








































Forensic Anthropology Vol. 2, No. 2: 129–138
DOI: 10.5744/fa.2019.1002

© 2019 University of Florida Press

Individual identification is a primary goal for forensic anthro-
pologists who work in medicolegal and humanitarian  contexts. 
However, when skeletal remains are commingled, analyses 
and individual identification can become more complicated 
and even impossible until commingling is at least partially 
resolved. A collection of bones, or assemblage, with  duplicated 
elements or overlapping portions indicates that commingling 
is present; however, it should be noted that the lack of dupli-
cated elements does not always indicate that only one individ-
ual is present. Recovery context and methods, scale of the 
incident, and the degree of commingling affect which meth-
ods may be appropriate to identify and resolve commingling 
within an assemblage (Byrd & Adams 2014; Ubelaker 2002).

One of the first steps in analyzing commingled assem-
blages is to determine the number of individuals represented, 
which can be done using a variety of methods. This step is crit-
ical for determining the scale of an incident and assessing 

whether all individuals have been recovered, as well as for 
planning purposes, to include resource allocation, analyti-
cal approaches, and the direction of future recovery efforts.

Assemblages vary based on recovery context and tapho-
nomic conditions affecting the associated elements, and the 
various individuating methods can provide different esti-
mates. Few methods have been adequately compared or 
tested on assemblages of known size (Lambacher et al. 2016; 
Lyman 2018; Mack et al. 2016; Marean & Spencer 1991; 
Nikita & Lahr 2011; Robb 2016; Sussman 2017), and esti-
mates can be hindered by inconsistent inventory terminol-
ogy. Although the exact number of individuals represented 
in this assemblage is not known, there is an expected num-
ber of individuals based on historical information, which can 
be compared with the various estimates presented in this arti-
cle. This article presents the results of several MNI methods, 
highlights a standardized way to inventory skeletal elements, 
and assesses the most likely number of individuals (MLNI) 
from a large commingled sample.

Individuating Methods

MNI and MLNI are methods used to quantify the number of 
individuals represented in an assemblage. The MNI meth-
ods were developed to ensure that each individual is only 
counted once in an assemblage with the intent of avoiding 
overestimations. Minimum number quantifications have 
roots in zooarchaeological analyses, which typically include 

aIndiana University of Pennsylvania College of Humanities and Social 
Sciences— Anthropology, Indiana, PA, USA

bDefense POW/MIA Accounting Agency, Joint Base Pearl Harbor–
Hickam, Hawai’i, USA

cDefense POW/MIA Accounting Agency— Laboratory, Offutt AFB, 
NE, USA

dUniversity of South Florida, Anthropology, Tampa, FL, USA
*Correspondence to: Andrea Palmiotto, Indiana University of 

Pennsylvania College of Humanities and Social Sciences— Anthropology, 
McElhaney Hall Rm G1 1011 South Drive, Indiana, PA 15705, USA

E-mail: apalmiot@iup.edu

Received 7 June 2018; Revised 9 August 2018;  
Accepted 29 August 2018

TECHNICAL NOTE

Estimating the Number of Individuals in a Large 
Commingled Assemblage

Andrea Palmiottoa,b* ● Carrie Ann Brownc ● Carrie B. LeGardec,d

ABSTRACT: Estimation of the number of individuals in an assemblage is critical to determine the scale of an incident and whether all 
expected individuals have been accounted for. However, estimates are affected by recovery and other taphonomic factors inherent to the 
assemblage, as well as the estimation methods themselves. This study examines several quantification methods using data from the com-
mingled remains of individuals who were aboard the USS Oklahoma at the time of its sinking. Alternatives to traditional minimum number 
of individuals (MNI) quantifications are presented, to include MNI by duplicated elements per mitochondrial DNA (mtDNA) sequence and 
use of the zonal inventory procedure. These methods are assessed to determine which provides an estimate closest to the expected popu-
lation size. Potential advantages and limitations associated with each method are discussed. For this assemblage, a combination of mtDNA 
sequencing using hypervariable regions 1 and 2 (HV1, HV2) and element duplication provides the most accurate estimate of individuals, 
and methods that employ pair- matching perform better than those that do not.

