Varnish violin

String instruments made throughoutEurope from the earliest times were var-nished in a way that violinmakers in mod-

ern times have been unable to reproduce. Thesevarnishes were not all the same everywhere, butthey show certain common features that suggestthat they were made broadly in the same way.After 1700 the use of these varnishes began todie out, surviving longer in some places thanothers, and by about 1750 their use had all butdisappeared. Despite persistent efforts sinceabout 1800, no recreation of this classical var-nish, especially the ground layer, has beenattained.

There was always a ground varnish and thiswas probably overlaid with a finishing varnishthat was clear for many makers, while for othersit was colored. Where the top layer is colored, itsloss by wear exposes the ground varnish, oftenwith a very attractive appearance. Photographstaken with an electron microscope reveal thatthe ground layer varnish is distinctly differentfrom the top layer varnish. Some of these pho-tographs show a ground varnish with no topvarnish, but the absence of the top layer is veryprobably due to it having worn off. This paper isonly concerned with the ground layer varnish,which is considered by many to have the greatestinfluence on the tonal quality of the instrument.1

Several researchers have discovered usefulfacts about the nature of the ground layer ofclassical violin varnishes. This paper begins withan examination of this knowledge, to which weadd some of our observations about the acousticrequirements of violin varnish. Assembling allthis information points to the varnish being apre-polymerized oil-resin mix, emulsified withglue, stabilized by the addition of clay particles,and diluted with water until it can be spreadonto the instrument with the thumb. An emul-sion is a stable combination of oily and aqueousingredients whose close integration is main-tained by emulsifying and stabilizing agents. Inan emulsion, fine droplets of the aqueous medi-um can disperse into the oily medium to form awater-in-oil emulsion, or fine droplets of oil candisperse into the aqueous (water) medium toform an oil-in-water emulsion.

WHAT DO WE KNOW ABOUTTHE GROUND VARNISH?

What does it look like?Classical ground layers come in a variety of col-ors and these can all be reproduced by the tradi-tional methods of pre-polymerizing an oil-resinmix. When pre-polymerization is achieved bylow-temperature cooking or sun thickening, it

1

J. Violin Soc. Am.: VSA Papers • Summer 2007 • Vol. XXI, No. 1

A Recreation of the Particulate Ground Varnish Layer

Used on Many Violins Made before 1750NIGEL HARRIS, ROGER SHELDON, AND JAMES JOHNSTON

53 Aberdeen Ave., Cambridge, CB2 8DL. [email protected]

AbstractThe results of research by a number of writers are summarized and additional research by the authors is presented. Theacoustic function of the ground varnish is discussed. A case is made that the ground varnish should have the lowestpossible penetration into the wood. Emulsion varnishes have lower penetration than solvent-based varnishes. It is pro-posed that the ground varnish layer widely used before 1750 consisted of an oil-resin mix, emulsified with glue andstabilized by the addition of clay particles. This was applied directly to the wood and acted as both sealer and ground.The method of making and applying such a varnish is described. The physical and chemical properties of this varnishare similar to those measured on classical instruments.

VSAP Harris 9/8/08 5:22 PM

appears to have a pale yellow color. This color isseen in the Amati varnish and in the ground lay-er used by Stradivari, Guarneri del Gesù, and theCremonese school in general. Cooking at highertemperatures produces the darker colors seen onsome old instruments.

Many writers have described the groundvarnish used by classical makers as havingexceptional transparency. While we agree thatthe ground has distinctive visual characteristics,we do not consider the transparency to be excep-tional. Fry [1] describes the ground used by theCremonese makers as being dichroic: appearingyellowish in thin areas and reddish-amber inthick areas, depending on the method of woodpreparation and the viewing angle. This hasbeen said to result from the varnish being bire-fringent, which means it has two refractiveindices. Condax [2, 3] used the term “dichro-ism,” but he demurred from saying the varnish isbirefringent. We consider this still to be a matterof uncertainty, although again we do not denythat the varnish is very attractive.

What does it look like microscopically? Barlow and Woodhouse [4] have published sev-eral electron microscope pictures of cross sec-tions through the varnish films. In Fig. 1, thewood is split along the grain. Figure 1a, anexample from P. Rombouts of Amsterdam,shows that two different types of varnish havebeen applied. Figure 1b shows a cross sectionthrough the ground varnish on a Stradivari. Theappearance of these ground layers is typical ofinstruments made in Italy and northern Europefrom 1550 to 1750. Not all the samples testedhad the particulate layer, but significantly, thethree Stradivari samples tested did have it. Oth-ers may have lost it by wear. The thin upper lay-er (on the Rombouts sample) has a cross sectionrather like a piece of broken glass or toffee. Allconventional oil or spirit varnishes break, expos-ing a fracture surface similar to this. Figure 2shows a cross section through a solvent-bornevarnish film, i.e., a solvent-thinned oil varnish ora spirit varnish film.

