National Gallery of Art

RUBENS'S: The Gerbier Family: Technical Examination of the Pigments and Paint Layers
Author(s): ROBERT L. FELLER
Source: Studies in the History of Art, Vol. 5 (1973), pp. 54-74
Published by: National Gallery of Art
Stable URL: http://www.jstor.org/stable/42617886
Accessed: 17-12-2015 19:16 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/42617886?seq=1&cid=pdf-reference#references_tab_contents

You may need to log in to JSTOR to access the linked references.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

National Gallery of Art is collaborating with JSTOR to digitize, preserve and extend access to Studies in the History of Art.

http://www.jstor.org

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
 RUBENS'S

The GerUer
Family:

Technical Examination of the

and Paint
Pigments Layers

ROBERT L. FELLER

INTRODUCTION

When a paintingis consideredfor treatment, a veritableflood of

questions soon confronts the curator, historian,and conservator.
What is the conditionof thesupportand paintlayersin relationto
what can and should be done for preservation?If the varnishis
tinted,can evidencebe foundto fixthe date of thisvarnish?Are
there additionsto the design by a hand other than that of the
originalartist?
Many of thesequestionscan be answeredfollowinga thorough
examinationby a trainedand experiencedconservator.However,
advancedscientific methodsof analysisare beingcalled upon today
to supplementthe analytical tools traditionallyemployed. Dr.
1 B. KeischandRobertC. Callahan, Keisch1and I havepreparedreportson thetechnicalexamination
"Investigation of
byLeadIsotopeMassSpectrometry," pp.75-78. The G erbier Family in which advanced technicaltechniquesof
analysishave been applied thatprovideanswerswhich are poten-
tiallymore preciseand conclusivethanwould have been possible
onlya few yearsago.
As a model of completenessin the scholarlyconsiderationof a
painting,one can scarcelyimproveupon the reportpreparedby
theInstitutRoyal du PatrimoineArtistiquein Brusselson thetreat-
2 "La descentdecroixdeRubens, Etudepréalablementof Rubens's Descent fromthe Cross at Antwerp.2Therein,
Bulletin
autraitment," , Institut
RoyalduPatri- are presentedin separateessays:his-
moineArtistique pp.6-187. complementary
5,1962,
, Bruxelles, investigations
constitution
tory,stylisticanalysis, of the paintand supports,con-
dition,conservation.In conformation withthisprecedent,mycon-
tributionto thestudyof The G erbierFamilywill considerthepig-
mentsand paint layersfoundduringthe technicalexamination.
The information being reportedby Dr. Keisch and myselfwas
gathered supportdecisionsto be made by the conservatorin the
to

54

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
programof treatment.In the process of seekinganswersto the
questionsposed by the conservatorand curators,we have gained
valuable insightsconcerning the constructionof the painting.
Nevertheless,as valuable and fascinatingas such information may
be, the principlethatguidessuch investigations mustnot be over-
looked: the primarypurposeof technicalexaminationis to aid in
the preservationand treatment of the work in orderto revealit as
as
faithfully possible as an artistic
and historicaldocument.

PRINCIPAL
PROBLEMS

Upon firstbecomingacquaintedwith The GerbierFamily, I found

the piecesof canvasadded on all sidesto be particularlyintriguing
3
althoughI knew thattheoccurrencewas not unusual.3In addition, R. D. Buck,"Examination andTreatment,"
the reddish-brown varnish,especially over the skirt,seemed to be pp.32-53.
remarkablydisfiguring. These featuresposed the two major ques-
tions for which technicalassistancewas needed. Could objective
evidence be provided that would determinebeyond reasonable
doubt: (a) whetherthe added pieces of canvas and paintwere an
integralpart of the originalwork and (b) whetherthe reddish-
brownvarnishwas a lateaddition?
To answerthesequestions,the simplestanalyticalapproachis to
determinewhetheror not a discontinuity existsin the materialsor
methodsused in the work. This is the principalinvestigative path
chosen.
A second pointof view undoubtedlywill have occurredto the
reader:can objectiveevidencebe obtainedto show thatthe paint-
ing agrees with the known workingmethodsof Rubens and his
atelier?Unfortunately, herewe are on muchlessfirmground;little
reliableinformationis available.4Althougha numberof authorities 4 Technicalinformation regardingtheworkof
have writtenon the methodsand materialsused in the oeuvre of morethanoneortwoindividual hasnot
artists
yetbeensystematically
assembled.A notable
Rubens, few such pronouncementsin the past have been based exception istheoeuvreofVermeer, seeH. Kühn,
upon adequatetechnicalanalysis.We will reviewthe mostreliable "A StudyofthePigments andtheGrounds Used
data thathas been publishedthus farand comparethemwith the byJanVermeer," NationalGalleryofArt,Report
information
andStudiesintheHistory
uncoveredin the examinationof the presentpainting. 202.SeealsoH. Kühn, pp.155-
ofArt, 1968,
"Untersuchungen zu den
Malgründen Rembrandts,
Jahrbuch derStaat-
SUMMARY OFTHERESULTS OFTECHNICAL EXAMINATION inBadenWürtemburg
lichenKunstsamlungen ,
2,Berlin, pp.189-210.
1965,
As a resultof technicalexamination,thelaboratoryis able to report
that
( 1) the pigmentsused in the painton the added pieces are similar
in all practicalrespectsto thoseemployedin the centralportion
(2) syntheticultramarine, a pigmentnot developeduntilthenine-
teenthcentury,was presentin the reddish-brown surfacevarnish.
Evidencesuggeststhatthissurfacecoatinghad been appliedsome-
timebetween1860and 1927.Consequentlyit was removed.
Furthertechnicalanalysisof thestructureof thepaintingreveals
that
( 1) a size,probablyglue,was appliedto the originalcanvas
(2) over the sized canvas, a sandy-coloredor lighttan ground,
pigmentedwithchalk,lead white,and a yellow earth,was applied
( 3) an imprimatura layer,grayin the centralsectionand whiteon

55 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
 the added pieces,was placed over the sandy-coloredground
(4) the mediumused in the sandy-coloredground,imprimatura
layer,and paintwas largelybased on a dryingoil, perhapslinseed
oil
(5) the pigmented,shellac-basedsurfacecoatinghad been applied
moreheavilyin someareasthanothers

ANALYTICAL TOBEREPORTED
INFORMATION

The size, ground-imprimatura layer,medium,pigments,and var-

nish will be discussedin this order. Pigments,grouped by color,
will each be introducedwitha briefdescriptionof itschemicaland
physicalcharacter.This will be followed by a discussionof the
particularcharacterof the pigmentencounteredhere and con-
cludedwithsuch factsas are availableconcerningitsuse by Rubens.

