A joint effort: understanding kneecaps past and present

Researchers from the University of Southampton recently spent three days at MOLA recording patellae (kneecaps), as part of their project 'A Joint Effort', which explores knees past and present through 3D imaging. In this blog, Emma van der Velden and Dr. Christian Hoggard explain how the data recorded will help them to generate models to improve sex classification scores in osteology...

In osteoarchaeology, the first stage of any skeletal analysis is to establish a population’s demography and structure, with biological sex and approximate age at death being two of the first things to be identified. Determining the sex of an individual is usually performed on areas of the skeleton most influenced by sexual dimorphism – regions where differences in bone shape tend to develop during puberty – typically the skull and pelvis.

One of the main limiting factors in estimating the sex of an individual is where these areas of the skeleton are either damaged or absent. In these circumstances, how do you establish the biological sex of an individual? The answer might be surprising… but one option is the kneecap (patella). In fact, existing methodologies allow the sex of an individual to be estimated using a small number of basic linear measurements, providing the results are restricted to the population represented by the specific skeletal assemblage under investigation. One example of this method is the measurement of patellar height and width. However, accuracy is currently quite variable, perhaps due to the lack of appropriate training datasets.

Current research at the University of Southampton is focused on improving on this success rate by recording and analysing shape data of three-dimensional patella models, created using structure from motion (SfM), which is the process of estimating a three-dimensional structure from a set of two-dimensional images. Through this analysis these methods can, in comparison to previous studies, capture a greater degree of shape complexity, allow more powerful statistical methods to be implemented and provide better accuracy rates.

We hypothesise that this difference in shape likely has its origins in the morphology of the pelvis. We suggest that the resulting influence on the position and angle of musculature in the thigh creates different stresses and strains on the patella, which in turn results in slight but indicative differences in patella shape. Like any other bone, the patella is not guaranteed to be recovered from an archaeological context. However, where it can be recovered, it is often in reasonable condition and has maintained sufficient shape to enable assessment.

At the heart of any new method there should always be a sample of individuals of known sex, this allows for a training dataset (control) to be produced, against which other patellae can be assessed. We recently spent three days at MOLA recording 58 patellae from the 19th-century Mare Street Baptist Chapel site, which was excavated in 2014. The selected patellae were from skeletons for which we had some biographical data (notably documented gender) as a result of consulting associated coffin plates and/or burial register entries.