*3.2. The Purple Pigment Find from Trianda, Rhodes*

A mass of purple pigment (TRI-PR13) was found at a depth of 3.40 m below the present surface in a layer of debris discarded in the northeast open space outside the monumental building located north of an E–W street crossing the southern zone of the prosperous Late Bronze Age IA city at Trianda [35]. The building (partially revealed in max. dimensions: 16.10 × 13.80 m) is a Xeste (of ashlar masonry), similar to the Xestai at Akrotiri, Thera [35,36]. The pigment was recovered from the earth filling of the debris bordering the west facade of the Xeste, partly covered with tephra from the Thera eruption. It was found just beneath a small fresco fragment (APX 2098) [37] depicting an elaborate black running spiral on a shiny whitish surface, among other fragmentary monochrome, polished dark grey, reddish, orange, or whitish frescoes; pieces of round tripod offering tables made of lime plaster; a reddish schist slab; and fragments of conical or some semiglobular cups and painted Late Bronze Age IA pottery, as well as animal teeth and bones. The pigment had the appearance of a cohesive dry mass, most probably once kept in a container of perishable material (see Figure 1).

The context comprised a secondary deposition of several materials fallen among the pre-eruption earthquake ruins, which were moved to outside the ruined building in order to clean it and prepare for its reconstruction. Similar contexts appear very often in the Late Bronze Age IA layers, mostly outside buildings, and they are indicative of the intensive cleaning activities of the earthquake ruins at Trianda before the tephra fall [36] (pp. 84–85), [38] (pp. 31–32), [39] (pp. 103–104, 107, 109). The excavation in the Paraskeva plot yielded abundant evidence of such intensive activities, since the exterior areas of the Xeste north of the E–W street was covered almost entirely with such secondary depositions containing many important materials [36] (pp. 83–85) and provided strong contextual evidence for the explanation of several aspects of the settlement and the activities of the people.

#### *3.3. Purple Paint Details on a Wall Painting Excavated at Raos, Thera*

The excavations at Raos brought to light a significant Late Cycladic I/Late Minoan IA complex atop a long crest consisting of a building surrounded by a spacious courtyard bounded by a robust stone-built enclosure wall [40–42]. Despite the poor state of preservation by Theran standards, the evidence shows that the Raos "villa" would have been quite impressive, with an upper story, high-quality wall paintings, and rather luxurious household equipment. It was situated on the outskirts of the Late Cycladic I/Late Minoan IA city at Akrotiri, less than 1 km from its northwest limit, on the natural land route leading from the coast to the hinterland of the Akrotiri Peninsula, where there are extensive tracts of arable land.

The fresco fragment from which the sample was taken derived from the north part of the southeast wing of the Raos residential building. It belonged to a heap of fresco fragments recovered on the floor of the ground floor. The heap covered an area of approximately 2 m2. Most of the fresco fragments were found on entire or broken mud bricks and must have come from a collapsed mud-brick wall.

Near the heap of fresco fragments, two dressed tuff bases of a pier-and-door partition were found in situ. The evidence shows that a pier-and-door partition provided access to the southeast wing via a paved corridor at the heart of the building. Almost no moveable finds, neither in situ nor fallen from an upper story, were excavated in the north part of the southeast wing, i.e., in the space where the wall paintings were found. The exact plan and arrangement of the wing is difficult to grasp because of its almost complete destruction. However, both the particular architectural features (pier-and-door partition, frescoed mud-brick wall) and the absence of moveable finds indicate that this area was possibly used for rituals or formal receptions.

All the fresco fragments belonged to one composition, which we called the "Wall Painting with the Rosettes and Lilies." This is tripartite in structure. An upper narrow frieze of polychrome rosettes, some of which have purple petals, crowns a wide middle zone (Figure 2). In this main zone, pendent red triangular surfaces alternate with upright triangles, covered with scattered red lily flowers on a white ground. A lower zone of oblong panels embellished with colored wavy bands imitates a polychrome veined marble dado [41] (pp. 212–213), [42] (pp. 137–139).