SIZEANDGROUND

Gettensand Stout definethe groundas the "stratumin a painting

r>R.J.Gettens andG. L. Stout,Painting Mate- which lies betweenthe supportand the paintor designproper;"5
NewYork,rev.ed.1966,
rials, pp.329,331.Foran is to preparethe surfaceof a supportto receive the
extensive seePhilipHendyandA. S. its purpose
discussion
Lucas,"Theground inpictures,"Museum ,27, artist'spainteddesign.A customaryprocedurewith paintingson
1968,pp. 266-276.canvas is to apply an aqueous size and pigmentedgroundover the
textilein order to isolate the fibers.The size, in particular,is ap-
plied both to controlthe absorbanceof the oil and to minimizeits
contribution to the deteriorationof thecanvas.In a numberof pas-
sages, Mayerne, contemporaryof Rubens,recommendedthe'
de a
6 T. T. deMayerne, Victoria Sculptoria & quae applicationof size to canvas.6
subalternum artium1620.Trans,byF. Berger, A
Beitrage zurEntwickelungsgeschichte derMal- proteinaceousglue size seems to have been employed in
technik, Munich,1901,[Ms.pp.5,11,87,96,98 The Gerbier Family. In crosssectionsof severalsamplestakenfrom
verso,154.] theedgesof thepainting,Coomassieblue,a dye withan affinity for
protein, stained an otherwise scarcelynoticeablepelliclethatwas in
7 MerylJohnson andElizabeth Packard, "Meth- contactwiththefibers(fig. 1) .7
odsUsedfortheIdentification ofBinding Media
inItalian oftheFifteenth
Paintings andSixteenth Over thesized canvasin theGerbierpaintingliesa tan-or sandy-
Centuries,"StudiesinConservation ,16, 1971, colored ground,pigmentedwith chalk, lead white and a yellow
pp. 145-164.earth. Apparentlyan oil or resin-oilvehicle,ratherthan animal
glue, was used for the binderbecause no evidenceof proteinwas
detectedin this layer when an attemptwas made to stainit with
Coomassieblue.
Plestershas reported fìndingan oil-based vehicle, containing
principallyochre and lead white,in the groundsof fourpaintings
on canvasby Rubens;in contrast,thegroundsin fourteenpaintings
s JoycePlesters,inGregory Martin, TheFlemishon panel were based on aqueous glue.8Coremansand Thissende-
School, TheNational Gallery, London,1970, scribedthe
Appendix II,pp.289-290. ground on the wood panels of The Descent fromthe
Crossas a layerof chalkand animalglue "impregnated in partwith
a dryingoil" and having a thicknessof about 200-300 microns
9 P. Coremans andJ.Thissen, "Composition et (0.008 to 0.012 inch).9 My impressionis that the groundin The
Structure desCouchesOriginales" Bulletin, In- Gerbier
stitutRoyalduPatrimoine Artistique, Bruxelles, Familyis thinnerthanthis,perhapsfrom100-200microns
5, 1962,pp. 119-129. (0.004-0.008 inch).
Over the sandy-coloredground layer in the Gerbier painting
thereoccurs an oil-basedpaint that has been variouslycalled the
imprimatura , as it is in the de Mayerne manuscript,primingor

56 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
 ofpaintattopleftcorner
Fig.1.Crosssection showingthinlayerof
whiteimprimatura
paint, layer, tan-colored
groundandlayerofproteinaceous
glueshestainedwithCoomassie lOOx.
Blue.SampleAl,Magnification:

couche d'impression . The last expression,used by Coremansand

Thissen,is mostdescriptivein the literalsense,forthisis a layerof
overalltonalityupon which the artistmay place his design.None
of theseexpressionsis completelysatisfactory, but forthe present,
Buck and I havesettledupon theword imprimatura .
In thecentralpiece of canvas,theimprimatura is gray,a layerof
about 10-30 micronsthickness(0.0004 to 0.0012 inch) pigmented
withlead whiteand plantblack. A highproportionof lead white
causesthiscoat to be of greateropacitythanthe groundbelow, as
10 10A photomicrograph incolorofa crosssection
maybe clearlyseenin thecrosssections(fig. 1; Buck, fig.7). The
color of the imprimaturaoftenis used to advantageas a middle ofTheDescentfromtheCrosspublished by
Coremans andThissen showsa graylayerof
valuein modeling;Buck pointsout thatthegrayis employedin this similarappearance ofcourse,thevehicle
although,
mannerin the faces of Mme Gerbier and her eldest son (Buck, inthiscaseissaidtohaveanaqueousbase.
fig.9).
In contrastto the centralsectionof canvas,the pieces added on
the sides bear a white layer immediatelyover the tan-colored
ground; the particlesof black that characterizethis layer in the
centralsectionare clearlyabsent(comparefig.1 withBuck,fig.7).
A gray layer does not extendoverall,but instead,gray, red, or
otherlocal color has been applied directlyover the white as re-
quired.
The findingof whiteimmediatelyover the sandy-coloredlayer
on the added pieces,ratherthangray,suggeststhatthe canvaswas
enlargedat a timedistinctfromthepreparationof thecentralpiece.
This suggestionis reinforcedby the observationin the X-radio-
graphsthatcuspingof the yarnsoccurs at the edges of the central
section.Cusping along the edges would result fromthe central

57 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
 piece havingbeen stretchedand groundedas a separateunit.A cross
sectionmade by Mr. Buck of a specimenof thered draperyjustto
theinsideof theseamat thetop confirms theorderof construction:
tan-coloredground,grayimprimatursred paintover whichthere
is a layer of white which must have been pulled across the seam
onto thered of the centralpiece whilethewhitegroundwas being
applied to the added strip;thisis followedby an uppermostdesign
layerof red (Buck, fig.3) .
Evidence suggeststhatthe added pieces are contemporarywith
the centralsection,perhapsadded followingthe layingout,if not
the completion,of the compositionin the centralsection.Cross
sectionsof samplesfromtheoutermostedgesindicatethatthecom-
positionon the added portionswas executed according to plan;
thereis littleindicationof trialsin designor local color exceptin
portionsof thedraperybehindMme Gerbier(Buck, fig.8). More-
over,as will be describedbelow, manyof thepigmentsare similar;
in particular,lead-isotope-ratiostudiesindicatethatthe lead in the
lead whiteat the edges is identicalto thatin the centralportion.
Some years ago, added sectionsin Rubens's Chapeau de Faille
were foundto have whiteunderthedesignlayer,whereastheprin-
11PaulCoremans,"TheRecentCleaning ofRu- cipal panel had been coated with a thinbrownimprimatura.11 One
bens'ChapeaudePaille"Burlington
Magazine, resultof thisdifference became noticeablefollowingcleaningthat
1948,pp. 257-261.
took place at that time; partsof the designpaintedon the added
sectionsare now lighterin color thanthepassageson themajorsec-
tionsof panel.The whiteimprimatura layeron theadded piecesof
The G erbierFamilymaylikewisepartiallyexplainthelightercolor
of someof thered draperyat theupperleft.

