Identification and analysis of human remains recovered from wells from the 1991 War in Croatia.

Abstract

From 1996 to the present, the remains of 61 individuals killed during the 1991 War in Croatia were recovered from both dried out and functioning wells. Positive identification was established in 60.7% or 37/61 cases. Remains recovered from the same geographical region but from non-well settings were identified in 77.4% or 1256/1623 cases. The purpose of this paper is to report on the taphonomic, demographic and trauma characteristics of remains recovered from wells and identify factors responsible for the discrepancy in the identification ratios. The age and sex distributions in the well and non-well series were similar, as were the frequencies of recovered personal documents, jewelry and other artifacts. The taphonomic features of the remains were, however, significantly different. Preservation of remains was considerably better in non-well settings (1400/1623 or 86.3% compared to 40/61 or 65.5% in wells). Congruently, commingling of remains was more frequent in wells (26/61 or 42.6% compared to 77/1623 or 4.7% in non-well settings). In bodies recovered from non-well settings the preservation, state and commingling of the remains were strongly correlated with positive identification. None of these features were correlated with the identification of bodies from wells. Instead, identification of remains from wells was significantly affected by the presence or absence of water in the well. As both series have similar frequencies, and identical rankings of identification factors, the reason for the discrepancy in the identification ratios lies in the fact that by themselves, these factors were rarely sufficient for positive identification. In both series the majority of identifications (51.4% in the well, and 58.1% in the non-well series) were established through a combination of biological and non-biological features. The significantly lower identification ratio in the well series resulted from the difficult recovery conditions in wells with significant amounts of water, and the negative effect that water had on the preservation of clothing, personal artifacts and some dental features. Significant differences were also noted in the types and locations of perimortem traumas. There is a significantly higher frequency of entrance gunshot wounds to the back of the head, suggestive of execution style murders, in the well series.

Full text

Page 1

Identification and analysis of human remains recovered from wells
from the 1991 War in Croatia
Mario Sˇlausa,b,*, Davor Strinovic ´b, Nives Pec ´ina-Sˇlausc, Hrvoje Brkic ´d,
Drinko Balic ˇevic ´e, Vedrana Petrovec ˇkib, Tatjana Cicvara Pec ´inaf
aDepartment of Archaeology, Croatian Academy of Sciences and Arts, A. Kovac ˇic ´a 5, 10000 Zagreb, Croatia
bDepartment of Forensic Medicine and Criminalistics, School of Medicine, University of Zagreb, Sˇalata 11, 10000 Zagreb, Croatia
cDepartment of Biology, School of Medicine, University of Zagreb, Sˇalata 3, 10000 Zagreb, Croatia
dDepartment of Dental Anthropology, School of Dental Medicine, University of Zagreb, Gundulic ´eva 5, 10000 Zagreb, Croatia
eDepartment of Pathology, Sisters of Mercy Hospital, Vinogradska 29, 10000 Zagreb, Croatia
fRegional Health Center Zˇeljeznic ˇar, Grgura Ninskoga 3, 10000 Zagreb, Croatia
Received 13 March 2006; received in revised form 11 May 2006; accepted 6 October 2006
Available online 15 November 2006
Abstract
From 1996 to the present, the remains of 61 individuals killed during the 1991 War in Croatia were recovered from both dried out and
functioning wells. Positive identification was established in 60.7% or 37/61 cases. Remains recovered from the same geographical region but from
non-well settings were identified in 77.4% or 1256/1623 cases. The purpose of this paper is to report on the taphonomic, demographic and trauma
characteristics of remains recovered from wells and identify factors responsible for the discrepancy in the identification ratios.
Theageandsexdistributionsinthewellandnon-wellseriesweresimilar,aswerethefrequenciesofrecoveredpersonaldocuments,jewelryand
other artifacts. The taphonomic features of the remains were, however, significantly different. Preservation of remains was considerably better in
non-well settings (1400/1623 or 86.3% compared to 40/61 or 65.5% in wells). Congruently, commingling of remains was more frequent in wells
(26/61 or 42.6% compared to 77/1623 or 4.7% in non-well settings). In bodies recovered from non-well settings the preservation, state and
commingling of the remains were strongly correlated with positive identification. None of these features were correlated with the identification of
bodies from wells. Instead, identification of remains from wells was significantly affected by the presence or absence of water in the well. As both
series have similar frequencies, and identical rankings of identification factors, the reason for the discrepancy in the identification ratios lies in the
fact that by themselves, these factors were rarely sufficient for positive identification. In both series the majority of identifications (51.4% in the
well, and 58.1% in the non-well series) were established through a combination of biological and non-biological features. The significantly lower
identification ratio in the well series resulted from the difficult recovery conditions in wells with significant amounts of water, and the negative
effect that water had on the preservation of clothing, personal artifacts and some dental features.
Significant differences were also noted in the types and locations of perimortem traumas. There is a significantly higher frequency of entrance
gunshot wounds to the back of the head, suggestive of execution style murders, in the well series.
# 2006 Elsevier Ireland Ltd. All rights reserved.
Keywords: Forensic anthropology; Mass graves; Positive identification; Wells; 1991 War in Croatia
1. Introduction
The process of recovering and identifying human remains
from individual and mass graves has proven to be the most
effective method of resolving the fate of missing individuals in
the former Yugoslavia [1–7]. The primary objective of these
efforts is two-fold: first, to identify and determine if possible the
manner of death of the recovered individuals, and second, to
bring closure to living family members, thus supporting the
human rights of both the living and the deceased. The 1991 war
between Croatia and Serbia that followed the dissolution of the
formerYugoslaviacausedextensivematerialdestructionandloss
of life. The conflict was characterized by executions and mass
www.elsevier.com/locate/forsciint
Forensic Science International 171 (2007) 37–43
* Correspondingauthorat:DepartmentofArchaeology,CroatianAcademyof
Sciences and Arts, Ante Kovac ˇic ´a 5, 10000 Zagreb, Croatia.
Tel.: +385 1 4698 256; fax: +385 1 4856 211.
E-mail address: mario.slaus@zg.htnet.hr (M. Sˇlaus).
0379-0738/$ – see front matter # 2006 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.forsciint.2006.10.003

