BENEATH THE SURFACE

The inclusion of minerals in Italian varnishes from the 16th to mid-18th centuries has long been a source of speculation. Balthazar Soulier, Stefan Zumbühl and Christophe Zindel present the first results of a long-term study showing that key answers can be found in early German recipes

Set-up for examination and sampling of the instrument coatings at Atelier Cels

Since the 1950s, researchers have applied modern analytical tools to elucidate the composition of ‘classical Italian varnishes’, in particular those of Antonio Stradivari. Despite the advanced technologies used over the past few decades, the investigations conducted by research groups across the globe have provided results that are only partial and often contradictory, at least in their interpretation. In most studies, the varnishes of only a small number of instruments were examined, using complementary analytical techniques. Most of them focused only on Stradivari’s instruments, which allows no contextualisation.

Nevertheless, some significant and consistent findings have been obtained. In particular, the varnish layer of instruments made before the mid-18th century has been unanimously identified as a mixture of drying oil and resins of the pine family. Only in a few cases have other resins been detected, such as mastic, sandarac and shellac.

The main divergences concerned the nature of the ground layer and the interpretation of the layering (stratigraphy). While the most comprehensive studies on Italian varnishes have reported a bi-layered structure based on oil–resin mixtures, other researchers have considered more complex systems involving protein and mineral particulate sub-layers, as well as special wood treatments.

One of the most intriguing findings of a large number of studies has been the discovery of a significant content of inorganic elements (mostly calcium, silicium, sodium and potassium) within classical Italian varnishes. This inorganic content was ascribed mostly to silicates and calcium salts by Michelman and Condax as early as in the 1950s, and to a volcanic rock (pozzolana) in the 1980s by Barlow and Woodhouse. Although the distribution of the mineral compounds within the layers were not established, the researchers related them almost solely to the composition of the ground layer. These analytical results had a considerable influence on luthiers, notably through Simone Sacconi in the US and David Rubio in the UK, who both developed different types of ‘mineral grounds’ and gave rise to a wide variety of speculations about the origin and role of mineral content in violin varnishes.

Thus, many fundamental aspects of the composition and elaboration of historical varnishes remain to be discovered. In particular, matters such as the origin and role of the mineral compounds; the precise type of resins and their relative proportions; the different types of pigments and dyes; and the effects of the cooking process still need to be studied in depth.

To address these complex issues, in 2018 we began a long-term research project at the art technological laboratory of the Bern University of Applied Sciences (BFH). This ambitious project, undertaken in collaboration with Atelier Cels in Paris and with the essential support of the Karolina Blaberg Stiftung, offered a unique opportunity to examine a large group of historical instruments over several years. In the first phase we focused on the main composition of the varnish stratum only. Rather than focusing on in-depth characterisation of single instruments, we favoured a statistical approach. The aim was to obtain as much information as possible on both organic and mineral compositions of a larger group of Italian instruments from the 16th to the mid-18th centuries and different schools, in order to recognise common characteristics and discern any regional or epochal specificities.

One of the main analytical challenges is to be able to distinguish original material from contaminants. In situ examination using a stereo microscope with visible and UV light sources is a prerequisite, but is not sufficient in itself to ensure the sample material is genuine and representative of the whole coating. Collecting micro-samples from multiple locations on each instrument considerably improves the reliability of the results. Such a strategy requires the miniaturisation of sample size.

One must be aware that the entire surface of an old instrument will inevitably have been subjected to contamination throughout its history, whether by environmental pollution, use, maintenance, or conservation treatments. As early as the late 18th century, Count Cozio di Salabue reports the common practice of cleaning and polishing the surface of instruments with clarified linseed oil and pumice. Since then, most instruments have been frequently lustred with alcohol varnishes based on shellac, sandarac, mastic, elemi or benzoin.

As the thickness of the stratum that forms the original varnish on top of the wood rarely exceeds 50µm (0.05mm) on ancient instruments, one should be able to extract and analyse varnish mini-fragments of no more than 20–30µm 3 , in order to avoid measuring the impregnated wood strata or external contamination (figure 1). Thanks to the recent development of a specific sample pre-treatment technique at the BFH, we were able to identify most of the organic compounds with micro-infrared spectroscopy (µFTIR) even in the smallest samples (c.20µm). As this technique is not destructive, the inorganic composition of the same micro-samples could be further determined with energy-dispersive X-ray spectroscopy in a scanning electron microscope (SEM-EDX). Chemical imaging characterisation of embedded cross-sections will only be carried out on a few exemplar instruments, as they require much larger samples and a much longer analytical process.