KEYWORDS: forensic anthropology, commingled skeletal remains, MNI, zonation method



130 Estimating the Number of Individuals in Commingled Assemblages

commingled assemblages from multiple species (Reitz & 
Wing 2008).

MNI determines the smallest number of individuals 
needed to account for the physical remains that are recov-
ered in an assemblage (Lyman 1994; Reitz & Wing 2008). 
This is typically done by determining the minimum number 
of elements (MNE) for each element, which is generally 
defined in this application as a single complete bone or tooth 
(e.g., sacrum, left humerus, etc.), but may include skeletal 
portions (e.g., proximal left humerus), and therefore must be 
defined at the beginning of analysis. The most abundant ele-
ment per taxon represents the MNI estimate for that species. 
Variations in MNE within an assemblage provide the addi-
tional benefit of depicting differential recovery and under- 
represented elements. A more robust MNI estimate can be 
calculated when age or other unique indicators are consid-
ered. Overall, however, calculations of MNE and MNI are 
relatively easy, provided accurate and consistent documen-
tation of the skeletal elements.

In cases of fragmented remains, MNI is traditionally 
estimated based on the reported completeness of the most 
common element (Ubelaker 1974; White 1953). Present por-
tions of bone (e.g., proximal or distal) are used as the basis 
of quantification with the goal to identify anatomically over-
lapping portions. In this study, “traditional” MNI and MNE 
are calculated based on simple designations of present por-
tions. However, this method is most effective when assem-
blages are small, remains are not highly fragmented, and/or 
only a single analyst consistently assesses the remains. When 
multiple analysts are involved in the inventory process, or 
when assemblages are large and/or highly fragmented, dif-
ferences in interpretations of remains and inventory pro-
cesses can skew MNI estimates.

The number of individual specimens per taxon (NISP) 
is another quantitative measure that in essence is simply a 
raw count of all of the elements and fragments within an 
assemblage. This measure should not be used to account for 
the number of associated individuals, because it may lead to 
considerable overestimation, particularly when fragmenta-
tion is high (Dominguez- Rodrigo 2012; Gifford- Gonzalez 
2018; Grayson 1984; Kline & Crub- Urize 1984; Lyman 2018; 
Marshall & Pilgram 1993).

In zooarchaeological cases, the entirety of fauna at a site 
is rarely, if ever, recovered. Additionally, taxa representation 
must be accounted for within an assemblage, especially for 
cases where fauna contain different numbers of bones per 
skeleton (e.g., fish versus mammal skeletons). In order to 
explore activity and site use patterns within and across sites, 
MNI is, therefore, a valuable tool. However, MNI tends to 
represent ordinal- scale relationships between species, and it 
is not intended to provide absolute counts of individuals 
(Lyman 2018; Reitz & Wing 2008). When applied to foren-
sic anthropology, the context for MNI assessments differs. 
Only one species is emphasized in forensic anthropology 

cases, and forensic anthropologists often aim to recover the 
entirety of a given skeletal assemblage, not subsets of a larger 
population. For forensic anthropologists, absolute counts are 
necessary to determine if all expected individuals have been 
recovered or if additional recovery operations are required.

To address one of the issues in MNI calculation— the 
subjectivity of completeness descriptors in initial inventory 
procedures— Knüsel and Outram (2004) developed a stan-
dardized and replicable method to estimate MNI from human 
remains in bio archae ol o gi cal assemblages. Each skeletal ele-
ment is divided into zones based on expected postmortem 
breakage patterns as observed in non- human archaeological 
exemplars. The number of zones per element ranges from 1 
(e.g., carpals) to 15 (e.g., cranium) and is often characterized 
by diagnostic traits that allow for the identification of isolated 
fragments. Additionally, at least one zone must be described 
per identified fragment. Knüsel and Outram (2004) indicate 
that zones are counted even if the zone is only partially rep-
resented; they do not give any indication how much of a zone 
should be observable in order to be marked as present.

Methods that recognize paired elements in an assem-
blage include grand minimum total (GMT; Horton 1984) 
and most likely number of individuals (MLNI; Adams & 
Konigsberg 2004). The former produces an MNI estimate 
based on counts of paired and unpaired elements, and the lat-
ter calculates the maximum likelihood estimate of the total 
number of individuals that may have originally comprised an 
assemblage. Additionally, the highest density region (HDR; 
Adams & Konigsberg 2004) can be calculated to determine 
an associated confidence interval for MLNI. Pair- matching 
ideally is conducted on the most frequently recovered antim-
eres. An element must be compared to all contralateral ele-
ments and accurately matched. Uncertainties in pair- matching 
will have a negative impact on the accuracy of GMT and 
MLNI. Therefore, accurate inventories and pair- matching are 
critical to estimate the number of individuals represented.

