Infanticide in Roman Britain

Article excerpt

Age-distributions of perinatal infants from Romano-British sites and a medieval site are different and may reflect different major causes of death. Whilst the medieval infants probably represent natural deaths, the Romano-British infants, from both cemetery and non-cemetery sites, may mainly represent victims of infanticide.


Infanticide, the killing of unwanted babies, usually at or soon after birth, has been shown to be practised in human societies on every continent and at every level of social complexity from hunter-gatherers to urbanized industrial societies (Williamson 1978; Langer 1974a; 1974b; Scrimshaw 1984). Although infanticide has received much attention from historians and anthropologists (e.g. W.V. Harris 1982; papers in Hausfater & Hrdy 1984), it has been less discussed by archaeologists.

In Roman Britain infanticide has sometimes been suggested as a possible explanation for the finding of infant burials outside recognizable cemetery areas, on villa and settlement sites (discussion in Merrifield 1987: 51; Watts 1989; Collis 1977).

The interpretations of infant burials as evidence for infanticide in Roman Britain simply because they were interments outside recognizable cemetery or 'religious' areas is somewhat speculative, as ethnographic evidence shows that in many societies infants may be accorded different burial treatment from the rest of the population (Ucko 1969; O. Harris 1982; Woodburn 1982). Thus simply that Roman infants were often treated in a different manner in burial does not necessarily suggest that they were victims of infanticide.

A recent paper by Smith & Kahila (1992) studying infant bones from a late Roman-early Byzantine sewer at Ashkelon, Israel suggests infanticide. They found, using a study of long-bone lengths, that the infants were all of approximately the same age at death (around full term); they argued that this pattern is consistent with infanticide, as it is generally carried out immediately after birth. The present paper reports a parallel study from the other end of the Roman Empire, using ages at death of perinatal infants from British sites. The results are not precisely comparable with those from Ashkelon as I was unaware of Smith & Kahila's work when I was preparing this paper, but, as will be shown below, the results are clearly similar.


The regression equations of Scheuer et al. (1980) allow gestational age (from the first day of the last menstrual period -- also known as menstrual age) of perinatal infants to be estimated to within about 2 weeks using long-bone lengths. Their linear regression equations are used here.

The data used in the present work come from late Romano-British sites. Sites were selected on the basis that they had reasonable numbers of perinatal infant burials and that these were aged using the regression equations of Scheuer et al. (1980) in the original report, or data on long-bone lengths were presented in such a way that age estimates could be made by the present writer. Six Romano-British sites were chosen: two cemeteries (Poundbury, Dorset (Molleson n.d.) and Ancaster, Lincolnshire (Cox 1989)); two villas (Winterton (Denston 1976) and Rudston, Humberside (Bayley 1980)); and two settlements (Old Winteringham, Humberside (Denston 1976) and Thistleton, Leicestershire (Powers et al. n.d.)). In order to provide archaeological data to compare with the Romano-British sites, the perinatal infants from the large assemblage of human skeletons excavated from the churchyard at the deserted medieval village of Wharram Percy, North Yorkshire (Mays forthcoming) were also examined. The results are shown in TABLE 1 and FIGURES 1a-c. The average length of gestation, measured from the first day of the last menstrual period is 40 weeks, but there is some variation about this mean -- the limits of normal gestation are about 38-41 weeks (Tanner 1989: 43).

FIGURES 1a & b reveal a striking similarity between the age distributions of infants from TABULAR DATA OMITTED Romano-British cemetery and non-cemetery sites, both distributions showing a strong central tendency at around full term. …