Threads of War: Scientific Analysis of the Dyes, Fibres and Mordants Used in the Production of Afghan War Rugs

Abstract

So-called ‘war rugs’ started being produced in Afghanistan after the Soviet invasion in 1979. These textiles have sparked debate as symbols of resilience and political commentary but also as controversial commodification of human suffering. However, their manufacture and materiality have not been studied so far. In the framework of the British Museum exhibition “War rugs: Afghanistan’s knotted history”, a scientific investigation was conducted on nine rugs from the collection. Approximately 65 samples were analysed by optical microscopy (OM), scanning electron microscopy coupled to energy dispersive X-ray spectroscopy (SEM-EDX) and high-pressure liquid chromatography coupled to diode array detector and tandem mass spectrometry (HPLC-DAD-MS/MS) to study the fibres, mordants and dyes used in the production of the rugs. Scanning X-ray fluorescence (MA-XRF) and multiband imaging (MBI) were also used directly on the rugs to map the distribution of specific mordants and dyes, respectively. The results revealed the intentional use of white or dark wool as the substrate for dyeing, to obtain specific colour shades. A wide range of synthetic dyes was detected, including Acid Orange 7, Acid Red 88, Basic Green 4, Acid Blue 92, Acid Black 1 and Direct Black 38 in the earlier rugs, whereas Direct Yellow 1, Direct Brown 1, Direct Yellow 12, Acid Green 25, Acid Blue 113 and Direct Blue 15 were identified in the later rugs. Some synthetic dyes remained unidentified. Additionally, natural dyes were used in three rugs. An emodin-based colourant, possibly obtained from dock or sorrel (Rumex spp.), was detected in two light brown areas. A berberine-based colourant consistent with barberry (Berberis spp.) was detected in a yellow area. These results represent the first scientific study of these objects and enable preliminary insights into the details of this complex craft that has evolved from centuries of carpet making in this area.

1. Introduction

The history of Afghanistan has been full of conflict. Geographically situated on major trade routes between China, India, Central Asia and Iran, and rich in natural resources, Afghanistan has often been targeted by forces seeking to extend their power in the region. Its borders have fluctuated across time as its territory was conquered and ruled by numerous empires and dynasties, which reflect in the rich diversity of Afghan carpet-weaving. The practice was traditionally carried out by women of nomadic groups, such as the Turkmen and Baluch, who used their own styles and motifs. They made knotted rugs with tufted piles as well as flatweave rugs, which were also made by groups such as the Hazara, Uzbek and Kyrgyz. Most knotted rugs from Afghanistan were weaved on horizontal looms, using the asymmetric Persian knot and including two weft threads between each row of knots [1].

The rise of the Russian and British empires in the 1700s and 1800s turned Afghanistan into a zone of political and economic rivalry. In December 1979, a few months after the coup-d’état led by the filo-American Mujahideen movement, the Soviet Union invaded Afghanistan. By the fall of the USSR in 1990, Afghanistan was left in chaos. Islamist groups, such as the Taliban and al-Qaeda, became increasingly powerful. Al-Qaeda’s involvement in the 9/11 attacks ultimately led to military intervention by the United States and its allies [2]. In this complex framework of war and conflict, weavers started introducing military imagery into their rugs. Birds, flowers and centuries-old symbols were replaced by helicopters, guns, tanks and artillery, creating the new artform of Afghan war rugs. However, the imagery of war rugs extends beyond representations of military equipment. Some motifs relate to specific local, ethnic or religious traditions. Others are drawn from the rich literary and cultural heritage that connects Afghanistan to Iran, Central Asia and Northern India [3]. Over the last few decades, vast numbers of Afghans have been displaced to refugee camps, both within Afghanistan or to neighbouring Iran or Pakistan. Here, weavers from various ethnic backgrounds have met, combining different rug-making aesthetics and adapting to workshop conditions, resulting in a general homogenisation of traditional motifs and styles in these rugs [3].

Nevertheless, stylistic differences often enable war rugs to be broadly classified according to their production period. First generation rugs date to the Soviet–Afghan War (1979–1989). Traditional imagery is mixed with war motifs discreetly woven into the design, including weapons and aircraft often hidden among other elements. In some cases, war motifs are more evident, especially Kalashnikov rifles, as symbols of the Mujahideen resistance and heavy Soviet weaponry (tanks, ships and helicopters organised in columns and rows). Second generation rugs are associated with the civil war and Taliban period (1990–2001). In these, traditional rug symmetry blurs and English inscriptions appear, emphasising the growing commercialisation aimed at Western markets. Themes include Soviet troops leaving Afghanistan and strong political satire. Third generation rugs correspond to post-9/11 attacks (2001 to present). Following the U.S. invasion of Afghanistan, war rugs began depicting the events of the attack on the Twin Towers, as well as U.S. flag and military hardware and slogans echoing Western perspectives [4].

Early war rugs appealed to soldiers fighting in Afghanistan and to humanitarian workers, journalists and diplomats, fuelling two main markets, aimed mostly at foreigners: small pieces that could be rolled into a soldier’s backpack, and larger, more complex rugs for international collectors. Traders and workshop owners quickly began direct production, providing a variety of templated designs and lowering the quality of materials and production methods. As a result, war rugs are often criticised as inauthentic and manufactured under exploitative conditions, although they continue to provide Afghan weavers with a source of income in dire times [5].

Despite their relative popularity as heritage objects, the materiality of war rugs has not hitherto been investigated, especially with attention to the changes in materials, such as dyes, fibres and mordants over time. Within this framework, nine rugs from the British Museum’s collection were scientifically studied using a combination of techniques. Optical microscopy (OM) was used to investigate the colour distribution in the threads [6,7]. Scanning electron microscopy coupled to energy dispersive X-ray spectroscopy (SEM-EDX) was used on samples to study the fibre morphology [8,9,10] and gather indications on the use of mordants [11]. Scanning X-ray fluorescence (MA-XRF) [12,13,14] and multiband imaging (MBI) [11] were adopted to map the distribution of chemical elements and specific dyes, respectively. High-pressure liquid chromatography coupled to diode array detector and tandem mass spectrometry (HPLC-DAD-MS/MS) is the state-of-the-art technique for dye identification [15,16,17,18,19,20]. This thorough investigation was carried out with the aim to highlight the connections between modern and traditional practice, shed light into material availability and market trade, and reveal the details of this complex craft that has evolved from centuries of carpet making in Afghanistan and beyond.

2. Materials and Methods

2.1. Afghan War Rugs

The rugs studied are presented in the following sections in order of production date. Their accession numbers are included, but for conciseness the rugs will be henceforth referred to as “RX” (X = last digits of the accession number). Samples are referred to as “RX_colour” (colour being the colour of the sample).