MLNI recognizes an inherent bias in MNI: because 
MNI is calculated using only one element per taxon, any indi-
viduals that are not represented by that element will not be 
accounted for. MLNI is based on capture- recapture studies 
of living animals and has been modified for osseous collec-
tions. MLNI considers the unlikelihood of recovering 100% 
of an assemblage and attempts to account for individuals that 
may not be present in the pair- matched elements. Previous 
assessments of MLNI typically have focused on smaller 
assemblages (<50 individuals) where the size of the original 
loss population was unknown (Konigsberg & Adams 2014). 
MLNI has rarely been tested on larger assemblages of known 
size; Konigsberg and Adams (2014) assessed variable recov-
ery rates and compared MLNI and MNI estimates on simu-
lated populations containing up to 150 individuals. They 
discussed the high accuracy of MLNI versus MNI (GMT) 
estimates when recovery rates reached or exceeded 50%. 
Konigsberg and Adams (2014) suggest that the remains must 



Palmiotto et al. 131

be well preserved and at least 50% of an assemblage must be 
recovered to adequately assess MLNI.

The application of any of these methods is dependent 
on the condition of the assemblage. Taphonomic factors can 
affect the integrity of bone, such as the amount of fragmen-
tation and degree of preservation. Additional concerns include 
resource availability, allotted analytical time, and expertise of 
the analysts. Not every method can be applied for every situ-
ation; therefore, the selection of methods should be carefully 
considered.

Materials and Methods

The assemblage used for this study consists of the commin-
gled skeletal remains recovered from the USS Oklahoma. 
Following identification efforts in the 1940s, the number of 
unidentified Oklahoma casualties was established as 394. For 
information on this assemblage see Brown (2019).

All elements in the assemblage were inventoried, mea-
sured, and labeled prior to other analyses, and all data were 
compiled in Microsoft Excel spreadsheets. The elements are 
largely intact with limited fragmentation. Observed tapho-
nomic alterations include adherent oil residue, rust staining, 
and cortical erosion. Inventory of the assemblage includes a 
traditional description of the completeness and condition of 
the remains. Traditional descriptors were reported; elements 
were characterized as “partial,” “mostly complete,” or “com-
plete,” while the present portions (e.g., proximal, distal, and 
shaft) were additionally reported for long bones. At the same 
time, the presence of zones was recorded. For this research a 
zone was counted as present if at least 50% of the zone was 
present.

Traditional MNI and MNE estimates are based on the 
traditional descriptors and present portions. In contrast to tra-
ditional MNI and MNE estimates, we calculate zonation 
MNI and MNE wherein the most common zone is identified 
per element. The zonation method limits the bias of interob-
server variation (e.g., what constitutes proximal one- third or 
distal one- fourth) that may exist. Because of the low degree 
of fragmentation in this assemblage, however, we expect that 
these traditional and zonal MNE numbers will be similar.

Where applicable and during the initial inventory, ele-
ments were assessed for pair matches and articulations within 
each “bundle”1 providing the basis of an anthropologically 
informed DNA sampling strategy (e.g., if an analyst assessed 
contralateral elements as pair matches, only the left side was 
sampled for DNA testing; see Brown 2019). Exceptions to 
this practice include the sampling of all complete cranial 
remains and all left and right humeri and tibiae to aid 

1. Remains were interred in bundles, with multiple bundles of 
remains per casket. For more information on the historical analyses and 
interment of the remains, see Brown 2019.

calculations of MNI. Nearly 5,000 elements were sampled 
for DNA analysis. Sequencing of the DNA focuses primar-
ily on HV1 and HV2.

The initial inventory recorded nearly 13,000 elements. 
Based on the percentages derived from computed tomography 
volume rendering (Rowbotham et al. 2017), approximately 
55% of the expected remains are represented. Traditional and 
zonal MNE were calculated for the assemblage based on the 
cranium, humerus, and femur. These elements are included in 
this study because all mtDNA analyses have been completed 
for cranial and dental remains, both left and right humeri 
were sampled for DNA analyses, and the femur is commonly 
considered one of the most durable bones in the human body.

