Identification of Yellow Dye Types in Pre-Columbian Andean Textiles

Jan 10, 2007 - Samples of textile fiber yarns (usually 0.2−0.8 mg) were extracted in with 5% formic acid in methanol as described by Zhang and Laurs...
37 downloads 12 Views 290KB Size
Download PDF
Anal. Chem. 2007, 79, 1575-1582

Identification of Yellow Dye Types in Pre-Columbian Andean Textiles Xian Zhang,† Ran Boytner,‡ Jose´ Luis Cabrera,§ and Richard Laursen*,†

Department of Chemistry, Boston University, Boston, Massachusetts 02215, Cotsen Institute of Archaeology at UCLA, Los Angeles, California 90095, and Departamento de Farmacia, Facultad de Ciencias Quı´micas, Universidad Nacional de Co´ rdoba, Ciudad Universitaria, 5000 Co´ rdoba, Argentina

A number of pre-Columbian textiles, most discovered in northern Peru and dating to the Late Intermediate Period (ca. 1050-1200 AD), were analyzed by high-performance liquid chromatography with diode array and mass spectrometric detection (LC-DAD-MS), after extraction of the dyes with formic acid and methanol. The focus of this work was yellow dyes, most of which are present as glycosides of flavonoids and related compounds, with the objective of identifying the plants originally used for dyeing. Two major types of dyes were found in this set of specimens. The first type is characterized by the presence of flavonol 3-O-sulfates (never before reported as being present in dyes) and 3-O-glycosides; this type was probably derived from the plant Flaveria haumanii or a close relative. The second type is characterized by the presence of both chalcone (heretofore not reported in pre-Columbian textiles) and luteolin glycosides, though a specific plant source could not be identified. Two other yellow dye types appeared to be present, but there were not enough examples to allow conclusions to be drawn. Also present in some extracts were various hydroxybenzoic acids, which appear to be oxidation products of the respective unsubstituted flavonol (3-hydroxyflavone) dyes. Most yellow dyes are synthesized in plants as glycosides (or other derivatives), which are incorporated more or less intact into textile fibers during dyeing. Extraction of these derivatives and analysis by LC-DAD-MS yields distinctive profiles that, with appropriate plant reference materials, can aid in the identification of the original plant dyestuffs. In many early cultures, including those of the Andes, textiles served more than utilitarian purposes, such as garments. For example, they also were used in rituals, for payment of tribute, and to serve as an indicator of the owner’s status in society; textiles * Corresponding author. E-mail: [email protected]. Tel: 617-353-2491. Fax: 617353-6466. † Boston University. ‡ Cotsen Institute of Archaeology at UCLA. § Universidad Nacional de Co ´ rdoba. (1) Barber, W. E. Women’s Work: The first 20,000 Years; Norton: New York, 1994. (2) Boytner, R., Ph.D. Dissertation, UCLA, Los Angeles, 1998. (3) Drooker, B. P.. Webster, L. D. Eds. Beyond Cloth and Cordage: Archaeological Textile Research in the Americas; University of Utah Press: Salt Lake City, UT, 2000. (4) Oakland-Rodman, A. Latin Am. Antiquity 1992, 3, 316-340. 10.1021/ac061618f CCC: $37.00 Published on Web 01/10/2007

