Romanino’s Colour Palette in the “Musicians” Fresco of the Duomo Vecchio, Brescia

Abstract

This study examines the pigments and materials used in Girolamo Romanino’s Musicians fresco (1537–1538), located in the Duomo Vecchio in Brescia, with the aim of identifying and analyzing the artist’s colour palette. Ten samples of the pictorial layer and mortar were collected from two frescoes and characterized using microscopic and spectroscopic techniques. Confocal laser scanning microscopy (CLSM) was used to define the best positions where single-point, spectroscopic techniques could be applied. Raman spectroscopy and micro-Fourier transform Infrared spectroscopy (micro-FTIR) were used to detect pigments and organic binders, respectively. X-ray powder diffraction (XRPD) provided additional insights into the mineral composition of the pigmenting layers, in combination with environmental scanning electron microscopy equipped with energy-dispersive spectroscopy (ESEM-EDS). The analysis revealed the use of traditional fresco pigments, including calcite, carbon black, ochres, and copper-based pigments. Smalt, manganese earths, and gold were also identified, reflecting Romanino’s approach to colour and material selection. Additionally, the detection of modern pigments such as titanium white and baryte points to restoration interventions, shedding light on the fresco’s conservation history. This research provides one of the most comprehensive analyses of pigments in Romanino’s works, contributing to a deeper understanding of his artistic practices and contemporary fresco techniques.

1. Introduction

Girolamo Romanino (1484–1566), one of the well-reputed artists in the Italian Renaissance, is renowned for his innovative techniques and distinctive style. His work spans a variety of commissions, including frescoes, altarpieces, and decorative panels. Romanino’s family, originally from Romano di Lombardia, settled in Brescia in the 15th century [1]. Among Romanino’s notable works is the Musicians fresco (Figure 1) [2], located in the Duomo Vecchio, a historic cathedral dating to the 11th and 12th centuries [3]. This fresco captures a vivid scene of musicians, their instruments, and spectators, reflecting Romanino’s artistic mastery. Rediscovered during 19th- and 20th-century restoration efforts, this fresco demonstrates the artist’s unique style and contribution to Renaissance art in Brescia [4,5].

Originally, the fresco was part of a broader decorative scheme that included The Marriage of the Virgin painted on the organ doors [2]. The organ, constructed by Giovan Giacomo Antegnati in 1536 and inaugurated in 1538, was commissioned as part of a project to enhance the cathedral’s interior [2,3]. Preservation and restoration efforts, supported by institutions such as Intesa Sanpaolo, have ensured the longevity and appreciation of these artworks [4]. The rediscovery of the frescoes and their recontextualization within the Duomo Vecchio emphasize their importance in the cultural and artistic history of Brescia’s Renaissance [5,6,7].

This research investigates the pigments and fresco technique used in Girolamo Romanino’s Musicians fresco in order to have a reliable identification of the materials. The study crosses the results from several analytical methods (μ-Raman, μ-FTIR, XRPD, and ESEM-EDS). By comparing these results with historical references to 16th-century materials, the research aims to define Romanino’s colour palette, thus enhancing the present limited knowledge of the artist’s practices and providing valuable insights into Renaissance fresco techniques [7,8].

2. Materials and Methods

2.1. Sample Description

Ten samples of pictorial layer and mortar were taken from two frescoes by Romanino (dating to around 1537–1538), located to the left and right of the pipe organ in the church of Duomo Vecchio in Brescia (Figure 1). The samples were selected by the restorers based on availability in the areas of the fresco in need of conservation treatment. Due to the intrinsic fragility of the samples, in most cases, only small fragments were available (sub-millimetre size, labelled “a”, “b”, etc., as in

Table 1. Description of the investigated samples.