2.1.1. Accession Number 2010,6013.7—R7

This rug (Figure 1) represents a scene of an attack on a small town or village with a mountain range. The depiction includes houses, a mosque, trees, a river, a roadway and flowers, as well as jets with exploding bombs overhead. Explosions are depicted along the lower section of the scene as well. The borders have abstract and floral motifs. The combination of motifs and the inscriptions in Farsi probably place this rug as an early production (1980–1989). The colours are orange, bright orange, red, beige, brown, black, light blue and blue. Samples were taken for all colours.

2.1.2. Accession Number 2010,6013.10—R10

The main design features a building, possibly a mosque, with two tower minarets, a large vase with flowers, animals, a convoy of military vehicles and helicopters (Figure 2, left). Borders of multi-coloured flowers and geometrical patterns frame the design. The V-shaped pattern topping the building resembles the rams’ horns motif, which is symbol of prosperity and protects against evil spirits in Turkmen culture. The floral vase with symmetrical paired felines is often found in Baluch rugs and picks on decorative patterns in paintings, carpets and architecture from Iran, India and Central Asia. The combination of traditional motifs with a few images connected to war subtly inserted in the design probably places this rug as an early production (1980–1989). It has a beige ground, and the colours are orange, bright orange, red, light brown, brown, dark brown, aqua and blue. Samples were taken for all colours.

2.1.3. Accession Number 2010,6013.12—R12

The central image shows an Afghan soldier attacking a Russian soldier, surrounded by onlookers, helicopters and soldiers leading camels (Figure 2, centre). The Russians are depicted as horned demons (divs in Persian) in the style of traditional Persian miniature paintings of scenes from the Shahnameh (Book of Kings) depicting the hero, Rustam, slaying the evil White Div. The wide border around the carpet features a convoy of tanks with machine guns. Several inscriptions in Farsi are placed around the main scene. Some are reversed, suggesting that they were woven from a template. The references to traditional motifs suggest that this rug might be an early production (1980–1989). The colours are orange, red, dark red, yellow, brown-grey, dark brown and dark blue. Samples were taken for all colours.

2.1.4. Accession Number 2010,6013.13—R13

The complex iconography (Figure 2, right) includes a large tank and a four-barrel anti-aircraft gun in the central field flanked by tall poppy plants, probably referring to the opium trade to finance the war effort. The anti-aircraft gun bears similarity with hand-shaped standards displayed during processions to commemorate revered individuals. The corners of the central field include a grenade, a rocket launcher and other unidentified weapons. The wide border includes a train of tanks and other armoured vehicles. The inscriptions are in Farsi and Russian indicating that the rug is probably an early production (1980–1989). The colours are orange, red, dark red, beige, grey, brown, light blue and dark blue. Samples were taken for all colours.

2.1.5. Accession Number 2010,6013.14—R14

This almost-square rug features a large body of water with military boats surrounded by buildings, trees and a mountain range (Figure 3, top). Russian block houses are depicted on the mountain tops, and the sky is full of military helicopters and planes dropping star shells. The middle and widest decorative border includes a variety of tanks and armoured vehicles, whereas the inner and outer borders feature geometrical patterns. The inscriptions are in Farsi and the rug is probably an early production (1980–1989). The colours are orange, red, dark red, beige, brown, aqua, light blue and blue. All colours of the design were sampled.

2.1.6. Accession Number 2010,6013.15—R15

The rug features five niches and five buildings with minarets at the top (Figure 3, left), suggesting that it may have been used as a prayer carpet. The two large central motifs appear to be stylised vases with flowers with images of teapots above. Teapots and water ewers were often included in Central Asian embroideries to represent hospitality and generosity. The colours are orange, red, dark red, black, light blue and navy blue, which are typical of traditional Baluch rugs. However, the floral vases recall ikat patterns on Uzbek and Tajik textiles. Light and dark blue tanks frame the main design. The mixture of motifs suggests an early production of this rug (1980–1990). Samples were taken for all colours.

2.1.7. Accession Number 2010,6013.17—R17

Referred to as the “garden carpet” (Figure 3, right), this rug features a main design on a peach ground with a repeating abstract floral pattern. Blue helicopters are subtly incorporated in the corners, whereas a conventional pattern might have included birds. A small inscription in Farsi has been incorporated on one corner of the rug (possibly the weaver’s signature). The multiple blue floral borders frame the peach ground. A flatweave blue border with bright pink patterns is included along the weft edges. Reminiscent of Baluch “zakani” rugs, the textile is probably of early production (1980–1990). The colours are orange, bright pink, brown, olive green, green, light blue and blue. Samples were taken for all colours.

2.1.8. Accession Number 2010,6013.24—R24

The main design features a map of Afghanistan in green with rocket launchers and Kalashnikov machine guns depicted inside the map (Figure 4, left). These point at several armoured vehicles and helicopters shown leaving Afghanistan. These rugs are sometimes referred to as ‘exit rugs’, ‘protest rugs’ or ‘mujahideen rugs’, as they depict the Soviet forces leaving Afghanistan. The rug also includes a large poppy plant, a green flag on the right and a flowering shrub, which traditionally refers to happiness and well-being and may have been an expression of joy at the Soviet withdrawal. The lower third of the rug features various types of anti-aircraft guns, which are labelled in English script. Several bands of inscriptions in English and Farsi are spaced around the rug, describing the images. The inscription in the lowest part of the rug states that it was made in Afghanistan by a Turkmen weaver. This is probably a second-generation rug produced between 1989 and 2000. The colours include a cream background, as well as red, green, olive green, light blue, brown and black. Samples were taken for all colours.

2.1.9. Accession Number 2010,6013.27—R27

A map of Afghanistan is depicted in the centre in olive green with a scene of military action taking place against terrorists in the Tora Bora cave complex (situated in the White Mountains or Safed Koh of Eastern Afghanistan) in 2001 (Figure 4, right). Two jet planes with laser-guided missiles are shown bombing the caves where Osama bin Laden and the Taliban insurgents were thought to be based. The British and American flags, three fighter jets, several inscriptions in English and a date, probably indicating the production year as 2002, are also included. These rugs are sometimes referred to as ‘War on Terror rugs’; ‘Alliance rugs’; or ‘Tora Bora rugs’. The colours include a cream background, as well as orange, red, grey, olive green, light blue and dark blue. Samples were taken for all colours.

2.2. Instrumentation

2.2.1. Optical Microscopy (OM)

OM was performed after placing the samples on a microscope glass slide without any pre-treatment., A Leica DM 4000M microscope with Leica EL6000 light source (Leica Microsystems GmbH, Wetzlar, Germany) was used to image the fibres under visible and UV light, respectively. The images were processed by means of the Leica LAS X software (v. 4).

2.2.2. Scanning Electron Microscopy—Energy Dispersive X-Ray Spectroscopy (SEM-EDX)

Samples were placed, uncoated, on an adhesive carbon disc mounted onto an aluminium SEM stub. Examination was carried out with a variable pressure VP SEM S-3700 N (Hitachi, Hitachinaka, Japan) using the backscatter electron (BSE) detector, mostly at 16 kV. The working distance was 10 mm. The SEM chamber was only partially evacuated (mostly 40 Pa). The EDX spectra were collected using an AZtec EDX spectrometer (Oxford Instruments, High Wycombe, UK) with a 0 ± 20 keV spectral range, 150 s live time, and 2048 channels. AZtec-Energy analysis software (Oxford Instruments, v. 3.1) was used to process the data.

