DVI from the Osteology Angle: MNI, MNE and the Commingled-Remains Problem

The Minimum Number of Individuals is the smallest number of people that must have been present to account for all the skeletal elements recovered. The calculation depends on one observation: any given bone, for any given individual, appears at most twice in a complete skeleton (left and right for paired elements) or once (for unpaired midline elements). If three left humeri are recovered, at least three individuals contributed them. If five right femora are recovered, at least five individuals are represented. The MNI for any element is therefore the count of the most-represented side (for paired elements) or the total count (for unpaired elements) of that element in the entire assemblage.

In practice, MNI is calculated element by element across the full skeletal inventory, and the highest single-element count becomes the global MNI for the assemblage. If the most-represented element is the left temporal bone at count eight, then MNI equals eight, regardless of whether eight right humeri or eight right femora were also found. The logic is conservative by design: MNI never over-counts.

The calculation must account for age and sex when a single element count would otherwise be misleading. If seven right femora are recovered but two are clearly sub-adult (with unfused distal epiphysis) and five are adult, the MNI from the femur alone is still seven, but the assemblage now contains at minimum two sub-adults and five adults. A large assemblage may be stratified further: right femora from females estimated 35 to 55 years, right femora from males estimated 20 to 40 years, and so on. Each stratification tier contributes separately to MNI, and the final figure is the highest count in any single tier.

The Aviation accident context illustrates the limits. In the ET302 recovery and in the MH17 recovery (298 victims, July 2014, eastern Ukraine), the degree of fragmentation meant that many individual bones were themselves fragmentary. A right femoral shaft fragment with no proximal or distal epiphysis can be side-determined but not reliably size-matched to another fragment without additional criteria. The MNI from right femoral shaft fragments in a high-energy impact may undercount if multiple short shaft segments from the same individual are counted as separate elements.

The Minimum Number of Elements (MNE) complements MNI. Where MNI counts individuals, MNE counts the elements that are present in the assemblage. MNE for a given skeletal element is the estimated number of complete elements represented by all the fragments of that element. A complete right femur is MNE = 1. Three non-overlapping shaft fragments that together cover the full diaphysis but lack epiphyses are also MNE = 1 (they represent one femur, though fragments were produced by fragmentation). Three shaft fragments that do overlap (meaning two or more could belong to the same individual) are harder: MNE is determined by finding the minimum number of complete femora required to account for all fragment positions.

MNE is used alongside MNI in reports because the two together communicate completeness: if MNI is 157 (an aviation disaster) but MNE for the femur is only 80, then nearly half the individuals are represented without a femur, and those individuals will need to be identified through other means (DNA from tooth roots, dental charts, fingerprints).

The Most Likely Number of Individuals (MLNI) was proposed by Adams and Konigsberg in 2004 specifically to address the systematic undercount in MNI. MNI is conservative because it assumes maximum redundancy: it asks whether any of the counted elements could represent the same individual. But in a large commingled assemblage, pair-matching (identifying left and right elements that belong to the same individual based on morphological compatibility: matching size, cortical thickness, bone texture, developmental stage) can elevate the count above MNI. MLNI uses the pair-match data to calculate a maximum-likelihood estimate of the true number of individuals, accounting for the probability of recovery for each element.

The Adams-Konigsberg 2004 formula produces MLNI as a function of MNI, the number of successfully pair-matched elements, and the total element count. In practice, for a well-recovered assemblage where pair-matching is feasible, MLNI tends to be 10 to 30 per cent higher than MNI. For the Srebrenica primary graves (documented by the ICTY OMPF, the Office of the Missing Persons and Forensics of Bosnia-Herzegovina, and its successor ICMP, the International Commission on Missing Persons), the difference between MNI and MLNI in excavated grave complexes was used to argue that secondary grave disturbance had fragmented original burial counts. When Bosnian Serb forces disturbed the primary graves in September 1995 and moved remains to secondary and tertiary locations, the MLNI analysis of the secondary graves showed that the secondary graves contained only a fraction of the individuals whose primary graves had been emptied. This argument was presented in the Krstic trial before the ICTY in 2001 and in subsequent Srebrenica-related proceedings.

Once MNI is established, the osteologist attempts to sort the commingled assemblage into groups of bones that are likely to represent single individuals. The goal is not a perfect reconstruction of each skeleton (that is rarely achievable in high-energy disasters) but the creation of bone groups that can be submitted for DNA typing as single-contributor samples, and that carry a coherent partial biological profile for reconciliation purposes.

Four segregation criteria are applied, in a priority sequence that reflects their discriminating power.

