Journal of Archaeological Science (1998) 25, 861–865
Article No. as970231
Determining the Sex of Infanticide Victims from the Late
Roman Era through Ancient DNA Analysis
Marina Faerman and Gila Kahila Bar-Gal
Dental Division of Anatomy & Cell Biology, Hebrew University—Hadassah Faculty of Dental Medicine,
Jerusalem 91010, Israel
Department of Hematology, Hebrew University—Hadassah Medical School, Jerusalem 91010, Israel
Charles L. Greenblatt
Department of Parasitology, Hebrew University—Hadassah Medical School, Jerusalem 91010, Israel
The Semitic Museum, Harvard University, Cambridge, MA 02138, U.S.A.
Department of Hematology, Hebrew University—Hadassah Medical School, Jerusalem 91010, Israel
Dental Division of Anatomy & Cell Biology, Hebrew University—Hadassah Faculty of Dental Medicine,
Jerusalem 91010, Israel
(Received 24 March 1997, revised manuscript accepted 18 November 1997)
Infanticide has since time immemorial been an accepted practice for disposing of unwanted infants. Archaeological
evidence for infanticide was obtained in Ashkelon, where skeletal remains of some 100 neonates were discovered in a
sewer, beneath a Roman bathhouse, which might have also served as a brothel. Written sources indicate that in ancient
Roman society infanticide, especially of females, was commonly practised, but that females were occasionally saved and
reared as courtesans. We performed DNA-based sex identiﬁcation of the infant remains. Out of 43 left femurs tested
19 specimens provided results: 14 were found to be males and 5 females. The high frequency of males suggests selective
preservation of females and that the infants may have been oﬀspring of courtesans, serving in the bathhouse, supporting
its use as a brothel. 1998 Academic Press
Keywords: ANCIENT DNA, SEX DETERMINATION, INFANTICIDE, LATE ROMAN PERIOD,
Today archaeologists are paying increasing
attention to examining social structure within
past societies. While gender diﬀerences have
been traditionally explored through identiﬁcation of
grave goods considered indicative of female or
male roles, physical anthropology enables archae-
ologists to develop the study of mortuary practices
through identiﬁcation of gender in relation to
burial type, even when grave goods are absent.
However, the problem of sex determination remains
0305–4403/98/090861+05 $30.00/0 1998 Academic Press
in dealing with fragmentary and/or infant burials. The
reliability of morphometric analyses for gender identi-
ﬁcation in infants is low, especially in the case of
Infant burials from the Neolithic to recent periods
frequently occur in Israel in diﬀerent archaeological
contexts from those of older children or adults. These
infant remains may have been treated with great care,
as for example, the jar burials with grave goods found
at Middle Bronze Age Kabri (Kempinski & Niemeier,
1992), or alternatively treated with complete disregard
like the infants thrown into rubbish pits at Chalcolithic
Shiqmim (Levy et al., 1991), or into sewers in Late
Roman Ashkelon (Stager, 1991;Smith & Kahila,
1992). Knowing the gender of infants found in diﬀerent
archaeological contexts has implications not only for
the type of burial accorded, but also the possible role
of gender in relation to the question of infant sacriﬁces
Human settlement of Ashkelon dates back over 5000
years, and during most of this period Ashkelon was a
major seaport (Stager, 1993). It served the Canaanites
from c. 2000–1200 , and was one of ﬁve main
centres of the Philistines until 604 . Under Persian
hegemony Phoenicians from Tyre colonized the sea-
port from 525–300 . They, in turn, were successively
replaced by the Jews, the Greeks and ﬁnally the
Romans in the 1st century .
Skeletal remains of more than 100 neonates were
found during archaeological excavations by the Leon
Levy Expedition to Ashkelon (Stager, 1991;Smith &
Kahila, 1992;Faerman et al., 1997). The infant re-
mains were found in the sewer beneath a bathhouse,
built in the 4th century and used until the 6th century.