The classical ground varnishes have a verydifferent appearance at the broken faces than dosolvent-borne varnishes. The surface looks morelike a broken biscuit and appears to contain a lot

of discrete particulate material. These particlesappear to be held together with some sort ofbinder.

Another characteristic of the ground layer,which is not clearly demonstrated in these pic-tures but which was noted by Barlow and Wood-house [4] after studying many examples, is thelack of penetration of the varnish into the wood.It has been suggested that the penetration of sol-vent-borne oil varnish into unsealed wood varieswith the formulation of the varnish, but fromour observation of electron microscope pictures,this penetration is comparatively enormous.

It is not possible to see any evidence of thepresence of a sealer under the ground in any ofthese pictures. Had a sealer been used, onewould expect to see some discrete layer of itsomewhere, albeit very thin. None of these pic-tures shows any such layer.

What is the particulate material?By using Energy Dispersive Analysis by X-rays(EDAX), Barlow and Woodhouse [4] deter-mined that the particles contained elements con-sistent with them being mineral material. Theelemental composition of this mineral materialvaried widely from one sample to another, evenwithin samples of varnish by the same maker.Barlow et al. [5] reported that most of the parti-cles have diameters less than 4 µm and are roughin shape, some being nearly spherical. From ourobservation of the pictures made by Barlow andWoodhouse [4] and the accompanying EDAXanalysis, we suspect that the particles are of soil-type minerals that have undergone weathering,possibly various clays and siliceous material.Although clay particles, particularly kaolinite,generally have a flat, plate-like morphology,weathering tends to blunt the corners of theplates. The wide variation in the mineral contentof the particulate matter suggests that its func-tion is not likely to be a chemical one.

Mayer [6] noted that emulsions can be stabi-lized by the addition of finely divided particulatematerial. It is, for example, commonly knownthat ground pepper will help stabilize a dressingmade as an emulsion of olive oil and vinegar. Ifthe components of the ground layer were boundtogether by an emulsion, the presence of the par-ticulate material might be an effective stabilizer.

2

J. Violin Soc. Amer.: VSA Papers • Summer 2007 • Vol. XXI, No. 1


3


Figure 1. Cross sections through classical varnish treatments measured with an electron microscope: P. Rombouts,Amsterdam, width of photo: 160 µm (top); and A. Stradivari, Cremona, width of photo: 80 µm (bottom). (Pho-tographs supplied by Dr. C. Barlow, University of Cambridge, UK.)


Condax [2, 3] had earlier found a lot of inor-ganic material in the varnish. He also identifiedthe presence of cellulose (perhaps from sandingdust), filaments of cotton (perhaps from applica-tion by a cotton pad), plant spores (perhapsfrom outside drying), starch grains (the use ofwhich is discussed below), and a great deal ofepithelia (flakes of skin, suggesting applicationby the hand).

What is the binder?The EDAX analysis by Barlow et al. [5] was notcapable of identifying the presence of elementsof low atomic weight, particularly those lighterthan sodium; hence, it could not be used to mea-sure the carbon, nitrogen, and oxygen content ofthe organic components. This was unfortunatebecause the common varnish ingredients ofresins and oils are all made from such elementswith low atomic weight. Had there been a recur-ring pattern of elements with higher atomicweight, it might have suggested the presence ofan inorganic binder such as waterglass (sodiumsilicate), but there was none. The likely conclu-sion is that the binder comprised the low atomicweight elements—carbon, nitrogen, and oxy-gen—found in common drying oils and resins.

The normally cautious Hill brothers [7] saidwithout reservation that the ground layer was

basically an oil varnish. This is a view that hasbeen fairly generally accepted by British makers,although many Italian makers have maintainedthat it was a spirit varnish. White [8] tested var-nish samples from a Serafin violin top, a Zanet-to viola top, and a Tononi cello rib. The sampleswere collected by scraping, and if there were top-coat varnishes they would have been included.The Serafin varnish had a resin componentmainly from a pinus source with small amountsof copal and sandarac. The oil was walnut, butonly enough was used to plasticize the varnish.The Zanetto varnish consisted of pine resin andwalnut oil, and the Tononi varnish containedmainly pine resin with a little mastic, somebeeswax, and some form of dammar with wal-nut oil. In a subsequent article, Dilworth [9]reported the results of White’s examination ofthe varnishes on a violin by Giuseppe Guarnerifilius Andrea and a Montagnana cello. TheGuarneri varnish was found to be mainly linseedoil with small proportions of (copal) resin andmastic added. The Montagnana sample was sim-ilar, with the copal being replaced by pine resin.