GMT and MLNI were calculated for the humerus 
because this element was subjected to large- scale pair- 
matching. The equations used to calculate GMT and MLNI 
are presented below, where L is the number of left humeri, R 
is the number of right humeri, and P is the count of the num-
ber of pairs (Adams & Konigsberg 2004):

 GMT = L + R − P

MLNI = (L +1)(R +1)
P +1

−1
⎢

⎣
⎢

⎥

⎦
⎥

The pair- matching numbers used in this study are from 
a test conducted by a single, experienced analyst (LeGarde 
2019). Since all humeri were sampled for DNA, the pair- 
matching results are verified based on mtDNA results to 
date. Finally, MNI was calculated based on the most numer-
ous duplicated element per mtDNA sequence in order to 
assess how additional information, in this case mtDNA data, 
can aid in more accurate approximations of the population 
size (e.g., for an mtDNA sequence that has no associated 
skeletal duplication, MNI = 1; however, for a sequence that 
includes three right femora, MNI = 3).

In attempt to describe the amount of fragmentation pres-
ent per element within the assemblage, the MNE is divided 
by the NISP (Table 1). The resulting value represents the 
estimated frequency of complete elements, or completion rate. 
The idea behind it is that an assemblage with no fragmenta-
tion will have a 100% completion rate (i.e., NISP = MNE), 
while a highly fragmented assemblage will have a lower 

TABLE 1—Summary of USS Oklahoma osseous remains 
discussed in text.

Element NISP Percent of Complete Elementsa

Cranium 421 86.46
Left femur 359 95.54
Right femur 363 94.49
Left humerus 306 95.42
Right humerus 304 96.05
aPercent of complete elements calculated using NISP divided 
by zonal MNE (see Table 2).



132 Estimating the Number of Individuals in Commingled Assemblages

completion rate. This frequency can be used to quickly assess 
the overall fragmentation of an assemblage once MNE has 
been calculated.

Additionally, an accepted MNI is established based on 
duplicated elements per mtDNA sequence obtained. Although 
there is an expected number of individuals based on historical 
context (394), theoretically, the results from the duplicated ele-
ments per mtDNA sequence provide the most accurate repre-
sentation of the number of individuals because (1) the expected 
number is based on the available historical information, 
which may be incomplete; (2) extensive mtDNA sampling 
was conducted for the project; and (3) more than a single ele-
ment type is considered by the results based on duplicated 
elements per mtDNA sequence. Therefore, the MNI based on 
duplicated elements per mtDNA sequence is considered the 
accepted MNI and is used to assess the assemblage in terms 
of the historical MNI and the other MNI methods.

Although the accepted MNI is a robust calculation that 
should account for the majority of individuals, it should not 
be assumed as the actual number of individuals, because (1) 
not all elements were recovered for this assemblage; (2) not 
all individuals may be represented by the samples taken for 
DNA analysis; and (3) even if no duplicated elements are 
present in a sequence, the available elements could still rep-
resent more than one individual, and this study does not 
attempt to determine MNI based on other methods (e.g., 
osteometric sorting, pair- matching [of elements aside from 
the humerus], or articulation).

Results

A total of 421 crania, 722 femora, and 610 humeri are included 
in the Oklahoma inventory (see Table 1). These numbers 
reflect the NISP for each element and include fragments, 
which may represent overlapping portions, so these raw 
counts cannot be used to determine the number of individuals. 
Additionally, the long bone NISP does not include elements 
that could not be sided during the inventory process. From 

the humeri, 287 left and 293 right humeri were used for the 
pair- matching study (representing an approximate 73% 
humeri recovery rate); the remaining humeral fragments 
were too incomplete for pair- matching.

MNE, MNI, and MLNI estimates are represented in 
Table 2 and Figure 1. Based on the inventory, 357 mostly 
complete or complete crania are present, but cranial MNE is 
364 individuals based on the presence of zone 5 (occipital 
region; Fig. 2). The right distal femur is the most common 
portion based on traditional reporting (336 individuals), but 
femoral MNE is 343 based on the presence of zone 2 (lesser 
trochanter region; Fig. 3). The right distal humerus is the 
most common portion based on traditional reporting (289 
individuals), but humeral MNE is 292 based on the presence 
of zone 5 (lateral epicondyle region; Fig. 4).