**Figure 2.** (**a**) Photograph of the wall-painting detail from Raos, with an indication of the sampling site on the purple petal of the rosette and (**b**) microphotograph of the cross-section of sample RAOS-KAL1.

#### **4. Analytical Results and Discussion**

#### *4.1. Identification of the Brominated Indigoid Chromophores in the Purple Pigment*

microRaman spectra were acquired in all the examined archaeological purple samples, on grains set aside from the three finds of purple pigment lumps (AKR-10882, AKR-10891, and TRI-PR13) and on the microsample from the purple paint detail in the Raos wall painting (RAOS-KAL1) prepared in a polished section, shown in Figure 2. The spectra are presented comparatively in (Figure 3); they present a perfect match of their peaks in both position and relative intensity for most of the characteristic vibrational bands of 6,6 -dibromoindigotin (DBI), the main chromophore compound of murex purple. For comparison, the spectrum taken from a reference 6,6 -DBI substance, synthesized according to previously described methods [43], is given in the same graph.

**Figure 3.** MicroRaman spectra acquired on samples from the purple pigment finds from Akrotiri and Trianda, (AKR-10882, AKR-10891, and TRI-PR13), as well as from the purple paint detail on the wall painting from Raos (RAOS-KAL1). The spectral signature of 6,6 -DBI is displayed in all the spectra acquired on the archaeological samples, shown in comparison with the spectrum taken on a synthetic 6,6 -DBI substance, [43].

The band positions and assignments to the vibrational modes for the 6,6 -DBI, the main chromophore compound of Tyrian purple, were previously published in comparison with indigo and its corresponding dihalogenated species 6,6 -DXI (where X: I, Br, F, Cl) [44]. A closer look at the spectra of the indigo and the dihalogenated indigoids, comparatively, allows for many similarities to be noticed, which is to be expected given that the majority of the detectable Raman bands are attributed to the indigoid structure. However, the exact position and the relative intensities of the major bands are proper to each halogenated indigoid compound, as the different halogeno-substituents induce differential changes (wavelength shift or relative intensity variance) in the characteristic vibrations that are in common, which makes the Raman spectrum a fingerprint for its identification. The specificity of each Raman spectrum extends even to differentiate two DBI positional isomers, as demonstrated in a recent paper comparing 5,5 -DBI to 6,6 -DBI and showing that the position of the bromine substituents affects many of the key bands of DBI and especially the wavelength position and the relative intensity of the band assigned to the ν(CBr) stretching vibration [45]. The 6,6 -DBI in particular presents the Raman bands corresponding to the benzene ring quadrant stretching vibrations ν(C=C) at ~1581 (s) cm−<sup>1</sup> and ~1625 (ms) cm−1, the band assigned to the carbonyl stretching ν(C=O) at ~1700 (m) cm−1, the band corresponding to the bending δ(NH), δ(CH) modes at ~1445 (m) cm−<sup>1</sup> and ~1360 (m) cm<sup>−</sup>1, the stretching ν(C–N) modes at ~1300 (m) cm−1, and a strong band at ~1250 (s) cm−<sup>1</sup> corresponding to δ(CH), δ(C=O) bending modes. In addition, the band proper to the brominated indigoid compounds, corresponding to the stretching modes of the Br-C bonds, which for the 6,6 -DBI are at the symmetrical positions 6,6 , is strongly resolved at ~305 (s) cm<sup>−</sup>1.

The identification of the 6,6 -DBI in the purple pigment through a well-resolved microRaman spectrum answers, satisfactorily in most cases, the primary archaeological question: the recognition of the murex identity of the dye, which automatically ascertains its added value. However, with Raman analysis, it is unlikely to detect any other than the DBI chromophore compound possibly present in the dye.