© 2007 American Chemical Society

were often the most valuable objects an individual possessed.1-4 For this reason, much care went into the production of textiles, e.g., in terms of design, weave, and colorsthe latter being achieved by dyeing. Until the mid-19th century, the only sources of dyes for textiles and other fibers were natural materialssplants, mainly, but also some insects and animals. In general, all shades of color, except for browns and blacks, were made from combinations of the primary colors, blue, red, and yellow. In the Andes, as well as other parts of the world, there was one blue, indigo, that could be isolated from a number of plants; five or six reds, from plant and insect sources; and potentially hundreds of yellows, all from plants (for a more detailed discussion of dye sources, see Boytner5). Of the various classes of dye, the yellows offer the best opportunity for relating the dye to the plant that produced it and, consequently, relating the culture that produced the textile to its environment. Not only are there more potential yellow dye-yielding plants, but most yellow dyes are synthesized in plants as glycosides or other derivatives, which are incorporated more or less intact into the textile fibers. Furthermore, each plant produces a number of yellow-colored molecules, e.g., a “fingerprint,” so in principle the fingerprint of a textile dye extract can be matched with that of a particular plant and the dyestuff so identifieds provided that a sample of the plant extract is available for reference. In their pioneering work on Andean dyes, Wouters and Rosario-Chiniros6 analyzed textile dyes of all colors using highperformance liquid chromatography with diode array detection (HPLC-DAD). This work demonstrated both the high-resolution separations that can be obtained by HPLC (compared with thinlayer chromatography) and the utility of DAD for identifying chromophores based on their UV-visible absorption characteristics. (For more examples of the applications of HPLC-DAD to dye analysis, see Hofenk de Graaf.7) However, the identification of yellow dyes was limited in that the HCl extraction procedure used cleaved most of the glycosides, and thus, some of the identifying information was lost. Recently, Zhang and Laursen8 developed extraction procedures that preserve glycosidic linkages and combined HPLC-DAD with (5) Boytner, R. In Andean Textile Traditions; Young-Sa´nchez, M., Simpson, F. W., Eds.; Denver Art Museum, Denver, CO, 2006; pp 43-74. (6) Wouters, J.; Rosario-Chirinos, N. J. Am. Inst. Conserv. 1992, 31, 237-255. (7) Hofenk de Graaf, J.H. The Colourful Past: Origins, Chemistry and Identification of Natural Dyestuffs; Archetype Publications, London, 2004.

Analytical Chemistry, Vol. 79, No. 4, February 15, 2007 1575

mass spectrometric (MS) detection, which particularly facilitates the identification of flavonoid (and other) glycosides. In this report, we describe the application of HPLC-DAD-MS to the analysis of yellow dyes from ancient Andean textiles. This work has revealed two previously unsuspected categories of yellow dye and the probable plant source of one of them. MATERIALS AND METHODS Source of Analytical Samples. Textile samples were collected by Boytner from various archaeological sites and museum collections in Peru. Specimens of Flaveria haumanii Dim. et Orf. (Asteraceae), Cotinus coggygria Scop. (Anacardiaceae), and Coreopsis spp. (Asteraceae) were collected locally in Argentina and in the Boston area. Extraction of Dyes from Textile. Dyes were extracted from wool yarns or threads (estimated to weigh 0.2-0.8 mg in most cases) using formic acid/methanol (5:95, v/v) or, for comparison, HCl/methanol/ water (2:1:1, v/v) as described by Zhang and Laursen.8 The extracts, including the textile fibers, were dried under vacuum over NaOH pellets. The residues were suspended in 50 µL of methanol/water (1:1, v/v) and then centrifuged to separate particulate matter. The upper 30 µL of solution was removed with a pipettor for HPLC-DAD-MS analysis (20 µL was injected). Analysis of Extracts. Analysis of either textile or plant extracts was performed with an Agilent 1100 liquid chromatography system (HPLC) consisting of an automatic injector, a gradient pump, a HP series 1100 DAD, and an Agilent series 1100 on-line atmospheric pressure electrospray ionization MS. Dye components were separated on a 2.1-mm-diameter Vydac C4 reversed-phase column as previously described.8 Mass spectra were usually obtained in the negative ion mode, which gives the greatest sensitivity, and records [M - 1]- ions. In some cases, positive ion spectra were obtained. Positive ions, [M + 1]+, tend to produce fragments resulting from sequential loss of sugar units, which aids in structure elucidation. RESULTS Since the focus of this work was the analysis of yellow dyes, most of the samples examined had a yellow appearance, though a few specimens of other colors were analyzed, as well (Table 1). Not surprisingly, a green sample (#2b) was found to have been dyed with blue (indigo) and a yellow, while blue specimen 3 contained only indigo. In addition, one of the reds (#2a) contained both red (purpurin) and yellow dyes, while another (#25) contained only carminic acid. In total, 32 specimens were analyzed. However, several gave signals too weak to be interpreted, so they are not included in Table 1. The 23 yellow dye-containing samples analyzed here were collected from 8 different sites, located primarily on the North Coast of Peru, and represent five cultures, primarily Lambayeque and Chimu, The exceptions are the samples #2a-c from Algodonal on the South Coast of Peru.9 Though none of the textiles was dated directly, stylistic comparisons indicate that all were from the Late Intermediate Period (ca. 1050-1200 AD)swith the exception of sample #3, which was much older. However, since (8) Zhang, X.; Laursen, R. A. Anal. Chem. 2005, 77, 2022-2025. (9) Wallert, A.; Boytner, R. J. Archaeol. Sci. 1996, 23, 853-861.