Page 2

burials of civilians and military personnel, extensive destruction
of private and government property, and the intentional
destruction of schools, hospitals, museums and cultural objects.
Alltogether,from1991to1995,thereweremorethan14,000war
related deaths in Croatia. To deal with this humanitarian crisis,
theCroatianGovernmentformedtheCommittee forImprisoned
andMissingIndividuals.Followingthereintegrationofoccupied
territories in 1995 the Committee recoveredthe remainsof 3398
individuals. These individuals were recovered from individual
burials, burials containing two individuals, and mass graves—
defined as graves containing three or more individuals. Apart
from the number of individuals present in a burial, further
subcategories relevant to the recovery or state of the remains
could also be identified. Some of the victims were partially or
completely incinerated, a small number were recovered from
septic tanks, several were recovered from fish-ponds, and 61
individualswererecoveredfromwells.Toourknowledge,thisis
the largest number of individuals recovered from this type of
setting related to a specific conflict. So far, the remains of 3459
previously missing individuals have been recovered, of which
2878havebeenidentifiedgivinganidentificationratioof83.2%.
However, significant differences were noted in the identification
ratios of individuals recovered from non-well settings (2841/
3398 or 83.6%) and those from wells (37/61 or 60.7%;
x2= 20.97, P = 0.000004). Avalid question that can, therefore,
be asked is in what way did the remains recovered from wells
differ from remains recovered from non-well settings, and what
factors are responsible for the discrepancy in the identification
ratios.Toanswerthesequestionsdatacollectedfromindividuals
recovered from wells were compared with those from a large
sampleofindividualsrecoveredfromnon-wellcontextsfromthe
same geographical region. This sample consisted of 1623
individuals recovered from 51 individual and mass graves in
continental Croatia. The identification ratio in this sample—
1256/1623(77.4%)wassimilartotheoverallidentificationratio,
and also significantly higher than the identification ratio of
individuals recovered from wells (x2= 8.31, P = 0.003). The
factors that were analyzed included the preservation, state
and commingling of the remains, characteristics of the
wells including the presence or absence of water, and the
contributionofdifferentbiologicalandnon-biologicalfeaturesto
identification.
2. Materials and methods
Human remains were recovered from 13 wells that yielded a total of 61
individuals. Six wells contained the remains of one individual, one well the
remains of two, while six contained three or more individuals.
Standard recovery procedures were used in non-well settings. Recovery of
victimsfromwellswasconsiderablymorecomplicatedbecauseoffactorssuchas
the depths of the wells (which varied from 11 to 31 m), the presence of water in
some wells, the danger of the well walls collapsing in some of the dilapidated
wells, and the presence of unexploded explosive devices (recovered from two
wells).Slightlymorethanhalf(7/13)ofthewellscontainedsignificantamountsof
water,fourhadcompletelydriedout,whiletwocontainedsmallamountsofwater.
All of the wells contained large amounts of garbage that hindered recovery
of the remains. This garbage was in the form of debris from destroyed houses,
earth,bricks,brokenappliances,furniture,andautomobileandengine-partsthat
had been used to cover up the bodies. In all but two wells, the bodies were
locatedinonelayer,inthedeepeststratumofthewell,withthedifferenttypesof
garbage thrown on top to hide or deny access to the bodies. In one well the
bodies were distributed in two layers separated by an approximately 1.5 m deep
layerofgarbagewhileintheotherthebodiesandgarbagewereevenlydispersed
along the complete depth of the well.