LeGarde identified 211 humeri pairs in the pair- matching 
study (see LeGarde 2019). Based on this assessment, the 
GMT is estimated as 369. When the pair- matching data are 
used to calculate the MLNI, the result is 398 with the 95% 
HDR encompassing 385– 414 individuals (Fig. 5). Addition-
ally, a total of 310 unique mtDNA sequences based on HV1 
and HV2 results are associated with the project as of Febru-
ary 2018 (2,980 samples processed). When duplicated ele-
ments per mtDNA sequence are considered, MNI is 400.

The results indicate variability between MNE, MNI, and 
MLNI estimates (see Table 2, Fig. 1). MLNI (398) and MNI 
based on the number of duplicated elements per mtDNA 
sequence (400) provide the highest estimates, while tradi-
tional MNI reporting results in the lowest estimates.

Discussion

In this assemblage, MNI by duplicated elements per mtDNA 
sequence and MLNI provide estimates that are nearly iden-
tical to the expected historical population size. GMT is the 
next most accurate method, followed by MNE based on 
zones. As expected, the low degree of fragmentation results 
in similar estimates for traditional and zonal MNE; however, 

TABLE 2—Estimated Number of Individuals (MNE, MNI, and MLNI) for the USS Oklahoma Assemblage.

Element Method Description
Estimated Number of 
Individuals

Duplicated elementsa MNI: mtDNA — 400
Humeri MLNI — 398 (95% HDR 385– 414)
Humeri MNI: GMT — 369
Cranium MNE: Zonation Zone #5 364
Cranium MNE: Traditional Mostly complete 357
Right femur MNE: Zonation Zone #2 343
Right femur MNE: Traditional Distal portion 336
Right humerus MNE: Zonation Zone #5 292
Right humerus MNE: Traditional Distal portion 289
aBased on duplicated elements per mtDNA sequence (310 sequences from 2,980 samples, as of 
February 2018).



Palmiotto et al. 133

FIG. 1—Estimated number of individuals (MNE, MNI, and MLNI) for the USS Oklahoma assemblage, where DE = duplicated elements, Z = zonation 
method, and T = traditional method.

FIG. 2—Cranium divided into zones, following the Knüsel and Outram method (illustration by Nandar Yukyi and Emily Streetman).



134 Estimating the Number of Individuals in Commingled Assemblages

FIG. 3—Femur divided into zones, following the Knüsel and Outram 
method (illustration by Elizabeth Lockett).

FIG.4—Humerus divided into zones, following the Knüsel and Outram 
method (illustration by Elizabeth Lockett).

FIG.5—The most likely number of individuals (MLNI) with the 95% 
HDR for the USS Oklahoma assemblage.

zonal MNE is consistently higher than traditional estimates 
among all elements.

The accepted MNI, determined by duplicated elements 
per mtDNA sequence, provides the most accurate number of 

individuals associated with this assemblage when consid-
ering the number of individuals unidentified following the 
incident. Because this number is slightly higher than the 
expected loss number of 394, it could indicate that some indi-
viduals were mistakenly associated with the Oklahoma upon 
recovery, individuals not listed as Oklahoma casualties were 
on board the ship at the time of the incident, and/or portions 
of previously identified individuals are present in the cur-
rent assemblage.

While MNI by duplicated elements per mtDNA sequence 
is the most accurate, the DNA testing for the project is not 
yet complete; approximately 60% of the total sampled ele-
ments were used to estimate the original population size. It 
will likely take at least another year to process the remaining 
samples and conduct additional nuclear DNA testing. Thus, 
MNI by duplicated elements per mtDNA sequence does not 
represent the timeliest or most cost- effective strategy, and 
given that samples are still in analysis, MNI may fluctuate as 
more DNA results are received.



Palmiotto et al. 135

A challenge for both MLNI and GMT is that these meth-
ods necessitate accurate pair- matching across the entire 
assemblage. LeGarde’s study indicates that accurate large- 
scale pair- matching is feasible; therefore, this method can be 
used to estimate GMT and MLNI. However, this may not be 
possible for all assemblages, and if assemblage size, time lim-
its, and/or condition of the remains precludes the use of pair- 
matching, the zonation method provides a reliable, easily 
applied alternative.

