Home Choices How to Buy Biography About Tone Quality
Translate Web Site / Contact Us

Violins, Violas, and Cellos based on 28 years of research on Stradivari, Guarneris, and Amatis.

Nagyvary Violins
Review of "Stradivari Varnish" by Brandmair and Greiner

This monograph was released through commercial channels at a relatively high price, and so it is not easily accessible to individuals through libraries. What follows below is a review, the kind one would write for a scientific journal as an anonymous referee. Since the language might be too technical for some violin makers, I also added my thoughts on how the results could be interpreted.

The heavy-weight large-format coffee table book has two authors, B. Brandmair and S.-P. Greiner, who listed as primary scientific collaborators Dr. H. Dünnwald, Prof. Dr. Elisabeth Jägers, and Dr. Erhardt Jägers; also credited in the introduction were some 22 other academic contributors, mainly from German universities, and one from Italy. Considering the projected audience of the monograph and having only a minimal experimental section, it cannot be reviewed by the usual scientific criteria. One cannot know if the conclusions of the study were shared by the PhD scientists who did the analysis, or they represent the interpretation of the two authors. The results on the Stradivari varnishes were explained in the opening article by Stefan-Peter Greiner, who is a prominent German violin maker. The part dealing with the actual sample analysis was written by Dr. Brigitte Brandmair, a conservation scientist. They have taken varnish samples (with the wood attached) from 5 violins of Stradivari dated 1730 ( sample from the back), 1720 (rib), 1706 (rib), 1687 (back), and 1677 (rib) and also one cello of uncertain date (17xx), which could have been partially re-varnished. Additionally, a cello dated 1707 was sampled only in its interior of the back, and it was analyzed only for proteins by the ELISA method. Two violins of Guarneri del Gesł dated 1737 (rib) and 1735 (site of sample origin unknown), and one from N. Amati dated 1644 (rib) were also included in a limited number of analyses. The cross-sections of 7 instruments were stained by fuchsin that revealed the presence and location of proteins; 6 of them were subjected to EDX mapping of minerals and to FTIR analysis; 8 of them were photographed in a cross-sectional view by UV fluorescence and visible microscopy; and 3 of them were analyzed by GC/MS mainly to determine the amino acids. The entire body of the varnish was analyzed by GC/MS only for the del Gesł dated 1737 and the cello 17xx. Clearly, this was an extraordinary effort, which required a considerable amount of organizational skills in addition to the technical-analytical skills of the individual participants who all should be congratulated for their contributions. At this place, I can discuss only the most important conclusions pertaining to the stratigraphy and the composition of the varnish.

According to Greiner's essay Stradivari's process began with the application of a thin coat of protein as a sealer, and the protein could have derived from milk, egg white, or animal glue. This was in most instances stained brown by an unidentified stain. Then came the first major stratum-forming layer of colorless "lean" oil varnish, which was made up of linseed oil and spruce colophony in the weight ratio of 1 to 4; it also had a strong whitish fluorescence. In the early Strads, and in the Amati, this first stratum already contained coloring matter. The second main stratum in the 18th century Strads was the color varnish, which was probably made up by adding colored matter to the colorless varnish. A variety of colored particles were identified, among them the lake of carminic acid, madder lake, and also small amount of cinnabar (HgS). The cello 17xx was unusual because it had no stain, but the first stratum was already brown colored; it contained copper, and numerous colored particles. This part was considered the original work of Stradivari, and the further coatings were assigned to restorers. In spite of certain similarities, there were many individual variations among the instruments studied. The authors concluded in agreement with Échard et al. that Stradivari could have made many adjustments in his system of varnishing, and one cannot speak of Strad's varnish as a specific composition. Furthermore, Stradivari made use of only a few most ordinary materials, all of which have been used by many makers in the past and the present.

