The Accuracy of the Osteological Sexing of Cremated Human Remains: A Test Based on Grave Goods from East Lithuanian Barrows.

The accuracy of the sex determination of cremated human remains is one of the vital parameters for archaeologists and skeletal anthropologists dealing with cremations. Few studies have so far aimed at testing it. In the present paper, the accuracy of the sexing techniques was assessed by paralleling biological sex (identified morphologically) to gender (presumed on the basis of the grave goods which accompanied the deceased). A collection of cremated bones from East Lithuanian barrows (c. fourth/fifth--eleventh/twelfth centuries AD) was employed. The fragmentary nature and poor state of the bones generally represents cremains from similar archaeological contexts. The database inevitably underwent several stages of filtration. Out of 364 cremations with a minimum of 445 individuals, only 157 were sexed single adult burials, of which only 81 had "gender-related" grave goods. The relationship of artefact type to gender was defined statistically, revising the results in line with the chronological and typological differences and the probable symbolism of the grave goods. Sex and gender coincided in 56 cases (69.14%), but a considerable distance between the results for both sexes was observed. Biological females displayed a fairly high match level, i.e. 35 out of the 41 (85.37%) individuals osteologically identified as females had been given "feminine" items. The burials of biological males, on the other hand, yielded a surprisingly low match rate, i.e. only 21 out of 40 (52.50%). This disparity suggests a possible misinterpretation of grave goods as gender markers, rather than (only?) erroneous sexing. It is thus argued that for females, the mean value for the accuracy of sexing is 85.5%. In most cases, such precision is entirely satisfactory for the analysis of a poorly-preserved osteological material. For males, however, the accuracy is likely to fall somewhere in the range between 52.5 and 85.5%, with the applied methodology so far failing to contribute to higher precision.