1576

Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

#3 did not contain a yellow dye (only indigo), it was less relevant to this study. All samples came from decorated textiles, but only one could be identified, as the rest were fragments. The sample from Cerro Huarango was a small fabric square decorated at its edges with dyed bands having geometric designs. Such squares are known throughout the Andes to be associated with funerary practices and are typically placed on the head of the deceased before burial. Inasmuch as one purpose of this work was to test the analytical method using archaeological specimens that had been at least partially characterized in terms of age and site of discovery, we analyzed relatively few samples. Furthermore, by examining several samples from a particular localesrather than one or two from each of many widespread sitesswe obtain better statistical results. As will be shown, the analysis of a single sample can give ambiguous results as to which type of plant may have been used to dye the fibers. However, analysis of several samples can lead to patterns that lead to more solid conclusions. Analysis of Textile Fibers. Samples of textile fiber yarns (usually 0.2-0.8 mg) were extracted in with 5% formic acid in methanol as described by Zhang and Laursen.8 If sufficient sample was available, which was usually not the case, a second portion of the same fiber sample was extracted using the HCl method. Otherwise, the residue from formic acid extraction was extracted with HCl. Most yellow dyes, both in plant material and in dyed textiles, are glycosides. The formic acid procedure preserves the glycosidic linkages, whereas HCl extraction hydrolyzes them. Both extraction procedures give valuable information as to the nature of the dye components. Analysis by HPLC-DAD-MS permits dye molecules to be characterized by four criteria: HPLC retention time, UV-visible spectrum, relative abundance, and molecular mass. Presenting HPLC-DAD-MS data concisely is difficult. A particular extract may contain up to a dozen or more peaks, and some peaks may contain more than one molecular component. Furthermore, each molecular component can be described by HPLC retention time, relative abundance, UV-visible absorption properties (λmax), and molecular mass. In general, relative abundance is highly variable because it may depend on characteristics of the plant dyestuff (species, where grown, when harvested, etc.), the dyeing process used, the type of fiber (wool, silk) dyed, and the procedure used to extract the dye from the fibers. Retention time is also somewhat variable and may change over a period of time from column to column and depending on HPLC pump performance. However, when reversed-phase HPLC columns are used, the order of elution is generally the same, at least when mixtures of the same eluting solvents (e.g., acetonitrile and water) are used. Table 1 summarizes the analytical results for textile fibers containing yellow dyes. Each extract was separated on a reversedphase HPLC column and the eluate was monitored at 350 nm, which is close to the maximum absorbance for most yellow dyes. Then, UV-visible and mass spectra for each peak on the resulting profile were obtainedsif the peak was sufficiently large to obtain data. In Table 1, each observed peak having a particular combination of (λmax) and molecular mass [M] values is indicated by a circle (O). Stronger peaks are indicated with two or three circles. Within a particular dye type (e.g., chalcone or flavonol) the HPLC elution (retention) time increases from left to right. The higher

Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

1577

Table 1. Yellow Dye Component Patterns for Andean Textile Samples and for Selected Plant Dyestuffs

