Introduction

Albertus Pictor ( ~1440–1509) is the most well-known medieval mural painter in Sweden1. He is believed to have been of German descent and a first-generation immigrant to Sweden2. Pictor was the leader of an extraordinarily productive workshop to which 36 mural decorations in Swedish churches, mainly in the Lake Mälaren area, have been attributed. In some of the churches, there are readable signatures by Albertus Pictor, while in others the iconography, the technique and the outstanding workmanship justify an attribution to his workshop. His workshop also produced embroideries e.g., ecclesiastical textiles. This is supported by archival material where Albertus Pictor is mentioned alternately as a painter and as an embroiderer3.

One of the most well-preserved series of wall paintings by Albertus Pictor and his workshop is found in Härnevi church in Uppland (Sweden). They have been dated to around 14801,4. Härnevi church is one of only six churches with paintings attributed to Pictor and his workshop that has never been completely overpainted although the walls were covered with a lime wash in the eighteenth century (Fig. 1A, B)5. Research showed that these murals had only been subjected to one documented restoration and conservation intervention in 1956 during which the paintings on the walls were uncovered6.

Fig. 1: Views of Härnevi church.
figure 1

A Härnevi church (exterior) from the southeast. B Härnevi church (interior), view over the chancel. The vault as well as the walls is covered with wall painting.

Full size image

The murals are painted primarily al secco which means that the paint was applied on dry plaster7,8. It is the most common technique found in northern European mural painting. The plaster has been applied in one levelling coat and covered with a lime slurry. After the initial sketch was carried out, the colour was applied, often in several layers and with semi-transparent shadows and white highlights. Previous binding media analyses from churches decorated by the same workshop have shown that collagen proteins (most likely animal glue) was the most common binder used by the Pictor workshop followed by casein. Egg proteins and drying oils have also been detected9,10,11. Types of analysis used for these investigations included FTIR, GC-MS, HPLC and ion chromatography.

Lime and chalk were used as white pigments. Other pigments include red iron oxide, yellow ochre, carbon black, lead-tin yellow, red lead, massicot, malachite, atacamite, green earth, azurite and cinnabar12. Of particular interest is the recent discovery of fibrous red and blue organic lake pigments used in different areas of the murals in Härnevi church (Fig. 2a, b). This discovery led to an in-depth study of these pigments using the methodology of technical art history in order to gain a comprehensive understanding of how they were used, their properties, appearance, preparation and chemical composition. The methodology developed, and the understanding of materials gained, can be used by other researchers going forward. To date this specific type of organic pigment has rarely been identified in wall paintings but there is some reason to believe that it has been overlooked more than we think due to the often poorly preserved areas where it has been used.

Fig. 2: Results of optical microscopy and SEM on Sample Red 3 and Sample Blue 1.
figure 2

a Sample Red 3 embedded paint cross section. a(i) shows the mortar with aggregates (layer 1), thin slurry on the top (layer 2) followed by red paint layer pigmented with the fibrous red lake pigment (layer 3). a(ii) shows the same paint cross section under UV light, the top red layer (layer 3) exhibits a very noticeable orangey red fluorescence typical of madder lake pigments. a(iii) backscatter electron image shows the mortar with aggregates (layer 1). b Sample Blue 1 embedded paint cross section. b(i) shows the slurry (layer 1) followed by blue paint layer pigmented with the fibrous blue lake pigment (layer 2). b(ii) shows the same paint cross section under UV light, the blue fibrous lake pigment shows a dull bluish fluorescence. b(iii) backscatter electron image shows the slurry (layer 1) and blue paint layer with fibrous blue lake pigment (layer 2).

Full size image

Dyestuffs for lake pigments are obtained from plants or other organic material such as insects. This article will focus on red and blue plant-based lakes derived from madder and woad that could be obtained locally and were common in northern Europe during the medieval period.

Lake pigments are usually made by precipitating the liquid dyestuff onto an insoluble substrate in the form of hydrated alumina (Al(OH)3)13. However, lake pigments were not always prepared directly from insects or plants. A small number of findings also show that this type of red and blue lake pigment was prepared from either wool/ silk or textile shearings which had been dyed with madder and woad/indigo. This is supported very nicely by a contemporary early sixteenth-century coloured ink drawing from the Hausbuch der Landauerschen Zwölfbrüderhausstiftung (Nuremberg, Germany) which shows a man dyeing red and blue wool in a copper vat. The textiles used were proteinaceous, most commonly wool or silk14.

The textile material is boiled in an alkaline solution which creates a sticky protein-containing mass. The alkali used in the fifteenth century was commonly prepared from wood or plant ash extracted with water. Stale urine, wine lees and lime water could also be used14. The fibres dissolve to different degrees depending on the strength of the lye. The solution is dried, and the solid residue is ground with alum and the pigment is formed. The pigment is then boiled and rinsed to remove excess alum.