Timebetweendeath and recoveryof the bodieswasbetween4 and14 years,
allowing significant decomposition—up to complete skeletonization. In some
wellshuman remainswere, therefore,extensivelycommingled.Membersofthe
identification team were present on the surface and determined the number of
individuals present in the well. As each body was recovered it was assigned a
locationnameandnumber.Incasesofextensivecomminglingalloftheremains
were removed together as a single layer, given a series of numbers based on a
preliminary analysis of the probable number of individuals present, and sent to
the Department of Forensic medicine in Zagreb for re-individualization.
In both well and non-well settings sex was determined according to the
following criteria: in saponified remains by examination of the external
genitalia, in skeletonized and partially skeletonized remains by analysis of
pelvicandcranialmorphology[8,9]andbydiscriminantfunctionsforthefemur
developed for the modern Croatian population [10]. Adult age at death was
estimated using as many methods as possible including: ectocranial suture
fusion [11], pubic symphysis morphology [12–16], auricular surface morphol-
ogy [17], sternal rib end changes [18,19] and the Lamendin dental age protocol
[20]. The large number of methods used was dictated by differences in
preservation and causes of death, and the subsequent differential availability
of specific bone elements. In subadults, age-at-death was estimated using
epiphyseal fusion,diaphyseallengths,
[8,9,14,21,22].
To identify factors potentially responsible for the noted discrepancy in the
identification ratios between well and non-well contexts, data were collected on
the characteristics of the recovered remains, the manner of death when
determinable, and the factors responsible for positive identification. These
factorsincluded:sex,age-at-death, stature,bodyand facialhair, dentalremains,
evidence of antemortem surgery or fractures, tattoos or scars, clothing, personal
documents, personal jewelry and artifacts. In all cases a primary factor
responsible for positive identification was selected among the following
possibilities: forensic DNA testing, dental criteria, antemortem surgical inter-
ventions, antemortem fractures, tattoos or scars, clothing, personal documents,
personal jewelry and personal artifacts. If none of these were applicable, the
factor ‘‘a combination of biological and non-biological features’’ was selected.
All of the recovered remains were also analyzed for the presence of
perimortemtraumas.Thisincludedanalysesofthetypesofperimortemtraumas
(one or multiple gunshot wounds, shrapnel traumas, blunt force trauma, etc.),
the numberand distributionoftraumason the bodyor skeleton, and thelocation
and pathways of the gunshot wounds as determined by the locations of entrance
and exit gunshot wounds defined by the criteria of Di Maio [23], Smith et al.
[24,25], and Quatrehomme and I˙s ¸can [26,27].
and dental eruptioncriteria
3. Results
The taphonomic features of the remains recovered from
wells and non-well settings were significantly different. Well
preserved bodies were more common in non-well settings
(1400/1623 or 86.3%) than in wells (40/61 or 65.5%;
x2= 18.66, P = 0.00001). Commingling of the remains was
also much more frequent in wells (26/61 or 42.6%) than in non-
well settings (77/1623 or 4.7%; x2= 140.40, P = 0.000).
Surprisingly, no significant difference was noted in the
frequencies of skeletonized or partially skeletonized remains
(33/61 or 54.1% in wells and 961/1623 or 59.2% in non-well
settings, x2= 0.44, P = 0.506).
In bodies recovered from non-well settings the preservation,
state and commingling of the remains was strongly correlated
with positive identification (Table 1). Well-preserved, saponi-
fied bodies had a much better chance of being identified than
M. Sˇlaus et al./Forensic Science International 171 (2007) 37–4338