White’s tests for proteinaceous material,such as glue or albumen, all proved positive, butonly in small amounts. Not enough was foundto support the possibility of egg tempera as abinder. The Montagnana sample had the great-

4


Figure 2. Cross section through a solvent-borne varnish film. Width of photo: 80 µm.


est quantity of glue. When considering the Ser-afin, White concluded that there was ahydrophilic (water-based) ground with ahydrophobic (oil- or spirit-based) top layer var-nish. Condax [2, 3] examined samples from 60classical instruments. He also identified the pres-ence of small percentages of the colloid, glue (asdistinct from other proteinaceous non-colloidalmaterial like epithelia), in the samples, which hebelieved indicated that the varnish was appliedas an emulsion. White was unenthusiastic aboutit being an oil-glue emulsion because of the lackof a precedent for it. If there had been an obviousprecedent for this ground layer, it would havenot eluded rediscovery for over 200 years.

At one point in Fig. 1a there appears to havebeen a leak from the mixture of what Barlow etal. [5] believed was an oil or oil-like material thathad penetrated the wood. On some violins bysome makers, this leaking was not uncommon;on others it was not observed. This suggests thatthe oily material was a component of the binderthat was able to escape from the varnish into thewood, perhaps from an unstable emulsion.

Penetration into the woodPhotographs using an electron microscope donot reveal any sealer film below the ground. Asnoted above, Barlow and Woodhouse [4] foundthat, in general, the penetration of the groundinto the wood was very low. Greater penetrationcan occur at isolated points due to the leaking ofthe oily phase from unstable emulsions. Thiscould only have occurred if there was no sealerunder the varnish. It is possible that the groundlayer sometimes did not seal all the pores. Forexample, if one examines carefully the backs ofviolins by Stradivari and Guarneri del Gesù, oneoccasionally observes that where the pores in themaple are large, the ground did not bridge overthe holes and the colored top varnish passedthrough the holes in the ground into the wood,showing up as distinct colored dots. The posterof the Kreisler Guarneri del Gesù published byThe Strad demonstrates this. Where the top var-nish has worn off, the ground has holes in it thatare filled with red topcoat varnish. The sinkingof the topcoat varnish into these holes has leftholes in the top coat, which have filled with dirt.These holes occur on the line of the flames in the

maple. This suggests that the ground acted as asealer and has bridged over all but the largestwood pore holes. It also suggests that there is nosealer under the ground. Had there been a sealer,the colored varnish would not penetrate thewood at the holes in the ground. The Stradposter of the Heifetz Guarneri del Gesù showscomplete sealing of the pore holes, and wherethe ground has been worn off, dirt has enteredthe wood. Below, we show that, with correct for-mulation of the emulsion, all pore holes can bebridged.

If a ground that does not penetrate the woodcompletely wears off the surface, the poreswould be open to penetration. In areas that arein contact with musicians’ hands, such as thescroll and the neck, finger grease and oils willhave penetrated and fill the grain in a fairly col-orless way. In areas like the lower back, theremay be less grease and more scope for dirt. Thiswill undoubtedly vary also depending on howoften a violin has had polish applied to it.

The floor varnishes in use today are in mostcases emulsions of acrylic and other resins. In therecent past, floors were commonly varnishedwith oil-based varnishes or polyurethane. Floorswith oil varnish take a coat or two to seal thewood and have the effect of accentuating thegrain of the wood and strengthening its color.When the varnish wears off, there is no tendencyfor dirt to penetrate the wood. The emulsionvarnish, on the other hand, seals the wood on thefirst coat and begins to build up varnish on theoutside of the wood. The color of the wood isless affected, and when the varnish wears off, thewood gets dirty. The message is that emulsionvarnishes have much lower penetration into thewood than solvent-based varnishes.

In London in 1998 there was a very goodexhibition of British violin making through theages [10]. It was clear that the early English mak-ers used a varnish that, like the emulsion floorvarnish, did not accentuate the grain or color ofthe wood. Within a year or two of 1715 thesevarnishes completely disappeared (at least in theexamples on display) and varnishes that accen-tuated the wood grain and color were used. Wehave no idea why this change was so sudden anduniversal, but it is consistent with a change froman emulsion to a solvent-borne ground varnish.

We have investigated the tonal effect of vary-

5



ing the degree of penetration. We found that sol-vent-borne varnishes applied to unsealed wooddulled the tone of the bowed violin by effective-ly cutting off the high-frequency harmonics inthe sound spectrum. Some people may like thissound (for much the same reason as turningdown the treble control on an amplifier can givean illusion of increased bass), but there is noquestion that the projection and carrying powerof the instrument will be reduced significantly.This is described in greater detail in a paper byHarris [11]. Audio spectral analysis tests byRodgers [12] indicated that good modern violinsshow a high-frequency cutoff at ~4,500 Hz,while the Stradivari violin he tested did not cutoff until 9,000 Hz. This high-frequency differ-ence between modern violins and fine old Cre-monese instruments was also reported by Buen[13].

