Repatriation of the Mercer tsantsa to Ecuador required the generation of a report that 1) provided evidence of the provenance of the artifact and 2) compared the artifact to ceremonial tsantsas using criteria provided by the INPC. From this report, the INPC would ultimately decide whether or not to repatriate the tsantsa. To assist in authenticating the artifact, additional criteria were taken from the literature. The authenticity of tsantsas have previously been confirmed by qualitative assessment [10, 12, 23], morphometric analysis [24], and genetic testing [15, 25, 26]. These papers define ceremonial tsantsas as those produced by the SAAWC culture groups using traditional methods, while commercial tsantsas are a broadly defined category of artifacts originating in the nineteenth and twentieth centuries and produced for profit as opposed to meaningful cultural purposes. Commercial tsantsas may be constructed from a human corpse, an animal corpse, synthetic materials, or some combination thereof. The preparation of the commercial tsantsa does not follow traditional methods, but may be a convincing forgery. As a result of high-quality commercial tsantsas produced from human remains, some researchers have generated a third category of tsantsa referred to as “ambiguous” [2, 3, 24]. The term “ambiguous” does not designate a tsantsa as being commercial; rather, ambiguous tsantsas possess some, but not all, characteristics of ceremonial tsantsas with known origins.
Initial inspection of the Mercer Tsantsa
The examination of the tsantsa began with an inspection of its container and separation of the head from its mount (Fig. 1a). Prior to 1979, the tsantsa was stored in an airtight glass bell jar; the original container was discarded during the filming of the movie, and at some time in 1979 the tsantsa was mounted on a base made from a painted Styrofoam cup and stored in a plastic bell jar [22]. The neck was directly attached to the mount via an unknown adhesive. After the tsantsa was separated from the mount, remnants of the mount were gently separated from the neck. The container smelled of naphthalene, but no source was noted. The cavity of the head was filled with Ecuadorian newspaper from the 1940s. The newspaper was carefully removed with tweezers, and a fine black powder was found on the paper (Additional file 2: Figure S2). Charlier and coworkers [12] note that traditional preparation of a tsantsa is known to leave small amounts of charcoal or sand in the cavity. While this fine black powder may be an indication of remaining ash or charcoal, it also may have originated from degradation of the newspaper and/or ink.
Comparison of the Mercer Tsantsa to Ceremonial Tsantsa characteristics
Table 1 provides the checklist of criteria applied to the tsantsa, generated from the peer-reviewed literature and requirements from the INPC. This investigation was not designed to be scholarly in nature; the purpose was to repatriate the tsantsa to Ecuador and documenting the characteristics for authentication provided by the INPC was essential. The table is organized into the following categories: skin; structure and facial anatomy; modifications, and hair.
Analysis of the skin
Because ash was rubbed into the skin during the final stages of curing, the outer surface has a mahogany-black color (Fig. 1b). The skin was substantially thick and leathery, particularly noticeable at the neck, with no facial painting observed (Fig. 1b). The nose tip and cheeks were relatively smooth and lighter in color than the rest of the face, and downy hair was absent on the cheeks (Fig. 1b). The lack of hair on the cheeks is known to result from the use of hot rocks to “iron” the facial skin. The lighter shaded areas are consistent with a polished appearance accompanied by the breakdown of the outer epidermal layer. The combination of ironing and the smoking of tsantsas over open fires to discourage and prevent insect activity may explain the darkening of the skin [8, 12].
Analysis of the structure and facial anatomy
Assessment of the facial anatomy of the Mercer tsantsa was conducted during the initial inspection as well as through the use of computerized tomography (X-Ray CT). The tsantsa had the structural hallmarks of a ceremonial artifact. After processing, the size of the head was appropriate for a ceremonial tsantsa, being “no larger than a clenched adult human fist.” The neck’s opening had an oval shape, though somewhat distorted from the adhesive used to fix the tsantsa to the Styrofoam base. The tissue of the ears and nose were appropriately shaped presumably due to the presence of elastic and hyaline cartilages in these respective locations. Although smaller in size (like the head) the details of the ear were maintained (Fig. 1b, c). Both eyes were tightly closed, and no stitches, pitch, or adhesive was noted on the eyelids. The upper eyelid appears to be positioned over the lower eyelid. If the eyelids were indeed stitched shut then it might have been because of fine stitching known to be associated with commercial production. If this were determined to be true, it would not be consistent with authentic ceremonial production methods.