2.2.3. Multiband Imaging (MBI)

R12 and R17 were imaged using a modified Nikon D7000 camera (Nikon, Ayutthaya, Thailand) body with a maximum resolution of 4928 × 3264 (16.2 megapixels). The modification consists of the removal of the inbuilt UV-IR blocking filter, to exploit the full sensitivity of the CMOS sensor (c. 300–1000 nm). The lens used is a Canon EF 50 mm f/1.8 II. The camera is operated in fully manual mode. A reference grey scale, comprising a set of Lambertian black, grey and white tiles, is placed in the same plane as the object under investigation. In each case the object is illuminated by two radiation sources symmetrically positioned at approximately 45° with respect to the focal axis of the camera and at about the same height. A filter, or combination of filters, is placed in front of the camera lens in order to select the wavelength range of interest. Combinations of radiation sources and filter(s) enable six sets of images to be acquired: visible-reflected (VIS), ultraviolet-induced visible luminescence (UVL), infrared-reflected (IRR), ultraviolet-reflected (UVR), visible-induced visible luminescence (VIVL) and multiband-reflected (MBR) images. Details are reported in [7,21].

All images are acquired as RAW images and transformed into 4928 × 3264 pixel resolution images in 16-bit TIF (tagged image file) format, and by turning off all enhancements (e.g., recovery, fill light, blacks, contrast, brightness, clarity, vibrance, and saturation), as well as setting the tone-curve to linear. This procedure can be carried out using the camera software or external programs such as Adobe Photoshop. Post-processing procedures for the standardisation and calibration of the VIS, IRR, UVL and UVR images and the creation of infrared-reflected false colour (IRRFC) and ultraviolet-reflected false colour (UVRFC) images are then carried out using “BM_workspace”, a plug-in for Nip2, the open-source graphical user-interface of VIPS, a free image processing software. For details regarding the post-processing of these images, see the manual on multispectral imaging techniques [22].

2.2.4. Scanning X-Ray Fluorescence (MA-XRF)

Representative areas of R10, R12, R13, R15, R17, R24 and R27 were scanned with no preparation required to be carried out on the rugs. Elemental mapping was carried out using a M6 JETSTREAM large area micro-XRF scanner (Bruker, Berlin, Germany). The X-ray tube was set to 35 kV and 800 µA and the analysis was carried out in air. The maps were taken using a spot size of 100 µm and a pixel time of 25 ms/pixel, and two silicon drift detectors were used. The overall size of the map, pixel size, measurement time and stage speed varied between each map, and these details are listed in Table S1 (Supplementary Information).

2.2.5. High Pressure Liquid Chromatography Coupled with a Diode Array Detector and Tandem Mass Spectrometry (HPLC-DAD-MS/MS)

The dye extraction was performed using a method previously published [23], which briefly consists of a double mild extraction procedure, using dimethylsulphoxide (DMSO) first, and secondly a mixture of methanol/acetone/water/0.5 M oxalic acid 30:30:40:1 (v/v/v/v). The instrumentation consisted of a 1260 Infinity HPLC (Agilent Technologies, Waldbronn, Germany), coupled to a 1100 DAD detector (Agilent Technologies, Waldbronn, Germany) and to a Quadrupole-Time of Flight tandem mass spectrometer 6530 Infinity Q-ToF detector (Agilent Technologies, Waldbronn, Germany) equipped with a Jet Stream ESI interface (Agilent Technologies, Waldbronn, Germany). Separation was achieved using a Zorbax Extend-C18 column (2.1 mm × 50 mm, 1.8 μm particle size) with a 0.4 mL/min flow rate, a 40 °C column temperature, and a gradient of water with 0.1% formic acid (eluent A) and acetonitrile with 0.1% formic acid (eluent B). The elution gradient was programmed as follows: nitial conditions 95% A, followed by a linear gradient to 100% B over 10 min. Then, 100% B was held for 2 min. Re-equilibration time for each analysis was 10 min. A 5 μL injection volume was adopted for MS experiments and a 10 μL for MS/MS experiments. The DAD detector (cell volume 50 μL) scanned in the range 190–700 nm with 2 nm resolution. The ESI operating conditions were drying gas (N₂, purity > 98%) temperature 350 °C and 10 L/min flow; capillary voltage 4.0 kV; nebulizer gas pressure 40 psig; and sheath gas (N₂, purity > 98%) temperature 375 °C and flow 11 L/min. High resolution MS and MS/MS spectra were acquired in both negative and positive mode in the range 100–1700 m/z. The fragmentor was kept at 100 V, nozzle voltage 1000 V, skimmer 65 V, and octapole RF 750 V. For the MS/MS experiments, different voltages (from 10 to 50 V) in the collision cell were tested for Collision Induced Dissociation (CID) to maximize the information obtained from the fragmentation of the single molecules. The collision gas was N₂ (purity 99.999%). The data were collected by targeted MS/MS acquisition with an MS scan rate of 1.0 spectra/s and a MS/MS scan rate of 3.0 spectra/s. Auto-calibration was performed daily using Agilent tuning mix HP0321 prepared in 90% water and 10% acetonitrile. MassHunter Workstation Software (v.10) was used to carry out diode array detector and mass spectrometer control, data acquisition, and data analysis. Molecular identification was based on comparison with in-house molecular databases of natural [23] and synthetic dyes [24], as well as data available in the relevant literature (see corresponding sections). Calculations based on mass fragmentations were carried out in the cases where database matches were not obtained [15]. Additionally, complex MS/MS spectra were further processed using the SIRIUS software (v.6), which has been developed for the structure elucidation of novel molecules. It combines the analysis of isotope patterns in MS spectra with the analysis of fragmentation patterns in MS/MS spectra, and uses CSI:FingerID as a web service for searching molecular structure databases [25].

3. Results

3.1. Optical Microscopy (OM)

Micrographs of all samples are included in Table S2, showing a certain variability of colour distribution on the fibres.

Generally, light shades, such as yellow, orange, red and beige, show a relatively homogeneous distribution of colour, whereas dark shades, such as burgundy, brown, blue and black, appear to contain fibres of different colours in the same thread. These observations are summarised in Figure 5a–d, which shows the red, blue, brown-grey and dark red samples of R12. These samples were re-observed under the microscope after the dye extraction (Figure 5a’–d’). The red sample, which showed a bright homogeneous red dyeing before extraction (Figure 5a), almost completely lost its colouration (Figure 5a’), as expected after a successful dye extraction. By contrast, the other samples retained light and dark brown fibres. This is a clear indication that naturally pigmented wool was used to obtain these colours. The melanosomes (also called pigment granules), typically observed in brown wool [6], are visible in the images obtained after extraction. The use of naturally pigmented wool is consistent across all rugs, but selectively reserved to dark shades of colour, indicating that its use is intentional and functional to the obtainment of a precise colour. The use of white or dark/mixed wool is indicated in

Table 1. Summary of the results obtained from the observation and analysis of all samples taken from the rugs under investigation.