Other treatments, such as oiling the woodwith linseed oil or giving it a coat of sodium sili-cate as advocated by Sacconi [14], were found tobe equally damaging to the carrying power. Thisled us to try some of the sealers. We tried eggwhite, Sacconi’s White Varnish [14], dilutegelatin, and casein. These help to reduce pene-tration and improve the carrying power, but fur-ther experience led us to conclude that sincethese sealers all penetrate the wood to someextent themselves, they are still not goodenough. Eventually we began sealing the woodwith a modern polymer. This hydrophilic mate-rial placed a very thin film on the surface. It hasvery low penetration, apparently unlimitedbridging power over pore holes, and a modulusof elasticity the same as the wood itself. Wereached the conclusion that the coating wasacoustically invisible. The usual varnishes wereapplied over this sealer. The brilliance and pro-jection of the violins were remarkable. Acousticanalysis showed that the frequency responsefunction of the radiated sound of the bowedinstrument dropped off at a much lower ratewith rising frequency than was found with allother varnish treatments we tried. We were wellaware that this coating had nothing to do withthe historical varnishes pre-1750, but it did con-vince us that the varnish should not be allowedto penetrate the wood.

In an early effort to minimize penetration,we applied a layer of amorphous silica, which

had high oil absorbency, directly onto the woodsurface. This was then followed by the varnish.We expected that the amorphous silica wouldabsorb the varnish at the wood interface andprevent penetration into the wood. However, asthe silica was hydrophilic in nature and pos-sessed a high density of surface silanol groups, itwas able to chemically bind closely to the woodsurface. This had the effect of engendering aclose association of the varnish with the woodsurface rather than providing a barrier, and thusresulted in poor projection of the sound. The netresult reinforced our developing understandingthat nothing must be allowed into the wood.

Further evidence of this emerged accidental-ly. We had performed many acoustic tests onthree violins in the white. When these tests werefinished we carefully dusted off the rosin on theviolins and washed them with isopropyl alcoholto remove other slightly grubby marks. Anymicroscopic traces of rosin remaining on the sur-face would be washed into the wood by this pro-cess. When varnished and finished, these violinsall lacked the clarity and brilliance of our usualwork. One test violin was used more than theothers in the white and would have accumulatedmore rosin dust. It was the most affected tonally.Fry [1] quoted Vidal, Vuillaume, Savart, andLupot as all contending that if a violin is not var-nished, the tone is alright for a while but gradu-ally it loses clarity and brilliance, becomingsteadily duller. This is probably caused by thewood pores becoming filled with dirt, grease,and rosin.

On three occasions we have found it neces-sary to remove very poor over-varnish from aviolin. We first removed the varnish from thesides, then played it and found no tonal differ-ence. We then removed the back varnish andagain noticed no tonal difference. We thenremoved the varnish on both sides of the finger-board, bridge, and tailpiece line, and again, nodifference was noticed. But when the varnishwas removed from the area between the sound-holes and between the end of the fingerboardand the tailpiece, the tone was clearly different!We reached the conclusion that the varnish onlyin the area between the soundholes really affectsthe tone.

Wang and Burroughs [15] noted that veryhigh-frequency sound beams are radiated from

6



the motions of the belly in the general region ofthe bridge and f-holes. The effect of varnish pen-etration is to cut the high frequencies. We havebeen unable to find a convincing reason why thepenetration of varnish should have this effect,but we are certain that it does.

When the varnish wears off of a violin anddirt gets into the wood, there is probably littleeffect on the sound unless it happens in the areabetween the f-holes. Fortunately, this is an areawhere it seldom does happen because of the shel-tering effect of the strings, etc. Classical violinswere often finished with scrapers and the surfacewas slightly rough, particularly in the curls ofthe maple. When the varnish wears off, it isinevitable that traces remain in the troughs ofthe rough surface and are detectable with UVlight. The observation of these need not be anindication that the varnish has actually penetrat-ed the wood surface.

Summary of research studies

The classical ground-layer varnish contains adrying oil, a resin, glue, and comparatively largequantities of clay-sized mineral particles. It wasvery possibly an emulsion and may have beenapplied while in a fairly viscous state by spread-ing with the hand. The tonal benefit that it offersis consistent with its very low penetration intothe wood.

OUR EXPERIMENTS WITH EMULSION VARNISHESClassical ground layers appear to have used abinder made from an oil and a resin. (Availableevidence indicates as unlikely that there was anyrigid or routine formulation.) We decided tobegin by cooking together linseed oil and rosinuntil the firm pill stage is reached. The cookingpre-polymerizes the oil and makes it more sicca-tive. The resulting material was used in all thevarnish formulations that follow.