The infant bones had been discarded in the gutter of
the sewer along with animal bones, potsherds and
isolated coins; no signs of careful burial or associated
grave goods were observed. The casual method of
disposal contrasts sharply with the careful infant jar
burial from the same period discovered some 200 yards
away. Bone size, dental development and lack of
neonatal lines in the teeth indicated that they were all
neonates, 1–2 days old. The combination of early death
of so many infants and their mode of disposal implied
infanticide, rather than death from natural causes
(Smith & Kahila, 1992). None of the infants showed
evidence of disease or skeletal malformation, indicat-
ing that other factors, such as their gender, may have
been the motive for infanticide.
New developments in molecular biology, and
especially in analysing DNA recovered from ancient
bones, have provided reliable methods for gender
determination based on ampliﬁcation of DNA
sequences speciﬁc to the X and/or Y chromosomes
(Gill et al., 1994;Faerman et al., 1995;Lassen,
Hummel & Herrmann, 1996;Stone et al., 1996).
This paper brings together material previously
discussed by Smith & Kahila (1992) and Faerman et al.
Materials and Methods
DNA was isolated from the bone powder, obtained
from left femurs only, to avoid testing the same indi-
vidual twice (Figure 1). Altogether 43 left femurs were
available for the analysis (29 complete and 14 fragmen-
tary). Bones were cleaned with a soft brush. The
surface layer was removed by electric drill (large bit),
and bone powder was obtained by drilling in a freshly
uncovered surface with a sterile small burr. Approxi-
mately 0·5–1·0 mg of bone powder was used for each
DNA extraction. DNA from each specimen was
extracted twice following the chelex puriﬁcation pro-
cedure (Woodward et al., 1994). A third extraction was
performed and analysed at least 6 months later using a
silica-based puriﬁcation method (Hoss & Paabo, 1993).
Ancient DNA studies are prone to numerous arte-
facts (Paabo, Giﬀord & Wilson, 1988;Hagelberg &
Clegg, 1991). To eliminate contamination by DNA of
exogenous sources, stringent precautions were included
at every step. Disposable sterile tubes, ﬁltered tips and
aliquoted sterile reagents and solutions, kept only for
ancient DNA work, were used throughout. DNA
extraction and polymerase chain reaction (PCR) were
performed in diﬀerent hoods, sterilized by UV light,
and located in two diﬀerent rooms. Diﬀerent sets of
pipettes were used for DNA extraction, PCR ampliﬁ-
cation and analyses of the PCR products. Blank
extraction controls, containing no bone material, were
run in parallel with each set of experiments.
We have applied a highly sensitive method based on
PCR ampliﬁcation of the X and Y amelogenin alleles
(Faerman et al., 1995). The reaction yields distinguish-
able X- and Y-chromosome products by the simul-
taneous use of three primers (Figure 2(a)). We
subjected 3, 7 and 11 ìl of each DNA extract to PCR
ampliﬁcation along with a blank extraction control
and no DNA PCR control to monitor contamination
during the DNA extraction and PCR ampliﬁcation.
Conditions for the PCR and the three primers (M4,
M5 and M6) have been described previously (Faerman
et al., 1995). Primer M7 (5-GTGACTATCTTAGA
ATCAGGAG-3), designed during this study (see
Figure 1. Left femurs of Ashkelon infants. Scale is 1 cm.
862 M. Faerman et al.
Results), was used in part of the experiments instead of
primer M5. We analysed 18 ìl aliquots on 2% Nusieve
agarose gel, stained with ethidium bromide. To verify
the authenticity of the X and Y amelogenin alleles, the
respective bands were sequenced. For this purpose 5 ìl
of the PCR products along with the appropriate
controls were subjected to additional 25 cycles, and the
re-PCR products were puriﬁed by electrophoresis on
1% low melting agarose gel (Filon et al., 1995).
Sequence analysis was performed using the allele-
speciﬁc primers with Sequenase Version 2.0 (USB).
Ampliﬁcation was successful for 19 out of the 43
ancient specimens tested. Fourteen specimens were
found to be males and ﬁve females, giving a signiﬁ-
cantly higher frequency of boys than girls (P<0·05).