Sample, Fragments	Appx Size *	Colour	Sample Location
1 a,b	400 μm	Grey	Background on the left of the central flutist
2 a,b	200 μm	Brown	Background on the left of the central flutist, where white plaster is visible
3 a,b	500 μm	Brown	From the “tapped” area above the “lacerto” (remnant)
4 a–c	200 μm	Red	The red sock on the flutist’s left leg
5 a–c	200 μm	Yellow-Orange	Flutist’s yellow pants
6	200 μm	Gilded yellow	The “gold” brush on the wood-like column
7 a–c	200 μm	Yellow	The scarf of the old player, above the arm
8 a–f	200 μm	Blue, red	The left sleeve of the horn player (main Figure, white dressed)
9 a–c	200 μm	Yellow	Plaster from the area above the tympanum (not visible in Figure 1)
10	4 mm	Yellowish-Brown	Plaster from the balaustrade on the right, central height

* Maximum diameter of the fragment.

) for each sampling point. The details of the location of the samples are reported in Figure 1. The images and details of the analyzed fragments are reported in the Supplementary Materials. The fragments were analyzed using techniques that adopted special non-invasive protocols, thus preserving their original integrity and the possibility of measuring exactly the same areas using different methods.

2.2. Instruments and Techniques

2.2.1. Confocal Laser Scanning Microscopy (CLSM)

The confocal laser scanning microscopy was selected to obtain a comprehensive high-resolution overview of the samples and to facilitate a more in-depth analysis. Each fragment was imaged using the Olympus confocal microscope LEXT OLS-4000 (Evident Scientific, Tokyo, Japan) available at the Department of Geosciences, University of Padova. This instrument offers magnifications of 5×, 10×, 20×, 50×, and 100×, and includes a stitching mode in the software, allowing us to map the entire fragment in the form of mosaic images. From the low-magnification images (5×, 10×), the areas and points of interest on both the front and reverse sides of the fragments were identified and imaged at the highest magnification (100×) to serve as references for subsequent analyses. Sample 10, however, was too large for the CLSM instrument, and a portable professional digital microscope (Dino Lite Premier AM4113T) was used to obtain clear images of this sample.

2.2.2. Raman Micro-Spectroscopy

Raman spectroscopy was carried out at the Department of Chemical Sciences, University of Padova, using a Renishaw InVia Raman microscope. A 633 nm red laser was used for all samples, with laser power, acquisition settings, and magnification (10× to 50×) adjusted on each fragment for optimal results. A 514 nm green laser was employed for samples 8 and 9 due to insufficient excitation with the red laser. In both the Raman setups, a resolution of 1 cm⁻¹ can be estimated. Most of the spectra were acquired at 50× magnification, corresponding to a spatial resolution of about 1 μm at 633 nm and 1.5 μm at 514 nm. Sample 10 was too large for this method. The Raman scattering data were visualized using the Origin Pro 2025b software and interpreted using the available reference libraries for pigments and minerals (UCL Raman Spectroscopic Library: https://www.ucl.ac.uk/mathematical-physical-sciences/chemistry (accessed on 20 August 2025); RRUFF database: https://rruff.info/ (accessed on 20 August 2025)).

2.2.3. Fourier Transform Infrared Micro-Spectroscopy (μ-FTIR)

The instrument used for this analysis was the Bruker LUMOS II Micro-FTIR spectrometer, located in the Department of Chemical Sciences at the University of Padova. The instrument is equipped with a high-sensitivity MCT detector cooled with liquid nitrogen. Although the instrument can carry out analysis in ATR mode, with a micro-ATR tip, the measurements were carried out in external reflectance modality (micro reflectance), so as to avoid any contact with the very fragile samples. A spectral resolution of 2 cm⁻¹ can be estimated in this configuration.

All points previously examined with Raman spectroscopy were remeasured with FTIR, targeting areas with uncertain data and adding new points for comparison. Measurements were taken at 20× magnification (corresponding to a 50 μm × 50 μm squared area of analysis), with 64 accumulations typically used, increased to 128 for selected samples to enhance data statistics.