Despite the low level of fragmentation (see Table 1), 
zonal MNE outperformed traditional MNE estimates for all 
tested elements. The zonation method provides a replicable 
means to assess fragmented remains, regardless of the degree 
of fragmentation or the number of analysts working on a proj-
ect. Additionally, the zonation method is standardized, 
unlike estimates based on traditional and more subjective 
reporting techniques. Because the zonation method can be 
used accurately where other methods are precluded, it is valu-
able for forensic contexts. Compared to DNA, it is also rela-
tively inexpensive, requiring only the analysts’ time.

The MLNI assessment is consistent with the accepted 
MNI and the historical estimate, and it corroborates previ-
ous findings (Konigsberg & Adams 2014), but on a larger 
scale. With a humeri recovery rate of approximately 73%, the 
MLNI results of the USS Oklahoma are in general agreement 
with previous simulations (Konigsberg & Adams 2014). Con-
sidering that fewer humeri were recovered than femora 
(approximately 73% versus 85% element recovery, respec-
tively; see Tables 1 and 2), it is unknown how MLNI may 
have been affected if femora had been subjected to pair- 
matching instead. Furthermore, when using the MLNI for 
planning purposes, the MLNI value should be preferred over 
the associated interval. The associated interval may under-  
or overestimate, which can have negative impacts on plan-
ning, resource allocation, field recovery efforts, and analytical 
approaches.

Conclusion

Resolution of commingled assemblages is critical for indi-
vidual identification. An important first step in this process 
is to determine the number of individuals represented in an 
assemblage, and multiple methods exist to complete this task. 
Although it is necessary to have various methods to account for 
the vastly different types of assemblages and taphonomic con-
ditions affecting the associated elements, very few methods 
have been compared or tested on assemblages of known size.

This study assesses multiple methods of quantifying 
individuals using remains associated with the Oklahoma 
assemblage as well as introducing MNI by mtDNA sequence 
and testing the zonal inventory procedure. Traditional meth-
ods provided the lowest estimate in this study, while the 
alternative methods appear to be more accurate. All of the 

estimates discussed herein can be calculated fairly easily and 
provide more accurate results than traditional MNI estimates, 
although only MLNI and MNI by duplicated elements per 
mtDNA sequences provided estimates comparable to the 
historical estimate.

However, among the alternative methods, none is advo-
cated over another, because every osseous assemblage and 
associated context is different; methods should be considered 
and applied as appropriate per project. Most importantly, 
before any individuating estimates can be determined, it is 
critical to inventory remains in a consistent, standardized 
way. The Commingled Remains Analytics (CoRA) is a web 
application that can be used for inventorying commingled 
human remains assemblages and includes features such as a 
zone module to assist in MNI calculations. It was developed 
specifically to address challenges associated with ongoing 
commingled projects at the DPAA. For more information 
about CoRA, see Brown (2019).

If time and cost are not an issue, extensive DNA testing 
and anthropological analysis can be used effectively to esti-
mate the number of individuals present in an assemblage with 
a high degree of accuracy. When elements are mostly com-
plete, pair- matching can be used to determine GMT and 
MLNI, both of which can provide relatively accurate esti-
mates. This study did not include pair- matching of the most 
common contralateral elements, did not test differences in 
estimates when using more than one element type, and did 
not assess the feasibility of pair- matching in an assemblage 
with moderate- to- pronounced fragmentation.

Although it provided a lower estimate than the other 
alternative methods in this study, the zonation method is 
replicable, standardized, useful on complete or fragmented 
assemblages, and it can be implemented where pair- matching 
and DNA analyses are not options. The zonation method is 
especially useful in fragmented cases, and therefore can be 
applied where other methods are precluded. Zonal MNE out-
performed traditional MNE even with the low degree of ele-
ment fragmentation within the assemblage.