With the HPLC–DAD analysis, a more accurate determination of all chromophores present in the samples was possible. Three major peaks were detected in the HPLC profiles (Figure 4), which correspond to 6-bromoindigotin (MBI), 6,6 -dibromoindigotin (DBI), and 6,6 -dibromoindirubin (DBIR). Moreover, indigotin (IND), 6 -bromoindirubin (MBIR) and 6-bromoindirubin (MBIR) were detected in the chromatograms but corresponded to very small HPLC peaks, as illustrated in Figure 4 in the chromatogram of sample AKR-10882 shown in higher detail. Finally, indirubin (IR), which is usually found in molluscan extracts in small amounts [46–48], was not detected in any of the examined archaeological samples.

**Figure 4.** Chromatograms that were collected at 288 nm for the archaeological samples. The major HPLC peaks correspond to MBI, DBI and DBIR. Much smaller peaks were recorded for IND, 6'MBIR and 6MBIR as revealed for instance in the chromatogram of sample AKR-10882, shown in higher detail. The graphs for the samples AKR-10891 and AKR-10882 were adapted from [30] (with permission from Elsevier) and [32], respectively.

The aforementioned observations, which are visualized in the HPLC chromatograms, are clearly stated in the relative (%) values estimated from the integrated HPLC peak areas, as given in Table 1. The measurements were carried out using a monitoring wavelength of 288 nm [30]. Apparently, the given numbers are not the actual mass compositions of the samples. Based on the findings of a previous study it was to be expected that the % of DBI and IND reported in Table 1 would be an underestimation and overestimation, respectively, of the actual relative mass concentrations of the two compounds [48]. However, the semiquantitative results, which are summarized in the table, are a good approximation of the actual concentrations and a very useful approach to compare the samples.

**Table 1.** Relative (%) integrated HPLC peak areas measured at 288 nm. The results are provided separately for the major HPLC peaks (MBI, DBI and DBIR) and as a sum for the minor HPLC peaks.


The chromatographic analysis in Figure 4 and Table 1 confirmed the identification of molluscan purple in the four archaeological samples and revealed that the purple pigments and paints have similar compositions. DBI is the dominant compound in the four chromatograms of Figure 4, followed by MBI and DBIR. In the case it was detected, IND corresponded to an extremely small peak. These results suggest that, based on the identified composition of indigoid chromophores, the Late Cycladic purple pigment should have had a pure purple hue, with balanced proportions of red and blue components if not with a tendency to reddish purple. The actual color of the investigated samples, as appeared to the naked eye (see Figures 1 and 2a) and under microscope (see Figure 2b and Figure 6), as well as the measurements with VIS spectroscopy in the find (inv. no. 10891) [23], confirmed the HPLC-based assumption: There was no predominant bluish tint, which is often observed in molluscan pigments rich in blue IND [49].

As reported in previous published works, in the Mediterranean, there are three mollusc species that were used for dyeing and painting in antiquity: *Hexaplex trunculus* L. (*Murex trunculus*), *Bolinus brandaris* L. (*Murex brandaris*), and *Stramonita haemastoma* (*Thais haemastoma*). Studies on the significant amounts of shells found in several settlements on Crete (Palaikastro, Kouphonisi [5,50,51], Kommos [52], Chryssi, Pefka [53,54]), Kythera, Thera, Greece) and other insular and continental sites in Greece, as well as on the Eastern Mediterranean coasts, all agree on the almost exclusive majority of *M. trunculus*, with the exception of the Kythera case (a small settlement near Kastri), where *M. brandaris* is attested to be in abundance [55], whereas the incidence of *T. haemastoma* is always significantly lower.

The extraction of the Bronze Age Aegean purple principally from the *M. trunculus* as suggested by the archaeomalacological investigations is here supported by the HPLC findings with reference to related chemical studies [30]. These results about the biological source of the purple, although indicating the foremost exploitation of *M. trunculus* in purple dyeing, do not exclude the scenario of the addition of a regulated proportion of raw material from the *M. brandaris* species in order to obtain a unique, warmer shade of the expensive color.