Colour	Fibres	UV-Microscopy	Dyes	Elements *
Accession number 2010,6013.7–R7
Orange	White wool	-	Acid Orange 7 (C.I. 15510)	S, Ca, Si, Al, Na, Cl
Bright Orange	White wool	-	Acid Orange 7 (C.I. 15510) Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl
Red	White wool	-	Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl
Beige	White wool	-	Nitrobenzenes and nitrophenols	S, Ca, Si, Al, Na, Cl
Brown	Brown wool	-	Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl, Mn
Black	Brown wool	-	Acid Orange 7 (C.I. 15510) Direct Green 1 (C.I. 30280) Direct Black 38 (C.I. 30235) Direct Black 38 degradation products Unidentified molecules	S, Ca, Si, Al, Na, Cl, Mn
Light blue	Brown wool	-	Basic Green 4 (C.I. 42000)	S, Ca, Si, Al, Na, Cl, Mn
Blue	Brown wool	-	Acid Blue 92 (C.I. 13390)	S, Ca, Si, Al, Na, Cl, Mn
Accession number 2010,6013.10–R10
Orange	White wool	-	Acid Orange 7 (C.I. 15510)	S, Ca, Si, Al, Na, Cl
Bright orange	White wool	-	Acid Orange 7 (C.I. 15510) Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl
Red	White wool	-	Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl
Light brown	Brown wool	-	Rumex spp.	S, Ca, Si, Al, Na, Cl
Brown	Brown wool	-	Acid Orange 7 (C.I. 15510) Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Mn, Cl
Dark brown	Brown wool	-	Acid Orange 7 (C.I. 15510) Acid Black 1 (C.I. 20470)	S, Ca, Si, Al, Na, Mn, Cl
Turquoise	White wool	-	Basic Green 4 (C.I. 42000)	S, Ca, Si, Al, Na, Mn, Cl
Blue	Brown wool	-	Acid Blue 92 (C.I. 13390)	S, Ca, Si, Al, Na, Mn, Cl
Accession number 2010,6013.12–R12
Yellow	White wool	Yellow luminescence	Berberis spp.	S, Ca, Si, Al, Na, Cl
Orange	White wool	-	Acid Orange 7 (C.I. 15510)	S, Ca, Si, Al, Na, Cl
Red	White wool	-	Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl
Dark red	Brown wool	-	Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl, Mn
Brown-grey	Brown wool	-	Acid Orange 7 (C.I. 15510) Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl, Mn
Dark brown	Brown wool	-	Acid Orange 7 (C.I. 15510) Acid Red 88 (C.I. 15620) Direct Green 1 (C.I. 30280) Direct Black 38 (C.I. 30235) Direct Black 38 degradation products Unidentified molecules	S, Ca, Si, Al, Na, Cl, Mn
Blue	Brown wool	-	Acid Orange 7 (C.I. 15510) Acid Red 88 (C.I. 15620) Acid Blue 92 (C.I. 13390)	S, Ca, Si, Al, Na, Cl, Mn
Accession number 2010,6013.13–R13
Orange	White wool	-	Acid Orange 7 (C.I. 15510)	S, Ca, Si, Al, Na, Cl
Red	White wool	-	Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl
Dark red	Brown wool	-	Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl, Mn
Beige	White wool	-	Rumex spp.	S, Ca, Si, Al, Na, Cl
Brown	Brown wool	-	Acid Orange 7 (C.I. 15510)	S, Ca, Si, Al, Na, Cl, Mn
Grey	Brown wool	-	Acid Orange 7 (C.I. 15510) Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl, Mn
Light blue	Brown wool	-	Acid Orange 7 (C.I. 15510) Acid Red 88 (C.I. 15620) Basic Green 4 (C.I. 42000)	S, Ca, Si, Al, Na, Cl, Mn
Dark blue	Brown wool	-	Acid Orange 7 (C.I. 15510) Acid Red 88 (C.I. 15620) Acid Blue 92 (C.I. 13390)	S, Ca, Si, Al, Na, Cl, Mn
Accession number 2010,6013.14–R14
Orange	White wool	-	Acid Orange 7 (C.I. 15510)	S, Ca, Si, Al, Na, Cl
Red	White wool	-	Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl
Dark red	White wool	-	Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl
Brown	Brown wool	-	Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl
Blue	Brown wool	-	Unidentified blue	S, Ca, Si, Al, Na, Cl
Light blue	White wool	-	Acid Orange 7 (C.I. 15510) Acid Red 88 (C.I. 15620) Basic Green 4 (C.I. 42000) Unidentified blue	S, Ca, Si, Al, Na, Cl
Accession number 2010,6013.15–R15
Orange	White wool	-	Acid Orange 7 (C.I. 15510)	S, Ca, Si, Al, Na, Cl
Red	White wool	-	Acid Orange 12 (C.I. 15970) Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl
Dark red	White wool	-	Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl
Black	Brown wool	-	Direct Black 38 (C.I. 30235) degradation products	S, Ca, Si, Al, Na, Cl
Light blue	Brown wool	-	Direct Blue 15 (C.I. 24400) maybe something else	S, Ca, Si, Al, Na, Cl
Blue	Brown wool	-	Acid Orange 7 (C.I. 15510) Acid Red 88 (C.I. 15620) Unidentified blue (same as R14)	S, Ca, Si, Al, Na, Cl
Accession number 2010,6013.17–R17
Orange	White wool	-	Acid Orange 7 (C.I. 15510)	S, Ca, Si, Al, Na, Cl
Pink	White wool	Bright orange luminescence	Basic Violet 10 (C.I. 45170)	S, Ca, Si, Al, Na, Cl
Brown	Brown wool	-	Acid Orange 7 (C.I. 15510) Acid Red 88 (C.I. 15620) Sulphonated Fast Red	S, Ca, Si, Al, Na, Cl, Mn
Olive green	Brown wool	-	Acid Orange 7 (C.I. 15510) Acid Red 88 (C.I. 15620) Basic Violet 10 (C.I. 45170) Sulphonated Fast Red	S, Ca, Si, Al, Na, Cl, Mn
Green	Brown wool	-	Acid Blue 7 (C.I. 42080)	S, Ca, Si, Al, Na, Cl, Mn
Light blue	Brown wool	-	Acid Orange 7 (C.I. 15510) Acid Blue 83 (C.I. 42660)	S, Ca, Si, Al, Na, Cl, Mn
Blue	Brown wool	-	Acid Orange 7 (C.I. 15510) Acid Red 88 (C.I. 15620) Acid Blue 7 (C.I. 42080) Acid Blue 260	S, Ca, Si, Al, Na, Cl, Mn
Accession number 2010,6013.24–R24
Yellow	White wool	-	Direct Yellow 1 (C.I. 22250) degradation product or synthesis by-product (317.0926)	S, Ca, Si, Al, Na, Cl
Olive green	White wool	-	Direct Yellow 1 (C.I. 22250) degradation product or synthesis by-product (317.0926) Direct Brown 1 (C.I. 30045) Tentatively Mordant Orange 6 (C.I. 26520)	S, Ca, Si, Al, Na, Cl
Red	White wool	-	Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl
Brown	White wool	-	[M-H]− = m/z 409.034 794.0769 734.0276 874.9910 (2 isomers)	S, Ca, Si, Al, Na, Cl, Co, Cr
Green	White wool	-	Direct Yellow 12 (C.I. 24895) Acid Green 25 (C.I. 61570) and degradation products or synthesis by-product	S, Ca, Si, Al, Na, Cl
Light blue	Brown wool	-	[M-H]− = m/z 876.1070	S, Ca, Si, Al, Na, Cl, Mn
Black	Brown wool	-	Acid Blue 113 (C.I. 26360) Acid Orange 7 (C.I. 15510) Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl, Mn
Accession number 2010,6013.27–R27
Orange	White wool	-	Acid Red 88 (C.I. 15620) Direct Yellow 1 (C.I. 22250) and degradation product or synthesis by-product Direct Brown 1 (C.I. 30045)	S, Ca, Si, Al, Na, Cl
Red	Brown wool	-	Acid Red 88 (C.I. 15620)	S, Ca, Si, Al, Na, Cl
Grey	Brown wool	-	Unidentified	S, Ca, Si, Al, Na, Cl
Olive green	Brown wool	-	Direct Yellow 1 (C.I. 22250) and degradation product or synthesis by-product Direct Brown 1 (C.I. 30045)	S, Ca, Si, Al, Na, Cl, Mn, Fe
Light blue	Brown wool	-	Direct Blue 15 (C.I. 24400) degradation products or synthesis by-products	S, Ca, Si, Al, Na, Cl
Dark blue	Brown wool	-	Direct Blue 15 (C.I. 24400) degradation products or synthesis by-products Acid Red 88 (C.I. 15620) Acid Orange 7 (C.I. 15510)	S, Ca, Si, Al, Na, Cl, Mn, Fe