MEDIUM

The analysisof artists'dryingoils and mediumsbased on drying

oils heat-processedwith resinsis stillin its infancy,althoughrapid
progressis beingmade by meansof gas chromatography, thinlayer
chromatography, and mass spectrography. In thepresentcase,small
samples taken from the edges of the painting were set aside for
detailedinvestigation by thesemethods at a laterdate. At thistime
we mustbe contentsimplyto say thatthe vehiclein the paintbe-
haved like a dryingoil in severalrespects:microscopicbitsof the
medium disintegratedunder the action of dilute alkali and the
layersin cross sectionsaccepted an oil-solublestain.In addition,
the hydrogenperoxidetest for the presenceof dryingoil, a test
12H. Kühn,"A StudyofthePigments andthe routinelyusedby Kühn,provedto be positive.12
Grounds UsedbyJanVermeer," and
pp.155-202; At a meetingin Lisbon, October 1972,JohnMills of the Na-
Kühn,"Untersuchungen zu denMalgründen
Rembrandts,"pp. 189-210.
tional Gallery,London, reportedthatthe dryingoil provedto be
linseed oil in the only paintingby Rubens that he had thus far
13J.S. MillsandR. White,"TheGas-Chromato-examined
ExaminationofPaintMedia,"Con- by gas chromatography.13
graphic Attempts to color thecrosssectionswitha proteinstainrevealed
servation
ofPaintingsand the Arts
Graphic , Pre-
oftheLisbonConference
prints London, only
1972, one layer that seems to be aqueous in character;a size im-
Institute
International forConservationofHis- mediatelyin contact with the canvas. As mentioned,the sandy-
toricandArtisticWorks,pp.701-703.colored
groundgave no indicationof protein.
We tentatively concludethatthemedium,or paintvehicle, in the
ground-imprimatura and designlayersis a dryingoil.

58 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
PIGMENTS

If we may judge fromthe numberof pigmentsfoundin The De-

scentfromtheCrossand fromthenumberreportedto us by Kühn
following his examinationof about thirtypaintings,Rubens's
palettemay be characterizedby the employmentof a wide variety
A listof thepigmentsfoundin thepresentinvestiga- 14H. Kühn,Doerner
of pigments.14 Munich,
Institute, private
tion and those previouslyreported by others is presentedin communication.
Table I.15 15E. Raehlmann, derMalerei
ÜberdieFarbstoffe ,
deWild,TheScientific
andA. Martin
Leipzig,
Blue ofPictures
Examination ,London,1929.
Four blue pigmentsare presentin The G erbier Family. Of these,
smalt,azurite,and verditerwere the artist'sprincipalchoices,with
a traceof lapislazuli (naturalultramarine)in thefoliageoverhead.
The sky consistsprimarilyof lead white colored with black-

TABLE I
byRubensCompared
Employed
Pigments to ThoseFoundin The Gerbier
Family
Investigator

I
s
M-1
RS! s Í».
§ ^ g -a §.
■ss le B 2
$ O fci n 3
gooses
Blue Smalt • • • •
LapisLazuli(Natural
Ultramarine) • • •
Azurite • • • •
(var:Verditer) •
Indigo • •
Green Malachite •
ofBlueandYellow
Mixtures • • •
Verdigris • •
Yellowand Brown YellowOchre • • • •
Lead-TinYellow • •
YellowLake • •
Red Ochre(RedEarth) • •
Van DykeBrown •
Red Vermilionor Cinnabar • • • •
RedLead • • •
Red Lake • •
Madder • • •
White LeadWhite • • • •
Lead Carbonate •
Chalk • • • •
Black PlantBlack • • • •
Asphaltum •
IvoryorBoneBlack •

59 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
 Forcolor , seepage47.
Fig. 2.Azurite indarkgreenshadowofskirt . Atleftisa particle
ofplant
black , atright a particle
ofpalesmalt.Yellowcolorbetween theparticles
of
azurite maybelake.Tra?tsmitted ,inPermount®,
illumination ND1.52.
H 23inches , W 31inches. SampleGP10.Magnification: 1660x.
Fig.3. Verditer,yellowlakeandchipofsmalt from greenofdress.Trans-
mitted lightinPermount®, ND1.52.H 24%inches , W 14%inches. Sample
GP8.Magnification 1660x.

