doi: 10.1111/cote.12169 Armenian cochineal (Porphyrophora hamelii) and purpurin-rich madder in ancient polychromy Coloration Technology Dimitrios Mantzourisa and Ioannis Karapanagiotisa,b,* a Art Diagnosis Centre, Ormylia Foundation, Ormylia, 63071, Greece b Department of Management and Conservation of Ecclesiastical Cultural Heritage Objects, University Ecclesiastical Academy of Thessaloniki, Thessaloniki, 54250, Greece E-mail: y.karapanagiotis@aeath.gr Society of Dyers and Colourists Received: 10 January 2015; Accepted: 11 July 2015 Samples removed from funeral figurines, dated the third to the second century BC (Hellenistic period) and found in Macedonia, Greece, are investigated using high-performance liquid chromatography. Two results are reported that provide new insights into ancient polychromy. Firstly, high-performance liquid chromatography results, in combination with historical information, indicate that insect species of cochineal, most probably Porphyrophora hamelii Brandt, was used in Hellenistic objects. Secondly, madder detected in the archaeological samples contains high amounts of purpurin (alizarin is either not detected or detected in trace). This relative composition, which is recorded using the hydrochloric acid method for dyestuff extraction, is consistent with that prepared from wild madder (Rubia peregrina L.), which is a common plant in the Mediterranean region. Introduction The Hellenistic historical period ranges between the end of the Classical Greek and the rise of the Roman periods and corresponds probably to the climax of the famous ancient polychromy. However, reports describing the identification of organic colourants in Hellenistic objects are extremely rare [1–5]. High-performance liquid chromatography (HPLC) is used herein to investigate samples that were extracted from four funeral figurines found in two tombs of the Hellenistic period in Thessaloniki, Macedonia, Greece. The HPLC data are carefully processed and discussed, leading to the identifications of (i) cochineal (most probably Porphyrophora hamelii Brandt) and (ii) purpurin-rich madder in Hellenistic funeral figurines, thus expanding existing knowledge on the original ancient polychromy. Experimental Solvents and chemicals The following solvents and chemicals were used for chromatography and sample treatment processes: HPLCgrade water (Chem-Lab NV, Belgium), HPLC-grade acetonitrile and methanol (CH3CN and MeOH) (J.T. Baker, USA), trifluoroacetic acid (TFA) (Riedel-de Ha€en, Germany) of 99% purity, HPLC-grade dimethyl sulfoxide (DMSO) (Sigma-Aldrich, USA), and analytical-grade hydrochloric acid 37% (HCl) (Merck, Germany). Samples for analysis Figurines were found in two Hellenistic tombs (numbers 641 and 767, dated the third to the second century BC) excavated recently in the city centre of Thessaloniki, which is located in central Macedonia, Greece. Four samples removed from four different figurines were analysed using HPLC coupled to a diode array detector (DAD). The same HPLC method, described below, was used to analyse 370 samples of Mexican and Armenian cochineal, which were kindly provided by R Karadag, G Kremer, C Porter, and L Puchinger. Sample preparation Archaeological and cochineal reference samples (Armenian and Mexican cochineal extracts prepared from powders consisting of many insects) were treated using the standard hydrochloric acid (HCl) method [6], as follows: (i) the samples were immersed in 400 ll of H2O:MeOH:37% HCl (1:1:2, v/v/v) kept at 100 °C for 10 min; (ii) the samples were then evaporated (65 °C) under a gentle nitrogen flow; (iii) the dry residues were dissolved in 300 ll of DMSO; (iv) the samples were centrifuged, and the upper liquid phase was immediately submitted to HPLC. Chromatography The HPLC–DAD system (Thermoquest, USA) consisted of a 4000 quaternary HPLC pump, a SCM 3000 vacuum degasser, an AS3000 autosampler with a column oven, a Rheodyne 7725i injector with a 20 ll sample loop, and a UV 6000LP diode array detector. Analyses were carried out with an Alltima C18 (Alltech Associates, USA) column (5 lm particle size, 250 mm 9 3.0 mm) at a stable temperature of 35 °C. For gradient elution, two solvents were used, consisting of (A) 0.1% (v/v) TFA in water (pH = 2) and (B) 0.1% (v/v) TFA in acetonitrile. The flow rate was 0.5 ml min 1. The following gradient programme was applied: 0–1 min 5% B isocratic; 1–15 min linear gradient to 30% B; 15–20 min linear gradient to 60% B; 20–23 min 60% B isocratic; 23–33 min gradient to 95% B; 33–35 min 95% B isocratic. Data were received and analysed using the Xcalibur software (Thermoquest). Results and Discussion The results obtained for the archaeological samples are summarised in Table 1. Figure 1 shows the HPLC graph for © 2015 The Authors. Coloration Technology © 2015 Society of Dyers and Colourists, Colora. Technol., 131, 370–373 Mantzouris and Karapanagiotis Cochineal and madder in ancient polychromy Table 1 Colouring compounds identified in the archaeological samples: carminic acid (CA), kermesic acid (KA), flavokermesic acid (FL), alizarin (AL), purpurin (PU), and rubiadin (RU). HPLC peak areas