Page 3

poorly preserved skeletonized remains that needed to be re-
individualized. None of these features were correlated with the
identificationofbodiesfromwells.Afeaturethatwas,however,
stronglycorrelatedwiththeidentificationofremainsfromwells
was the presence or absence of water in the well. Only 7/22
(31.8%) individuals recovered from wells containing water
were positively identified, compared to 30/39 (76.9%) of
individuals recovered from dry wells. This difference is
statisticallysignificant(x2= 10.17,P = 0.001).Thepresenceof
water also significantly affected the state and commingling of
the recovered remains (Table 2). Wells with water contained a
considerably higher frequency of skeletonized or partially
skeletonized bodies than dry wells (the difference was
marginally not significant), and commingling of remains was
significantly (x2= 10.89, P = 0.0009) more frequent in wells
with water than in dry wells.
The age and sex distributions in the two series were similar.
Identified individuals from the well series consisted of 25
males and 12 females giving a male/female ratio of 2.08. The
mean age-at-death for the complete sample was 47.1 years
(S.D. = 16.7). The average age-at-death of identified females
was53.6years (S.D. = 15.6),
(S.D. = 16.6). The non-well series consisted of 991 identified
males and 265 identified females giving a male/female ratio
of 3.73. The mean age-at-death for the complete sample was
47.0 years (S.D. = 18.62). The average age-at-death of
identified females was 58.9 years (S.D. = 15.74), of males
42.8 years (S.D. = 16.71). Sex did not play a role in
identification in either series. In the well series 25/44
(56.8%) males were identified compared to 12/17 (70.6%)
females. This difference was not statistically significant. In
the non-well series the frequencies were almost identical—
ofmales44.0years
Table 1
The preservation, state and commingling of remains cross-tabulated with identification status for wells and non-well contexts
Na
Id/unidb
% Idc
x2
ðYatesÞ
P
Wells
Preservation of remainsPartially or poorly
preserved bodies
Well preserved bodies
2113/8 61.9 0.00 1.00
40 24/16 60.0
State of remainsSkeletonized or
partially skeleton.
Saponified
3319/14 57.60.07 0.78
2818/10 64.3
Commingling of remainsCommingling
No commingling
26
35
13/13
11/24
50.0
31.4
1.440.23
Non-well settings
Preservation of remainsPartially or poorly
preserved bodies
Well preserved bodies
223125/9856.0 65.83
<0.00001
14001131/26980.8
State of remainsSkeletonized or
partially skeleton.
Saponified
961 706/25573.5 20.16
<0.00001
662550/11283.1
Commingling of remainsCommingling
No commingling
7733/44
1223/323
42.9
79.1
53.02
<0.00001
1546
aNumber of bodies.
bIdentified/unidentified individuals.
c% of individuals identified.
Table 2
The preservation, state and commingling of remains in wells with water and dry wells
Wells with water Dry wells
x2
ðYatesÞ
P
Na
%b
N%
Preservation of remainsPartially or poorly
preserved bodies
Well preserved bodies
1150.0 1025.6 2.69 0.100
11 50.029 74.4
State of remainsSkeletonized or
partially skeleton.
Saponified
16 72.7 1743.63.700.054
627.3 2256.4
Commingling of remainsCommingling
No commingling
16
6
72.7
27.3
10
29
25.6
74.4
10.890.0009
aNumber of bodies recovered.
b% of bodies recovered.
M. Sˇlaus et al./Forensic Science International 171 (2007) 37–43 39