2.2.4. X-Ray Powder Diffraction (XRPD)

To further investigate the mineralogy of the pigments and support the spectroscopic observations, X-ray powder diffraction (XRPD) was performed at the Department of Geosciences, University of Padova. The tiny fragments were measured in non-invasive mode, using a special mounting that involved precise placement of the fragment at the centre of a rotating sample holder using a stereomicroscope and soft adhesive to ensure stability during X-ray exposure.

The data were measured using a Bragg–Brentano θ-θ diffractometer (PANalytical X’Pert PRO, Cu Kα radiation, 40 kV and 40 mA), equipped with a real-time multiple strip (RTMS) detector (X’Celerator, by Malvern PANalytical, Almelo, The Netherlands). The data acquisition involved a continuous scan over a range of 5–70° 2θ, with a virtual step scan of 0.02° 2θ. Diffraction patterns were interpreted using PANalytical’s X’Pert High Score Plus 4.0 software, which allowed us to qualitatively identify the crystalline components by search–match procedures based on crystallographic databases (PDF-2, International Centre for Diffraction Data; COD, Crystallography Open Database (https://www.crystallography.net/cod/ accessed on 20 August 2025)).

2.2.5. Environmental Scanning Electron Microscopy with Energy-Dispersive Spectroscopy (ESEM-EDS)

ESEM-EDS was selected for the analysis of the fragile fresco samples to preserve their suitability for the spectroscopic analyses. Unlike SEM, ESEM eliminates the need for coating, allowing us to obtain high-quality surface and elemental data without compromising the sample integrity. This method enabled the identification of the elemental composition of the selected grains and areas in each fragment, contributing to a comprehensive understanding of their nature while ensuring their integrity for further analysis.

The ESEM analysis was conducted at the CEASC Centre at the University of Padua, using the FEI Quanta 200 model equipped with a tungsten filament. The instrument configuration included several detectors: the Everhart–Thornley secondary electron detector (SED) in high vacuum (HV), the large field secondary electron detector (LFD) in low vacuum (LV), and the backscattered electron/gaseous analytical detector (BSED/GAD) in both high and low vacuum modes. Additionally, the energy-dispersive X-ray detector (EDX) used was the EDAX Element-C2B model, which allowed for elemental mapping across the sample.

3. Results of the Analyses and Discussion

3.1. Raman Analysis

Samples 1 (grey) and 2 (brown) both contained calcite and carbon black. In sample 1, the extensive black area (Figure S1) revealed bands of graphitic carbon black (1360, so called D1-band, and 1598 cm⁻¹, so-called G-band), while sample 2 (Figure S2) showed bands relative to a more disordered carbon black pigment (G-band at 1604 cm⁻¹ and several overlapping “D-bands” between 1220 and 1500) [9]. Both contained red ochre, with sample 1 also exhibiting chlorite-type minerals (241, 434, and 683 cm⁻¹) and albite-type feldspar. The latter are clearly minerals from the mortar preparatory layer of the fresco, together with calcite.

Sample 3 (fragments “a” and “b”, both brown) also showed calcite and carbon black in black areas (Figure S3). Fragment “b”, in an orange area, revealed the typical red ochre peaks (hematite: 221, 288, and 408 cm⁻¹) with other peaks likely attributable again to iron-oxide pigments (Figure 2).

Samples 4 (red), 5 (yellow-orange), 7 (yellow), and 9 (yellow) share common pigments, including yellow ochre, calcite, and carbon black. Sample 4 additionally contains red ochre (hematite), with peaks around 224 cm⁻¹ and 412 cm⁻¹, and goethite (around 293 and 392 cm⁻¹) (Figures S5–S7). Concerning the green spot of sample 4, sharp Raman signals at 1571, 1454, and 745 cm⁻¹ match with expected Raman peaks of azurite, while peaks at 1341 and 1049 cm⁻¹ are close to the expected ones for malachite. On the other hand, the Raman spectrum of azurite should show a strong peak around 400 cm⁻¹, while a strong peak around 430 cm⁻¹ is expected for malachite, which is lacking in the Raman spectrum of Figure S6. We can speculate that the signals are likely related to some alteration by-product deriving from an original azurite/malachite pigment.