Formulation 1: A simple oil-in-wateremulsionThe first experiment was to try emulsifying theoil-rosin mix with ordinary violinmakers’ glue.It was easy to form the emulsion, but there wasthe obvious disadvantage that the mixture had

to be kept hot, and applied hot, to keep the glueliquid. The exact formulation was adjusted togive the least penetration into the wood. Duringthe development of the emulsion, a small drop ofthe emulsion was put on paper and spread out alittle with the finger. By holding the paper up tothe light and viewing its transparency, we couldeasily assess the amount of penetration into thepaper, both immediately and after a day. Thelower the penetration, the more opaque thepaper remained. Later, we tested samples ondiaphanously thin plane shavings of spruce.Finally, we arrived at a formulation that hadexcellent “hold out.” Some of this was applied towood and examined under an electron micro-scope. Figure 3 shows a view of this.

Figure 3 shows no evidence of penetration.Also, the cut surface of the emulsion varnish hasa coarser appearance than that of the solvent-borne varnish shown in Fig. 2. It also has aslightly particulate appearance, although it didnot contain added particulate material.

Formulation 2: A water-in-oil emulsionTo overcome the problem of having to apply themixture hot, we consulted old recipe books tofind traditional methods of making glue liquid atroom temperature. The most common methodwas to add nitric acid. Nitric acid was used toseparate gold and silver, according to an earlywork on assaying published in 1510. We evenfound a contemporary woodcut illustration(published in 1574) of a nitric acid productionplant in Venice, showing it being manufacturedon an industrial scale [16]. The addition of nitricacid does tend to make the glue, and hence theemulsion, hygroscopic. Fry [1] contended thatthe action of nitric acid on the resin in varnishpromoted the dichroism he observed.

Emulsions require an emulsifying agent (inour case, the protein-based glue), but in addi-tion, stabilizers may be needed to hold the emul-sion together for the full drying time of the film.The stability of an emulsion is dependent on itsenvironment and the nature of the substrate towhich it is applied. For example, an emulsionthat is stable enough to dry on paper can comeapart on spruce. When an unstable emulsioncomes apart, the separate phases can leak intothe wood, as may have been observed by Barlow

7



and Woodhouse [4]. In this formulation we usedstarch as the stabilizer and found it to be effec-tive. The motivation to try starch came fromCondax’s report [3] that starch was present insome varnishes. The resulting emulsion was awater-in-oil emulsion, and its ability to hold outwas very sensitive to changes in the recipe.

Figure 4a shows the excellent cell bridging.The structure showed fine holes like meringue.The varnish contained no particulate material. Ithad a lava-like appearance and looked evenmore particulate than formulation 1, but it stilllacked the visual effect of the particulate materi-al present in the old ground varnishes. The var-nish was applied to a cello using the hand,probably adding epithelia to the brew. Tonally,the result was indistinguishable from the instru-ments varnished by our standard method usingour polymer sealer. The varnish film was, how-ever, slightly hygroscopic, although it was cer-tainly not water soluble. Somehow, we stillneeded to reduce the glue content.

Formulation 3: An improved oil-in-wateremulsionWe were reluctant simply to throw in clay parti-cles just to make it look right under an electronmicroscope and give an EDAX analysis thatagreed with that of the classical varnishes. Wehad to know why we were adding it and what itsfunction was. Since particulate material is reput-ed to aid in the formation of emulsions [6], wedecided to add some to the emulsion to see if wecould reduce the amount of glue and hencereduce its hygroscopicity. We believed that thechemistry of the particulate material was unim-portant, so we had a free choice. We chosekaolin (China clay) because its refractive indexmost closely matches that of a dried linseed oil-resin varnish film and would therefore result inthe least loss of transparency. The kaolin wasground into the rosin-oil mix before the emulsi-fication. Initially, a nitric-glue medium wasadded and worked into the oily medium. Awater-in-oil emulsion was formed, but on fur-ther manipulation it flipped over to an oil-in-

8


Figure 3. An oil-in-water emulsion of cooked oil-rosin, emulsified with hot glue with no added stabilizer, applied tospruce cut across the grain. Width of photo: 80 µm.


9


Figure 4. A water-in-oil emulsion of cooked oil-rosin emulsified with nitric-treated glue (~18% glue content) andstabilized with starch. Applied to spruce and split along the grain. Width of photos: 160 µm (top) and 25 µm(bottom). Note the particulate appearance even though there is no added particulate material.


water emulsion. This emulsion required sub-stantially less glue than an emulsion without thekaolin. Also, we discovered that, as the propor-tion of kaolin was increased and the glue contentreduced, it was no longer necessary to make theglue liquid at room temperature. So we laterchanged to using normal violinmakers’ glue asthe emulsifying agent at room temperature withno nitric acid. The addition of the kaolin and theremoval of the nitric acid together made possiblea big reduction in the amount of glue. The kaolinclay behaved as an effective stabilizer. Thisresulting oil-in-water emulsion was more robustin its nature, and the resulting film showed nosign of being hygroscopic.