These insights are also supported by a group of fifteenth-century German recipes which describe the preparation of red and blue woolly lake pigments from textile shearings. The name Parisrot (Paris red) is often used for the red pigment in the instructions, whereas indich or endig (indigo) is used for the blue lake. The recipes differ from each other in that sometimes it is recommended to extract the dyestuff from the textile and most likely discarding the residue after filtering (although this is not explicitly mentioned) as is described in the recipe found in the late fifteenth-century ‘Colmarer Kunstbuch’ (Bern, Bürgerbibliothek, Cod. Hist. Helv. XII 45), while the majority of the recipes recommend different levels of dissolving of the fibres and then utilising the residue itself as the pigment as can be found in different recipes including the mid fifteenth-century ‘Nürnberg Kunstbuch’ (see Table 1)15.

Table 1 Selection of fifteenth-century German recipes for red and blue lake pigments prepared from wool shearings
Full size table

While our findings cannot completely eliminate other recipes, the texture and brittleness of the pigment produced with our own reconstructions indicates that both the red and the blue pigment found in Härnevi seem to be of the latter version on account of the undissolved fibres used directly as pigments (see Supplementary Figs. 2, 3 and 4).

The colour and the hue of the resultant pigment can differ depending on the recipe. Lake pigments prepared from wool shearings are primarily opaque but varying somewhat in translucency depending on the binder. They are sensitive to the caustic action of uncarbonated lime which renders them dark in colour so for the use on wall paintings they can only be used in secco techniques16.

There are a number of reasons given for why shearings are used to produce a lake pigment. One reason for using textile shearings as a basis for red lake pigment is because it is easier to obtain a desired colour than from the raw material itself. The range of anthraquinones present in the root and the extraction method gives different hues; thus, by using already dyed textile material the most difficult step in the process has already been undertaken by the dyer17. A secondary reason is that this method could be considered more economical. It has been reported that various textile scraps could be combined, perhaps dyed with different dyestuffs18,19. Mixed dyestuffs in a single pigment layer have therefore been used as confirmation of the connection between textiles and pigment manufacturing18. As indigotin is insoluble the use of the blue textile shearings was an easy option to obtain the pigment, therefore indigo dyed textiles were probably reused as pigment sources to reduce cost.

Although red lake pigments were infrequently used on large surfaces such as wall paintings, madder lakes were considerably cheaper than other types of red lakes (e.g., insect dye derived pigments) because it was native to Europe20. Late fifteenth and early sixteenth-century pharmacy price lists from Munich (Germany) suggest that madder lakes were almost 40 times cheaper than lakes prepared from scale insects21.

Indigo is generally not recommended for mural painting on account of its inadequate light fastness22. Even so, Cennini recommends indigo for use on walls. He also suggests that the colour of azurite can be improved by the addition of woad23. Indigo was commonly used in pigment mixtures, as it is naturally almost black in colour therefore other pigments were used in order to modify it. Recipes for mixed purples and blues containing indigo can be found on pages 100, 105, and 121 in the fourteenth-century Liber diversarum arcium24. Theophilus also mentions indigo for the use on walls in his treatise25. It has been identified in wall paintings in Asia and India dating from the sixth to tenth century. In Europe, it has been identified in the Moritz Chapel in Freiberg Cathedral dating to 1560 and it was commonly used in English medieval wall painting from the thirteenth century onward18. The use of woad in wall paintings was rarely recorded but there is an entry in the fourteenth-century accounts of Countess Artois in France that could be interpreted as woad used for murals. The accounts mention both indigo from Bagdad and “inde” for use in interior decoration. The indigo is listed as seven times the price for “inde” and therefore “inde” has been interpreted as based on native species such as woad18.

The most common dyestuff for red lake pigments as well as for textiles in northern Europe during the late fifteenth century was the madder root, both cultivated and wild (Rubia tinctorum L. and Rubia peregrina L.)14,26. In Sweden, wild plants of the same family (Rubiaceae) were also used for example, Northern bedstraw (Galium boreale) and Dyer’s woodruff (Asperula tinctoria L.)27. Madder root is believed to have been imported to Sweden from as early as the thirteenth century and during the fifteenth and sixteenth centuries the most frequently imported goods were luxurious textiles made of materials such as silk and wool28. However, it should also be noted that customs documents show that dyestuffs were imported at this time as well29.