In sample 5 (yellow-orange), a magenta point showed peaks at 1604, 1489, 1366, 1265, 1184, 967, and 743 cm⁻¹, closely matching those of pigment red 48:2, a synthetic azo red (Figures S8 and S9).

Sample 6 (gilded yellow), taken from a gilded area, showed a red pigment (weak peaks around 224, 400, and 630 cm⁻¹ related to hematite), likely a red bole, in the preparation layer. Analysis also identified carbon black, calcite, quartz (peak at 465 cm⁻¹), and some features may be related to organic compounds (Figures S10–S12). The gilded layer is, of course, invisible to Raman spectroscopy.

Sample 7 (yellow) shows the likely presence of red and yellow ochres, mixed with carbon black, similar to sample 3 (Figures S13 and S14), although a clear identification is not easy, due to the broadness of the Raman bands.

In sample 8 (blue and red-brown areas), fragment “a”, the Raman of a blue area showed bands at 529, 720, 832, 976, 1082, 1229, and 1365 cm⁻¹ (Figure S15). The characterization as smalt of the blue area is mainly based on the chemical composition measured by ESEM-EDS (see the ESEM-EDS Section 3.4), but we can suppose a posteriori that these observed Raman bands are related to smalt (see the discussion in the ESEM-EDS Section 3.4). In fragment “b”, the spectrum on a blue point (Figure 3 left) suggests the possible presence of a proteinaceous binder, likely gelatine (Amide I and II peaks at 1637 and 1539 cm⁻¹). The red-brown area revealed the presence of red ochre (Figure 3, right).

3.2. Micro-FTIR Results

To detect organic compounds not identified by Raman due to fluorescence, μ-FTIR analysis was adopted. Since there are virtually no reference spectra in the databases collected in external reflectance mode (ER-IR), several standard compounds were collected for reference using the same experimental protocol adopted for Romanino’s sample. These spectra are reported in the Supplementary Materials (Figures S18–S21), and they may prove to be very useful for future studies.

Samples 1 (grey) and 2 (brown) show calcite, organic compounds, and ochres as common components. For Sample 1, fragment P1a showed the “overtones” peaks of calcite (2515 cm⁻¹, 1796 cm⁻¹), and peaks likely attributable to organic substances (1335, 1532, 1683 cm⁻¹) (Figure S2). Fragment P1b displayed similar results, with an orange point contributing faint red and yellow ochres (overtones peaks in 2013, 2200 cm⁻¹) (Figure 4). For sample 2, in fragment P2a, the ER-IR spectrum of a brown area shows additional low-wavenumber peaks, likely indicating silicates (Figures S22 and S23).

Samples 3 (brown), 4 (red), and 5 (yellow-orange) all contain calcite, organic compounds, and ochres. For sample 4, fragment P4a showed malachite (1556, 1438, 1073, 831 cm⁻¹) in a green area and red ochre (1915, 2017, 1623 cm⁻¹) in a red area. P4b’s white areas displayed pure calcite, while red areas revealed red ochre alongside calcite and carbon black. For sample 5, fragment P5a showed weak ochre signals overshadowed by strong calcite. In P5b, a green grain showed no malachite peaks in the IR spectrum, although calcite, carbon black, and minor ochre signals were present (Figures S25–S29). It was not possible to acquire a Raman spectrum on this grain due to fluorescence problems. We can just speculate that the grain is due to a green earth, since this was a common green pigment used in fresco.

Sample 6 (gilded yellow), taken from a gilded area, showed protein-based glue in the preparation layer with peaks at 1665, 1565 (Amide I and II), and 1452 cm⁻¹. Minor signals of red ochre or other iron oxides and calcite were also identified (Figure S30).

Samples 7 (yellow), 8 (blue, red), and fragment P9c of sample 9 (yellow) share red ochre as a common pigment, identified by peaks at 1610–1614, 1870, and 1950 cm⁻¹. In sample 7, additional signals indicate yellow ochre and possible carbonyl compounds (peak at 1737 cm⁻¹), suggesting the presence of lipids (Figure S31). Sample 8 exhibits similar features to sample 7, with additional weak amide I and II peaks (1542 and 1661 cm⁻¹), indicating the potential presence of proteins (Figure S32).

