**6. Purple but not True Purple**

Shellfish purple was held in high esteem, as it is hard to be produced in large amounts from the molluskan raw source. In his pioneering work, Friedländer collected 12,000 *M. brandaris* extracting only 1.4 g of dry pigment [24]. It is estimated that 10,000 *M. trunculus* mollusks are needed to dye a kilogram and a half of wool [122]. Consequently, shellfish purple had always been an expensive material, which was sparingly used. Alternative procedures to mimic true purple had been developed and therefore "not all purples were equal, and not all purple was purple" [123].

For example, mixtures of Egyptian blue with red pigments were used in painting plasters to achieve a purple hue [124]. For the same reason, blue pigments and paints of Lapis lazuli [125] and indigo/woad [121] had been used in painting backgrounds since the Mycenaean period. Egyptian blue mixed physically with a red lake, which however is unidentifiable, was revealed in wall paintings fragments in Egypt [63]. Mixture of Egyptian blue and a red-pink lake was also found in 3rd century BCE oinochoe (British Museum) [45]. According to HPLC results, the lake was derived from purpurin-rich madder, cochineal (*Porphyrophora* spp.) and, unexpectedly, lac (*Kerria Lacca* Kerr) [45]. Mixtures of purpurin-rich madder and Armenian cochineal *Porphyrophora hamelii* Brandt were found in funeral figurines dated to 3rd–2nd century BCE [126,127].

A common practice for the Egyptian dyers was to mix madder and indigo/woad [8,34,93] achieving a remarkable purple hue. The purple mixture was sometimes enriched with a coccid dye, such as, kermes [128]. An analogous practice was revealed in Roman-Egyptian mummy portraits (2nd century CE) in which mixtures of madder lakes and indigo were identified and were probably applied as cheaper substitutes for shellfish purple [129]. Kermes in mixture with indigo/woad—but not madder—was identified in purple parts of a textile from the Topkapi Palace [130].

Folium and particularly orchil were widely used as alternatives to shellfish purple in the coloring of parchments in which the use of true purple was extremely rare [112,131,132], as discussed previously. Orchella weeds, when properly processed, provide a bright purple, which has been often detected in illuminated manuscripts [108,133].

#### **7. Dyeing with Shellfish Purple: From Purple to Blue**

Dyeing with shellfish is complicated. The colored water-insoluble extract must be chemically reduced in the dye vat to give the water-soluble and colorless leuco-forms of the compounds of Figure 3c. Reduction is induced by bacteria according to the mechanism that was elucidated for the reduction of woad [134]. The dyeing process can have an enormous effect on the composition and therefore the color of the attached dye considering that several parameters have to be adjusted in the vat (treatment time, temperature, pH, etc.). Moreover, sunlight can induce debromination of the leuco forms, which, moreover, do not have the same affinity for textile fibers [29].

The dramatic effect of the dyeing conditions on the composition of the attached dye is demonstrated in the following experiment. *M. trunculus* mollusks were collected by V. Gatsos from the sea of Hermione, Greece. The city of Hermione was famous for the fine dyeing with shellfish purple, an industry that had been flourished for more than 1000 years, from 6th century BCE to 6th century CE [12]. It was the Hermione purple that caused the admiration of Alexander the Great when he took Susa in 331 BCE and found purple garments in the palace of Darius III [7,12]. Wool and cotton were dyed by V. Gatsos following two recipes. Large mollusks (*M. trunculus*) were collected at a depth of 2–3 m, from the sea, just next to the ancient wall of Hermione.

Recipe I: The shells were crushed using a stainless steel tod at the third helix where the gland of the mollusk could be easily reached. The glands were quickly removed to preserve the secretions and placed in a flask containing 30 ml of water and 2 g of common salt. The flask was left open and was rigorously agitated 3–4 times per day for 15 days. After this process the pulp became purple. For the dyeing process, the purple pulp was transferred to a glass container. Ten (10) g of honey and 3 g of salt were added. The container was airtight sealed, agitated and remained into a water bath which was heated at 45 ◦C for two days. Then a 1 cm × 3 cm piece of fabric (wool or cotton) was immersed into the mixture and additional amounts of honey (2 g) and salt (1 g) were added. The container was airtight sealed, agitated and remained into a water bath which was heated at 45 ◦C for one day. The fabric was removed and washed with warm water (70 ◦C) and soap solution.

Recipe II: The shells were crushed and the glands were placed in a flask containing 80 mL of water and huge amount (20 g) of common salt. The flask was left in the sunlight for five days. Then, the dyeing procedure described in recipe I was followed.

Consequently, in recipe II, extreme conditions related to the quantity of salt and the duration of sunlight exposure were selected. The possible role of salt in the dyeing process is critically discussed in detail by C. Cooksey [18]. The four samples, two wools and two cottons, dyed with the two recipes, were analyzed using HPLC. A DMSO bath at 80 ◦C was used to extract the purple dyes [42]. The % relative integrated HPLC peak areas were measured and the results are provided in Table 3. The two recipes gave totally different results, as a dominant debromination process was developed in recipe II, resulting in reduced amounts of brominated indigoids (MBI and DBI) attached to the fibers. IND is clearly the major coloring compound in the samples, which were prepared using recipe II. On the contrary, large amounts of MBI were attached to fibers dyed with recipe I. The difference of the two recipes in the dyeing results was visible by naked eye, as samples prepared using recipe I were purple whereas samples that were dyed using recipe II were blue.


**Table 3.** Relative (%) integrated HPLC peak areas measured at 288 nm for wool and silk, which were dyed with two recipes, as described in the text.

### **8. Solubility Issues**

Solubility data is available in the open literature for indigo, which contains indigotin and indirubin but not their brominated derivatives. Table 4 shows the solubility of indigo in ten solvents [135]. The results of Table 4 were calculated using the COSMO–RS (conductor-like screening model for real solvents), which is a quantum-mechanical approach [135]. The best solvent found is sulfuric acid, in which COSMO–RS predicts complete miscibility with indigo. On the other hand, water practically does not dissolve indigo.

**Table 4.** Results of the conductor-like screening model for real solvents (COSMO–RS) solubility screening for indigo. The results were adapted from elsewhere [135].


Efficient extraction of dyes from fiber samples is important for successful HPLC analysis of textiles of the cultural heritage. Among the solvents included in Table 4, DMSO [42,62] and pyridine [42,97] were suggested for the extraction of shellfish purple from archaeological samples. Moreover, N,N–dimethylformamide (DMF), which is not included in Table 4, was also used for the extraction of the purple material [42,78]. DMSO and DMF have similar solubility properties considering that they have comparable Hansen solubility parameters [136]. The latter are summarized in Table 5 for the three aforementioned solvents.

**Table 5.** Hansen solubility parameters: δd, δ<sup>p</sup> and δhb are the dispersion, polar and hydrogen bonding parameters, respectively [136].


The efficiencies of DMSO, DMF and pyridine to solubilize shellfish purple were compared using HPLC [42]. The experimental results showed that DMSO and pyridine result in very good and poor yields, respectively [42], which is in agreement with the prediction for indigo provided by the COSMO–RS (Table 4). Moreover, the experimental study showed that DMSO and DMF have comparable efficacy in solubilizing shellfish purple [42], which is in agreement with the similar Hansen solubility parameters of these two solvents (Table 5).