The nose itself was upturned and worn, indicative of handling as seen with traditional processing of ceremonial heads (Fig. 2a, c). Similarly, indents at the temples and protruded lips were observed and are evidence of handling during fabrication. However, the moderate degree of lip protrusion is not as exaggerated as in the fashion sometimes performed by SAAWC culture group members. The overall shape of the head is consistent with the description of being laterally pinched at the temples (Fig. 2d). These characteristics, minus the protruded lips, contribute to the Mercer tsantsa’s narrow head and elongated profile (Fig. 2c), typical of ceremonial tsantsas.
CT scanning allowed for viewing intricate details of the structure of the head not discernable by simple inspection, and has been previously used to identify characteristics of ceremonial tsantsas [2, 12]. The CT scan of the tsantsa had a voxel resolution of 115 \(\mu\)m. At this resolution, the majority of the anatomical features of the tsantsa were detected; however, very fine filamentous features such as some facial hairs and eyelashes were at the limits of detection. Micro-CT scanners, can yield voxel resolution less than 100 \(\mu\)m, but we did not have access to such an instrument.
The scan revealed that the head was self-supporting, and no wires or synthetic supports were used during construction of the tsantsa. Similarly, no bones or teeth were present in the head. CT-scans also allow for examination of the layers in the skin under the epidermis, if present. Houlton and Wilkinson [2] identify that the papillary layer often detaches from the reticular layer of the dermis during traditional processing. This “double hiding” is common in the leather industry, and was present in all ceremonial tsantsas analyzed in their study. The Mercer tsantsa does not have evidence of double hiding. Image processing revealed a single layer of integumentary tissue, presumably the papillary layer of the dermis, which is visible in the 3D volume rendering of a sagittal section (Fig. 3i) and in 2D CT scan images (Fig. 3f).
Of note, the head possesses a unique anatomical feature: a distinct discrepancy between the height of head surface on the left of the posterior stitching line and the height of the surface on the right of the line (Fig. 3g). Due to this tissue height difference, when the 2D sagittal views of the CT scan are generated at low resolution they have a similar appearance to double hiding (Fig. 3a–c). This initially misled investigators here, but what appeared to be double hiding is not observed when the 2D sagittal views of this unique anatomy are generated at peak resolution (i.e. 625 μm slices). Care must be taken to select appropriate data processing parameters to yield as precise image outputs as possible, otherwise unique anatomical features of a tsantsa may be incorrectly identified as double hiding as a result of image processing.
Analysis of traditional fabrication and modification
Modifications of the skin and anatomy of the tsantsa are incurred both during the life of the victim and the processing of the tsantsa. The most obvious modifications to the Mercer tsantsa are in line with those found in ceremonial tsantsas. Three holes in the upper and lower lips, joined together using a vegetal fiber, are also present in ceremonial examples (Fig. 1a). The mouth was slightly opened (Fig. 2c). During examination it became clear that the vegetal fibers strung through the lips had been damaged and degraded over time. Sections of the fibers that broke off during inspection were collected and stored separately from the head. It is difficult to determine if the mouth had been bound into a tighter closure when it was originally collected, as to be expected in an authentic artifact. Larger chonta pins (wooden thorns) were not observed and there were not significantly large perforations or distortions around the holes. Neither were there lengthy tassels made of more substantial vegetal fibers. However, it is possible that these indicators do not reflect the original state of the collected tsantsa. There are no photographs of it prior to it being used in filming a motion picture.