* Results obtained from both EDX and XRF. In bold are the most abundant elements.

for all samples/colours.

When observed under UV light, R12_yellow and R17_pink exhibit UV-induced visible luminescence (Figure 6). The luminescence produced by the yellow samples is not particularly useful to narrow down the type of dye used, considering that numerous natural and synthetic yellow dyes exhibit this characteristic [21]. By contrast, the bright orange luminescence produced by the dark pink sample of R17 may be taken as indication that a synthetic xanthene dye, probably of the rhodamine family, was used for this colour [11].

3.2. Multiband Imaging (MBI)

In the light of the results obtained by optical microscopy and with the aim to better study the distribution of materials and their behaviour when exposed to different light wavelengths, R12 and R17 were investigated using MBI.

Infrared reflected images (IRR) highlight areas of slight absorption of the infrared radiation. These areas appear dark, as exemplified in Figure 7a’, while other areas are infrared transparent, thus appearing white. This was in correspondence with the use of naturally pigmented wool versus white wool. Melanin is known to strongly absorb electromagnetic radiation particularly in the UV–Vis range, with absorption gradually declining in the near-IR [26,27]. However, in IRR images, where the absorbance of most dyes is low, and hence does not interfere with the response of the wool substrate, sufficient absorbance from the melanin is observed, resulting in a distribution map of where naturally pigmented wool is used.

UV luminescent dyes were also confirmed. However, only a weak yellowish luminescence is observed in the UVL images of R12, corresponding to the outline of the soldier in the left corner and the rounded decorations of the camel saddle, suggesting that natural ageing may have attenuated the response of the luminescent dye over time.

By contrast, a strong orange luminescence is produced by all bright pink areas of R17, suggesting that the same luminescent dye, likely a synthetic xanthene dye of the rhodamine family [11], was used for this colour (Figure 7b’).

3.3. Scanning Electron Microscopy and Energy Dispersive X-Ray Spectroscopy (SEM-EDX)

The use of wool as sole fibre used in the rugs was confirmed by SEM. Images clearly show the typical surface scale pattern of wool (Figure 8). Fibres of various diameters are observed in most samples, ranging from ca. 5 to 50 μm in width. Such variability is common in sheep fleece and does not point directly to a specific type of breed being used [9,28].

EDX was used to explore the possible presence of mordants. However, no clear indication was obtained for any of the samples. Generally, S, Ca, Si, Al, Na and Cl were detected in the EDX spectra. These elements are quite ubiquitous in wool textiles, with S partially deriving from the wool itself and the other elements being part of environmental contamination. NaCl and CaCO₃ are sometimes added to dye baths [29], but their presence as a result of intentional use cannot be proved based solely on these results. An exception was represented by sample R13_beige. Although the fibres are particularly contaminated, areas clear from particles retained relatively high levels of Al, possibly pointing toward alum being used as a mordant [11]. Sample R10_light brown also showed slightly higher Al levels compared to the other samples.

3.4. Scanning X-Ray Fluorescence (MA-XRF)

To further investigate the presence of chemical elements and their distribution, representative areas of R10, R12, R13, R15, R17, R24 and R27 were scanned. In all cases the maps of Ca and Mn appear to correlate with dark colours, such as dark brown and dark blue, whereas the elements are significantly less abundant in light or bright colours, as shown in Figure 9 and Figures S1–S4 (Supplementary Information). Based on optical microscopy, dark colours are associated with the use of naturally pigmented wool. Ca and Mn are reported as elements associated with melanin in dark wool [30]. Therefore, the results can be interpreted as an indication of the use of dark as opposed to white wool. Nevertheless, other elements, such as Cu and Zn, seem to be associated with melanin [30], but they were not significantly detected in our analyses. Thus, the presence of Ca and Mn in association with dyeing practices cannot be ruled out and a conclusive interpretation cannot be made based solely on these results.

The maps obtained for R24 revealed the clear presence of Co and Cr in association with the brown dye (Figure 10). These metals are known for being used in so-called metal complex dyes or pre-metallised dyes [31,32]. These dyes are usually of the azo acid family and are structured so that a central metal ion, such as Cr³⁺ or Co³⁺, forms a co-ordination complex with one or two organic dye molecules. They started being produced in 1912 and were popular in the 1950s, when they were commonly used for dyeing wool, silk, and nylon to produce colourfast shades [33]. The maps obtained for R27 reveal some differences in the distribution of Ca and Fe, which are less straightforward to interpret (Figure S4, Supplementary Information), but seem to be again associated with the darkest colours.