Sex segregation uses the pelvis as the first indicator. The greater sciatic notch width and sub-pubic angle reliably separate male and female adult remains where the pelvis is present and intact. In fragmentary assemblages, the ischiopubic ramus ridge (present in females, absent in males) is identifiable even on small fragments. The skull contributes: mastoid process dimensions, glabella prominence, and supraorbital ridge robustness can be scored on cranial fragments. For postcranial elements without a pelvis, the femoral head diameter (threshold 43 to 45 mm, population-dependent) and humeral head diameter discriminate sex with accuracy above 85 per cent when population-appropriate reference data are used. All bones tentatively assigned as female are segregated from all bones assigned as male; bones that cannot be sexed are held separately.

Age segregation is anchored in epiphyseal fusion. Any bone with an open or fusing epiphysis belongs to a sub-adult. The distal radial epiphysis fuses at approximately 17 to 20 years; the medial clavicular epiphysis (the last to fuse in the body) fuses at approximately 22 to 28 years. Any bone showing these features is segregated into a sub-adult group. Within adults, age estimation from the pubic symphysis (Suchey-Brooks) and auricular surface (Buckberry-Chamberlain) can assign adult remains to broad age brackets (young adult 18 to 35, middle adult 35 to 50, older adult 50+). This bracketing separates, for example, a 25-year-old flight attendant from a 65-year-old passenger even when both are fragmentary.

Pathology segregation is among the most powerful tools. Healed fractures (with cortical remodelling), surgical implants (plates, screws, joint replacements with manufacturer serial numbers on the surface), and skeletal pathologies (osteosarcoma, Paget's disease, severe osteoarthritis) are individually unique markers. A right femur with a well-healed mid-shaft fracture callus, a left hip with a cemented total hip replacement, and a cervical vertebra with a surgical fusion cage all represent unique skeletal signatures. Any other bone that matches the physical size and morphology of a bone associated with one of these unique markers is a candidate for assignment to the same individual. In the MH17 recovery (Netherlands Forensic Institute, Dutch National Police, Belgian Federal Police, working under Dutch leadership from 2014 to 2020), multiple victims were initially identified through surgical implants before DNA confirmation was available. The implant serial numbers were traced to specific hospital records, producing a pre-DNA identification that was then confirmed by DNA typing.

Size segregation uses long-bone length and cortical thickness to group remains. Where a humerus and a femur cannot be sexed or age-estimated independently but share a consistent size and cortical density, they are provisionally assigned to the same size group. This is the weakest of the four criteria, used only after the other three have been applied, because size overlap between individuals is common. It is most useful for separating extreme ends of the size distribution: a femur measuring 480 mm (corresponding to a stature estimate of approximately 190 cm) belongs in a different group from a femur measuring 380 mm (corresponding to approximately 155 to 160 cm).

1. Complete skeletal inventory
   Lay all recovered material on a padded table in anatomical position. Record every element present, its completeness (percentage), side, and any notable features using the Buikstra-Ubelaker 1994 inventory standard. Calculate MNI from the most-represented element.
2. Sex segregation
   Assess every pelvic fragment for sciatic notch and ischiopubic ramus ridge. Score every cranial fragment for mastoid and glabella dimensions. Apply femoral and humeral head diameter thresholds (population-appropriate). Assign all bones to male, female, or indeterminate groups.
3. Age segregation within sex groups
   Within each sex group, identify all sub-adult elements by open or fusing epiphyses. Apply pubic symphysis and auricular surface aging to adult elements. Assign adult elements to broad age brackets. Note that sub-adult elements may require re-sexing after this step because age and sex interact.
4. Pathology and unique-marker identification
   Document all healed fractures (note side, anatomical location, callus morphology), surgical hardware (photograph, record serial numbers where visible), and distinctive pathologies. Use these as anchors: assign morphologically compatible adjacent bones to the same individual group.
5. Size matching and pair-matching
   Within each sex-age group, rank elements by long-bone length. Attempt pair-matching (left-right for all paired elements) using size, cortical thickness, and bone texture. Record successful pair-matches for MLNI calculation. Calculate MLNI from the pair-match data per Adams-Konigsberg 2004.
6. DNA sampling plan
   Select one element per provisional individual group for DNA sampling. Prefer dense cortical bone: the petrous portion of the temporal bone (highest DNA yield in degraded remains, per Pinheiro 2006 and subsequent studies), the mid-shaft femur, or the root of a tooth. Avoid cancellous bone. Submit each group's sample separately.

In mass-disaster contexts, the ideal of a complete skeleton is rarely achieved. The practical question is: given a set of partial, fragmentary, or isolated elements, what biological profile can be defended in court or in an identification board? The answer depends on which elements are present, the quality of preservation, and the population-specific reference data available.

The four-part biological profile (sex, age, ancestry, stature) does not require equal amounts of skeletal material for each component. Sex estimation from a single intact pelvis or a well-preserved pubic symphysis is achievable with 95 per cent accuracy (Walker 2005 review). Sex from the skull alone is achievable at 80 to 90 per cent (Phenice 1969 and successive studies). Sex from a single long-bone metric is achievable at 70 to 85 per cent, population-dependent. A forensic anthropologist working with only a femoral shaft and a right temporal bone can generate a sex estimate, though the confidence interval will be wider than if the pelvis were present. The report must quantify the uncertainty: "Sex estimated as female based on femoral head diameter consistent with female range for the South Asian reference population (Kanchan-Krishan 2011); pelvis not recovered; sex estimation accuracy at this level of evidence approximately 80 per cent."