For this first study, we selected Italian instruments from different schools and periods that are unequivocally considered authentic, and that retain well-preserved parts of their original coatings. Each instrument was examined in detail using visible and UV light sources. The samples were taken under stereomicroscopes with fluorescence contrast with special micro-tools. To exclude contamination as far as possible, at least two samples were taken in different areas of each instrument. Before the analytical measurements, the purity of the collected micro-fragments was again checked under a high-magnification light microscope and dissected further if necessary.

ALL IMAGES COURTESY BALTHAZAR SOULIER

RESULTS

So far, we have conducted analyses on the varnish layer of about 30 instruments before c.1750, from 16th-century Bolognese lutes to instruments from the main Cremonese workshops of the 17th and first part of the 18th century, and up to those of the Neapolitan, Turinese and Venetian schools. The varnish of all instruments studied have revealed surprisingly similar chemical compositions (Table 1).

All the analysed samples are mainly composed of drying oil and a resin of the pine family (pine, fir, larch or spruce). Other resinous components that might have been expected, such as sandarac, amber, mastic or shellac, were not detected in any sample. The same basic raw materials appear to be present in similar relative proportions. We also noticed that most samples were highly saponified, due to the significant presence of calcium carboxylates.

As well as the oil resin compounds, we found a significant and remarkably consistent amount of minerals in all samples using SEM-EDX. We also detected a heterogeneous distribution of calcium (Ca), silicon (Si), aluminium (Al), sodium (Na), potassium (K) and minor elements such as phosphorus (P), magnesium (Mg), iron (Fe), copper (Cu), titanium (Ti), manganese (Mn), chlorine (Cl) and barium (Ba), with small variations in the trace elements.

The interdependent concentration of Ca and P indicates the presence of calcium phosphate, although other calcium compounds cannot be excluded. Calcium phosphate was commonly sourced from calcined bones. The other elements, mainly silicates and (Na, K) aluminosilicates, associated with the traces of Ti and Mn, are characteristic of volcanic rocks, notably pumice.

TABLE 1 Overview of the analytical results by instruments and components. Elements in bold text showed high concentration; those in normal text, medium concentration; and those in brackets showed low concentration/traces. Where (Cu) is noted, the copper traces might have come from cooking utensils. Apart from the three 16th-century lutes, which belong to the collection of the Kunsthistorisches Museum in Vienna, all of the examined instruments come from private collections. Their specifications are available on request.

FIGURE 2 Top Different kind of pumice and volcanic rocks (pozzolanes) Above Oil–resin varnishes containing 10 per cent of the corresponding volcanic powders

Pumice and pozzolanes (figure 2) have very close refractive indices to the aged oil–resin mixture, so it is almost completely transparent within the varnish. Moreover, as most of their compounds have an amorphous or crypto-crystalline structure, only a few mineral particles are visible by polarisation microscopy. In addition, as the calcium phosphate undergoes chemical dissolution in the increasingly acidic oil–resin mixture over time (forming calcium salts), the initial form of bone ash is no longer recognisable. The minerals, even if highly concentrated, are thus almost invisible in the oil-binding media, even at high magnification. This explains why they may have been overlooked in many previous studies.

WRITTEN SOURCES

To test whether the uses of the detected mineral compounds were historically documented, and to understand their attendant functions in oil varnishes, we conducted a comprehensive study of written sources. The extensive research proved to be extremely fruitful, as we were able to uncover a long tradition of using pumice stone and calcined bones as additives for varnishes in German sources (Table 2).

The reference to using pumice powder and white bone ash to improve the drying property of oils could be traced back to the early 15th century. In the so-called Strasbourg Manuscript of c.1400 we found a specific recipe for the preparation of painting oils:

Take linseed oil or hempseed oil or nut oil as much as you want and add burnt bone white and a lot of pumice and let the oil boil.

In the Mittelrheinischen Malerbuch (c.1445), a similar recipe is explicitly mentioned for the preparation of varnishes:

If you want to make a good varnish: take old hemp oil and make it hot […] put a lot of pumice and a lot of burnt bones […] if you want it thicker take four lots of mastic’.