The emulsions presented in Figures 5 and 6show the effect on the appearance (under theelectron microscope) of reducing the glue con-tent and increasing the clay content. Withincreasing kaolin content, the glue content canbe reduced to the point where there is no need toadd nitric acid. The coating images of Fig. 6show closer agreement with the varnish samplesfrom classical violins that Barlow and Wood-house [4] examined. It should be remembered,while making this comparison, that we used apure form of unweathered kaolin, whereas theviolinmakers of the classical era were likely tohave used quite a mixture of weathered clay par-ticulates and siliceous material.

The particles in our emulsion were exam-ined from many angles and were found to beparticles of the oil-rosin mix that had formedaround the China clay particles. They varied insize and shape depending on the nature and sizeof the agglomerates of the clay particles in themix. It seems likely that grinding the clay intothe oil-rosin mix more thoroughly could breakup some of the agglomerates, reduce the effectiveparticle size, and increase dispersion according-ly. These emulsions were all formulated for“holdout” and “pore bridging.” This is welldemonstrated in Fig. 5 (top), where it can be seento be “reluctant” to come through the openpore. There is a visible crack at the cutoff pointbetween the varnish and the wood. It is probablethat this crack was caused by the stress of break-ing the sample for microscopy, but it onlyoccurred in the samples of virtually zero pene-tration. It is not apparent in all the picturesabove because of the viewing angle. The samples

using our polymer sealer showed the same crack.Claire Barlow has mentioned, in both papersand lectures, that she observed this apparentcrack at the interface between the wood andground varnish on some of the classical samples.

While any emulsion, regardless of how care-lessly it is formulated, will have low penetration,it is possible by careful formulation and testingto reduce the penetration to almost zero. It isquite possible that the Cremonese makers, real-izing the tonal importance of very low penetra-tion, did formulate their ground emulsionscarefully. Conversely, some of the old varnishgrounds may well have been simply unemulsi-fied mixtures of resin-oil with a handful of clayadded. We have experimented with suchunemulsified mixtures to investigate the effect ofadding particulate material.

Figure 7 demonstrates how increasing theparticulate content of an emulsion formula (witha fixed quantity of glue) affects the holdout onpaper. The paper shown here is of poor quality,being thin and poorly coated (or uncoated),which places a more demanding test on theemulsion than would be the case with high-qual-ity coated paper. The paper is shown againstback lighting: where the penetration is greatest,the color is lightest. The top row of samples hadnot been emulsified and was a simple oil-resin-particulate mix. The second row of samples hadbeen emulsified. The percentage of particulatesincreases from left to right. In comparing thetwo rows, it is evident that the penetration wasmuch greater for the unemulsified samples thanfor the emulsified samples. But also going fromleft to right in both rows, as the particulate con-tent increased, the penetration was reduced. Sothere would be some benefit from throwing ahandful of particulates into an unemulsified oil-resin mix, but there would be greater benefitfrom emulsifying the brew.

The formulation of the Harris & Sheldonemulsion ground varnishWe suggest that emulsion varnishes made of asiccative oil-resin mix, emulsified with glue(without nitric acid treatment) and stabilized bythe addition of clay-sized mineral particles, aspresented above, have similar properties as theground layers widely used by violinmakers

10



11


Figure 5. Electron microscope photographs of an oil-in-water emulsion of cooked oil-rosin emulsified with glue andstabilized with kaolin clay, applied to spruce and split along the grain. Kaolin and glue content: 12.6% and 9.2%plus nitric acid (top); 17.5% and 2.9% (bottom).


12


Figure 6 Electron microscope photographs of oil-in-water emulsions of cooked oil-rosin emulsified with glue andstabilized with kaolin clay, applied to spruce and split along the grain. Kaolin and glue content: 24.2% and 2.6%(top); 29.8% and 2.4% (bottom). Width of photos: 80 µm.


before 1750. In this paper, we refer to this fami-ly of varnishes as the “Harris & Sheldon emul-sion ground varnish.” Within this family ofvarnishes it is possible to give a more specific for-mulation that optimizes the acoustic and visualqualities.