It is noteworthy that EDX analysis could not detect significant differences among the samples in terms of elemental content. Co and Cr were also not detected in the R24_brown. The difficulties of detecting mordants in dyed fibres using EDX have been partially addressed in the literature [11,13,34], as well as the need to often use both EDX and XRF to obtain more accurate results [13]. This study reiterates this important point, highlighting that, while EDX can be better exploited to suggest the presence of light elements such as Al, other heavier elements, such as Fe, Cu, Co, Cr, etc, are often better detected by XRF. This is partially related to intrinsic limitations of the techniques, e.g., EDX being sensitive to the surface of the sample and not the bulk. Thus, the detection of mordants and metals in textile fibres remains an area that requires careful and well-controlled experimental investigation.

3.5. High Pressure Liquid Chromatography Coupled with a Diode Array Detector and Tandem Mass Spectrometry (HPLC-DAD-MS/MS)

The results of the dye analyses are summarised in Table 1. For all dyes identified or discussed, the molecular details are reported in Table S3 (Supplementary Information).

3.5.1. Natural Dyes

Out of all samples analysed, three revealed the use of natural dyes. The yellow areas of R12 were obtained using a protoberberine dye rich in jatrorrhizine, berberine and palmatine (Figure 11a). Various alkaloid dyes of this family are available throughout Asia. These include Phellodendron, Berberis, Coptis, Mahonia and Fibraurea species [35,36]. Distinguishing among these genera is not always straightforward. However, the molecule coptisine is found in Coptis spp. [23], while Fibraurea spp. do not contain berberine [37], thus excluding these sources. The relatively high content of jatrorrhizine also seems to exclude Phellodendron spp., which are richer in berberine [23,37,38,39]. Mahonia has recently been reclassified under the Berberis genus [40]. In fact, their dye composition is also reported to be very similar [37]. The results constitute a good match with data reported on Berberis asiatica, aristata, vulgaris and thunbergii, all showing relative high content of jatrorrhizine compared to berberine and palmatine [23,37,38,39]. Although a precise species identification is not possible and unpredictable changes in compound ratios due to ageing and plant variability should be considered, Berberis spp., commonly referred to as barberry, emerges as the most likely source of this yellow dye. Barberry is reported as a local plant in Central Asia [23,41]. Protoberberine dyes exhibit bright yellow UV-induced luminescence [21], in agreement with the UV-microscopy and MBI observations.

The samples taken from the light brown areas of R10 and the beige background of R13 revealed emodin and emodin-8-O-glucoside as the main components, as well as nepodin-1-O-glucoside, chrysophanol-8-O-glucoside and physcion-8-O-glucoside (mass fragmentation spectra are reported in Figure S5, Supplementary). Emodin is a relatively common anthraquinone found in various plants [42] and fungi [43] that is used for dyeing. In the context of Asian dyes, the most common sources of emodin-containing dyes discussed in the literature are Rheum (rhubarb) and Rumex (dock or sorrel) plants [44,45,46,47]. Some Rhamnus plants (buckthorn) and Frangula alnus (alder buckthorn) are also mentioned as sources of emodin-rich colourants [44,48]. However, in the case of Rhamnus spp., various flavonoids (e.g., rhamnetin glycosides) are present; in the case of Frangula alnus, various emodin glycosides (e.g., frangulin A and B) are present [49].

By contrast, Rheum and Rumex plants are reported to contain other anthraquinones, such as chrysophanol and physcion, as well as their glycosides [23,46,50], in agreement with our results. Nevertheless, the composition of rhubarb extracts is generally more complex, including other molecules, such as stilbenes [23], whereas good correspondence with reference samples from Rumex spp. is observed [51]. In particular, nepodin-1-O-glucoside was identified in our samples. This molecule has been specifically reported as a component of some Rumex species [52]. As various species of Rumex are native or available in Central Asia [53], this plant emerges as the most probable source of the dye.

3.5.2. Synthetic Dyes

The two main orange and red colourants in all rugs are identified as Acid Orange 7 (C.I. 15510) and Acid Red 88 (C.I. 15620), as shown in Figure 12. The two dyes are often combined to obtain bright orange colourations. Dark red and brown shades are often obtained by using one or both dyes on naturally brown wool, as summarised in Table 1.

In addition to orange, red and brown shades, most rugs feature light and dark blue colours, as well as black. The dyes used to obtain these colours in R7, R10, R12 and R13 are similar. The light blue/turquoise shades in these rugs, as well as R14, contain Basic Green 4 (C.I. 42000) as the main colourant. The dye is identified thanks to its distinctive molecular pattern (Figure S6, Supplementary Information), and the presence of the leuco-components and degradation products is observed, as often reported in the literature [11,24,54]. If applied on white wool, as in R10 and R14, the colour appears as a light aqua, whereas on naturally-brown wool, as in R10 and R13, it appears as a light blue. In R13 and R14, the dye is also mixed with orange and red in the former and with an unidentified blue colourant in the latter. These mixtures need to be observed with caution, as cross-sample contamination is particularly easy in carpets and rugs due to their use. Nevertheless, the intentional use of red/orange to darken the blue colour is a possibility.

The dark blue shades in R7, R10, R12 and R13 contain Acid Blue 92 (C.I. 13390) as the main colourant applied on naturally brown wool. The identification was challenging. As shown in Figure 13, the analyses revealed two main components. The main one produced a doubly charged ion [M-2H]²⁻ = m/z 313.5044 and a very weak singly charged ion [M-H]⁻ = m/z 628.0142, enabling the molecular formula to be assigned to C₂₆H₁₉N₃O₁₀S₃ (dppm = −2.83). The study of the MS/MS fragmentation enabled some key fragment ions to be identified, ultimately allowing the structure of the molecule to be elucidated, as shown in Figure 13. The second molecule was also identified as a fragment of the main molecule, most likely a degradation product or synthesis by-product, further strengthening the identification. However, a standard was not available to corroborate the identification. This is a common scenario in the analysis of historic synthetic dyes. However, high-resolution tandem mass spectrometry often enables molecular structure elucidation with a high level of confidence, as in this case [15,16,24].

The dark brown/blackish colour of R10 is obtained with a mixture of Acid Orange 7 (C.I. 15510) and Acid Black 1 (C.I. 20470) on naturally brown wool (Figure S7, Supplementary). However, the dark shades in R7, R12 and R15 produced more complex results. As shown in Figure 14 for R7_black, various molecules were detected. Except for Acid Orange 7, none of these molecules produced a match with available mass spectrometric databases of dye molecules.