Madder root contains 15 anthraquinone molecules capable of dyeing. These include alizarin, pseudopurpurin and purpurin30. The anthraquinone compounds have different hues that vary between reds with an orangey tone to reds with a more pinkish hue which enables an extensive range of colours to be obtained from the root by variations in the dyeing process 17. The root is cut or grinded into small pieces that are put into a water bath and the dye is extracted by soaking or boiling the mixture. After extraction, the remaining plant material is filtered off. The extracted dye can then be used for dyeing or for pigment making.

Blue dyes can be obtained from a number of plants but for medieval Europe woad (Isatis tinctoria L.) was the main source. It was cultivated on a large scale particularly in France, Germany and Italy where it was of considerable economic importance14. In fourteenth and fifteenth-century Germany, the growing and trading of woad for textile purposes was a flourishing business in around Erfurt and Cologne31,32. Woad was not replaced until the sixteenth and seventeenth century by indigo (Indigofera tinctoria L.) imported from India and the far east. However, the indigo plant was probably the source for the solid indigotin used as a pigment in Europe (due to its higher pigment content compared to woad) during and before the medieval times but its use as a dye was prohibited in much of Europe in order to protect the local woad trade. Chemically, there is little difference between the blue made from woad and the blue made from indigo and it has been reported that separating the two by analytical means is not possible. However, recent work by Laursen and Mouri has suggested that there are markers to differentiate these molecules (‘pseudoindirubin’)33.

The colouring matter for these blue lakes is found in the leaves of Isatis tinctoria L. and Indigofera tinctoria L. in the form of soluble, colourless precursor glycosides. Woad contains three of these precursor glycosides, indican, isatan A and isatan B, which are converted by enzymatic hydrolysis to indoxyl. The indoxyl molecules are combined by air oxidation and forms indigotin14. Indigoids need to be converted into water-soluble form by fermentation before dying. After fermentation the paste of the leaves turns into a yellowish green. The textile can then be added. When it is removed from the bath and exposed to the air the reduced indigotin is reformed to its blue colour through oxidation. Blue indigotin is also formed on the surface of the dye bath and can be skimmed off and used as a pigment14.

Methods

The surface and stratigraphy of the wall paintings were examined with a set of complementary scientific examination techniques. The focus of this paper is on the characterisation of the unusual red and blue lake pigments present in the murals.

Macroscopic and microscopic observations

The investigation started with non-invasive scientific examination methods: technical photography with visible light, UV fluorescence photography and macro-photography. A Canon EOS 60D was used for both VIS and UVF images but with an addition of a Kodak Wratten 2E filter for UVF. The UV light has been provided by UVA Spot 400/T Blacklight Large area UV lamp (315–400 nm). All UVF images are unprocessed.

Macro-photographic documentation in situ of both the red and blue pigments was performed with a portable DINO Lite microscope model AM4013MT with 1,3 megapixels and up to 200x magnification. All macro images are unprocessed.

Light microscopic studies of the unmounted samples and the paint cross-sections of the red and blue lake were performed on an Olympus BX41 analytical microscope. A 100 W Halogen projection lamp provided normal reflected light in bright field and in dark field illumination. An Osram HBO 50 high-pressure mercury lamp and Olympus filters U-M11011v2 (excitation 355–425 nm, emission 470 nm) and U-MWB2 (excitation 460–490 nm, emission 520 nm) were used for UV fluorescence microscopy. Normal reflected light in bright field and in dark field illumination using an Olympus LED fluorescence excitation light source (range 365–635 nm) was used on the samples (labelled Blue1 and Blue2) where the blue fibrous pigment was present. The same Olympus UV filters were also used with this light source.

Sample description and preparation

Micro-samples were taken for analysis from the mural (see Table 2). Three of these samples (labelled Red1, Red2, and Red3) were prepared as paint cross-sections by mounting them in a 1-component methacrylate (Technovit 2000 LC, Heraeus Kulzer, Germany) that polymerizes under UVa light (315–400 nm). Two further samples were collected for analysis (labelled Blue1 and Blue2,). Both samples were mounted in the light curing resin (CEM4000 Lightfix). After preliminary grinding with silicone-carbide (SiC) paper, the surfaces of all five paint cross-sections were dry polished with sheets (Micro-Mesh®) using grades 1800 through to 12,000.