The basic criterion in the formulation of thisemulsion is its ability to hold out from penetrat-ing a very thin sample of belly wood. But, as Fig-ures 5 and 6 show, this can be achieved with a lotof glue and little or no particulate material rightup to very little glue and a lot of particulatematerial. The high-glue option may requireaddition of nitric acid to achieve room tempera-ture workability, and this could lead to problemsof hygroscopicity. The low-glue/high-particulateoption may cause a loss of transparency and leadalso to poor bridging over pore holes in themaple backs of some instruments. (This mayhave happened to the Kreisler Guarneri del Gesùviolin.) The option we most prefer has a gluecontent of 2.6% and a particulate content of25%. Examination of the few pictures of Cre-monese varnish films available to us suggeststhat the Cremonese ground erred on the side of a

slightly higher particulate content than this. Working with emulsions can be a time-con-

suming and messy business. Each ingredient hasan effect on the balance of the other ingredients.If one makes a mistake or overshoots, it is verydifficult to “recover” the emulsion. It is better toclean up and start again. As the emulsion is anoil-in-water type, the glassware can be cleanedwith water. We make the emulsion using a glasspestle and mortar. The resulting consistency islike thick mayonnaise, best applied with thehand or thumb. It is easy to apply an even firstcoat over the surface of the wood. Happily, thisseems to deposit a thickness of approximately35 µm, which is quite close to the thickness ofthe samples shown by Barlow et al. [5].

As a guide to formulating an emulsion withthe lowest penetration, we give the controllingcriteria and suggest a recipe that will optimizethe performance of our emulsion ground var-nish.

The oil-resin mix. Historic examples probablyvaried widely in the choice of oil and resin andtheir relative proportions. We cook 40 g of lin-

Figure 7. Samples of unemulsified (top row) and emulsified (bottom row) varnish, with increasing content (left toright) of particulate material, laid on paper to demonstrate penetration. The samples were lit from behind. Thedarker the color, the less penetration there is (the lines are the usual lines printed on writing paper). The added clayreduced the penetration, but emulsification had an even greater effect.

13



seed oil and 30 g of rosin until the “firm pill”stage is reached. After preparation the mix mayneed to be diluted with turpentine to be work-able at room temperature, but no more than nec-essary. Increasing the dilution increases thepenetration of the emulsion. We add 15 ml ofturpentine while cooling, but not before theoil/rosin temperature has dropped well belowthe boiling point of the turpentine.

The clay. Kaolin, aka China clay, has a refrac-tive index that is close to that of the dried oil-resin film and will therefore minimize any loss oftransparency. The optimum amount of clay canbest be determined by testing the holdout of theemulsion on paper and thin spruce shavings. Wehave found that approximately 25% kaolin wassufficient to maximize the holdout. If the claycontent is increased beyond this, the emulsifiedfilm is more likely to leave a crater around anopen pore. So, for 5 ml of the above varnish wethoroughly grind in 1.7 g of kaolin.

The glue. We use 6 g of dry rabbit skin gluesoaked in 30 ml of water heated in a water bath.Just enough glue is added to form an oil-in-wateremulsion. For 5 ml of varnish, we mix in 30drops of the glue mix.

The water. After making the emulsion, watercan be added to lower its viscosity to a workableconsistency. The water may be added, a coupleof drops at a time, until the emulsion can beapplied by the thumb. We found that about 10drops was sufficient for the above quantities.

This recipe gives the following approximateproportions: 24.7% of China clay and 2.6% ofglue by weight in the dried varnish film. Theseare the minimum percentages required to per-form their function. The varnish has poor self-leveling, but it is not difficult to apply an evencoating with the thumb. When dry, the appliedfilm has a matte surface, or possibly a very lowsheen, not unlike a meringue, and a slightlychalky appearance.2 If the cooking is done at lowtemperature, the color of the dried film will be apale yellow. The ground takes over-coating verywell and is transparent.

CONCLUSIONS

Emulsions of the pre-polymerized oil-resin var-nish can be made using violinmakers’ glue as anemulsifying agent and clay-sized mineral materi-al as a stabilizer. These varnishes greatly reducepenetration into the wood and thereby increasethe frequency range of the radiated sound, aswell as its projection. They function as both asealer and ground layer. Of the emulsions weformulated, we found that the oil-in-water typewas the easiest to manufacture and it exhibitedbetter holdout and stability. Because of itsreliance on particulate material, we think that itcould be the most authentic. We call this formu-lation the “Harris & Sheldon ground varnish,”and we believe that it is a recreation of the fami-ly of ground varnishes used by many of the clas-sical violinmakers. Their varnishes varied, butthe Harris & Sheldon ground varnish can beadjusted in oil and resin type and in the natureand quantity of the mineral content to replicatemost classical ground varnishes. A specific for-mulation is given that has been designed to opti-mize the acoustic and visual properties.

The addition of clay-sized mineral materialhas an essential part to play in the constructionof this oil-in-water emulsion. Not only does itreduce the amount of water-borne emulsifyingagent (glue) required, but it also stabilizes theresulting emulsion. Similar ground varnish lay-ers may have been widely used before 1750, andbecause of the low penetration of emulsions intowood, they would have bestowed certain tonalbenefits. However, it may have been only in Cre-mona that the emulsion was specifically formu-lated to minimize penetration into the wood,perhaps explaining the superiority of many Cre-monese violins.