mass glycosides tend to elute earlier (left side of table) than the aglycones (right side). Also included for comparison, at the bottom of Table 1, are dye components found in the extracts of selected plants. Unexpectedly, most of the extracts in this study seemed to fall into two categories, which we term loosely as “chalcone”-type and “flavonol”-type dyes, even though some other dye types are included in these categories. Some representatives of various dye classes are shown in Figure 1. Many other examples are listed by Cardon.10 Chalcone Dyes. The chalcone-type dyes, which are indicated generally in Table 1, and with a specific example in Figure 2, are characterized by the presence of components with λmax ∼380 nm, one of which has a relatively strong peak with M ) 450 Da. This corresponds to the hexoside of pentahydroxychalcone (e.g., okanine 4′-glucoside or an isomer; see Figure 1). Also present in some extracts (cf. Figure 2) were chalcones with M ) 434 Da (butein glucoside, or an isomer) and M ) 612 Da (okanine diglucoside or an isomer). A second marker in many of these samples was a peak with λmax ∼342 nm and M ) 270 Da. We have tentatively identified this peak as 5-deoxyluteolin (7,3′,4′trihydroxyflavone) by comparison of its spectral properties and retention time with those of an authetic reference sample. By the same criteria, we were also able to eliminate apigenin, 5-deoxykaempferol, and sulfuretin, all of which have M ) 270 Da, as possibilities. However, the major peak in all of the chalcone-type extracts had λmax ∼348 nm and M ) 448, which, based on these properties and retention time, is probably the flavone, luteolin 7-glucoside. This compound is one of the most common of yellow dyes and occurs in many species of plant dyestuff. Acid hydrolysis also yields the aglycone, luteolin. Luteolin 7-glucoside itself is not distinctive but paired with the 450-Da chalcone glycoside does serve as a marker for for what we refer to as chalcone dyestuffs. Extraction of sample #16 by heating with HCl gives a very simplified HPLC profile (Figure 2), showing only luteolin (the largest peak) sulfuretin, and the putative 5-deoxyluteolin. It is of interest to note that no chalcone aglycones are seen. We do not yet have an explanation for their absence, but unpublished experiments with reference compounds suggests that they are stable to the conditions of HCl extraction. We have not yet been able to identify a specific plant that may have been used to produce the chalcone-type dyes, but we have included in Table 1 some data for an extract of a yet-unidentified species of Coreopsis, which contains predominantly marein, the 4′-glucoside of okanine (Figure 1), as well as small amounts of the glucoside of butein and other components.11 Our Coreopsis sample contained no detectable luteolin glycoside or deoxyluteolin, however. A detailed analysis of the coloring principals of Coresopsis tinctoria by Shimokoriyama12 also showed the presence of marein, primarily, but no luteolin (or any other flavone). Flavonol Dyes. Flavonols generally refer to 3-hydroxyflavones (see Figure 1), of which the most common dye members are derivatives of quercetin. However, the most distinctive feature of the flavonol-type dyes listed in Table 1 is the presence in many of (10) Cardon, D. Le monde des teintures naturelles: Belin: Paris, 2003; pp 525531. (11) Cardon, D. Le monde des teintures naturelles: Belin: Paris, 2003; pp 192 and 530. (12) Shimokoriyama, M. J. Am. Chem. Soc. 1957, 79, 214-220.

1578 Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

Figure 1. Structures of some typical yellow plant dye types. Usually these compounds are glycosylated, most frequently on the hydroxy groups shown in boldface type. The most common glycosyl moiety is glucose, which results in addition of 162 Da to the molecular mass. The UV-visible absorption maximums (λmax) of these compounds tend to be more or less constant within a class and are useful for characterizing the type of dye.

the samples of quercetin 3-sulfate and isorhamnetin 3-sulfate (M ) 382 and 396, respectively), since sulfates have never before been reported as occurring in any natural dye. All of the aglycones in these dyes (quercetin, isorhamnetin, and kaempferol) are 3-hydroxyflavones and all seem to have glycosyl or sulfate groups on the 3-hydroxyl group, as evidenced by λmax ∼352 nm, which is typical of 3-O-substituted flavones. By contrast, the corresponding aglycones, which are produced by acid hydrolysis of glycosides

Figure 2. HPLC profiles of extracts and spectroscopic data for a textile sample (#16) colored with a chalcone-type dye. (A) Formic acid extract showing glycosides; (B) HCl extracts showing aglycones only. The structures proposed in the table are based on UV-visible and mass spectral data.