60 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
 -
probablyplantblack and smalt.The latteris essentiallyground-
16B. Muhlethaler
andJ.Thissen, Studies
up cobalt-blue glass.16The chips of smalt are in many cases so "Smalt,"
that , 14, 1969,pp.47-61.
in Conservation
verypale theyare in dangerof beingoverlookedwhensamples
are dispersedfor microscopicexaminationby transmittedlight.
Nonetheless,the chips and flakesmay be clearlyseen in the cross
sections(Buck, figs.7 and 8).
Cobalt was detected by emission spectrographyin "trace"
amountsin two samplestaken fromthe sky,and at a highercon-
centrationin a thirdsampleanalyzedby Keisch by meansof neu-
tronactivation.Through comparisonwith specimenspreparedat
the Doerner Institute,the compositionof many particlesof smalt
in the paintingmay be judged to be about 1 to 3 per cent cobalt
oxide (CoO) and 80 to 85 per centsilica (Si02), perhapssimilarto
thevarietydesignatedby Riedereras "eschel."17Many of thesmalt- 17J.Riederer,"Die Smalt,"
DeutscheFarben-
likeflakesare of such low color thatMr. Buck and I began to won- ,9, 1968,
Zeitschrift We aregrateful
pp.386-395.
toDr.Kühnforproviding a number
ofsamples
deriftheyhad not been added at timesmoreto provide"tooth"or ofsmaltprepared byDr.Rie-
inthelaboratory
"body" to the paint than color. This possibilityparticularlyoc- derer.
curredto me when colorlessflakesoccasionallywere observedin
the imprimatur a layer and the sandy-coloredground (Buck, fig.
8).18 18Dr.EinbeckoftheIntermuseum Laboratory
Smalthasbeen reportedin two of Rubens'spaintingswhichhave (Oberlin)haskindly drawnourattention
topas-
beensubjectedto detailedexamination sagesindéMayerne thatrecommendtheaddition
by Kühn. Coremanswentso ofpowdered glasstoincrease
easeofdryingof
faras to say that"it was a favoritepigment"of the artist;thiscer- certain E. Berger,
pigments. Ms.pp.18,153verso.
tainlyseemsto be the case in The G erbier Familywhere smaltis
foundin practicallyall of thecool colors.
Azuriteis the name givento the blue-coloredbasic copper car-
bonate compound having the chemical formula, Cu(OH)2*
2CuC03; it is also the name assignedto the naturalmineralof this
Identifiedby microscopic examinationand con- 19R. J.Gettens "Azurite
andE. W. FitzHugh,
composition.19
firmedby X-raydiffraction, thiscompoundwas foundin thegreen Studies
andBlueVerditer," ,11,
inConservation
skirtand in the dark green detailsof the background.Figure 2 1966,pp.54-61.
shows an excellentexample,in all likelihoodthe naturalmineral
taken fromthe bluish-greenlandscape (at 63 y2 inches W, 26%
inches H).20 Azurite has been observed in thirteenof Rubens's 20Locationofsamplesisgivenintermsofthe
X andY Cartesian
coordinates in
withtheorigin
paintingsexaminedby Kühn, in The Descent fromthe Cross, and thelowerleftcorner:
X (width),Y (height).
in two of Rubens'spaintingsexaminedby de Wild.
Blue verditeris the manufactured varietyof azuritewhich gen-
erally possesses a particleshape: rounded nodules or
distinctive
rosettes(fig. 3). Harley statesthatverditerwas a fairlynew pig-
mentin 1600.21She also points out that the naturalmineraland 21R. D. Harley, '
Artists c. 1600-183
5,
Figments
verditeroftenexhibita greenish-bluehue. Many of the particles London,1970.
observedin our sampleswere in factso greenor greenish-blue by
transmitted lightit was some timebeforeI became convincedthat
theywere azurite.
Evidence forthe presenceof lapis lazuli,a naturalblue mineral
of thezeolitefamily,is based upon microscopicexamination.22 Iso- 22JoycePlesters, Blue,Natural
"Ultramarine and
lated bits of isotropicblue particleswhich had the color, shape, Studies
Artificial," ,11, 1966,
inConservation
refractive to acid similarto thatof lapislazuli pp.62-91.
index,and sensitivity
were observedin a sampletakenfroma leafat thetop of thepaint-
ing (52 inchesW, 53% inchesH) and froma blackish,glazelike
paintin the rightbackground(10 inchesW, 22% inchesH). As
will be noted later,the syntheticproduct,ultramarine, was found

61 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
 in the surfacevarnish;itsphysicalcharacteristics
are distinctly
dif-
ferent.

Green
The principalgreenpassagein thepaintingis Mme Gerbier'sskirt,
althoughdark greens also appear in the background.The artist
achievedthiscolor by mixturesof blue and yellow,thelatterbeing
providedby ochre,lead-tinyellow,and a lake.
Initial inspectionof samplestaken fromthe skirtrevealedthe
major pigmentsof highchromato be azurite,smalt,and verditer,
with lead-tinyellow used in the highlights. We were perplexedat
firstto findso much thatwas blue in such a decidedlygreenpaint.
However, when additionalcrosssectionswere prepared,and when
bits of materialwere examinedunder the microscope,we soon
sensedthata yellow lake was present.More will be said about this
yellow below.
A crosssection(Buck, fig.7) revealedthatthegreenskirtconsists
essentiallyof one layer of paint,about 76 micronsin thickness
(0.003 inch), over the ground and gray imprimaturalayer. The
principalpigmentsin the green paint are coarse, dominatedby
blackish-appearing "chips"thataresurroundedby smallergreenish-
blue particles.Because the blue color of manyof the dark"chips"
was not obviousand because the yellow color- whichwould con-
tributein a majorway to thegreenhue- was notuniformly intense
the
throughout layer, it was decided to examine a cross section in
the electron-beammicroprobeand the scanningelectronmicro-
scope (SEM) to investigate thecharacterof thepigmentsin greater
detail.
In the electron-beammicroprobe,a beam of electronsis made to
scan acrossthe sampleand to generateX-rayscharacteristic of the
23Useoftheelectron-beam microprobe inthe elements that are The can be "tuned" to detect
present.23 probe
ofartobjectswasintroduced
analysis byR. E. X-raysemittedby a particularelementand to displaytheirlocation
The
Ogilvie. application ofthemicroprobe and
theSEM isnowso common thatI shallnotat- on a screensimilarto thatin a televisionset.
tempttocitethemanyreferences toitsusein The resultsof the examinationof the particularcrosssectionof
museum work.Foranearlydescription ofthe the skirtshownin color in Buck
techniqueseeR. E. Ogilvie,"Electron Micro- figure7 are assembledin figures4
ofPaintSamples from theBershah Sar- and 5. Firstof all, the microprobescan revealsthatcopperis present
analysis
cophagus"inApplicationof Science inExamina- in the specimen at a concentrationjust as greatat the top of the
tionofWorksofArt, Boston, Museum ofFine as it is at the
n.d.(c. 1971),pp.223-229.greenlayer bottom,a factsuggestingthatthe original
Arts,
color may have been much the same at the top as at the bottom.
Lead is clearlyconcentratedin theimprimatura layer.Silica,which
one expectsto findin smalt,is concentratedin the vicinityof the
large "chips."
The selectivity of theinstrument was notsufficient in thisanalyti-
cal mode to pin down the locationof cobalt.However, itslocation
was preciselydeterminedwhen the electronbeam was scanned
acrossthe paintlayerin the regionmarkedA-A in whatis called a
line-scanmode. The resultantdisplay,shown in figure5, conclu-
sivelydemonstrates thatsilicaand cobalt occur together,as would
be expectedif manyof the dark"chips" were smalt.A supplemen-
taryscan forcopperindicatesthatpigmentscontainingthiselement
occur on eithersideof theflakesthatcontaincobalt.
A photographof thissamecrosssectiontakenwiththescanning

62 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
Fig. 4.Examination ofright-hand portion ofBuck,figure 7 byelectronbeam
microprobe andscanning electron microscope : ( a) thecrosssectionviewed
byreflectedlight ,magnified 340x, ( b) practically thesameareaviewed
byscanning electron microscope, ( c) absorbed-electron displayofsample
underelectron beammicroprobe , ( d) location ofsiliconinthesample, an
element insmalt , ( e) location
ofcopperinthesample , anelement inazurite
andverditer, ( f) location ofleadinthesample . d,e andfaredisplays of
intensityofresponse toX-radiation from thechosen element intheelectron
beammicroprobe .