FURTHER RESEARCH

Recent advances in analytical methods used inmaterials science offer the possibility of further-ing our research. What we need are small sam-ples of varnish of unquestioned authenticityfrom instruments made by distinguished Cre-monese makers. Please contact us if you canoffer such samples.

14



ACKNOWLEDGMENTS

We wish to thank David Flynn of the School ofChemical and Physical Sciences, Victoria Uni-versity of Wellington, New Zealand, for themany electron microscope images and EDAXelemental maps he made for us.

NOTES

1. The various substances applied to violinshave been varyingly referred to as fillers, sealers,grounds, and varnishes. Terminology can raisestrong feelings, so we have given careful consid-eration to the wording of the title “particulateground varnish layer.” The coating material dis-cussed in this paper was very carefully formulat-ed to not penetrate the wood. To call it a “filler”would be very misleading. It certainly seals thewood against penetration by subsequent coatsand in that sense it is a “sealer,” but it is muchthicker than merely a sealer and forms a ground.A ground is the stuff one sees when the top var-nish layer wears off of an instrument made in theclassical era. Writers have traditionally referredto this layer as a “ground,” and we adopt thisterm. To distinguish it from alternative types ofground layers, we add the word “particulate” inrecognition of the fact that it contains particu-late material. It is made essentially as an oil var-nish. Hence, we describe it as a “particulateground varnish,” but it does act also as a sealer.

2. We have seen the back from a violin madeby Andrea Amati (in the collection of CharlesBeare) that has had the top varnish removedwith solvents to expose the unworn virginalground varnish. The surface of this ground hasvery low sheen and a slightly chalky appearance.The appearance of our Harris & Sheldonground varnish is similar in this regard.

REFERENCES

[1] G. Fry, The Varnishes of the Italian Violin-Makers of the Sixteenth, Seventeenth andEighteenth Centuries and their Influence onTone (Stevens and Sons, London, 1904).

[2] L.M. Condax, Italian violin varnish anddichroism, Catgut Acoust. Soc. Newsl., No.12, pp. 10-12 (Nov. 1969).

[3] L.M. Condax, Final summary report of vio-lin varnish research project, Catgut Acoust.Soc. Newsl., No. 37, pp. 31-36 (May 1982).

[4] C.Y. Barlow and J. Woodhouse, Of old woodand varnish: Peering into the can of worms,Catgut Acoust. Soc. J., Vol. 1, No. 4 (SeriesII), pp. 2-9 (Nov. 1989).

[5] C.Y. Barlow, P.P. Edwards, G.R. Millward,R.A. Raphael, and D.J. Rubio, Wood treat-ment used in Cremonese instruments,Nature, Vol. 332, p. 313 (24 Mar. 1988).

[6] R. Mayer, The Artist’s Handbook of Materi-als and Techniques, 4th ed. (Faber and Faber,London, 1981).

[7] W.H. Hill, A.F. Hill, and A.E. Hill, AntonioStradivari: His Life and Work (1644-1737),repr. of 1902 ed. (Dover Publ., New York,1963).

[8] R. White, Varnish—Eighteenth centuryinstruments examined, The Strad, Vol. 95,No. 1132, pp. 258-259 (Aug. 1984).

[9] J. Dilworth, The real thing, The Strad, Vol.95, No. 1134, pp. 437-438 (Oct. 1984).

[10] J. Milnes, Ed., The British Violin—The Cat-alogue of the 1998 Exhibition ‘400 Years ofViolin & Bow Making in the British Isles’(British Violin Making Assoc., Oxford,2000).

[11] N. Harris, What sort of sound should we beaiming for? British Violin Making Assoc.Newsl., Spring, 2006.

[12] O.E. Rodgers, Tonal tests of prizewinningviolins at the 2004 VSA Competition, VSAPapers, Vol. 1, No. 1, pp. 75-95 (Summer2005).

[13] A. Buen, Comparing the sound of goldenage and modern violins: Long-time-averagespectra, VSA Papers, Vol. 1, No. 1, pp. 51-74(Summer 2005).

[14] S.F. Sacconi, The “Secrets” of Stradivari,Engl. transl. by A. Dipper and C. Rivaroli(Libreria del Convegno, Cremona, 1979).

[15] L.M. Wang and C.B. Burroughs, Directivitypatterns of acoustic radiation from bowedinstruments, Catgut Acoust. Soc. J., Vol. 3,No. 7 (Series II), pp. 9-17 (May 1999).

[16] F.S. Taylor, A History of Industrial Chem-istry (Heinemann, London, 1957).

15



Varnish violin

Documents

Transcript of Varnish violin