Table 2 Sampling areas of wall paintings in Härnevi church (Sweden)
Full size table

Scanning electron microscopy coupled with energy dispersive X-ray microanalysis (SEM-EDX)

Cross-sections of the red and blue lake pigments were analysed with scanning electron microscopy coupled with energy dispersive X-ray microanalysis (SEM-EDX) in order to obtain elemental analysis of substrate and mordant as well as possible presence of inorganic pigments within the layer build-up. Sample imaging and analysis with chemical mapping and spot analysis was conducted on polished paint cross sections with a ~ 20 nm carbon coating using a Carl Zeiss Sigma variable pressure field-emission-gun scanning electron microscope (VP-FEGSEM) equipped with an Oxford Instruments X-Max 80 mm2 Silicon Drift Energy-dispersive Detector. Backscattered electron (BSE) imaging and Energy-dispersive X-ray Spectroscopy (EDX) were conducted in high vacuum using high current mode and accelerating voltage of 20 keV, working distance of 8.0 mm and aperture of 60 µm. EDX mapping data were acquired and processed using the Oxford Instrument Aztec software. All EDX maps shown have had the Aztec TruMap function applied, which removes background and artifacts, and resolves element peak overlap issues.

Fourier-transform infrared spectroscopy with attenuated total reflection (ATR-FTIR)

Fourier Transform Infrared Spectroscopy with attenuated total reflection (ATR-FTIR) was carried out using a Perkin Elmer Spectrum One FTIR Spectrometer with Spectrum software version 5.0.1 and fitted with a Universal ATR Sampling Accessory. The ATR crystal used was a diamond /thallium-bromoiodide (C/KRS-5) with a penetration depth up to 2 µm. 32 accumulations were used at a resolution of 4cm-1. Individual red fibres were isolated from the loose pigment samples under stereomicroscope and washed on a glass depression slide. The fibres were initially washed in water (a small amount of alcohol was added to reduce water tension) to remove excess material. The water was allowed to evaporate, and the fibre transferred to a clean depression slide. Then it was washed in a solution of MeOH and I M HCl to remove the coating of lime and possibly traces of aluminium potassium sulphate. The washed and dried fibre was placed directly onto the ATR crystal. The recorded spectra from the Härnevi samples were compared with reference spectra from the database of aged silk and wool fibres. Individual fibres were also isolated from reconstructed madder lake pigment on a wool substrate as well as a cochineal lake on a silk substrate. These fibres were also analysed with ATR-FTIR.

Fourier transform infrared spectroscopy (FTIR) using Perkin Elmer Frontier FTIR spectrometer with Perkin Elmer Spectrum Spotlight 200 FTIR microscope and Specac diamond compression cell was used to identify the blue fibre substrate and the colourant. 16 accumulations at transmission mode were used. The aperture window ranged between 20 × 20 µm and 100 × 100 µm. Scan range between 4000 cm−1 and 600 cm−1. Individual blue fibres were isolated from the pigment samples and washed in water with a small amount of alcohol according to the method described for the red pigment fibres. The washed fibre was placed in the glass depression slide. The recorded spectra were compared with reference spectra of aged blue wool fibres and indigo.

Raman micro-spectroscopy

Raman micro-spectroscopy was used for the verification of the indigo component in the blue fibre sample. Measurements were carried out with a BWTek i Raman Plus coupled with a with BWTek optical microscope. The spectra were recorded at 40× magnification using a 785 nm (λ0) near-infrared diode laser. Acquisition time was an average of 6 s with the laser operating at 100% of its maximum power (495 mW) and an integration time of 2000 ms. The scan range was between 0 cm−1 to 4000 cm−1. Raman shift and baseline correction were employed.

A reference spectrum of natural indigo from L. Cornelissen & Son (London) was used for comparison. Raman spectra of the reference sample were obtained using an inVia™ Confocal Raman Microscope (Renishaw, Wotton-under-Edge, Gloucestershire, UK). The spectra were recorded at 50x magnification with a 785 nm (λ0) near-infrared diode laser. Raman spectra were obtained with a 15 s exposure on the sample with a scan range between 0 cm−1 and 4000 cm−1. A 1200 L/mm (633/780) diffraction grating was used to measure the Stokes Raman scattering. The laser was operated at 0.05% of its maximum power, ensuring that the sample was not overexposed and damaged by the laser source during scanning. Raman shift and baseline correction were employed.

High-performance liquid chromatography with diode array detection (HPLC-DAD)

The dyestuff in the loose pigment samples was analysed using a gradient high pressure chromatography (HPLC) system with a UV diode array detector. The Merck-Hitachi HPLC system comprised: L-7200 autosampler, L-7100 gradient pump, Jones Chromatography Genesis C18 4 µm 250 by 4.6 mm column, L-7350 column oven, L-4500 diode array detector and L-7000 HPLC system manager software. A multi-step extraction approach was employed. The first extraction of the dyestuff was made with 200 microliters MeOH 50% + 0,1 M HCl. The samples were then placed into a water bath of 80 °C for 30 min. After the dye had been extracted the sample solution was filtered and the filtered extract analysed in the HPLC. The second extraction was made from the residue from the first extraction. This time 200 microliter extraction fluid of MHCl + 50% MeOH was used. The sample was placed in a water bat of 90 °C for 30 min and then filtered and analysed with HPLC.