Page 4

991/1279 (77.5%) males, and 265/344 (77.0%) females were
positively identified.
Cause of death was determined in 34/37 or 91.8% of
identified individuals from wells, and in 1111/1256 or 88.4% of
identified individuals from the non-well series. In the well
series the most frequent type of trauma causing death was a
single gunshot wound (23/34 or 67.7%), followed by multiple
gunshot wounds (6/34, or 17.6%), shrapnel injuries (4/34 or
11.8%), and blunt force trauma with one or more gunshot
wounds (1/34 or 2.9%). In the non-well series the most frequent
type of trauma causing death was multiple gunshot wounds
(698/1111 or 62.8%), followed by shrapnel injuries (212/1111,
or 19.1%), a single gunshot wound (189/1111 or 17.0%), and
blunt force trauma with one or more gunshot wounds (12/1111
or 1.1%).
In individuals in whom the cause of death was a single
gunshot wound there is a clear preference for the location of the
entrance wound in the well series. This location is the back of
the head defined as the occipital bone, and the posterior parts of
the parietal and temporal bones (posterior to the external
auditorymeatus).Thefrequencyofentrancegunshotwoundsto
the back of the head in the well series is 15/34 or 44.1%. In the
non-wellseriesentrancegunshotwoundsintheposteriorpartof
the skull were considerably less frequent (168/1111 or 15.1%).
This difference is significant (x2= 18.55, P = 0.00001).
Basic biological data—sex, age-at-death, and stature
contributed to positive identification in all cases in both the
well, and non-well series. Other factors contributing to
identification in wells and non-well contexts are presented in
Table 3. No significant differences were noted in the two series.
Of the biological criteria, dental remains contributed the most
inbothseries.Intheabsenceofhardantemortemdatathedental
criteria used by the identification team included the presence of
amalgam or other types of fillings, antemortem tooth loss, and
the presence of fixed or mobile prosthetic devices—corrobo-
rated by family members or photographs. Of the non-biological
criteria, clothing and personal artifacts (including items such as
folding knives, unusual belt buckets, toys, rosary beads, etc.)
contributed the most in both series.
Primary factors responsible for identification in the two
series are presented in Table 4. Once again, no significant
differences were noted in the two series. In slightly less than
half of the cases (18/37 or 48.6% in the well series, and 526/
1256 or 41.9% in the non-well series) a single primary factor
responsible for positive identification could be determined.
Forensic DNA testing followed by the presence of personal
documents established identification most frequently in both
series. Unique biological features (DNA, dental remains,
skeletal evidence of antemortem fractures or surgery, and
tattoosinsaponifiedremains) establishedpositiveidentification
in 35.1% (13/37) of individuals recovered from wells, and in
26.7% (336/1256) of individuals recovered from non-well
contexts.
4. Discussion
Identification of victims of mass disasters is carried out by
different means. The most simple are visual recognition and
fingerprint analysis. In poorly preserved remains, or in the
absence of useable dermatogliphics, the identification team
tries to narrow the range of possible matches using basic data
such as age-at-death, sex, and stature, and then to establish or
reject positive identification based on forensic DNA testing
[5,28], dental records [3,4,29,30], or comparisons of ante-
mortem and postmortem X-rays [31–34].
The identification process in Croatia was complicated by
two factors: the large number of missing individuals, and the
absence of reliable antemortem data. Because of the
destruction of hospitals and regional medical centers ante-
mortem hospital and dental records of local inhabitants were
unavailable for comparison [1–3]. Similarly, because the
Croatian Army was formed in 1991, and consisted of
volunteers from different parts of Croatia as well as from
other countries, limited antemortem medical and dental data
were available for military personnel. For instance, no
antemortemmedicalordentalX-raysorrecordswereavailable
for comparison for 46 individuals exhumed from four mass
gravesinthevicinityofPetrinjain1995[3].Foralargersample
of1000individualsexhumedfrommassgravesinCroatiafrom
1995 to July 1998, no antemortem medical records or X-rays
Table 3
Factors contributing to identification in wells and non-well settings
Wells Non-well settings
Cont.a
Nb
%Cont.N%
Facial or body hair
Dental remains
Antemortem surgery
Antemortem fractures
Tattoos/scars
Clothing
Personal documents
Personal jewelry
Personal artifacts
18
30
1
2
2
35
7
3
17
37
37
37
37
37
37
37
37
37
48.6
81.1
2.7
5.4
5.4
94.6
18.9
8.1
45.9
653
1120
1256
1256
1256
1256
1256
1256
1256
1256
1256
52.0
89.2
3.5
8.4
3.8
85.5
15.3
11.6
44.0
44
105
48
1074
192
146
553
aNumber of cases in which feature contributed to identification.
bNumber of individuals identified.
Table 4
Primary factors responsible for positive identification in wells and non-well
settings
Wells Non-well settings
Cont.a
Nb
% Cont.N%
Forensic DNA testing
Dental remains
Antemortem surgery
Antemortem fractures
Tattoos/scars
Clothing
Personal documents
Personal jewelry
Combination of
biological and
non-biological features
7
2
1
2
1
0
5
0
37
37
37
37
37
37
37
37
37
18.9
5.4
2.7
5.4
2.7
0.0
13.5
0.0
51.4
211
75
13
31
1256
1256
1256
1256
1256
1256
1256
1256
1256
16.8
6.0
1.0
2.5
0.5
0.5
13.9
0.7
58.1
6
6
175
9
19730
aNumber of cases in which feature contributed to identification.
bNumber of individuals identified.
M. Sˇlaus et al./Forensic Science International 171 (2007) 37–4340