In the 1549 Book of Illumination by Boltz von Ruffach, a further recipe also describes the preparation of bone ash and its incorporation into the varnish:

If you want to have a varnish that dries quickly, take sheep bones, put them in a new oven, and seal the lid with clay, put it in a strong [fire] for two hours, then put the oven down and let it cool. Take the burnt [bones] and crush them as fine as flour so that they are no longer coarse. Sieve it through a hair sieve and stir an amount similar to the size of a walnut into the hot varnish [...] so it dries quickly on whatever you spread it on’.

In general, the relative amount of minerals is significant. Cröker (1719) indicates equivalent amounts of minerals and resins! Only in rare cases, as in the Solothurn Manuscript of c.1500, should the mixture be filtered after cooking. Most recipes explicitly state that pumice and burnt bone should be ground very finely ‘like dust or flour’.

We found mention of the use of pumice and bone ash as additives in oil varnishes in sources up to the late 18th century. Thus it appears that this particular technique was continuously in use for a period of five centuries in many regions north of the Alps. Pumice and bone had the advantage of being widely available and inexpensive. In addition, because they are transparent in oil, they don’t need to be filtered like lead or zinc driers, and thereby enable the amount of oil resin to be reduced.

It is astonishing that this widespread practice north of the Alps is not reported in any Italian sources. We have examined in detail at least 60 Italian manuscripts and books, from 1420 to 1760, from Rome to Venice via Florence and Cremona, but we have never found any mention of ‘the northern mineral siccative’. The fact that drying agents are generally less addressed in Italian written sources may be because they were less needed – as there is simply more sunshine!

Nevertheless, even if the recipes are not specifically intended for violins and do not come from Italy, the ancient Germanic tradition of using pumice and bone ash as solid dryers in oil varnishes corresponds in every respect (composition, proportions and form) to the analytical results obtained on all the investigated Italian varnishes. These remarkable intercorrelations have led us to assume that the use of mineral additives in oil varnishes may well have been introduced to Italian luthiers by the numerous South German lute makers who migrated to all of Italy’s major instrument making centres during the 16th century. The analysed varnishes of the lutes of Hans Frei (figure 3) and Wendelin Tieffenbrucker (alias Venere) workshops already show the same basic composition as 17th-and 18th-century Italian violins.

TABLE 2 Overview of the main relevant written sources, in chronological order

Obviously, this German method of varnish preparation was passed on by the successive generations of Italian makers until the abandonment of oil-based varnishes. The fact that the minerals must be incorporated into the varnish shortly before application is a further indication that this technique was mastered by the violin makers themselves.

PROPERTIES OF THE MINERAL ADDITIVES

On the basis of the analytical data and the corresponding historical recipes, we carried out first experimental reconstitutions in order to evaluate the effects of pumice and bone ash on the oil–resin mixture. We produced oil–rosin varnishes (2:1 ratio) containing different amounts of pumice and bone ash (2.5 to 20 per cent by weight). We observed that the drying rate could be substantially increased with the addition of finely ground pumice and bone ash. They act on two levels: first, through the catalytic effect of the trace elements contained in pumice; and second, through their filling properties, which allow us to reduce the oil/binding media content. As specified in the sources, the minerals must be ground to a very fine powder to increase the active surface area and allow a homogeneous dispersion throughout the varnish. According to the analytical data and the first experimental reconstitutions, it can be assumed that the mineral content represents at least 10 to 20 per cent of the complete mixture.

This high mineral content proved to have many side effects on the properties of the varnishes, notably as fillers but also through the chemical reactions that occur over time between the different compounds.

Among other things, we could observe that the mineral fillers substantially improve the mechanical strength of the coating. In addition, the inclusion of bone ash as a source of calcium ions ensures a long-term chemical stabilisation of the natural degradation processes of the oil–resin mixture (owing to the persistent formation of saponification products). The incorporation of minerals has further shown very interesting effects on the visual appearance of the varnish. Depending on the origin of the volcanic rocks, we have observed significant colour variations. The small differences of reflection indices between minerals and binders, as well as the thin silicate structures of pumice, could also contribute to the special shimmering appearance of the classic Italian varnishes.

These new findings represent a paradigm shift in our understanding of historical violin varnishes and open up exciting new perspectives. Although composed of relatively common ingredients, the basic composition of classical Italian varnishes appears to be more complex than previously assumed. The varnish layer itself, and not only the ground layer, must be recognised as a subtly balanced technological system based on both organic and mineral compounds.