electronmicroscoperevealsthethree-dimensional characterof the

paint;the coarse of
particles smalt and azuritestand out in marked
contrastto the finerpigmentsin the imprimaturalayer (fig.4).
The nodularshape of the particlesof verditeris also clearly dis-
played.
The examinationof the greenlayerby meansof themicroprobe
demonstrated two principalpoints: (1) thatthe concentrationof
copper is much the same at the top of the layer as at the bottom
and (2) that many of the blackish-appearingflakesare indeed
smalt,because cobalt and silica occur in the same particle.Smalt
appearsblackratherthanblue whenthecrosssectionsare examined
by reflectedlightbecause it is relativelytransparent;lightis re-
flectedso much less by thismaterialof low refractiveindexthatit
appearsblack in relationto many of the neighboringpigments.
Yellow and Brown
At leastfouryellow and brown pigmentsare present:yellow and
brown earthcolors (probablyochres), lead-tinyellow,and a yel-

63 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
 Fig.S.Line-scan
traces
ofscansacrossthecrosssection 4
GP26infigure
alonglineA-A todetectthelocation ofcobalt andcopper.
,silicon The
presenceofbothcobaltandsilicon intwolocations thepresence
implies of
smalt(a cobalt-containing
glass).

low lake. Yellow ochre seemsto have been used ratherextensively

throughout,althoughthis finelydivided pigmentis not easy to
identifywith assurancewhen it is mixed with severalotherpig-
ments.The detectionof iron,silicon,and aluminumby emission
spectrographicanalysisprovides supportingevidence that earth
color is presentin samplestaken fromthe sky,the red drapery,
and darkgreensin thebackground.Ochres are oftenfurther char-
acterized by the presence of certainassociatedelementssuch as

64 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
titanium;traceamountsof thiselementwere detectedin a sample
fromtheochre-coloreddetailat thetop ofthepergola.
Authorshave widely assumedthat umber,an earthcontaining
severalper centof the elementmanganese,has been popularlyem-
ployedin oil paintingowing to its usefulnessin dark shadowsand
as a pigmentthathastensthedryingof oil paints;however,we have
no indicationthatRubens employedumber.Kühn did not report
findingumberin any of the paintingsthat he examined,nor did
Coremansand Thissen findmanganesein any of their emission
spectrographic analyses.Likewise,we were not able to detectman-
ganese in the spectrographic analysisof morethana dozen samples
examined.On the otherhand,iron was foundin varioussamples
and isolatedbitsof particleswere observedthatresembledthe red
oxide of iron,haematite(Fe203), evidencesuggestingthatred or
brownochre,or perhapsburntsienna,was includedin the palette.
Severalhighlights in theskyprovedto be practicallypureyellow
pigment, which was identifiedby X-raydiffraction as lead stannate
(Pb2Sn04), the pigment lead-tinyellow.24 Attention was also given 24H. Kühn,"Lead-tin Studies
yellow," inCon-
to the yellow of the brocade in the skirt,a detailof interestto art servation , 13, 1968,pp.7-33.
historian colleaguesbecauseit does not appearin therelatedWind-
sor picture(Stechow, fig.7). The questionmay be raisedwhether
the brocade could have been added at a much laterdate,although
this possibilityseems remote because the patternoccurs in the
gouache copy of 1749 and the mezzotintof 1755. Moreover,as
Buck has noted,thisappearsto be thesamedraperythatis wornby
the figureidentifiedas Wealth in Rubens's War and Peace at the
NationalGallery,London.
We were pleased to confirmby X-ray diffraction analysisthe
presence of lead stannate, along with traces of lead white, in the
on the skirt.25
The 25Examination
yellow detailsof the brocade opaque yellow in undera binocular microscopeat
the fringeof the draperybehind the chair,a color that at first about20xmagnification thatthisopaque
indicates
yellowpigment isalsousedinthehighlights
of
glancemay appearto be based upon ochre,also provedto be lead- theskirt.
tinyellow.
This pigmenthas a fascinating story;authorities on paintingma-
terialswere not even aware of its existenceuntil 1941.26All evi- 26R. Jacobi, "ÜberdeninderMalerei verwende-
dence suggeststhat lead-tin yellow entered the artist'spalette tengelben Farbstoff derAltenMeister,"
Ange-
wandteChemie , 59, 1941,
pp.28-29.
around 1300,but forsome unknownreasonits use was abandoned
about the middleof the eighteenthcentury.Hence, the findingof
lead-tinyellowin a painting - a in boththe
pigmentlong forgotten
artist'sand conservator'sworkshop- is generallytaken to imply
thatthepainthad been appliedpriorto themiddleof theeighteenth
century.27 The presenceof this pigmentin The Gerbier Family 27Unless, ofcourse, maybeevidence
there to
thus makes the particularpassages that contain it very much in suggest thatthepainting mayhavebeenfabricated
since1941.
keepingwith what one would expect to findin the work of a
seventeenth-century artist.Furtherconfirmation thatthesedetails
are an integralpartof the originalworkwas providedby an analy-
sis of the lead-isotopecontent- the lead in the lead-tinyellow
was shownto be close to thecompositionof thatfoundin boththe
lead whiteand thered lead.
Kühn reportedfindinglead-tinyellow in thirteenworksby Ru-
bens. There are two types,in his opinion,one varietyoftencon-
taininga significant portionof stannicoxide. Three samplesana-