Fragment P9c shows yellow ochre peaks at 805, 919, and 1044 cm⁻¹, along with signals at 1472 and 1692 cm⁻¹, suggesting contributions from organic compounds (Figure S33).

3.3. X-Ray Powder Diffraction (XRPD) Analysis

The Raman and micro-FTIR analyses were combined with XRPD measurements to obtain additional complementary mineral data.

The XRPD analysis of samples 1 (grey) and 2 (brown) expectedly reveals calcite (CaCO₃) as the dominant mineral, along with other carbonates (dolomite, CaMg(CO₃)₂) and well-detectable carbon-based material (graphite) (Figures S34 and S35), aligning with the findings from Raman spectroscopy. Additionally, sample 2 shows sylvite (KCl) as a secondary alteration phase.

Samples 3 (brown), 4 (red), and 5 (yellow-orange) contain calcite (CaCO₃) and graphite (C) as common minerals (Figures S36–S38). Sample 3 also contains hydrobasaluminite (Al₄(SO₄)(OH)₁₀·12–36H₂O), quartz (SiO₂), anatase (TiO₂), and graphite (C), complementing the Raman findings. Sample 4 includes sylvite (KCl), ankerite (Ca(Fe,Mg,Mn)(CO₃)₂), hematite (Fe₂O₃), and graphite (C). Hematite was also confirmed by Raman and micro-FTIR. Interestingly, sample 5 features clinochlore ((Mg,Fe)₅Al(Si₃Al)O₁₀(OH)₈), todorokite (Na,Ca,K)₂Mn₆O₁₂·3–4H₂O), dolomite (CaMg(CO₃)₂), birnessite ((Na,Ca)₀.₅Mn₂O₄·1.5H₂O), and quartz (SiO₂). The mineral assemblage clearly indicates the use of Mn-rich brown earths [10,11,12].

The XRPD spectrum of sample 6, the gilded sample, shows dominant calcite (CaCO₃) and dolomite (CaMg(CO₃)₂) peaks, along with iron oxide minerals, including hematite (Fe₂O₃), confirming the presence of red ochre as found in Raman spectroscopy. Additionally, two gold peaks are clearly detected, confirming the use of gold leaf or a gold-based material in the fresco (Figure 5).

The XRPD spectra of samples 7 (yellow), 8 (blue, red), and 9 (yellow) reveal common minerals, particularly calcite (CaCO₃) and graphite (C). Samples 7 and 8 both contain ankerite ((Ca,Fe)(CO₃)₂), a calcium-iron-rich carbonate mineral, along with graphite and calcite (Figures S38 and S39). Additionally, sample 7 shows goethite (α-FeO(OH)), an iron oxide hydroxide common in yellow ochre pigments [11,13], and baryte (BaSO₄). Sample 9, while sharing calcite and graphite, is characterized by the presence of magnesian calcite (CaMgCO₃) and quartz (SiO₂) (Figure S41).

Sample 10 (yellowish-brown) reveals a very complex composition with a wide range of phases, including todorokite ((Na,Ca,K)₂Mn₆O₁₂·3–4H₂O), birnessite ((Na,Ca)₀.₅Mn₂O₄·1.5H₂O) and/or chalcophanite ((Zn, Mn²⁺) (Mn⁴⁺)₃O₇·3H₂O), that is, several hydrous manganese oxides (Figure S42). Goethite (α-FeO(OH)), magnesium calcite (CaMg(CO₃)₂), and a spinel-type oxide are also present in the mixture. These findings once more suggest the use of manganese-rich earths and ochres in the fresco.