Nano-liquid chromatography coupled to electrospray quadrupole time-of-flight mass spectrometer (nano-LC-ESI-Q-TOF)

The binding media analysis of the cross sections and a loose sample was carried out using nano-liquid chromatography coupled to electrospray quadrupole time-of-flight mass spectrometer (nano-LC-ESI-Q-TOF). 40 µL of 50 mM NH4HCO3 containing approximately 10 µg/mL of trypsin was applied to the samples and let to react at room temperature for 2 h. After the trypsin digestion, the solutions were taken from the surfaces and purified on reverse phase C18 (ZipTip). After equilibrating, binding and washing steps, target compounds were desorbed from the stationary phase. The solutions were consequently used for analysis. Measurement was carried out using Dionex UHPLC Dionex Ultimate3000 RSLC nano connected with mass spectrometer Bruker ESI-Q-TOF Maxis Impact. 10 µL of peptide solution were previously dried and then dissolved in 97:3:0.1% mixture of water:acetonitrile:formic acid. Consequently, they were loaded on trap column Acclaim PepMap 100 C18 (100 μm x 2 cm, size of reverse phase particles 5 μm, Dionex with flow rate of mobile phase A 5 μL/min for 5 min. The peptides were eluted from trap column to analytical column Acclaim PepMap RSLC C18 (75 μm x 250 mm, size of reverse phase particles 2 μm) using following gradient: 0 min 3% B, 5 min 3% B, 85 min 50% B, 86 min 90% B, 95 min 90% B, 96 min 3% B, 110 min 3% B. Mobile phase A was 0.1% formic acid in water and mobile phase B was 0.1% formic acid in acetonitrile. The flow rate during gradient separation was set to 0.3 μL/min. Peptides were eluted directly to the ESI source – Captive spray (Bruker Daltonics). Measurement was carried out in positive ion mode with precursor selection in the range of 400–2200 Da; from each MS spectrum up to ten precursors were selected for fragmentation. Peak lists were extracted from raw data by Data Analysis (Bruker Daltonics). Proteins were identified using Mascot version 2.2.04 (Matrix Science) by searching protein database Uniprot version 20110-12. Parameters for database search were set as follows: oxidation of methionine and hydroxylation of proline as variable modifications, tolerance 50 ppm in MS mode and 0.05 Da in MS/MS mode.

Results and discussion

Two types of fragmentarily preserved pigments with characteristic elongated fibres have been observed in the wall paintings, one red and one blue pigment. The red pigment is usually painted in two layers on garments, one pink followed by a deep red layer. In floral decorations it has been painted in only one thick layer. The blue pigment is even more fragmentarily preserved but seems to have been painted in one layer and has been mixed with another blue pigment and the same red lake pigment to obtain a purple hue. Both lake pigments were investigated with optical microscopy, UV fluorescence microscopy, scanning electron microscopy coupled with energy-dispersive X-ray microanalysis (SEM-EDX), Fourier transform infrared spectroscopy with attenuated total reflection (ATR-FTIR) and high-performance liquid chromatography coupled with diode array detection (HPLC-DAD). In addition, Raman micro-spectroscopy was employed to analyse the fibrous blue lake pigment (see Table 3, summary of analytical findings).

Table 3 Summary of analytical findings
Full size table

Red lake

Scientific analysis with optical microscopy, SEM-EDX, and ATR-FTIR resulted in a better understanding of the layer stratigraphy and the composition of each layer as well as the specific particle morphology of the red lake pigments (Fig. 2a (i-iii)). The results for the mortar show a high lime content in relation to the aggregate which is quite small in particle size. On top of the mortar is a layer of lime slurry that has been applied generously. The ATR-FTIR results also show the presence of dolomite (CaMg(CO₃), which was also confirmed with element mapping. All techniques show that the pink layer is a mixture of the fibrous red lake and a calcium-rich substance. The calcium-rich substance is probably chalk and not gypsum (CaSO4 · 2H2O) because of the low sulphur content (see also discussion of blue pigment below). Red lake pigments are sensitive to the alkalinity of lime, therefore the admixture of chalk that has a lower pH is more likely. The red layer on top of the pink layer has not been mixed with any other pigment according to the SEM-EDX analysis. Figure 3a, b demonstrate this clearly through the microscopic investigations where a darker red layer from the drapery is shown.