Although distorted by the adhesive used to fix the head to the base for display, the head was clearly separated from the body at or near the base of the neck at the level of the clavicles. The margin of the neck had no fibers sewn into it. While there was evidence of sawing on the internal surfaces at the base of the neck, the edge was relatively smooth and comparable to photos of ceremonial tsantsas in the literature [2]. We cannot eliminate the possibility that any fibers sewn into the neck’s edge may have been simply cut off at some point after the tsantsa was preserved, or perhaps damaged when attached to the mount for display. As noted, this tsantsa was used as a prop for a motion picture. During filming, the head was clearly attached to a small body. We can only assume that certain authentic tsantsa characteristics might have been damaged as a result of this type of use.
Internally, the observable dermis around the neck opening showed evidence of scraping. Stitches of vegetal fiber were visible down the posterior vertical midline of the head and neck (Fig. 4c, d). The stitching pattern was an uneven “baseball stitch” (Fig. 4d), so named for its similarity to the figure-8 stitching on a baseball [2]. Again, the images derived from the CT-scan proved invaluable to the analysis. The length and density of the hair on the Mercer tsantsa complicated visualizing the stitches. Preliminary examination indicated that stitching at the base of the neck was looped through the skin in an “over and over” pattern. However, the CT scan clearly indicates a “baseball” stitch pattern beneath the hair, not found in any of the ceremonial tsantsas described in the Houlton and Wilkinson study [2]. This stitch pattern was confirmed by careful separation of the hair on the back of the skull to examine the seam.
The surface rendering of the CT scan data also indicated a hole at the top of the head, hidden by the hair of the subject (Figs. 2c, d and 3b). This feature is observed in ceremonial tsantsas and it typically corresponds to where a string was attached. No vegetal fibers or string were observed in our tsantsa associated with this hole but these vegetal fibers, when present, are fragile and easily damaged/removed. Charlier and coworkers note that traditional preparation of a tsantsa is known to leave small amounts of charcoal or sand in the cavity, which may be visible in the CT scan [12]. No such debris was noted in the scan of the Mercer tsantsa, although a charcoal like residue was found in the newspaper that was removed from the cavity (Additional file 2: Figure S2).
Analysis of the hair
Hair is a useful factor for distinguishing between ceremonial and commercial tsantsas, and particularly identifying commercial tsantsas constructed of non-human materials [15]. The Mercer tsantsa possessed the distinctive, 3-tiered hairstyle that is consistent with hair worn by SAAWC culture group members. Small amounts of hair were present on the upper lip, between eyebrows, and within the nose. The fact that small amounts of facial hair were present in some locations (around the lips) and not in others (cheeks) is consistent with manufacturing steps that involve hair removal with hot stones during the preservation process [9,10,11]. Hair might remain in hard-to-access places around the lips where there are folds, or where large chonta pins might have been positioned around the mouth.
Microscopic examination of hair shafts enabled accurate measurement of cross-sectional diameter of both the hair shaft and the inner hair medulla (Fig. 4e). A sample of hair shafts and their inner medulla demonstrate that the average ratio of medulla width to hair shaft width is 0.23 ± 0.06 (medulla width 23.3 μm ± 8; hair width 98.4 μm ± 18.1). These samples demonstrate an average hair medulla/hair width ratio less than 0.33, a known forensic characteristic of human hair [27]. Lice nits were also observed (Fig. 4f), providing supporting evidence that the hair is human [28].
3D printed replica from CT scan
A to-scale 3D replica of the tsantsa was printed using a SLA printer, as they are capable of 25 μm – 100 μm print resolution (per the resin choice), and they are the preferred printing modality for anatomical reproductions [29]. The 3D rendering of the DICOM file was used to generate a printable .stl of the tsantsa. SLA printing of filamentous structures longer than a few millimeters and less than 100 μm in diameter is difficult, and if desired, requires extensive support scaffolds. The tsantsa’s hairs, particularly its scalp hairs, posed a significant challenge for printing. Because SLA printers produce objects with continuous smooth surfaces, conversion from a 3D surface rendering into a .stl file requires a program that identifies where all the 3D features connect. In the case of scalp hairs there are literally thousands of instances where the features do not connect, so the program attempts to fill in the missing connections. This connected scalp hair prevented any visualization of the features or surface of the scalp so we applied a filter to remove the scalp hair prior to generating the .stl file. The .stl file was then printed in Formlabs white resin using an SLA printer (Additional file 1: File S1).