3.4. Environmental Scanning Electron Microscopy with Energy-Dispersive Spectroscopy (ESEM-EDS) Analysis

Samples 1 (grey), 2 (brown), and 3 (brown) share common pigments, including calcite (CaCO₃) and carbon black, identified by calcium and carbon as major elements. In these samples, the distribution map of calcium shows the dominating prevalence of calcite as the major binding phase of the lime mortar used in the preparation layer of the fresco. High local concentration of Ca indicates the presence of small lime lumps and limited mixing of the mortar (Figures S52 and S53). The mineral phases embedded in the mortar as inert additives are mainly identified by the maps of silicon, clearly marking the presence of quartz, feldspars, and phyllosilicates, the latter characterized by Al, K, Na, Al, and Fe, respectively. The combined maps of (Si,Al)+(Na,K) and Si+(Al,Fe) therefore show without ambiguity the distribution of the mortar phases.

Concerning pigments, areas with higher iron content neatly indicate the presence of iron-rich pigments in the mortar. They commonly appear as very finely ground grains embedded in the pigmenting layer: see, for example, the Fe map in Figure 6. The Fe emission peaks are especially well measured in the spot analyses performed at higher resolution on single grains (i.e., Figure S47, right: EDS spectrum 21 of sample 3a; Figure S51, right: EDS spectrum 44 of sample 8b). In most samples, the presence of calcite and ochre is confirmed, as indicated by the calcium and iron concentrations, respectively. The assessment of carbon black is hampered by the limited sensitivity of the detector at low energy and by the diffuse presence of carbonates.

In sample 4 (fragment “a”, Figure 6), one isolated grain rich in copper is present. The EDS spectrum of the grain essentially shows a composition based on Ca, Cu, and O (Figure 7), compatible with a Cu carbonate with lime contamination. It is clearly visible in the combined colour map as the blue grain surrounded by calcite. It is the same grain identified visually in Figure S5 (right) as a blue-green compound and interpreted as originating from an original malachite/azurite pigment, based on the IR analysis, with a surface alteration (as detected in Raman, Figure S6, right). The EDS analysis once more confirms and supports the spectroscopic interpretation.

Sample 5, fragment “b”, contains an appreciable amount of titanium (2.0%), likely from recent restoration. Traces of Ti have been found in the EDS spectra of other samples and support the identification of anatase (TiO₂) observed in sample 3 by XRPD. Anatase is one of the polymorphs of titanium dioxide (the others are rutile and brookite), and it has been massively introduced as a modern whitener in pigments and varnishes because of the high tinting power. It is normally assumed as a modern contaminant or introduced by recent restorations for its hiding power, although this assumption has been challenged by its identification in ancient pigments [14]. However, in the same sample, fragment a, a modern synthetic dye (PR48:2) was found by Raman. Combined with the presence of anatase, this suggests that a recent restoration intervention was carried out.

Sample 6 (gilded yellow) contains gold (Au), consistently observed by ESEM-EDS and XRPD. Appreciable amounts of copper (Cu) and silver (Ag) are present in the metal, together with even lower impurities of Fe and Mn, as they are common impurities in unrefined gold (Figure 8). As discussed previously, cracks in the gold layer allowed for testing the presence of glue beneath the foil by μ-FTIR, which was likely used for the adhesion of the gold foil.

In sample 7 (yellow), fragment “b”, the analysis shows a composition dominated by barium and sulphur (Figure S50), indicating the presence of barium sulphate already detected by XRPD (Figure S39), suggesting a stable compound possibly formed during restoration intervention. Although baryte can naturally be formed by bacterial activity, it is normally formed in mural paintings during conservation treatments by the so-called “barium hydroxide method”, aimed to remove calcium sulphates (mainly gypsum) and to consolidate highly degraded wall paintings [15]. The treatment induces the transformation of gypsum (CaSO₄·2H₂O) into calcium carbonate (CaCO₃) and stable baryte (BaSO₄), with the simultaneous removal of soluble ammonium sulphate ((NH₄)₂SO₄), the baryte being an almost insoluble salt which also has the role of consolidating the paint layer. In the present case, the baryte is abundant and distributed almost ubiquitously in the SEM chemical maps, clearly indicating precipitation during a previous consolidation treatment.