integrated at 275 nm for cochineal and at 254 nm for madder components are provided in parentheses. For the peak areas measurements absorbance was taken in AU. Cochineal components Madder components Object/sample Tomb Marker Other Marker Other 1 2 3 4 767 767 641 641 CA (79.3) CA (12.8) CA (12.4) KA + FL (0.9) PU PU PU PU AL (0.9), RU (0.3) (17.3) (0.8) (1.4) (5.6) 14 × 103 Archaeological sample Absorbance, µAU CA 12 × 103 Compound 10 × 103 CA 223, 275, 309, 493 KA 223, 273, 309, 489 FL 223, 283, 343, 431 AL 213, 247, 277, 429 PU 215, 255, 293, 479 RU 217, 243, 277, 411 8 × 103 6 × 103 4 × 103 KA Absorbance maxima, nm FL AL PU 2 × 103 RU 0 10 15 20 25 30 Time, min Figure 1 HPLC chromatogram collected at 410 nm for archaeological sample 1. The following compounds are identified: carminic acid (CA), kermesic acid (KA), flavokermesic acid (FL), alizarin (AL), purpurin (PU), and rubiadin (RU). Absorbance maxima of the detected compounds are included in the table sample 1, which according to Table 1 is the richest in terms of detected organic compounds. Carminic acid was identified (Figure 1), leading to the conclusion that cochineal was used in sample 1. The cochineal pigment found in the archaeological sample must have been prepared with scale insects that are native to Eurasia and have been known since antiquity, such as Porphyrophora hamelii Brandt (Armenian cochineal) and Porphyrophora polonica L. (Polish cochineal) [6]. Cochineal imported into the European continent from the New World could not have been used in objects of the Hellenistic period. Wouters and Verhecken [6] suggested that Armenian and Polish cochineal can be distinguished in a graph that takes into account the HPLC peak areas of carminic acid (CA), the marker compound for the identification of cochineal, 2-Cglucoside of flavokermesic acid [7], formerly labelled dcII, kermesic acid (KA), and flavokermesic acid (FL), integrated at 275 nm. The Wouters and Verhecken method is reconstructed in Figure 2 and evaluated by analysing three samples of Armenian cochineal using our chromatographic method. Polish cochineal was not available. Figure 2 shows that the results collected herein for the three Armenian cochineal samples are in agreement with the regimes specified previously by Wouters and Verhecken [6] for Porphyrophora hamelii. Further evidence that supports the agreement between the Wouters and Verhecken method and the results collected from our chromatographic system is provided by the analyses of three samples of Mexican cochineal. The three analysed Mexican cochineal samples fall within the corresponding regime specified by Wouters and Verhecken, as shown in Figure 2. Consequently, Figure 2 suggests that our HPLC analyses of the reference cochineal samples are in agreement with the previously published report [6]. CA, KA, and FL were detected in sample 1 (Figure 1 and Table 1), and therefore it was possible to include the HPLC results for this sample in Figure 2. It can be seen that the archaeological cochineal pigment falls within the Armenian cochineal regime. It is stressed, however, that cochineal pigments were not included in the Wouters and Verhecken investigation [6], where insect and dyed yarn extracts were analysed. Similarly, the cochineal reference samples, analysed herein to examine the agreement between our results with the Wouters and Verhecken method, were insect © 2015 The Authors. Coloration Technology © 2015 Society of Dyers and Colourists, Colora. Technol., 131, 370–373 371 Mantzouris and Karapanagiotis Cochineal and madder in ancient polychromy 4.0% Mexican cochineal (analysed herein) % 2-C-glucoside of FL 3.5% 3.0% Armenian cochineal (analysed herein) 2.5% Mexican cochineal (according to [6]) 2.0% 1.5% Armenian cochineal (according to [6]) 1.0% Archaeological sample (analysed herein) 0.5% 0.0% 0% 2% 4% 6% % KA+FL Figure 2 Relative compositions of cochineal materials, measured as percentage integrated HPLC peak areas at 275 nm, with respect to the 2-C-glucoside of FL (formerly known as dcII) [7], KA, and FL. Regimes specified by boxes correspond to Mexican and Armenian cochineal according to the analysis of Wouters and Verhecken [6]. For Polish cochineal it was reported that % KA + FL ranges from 12 to 38% [6], which is beyond the limit of the x-axis of the graph. HPLC results obtained herein for reference samples of Armenian and Mexican cochineal and archaeological sample 1 are included in the graph [Colour figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.] extracts. The preparation procedure of a cochineal pigment may affect the relative composition of the key compounds detected by HPLC. Consequently, the Wouters and Verhecken regimes specified in Figure 2 may not be appropriate for identifying the origin of a cochineal pigment. This concern is important considering that, according to Figure 2, the separation of Armenian and Mexican cochineal is mainly based on a slight difference in the amount of 2-C-glucoside of FL. Therefore, a potential slight effect of the pigment production process on the relative composition of cochineal could have a major impact on the separation