Future research, however, should examine the utility of 
the cranial zones (see Fig. 2). The majority of cranial bones 
are designated as a zone in and of themselves (e.g., the right 
and left parietal represent zones 3 and 4, respectively, of the 
15 cranial zones). Because we determined that half of a zone 
must be present to be counted, two halves of the right pari-
etal have the possibility of being counted as two individuals, 
even if they originate from the same individual. It is possi-
ble, therefore, in cases where cranial remains are highly frag-
mented, for the zonal method to overestimate the number of 
individuals based on cranial fragments. To ensure a standard-
ized and accurate cranial inventory, future research should 
explore refining cranial zones or possibly testing other meth-
ods, such as the Landmark method (Mack et al. 2016).

Some slight modifications are recommended to the zones 
from the original article (Knüsel & Outram 2004) based on 



136 Estimating the Number of Individuals in Commingled Assemblages

practice with the method and applicability for commingled 
assemblages (Appendix A): we renumbered the mandible so 
that the zones are specific to left and right sides; provided fig-
ures for the cranium and mandible separately; updated 
radius zone “J” to 11 and all ulna letters to numbers for con-
sistency with other elements; changed the carpal and tarsal 
(except calcaneus and talus) letter scheme to numbers (zone 
1 for all carpals); condensed phalanges into a single zone (1) 
as they are not as probative as other elements and analysis 
does not greatly benefit from the level of detail that multiple 
phalanges zones provides; and divided ribs into four zones, 
breaking zone 3 into shaft (zone 3) and sternal end (zone 4).

The Oklahoma project presents a robust assemblage with 
a low degree of fragmentation, which provides an optimal 
platform to assess individuating methods. Additional testing 
of assemblages with different recovery rates and/or degrees 
of fragmentation, however, would provide invaluable insight 
and comparison. Future research should focus on the rela-
tionship of recovery rate and estimation of the original pop-
ulation size, including the consideration of what elements are 
potentially more useful for the resolution of commingling 
and analyses leading to identification.

Acknowledgments

The views herein are those of the authors and do not represent 
those of the Defense POW/MIA Accounting Agency, Depart-
ment of Defense, or United States government. This research 
was supported in part by appointments to the Postgraduate 
and Student Research Participation Programs at the Defense 
POW/MIA Accounting Agency administered by the Oak 
Ridge Institute for Science and Education through an inter-
agency agreement between the U.S. Department of Energy 
and DPAA. Portions of this paper were presented at the 
70th Annual American Academy of Forensic Sciences meet-
ing in Seattle, WA. The authors would like to acknowledge 
the contributions of the Armed Forces DNA Identification 
Laboratory: Franklin E. Damann, Jeffery Lynch, Sarah C. 
Kindschuh, Sachin Pawaskar, Nandar Yukyi, Emily Street-
man, Elizabeth Lockett, Yangseung Jeong, Lara McCormick, 
Petra Banks, Ashley Burch, and Sarah Shankel.

References

Adams BJ, Konigsberg LW. Estimation of the most likely number 
of individuals from commingled human skeletal remains. 
American Journal of Physical Anthropology 2004;125(2): 
138– 151.

Byrd JE, Adams BJ. Analysis of commingled human remains. In: 
Blau S, Ubelaker DH, eds. Handbook of Forensic Anthropol-
ogy and Archaeology. 2nd ed. New York: Routledge; 2014: 
226– 242.

Dominguez- Rodrigo M. Critical review of the MNI (minimum 
number of individuals) as a zooarchaeological unit of quanti-
fication. Archaeological and Anthropological Sciences 2012; 
4(1):47– 59.

Gifford- Gonzalez D. An Introduction to Zooarchaeology. New 
York: Springer; 2018.

Grayson DK. Quantitative Zooarchaeology: Topics in the Analy-
sis of Archaeological Faunas. Orlando: Academic Press; 1984.

Horton D. Minimum numbers: A consideration. Journal of Archae-
ological Science 1984;11(3):255– 271.

Klein RG, Cruz- Uribe K. The Analysis of Animal Bones from 
Archeological Sites. Chicago: University of Chicago Press; 
1984.

Knüsel CJ, Outram AK. Fragmentation: The zonation method 
applied to fragmented human remains from archaeological 
and forensic contexts. Environmental Archaeology 2004;9(1): 
85– 98.

Konigsberg LW, Adams BJ. Estimating the number of individuals 
represented by commingled human remains: A critical evalu-
ation of methods. In: Adams BJ, Byrd JE, eds. Commingled 
Human Remains: Methods in Recovery, Analysis, and Identi-
fication. San Diego: Academic Press; 2014:193– 220.

