1. Introduction

A large number of Greek and Roman marble portraits survive today in the stately homes of England: products of the Grand Tour or later antiquarian collecting activity conducted around the Mediterranean (Michaelis 1882; Scott 2003; Bignamini and Hornsby 2010). Few of these pieces survive in anything like their original condition, having been aggressively cleaned or actively 'restored' in the 17th or 18th century through the application of missing fragments or alternative body parts. Such processes are relatively well understood and were described as long ago as the late 19th century by Adolf Michaelis who noted 'for the most part the newly discovered marbles came out of the earth in a wretched plight, broken, mutilated, corroded, or encrusted with the dirt of centuries' and that, for the purposes of display it was 'always made a great point that they should be properly cleaned – often to the great detriment of their freshness – and vamped up with old, or new, additions.' (Michaelis 1882, 83).

As a result of such widespread restorative practices, it is often difficult to either establish identity for a particular marble portrait or identify which pieces of sculpture originally belonged to which statue. Further complications arise with regard to context, for it is frequently impossible to ascertain the origin of pieces collected during the 18th or 19th centuries, items sometimes having been acquired in circumstances that today would be considered haphazard, criminal, underhand or at best dubious (Bignamini and Hornsby 2010, 17-30). Deprived of a precise name, context or place of origin, many sculptures languish in a form of portrait limbo, their significance and importance lost.

Two such cases of identity loss, from damaged portraits housed in the great antiquarian statue collections of Petworth and Wilton, in the south of England, are described here. The sculptures in question are today on public display, but both have been significantly altered through restoration to the extent that it is difficult to see their true and original selves; respective personalities and distinctive iconographies having been masked through the judicious application of modern marble and plaster. While the existence of both portraits has long been known (Michaelis 1882, 614, 712; Peter Stewart pers. comm. 2009), their precise identity has not, the Petworth statue being simply described as a 'boy in a toga' (Jackson-Stops 1978, 48), the Wilton bust being 'falsely named Constantinus Magnus' (Michaelis 1882, 712).

In order to create a detailed record for the damaged portraits that could be objectively studied, interpreted and cross-compared, both portraits were three-dimensionally (3D) laser scanned by the authors as part of a pilot study established in order to create a digital database of Roman portrait typology. A major advantage of using 3D data is that the portraits are captured without either photographic distortion or the vagaries of light and shade which, in more traditional forms of recording, can sometimes significantly alter the overall appearance of the object. Visualising the 3D model in a uniform greyscale further prevents the eye of the observer being drawn to facial areas that have a high colour contrast (for example where modern repairs have been conducted), allowing a far more objective analysis to be undertaken. A 3D dataset also enables the viewer to move a theoretical light source across the surface of the image, creating a shadow effect that can enhance discrete aspects of the carving. Precise metrical analysis can be carried out in order to enhance our interpretation and aid detailed comparison with other portraits. The initial findings of this pilot study have already been published within this journal (Russell and Manley 2013a) and elsewhere (Russell 2013; Russell and Manley 2013b; 2015).

At the time of the pilot study inception, three-dimensional (3D) laser scanning was widely utilised as an archaeological research tool on surfaces ranging in size from entire landscapes to small, individual artefacts. Short-range scanners, where the distance from the laser source to the object is typically <1m, have been extensively used by archaeologists, museum conservators and human osteologists for artefact monitoring and morphometric analysis, as well as for creating 3D models for public presentation (for example Kuzminsky and Gardiner 2012; Lobb et al. 2010; Lin et al. 2012). However, more recently, new technologies have become available that also can create highly accurate 3D models. Multi-imagery photogrammetry (sometimes referred to as Structure from Motion-SfM) uses multiple digital images captured from 2D cameras to create three-dimensional point clouds (Lerma and Muir 2014; Spring and Peters 2014) These point clouds can then be meshed to create solid objects for study, analysis and reproduction as 3D prints. The principle advantage of SfM is that standard, non-specialist digital cameras can be used to capture the images which can then be processed using free or low-cost software into the 3D models (e.g. Autodesk 123D; AgiSoft PhotoScan; VisualSfM). This has enabled a low-cost approach to be taken for 3D data capture and has significantly improved the ability for new interpretative methods to be used on portraiture and other archaeological artefacts.