds · Chapters 4–6
Test association · Chapters 7–9
Transfer and report · Chapters 10–12
CHAPTER 1 OF 12 · Define the units
1. Commingling breaks the easy link between bone and person
In an articulated burial, several observations can often be assigned to one skeleton. In a commingled deposit, fragments and elements from multiple people may be spatially mixed through burial practice, disturbance, recovery or curation.
A lesion can therefore be real at the bone level while its person-level meaning remains uncertain. The analysis must not count every affected fragment as a different individual or collapse all affected bones into one person.
| Unit | What it counts | Main risk |
|---|---|---|
| Fragment | Recorded piece of bone | One element may appear several times |
| Element | Anatomical bone or portion | Left and right sides may belong to one person |
| Individual | Distinct person | Association may be uncertain |
| Context | Deposit or analytical group | Mixing may join different episodes |
CHAPTER 2 OF 12 · Define the units
2. Begin with context, inventory and observability
Record provenience, recovery history, fragmentation, preservation and whether refitting or articulation survives. Then inventory elements by side, portion, age class and diagnostic landmarks.
For pathology, define which surfaces must be observable. An absent lesion on an unobservable surface is missing evidence, not a negative observation.
The minimum number of elements consolidates fragments that could belong to the same anatomical element. The minimum number of individuals then uses the most abundant compatible side, age or other non-duplicating category.
Both are minimums. They establish that at least a certain number is present; they do not prove that no additional people contributed poorly represented bones.
| Inventory clue | Contribution | Caution |
|---|---|---|
| Repeated left elements | Raises the minimum count | Fragment overlap must be checked |
| Age-incompatible portions | Separates possible individuals | Age estimates carry uncertainty |
| Refits | Reduces double counting | A failed refit is not proof of difference |
| Taphonomic pattern | Explains mixing or loss | Does not identify a person by itself |
CHAPTER 4 OF 12 · Build the bounds
4. Worked example: calculate person-level bounds
A commingled adult assemblage has an MNI of 24 based on femora. Eight observable left femora and five observable right femora show the same diagnostic lesion. No reliable left–right pair matching is available.
At least eight people were affected, because the eight left lesions cannot come from fewer than eight individuals. At most 13 people were affected if every affected left and right femur belonged to a different person. Relative to the MNI, the person-level lesion prevalence is therefore bounded from 8/24 = 33.3% to 13/24 = 54.2%.
| Quantity | Value | Meaning |
|---|---|---|
| MNI | 24 | Minimum people represented |
| Affected-person minimum | 8 | Largest affected count on one side |
| Affected-person maximum | 13 | All affected left and right elements belong to different people |
| Prevalence bound | 33.3%–54.2% | Conditional on the MNI denominator and stated pairing assumptions |
Practical check: the true number of people represented may exceed the MNI. The 33.3%–54.2% range is a conditional bound for this analytical denominator, not the cemetery’s final disease prevalence.
If only 15 of 24 left femora preserve the relevant surface, prevalence among all 24 left femora understates the observable evidence. If only the best-preserved bones are selected, the result may also overstate or understate disease depending on the lesion and preservation process.
Report observable elements, affected observable elements and the rules used to classify uncertain changes. Keep indeterminate observations separate from absent lesions.
Morphology, size, preservation, refitting, pathology pattern, osteometrics, isotopes or genetic evidence may support association between left and right elements. Each method has a different error structure and should be documented.
A confident pair of two affected femora reduces the maximum number of affected people by one. A possible pair should not be treated as definite simply because it produces a tidier prevalence.
CHAPTER 7 OF 12 · Test association
7. Use bounds when attribution remains unresolved
Bounds are an honest result, not a failed estimate. They show what every permissible reconstruction shares and where reconstructions diverge.
The lower bound answers how many people must be affected. The upper bound answers how many could be affected under the most dispersed compatible assignment. Additional association evidence can tighten either side.
| Statement | Supported? | Reason |
|---|---|---|
| Eight or more people had the lesion | Yes | Eight affected left femora require eight people |
| Exactly thirteen people had the lesion | No | Affected sides may pair |
| The lesion affected 33.3%–54.2% of the MNI | Conditionally | Depends on MNI and pairing assumptions |
| The whole living population had this prevalence | No | Burial, survival, recovery and MNI filters remain |
CHAPTER 8 OF 12 · Test association
8. Do not mix person-level and element-level rates
Element-level prevalence can compare the proportion of observable bones affected. Person-level prevalence describes people. Both can be useful, but they answer different questions.
A study may report element-level rates when individualisation is impossible, then add person-level bounds for lesions that can be mapped through sided elements. The labels must remain visible.
CHAPTER 9 OF 12 · Test association
9. Aggregation choices can create or erase patterns
Combining contexts can raise MNI, change observability and mix burial episodes. Splitting a deposit too finely can count the same individual more than once if bones moved between contexts.
Run the calculation at justified contextual scales and show how the result changes under plausible aggregation. Stability across scales strengthens the pattern; large shifts become part of the interpretation.
Suppose 18 loose lower first molars show enamel defects. Without secure association, two affected teeth from opposite sides may belong to one person or two. Tooth position, side, wear, development and refitting alter the possible assignments.
The logic transfers: define the anatomical unit, establish observable specimens, identify minimum and maximum affected persons, and keep tooth-level frequency separate from person-level prevalence.
Transfer question: what changes when the lesion can occur multiple times on one bone? The recording unit must include lesion location or episode so repeated marks are not mistaken for additional people.
MNI is not the true population size. A high MNI can still omit people represented by non-diagnostic fragments, and a low MNI may reflect recovery or aggregation choices.
DNA or isotopes do not automatically solve commingling. Sampling is partial, preservation varies and laboratory matches have uncertainty. Association evidence must remain proportional to its quality.
CHAPTER 12 OF 12 · Transfer and report
12. Report the reconstruction so it can be challenged
Publish the inventory rules, observable surfaces, lesion criteria, MNE and MNI method, association evidence, aggregation scale and calculations for lower and upper bounds.
A reader should be able to rebuild the range and see which new evidence would narrow it. That is more useful than a precise percentage whose person assignments are invisible.
Evidence check: preserve fragment, element and person tables separately; state every denominator; and label the final prevalence as observed, bounded or model-based.
Frequently asked questions
Why not divide affected fragments by all fragments?
Because fragmentation varies. One person’s bone can produce several pieces, so fragment frequency does not equal person prevalence.
Is the upper bound always affected left plus affected right?
Only under the stated anatomical and denominator assumptions. Other elements, age incompatibilities, duplication and association evidence can change the compatible maximum.
Can commingled remains support population comparison?
Yes, when recovery, observability, aggregation and uncertainty are comparable or modelled. Comparisons should not hide different analytical units.
Sources and further reading
- Adams and Konigsberg — Estimating individuals from commingled remains
- Lambacher and colleagues — Evaluating MNI methods in commingled contexts
- BABAO — Updated recording standards for human remains
- STARC — Excavation and study of commingled human skeletal remains
- Abegg and colleagues — Reconstituting paleopathological profiles from commingled populations
Connected eduKate reading: MNI, fragmentation and aggregation bias · pathology prevalence and lesion visibility · age profiles and survivorship · enamel defects and stress timing
Return to reading routes · Return to the Archaeology subject root