Image processing of the 3D rendering to remove the tsantsa’s hair also removed its vegetal fibers. Thus, the 3D printed model also lacked hair and fiber stitching. The proteins found in vegetal fibers and human hair have similar structures i.e. they are predominantly comprised of alpha helical molecular structures [30, 31]. This could result in similar signals on CT scans and make isolating one filamentous signal (hair) from the other (vegetal fiber) difficult during digital filtering. In some locations, filtering to fully remove the hair resulted in holes in the scalp. These instances were repaired in Meshmixer; however, they were easily identified on the scalp of the 3D printed model. Certain prominent structural characteristics were observable in the 3D printed model such as the temple indentations formed during the traditional construction of the tsantsa (Additional file 1: Figure S3A and D) and the shape of the mouth. The entire surface of the model was smooth and white, i.e. no textural characteristics of the tsantsa’s skin were retained in the translation from artifact to CT scan to 3D printed model. Overall, the size of the tsantsa head and the shape of its features were maintained. Thus, a 3D printed reproduction can be useful as a replacement for an authentic artifact. Prints of tsantsas could be valuable for use in cultural and historical education either in a museum or classroom setting. The use of SLA rapid prototyping could allow for the decolonization of museum collections by repatriating such artifacts [32, 33].
Although the 3D printed model was a reasonable likeness of the Mercer tsantsa, it lacked many of the features critical for accurate evaluation by experts in the field. With alternative image processing one could virtually remove the tsantsa’s hair while retaining the stitches. Other features important for characterization and authentication (e.g. skin coloring and texture and presence of facial hair) are not possible in an SLA-generated model. Whole prints of the tsantsa could be carried out where aspects such as opacity, heaviness, and overall accuracy are varied to help confer the actual qualities of the original specimen. Studies in the literature that attempt to understand what a typical museum-going audience might prefer in a replicated artifact demonstrate clearly that “verisimilitude” (the closeness of the printed model compared to the authentic artifact) is the quality of highest preference [29, 34, 35]. In total, the CT-scan and corresponding SLA print generated here allows assessment of 12 out of 33 characteristics required for authentication. The CT scan data can be easily shared with other experts for review, whereas the rapid prototype best serves as a visual teaching aid.
Repatriation of the Mercer Tsantsa to Ecuador
Most checklist indicators of authenticity (30/33) affirm that the Mercer tsantsa is ceremonial; this includes 100% of the criteria provided by the INPC and those established by Charlier and coworkers [12]. Based upon hair analysis and the presence of nits, it is concluded that the Mercer tsantsa is of human origin. The presence of nits is in and of itself not disqualifying; however, Houlton and Wilkinson comment that only minor nit infestations are present in ceremonial tsantsas (2). With only one tsantsa available, it was impossible to determine whether the Mercer tsantsa had a minor nit infestation, or something more significant. It is also important to note that the analysis of the neck was complicated by the improper mounting of the Mercer tsantsa and its potential rough handling during the production of Wise Blood.
The Mercer tsantsa did not meet 3 requirements established by Houlton and Wilkinson. Double hiding was not confirmed via CT scan, the stitching on the back of the head was not in the typical “over and over” pattern, and the eyelids were not stitched shut; each of these criteria were met by 100% of the ceremonial tsantsas in the Houlton and Wilkinson study [2]. The authors note that a few “ambiguous” tsantsas did not possess double hiding. Additionally, certain ambiguous tsantsas also possessed baseball stitching. Although the Mercer tsantsa may have been classified as an ambiguous tsantsa according to these criteria, it met the authentication criteria established by the INPC. An Ecuadorian archaeologist from the Ministry of Cultural Heritage reviewed our report and supporting images, and issued a separate document confirming the findings.