Nevertheless, the tandem mass spectra of two of the main molecules show great similarities with the fragmentation pattern of Acid Black 1 (Figure 15). In particular, the assignment of specific fragment ions (Figure 15a) suggests that these molecules share the diazotised 1-amino-8-naphthol-3,6-disulfonic acid core. The compound with [M-H]⁻ = m/z 722.1121 is assigned to a tentative chemical formula C₃₄H₂₅N₇O₈S₂ (dppm = −1.7). The compound with [M-H]⁻ = m/z 736.1387 possibly corresponds to C₃₄H₂₇N₉O₇S₂ (dppm = −2.05). The two isomers with [M-H]⁻ = m/z 748.1395 are tentatively assigned to C₃₅H₂₇N₉O₇S₂ (dppm = −0.95). Although these results suggest that these molecules are likely to be related, a full structural elucidation proved extremely difficult solely based on these data. Therefore, the software SIRIUS was used. SIRIUS exploits computational algorithms to interpret complex mass spectral fragmentation and proposes results based on matching each fragment ion to a sensible structure. It then searches all accessible databases (ChemSpider, PubChem, etc.) for possible molecular matches, ultimately proposing a chemical structure [25]. The software produced an excellent match for both [M-H]⁻ = m/z 722.1121 and [M-H]⁻ = m/z 736.1387, assigning these compounds to Direct Green 1 (C.I. 30280) and Direct Black 38 (C.I. 30235), respectively. The structures of these molecules are reported in Figure 15 and Table S3, confirming that they indeed share the same core as Acid Black 1, coupled to a diazotized benzidine. The two molecules are also related, with Direct Green 1 having a phenol side group instead of a m-phenylenediamine group. This suggests that Direct Green 1 may be present as an oxidation product of Direct Black 38. The two isomers with [M-H]⁻ = m/z 748.1395 did not produce a convincing match using SIRIUS. However, their tentative formula has an additional C atom compared to Direct Black 38. This enabled these compounds to be hypothetically assigned to degradation products of Direct Black 38. Various azo dyes with an amine group placed at the ortho position to the azo bond are susceptible to formaldehyde attack, resulting in the addition of a C atom and creation of a 5-ring structure [16,24,55]. Direct Black 38 has two of these positions, in agreement with the presence of these two isomers, whose structure is tentatively reported in Table S3. Although these results could not be confirmed by analysing standard molecules, the accuracy of the mass spectrometric results coupled with the capabilities of the SIRIUS software emerged as a powerful tool for structure elucidation.

R14 and R15 share similarities with R7, R10, R12 and R13. However, the blue colourant used in R14 and the dark blue shades of R15 remain unidentified. The dye was poorly soluble in the solvents used for extraction, producing weak signals. Nevertheless, a small signal for a molecule with [M-H]⁻ = m/z 435.1871 was detected. The signal was too small to obtain a satisfactory tandem mass spectrum, which prevented us from further exploring the identification.

R17 shows a different palette of colours compared to the rugs discussed so far, apart from the bright orange colour, which is obtained with Acid Orange 7. The brown colour is obtained by dyeing naturally brown wool with a mixture of Acid Orange 7 and a sulfonated derivative of Acid Red 88 (Figure S8, Supplementary). There are various dyes corresponding to this structure, including Acid Red 14 (C.I. 14720), Acid Red 102 (C.I. 14730), Acid Red 12 (C.I. 14835), Acid Red 3 (C.I. 14910), Acid Red 13 (C.I. 16045), Acid Red 25 (C.I. 16050), Acid Red 17 (C.I. 16180) and Acid Red 44 (C.I. 16250). These dyes differ by the position of the sulfonic acid groups on the molecule. Although the retention times of these dyes may vary slightly, distinguishing exactly among them is challenging. For the purpose of this study, this dye is annotated as an azo red of the Fast Red family [24].

R17 also exhibits a bright dark pink colouration, which contains Basic Violet 10 (C.I. 45170), also referred to as Rhodamine B. The dye is identifiable by its characteristic molecular composition (Figure S9, Supplementary), which is abundantly described in the literature [16,24,56,57,58]. The result is consistent with the observation of the strong UV-induced visible luminescence of this sample presented in Section 3.2, as well as the results of MBI [11]. The lighter shade of brown in R17, indicated as olive green, is obtained with a mixture of Acid Orange 7, Basic Violet 10, Acid Red 88 and its sulfonated derivative on naturally brown wool.

The dark blue colour of R17 contains various molecules. In addition to Acid Orange 7 and Acid Red 88, two blue dyes were identified (Figure 16a). Acid Blue 7 is present along with its de-ethylated and debenzylated derivatives, which is in agreement with molecular patterns observed for other triarylmethane dyes of the “patent blue” class [18]. The study of the mass fragmentation (Figure 16b) and the comparison with references of similar dyes [24] enabled Acid Blue 7 to be distinguished from isomeric formulations. For example, Acid Green 3 (C.I. 42085) is an isomer with sulfonic acids on the N-benzyl groups. However, N-benzyl sulfonic acids produce a distinct fragment ion at 170.0043 m/z [24], which is absent in the tandem mass spectrum obtained from sample R17_dark blue (Figure 16b), confirming that the sulfonic acids are on the triarylmethane core. SIRIUS also confirmed this identification. The two other main molecules detected were assigned to Acid Blue 260 components. These are substituted anthraquinoid molecules with one and two benzamide substituents, respectively. Benzamide groups are easily recognisable by the fragment ion [C₇H₆NO]⁻ = 120.0455 m/z (Figure 16c,d). The rest of the fragmentation is also straightforwardly assigned to bond cleavages between the anthraquinoid core and the substituents, as well as loss of SO₂.

The green bluish shade in R17 is obtained with pure Acid Blue 7, which is indeed described as a greenish blue dye [59]. A different blue dye is used in the light blue areas. This was identified as Acid Blue 83 (C.I. 42660), another “patent blue” dye with two N-benzyl sulfonic acid groups and one ethoxyphenyl substituent on the triarylmethane core (Figure 17). De-ethylated and desulphobenzylated derivatives were also detected as being commonly encountered with this category of dyes [24]. The tandem mass spectrum is consistent with N-benzyl sulfonic acid losses, and SIRIUS confirmed the identification.

R24 and R27 were produced mostly using different dyes compared to the rugs discussed so far. Both R24 and R27 exhibit an olive-green colour, which was obtained by mixing Direct Yellow 1 (C.I. 22250) and Direct Brown 1 (C.I. 30045) (Figure 18a). Direct Yellow 1 was identified by comparison with a historic reference sample of this dye [56]. In all cases, the intact molecule ([M-H]⁻ = m/z 481.115) is accompanied by a smaller one ([M-H]⁻ = m/z 317.093), corresponding to a degradation product or synthesis by-product (Figure 18b,c). The identification of Direct Brown 1 was obtained by using SIRIUS, which produced a convincing interpretation of the mass fragmentation of the singly charged molecular ion [M-H]⁻ = m/z 635.1431 (Figure 18d), dominated by cleavages at the azo bond positions. R24 contained an additional molecule with [M-H]⁻ = m/z 425.0547, for which a tandem mass spectrum was not obtained due to the low concentration. The accurate mass can be tentatively assigned to the molecular formula C₁₉H₁₄N₄O₆S (dppm = −3.35), which may correspond to Mordant Orange 6 (C.I. 26520), whose structure is depicted in Figure 18a. As a tandem mass spectrum or standard molecule are not available, the identification remains tentative. Isomeric configurations of this molecule could also be seen as degradation products of Direct Brown 1, highlighting the importance of checking against reference material, when available.