Page 5

wereavailable,whiledentalchartsorX-rayswereavailablefor
only 125 individuals [4].
To help identify missing individuals the Committee for
Imprisoned and Missing Individuals asked family members to
fill out a detailed questionnaire that listed biological data such
as year of birth, sex,stature, weight, length of hair and presence
of mustaches or beards, other distinguishing characteristics,
antemortem fractures, disease, or surgical procedures, and
dental characteristics including antemortem tooth loss or the
presenceoffixedorremovableprostheticappliances.Datawere
alsocollectedonthedatethemissingindividualswerelastseen,
their location at the time, the clothing and shoes they were
wearing, and personal jewelry or items that may have been in
their possession. Identification was carried out on the basis of
dataavailable fromthesequestionnaires,supplementedinsome
cases by additional data supplied by military surgeons,
individuals forced to bury bodies, witnesses of executions,
etc. When the identification team believed that probable
identification was indicated, family members were brought to
the Department of Forensic medicine in Zagreb to discuss the
findings. Family members, relatives and friends attending these
conferences were informed of the biological criteria that
indicated possible identification. They were then shown
photographs of clothing and personal items and asked if they
could recognize them. Positive identification was established
through this collaborative effort. If family members were not
satisfiedwith the biological data and could not identify clothing
or personal items, forensic DNA testing was attempted.
From 1996 to the present, the remains of 61 individuals
killed during the 1991 War in Croatia were recovered from 13
wells.Theageandsexdistributioninthewellseriesissimilarto
that of a much larger sample consisting of 1256 identified
individuals recovered from individual and mass graves in
continental Croatia. Important differences are, however,
present in the taphonomic features of the recovered remains,
and the types and locations of perimortem fractures in the
remains.
Bodies recovered from wells were less well preserved than
those recovered from non-well settings. Significant differences
between the two series were noted in the frequencies of well-
preserved bodies and the commingling of remains. However,
while these features significantly affected identification of
remains from non-well settings, they were not correlated with
the identification of remains from wells. A factor that strongly
influenced identification of remains from wells was the
presence or absence of water in the well. Data collected from
both series explains the reason for this.
Sex, age-at-death and stature contributed to positive
identification in all cases in both the well and non-well series.
No differences in sex and age-at-death accuracies were noted
between the series. Stature estimation was, however, proble-
matic in both. It is recognized that regression formulae for
estimatingstaturearepopulation-specific.RossandKonigsberg
[35] showed that the regression formulae for American Whites
developed by Trotter and Gleser [36], systematically under-
estimate the stature of Europeans from Croatia and Bosnia and
Herzegovina. Unfortunately, Ross and Konigsberg only
provided formulae for estimating male stature. A second more
serious factor complicating stature estimation was the accuracy
of antemortem statures reported in the questionnaires. Bias in
self-reporting of stature has previously been documented [37–
39], as have problems in the perception of stature by others
[40]. Komar reported [41] that less than a third (29.4%) of the
positively identified individuals recovered from mass graves in
theSrebrenicaregioninBosniaandHerzegovinahadareported
height that fell within the estimated range provided in the
postmortem report. The international identification teams
identifying these remains also used antemortem information
obtained through interviews with relatives Because of these
reasons stature estimates in both the well and non-well series
were given with a 10 cm margin of error.
Both series also have similar frequencies and identical
rankings of other factors that contributed to identification. For
instance, slightly less than 19% of the bodies recovered from
wells were found with personal documents, 8.1% had personal
jewelry, and 45.9% had personal artifacts. Similar frequencies
(15.3%, 11.6%, and 44.0%, respectively) were recorded in the
non-wellseriessuggestingthattherewasnoefforttodisguiseor
hide the identities of the individuals whose bodies were
deposited in wells. Data collected from both series also show
that, by themselves, these criteria rarely established positive
identification. The majority of identifications were established
through a combination of biological and non-biological
features (51.4% in the well, and 58.1% in the non-well series).
Typically this would include dental features, elements of
clothing, and personal artifacts. Although these variables
contributed significantly to identification in both series—they
were ranked second, first, and fourth in the well, and first,
second, and fourth in the non-well series, individually they
were not considered specific enough to establish positive
identification. In wells that contained significant amounts of
water, these features were frequently damaged or destroyed.
Analysis of the preservation of these features showed that no
fixed or mobile prosthetic devices were recovered in the dental
remains of 22 individuals recovered from wells with water,
compared to 12/39 (30.8%) fixed or mobile prosthetic devices
recovered from the dental remains of individuals from dry
wells. This difference is statistically significant (x2= 6.59,
P = 0.01). As there are no significant demographic differences
between the two groups, the likely reasons for this discrepancy
are the difficult recovery conditions in wells with water and the
greater likelihood of the prosthetic devices becoming detached
while submerged in water.
Clothing was alsoadversely affected by water. It contributed
to identification in 17/22 (77.3%) individuals from wells with
water, and 38/39 (97.4%) of individuals from dry wells. This
difference is also statistically significant (x2= 4.37, P = 0.036),
as is the difference between the frequencies of recovered
personal artifacts in wells with water (3/22 or 13.6%) and dry
wells (20/39 or 51.3%; x2= 6.95, P = 0.008).
These data suggest that the discrepancy noted in the
identification ratios between well and non-well contexts was
primarily the result of the negative effect that water had on the
preservation of clothing, personal artifacts and some dental
M. Sˇlaus et al./Forensic Science International 171 (2007) 37–43 41