The results for three specimens are shown in Figure
2(b). The success rates of the PCRs are given in Table
1. In total, data were obtained for 70 of 189 PCRs of
the 19 specimens. There were no inconsistencies or
conﬂicting data for any of the specimens. Furthermore,
the results for all the specimens, except specimen
No. 107, were reproduced on at least two separate
DNA extracts. The authenticity of the ampliﬁed
fragments was veriﬁed by direct sequencing of the
respective bands of male and female samples (not
Successful ampliﬁcation was obtained for 44% of
the specimens examined, despite the antiquity and
friable condition of the bones, and these included a
signiﬁcantly larger number of males than females.
In our experiments with contemporary DNA we
have noticed preferential ampliﬁcation of the Y allele
when less than 25 pg DNA was used per reaction
(Faerman et al., 1995). We considered the possibility
that for some reasons (diﬀerence in length of the
X- and Y-speciﬁc PCR products, or nature of the
primers) our test may miss females in highly degraded
DNA specimens. We therefore designed a new
X-speciﬁc primer, which together with the 5common
primer spans a smaller fragment (270 bp) than that
previously used (see Figure 2(a)). For the 24 bone
specimens which had not yielded ampliﬁable DNA, all
tests were repeated in full, including DNA extraction
and PCR ampliﬁcation, with the new set of primers.
No ampliﬁcation products were obtained in three
PCRs for each of two DNA extractions of each bone
specimen, a total of 144 PCRs.
The initial reasons for deciding that the individuals
analysed here were victims of infanticide were the lack
of infants aged more than 2 days and their casual
disposal in the sewer (Smith & Kahila, 1992). If the
Ashkelon sewer served as a public place for disposal of
infants who had died naturally, but were considered
too unimportant or too young for full burial rites, then
one would expect to ﬁnd infants of up to at least 3
months of age. Full burial rites were apparently rarely
carried out for infants of less than 6 months of age.
Similar considerations, namely age distribution and
location, were adduced by Mays (1993) as proof of
infanticide in Roman Britain.
M12 3456 7 M89101112131415 16
Figure 2. (a) Part of the amelogenin encoding gene showing the
location of the PCR primers. Fragment deleted in the Y chromosome
is indicated. (b) Sex identiﬁcation of Ashkelon infants: M—size
marker (1 kb DNA ladder); lanes 1, 8, ‘‘no DNA’’ PCR control; lane
2, ‘‘blank extraction’’ control; lanes 3–5, specimen No. 100; lanes
9–11, specimen No. 94; lanes 12–14, specimen No. 96; positive
controls: lanes 6, 15, modern female DNA; lanes 7, 16, modern male
Table 1. Data on DNA-based sex identiﬁcation of the Ashkelon infants
ResultsExtract 1 Extract 2 Extract 3
90 1 0 1 Male
91 1 0 1 Male
92 3 2 2 Male
93 2 2 0** Male
94 2 2 2 Male
95 3 2 1 Male
96 2 3 1 Male
97 2 1 0** Male
98 2 3 0** Male
100 2 1 1 Female
104 1 1 0** Male
107 2 0 0** Female
119 2 0 2 Female
120 1 0 1 Female
122 1 2 2 Male
123 1 1 2 Male
216 1 1 0** Female
221 1 1 0 Male
234 1 1 0 Male
*3 PCRs were performed for each DNA extract.
**6 PCRs were performed for each of these DNA extracts.
Determining the Sex of Infanticide Victims 863
Classical authors provide ample and credible
evidence for infanticide in Graeco-Roman society
(cited in Brunt, 1971;Eyben, 1980–81;Pomeroy, 1984;
Wiedemann, 1989;Harris, 1994). Recent archaeologi-
cal discoveries attest to infanticide from one end of the
Roman empire to the other, from Ashkelon in Roman
Palestine (Smith & Kahila, 1992) to sites in Roman
Britain (Mays, 1993).
Once accepted as a parental prerogative, oﬃcial
attitudes towards infanticide have changed over time.