Fig. 3: Results of optical microscopy and SEM on Sample Red 1.
figure 3

a Sample Red 1 embedded paint cross section, optical microscopy. a shows the light microscopic image (layer 4) of the top darker red paint layer (red glaze) with the fibrous red lake pigment. b Sample Red 1 embedded paint cross section, SEM-EDX. b shows the backscatter electron image (layer 4) of the same area of the top darker red paint layer (red glaze) with the fibrous red lake pigment.

Full size image

Investigations on mounted and unmounted samples with optical microscopy and scanning electron microscopy (SE and BSE images) resulted in a better understanding of the morphology of the fibrous pigment as well as aspects such as the diameter of the red and blue pigment which can indicate the type of fibre used. The lake pigment fibres (particles) appear often large and irregular in shape with ragged edges, something that is consistent with previous studies of this kind of pigment13. Due to the heavy degradation of the fibres it was not possible to use the traditional microscopic techniques to visualise scales which are characteristic of wools. The approximate diameter of the different fibres (particles) varied between 15-27 µm (some fibres as thick as 40 µm) which corresponds to wool and not silk which is thinner34. Under ultraviolet (UV) light the red particles in both mounted and unmounted samples exhibit a very noticeable orangey red fluorescence typical of madder lake pigments (Fig. 2a(ii)).

Elemental analysis with SEM-EDX (spot analysis and element mapping) of the pigment layer shows that aluminium, sulphur, magnesium, silicon, copper, chlorine and potassium are present in the fibrous particles (layers 3 and 4). The single element maps for aluminium and sulphur show very similar and clear results for the fibrous pigment. Therefore, it is difficult to deduce that the sulphur detected here can be solely associated with the molecular structure of wool as has been suggested in other research on woolly lake pigments14,35. In the woolly lake pigment analysed here the presence of both aluminium and sulphur is more likely to be associated with potash alum [KAl(SO4)2 . 12 H2O] (Fig. 4a, b). This finding is not unusual given the use of alum for the dying of wool fibres and the grinding of the jelly-like product (dissolved fibres) with alum on a stone slab which is described in some fifteenth and sixteenth-century German recipes, forming the red and blue pigment (see Table 1). However, it is also acknowledged that the sulphur detected could also include that from the wool keratin (the sulphur content of wool varies between 2 and 5%). Calcium and oxygen associated with the presence of lime or chalk was found in layer 3, whereas in layer 2 (slurry) magnesium was found in addition to these. Interesting is also the marked proportion of magnesium in all analysed particles but this is likely due to the presence of dolomite in the final slurry layer (see Supplementary Fig. 5). Noteworthy are also the small amounts of copper in all particles which could have its origin in the raw madder root, where trace amounts are present, but a more likely explanation is that the presence comes from the preparation of the wool where a copper or brass vat (vats often referred to as ‘coppers’) or utensil was used in the dyeing process36. This is supported by an early sixteenth-century coloured ink drawing from the Hausbuch der Landauerschen Zwölfbrüderhausstiftung (Nuremberg, Germany) which shows a man dyeing blue and red wool in what seems to be a copper vat (see Supplementary Fig. 1). A similar amount of copper was also determined in a fibrous pigment produced from textile shearings (colourant most likely madder) used to tint the original red glaze on Christ’s lips of the Holy Blood Altarpiece (dated 1499–1505) by Tilman Riemenschneider and his workshop in Rothenburg o. d. Tauber, Germany (see Supplementary Table 1).

Fig. 4: Results of SEM-EDX on Sample Red 1.
figure 4

a SEM-EDX of Sample Red 1. a(i) shows the backscatter electron image with the fibrous red lake pigment. a(ii) and a(iii) shows the element maps for aluminium and sulphur. b SEM-EDX of Sample Blue 1. b(i) shows the backscatter electron image with the fibrous blue lake pigment. b(ii), (iii) show the element maps for aluminium and sulphur.

Full size image

The pigment substrate has been identified as wool. The FTIR spectra shows organic sulphur at 1040 cm−1. The amide I band for wool is at 1635 cm−1 while for degraded silk it is closer to 1600 cm−1 (Fig. 5a, b)37. This molecular information was used to determine that the pigment substrate is wool and not silk.

Fig. 5: Results of analysis with ATR-FTIR.
figure 5

a FTIR spectrum of red fibrous pigment shows organic sulphur at 1040 cm−1. b FTIR spectrum of blue fibrous pigment shows organic sulphur at 1040 cm−1.