Page 6

features. The identification ratio in dry wells (76.9%) was
almost identical to that in the non-well series (77.4%). The
identification ratio in wells containing water (31.8%) was,
however, sufficiently low to result in a significantly lower
identification ratio for all individuals recovered from wells.
Another significant difference noted between the two series
is the high frequency of entrancegunshot wounds to the back of
the head in the well series. This pattern of trauma is consistent
with an execution style murder and has been noted in males,
females and subadults from the well series.
In conclusion, the collected data show no significant
differences in sex or age-at-death between individuals
recovered from wells and those from a large non-well series.
Individuals recovered from wells were significantly less well
preserved and had a significantly higher frequency of gunshot
wounds to the back of the head when compared to individuals
from the non-well series. The significantly lower identification
ratio in the well series is most likely the result of difficult
recovery conditions in wells with significant amounts of water,
and the negative effect that water had on the preservation of
clothing, personal artifacts and some dental features.
Acknowledgments
Theauthors wouldliketothankColonelIvanGrujic ´,head of
the Committee for Imprisoned and Missing Individuals for his
continued support and numerous insights. This study was
financially supported by the Ministry of Science and
Technology of the Republic of Croatia (Grant no. 0108249).
References
[1] D.W. Owsley, D. Strinovic ´, M. Sˇlaus, D.D. Kollmann, M.L. Richardson,
Recovery and identification of civilianvictims of war in Croatia, CRM 19
(1996) 33–36.
[2] D. Strinovic ´, J. Sˇkavic ´, I. Kostovic ´, N. Henigsberg, M. Judas, D. Clark,
Identification of war victims in Croatia, Med. Sci. Law 34 (1994) 207–
212.
[3] H. Brkic ´, D. Strinovic ´, M. Sˇlaus, J. Sˇkavic ´, D. Zec ˇevic ´, M. Milic ˇevic ´,
Dental identification of war victims from Petrinja in Croatia, Int. J. Legal.
Med. 110 (1997) 47–51.
[4] H. Brkic ´, D. Strinovic ´, M. Kubat, V. Petrovec ˇki, Odontological identifica-
tionofhumanremainsfrommassgravesinCroatia,Int.J.Legal.Med.114
(2000) 19–22.
[5] D. Primorac, S. Anpelinovic ´, M. Definis-Gojanovic ´, I. Drmic ´, B. Rezic ´,
M.M. Baden, et al., Identification of war victims from mass graves in
Croatia, Bosnia and Herzegovina by the use of standard forensic methods
and DSNA typing, J. Forensic Sci. 41 (1996) 891–894.
[6] J. Rainio, M. Hedman, K. Karkola, K. Lalu, P. Peltola, H. Ranta, et al.,
Forensic osteological investigations in Kosovo, Forensic Sci. Int. 121
(2001) 166–173.
[7] M.P. Djuric, Anthropological data in individualization of skeletal remains
from a forensic context in Kosovo—a case history, J. Forensic Sci. 49
(2004) 464–468.
[8] W.M. Bass, Human Osteology. A Laboratory and Field Manual of the
Human Skeleton, 3rd ed., Missouri Archaeological Society, Columbia,
1987.
[9] W.M. Krogman, M.Y. I˙s ¸can, The Human Skeleton in Forensic Medicine,
Charles C. Thomas, Springfield, IL, 1986.
[10] M. Sˇlaus, D. Strinovic ´, J. Sˇkavic ´, V. Petrovec ˇki, Discriminant function
sexing of fragmentary and complete femora: standards for contemporary
Croatia, J. Forensic Sci. 48 (2003) 509–512.
[11] R.S. Meindl, C.O. Lovejoy, Ectocranial suture closure: a revised method
for the determination of skeletal age at death based on the lateral-anterior
sutures, Am. J. Phys. Anthropol. 68 (1985) 57–66.
[12] S. Brooks, J.M. Suchey, Skeletal age determination based on the os pubis:
a comparison of the Acsa ´di-Nemeske ´ri and Suchey-Brooks methods,
Hum. Evol. 5 (1990) 227–238.
[13] B.M. Gilbert, T.W. McKern, A method for aging the female os pubis, Am.
J. Phys. Anthropol. 38 (1973) 31–38.
[14] T.W. McKern, T.D. Stewart, Skeletal Age Changes in Young American
Males. Analyzed from the Standpoint of Age Identification. Technical
Report EP-45, Quarter-Master Research and Development Center, Envir-
onmental Protection Research Division, Natick, MA, 1957.
[15] T.W.Todd,Agechangesinthepubicbone.I.Thewhitemalepubis,Am.J.
Phys. Anthropol. 3 (1920) 285–334.
[16] T.W. Todd, Age changes in the pubic bone. III. The pubis of the white