Currently illegal in most societies, the practice of
infanticide is still widespread for a variety of cultural
and economic reasons (Langer, 1974;Williamson,
1978;Tooley, 1983). The justiﬁcation for infanticide
rests in part on the assumption that new born infants
are not fully human, and in part on the importance of
controlling the size and the structure of the family and
society at large (Tooley, 1983).
Infanticide was often preferable to abortion as a
method of birth control and family planning: it allowed
for sex selection and birth order to be taken into
account, and it was less dangerous to the physical
well-being of the mother (Eng & Smith, 1976;Stager &
Wolﬀ, 1984). In Roman society according to Cicero,
malformed infants had to be destroyed, but many
healthy infants were also killed. The emperors
Augustus and Claudius issued edicts ordering the death
of infants born to members of their family accused of
The gender of a child was often an important factor
in deciding its fate. Most parents raised at least one
boy as an heir or support in old age. In contrast girls,
especially in patriarchal societies, were viewed as
burdens, especially if their marriage was dependent
on a dowry. In Roman society it was the father’s
decision alone that determined whether a new born
baby should be permitted to live. Under some circum-
stances girls who could be raised as performers or
prostitutes were preferentially kept (Fantham et al.,
The general consensus is that in both ancient and
modern societies more daughters than sons were se-
lected for infanticide (Pomeroy, 1983: 208). The most
vivid and explicit reference is from a letter, dated June
17, 1 , written by a certain Hilarion in Alexandria to
his expectant wife Alis in Oxyrhynchus. He writes ‘‘I
ask and beg you to take good care of our baby son, and
as soon as I receive payment I will send it up to you. If
you are delivered of child [before I get home], if it is a
boy keep it, if a girl discard it’’ (Papyrus Oxyrhynchus
744, translated in Lewis, 1985: 54). Thus, it comes as
something of a surprise to ﬁnd so many boy infants
discarded in the sewer of Late Roman Ashkelon. Males
were present at signiﬁcantly higher frequencies in our
subsample of 19, and there is only a very low prob-
ability (<0·001) that they were present in as few as 40%
of the entire sample (conﬁdence levels for small
samples quoted in Simpson, Roe & Lewontin, 1960:
Bathhouses, both public and private, proliferated
throughout the Roman empire. Until the time of the
emperor Claudius, men and women bathed separately.
During his reign a single set of bathrooms replaced
the double set as mixed bathing came in vogue, over
the objections of some Romans, who were concerned
that bathhouses were becoming bordellos (Ward,
Ovid, writing in the time of the emperor Augustus,
oﬀers this advice to a young woman on how to elude
the guardian and have sex in the baths: ‘‘While the
guardian keeps the girl’s clothes without, the numerous
baths hide furtive [i.e. sexual] sport’’ (Ars amatoria
3.939–40). Another author, writing in the time of Nero,
describes a father who went to the baths, leaving one
child at home, only to return from the baths a prospec-
tive father of two more (Ward, 1992: 134, citing
Nicharchus Anthologia Graeca 11.243). In his epi-
grammatic depiction of Roman life in the 1st century
, the poet Martial wrote: ‘‘The bathman lets you in
among the tomb-haunting whores only after putting
out his lantern’’ (Epigrams 3.93, translated by D. R.
Shackleton Bailey in Loeb Classical Library, vol. 1,
At the time when the Ashkelon bathhouse was in
use, the last great Roman historian Ammianus
Marcellinus (c. 330–395) records that when nobles,
‘‘each attended by ﬁfty servants, have entered the
vaulted rooms of a bath, they shout in threatening
tones: ‘Where on earth are our attendants?’ If they
have learned that an unknown courtesan has suddenly
appeared, some woman who has been a common
prostitute of the crowd of the city, some old strumpet,
they all strive to be ﬁrst to reach her . .. and extol her
with such disgraceful ﬂattery as the Parthians do
Samiramis, the Egyptians their Cleopatras . . .’’
(Ammianus Marcellinus 28.4.9, translated by J. C.
Role in Loeb Classical Library, vol. 3, pp. 141–143).