Full size image

From the HPLC analysis, chromatograms show peaks of alizarin, purpurin and a small amount of munjistin (Fig. 6a, b). Extraction with 0.1 M HCl in 50% methanol produced 3 peaks which were identified on the basis of their retention time (Fig. 6a) and confirmed from their UV-visible spectra by comparison with library spectra. The UV spectra of extracted dyestuff confirm the molecules of each of these components (see Supplementary Fig. 6). The small peak at 9.5 min produced a weak and noisy spectrum with a 66.7% match to munjistin. Peak 2 at 12.4 min gave a 99.3% match to alizarin and peak 3 at 14.4 min a 99.7% match to purpurin. The second extraction (Fig. 6b) with 1 M HCl in 50% methanol produced 2 smaller peaks for alizarin and purpurin. This shows that the pigment has a natural plant-based source.

Fig. 6: Results of analysis with HPLC-DAD.
figure 6

a, b HPLC chromatograms for samples Red1 and Red2 showing the presence of munjistin, alizarin and purpurin (see also UV-VIS spectra in Supplementary).

Full size image

Madder (Rubia tinctorum L.), wild madder (Rubia peregrina L.), Indian madder (Rubia cordifolia L.), dyer’s woodruff (Asperula tinctoria L.) and northern bedstraw (Galium boreale L.) all contain these anthraquinones30,38. Previous research on these types of pigments made by Kirby et al.17 has concluded that extraction of the dye with stronger alkali results in greater amounts of alizarin and purpurin17. The pigment found in Härnevi has a high degree of undissolved fibres and the alkali must therefore have been relatively weak, and this might be the reason for the low alizarin content that was detected. However, it has been reported that the high level of percentage variations can also be an explanation. The same study on madder lakes of the fifteenth to seventeenth centuries revealed that most samples fell within the range 30–65% alizarin and 35–70% purpurin17. Based on the analysis and findings reported in the literature, the result has been interpreted as being most likely madder (Rubia tinctorum L.). This is also supported by the strong orange-red fluorescence of the fibrous red lake pigment under ultraviolet (UV) light visible in the paint cross sections which is characteristic of madder lakes39.

Red lake pigments prepared from textile shearings have previously been identified on early fifteenth to early sixteenth-century panel paintings and polychromed wooden sculptures from Sweden, Germany, Austria, The Netherlands and Belgium [refs. 17,32,40,41,42; see Supplementary Table 1]. To the authors’ knowledge this type of fibrous pigment has only been observed once before in early fourteenth-century wall paintings in Frankfurt Cathedral (Germany) although scientific analysis still needs to be carried out on the red pigment used in this specific mural43. The workshop of Albertus Pictor seems to have been using this type of pigment frequently. Similar red pigments with a fibrous character have recently been found on other Swedish murals by the workshop such as in the churches of Kumla, Täby and Almunge among others although in these murals the pigments have been identified only by macro photography using a portable microscope (see Supplementary Table 1). It is interesting to note that in 1993 Nord and Tronner already analysed a fragmentarily preserved red organic pigment from the murals in Täby church44. The result of the analysis was inconclusive but the dyestuff was interpreted as more likely to be brazilwood than madder, kermes or cochineal12.

Previous researchers have not been able to identify both the dyestuff and the substrate of this pigment on a wall painting. Helen Howard became the first to identify red lake dyestuffs in English medieval wall paintings in the late 1990’s. Howard also identified kermes and brazilwood lakes in the same paint layer in fourteenth-century wall paintings in the Chapter house of Westminster Abbey. Due to the mixture of dyes in the same layer, the pigment was interpreted as based on textile shearings from textiles coloured with different dyestuffs18.

The paint binders were identified as animal glue and egg using nano-LC-ESI-Q-TOF as described in the methods section. The analysed cross section showed both bovine collagen and ovalbumin. The protein analyses further supported the manufacturing from wool shearings as keratin from sheep was also found in the cross-section sample. The binder might have served as protection for the pigment from the caustic environment. Contemporary historical documentary sources such as the fourteenth-century Liber diversarum arcium recommend that pigments sensitive to lime should be tempered with egg yolk to protect them24.

Blue lake

Analysis of the cross sections by SEM-EDX shows that the blue fibrous lake pigment has been mixed with a finely ground blue copper-containing pigment (Fig. 2b (i-iii)) and a calcium-rich substance to create a pale blue paint. The calcium-rich substance is probably chalk of marine origin due the presence of microfossils. The presence of copper in the light blue particles as well as the information gathered from optical microscopy suggests the use of azurite [Cu3(CO3)2(OH)2].

The pigment substrate of the blue lake has been identified as wool. The FTIR spectrum for the fibres showed a typical spectrum for protein with amide A and N-H stretching at 3292 cm−1, amide I (primarily C = O stretching of polypeptide chains) at 1646 cm−1, amide II (N-H bending and C-N stretching) at 1546 cm−1, CH2 vibrations (scissoring) at 1410 cm−1 and amide III (NH, CN and CC vibrations) around 1228 cm−1, as well as peptide C–C stretching at 1065 cm−1. The proteinaceous composition together with the fibre diameters observed in situ and with the aid of optical microscopy clearly indicate that the fibrous particles are wool35.