female. IV. The pubis of the female White–Negro hybrid, Am. J. Phys.
Anthropol. 4 (1921) 1–70.
[17] C.O. Lovejoy, R.S. Meindl, T.R. Pryzbeck, R.P. Mensforth, Chronolo-
gical metamorphosis of the auricular surface of the ilium: a new method
for the determination of age at death, Am. J. Phys. Anthropol. 68 (1985)
15–28.
[18] M.Y. I˙s ¸can, S.R. Loth, R.K. Wright, Age estimation from the rib by phase
analysis: White males, J. Forensic Sci. 29 (1984) 1094–1104.
[19] M.Y. I˙s ¸can, S.R. Loth, R.K. Wright, Age estimation from the rib by phase
analysis: White females, J. Forensic Sci. 30 (1985) 853–863.
[20] H. Lamendin, E. Baccino, J.F. Humbert, J.C. Tavernier, R.M. Nossintch-
ouk, A. Zerilli, A simple technique for age estimation in adult corpses: the
two criteria dental method, J. Forensic Sci. 37 (1992) 1373–1379.
[21] I.G. Fazekas, F. Ko ´sa, Forensic Fetal Osteology,
Budapest, 1978.
[22] C.F.A. Moorrees, E.A. Fanning, E.E. Hunt, Age variation of formation
stages for ten permanent teeth, J. Dent. Res. 42 (1963) 1490–1502.
[23] V.J.M. Di Maio, Gunshot Wounds. Practical Aspects of Firearms, Ballis-
tics, and Forensic Techniques,
London, Tokyo, 1993.
[24] O.C. Smith, H.E. Berryman, C.H. Lahren, Cranial fracture patterns and
estimate of direction from low velocity gunshot wounds, J. Forensic Sci.
32 (1987) 1416–1421.
[25] O.C. Smith, S.A. Symes, H.E. Berryman,M.M. Le Vaughn, Characteristic
features of entrancewounds from hollow-point bullets, J. Forensic Sci. 38
(1993) 323–326.
[26] G.Quatrehomme,M.Y.I˙s ¸can,Gunshotwoundstotheskull:comparisonof
entries and exits, Forensic Sci. Int. 94 (1998) 141–146.
[27] G. Quatrehomme, M.Y. I˙s ¸can, Characteristics of gunshot wounds in the
skull, J. Forensic Sci. 44 (1999) 568–576.
[28] M.M. Holland, L.D. Fisher, G.L. Mitchell, C.W. Rodriguez, J.J. Canik,
R.C. Merril, Mitochondrial DNA sequence analysis of human skeletal
remains: identification of remains from the Vietnam war, J. Forensic Sci.
38 (1993) 542–553.
[29] I.M. Sopher, Forensic Dentistry, Charles C. Thomas, Springfield, IL, 1976.
[30] J.J. Salley, F.J. Filipowitz, H.H. Karnitschnig, Dental identification of
mass disaster victims, J. Am. Dent. Assn. 66 (1963) 827–832.
[31] W.L.Culbert,F.M.Law,Identificationbycomparisonwithroentenograms
of nasal accessory sinuses and mastoid processes, J. Am. Med. Assn. 88
(1927) 1634–1636.
[32] H. Kade, H. Meyers, J.E. Wahlke, Identification of skeletonized remains
by X-ray comparison, J. Crim. Law Crim. Police Sci. 58 (1967) 261–
264.
[33] L. Atkins, M.S. Potsaid, Roentgenographic identification of human
remains, J. Am. Med. Assn. 240 (1978) 2307–2308.
[34] D.W. Owsley, D.H. Ubelaker, M.M. Houck, K.L. Sandness, W.E. Grant,
E.A. Craig, et al., The role of forensic anthropology in the recovery and
analysisofBranchDavidian compound victims:Techniquesofanalysis,J.
Forensic Sci. 40 (1995) 341–348.
[35] A.H. Ross, L.W. Konigsberg, New formulae for estimating stature in the
Balkans, J. Forensic Sci. 47 (2002) 165–167.
[36] M. Trotter, G.C. Gleser, Estimation of stature from long bones of Amer-
ican Whites and Negroes, Am. J. Phys. Anthropol. 10 (1952) 463–514.
Akade ´miai Kiado ´,
CRC Press, Boca Raton, Ann Arbor/
M. Sˇlaus et al./Forensic Science International 171 (2007) 37–43 42

Page 7

[37] E. Giles, D.L. Hutchinson, Stature and age-related bias in self-reported
stature, J. Forensic Sci. 36 (1991) 765–780.
[38] P.Willey,T.Falsetti,Inaccuracyofheightinformationondriver’slicenses,
J. Forensic Sci. 36 (1991) 813–819.
[39] C.C. Snow, J. Williams, Variation in premortem statural measurements
compared to statural estimates of skeletal remains, J. Forensic Sci. 16
(1971) 455–464.
[40] J.P. Baraybar, E.H. Kimmerle, Thousands dead: the use of stature in
individual identification, in: Proceedings of the 54th Annual Meeting of
the American Academy of Forensic Sciences, Atlanta, G, 230-1, American
Academy of Forensic Sciences, Colorado Springs, February 11–16, 2002.
[41] D. Komar, Lessons from Srebrenica: the contributions and limitations of
physical anthropology in identifying victims of war crimes, J. Forensic
Sci. 48 (2003) 713–716.
M. Sˇlaus et al./Forensic Science International 171 (2007) 37–43 43