Age from a partial skeleton follows the same logic. A pubic symphysis in Suchey-Brooks phase III places the individual between approximately 21 and 46 years (broad interval) but with a modal age of 26 to 35 years. If the auricular surface is also present and scores in Buckberry-Chamberlain stages 3 to 4, the combined estimate tightens. If neither pelvic age indicator is available but the fourth rib sternal end is present and scores in Iscan-Loth phase 4 to 5, an age of approximately 40 to 55 years can be defended. Each available indicator is reported with its individual uncertainty; the composite estimate is reported as the intersection of the intervals.

Ancestry estimation from fragments is the most methodologically constrained component. Full craniometric analysis (Howells measurements, FORDISC 3.1 discriminant function) requires a substantially complete calvarium. Many fragmentary assemblages cannot support ancestry estimation, and the report should state this clearly rather than speculate. Where the mandible is substantially intact, mandibular morphoscopic traits (particularly chin shape and gonial angle) contribute weak ancestry information. The modern practitioner reports ancestry as population affinity rather than biological race, per the 2020 AAA and SAA position statements and the parallel OSAC guidance.

Stature from a single long bone is straightforward where the bone is complete or can have its maximum length estimated from shaft fragment length using regression methods. Mukherjee-Bhattacharya (1955) and Pan (1924) equations are used for South Asian populations; Trotter-Gleser (1952, 1958 Korean War correction) for North American Black and White populations; Khanpetch (2012) for Thai populations. Where multiple long bones are present, stature estimates from each are reported and a composite mean or weighted estimate is provided. The ±4 cm standard error typical of regression-based stature estimation must appear in every report.

The interface between the osteological and DNA workflows in DVI is the bone sample. The osteologist selects which element or fragment to sample, which part of that element to sample, and how to handle the sample to preserve DNA yield. These decisions have significant downstream consequences.

DNA yield from bone correlates with the density and mineralisation of the sampled cortex, and inversely with postmortem interval and thermal exposure. The petrous portion of the temporal bone (the dense bone encasing the inner ear) consistently yields the highest DNA concentrations in degraded skeletal remains, often by one to two orders of magnitude compared to long-bone cortex from the same individual. Studies from the ICMP laboratory in Sarajevo (Corach 2001, Loreille 2007, Pinheiro 2009) and from the IRCGN laboratory in Paris have confirmed petrous bone superiority in severely degraded and thermally altered samples. For fresh to moderately degraded remains, mid-shaft femoral cortex is the standard sampling site: it is accessible, provides large amounts of compact bone, and is present in almost every adult recovery.

The osteologist's contribution to the DNA workflow goes beyond sample selection. The pair-matching data generated during segregation directly supports kinship analysis. If an osteologist has established, based on size and morphology, that a right humerus and a left femur are likely from the same individual, that provisional pair can be submitted for kinship comparison against two different family reference profiles. If both the right humerus and the left femur match the same family reference, the pair-matching inference is confirmed. If they match different families, the osteological pair-matching was incorrect, and the elements must be re-evaluated. The ICMP's DVI programme in Bosnia-Herzegovina, which has identified more than 7,000 Srebrenica victims since 2001, uses exactly this feedback loop: osteological pair-matching hypotheses are tested by DNA kinship comparison, and discordant results flag the bones for re-examination.

The sister Forensic Biotechnology subject (pcr-fundamentals-chemistry-primers-cycling-contamination and the STR typing and kinship analysis topics) covers the laboratory workflow in full detail: DNA extraction from bone, PCR amplification of STR loci, capillary electrophoresis, profile comparison, kinship LR calculation, and the database matching standards used by CODIS (US), NDNAD (UK), and the ICMP database in Bosnia. The osteological topic cross-links to that subject at every point where DNA analysis depends on a prior osteological decision.

In India, the Centre for DNA Fingerprinting and Diagnostics (CDFD) in Hyderabad and the national DNA Analysis Centre (NDAC) under the DNA Technology (Use and Application) Regulation Act 2019 carry the capability for mass-disaster DNA typing. The Uttarakhand floods of 2013, in which an estimated 6,000 or more pilgrims died in the Kedarnath valley and surrounding areas with most bodies lost to river transport and burial under debris, represented the largest Indian mass-casualty event in recent memory with a significant osteological component. No coordinated INTERPOL-standard DVI operation was mounted; identification relied on family statements and, for a small number of recoveries, local forensic pathology. The capacity gap was documented in the 2018 National Disaster Management Authority / National Institute of Disaster Management draft DVI framework, which explicitly called for INTERPOL DVI training for a dedicated Indian response team.