The Raman analysis (Fig. 7) showed the peaks associated with indigotin, the colour component of both woad (Isatis tinctoria L.) and indigo (Indigofera tinctoria L.). Baran et al.45 reported that the most intense signal of the ring stretching mode is observed near 1571 cm−1 with a pronounced shoulder at 1581 cm−1 due to the stretching vibrations of the conjugated system of C = C, C = O and N–H groups45. The vibration of this conjugated system gives rise also to bands near 1700, 1622, and 1363 cm−1. The band attributed to rocking vibrations of N–H groups is observed at about 1224 cm−1. Vibrations involving C–H rocking are recognized at 1481, 1459, and 1247 cm−1 whereas the vibrations of five- and six membered rings are identified at 1309, 1014, 756, and 674 cm−1. The differentiation between indigo and woad was not attempted, however, the early date of the wall painting, c. 1480, means that the origin of the dye is most likely from woad.

Fig. 7: Results of analysis with Raman micro-spectroscopy.
figure 7

Raman analysis of blue fibrous pigment from Sample Blue 1 (Spectrum ‘a’) and a reference sample of genuine indigo from L. Cornelissen & Son, London (Spectrum ‘b’) for comparison.

Full size image

It is possible that the woad/indigo lake was originally purple or blue with a violet cast judging from the mixture of colours, the fact that it has only been found on the garments of Christ who is commonly associated with blue and purple in medieval iconography further supports this interpretation46. However, since the pigment and paint is so fragmentarily preserved it is not possible to say for sure what the original colour might have been.

Interdisciplinary collaboration using the methodology of technical art history and a combination of analytical techniques have enabled us to characterise unusual and complex red and blue fibrous lake pigments which were intentionally used to enhance and nuance the colour scheme of specific areas of the wall paintings by Albertus Pictor and his workshop in Härnevi church (Sweden). It was determined that the fibrous pigments found in the murals are made from degraded wool shearings which have been dyed with madder root and woad or indigo. The blue lake pigment has also been mixed with the some of the madder lake and finely ground azurite resulting in a mauve colour, which would be supported from an iconographic perspective as Christ’s robe was either purple or blue.

The chemistry of these pigments is very complex and required a powerful suite of analytical techniques to fully understand their properties and chemical composition. This included optical microscopy, UV fluorescence microscopy, scanning electron microscopy coupled with energy-dispersive X-ray analysis (SEM-EDX), Fourier transform infrared spectroscopy with attenuated total reflection (ATR-FTIR), high-performance liquid chromatography coupled with diode array detection (HPLC-DAD) and Raman micro-spectroscopy. Through the analysis, insights were also gained on the preparation of this pigment. For example, the presence of copper being due to the use of copper vessels for the dyeing process and not solely a possible pigment such as azurite. This is especially clear with the red lake pigment. Another feature is the importance of understanding why certain elements have been detected in marked quantities. The presence of aluminium and sulphur combined with the art-technological primary sources allowed us to identify the presence of alum in the final pigment used.

Here the access to the samples from the murals in Härnevi church enabled us to develop a methodology to thoroughly investigate the materials used by this painter and his workshop. The pigment identifications were realised using a suite of elemental and molecular techniques in order to report robust findings of both the origin and use of the red and blue pigments. More specifically, the historical pigment preparation method used was identified through the use of material science analysis combined with insights gained from the study of contemporary art-technological recipes from appropriate primary sources. Previous studies reported the presence of sulphur to be associated with wool. The FTIR analysis proved conclusively that the fibrous material was wool and a small proportion of the sulphur in the fibres is likely to be have contributed in a small way to the proportion determined in the elemental analysis.

However, the identification of both aluminium and sulphur in both the red and blue lake particles using SEM-EDX confirmed that the pigment making process used alum in the final stage after the semi-degradation of the fibres. We also report for the first time the conclusive identification of a blue woolly lake pigment, which is a completely new pigment to the palette not only for the workshop of Albertus Pictor but for the medieval painter in northern Europe in general.

In this study, we have established criteria for the identification of woolly lake pigments used in wall paintings. It is likely that these pigments have been used more extensively in painting and polychromy than has been currently reported. This comprehensive approach has led to new findings in this understudied area of sacred art in churches of northern Europe.

The technical details of the execution of wall paintings in Sweden and in northern Europe in general are still not fully understood. Detailed knowledge about the materials and techniques used in the making of an artwork contributes to understanding about the processes of deterioration and it is essential when deciding upon compatible conservation interventions.