65 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
 lyzed fromthe Gerbierpaintingall containedstannicoxideas well
as lead stannate.
A lake is a pigmentpreparedby theabsorptionof a dye upon an
insolublesubstratesuch as chalkor clay. The organiccoloringmat-
ter (the dye) in a lake is not readilyidentifiableunlessthesampleis
of sufficient size to permitsome of the colored substanceto be
extracted.Yellows pose a furtherproblembecause theirprincipal
absorptionof lightis in the ultravioletor at the very edge of the
visibleregion;the resultis thatthe normaldiscolorationof oil and
varnishmayinterfere withpositiveidentification.
As mentioned,Mr. Richard Buck and I only graduallybecame
convincedthata yellow lake was present.One bit of evidencewas
knownin advanceto suggestthatRubensmay have employedsuch
a pigment.Hilaire Hiler, in discussingartist'spalettes,mentions
havingseen a collectionof pigmentsthatonce belongedto Rubens
28HilaireHiler,NotesontheT echniqueofFaint- in which a yellow lake had "faded almostentirelyaway."28We
rev.ed.1954,
,London,
ing, p. 137. were particularlypleased to have Dr. Kühn informus thathe has
observeda yellow lakein morethana half-dozenof Rubens'spaint-
ings.
Such a lake could have been derivedfromany numberof yellow
organic substancesthat are well describedby Harley and East-
29C. L. Eastlake,
MethodsandMaterials Paint-
of lake.29Perhapsthe mostlikelycandidateis the pigmentknownas
ingsoftheGreatSchoolsandMasters,NewYork, Dutch pink, a color made frombuckthornberries
vol.1,p. 438. (genus Kham-
nus) or a
possibly variety of broom ( Genestella tinctoria).Ac-
to
cording Harley, all seventeenth-century sources suggestthat
Dutch pink was a yellow,the mainuse of which was to providea
greenwhen mixedwith blue. Norgate,forexample,mentionedits
30Harley. mixturewith blue bice or verditer.30 De Mayerne also noted the
31Berger,Ms.pp.6 verso,
9, 22. mixture of azurite and Dutch in several
pink places.31
In partiallydispersedspecimenstaken fromthe greenskirt,we
foundthe yellow clusteredabout the lumpsof verditerand along
theedgesof thechipsof smalt(fig.3) .

Red
Thè reds are achieved by various combinationsof cinnabar,red
lead, and red lake, modulatedby the use of lead white and plant
black. Cinnabaris the mineralogicalname for the red varietyof
mercuricsulfide(HgS), vermilion , thenamegenerallyreservedfor
32R. J.Gettens,
R. L. FellerandW. T. Chase, the syntheticvariety.32 The cinnabarused throughoutthepainting
"VermilionorCinnabar,"StudiesinConservation
, containsbroken whichmayeitherbe thenaturalmineral
17, 1972, fragments
pp.45-69. or
pulverized bits of sublimed syntheticmaterial( dryprocessver-
milion). It is not yet possibleto distinguishthe two. In additionto
the broken fragments, finelydivided particlesless than about 1.5
micronsin diameterare also observablein almosteverysample.The
presenceof thesefineparticlesof ratheruniformdiameter,and the
lack of extensivecolorlessmineral"debris"in mostof thesamples,
suggeststhat the artistwas using synthetically preparedpigment
ratherthanthe naturalmineral.The Dutch had the reputationof
preparingexcellentdry-processvermilionand, by the seventeenth
century,the Netherlandshad become the principalEuropean cen-
ter formanufacture.
About a thirdof the samplesof red paintcontainedtracesof a

66 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
reddishlake.The coloringmattermaynothavebeensolelymadder,
thepigmentspecifiedin the findingsof Coremansand Thissenand
of Kühn,because fluorescence, a characteristicof madder,was not
invariably observedwhen the painting or an appropriatecrosssec-
tionwas examinedunderultravioletlight.Cochineal,lac, or kermes
could have been used forat leastsome of the lake. The artistem-
ployed the red lake as a glaze in some of the broad expansesof
drapery.In addition,its applicationin what may be considereda
"traditional"technique,thatis, as a glaze over a layerof the more
opaque vermilion, was seenin a crosssectionof theboy's shirt.33 33Examinationofthissamecrosssectionsuggests
Red lead (Pb304), at least in trace amounts,was found rather thattheboy'sreddish
shirtwasapplieddirectly
overthegrayimprimatura, notoverskyorfoliage
widely throughoutthe designlayers,althoughnot in the ground. asmighthavebeenthecaseifthefigure hadnot
The artistmixedvermilionand red lead withblack in the shadows planatanearlystage.
beenintheartist's
of thedrapery,and withlead whitein thelighterareasat theupper
left.
De Wild did not findred lead in the Dutch paintingsthat he
examined.We were,therefore, interestedin confirming the use of
thispigmentin paintingsby Rubens and were pleased to findthat
Plestershas reportedred lead in two of the groundsin the four
Rubens'spaintingson canvasthatshe examined.Kühn informedus
thathe foundredlead in The Slaughterof theInnocents, No. 572 in
theBayerischenStaatsgemaldesammlungen. Supplementing thisin-
formation,Keisch's measurementsof the lead-isotoperatios re-
vealedthatthe compositionof the lead in the red lead is similarto
thatfoundin thelead whitethroughoutthepainting.Thus, in spite
of de Wilďs not havingfoundred lead, we have no reasonto be-
lieve thatthe paintthatcontainsthispigmentin the draperyand
elsewhereis notpartof theoriginalwork.
Harleystatesthatred lead,althoughmentionedfrequently in the
early on
literature painting,was not recommended as highlyas
some otherreds; de Mayerneheld thatit lacked permanenceand
shouldnotbe usedin grounds.

Whitef igments
Lead whiteand chalkmay be cited as whitepigments,althoughin
oil paintthelatteris morein theclassof an "inert"fillerratherthan
a whiteof significant hidingpower. The lead whitepresentin the
whitesand in the lightercolors of the drapery,sky,and skirtap-
peared underthe polarizingmicroscopeto containtracesof lead
carbonate,a contaminantthatresultsfromthe ineffective conver-
sion of themetalto themorecommonbasic compound,2PbCOa
Pb(OH)2.34 The presenceof lead carbonatealong with the latter 34R. J.Gettens, H. KühnandW. T. Chase,"Lead
"lead white" compound was confirmedin the X-ray diffraction White"Studies inConservation, 22,1967,pp.125-
129;B. Keisch,
"X-Ray DiffractionandtheCom-
analysisof materialfromfourdifferent areas.
positionofLeadWhite,"1971,1972Studies inthe
Chalk is used extensivelyin the ground-imprimatura layer, as History ofArt,National GalleryofArt,Wash-
well as in some of the more transparentdark paintsof the back- ington, 1972,pp. 121-133.
ground.Chalkmay be differentiated frompulverizedfragments of
themineralcalcite (CaC03) by the presenceof coc eoliths,minute
calcariousremainsof unicellularmarinealgae thatarereadilyrecog- 35The first mention ofthepresence ofcoccoliths
nizableundera polarizingmicroscope.35 We may characterizethe inchalkgrounds wasbyP. Coremans, R.J.Get-
tensandJ.Thissen, "La techniquedesPrimitifs
chalk in The G erbier Family as containingan especially wide Flamands," StudiesinConservation , 7,1952-4,
variety of these ,
nannofossils (fig.6). pp. 1-29.