There were both public and private bathhouses in
Roman Ashkelon. The small bathhouse, where infants
were discarded in the sewer beneath, was probably one
of many private baths run for proﬁt in this seaport.
The proprietor welcomed sailors, merchants and any-
one else into the bathhouse with this enticing signpost:
‘‘Enter, enjoy, and . . .’’.
The bathhouse was built over earlier Roman villas,
including one with a room full of lamps decorated with
erotic images. The bathhouse was situated in what was
probably a well established part of the ‘‘red-light’’
district of Roman Ashkelon. The linkage of baths with
prostitution has been alluded to by classical authors
(cited in Dauphin, 1996) and reinforced by the archi-
tectural and epigraphic remains from Ashkelon. The
presence of both male and female victims in the gutter
beneath the bathhouse raises the intriguing possibility
that these infants may have been the unwanted oﬀ-
spring of courtesans serving in the bathhouse, thus
providing further supportive evidence for its use as a
864 M. Faerman et al.
At the same time this explanation may account for
the predominance of male infants discarded (assuming
that the limited subsample in which sex could be
determined is representative of the total population
of infanticide victims). Although both sexes were
recruited to work as prostitutes in the bisexual world of
the Romans, females were in greater demand. In the
Roman empire one of the primary sources of prostitu-
tion was abandoned children who had been rescued
and reared to work as prostitutes at an early age
(Rousselle, 1996: 299). We can imagine that the courte-
sans of Ashkelon selectively kept and reared some of
their illegitimate oﬀspring (mostly females) in the
profession and discarded others.
This research was supported by the Israel Science
Foundation, administered by the Israel Academy of
Sciences and Humanities, and by the National Center
for Cooperation between Science and Archaeology.
The archaeological excavations were sponsored by
the Leon Levy Expedition to Ashkelon (Israel) and by
the Semitic Museum, Harvard University. We also
acknowledge the comments of two anonymous
Brunt, P. A. (1971). Italian Manpower, 225 B.C.–A.D. 14. Oxford:
Dauphin, C. (1996). Brothels, baths and babes—prostitution in the
Byzantine Holy Land. Classics Ireland 3, 47–72.
Eng, R. W. & Smith, T. C. (1976). Peasant families and population
control in eighteenth-century Japan. Journal of Interdisciplinary
History 6, 417–445.
Eyben, E. (1980–81). Family planning in Graeco-Roman Antiquity.
Ancient Society 11–12, 5–82.
Faerman, M., Filon, D., Kahila, G., Greenblatt, C. L., Smith, P. &
Oppenheim, A. (1995). Sex identiﬁcation of archaeological human
remains based on ampliﬁcation of the X and Y amelogenin alleles.
Gene 167, 327–332.
Faerman, M., Kahila, G., Smith, P., Stager, L. E., Greenblatt, C. L.,
Filon, D. & Oppenheim, A. (1997). DNA analysis reveals the sex
of infanticide victims. Nature 385, 212–213.
Fantham, E., Foley, H. P., Kampen, N. B., Pomeroy, S. B. &
Shapiro, H. A. (1994). Women in the Classical World. Oxford:
Oxford University Press.
Filon, D., Faerman, M., Smith, P. & Oppenheim, A. (1995).
Sequence analysis reveals a â-thalassemia mutation in the DNA of
skeletal remains from the archaeological site of Akhziv, Israel.
Nature Genetics 9, 365–368.
Hagelberg, E. & Clegg, J. B. (1991). Isolation and characterization of
DNA from archaeological bone. Proceedings of Royal Society
London B 244, 45–50.
Harris, W. V. (1994). Child exposure in the Roman empire. Journal
of Roman Studies 84, 1–22.
Hoss, M. & Paabo, S. (1993). DNA extraction from Pleistocene
bones by a silica-based puriﬁcation method. Nucleic Acids
Research 21, 3913–3914.
Gill, P., Ivanov, P. L., Kimpton, C., Piercy, R., Benson, N., Tully,
G., Evett, I., Hagelberg, E. & Sullivan, K. (1994). Identiﬁcation of
the remains of the Romanov family by DNA analysis. Nature
Genetics 6, 130–135.