Direct Yellow 1 was also used in the yellow background of R24. A combination of Direct Yellow 1, Direct Brown 1 and Acid Red 88 was used in the orange colour of R27. Acid Red 88 was the dye used in the red areas of both rugs, establishing a similarity throughout all rugs studied.

The bright green colour of R24 was obtained with a mixture of Direct Yellow 12 and Acid Green 25 (Figure 19a). The former was identified through comparison with a reference sample [56], while the latter produced a convincing match using SIRIUS. The characteristic fragment ion [C₇H₈NO₃S]⁻ at m/z 185.0141 fully justifies the two substituents on the anthraquinone ring (Figure 19b). The desulphonated derivative of Acid Green 25 was also detected. This is probably present as a degradation product or synthesis by-product. Another molecule was detected, corresponding to a molecular ion with [M-H]⁻ = m/z 393.1351 and formula C₂₄H₁₈N₄O₂ (dppm = −1.52) (Figure 19d). This compound has been detected in a reference sample of Direct Green 6 (C.I. 30295) and interpreted as a degradation product or synthesis by-product. Therefore, the possible original presence of Direct Green 6 or a similar dye cannot be excluded.

The main dye in the black areas of R24 is Acid Blue 113 (C.I. 26360). The observation of the sample under the microscope showed a blue hue for this dye (Table S2). The HPLC results revealed three compounds, corresponding to the intact molecule and two degradation products or synthesis by-products (Figure 20a). Although a reference sample for this dye was not available, a reference sample of Acid Blue 120 (C.I. 26400), a methylated homologue of Acid Blue 113, produced very similar results [56]. In particular, the degradation product or synthesis by-product at [M-H]⁻ = m/z 326.0598 was detected in both cases. Additionally, Acid Blue 113 has an isomer, which is Acid Blue 116 (C.I. 26380). In this molecule, both sulfonic acids are on the naphthalene rings. By contrast, the fragment ion at m/z 170.999 proves that one of the sulfonic acids is on the benzene ring, in agreement with the molecular structure of Acid Blue 113.

The results of the light blue colour of the poppies represented in R24 did not produce a straightforward identification. The degradation product or synthesis by-product at [M-H]⁻ = m/z 326.0598, also observed in the black sample, was detected, indicating that the dye is probably also an azo blue dye (Figure S10, Supplementary). However, the main molecule with molecular ion [M-H]⁻ = m/z 876.1070 was not univocally assigned to a chemical formula and a complex mass fragmentation spectrum was obtained. SIRIUS did not manage to fully interpret this spectrum, giving 2/3 possible matches for formulas, but no match for chemical structures. Another minor compound with [M-H]⁻ = m/z 552.0029 and similarities in mass fragmentation to the main molecule suggests that this might also be a degradation product or synthesis by-product, but its structure was also not disclosed. Our hypothesis is that this dye might be a trisazo blue dye sharing part of the molecular structure with Acid Blue 113.

The brown colour of R24 remained unidentified as well. Five compounds were detected with clear molecular ions and fragmentation spectra (Figure S11, Supplementary). However, neither calculation nor SIRIUS enabled the structures of these molecules to be disclosed. Some clearly contain halogens, most likely Cl, as inferred by the isotopic patterns. The high molecular masses suggest possible complex multiazo structures for these dye molecules.

R27 shared the use of similar dyes to R24, specifically Direct Yellow 1 and Direct Brown 1. It also exhibited a grey colouration that remained unidentified. The R27_grey sample revealed weak signals (Figure S12, Supplementary), mostly ascribed to degradation products or synthesis by-products observed in other samples (Figure S12, Supplementary), leaving the identification of a grey dye unclear. These products have also been observed in reference samples of diazo diamine dyes, indicating that the dye might belong to this category [56].

Finally, the two shades of blue of R27 were obtained using the same dye. Two molecules were detected and identified (Figure 21) as degradation products or synthesis by-products of Direct Blue 15 (C.I. 24400). These molecules were detected in a reference sample of this colourant referred to as Diamine Sky Blue and, in this case, the expected intact molecule was also not detected [60]. The dark blue sample also contained Acid Red 88 and Acid Orange 7, probably added to darken the blue colour. These were not present in the light blue sample.

4. Conclusions

Dyes, fibres and mordants used to produce nine Afghan war rugs in the British Museum collection were investigated by optical microscopy, MBI, SEM-EDX, MA-XRF and HPLC-DAD-MS/MS.

Early production war rugs (R7, R10, R12, R13, R14 and R15) showed similarities in the synthetic dyes used. These were Acid Orange 7, Acid Red 88, Basic Green 4, Acid Blue 92, Acid Black 1 and Direct Black 38. A light brown natural dye, probably from dock or sorrel (Rumex spp.), was used in R10 and R13 with the aid of an alum mordant. The natural dye extracted from the barberry plant (Berberis spp.) was found in the yellow areas of R12. The use of natural dyes in such late textile productions is a significant observation, which resonates with the combination of traditional and war motifs still present in first generation rugs. The variety of shades present in the rugs was often obtained by using the same dyes on either white or naturally pigmented wool, especially to create browns, reds, greys, blacks and blues. Such craftmanship also echoes centuries-old practices in Baluch and Turkmen carpet-making, often drawing from limited available materials to create colourful textiles. Interestingly, naturally dark wool could also be mapped on the rugs by using both the Mn maps from MA-XRF measurements and IRR images from MBI.

R17 revealed a wider variety of dyes used, adding Basic Violet 10, Acid Blue 7, Acid Blue 83 and Acid Blue 260 to the previously mentioned Acid Orange 7 and Acid Red 88. Although this is probably also a first-generation rug, the intricate floral motifs reminiscent of Baluch zakani rugs may set this apart from the rest of the rugs in terms of production.

R24 and R27 are later productions and were entirely dyed with synthetic dyes, mostly different from the ones in the first-generation rugs. These include Direct Yellow 1, Direct Brown 1, Direct Yellow 12, Acid Green 25, Acid Blue 113 and Direct Blue 15. The identification of some of these dyes was challenged by the availability of molecular databases and commercial standards, which remains a key area of further development in this research field. Nevertheless, the SIRIUS software emerges as a powerful resource to tackle complex mass fragmentation spectra, often resulting in solid spectral interpretations and convincing structural assignments. Harnessing the depth of information of mass spectrometry data in the analysis of synthetic dyes is crucial to navigate the high number of formulations potentially present in modern textiles. Despite these advancements, some of the dyes remain unidentified. However, the detection of Co and Cr in correspondence to a brown dye in R24 enables this to be categorised as a metal-complex or pre-metallised dye.

Furthermore, the differences in dye use between first and second/third generation rugs may be the result of different supplies of synthetic dyes to Afghanistan between the 1980s and the 2000s or deliberate choices by the rug makers in adjusting to the demands of a shifting export market. However, the full contextualisation of these results would require additional study of the economic and anthropological aspects around the production of these fascinating and complex objects, as well as the analysis of a larger corpus of rugs.