or sulfate esters, all absorb at ∼372 nm. Glycosides of luteolin (a flavone; Figure 1), which lacks a 3-hydroxy group, have a λmax ∼348 nm which does not shift when the glycoside is hydrolyzed. Although flavonol sulfates have never been reported as dyes, such compounds have been known for some time,13,14 particularly in species of Flaveria. In 1999, Agnese et al.15 reported the presence of quercetin and isorhamnetin 3-sulfates in Flaveria bidentis var. angustifolia, which was subsequently renamed as a separate species, F. haumanii, in accordance with an earlier proposal by Dimitri and Orfila.16 Analysis of a F. haumanii specimen collected in Argentina revealed the presence of the same substituted 3-hydroxyflavone species seen in our archaeological samples (Table 1). In Figure 3, HCl (A) and formic acid (B) extracts of textile sample #6 are compared with an extract of F. haumanii (C). The same compounds are seen, albeit in different ratios. Also the plant extract contains a majority of aglycone, whereas the archaeological extract (B) has almost none. Although the HPLC profiles appear relatively simple, several of the peaks contain more than one compoundsspecifically pairs of flavonoids that differ by a mass of 30 Daswhich corresponds to replacement of a hydrogen atom (H) by a methoxy group (OCH3). We have noted (unpublished observations) that substitution of H by OCH3 does not lead to a change in retention time, although these pairs (13) Harborne, J. B., Mabry, T.J., Eds. The Flavonoids: Advances in Research; Chapman & Hall: London, 1982; pp 272-273. (14) Harborne, J. B., Ed. The Flavonoids: Advances in Research since 1980; Chapman & Hall: New York, 1988; pp 310-311. (15) Agnese, A. M.; Montoya, S. N.; Espinar, L. A.; Cabrera, J. L. Biochem. Syst. Ecol. 1999, 27, 739-742. (16) Dimitri, M. J.; Orfila, E. N. Tratado de Morfologı´a y Sistema´ tica Vegetal; ACME: Buenos Aires, 1985; pp 466-67.

probably can be resolved using a different or more efficient HPLC column than ours. Figure 4 relates the structures of these 3-hydroxyflavone structures. At present, the specific structures, 6-methoxyquercetin and its glucoside (Figure 4), must be regarded as tentative, but Barron et al.17 reported finding both patuletin (6-methoxyquercetin) and quercetin 3-sulfates in Flaveria chloraefolia. Although we cannot be completely certain which plants were used to dye the textile fibers discussed here, F. haumanii, or a related species seems like a good candidate. Oxidation of 3-Hydroxyflavone Dyes. We have noted in the present studies, and in others, that extracts of old or ancient textile fibers tend to contain relatively low amounts of aglycone compared with plant extracts. The absence of aglycones in the textile fibers could result from a dyeing process that excludes aglycones, to some process that selectively destroys the unsubstituted aglycones over time or to the fact that some plants, at least when fresh, contain only glycosides or other substituted aglycones. When certain chromatograms were monitored at 254 nm, profiles were obtained that showed early eluting peaks (Figure 5A) with masses corresponding to various hydroxybenzoic acids that could arise from oxidation of the C2-C3 bond in the C-ring of 3-hydroxyflavones (see Figures 1 and 4). Indeed, Ferriera et al.18 and Matsuura et al.19 have demonstrated that such products can arise by photooxidation of 3-hydroxyflavones provided that the 3-hydroxy (17) Barron, D.; Colebrook, L. D.; Ibrahim, R. K. Phytochemistry 1986, 25, 17191721. (18) Ferreira, E. S. B.; Quye, A.; McNab, H.; Hulme, A. N. Dyes Hist. Archaeol. 2002, 18, 63-72. (19) Matsuura, T.; Matsushima, H.; Nakashima, R. Tetrahedron 1970, 26, 435443.

Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

1579

Figure 3. HPLC profiles and spectroscopic data for flavonol sulfate-containing extracts. (A) HCl extract of sample 6, showing aglycones; (B) formic acid extract of sample 6, showing flavonol glycosides and sulfates; (C) formic extract of silk dyed with F. haumanii. The table lists data for each peak and the tentative assignments of structure. Note that several peaks contain two components which differ in mass by a 30-Da methoxy group (see discussion in text).