67 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
 Fig.6.Scanning
electron
microphoto ofcoccoliths
graphs inchalkground.
cretacea
LeftPrediscophaera (?) RightMaslovella
barnesae
(?). Magnifica-
tion:lOßOOx.
(Photocredit: M. Haller)
Coccoliths undercrossedNichols,Magnification:
photographed 2500x.(Photocredit:R. L. Feller)

68 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
Fig.7.Plantblackandleadivhite fromgrayinimprimatura
layerincentral
sectionofpaintingH 58inches , W 25V%inches.
SampleGP19B.Magnification:
1380X.(Photocredit: R. L. Feller
)
The preciseidentificationof nannofossilsin thechalk,madepossi-
ble todayby rapiddevelopmentsin the fieldof micropaleontology,
particularlytheapplicationof electronmicroscopy,promisesto be
a valuablemeansof characterizingthisubiquitousartists'material.
We are gratefulto Dr. B.L. Shafferof the Gulf Researchand De-
velopmentCompany, Houston, Texas for identificationof the
species presentin these samples.He kindlyinformedus that the
assemblageof nanofossilsfoundin samplesfromthe groundin this
paintingindicatethatthechalkdepositis mostlikelyof Campanian
36Thespecies byDr.Shaffer
identified bymeans
age (late Cretaceous).36
ofopticalmicroscopy were:Maslovella barnesae ,
Mietilastaurophora, Prediscosphaera cretacea ,
Black Z ygodiscusdiplogrammus, Z. lacunatus, Biscutum
The predominant black pigmentseemsto be plantblack,whichhas blacki
, Arkhangelskiellaparca, Eiffellithustur-
, Microrhabdulus
riseiffeli decoratus, Tetralithus
the appearance of opaque flakesunder the microscope (fig. 7). ovalis , Tranolithus , Bidiscus
phacellosus sp.,and
Kühn reportedonly thisblack in elevenof Rubens'spaintingsthat Cretarhabdus threearethemost
sp.The first
he had examined,althoughhe mentionedthatasphaltumwas also prominent inthesample.
Sincethepreparation ofthismanuscript, I have
detectedin severalsketches.Asphaltum(bitumen) would be diffi- learned from Dr.K. Perch-Nielsen oftheUniver-
cult to identifyin a well-dispersedmicroscopicsample unless it sitetets Mineralogisk-Geologiske Instituter, Co-
were presentin relativelyhigh concentration.Nonetheless,trans- penhagen, that shehasrecentlydescribed ina
Norwegian periodicaltheanalysesofcoccoliths
parentbrownish-black particleswere occasionallyobservedin our foundintwenty-seven medieval Norwegian art
in a from the dressof the oldest objects.I amverygrateful for thecopyofher
samples,particularly specimen girl. inEnglishwhichshekindly
These particlesmay well be asphaltumratherthan the similarly manuscript provided.
transparent(althoughnot similarlyshaped) ivory or bone black,
because phosphorous,an importantconstituentof the latter,was
notdetermined in anyof thespectrographic analyses<ofpaintspeci-
mensfromthedesignlayer.

69 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
 Fig.8.Infrared ofshellaccompared
spectra obtained
withspectrum from
pigmented foundonthepainting.
surface-coating

VARNISH

When the paintingwas received,it was coated with a pigmented

varnish,brown to reddish-brownin color. A number of areas
(Buck, fig.14) were pigmentedmuchmoreheavilythanothers.In
orderto allow forvariationsin the ease of removalof the surface
coatingand the sensitivity of a particularcolored paintbelow, the
varnishwas removedonly froma limiteddomainat any one session
of thework schedule.Test swabs were examinedregularlyand the
collectedswabs fromareasB, F, G, H, N, M and O (Buck, fig.14)
were sentto the laboratoryto have the mineralcontentseparated
and subjectedto detailedexaminationby emissionspectrographic,
X-ray diffraction, and microscopicmethods.By meansof infrared
spectroscopy, the principalsoluble materialwas demonstrated to
be shellac (fig.8).
Microscopic examinationof the separatedmineralsdiscloseda
light-coloredearthpigmentmixedwithpracticallycolorless"inert"
components.Emissionspectrographic analysisof themineralmatter
separated from five of the sets of swabs proved that,except for
slightvariations in relative
concentration, the sametwelveelements
were presentin all samples.This evidence suggeststhat the pig-
mentedvarnishwas basicallythesame over the entiresurface.The
detectionof iron,silicon,and aluminumalong with tracesof bar-

70 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
Fig.9.Spectral curveofvarnishes
reflectance ivhenplacedoverstandard
whitesurface
( bariumsulfate).

ium,chromium,nickel,and magnesiumis just as one would expect

if earthpigmentswere present.An infraredspectrogramof the
mineralsindicatedconsiderablesilicatecontent.
When thepigmentsextractedfromthevarnishwere collectedon
filterpaper,we were surprisedto findtheircolor to be primarily
gray.Even fromthe swabs takenfromthe decidedlybrown areas,
thecolor of theisolatedmineralswas nevermoreintensethandark
brown.Thus, the pigmentsdid not explainthe distinctlyreddish-
brownhue of the skirtand otherareas.It was thenthatthe partic-
ularly dark color of the shellac extractedfrom the swabs was
noticed.The extractswere subsequentlycollected,concentrated,
brushedout,and driedat twenty-five to thirty-eight
micronsthick-
ness (0.001-0.0015inch) on colorlessglass sheets.When dry,the
dark-coloredfilmwas placed over a barium-sulfate-white reflec-
tance standardand the spectralreflectancecurve measured.The

71 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
 results(fig.9) convincinglydocumentthe distinctly coloredchar-
acter of the varnish,farmoresaturatedin hue thanthe usual "dis-
colored picturevarnish."Accordingto the ISCC-NBS Method of
DesignatingColors and a Dictionaryof Color Names (NBS Cir-
by the
cular 553), the standarddesignationof the color illustrated
curve fortheshellac (fig.9) is "deep orangeyellow,"equivalentto
Munsellcolor 8.5YR/6/12.Obviously,theshellacresincontributed
to the color of the reddish-brownskirtand othersectionsof the
paintingas much as theadded pigmentsin thevarnish,if notmore.
Convincingevidence that thispigmentedvarnishhad been ap-
plied to thepaintingin moderntimeswas providedby thetracesof
artificialultramarine,a pigmentfirstplaced on themarketin 1828,
foundin the materialfromeveryset of swabs: rounded,intensely
blue isotropicparticles,easilydecolorizedwithmildacid and hav-
ing a refractiveindexslightlyabove thatof the mountingmedium
on themicroscopeslide (1.52).