One of the noteworthy findings in the analysis of the ground layer in all 8 Cremona violins was the absence of the silica-rich mineral ground layer. Actually, this was less of a surprise since Échard et al. came up with the same conclusion one year earlier. The absence of a dense particulate layer, which was reported some 25 years ago to exist in some Cremonese cellos, has some intriguing implications that require further discussion. Actually, the amount of minerals in the varnish of some violins was estimated to be up to 10%, which is certainly a significant quantity. Although much of the book consists of attractive photographic images, the absence of anything with higher magnification makes it impossible for the reader to assess the density of the granular particles—be they inorganic or organic—which appear to be present. The absence of magnification approaching 1000X and above (via SEM) is a major shortcoming of this book.

I regret that there were not enough experimental data to lend more credibility to the unexpected proposition concerning the composition of the colorless varnish, which was claimed to consist mainly of spruce colophony, with linseed oil being only a minor component. This conclusion appears to be supported only by FTIR spectra, and it stands in contrast to the one proposed by the French team. The similarity of the FTIR spectra of linseed oil and colophony—as shown in their Figs. 3.14 and 3.15—would render the accurate determination of their ratios problematic. This qualification applies to the conclusions of both Échard et al. and Brandmair et al., the former having interpreted similar spectra as an evidence of a high linseed oil varnish. Actually, FTIR spectra are unreliable for the identification of the resin component as deriving from spruce; they cannot rule out the presence of other terpene resins. The spectral data of colophony are almost identical to those of sandarac which contains bands at 2933 and 2873 cm–1 attributed to –CH2/–CH3 stretching modes, 1697 cm–1due to C=O carbonyl stretching and 1645 cm–1 from C=C stretching; also bands at 1449 and 1384 cm–1 from –CH2/–CH3 bending. Several resins have the same components and hence similar FTIR spectra. The identification of a resin, and the exclusion of others, can be done best by pyrolysis GC/MS and the detection of certain marker compounds that are specific to each resin. The list of marker compounds was originally developed at the Netherlands Institute for Conservation, and subsequently augmented by published data (Mills and White 1994, Koller and Baumer 1997, Niimura and Miyaykoshi 2000, Pastorova et al. 1997, Scalarone et al. 2002, Scalarone et al. 2003, van den Berg et al. 2000, and Heginbotham 2008).

It should be mentioned that FTIR was successfully used by M. Derrick (1989) for determining the relative amounts of components in a three-component resin system of antique furniture where the FTIR spectra of the individual components were already known; the method involved a complicated scheme of spectral subtractions and deconvolution. We do not know if the German scientists made use of these methods, whose applicability might be complicated by the highly reactive linseed oil. In absence of specific information, it appears they could have relied mainly on measuring the exact wave-number positions of the most relevant absorption bands. The book merely states that a number of mixtures of aged colophony and aged linseed oil were analyzed by FTIR and the 4:1 mix gave the best fitting results. This would be an acceptable approach only if the compositions of the two "aged" components were well defined. Unfortunately, rosin on aging can be oxidized to a large number of compounds, of which only a few were characterized so far by S. Prinz et al. (2002); their individual FTIR spectra could be all slightly different. Depending on the conditions, linseed oil can be polymerized by Diels-Alder reaction and formation of C-C bonds; it can also be oxidized to several compounds having carboxylic acid groups.

It would be hard to simulate 300 years of natural aging with a short term artificial aging. Partial/incomplete oxidation would undoubtedly yield different mixtures. Similar considerations apply to the aging of linseed oil, and we don't even know how the oil was processed in Cremona prior to the application by Stradivari. The stand oil type would age very differently from a sun-thickened high-peroxy linseed oil. It appears both from the present data on the del Gesł 1737 varnish and the published work of Echard et al. that a considerable amount of dicarboxylic acid were formed during aging. This means that the C=O stretching peak from the ester position around 1745 cm-1 would shift toward the lower value of the COOH not only because of the presence of the abietic-acid-derived COOH, but also because of the COOH derived from the fatty acid ester oxidation.

The alcohol solubility of the varnishes studied by the German team could be an argument for the preponderance of colophony, however a weak one, because oxidized and degraded linseed oil is also quite soluble in alcohol. Incidentally, the oil varnish of the Ruggeri cello had a slow and limited solubility in alcohol. The same would apply for the Strad cello 17xx.