Lambacher N, Gerdau- Radonic K, Bonthorne E, Valle de Taraz-
aga Montero FJ. Evaluating three methods to estimate the 
number of individuals from a commingled context. Journal of 
Archaeological Science: Reports 2016;10:674– 683.

Lyman RL. A critical review of four efforts to resurrect MNI in 
zooarchaeology. Journal of Archaeological Method and The-
ory 2018; Early View. doi: 10  .1007  /s10816  - 018  - 9365  - 3.

Lyman RL. Quantitative units and terminology in zooarchaeology. 
American Antiquity 1994;59(1):36– 71.

Mack JE, Waterman AJ, Racila A- M, Artz JA, Lillios KT. Apply-
ing zooarchaeological methods to interpret mortuary behav-
ior and taphonomy in commingled burials: The case study of 
the late neolithic site of Bolores, Portugal. International Jour-
nal of Osteoarchaeology 2016; 26(3):524– 536.

Marean CW, Spencer LM. Impact of carnivore ravaging on zooar-
chaeological measures of element abundance. American Antiq-
uity 1991;56(4):645– 658.

Marshall F, Pilgram T. NISP vs. MNI in quantification of body- part 
representation. American Antiquity 1993;58(2):261– 269.

Nikita E, Lahr MM. Simple algorithms for the estimation of the 
initial number of individuals in commingled skeletal remains. 
American Journal of Physical Anthropology 2011;146(4): 
629– 636.

Reitz EJ, Wing ES. Zooarchaeology. 2nd ed. New York: Cambridge 
University Press; 2008.

Robb J. What can we really say about skeletal part representation, 
MNI and funerary ritual? A simulation approach. Journal of 
Archaeological Science: Reports 2016;10:684– 692.

Rowbothan SK, Blau S, Hislop- Jambrich J. Recording skeletal com-
pleteness: A standardized approach. Forensic Science Inter-
national 2017;275:117– 123.

Sussman E. Comparing Methodologies for Documenting Commin-
gled and Fragmentary Human Remains [master’s thesis]. 
Greenville, NC: East Carolina University; 2017.

Ubelaker DH. Approaches to the study of commingling in human 
skeletal biology. In: Haglund WD, Sorg MH, eds. Advances 
in Forensic Taphonomy: Method, Theory, and Archaeological 
Perspectives. Boca Raton, FL: CRC Press; 2002: 331– 352.

Ubelaker DH. Reconstruction of Demographic Profiles from Ossu-
ary Skeletal Samples: A Case Study from the Tidewater Poto-
mac. Washington, DC: Smithsonian Institution Press; 1974.

White TE. A method of calculating the dietary percentage of vari-
ous food animals utilized by aboriginal peoples. American 
Antiquity 1953;18(4):396– 398.



Palmiotto et al. 137

FIG. A1—Left mandible divided into zones, following the Knüsel and 
Outram method (illustration by Nandar Yukyi and Emily Streetman).

FIG. A2—Right mandible divided into zones, following the Knüsel and 
Outram method (illustration by Nandar Yukyi and Emily Streetman).

FIG. A3—Radius divided into zones, following the Knüsel and Outram 
method (illustration by Nandar Yukyi and Emily Streetman).

FIG. A4—Ulna divided into zones, following the Knüsel and Outram 
method (illustration by Nandar Yukyi and Emily Streetman).

Appendix A. Updated Images of Zones for Use in 
Forensic and Bio archae ol o gi cal Cases, Based on 
the Knüsel and Outram Method



138 Estimating the Number of Individuals in Commingled Assemblages

FIG. A5—Bones of the left hand divided into zones, following the Knüsel 
and Outram method (illustration by Nandar Yukyi and Emily Streetman).

FIG. A6—Bones of the right foot divided into zones, following the 
Knüsel and Outram method (illustration by Nandar Yukyi and Emily 
Streetman).

FIG. A7—Rib divided into zones, following the Knüsel and Outram 
method (illustration by Nandar Yukyi and Emily Streetman).

FIG. A8—First rib divided into zones, following the Knüsel and Outram 
method (illustration by Nandar Yukyi and Emily Streetman).