Kempinski, A. & Niemeier, W. D. (1992). Excavations at
Kabri—Preliminary Report of 1991 of Tel Kabri Expedition.Tel
Aviv University and Israel Exploration Society.
Langer, W. L. (1974). Infanticide: a historical survey. History of
Childhood Quarterly 1, 353–367.
Lassen, C., Hummel, S. & Herrmann, B. (1996). PCR based sex
identiﬁcation of ancient human bones by ampliﬁcation of X- and
Y-chromosomal sequences: a comparison. Ancient Biomolecules 1,
Levy, T. E., Alon, D., Grigson, C., Holl, A., Goldberg, P., Rowan,
Y. & Smith, P. (1991). Subterranean settlement in the Negev
Desert, ca. 4500–3700 B.C. Research and Exploration 7, 394–413.
Lewis, N. (1985). Life in Egypt under Roman Rule. Oxford:
Marcellinus, Ammianus (1952). Ammianus Marcellinus. In (trans-
lated by J. C. Rolfe) Loeb Classical Library,3. London: W.
Heinemann, pp. 141–143.
Martialis, Marcus Valerus (1993). Epigrams. In (translated by D. R.
Shackleton Bailey) Loeb Classical Library,1. Cambridge: Harvard
University Press, p. 269.
Mays, S. (1993). Infanticide in Roman Britain. Antiquity 67, 883–
Paabo, S., Giﬀord, J. A. & Wilson, A. C. (1988). Mitochondrial
DNA sequences from a 7000-year-old brain. Nucleic Acids
Research 16, 9775–9787.
Pomeroy, S. (1983). Infanticide in Hellenistic Greece. In (A.
Cameron & A. Kuhrt, Eds) Images of Women in Antiquity.
London: Croom Helm, pp. 207–222.
Rousselle, A. (1996). The family under the Roman empire. In (A.
Burguiere, C. Klapisch-Zuber, M. Segalen & F. Zonabend, Eds) A
History of the Family 1, Distant Worlds, Ancient Worlds.
Cambridge, MA: Belknap, Harvard University Press, pp. 270–310.
Simpson, G. G., Roe, A. & Lewontin, R. C. (1960). Quantitative
Zoology. New York: Harcourt, Brace & World, p. 199.
Smith, P. & Kahila, G. (1992). Identiﬁcation of infanticide in
archaeological sites: a case study from the Late Roman–Early
Byzantine periods at Ashkelon, Israel. Journal of Archaeological
Science 19, 667–675.
Stager, L. E. (1991). Ashkelon Discovered. From Canaanites and
Philistines to Romans and Moslems. Washington, DC: Biblical
Stager, L. E. (1993). Ashkelon. In (E. Stern, Ed.) The New Encyclo-
pedia of Archaeological Excavations in the Holy Land 1. Jerusalem:
Israel Exploration Society, pp. 103–112.
Stager, L. E. & Wolﬀ, S. R. (1984). Child sacriﬁce at
Carthage—religious rite or population control? Biblical Archaeol-
ogy Review 10, 31–51.
Stone, A. C., Milner, G. R., Paabo, S. & Stoneking, M. (1996). Sex
determination in ancient human skeletons using DNA. American
Journal of Physical Anthropology 99, 231–238.
Tooley, M. (1983). Abortion and Infanticide. Oxford: Clarendon
Press, pp. 325–407, 441.
Ward, R. B. (1992). Women in Roman baths. Harvard Theological
Review 85, 125–147.
Wiedemann, T. (1989). Adults and Children in the Roman Empire.
Williamson, L. (1978). Infanticide: an anthropological analysis. In
(M. Kohl, Ed.) Infanticide and Value of Life.Buﬀalo, NY:
Prometheus, pp. 61–75.
Woodward, S. R., King, M. J., Chiu, N. M., Kuchar, M. J. &
Griggs, C. W. (1994). Ampliﬁcation of ancient nuclear DNA from
teeth and soft tissues. PCR Methods and Applications 3, 244–247.
Determining the Sex of Infanticide Victims 865