of the two cochineal species. For these reasons, we report that the potential use of Mexican cochineal in archaeological sample 1 cannot be excluded solely by the chemical results of Figure 2. The identification of Porphyrophora hamelii Brandt in the archaeological sample is achieved only if the chemical results of Figure 2 are interpreted in light of the historical information, which excludes the use of Mexican scale insects in Hellenistic objects. Furthermore, CA was identified in samples 2 and 3, but in amounts smaller than in sample 1 (according to Table 1). As KA and FL were not detected, the results of these two samples cannot be included in the graph of Figure 2. However, the potential use of Polish cochineal in samples 2 and 3 is extremely unlikely, as described in the following discussion. For Polish cochineal it was reported that the integrated HPLC peak areas at 275 nm are as follows: the % KA + FL ranges from 12 to 38, and the % CA from 62 to 88 [6]. If samples 2 and 3 contained Polish cochineal, then the smallest HPLC peak areas integrated at 275 nm for KA + FL should correspond to 1.75 for sample 2 and to 1.69 for sample 3, according to the CA peak areas given in Table 1. These values are calculated using the lower limit of 12% reported for KA + FL in Polish cochineal [6] and are almost double the % KA + FL measured in sample 1 (0.9), thus implying that KA + FL should have been detected by our chromatographic system if Polish cochineal was used in the two archaeological samples. Consequently, the absence of KA + FL in the results of samples 2 and 3 yields to the conclusion that a Eurasian cochineal species, poorer in 372 KA + FL, such as Armenian cochineal, was used in the corresponding archaeological figurines. The chromatogram in Figure 1 shows that purpurin (PU) was detected in sample 1, thus revealing the presence of madder. Alizarin (AL) is found in very small amount, and rubiadin (RU) is clearly detected in the graph of sample 1. HPLC peak areas integrated at 254 nm are summarised in Table 1. Only PU was detected in samples 2, 3, and 4. Consequently, Table 1 shows that only PU-rich madder pigments were used in the Hellenistic figurines. Several plants of the Rubiaceae family, such as Rubia cordifolia L. and Rubia peregrina L., contain primarily PU and/or considerable amounts of RU (Galium species) [8,9]. In the Mediterranean region, Rubia peregrina L. (wild madder) is the most common, along, of course, with Rubia tinctorum L., which, however, is rich in AL [10], and therefore its potential use in the Hellenistic objects is not supported by the results of Figure 1 and Table 1. However, as discussed previously for cochineal, the preparation procedure of an organic pigment may affect the relative composition recorded by HPLC, and therefore the unequivocal identification of the madder source used in the figurines is not possible. The identification of madder in Hellenistic objects, other than funeral figurines, has been scarcely reported [1,2,4,5]. In two studies, scientific evidence suggested the use of PUrich madder pigments in a basin [2] and wall painting pigment [4]. These results are in agreement with our findings. Two other studies mention the use of the AL-rich Rubia tinctorum L. species in archaeological objects, but without providing clear scientific evidence to support the presence of AL [1,5]. Consequently, we conclude that the unequivocal chemical identification of an AL-rich Rubiaceae pigment in ancient Hellenistic objects has not been reported to date. Conclusions The Wouters and Verhecken method [6] was reconstructed using our own chromatographic system and analytical © 2015 The Authors. Coloration Technology © 2015 Society of Dyers and Colourists, Colora. Technol., 131, 370–373 Mantzouris and Karapanagiotis Cochineal and madder in ancient polychromy procedures. This chemical method, in combination with historical information, indicated that some species of cochineal, most probably Porphyrophora hamelii Brandt, was used in funeral figurines dating back to the Hellenistic period and excavated in Macedonia, Greece. Rubiaceae pigments detected in samples removed from the archaeological objects, after treatment with HCl, contained high amounts of purpurin; alizarin was either not detected or detected in trace. References 2. A Wallert, Stud. Conserv., 40 (1995) 177. 3. P Maravelaki-Kalaitzaki and N Kallithrakas-Kontos, Anal. Chim. Acta, 497 (2003) 209. 4. I Karapanagiotis and Y Chryssoulakis, Ann. Chim. (Rome), 1–2 (2006) 75. 5. H Brecoulaki, E Fiorin and P A Vigato, J. Cult. Herit., 7 (2006) 301. 6. J Wouters and A Verhecken, Stud. Conserv., 34 (1989) 189. 7. K Stathopoulou, L Valianou, A-L Skaltsounis, I Karapanagiotis and P Magiatis, Anal. Chim. Acta, 804 (2013) 264. 8. D Cardon, Natural Dyes: Sources, Tradition, Technology and Science (London: Archetype, 2007). 9. C Mouri and R Laursen, Microchim. Acta, 179 (2012) 105. 10. J Wouters, Dyes Hist. Archaeol., 16/17 (2001) 145. 1. M Farnsworth, J. Archaeol., 55 (1951) 236. © 2015 The Authors. Coloration Technology © 2015 Society of Dyers and Colourists, Colora. Technol., 131, 370–373 373