Figure 4. Interpretation of the data in Figure 3. The molecular mass of each compound is shown. UV-visible spectra indicate that all the flavonols are substituted on the 3-OH group. The compound identified as 6-methoxyquercetin (quercetagetin) may be an isomer. The hydroxybenzoic acids are those observed in Figure 5.

group is unsubstituted. Although photooxidation seems unlikely in textiles that have been buried for hundreds of years, some Andean textiles have become uncovered as a result of looting or soil erosion, and thus, photooxidation may be possible. The idea that the hydroxybenzoic acids arise from 3-hydroxyflavonoids is strengthened by the observation that they are not observed in samples (e.g., #16; see Figure 5B) dyed with dyestuffs that do not contain flavonols. Furthermore, the hydroxybenzoic acids must have been produced after dyeing; otherwise they would have been washed away during the dyeing process. 1580

Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

Other Dyes. As indicated in Table 1, only two other types of yellow dyes were observed. Sample #2a, which was red, contained purpurin, indicating that it was dyed with a species of Relbunium,6,9 but it also contained a yellow that was distinctly different from the chalcone- and flavonol-type dyes. We do not have enough samples to make generalizations, but this dye appeared to contain sulfuretin and fisetin, which produce a deeper yellow than the other types. These compounds are also found in the heartwood of C. coggygria (see Table 1), which has not been reported as occurring in Peru, but they may be present in other plants. The

Figure 5. HPLC profiles, produced by monitoring at 254 nm, of formic acid extracts of textile samples showing (A) the presence of hydroxybenzoic acids in a sample (6) colored with flavonol-type dyes (cf. Figure 3) and (B) their absence in a sample (16) dyed with a chalcone-type dye (a few, so far unidentified peaks are seen, however). The peaks eluting after 25 min are flavonoids or chalcones, which do not absorb strongly at 254 nm.

second “other” yellow was found in sample #8, which, based on the UV-visible absorbance data (λmax 348 nm) indicates the presence primarily of flavone-type dyes, e.g., derivatives of luteolin. The components with M ) 462 and M ) 476 suggest glucosides of luteolin methyl ethers (see Discussion below). Such compounds have been reported6 in species of Baccharis and Bidens, which are reported20,21 to have been use as dyestuffs in Peru. Besides the yellows, we found carminic acid in samples #1 and #25 and indigo in #2b and #3. DISCUSSION Natural dyes are synthesized and are generally found in plants as glycosides or other conjugates (e.g., sulfates, as shown in this work). Contrary to common belief, most of these dyes (yellows, at least) are not hydrolyzed either during the dyeing process or during extraction for analysissas long as sufficiently mild conditions are used. In fact, many of the old to ancient (50-3000 years) yellow samples we have analyzed contain only glycosides. Blue and red dyes may be another matter, however. The production of indigo, by fermentation of plant dyestuffs, requires hydrolysis of the glycosidic precursor.22 Similarly, the production of red anthraquinone dyes (e.g., from madder) generally entails a (20) Antu´nez, de Mayolo, K. K. Econ. Bot. 1989, 43, 181-189. (21) Zumbu ¨ hl, H. Tintes naturals para lana de oveja; Kamaq Maki: Huancayo, Peru, 1979. (22) Balfour-Paul, J. Indigo; British Museum Press: London, 1998; pp 89-113, 234.

