13. Scrutinizing “Sarapon”: Investigating a Mummy Portrait of a Young Man in the Michael C. Carlos Museum, Emory University

In 2004 the Carlos Museum of Emory University acquired a mummy portrait that depicts a young, beardless man wearing a white tunic (fig. 13.1). He has dark curly hair, thick eyebrows, and full lips. Within a red, tabula ansata–shaped label at the subject’s right, an inscription in Greek provides his name, patronymic, and age at death. He is identified as “Sarap[i]on,”¹ son of Haresas, 25 (or 29) years.² Little more than 3 percent of the approximately one thousand mummy portraits worldwide include texts with the deceased’s name,³ making this example both rare and significant.

The portrait’s modern history began in the early twentieth century, when Philadelphia art collectors Vera and Samuel Stockton White III purchased the work from the Dikran Kelekian galleries. The Whites’ substantial art collection was donated to the Philadelphia Museum of Art in 1967; however, the mummy portrait was not included in that gift and instead was sold by Parke-Bernet Galleries, New York, presumably to its last private owner, businessman Jonas Senter.⁴

Mummy Portrait of Sarap[i]on, Romano-Egyptian, ca. second century AD. Wood, pigments, wax, and glue, 33.7 x 41.9 cm (13 ¼ x 16 ½ in.) Atlanta, Michael C. Carlos Museum, Emory University, Mohamed Farid Khamis / Oriental Weavers Fund, 2004.048.001.

fullscreen Figure 13.1

Upon acquisition by the Carlos Museum, the painting was treated to stabilize loose fragments, remove a historical but inappropriate frame, and minimize its fragmentary appearance by selectively inserting toned fills. The report from that treatment confirmed earlier observations: the portrait was assembled from more than one painting.⁵ The APPEAR project motivated this present reconsideration, which also benefits from improved technology.

The Carlos painting consists of more than fifty fragments of varying lengths and widths. The woods range in color, and their painted surfaces differ in thickness, ground, and brushwork. While the fragments can be differentiated by close visual inspection, this study also employed materials analysis and multispectral imaging to associate the fragments into groups.

Gas chromatography/mass spectrometry analysis of representative paint samples identified traces of aged beeswax on some fragments and animal glue on others; an egg coating might have been selectively applied. Analysis and examination are complicated, however, by the presence of modern animal glue used to secure the fragments to the plywood backing. Overpainting created most of the proper left eye and cheek, reshaped the proper right shoulder, and produced the highlights on the tunic’s folds. Additional media samples along with wood identifications may further differentiate fragment groups.

A complex puzzle emerges from combining close visual examination and media analyses with multispectral imaging and elemental mapping. Individual fragments and brushwork are highlighted by differences in radio-opacity in X-ray images. Fragments can be further distinguished and associated by their appearances under ultraviolet induced visible fluorescence (fig. 13.2) and false-color infrared imaging, revealing modern interventions. The distribution of elements present in various paints also suggests relationships among fragments. Elemental mapping by scanning X-ray fluorescence spectroscopy showed lead to be present in some whites as well as some reds, while iron is present in retouching on the face (fig. 13.3). Calcium is concentrated in the group of fragments above the ear, which have a visible, thick, white ground. Zinc is associated with modern reworking; the absence of zinc in the fragments that constitute the inscription is noteworthy. Scanning electron microscopy and energy dispersive X-ray spectroscopy analysis of a sample from the white letters revealed a highly crystalline compound of oxygen, aluminum, sulfur, and sodium, with trace inclusions of lead. Raman spectroscopy confirmed the presence of a sulfate compound, indicating that the letters are painted with (sodium-?) aluminum sulfate.

fullscreen Figure 13.2

fullscreen Figure 13.3

The inscription fragments are smoothly painted, probably in glue tempera, on dark wood with no visible ground layer. Those fragments and others adjacent to them appear similar by visual examination, multispectral imaging, and elemental mapping. Some fragments associated with this group depict carefully delineated black curls, indicating that the named deceased had dark hair with wiry curls. It is unlikely, however, that the face presents Sarap[i]on’s likeness. Examination and analysis reveal that the face fragments were instead painted in wax encaustic, and although probably ancient, they have been reworked. Thus, the fragments that now compose the Carlos portrait were likely taken from three or more ancient portraits, perhaps from among those found at Antinoöpolis, er-Rubayat, and/or Hawara.

In 1966, while it was still in the White collection, the portrait was described by Klaus Parlasca as “a heavily overpainted pastiche,”⁶ and in 1993 Dominic Montserrat pronounced that it was of “dubious authenticity.”⁷ Despite these reservations, however, Parlasca and Frenz did not list the Carlos portrait in the forgeries section of their most recent volume of the Repertorio—perhaps because they acknowledged that it “incorporates original fragments.”⁸

It could be argued that as a modern assemblage, retouched and intended to present a unified whole, the Carlos portrait constitutes a fake. Yet, there is historical accuracy to both the depiction and the object. The text and the image record and evoke the life of a young man who died in Roman Egypt and was memorialized according to contemporary religious practices and regional stylistic preferences. The Carlos portrait is, therefore, a modern construct that is representative of both the named deceased and the genre of ancient painting to which mummy portraits belong. In its re-presentation of ancient fragments, this object underscores the subtle distinctions that affect the assignment of “authenticity.”

Acknowledgments

We thank Melinda Hartwig for permission to publish. We acknowledge the contributions of Jessica Betz Abel, Madeline Beck, Ann Brancati, Miriam Cady, Caroline Cartwright, Claire Fitzgerald, Eric Formo, Stacey Gannon-Wright, Joy Mazurek, Brian P. Muhs, Tanya Olson, Michaela Paulson, Tina Salguero, Morgan Webb, Terry Wilfong, and the Andrew W. Mellon Foundation.