How shall makers interpret the diverse findings of several scientific laboratories?

The most original statement of Greiner & Brandmair that was stated with some conviction is the claim that the bulk of the varnish in all violins analyzed was a lean oil varnish, with only 20% linseed oil beside spruce colophony. Considering the insufficient support of this claim in the book, I would hold back my approval of this unusual composition. On the evidence alone, I would rather think a 50-50 mix would be more likely present. This problem calls for more experimentation with rosin varnishes and some degree of consensus on the best method. Makers who try out the lean 1:4 composition may get different results depending on how long the varnish layers were dried on sunlight. Some of the oxidation products of abietic acid (those with an aromatic C ring and opened up C-ring) have an oily consistency and would render the varnish probably too soft. In absence of much sunlight, the varnish could be hard and chippy. I doubt this kind of varnish would stand the test of time. Of course, there is plenty of evidence that Stradivari's varnish also lacked in durability.

An important question that arises is how much of what G&B have described for the stratigraphy was due to Stradivari himself. We are looking at varnishes no more than 30 µm thick. How many times was a typical Strad given a new protecting coat and polished by restorers during the last 250 years? Is 20 times a reasonable estimate? A lean oil varnish is extremely soluble in alcohol; it can be removed in less than one minute. More slowly, it is also soluble in essential oils and linseed oil. Therefore, each maintenance and beautification of a lean oil varnish would have inevitably led to a mixing of old varnish with new components, thus producing a major dilution effect, a mixing of the layers, and decreasing its original particle density. From what we know from the ex-Pawle varnish, Stradivari did not mix colors; he separated them with colorless varnish. He could have appreciated the optical consequences of mixing the colors or separating them. Having red, brown, and green particles all in one layer does not seem to me Stradivari's original intention.

However, it is unlikely that a solid particulate ground would have been broken up by the above mentioned dilution effect of restorations, unless the particles were made up primarily of chalk powder (CaCO3). Carbonates can be dissolved by the carboxylic acids present in the varnish over a long period of time. In some instances we can imagine the deliberate removal of the damaged old varnish and its refurbishing with new materials applied in a manner visually similar to the original appearance. It would be desirable to have also some Strads analyzed which did not go through the hands of French restorers, for example the instruments of the Spanish royal collection. The emerging compositional diversity of Stradivari's varnishes may something to do with the taste of the restorers who were privileged to carry out the required maintenance work.

Should we be shocked and disappointed that some Cremona varnishes contained a high density mineral/particulate ground, while the violins analyzed by the French and the German teams did not? Not really. It is feasible that Stradivari employed different varnishes for violins— even for the different parts of the violin—and cellos purely from acoustical considerations, and he obviously possessed a keen appreciation of their respective acoustical requirements. It makes sense that the use of mineral fillers was deemed necessary mainly in the case of large instruments whose wood was selected to be softer, with the grains wider and the pores larger; for them, an abundance of heavy mineral particles would help shifting the main wood resonances towards lower frequencies. The center of the violin belly might call for a strong mineral ground, while the flanks all over the perimeter would not. It is a difficult task to obtain deep-core varnish samples from the bellies of Strads, and one cannot blame B&G for not having any of those. Échard et al. reported perhaps one sample originating from the area under the tailpiece. Ribs of violins are supposed to be not overly stiff if the B resonances are already high; a solid mineral layer would be counter-indicated. One could argue the same way for many regions of the back. Sophisticated acoustical varnishing of the past, and of course, of the present, could have involved a changing topographical strategy.

Clearly, more work must be done by an alliance of owners, collectors, curators and a task force of scientists. Makers who find the price of the B&G tome high for their taste should consider the expenses of the underlying research, which considering the logistics, the time of the highly trained personnel, and use of instruments could be close to half a million dollars, unless the contributions were offered pro bono. In some form, the people who benefit from such research should find it fitting to pay for it.

| Home | Choices | How To Buy | Biographies | About Us | Tone Quality |
© 2001 - 2008 Joseph Nagyvary