fermentation process that hydrolyzes glycosides.23 Nevertheless, we have found anthraquinone glycosides as the primary red dye in some contemporary carpet fibers (unpublished observations). In the present studies, we have identified at least two types of yellow dye used in ancient times in the Andean region. One type is characterized by the presence of (A) chalcone glycosides, (B) a large amount of luteolin 7-glucoside, and (C) 5-deoxyluteolin and the absence, essentially, of any flavonol (3-hydroxyflavone). The second is characterized by the presence exclusively of flavonol sulfates and glycosides (see Table 1). It seems plausible that our samples were dyed using F. haumanii or a related plant. (Note that F. haumanii is a relatively new designation;15,16 in older literature, F. haumanii was named F. bidentis, even when it was listed systematically as F. bidentis var. angustifolia, a variety of F. bidentis var. bidentis () Flaveria contrayerba and other synonyms).] F. “bidentis” (i.e., F. bidentis or F. haumanii) has been reported to grow near the Lambayeque region where most of our samples originate,23,24 and it has also been reported as having been used for many years used as a source of yellow dye by indigenous peoples in Argentina and neighboring countries.26-28 Schweppe29 made reference to F. contrayerba as having been used as a dyestuff in Chile. Thus, although Flaveria species are not among the Andean dyestuff plants listed by Antu´nez de Mayolo20 or Roquero,21 there seems to be good reason to believe that many of the textile specimens analyzed here were dyed with these plants. Two other types of yellow dye were also detected, but we do not have sufficient data to draw conclusions as to their origin. One of these, sample #8, seems to be composed primarily of flavone glycosides, primarily glucosides of luteolin, as well as those of methylluteolin (chryseriol?) and possibly dimethylluteolin. These glycosides, as well as the corresponding aglycones produced by acid hydrolysis with HCl, may account for some of the late eluting “luteolin-like” peaks reported by Wouters and RosarioChiniros.6 More than a dozen mono- and polymethylated luteolins are known,10 and all of these could be predicted to elute later than luteolin itself from reversed-phase HPLC columns.31 Many of these are found in species of Baccaris and Bidens and may account for the post-luteolin peaks observed by Wouters and Rosario-Chiniros.6 The second “other” yellow dye was found only in a single, apparently red sample, and appears to contain only sulfuretin and fisetin, the principle coloring agents of C. coggygria. This plant has not been reported as having been used as a dyestuff in South America, but the same components may occur in other, so-far unidentified plants. (23) Derksen, G. C. H.; Naayer, M.; van Beek, T. A.; Capelle, A.; Haaksman, I. K.; van Doren, H. A.; de Groot, A. Phytochem. Anal. 2003, 14, 137-144. (24) Powell, A. M. Ann. Missouri Botanical Gardens 1978, 65, 590-636. (25) Brako, L.; Zarucchi, J. L. Monogr. Syst. Botany Missouri Botanical Gardens 1993, 45, 134. (26) Marzocca, A. Index de Plantas Colorantes Tinto´ reas y Curtientes: Manual de las especies Argentinas; Academia Nacional de Agronomı´a y Veterinaria; Buenos Aires, 1993; p 73. (27) Leal, A. R. Deserta 1972, 3, 115. (28) Roig, F. A. Cuaderno Te´ cnico 1981, 3, 120. (29) Schweppe, H. Handbuch der Naturfarbstoffe; Nikol: Hamburg, 1993; pp 4748. (30) Roquero, A. In Actas I Jornada Internacional sobre Textiles Precolombinos; Solanilla, V. D., Ed.; Universidad Auto´noma de Barcelona: Barcelona, 2000; pp 29-56. (31) Greenham, J.; Harborne, J. B.; Williams, C. A. Phytochem. Anal. 2003, 14, 100-118.

Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

1581

It is interesting that the primary dye classes identified in our study and in that of Wouters and Rosario-Chirinos6 do not seem to overlap. This may partly reflect the fact that the specimens examined by Wouters and Rosario-Chiniros cover many more cultures and a broader span of time than ours do. Wouters and Rosario-Chiniros also did not report the presence of any chalcones. This may be because the HCl extraction method used in those studies destroys chalcones. In our own studies (cf. Figure 2), formic acid extraction of sample #16 gave peaks that seem clearly to be chalcone glycosides, whereas HCl extraction yielded no chalcone aglycones, in confirmation of the foregoing report.6 We do not have an explanation for this, particularly in light of our own observations (unpublished) that acid hydrolysis does not seem do destroy the okanine present in Coreopsis extracts. Perhaps it has to do with the much smaller quantities of dye material present in textile samples compared with plant extracts or the presence of other substances.

1582

Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

ACKNOWLEDGMENT The authors acknowledge Bruce Owen, Carol Mackey, Luz Ramirez Bryson, and Christopher Donnan, as well as the staff members of both the Trujillo University Archaeology Museum and the Ica Regional Museum, in Peru, for access to their collections. We also acknowledge David Scott, Kym Faul, Alek Dooley, and Charles Stanish from UCLA for their support of the Andean Dye Research Project. SUPPORTING INFORMATION AVAILABLE UV-visible spectra of several of the flavonoid and other dye components discussed. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review August 30, 2006. Accepted November 20, 2006. AC061618F