In sample 8 (blue, red), fragment “b”, the background analysis revealed mostly calcium (i.e., calcite), with minor amounts of sulphur, aluminum, and iron. The red background contained high iron levels, supporting the identification of ochre. In the blue areas visualized by CLSM (Figure S15), which were unexpectedly problematic for Raman spectroscopy (see discussion above), the EDS spectra show high silicon, together with calcium, potassium, iron, and cobalt (Figure 9). The major element composition (Si, K) is consistent with a silica glass produced using potassium as a network modifier (fluxant). The typical composition and the presence of Co as chromophore strongly suggest the material to be Co-based glass, known as smalt [16,17]. The systematic presence of calcium and iron in the analyses of the blue grains, not usually present in the composition of the smalt, is likely due to contamination from the surrounding ochre and lime grains.

With hindsight, the presence of smalt is consistent with some peaks observed in the Raman spectra on the blue areas (bands at 973 and 832 cm⁻¹ in Figure 3, lower left, and bands at 529, 832, and 976 cm⁻¹ in Figure S15), although it should be noted that Raman spectra of smalt are difficult to interpret due to the glassy nature of the material and its propensity to degradation, making its composition very variable, as assessed by Robinet et al. [18]. Ground smalt has long been used as a pigment, although in the long run, it frequently undergoes discoloration, especially in oil paintings. The observed impurities of Fe, Bi, and Ni associated with Co in the glass indicate that the origin of cobalt is probably one of the skutterudite–pentlandite–bismuthinite mineral assemblages commonly used as cobalt sources in the past, most likely the Erzgebirge region in Germany, where cobalt was obtained as a byproduct of the Bi extraction [19,20]. Although discontinuously used since the 15th century [19], the blue smalt started to become a common substitute for the more expensive azurite and lapislazuli by Italian and Flemish painters in the 16th century [18,19]. As a matter of fact, the material traded as zaffre or zafara is present in the 1534 inventory of the stores of a Venetian colour-seller [21], and it is known that Romanino already used smalt layered over azurite in “The Nativity” around 1524 [22]. Romanino’s perusal of smalt parallels similar practices observed in the works of later Venetian artists such as Paolo Veronese and Tintoretto, demonstrating a shared interest in exploiting this pigment’s vibrant colour and economic advantages [18,19,23].

The analysis of sample 9 (yellow), fragment “c”, reveals a substantial amount of magnesium, mostly linked to dolomite, with small amounts of iron and calcium. Minor quantities of aluminum, potassium, and sulphur are also present.

4. Conclusions

This study utilized a number of non-invasive techniques (μ-Raman spectroscopy, μ-FTIR, XRPD, and ESEM-EDS) to analyze the pigments in Girolamo Romanino’s Musicians fresco. The adopted experimental strategies avoided even minimal preparation and manipulation of the available micro-fragments and allowed the very same specific areas of each sample to be analyzed by complementary techniques, thus providing a truly combined investigation, encompassing the mineralogical/structural nature and chemical composition of the pigmenting materials.

The integration and critical comparison of the experimental results provide a comprehensive understanding of the pigments that were employed by Romanino for his palette in the second quarter of the 16th century (

Table 2. The palette of Romanino in the “Musicians” fresco in the Duomo Vecchio of Brescia, resulting from the multi-technique investigation.