THEFORMERLY
BROWN
SKIRT

When the paintingwas received,the brown hue of the skirtcom-

mandedattention.This was a matterof particularconcernbecause
old documents,adduced elsewherein thisissueof Studies, spokeof
theskirtas beinggreen.Let us reviewthechainof evidencethatled
to the uncoveringof the greenskirt,as it is presentlyseen.Exam-
inationunderlow-poweredmagnification revealedthatthe domi-
nant color was achieved by a semi-transparent, almost reddish-
brown coating which proved principallyto be the varnishjust
described.At the time,the originalhue of the skirtcould be in-
ferredfromtracesof color thatcould be seen beneaththe varnish,
but when it was learnedfromProfessorStechow thatthe painting
had once been describedas portraying"a lady in a greenand white
dress",we immediatelysetabout to learnas muchas possibleabout
thepaintbelow thesurface.
A crosssectionindicatedthebrownto be verythinand thatblue
pigmentsexisted throughoutthe principaldesign layer. As de-
scribedearlier,a detailedinvestigation withtheaid of theelectron-
beam microprobedemonstrated thatcopper pigmentsin highcon-
centrationexistedrightto thetop of thatprincipallayer (fig.4). A
yellow lake,seen primarilyin the lower partsof the crosssection,
also contributedcolor (Buck, fig.7) ; perhapstheyellowwas at one
timemore prominentnearthe top but is now faded.There was no
evidenceof a coloredglaze overthegreenpaint- justthethinlayer
of reddish-brownvarnish.
When solvencytestswere made,Mr. Buck foundthatmuch of
the brown layer had a lower resistanceto solventaction thanthe
greenand had a much lower concentrationof pigment.Moreover,
when a testarea of thebrownvarnishwas removed,thegreenpaint
below was foundtö be completelysatisfactory in appearance.
Thus, we were confrontedwitha shellac-basedcoatingthatcon-
tainedsyntheticultramarine. The brown color did not correspond
to descriptionsof the paintingprior to Roger Fry's remarksin
37Buck,note16,p.51. 192737;moreover,it was
especiallyheavilytintedin specificareas
72 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
(Buck, fig.14). These observationsprovidedthe majorsupporting
evidenceto justifythe removalof the coating. This was accom-
plishedin stages,the curatorialand conservationstaffthoroughly
consideringtheappearanceand conditionof thepaintat each stage
beforeproceedingfurther.

SUMMARY
ANDCONCLUSIONS

The examinationsconducted by both the conservationand the

scientificlaboratorieshave proventhe paintingto be in excellent
condition,exceptforslightabrasionsin certainareasand minorre-
pairsalong the seams.A diligentsearchformajor areas of "over-
paint"or additionsthatmighthave been appliedin the nineteenth
and twentiethcenturiesfailedto disclosepigmentsthat were ob-
viouslyof thisdate,exceptforsyntheticultramarine in the decid-
edly reddish-brown surface varnish based on shellac.This, and the
supplementary evidence cited above, was considered sufficientto
justify the removal of the heavily tintedvarnish. The resultis a
visuallyrewarding return to the historicallydocumented green
skirtand the harmonizingdark greensin the background.
The investigations have revealedthatthe imprimaturalayer on
the added pieces is whiteratherthangray,as foundin the central
portion;moreover,theyarnsat theedgesof thecentralcanvasbear
the strainlines of prior stretching.These facts suggestthat the
pieces at the sides were added afterthe originalcompositionwas
established.The technicalevidenceindicatesnothing,however,to
suggestthatthe added pieces are not partof theoriginalwork pre-
sentedto the Gerbier family.For example,paint applied on the
added sectionsappearsto containthe same pigmentsas were used
in thecentralportion- particularly thesmalt,red lead,lead-tinyel-
low and lead white.Measurementsof lead-isotoperatiosprovide-
we believeforthe firsttimein art historystudies- extraordinarily
firmevidencethatall partsof thepaintingrepresenttheproductof
an individualstudio.
In the course of technicalexaminationour understanding of the
work of an artist'satelierof thisperiodhas been furtheradvanced.
The employment of a glue size is indicated,followedby a relatively
transparent pigmentedgroundand moreopaque imprimatura layer.
The size was detectedby use of a dye thatstainsprotein.The pig-
mentsare also of considerableinterest:verditer,red lead,and espe-
cially the lead white,now well characterizedby means of X-ray
diffraction and lead-isotopeanalysis;also the chalk,characterized
according the speciesof coccolithspresent.Pale smaltis noted,
to
employedso extensively thatit may have been used to enhancethe
drying of certain paints to givethembody.
or
Finally, we consider thatmanyof thegreenswere made through
mixturesof a yellow lake with mineralblue pigments,and have
learnedin thecourseof theseinvestigations thatKühn has observed
a yellowlakein a considerablenumberof paintingsand oil sketches
by Rubens.If it is indeedthecase,as manyof theoldertextsimply,
thatpaintersof severalcenturiesago frequentlyemployedorganic
yellows,it is importantthatthe historianfullyappreciatethisfact
73 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions
 because it is well knownthatsuch colorantstendto fade.Unfortu-
nately,the presenceof yellow lakeshas rarelybeen reportedin the
courseof technicalexamination, to detectand iden-
forit is difficult
tifythemwithassurance.Perhapsbetterthanany otheritemin the
report,our being obliged to statesimplythatunidentified yellow
and red lakeswere presentdramatizeshow muchstillremainsto be
learnedconcerningthe paintingmaterialsof the past.

ACKNOWLEDGMENTS

The technicalinvestigationswere carriedout by theNationalGal-

lery of Art ResearchProject at Carnegie-MellonUniversity, Pitts-
burgh,Pennsylvania.Measurements by electron-beam microprobe,
X-ray diffraction,emissionspectrographyand infraredspectrog-
raphywere made respectivelyby Drs. L. Vassamillet,S. Pollack,
E. Hodge and F. Dollish of the Research Services Department;
photographsby scanningelectronmicroscropywereby M. Haller.
Mary Curranpreparedthe cross sectionsillustratedhere and iso-
latedand identified
thevarnishcomponents.
We are particularlygratefulfor the interestand assistanceof
Drs. KennethReed and BernardShafferof the Gulf Researchand
DevelopmentLaboratoriesin characterizing thenannofossilsin the
chalk. We are also gratefulto Dr. HermannKühn forhisunpub-
lishedreporton the pigmentsthathe has foundin the examination
of paintingsby Rubens.Frequentconsultations and discussionswith
R. D. Buck and thestaffof theIntermuseum ConservationLabora-
tory,Oberlin,have been of the greatestpleasureand value.

74 FELLER

This content downloaded from 150.131.192.151 on Thu, 17 Dec 2015 19:16:55 UTC
All use subject to JSTOR Terms and Conditions