Sample	Main Colour	Pigment	Raman	Micro-FTIR	XRPD	ESEM
1	Grey	Carbon black	Calcite Carbon black	Calcite Carbon black Organics	Calcite	Calcite
2	Brown	Carbon black Red ochre	Red ochre Plant-based black	Carbon black Ochres Silicates	Calcite Dolomite Graphite Sylvite	Calcite Iron oxide
3	Brown	Carbon black Red/yellow ochre	Carbon black Red/yellow ochre	Calcite Carbon black Ochre	Calcite Basaluminite Anatase Graphite	Calcite Iron oxide
4	Red	Red ochre Malachite Azurite	Limonite Hematite Carbon black Malachite Azurite Ochre	Malachite Red ochre Calcite Carbon black	Calcite Sylvite Graphite Ankerite Hematite	Calcite Iron oxide
5	Yellow-Orange	Yellow ochre Green earths Black earths	Ochre Calcite Siderite Magnetite Azo dye	Carbon black Calcite Ochres Green earth	Chlorite Calcite Dolomite Quartz Hydrous Mn-oxides	Calcite Titanium oxide Barium sulphate
6	Gilded yellow	Gold Red ochre	Red ochres Organics	Red ochre Calcite Protein binder	Gold Calcite Dolomite Iron oxide	Gold Copper Silver
7	Yellow	Yellow ochre	Red ochre Yellow ochre Calcite Carbon black	Ochres Carbonyls (lipid)	Ankerite Goethite Calcite Graphite Baryte	Iron oxide Titanium oxide Barium Sulphate
8	Blue, red	Red ochre Blue smalt	Red ochre Protein binder (gelatine)	Red ochre Carbonyls Protein	Ankerite Calcite Graphite Sylvite	Iron oxide Cobalt Nickel Bismuth Potassium
9	Yellow	Yellow ochre	Yellow ochre Calcite	Yellow ochre Calcite	Calcite Graphite Quartz	Iron oxide
10	Yellowish-brown	Yellow ochre Black earths			Hydrous Mn-oxides Goethite

).

The general formulation used by Romanino was to add finely dispersed carbon black to the fresh slaked lime on the fresco surface to obtain the different shades of the background from white to dark. The greyish tint was variously enriched towards red, yellow, brown, and green by the major use of the appropriate earths. Red and yellow ochres (iron oxides), black earths (manganese oxides), and possible green earths (green clays) have been observed and commonly used to obtain the desired hues. The special use of black earths composed of manganese oxides (manganite, todorokite, birnessite, chalcophanite, etc.) is especially interesting and ought to be investigated further.

All shades from yellow to red to brown are easily prepared by a simple mixing of the basic fine-grained oxides, with the neat prevalence of Fe-based oxo-hydroxides (goethite, hematite, maghemite, lepidocrocite, etc.), whereas a wider variety of colours can be obtained by inserting more exotic and expensive pigments, often based on copper minerals (azurite, malachite) or the Co-based blue glass, which was very trendy in Romanino’s times. The need for fine-tuned colours is of great importance in Romanino’s artistic vision and practice, since he was using the colour shades rather than the colour strength to define the volumes and the space.

The analyzed sample 8, taken from the left sleeve of the white-dressed horn player (Figure 1), although sampled from a white area, shows how red was mixed with other dyes to create shadows on the garments, enhancing the sleeve’s depth and texture. The discussed presence of blue smalt in the red layer of this sample further indicates the efforts of Romanino towards sophisticated hues and effects. Moreover, the presence of a proteinaceous binder, likely gelatine, suggests that this area was applied a secco rather than in true fresco technique. This aligns with Romanino’s known methods, as by the 1530s, he had refined his use of binders such as glue and egg tempera for pigments like azurite, smalt, and malachite. The painter’s versatility is evident in his strategic use of binders to enhance colour intensity. However, unfortunately, the use of smalt and the fragility of these layers raise questions about the long-term stability of his pigment applications. The detected signs of previous consolidations and restoration, beyond the last successful intervention, witness the troubled history of the “Musicians” fresco.

Sample 6, taken from the “gold” brush on the wood-like column (Figure 1), confirmed the presence of gold, used as a sub-micrometre-thick leaf of unrefined metal glued to the painted surface through a proteinaceous binder. Gold is another mark of the very qualified position of the painter in the artistic scene of the time.

Although previous scientific studies on the pigments used by Romanino are scarce, limiting direct comparisons, our research represents one of the first comprehensive analyses of pigments in his works. This study should serve as a foundational step for further investigations into Romanino’s artistic materials.