Color changes and chemical reactivity in seventeenth

UvA-DARE (Digital Academic Repository)
Color changes and chemical reactivity in seventeenth-century oil paintings
van Loon, A.
Link to publication
Citation for published version (APA):
van Loon, A. (2008). Color changes and chemical reactivity in seventeenth-century oil paintings
General rights
It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s),
other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).
Disclaimer/Complaints regulations
If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating
your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask
the Library: http://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam,
The Netherlands. You will be contacted as soon as possible.
UvA-DARE is a service provided by the library of the University of Amsterdam (http://dare.uva.nl)
Download date: 20 Feb 2018
chapter | 
Da rk ening a s a result of Increased
Tr a nspa rency in 17 th C Oi l Painti ngs
abstract
– T h i s c h a p t e r fo c u s es on i ncreas ed t rans p arency and cons equent
da rk e n i n g o f se ve n t e e n t h - c e n t u r y oi l p ai nt fi lms , as yet anot h er as p ect of met al s oap
f o r m a t i o n t h a t d e gr a d e s t h e a p p e a r ance of p ai nt i ngs . Several p ai nt i ngs are p res ent ed
f ro m t h e c o lle c t i o n o f t h e Roy a l Pi ct ure Galler y Mauri t s h ui s Th e Hague and p ai nt ed
c eil in gs fro m t h e Tr i p p e n h u i s A m sterdam. Reduced s cat t eri ng/ h i di ng p ower, as a
res ul t o f t h e t r a n sfo r m a t i o n o f p a r ti culat e lead wh i t e or lead-t i n yellow p i gment i nt o
a m o r p h o u s le a d so a p s, e x p la i n s t h e i ncreas ed t rans p arency and darkeni ng obs er ved
o n the p a i n t su r fa c e s. T h i s c a n o c c u r i n underlyi ng and s ur face oi l p ai nt layers , but
c a n a l so o c c u r se le c t i ve ly i n sp e c i fic areas as s oci at ed wi t h t h e wood grai n res ult i ng
in dis t u r b i n g st r i p y p a t t e r n s o f d a rk li nes . Th e p h enomenon can affect p i ct ures t o
dif f ere n t d e gre e s, a lso d e p e n d i n g on t h e color of t h e affect ed layer and t h e p res ence
o f a d a rk u n d e r la ye r. A c o lo r c h a n ge due t o los s of t h e lead wh i t e comp onent i s
m o s t n o t i c e a b le i n m e d i u m b row n and gray t ones and, wh en mi xed wi t h s malt , i n
b l ue p a i n t s. A s w i t h t h e lo c a li ze d darkeni ng on t h e wood grai n, t h e t h i cknes s and
a b s o r b e n c y o f t h e c h a lk gro u n d , d e gree of s welli ng and t rans p arency of t h e p ai nt i ng
tec hn i qu e a lso a p p e a r t o b e d e c i si ve fact ors i n t h e fi nal ap p earance of t h e p i ct ures .
introduction
Already towards the end of the seventeenth century, the Dutch painter Gerard de
Lairesse observed the phenomenon of increased transparency with regard to oil paint
layers. He explicitly mentioned it in his Groot Schilderboeck published in 1707 [Van
Eikema Hommes 1998]. In the nineteenth-century literature, authors often remarked
that oil paint will become transparent and grow darker, and that “dark grounds will
devour the colors in time” [Carlyle 2001: 177-78, 261]. There are countless pictures,
where increased transparency of lead-white containing paint films allow underdrawing,
underpainting or artist alterations to become more visible than originally intended (most
pentimenti involve the use of lead white). Although in some cases underlying layers
were meant to be visible, in paintings such as Adriaen Coorte’s Still Life with Asparagus
(Rijksmuseum, Amsterdam inv. no. A-2099) from 1697, there is no doubt that the
underlying dark background that has become disturbingly visible through the asparagus’
tips is due to the lead-white- containing paint of the asparagus having become more
translucent.1
In much of the conservation literature, increased transparency of oil paint has been
ascribed to a rise in the refractive index (RI) of the oil binding medium with age (1.48
1
In addition to increased transparency, tiny craters filled with a whitish substance (corresponding to the phenomenon
of metal soap aggregates) were observed [Van Eikema Hommes 1998: 118 and 2004a: 38; Wallert 1999: 94].
206
for fresh linseed oil compared to 1.57 for a very mature paint film). This was largely
based on experiments in the 1920s and 30s by the British chemist A.P. Laurie [Laurie
1926]. However, for pigments with a high RI, such as lead white (RI: 1.9-2.1), this
rise is not enough to explain the increased transparency observed in many paint films.
Already in 1909, the German chemist Alexander Eibner, who was active with painting
studies in Munich, suggested that increased transparency of lead white paints was a
result of the formation of lead soaps [Eibner 1909: 121]. Gettens and Stout cite the
work of Eibner in this regard [Gettens and Stout 1966: 175]. In 1967, Laurie admitted
that the increase in transparency may in part be due to chemical change, also noting
that when there is darker paint below, the picture will decrease in tone [Laurie 1967:
110, 140-155]. Rees Jones in 1991 calculated the effect of the age-induced increase
in the transparency of oil paint, correctly observing that thinly painted pictures with
a dark ground or underlayer are worst affected [Rees Jones 1991]. Townsend in 1993,
pointed out that the change in refractive index is too small to account for the change in
appearance and that it must be the RI of the pigment that alters, possibly as a result of
a chemical change occurring on the surface of the particles from an interaction between
the oil and the lead pigment [Townsend 1993].
Metal soap-related oil paint degradation has been the subject of recent painting
studies (1998-2007), where the dissolution of lead white in association with lead soap
aggregates was noted in cross-sections of the affected ground layers of Rembrandt’s
Anatomy Lesson of Dr. Nicolaes Tulp from 1632 (The Hague, Mauritshuis inv. no. 146)
[Boon et al. 2002; Keune 2005a]. More recently, it was found that the conversion of
lead white to amorphous metal soap complexes also plays a major role in the increased
transparency of oil paint films [Noble, Van Loon and Boon 2005 and 2007]. Due to
saponification of lead white the light scattering properties of the paint are reduced
and light can penetrate deeper into the paint layer, creating a localized darkening
effect particularly when a non-reflecting paint or ground (or panel) is present below.
In medium to dark paint mixtures, color change is also noticeable due to loss of the
white component in the paint mixture. These results were also supported in a recent
study on metal soap-related color and transparency changes in two nineteenth-century
paintings where localized darkening was ascribed to the formation of lead and zinc
soaps. Here, the degree of darkening was found to be not only dependent on the degree
of saponification, but also on the color of the affected layer and the presence of a dark
underlayer [Shimadzu and Van den Berg 2006]. Furthermore, darkening can occur
selectively in specific areas where it is associated with the ground that fills the uneven
structure of the support. Various causes and names have been touted to describe this
phenomenon, including linear staining, ground staining and variable translucency
[Zucker and Boon 2007]. In panel paintings, it is often generally seen as abrasion or
possibly even as part of the artist’s technique. However, after microscopic examination
it becomes clear that the darkened lines, that are clearly associated with the wood grain,
are not due to exposure of the wood itself, but rather to alteration of the ground layers
that had become transparent and darkened. Not only do these darkened grounds have
a strong visual impact when they shimmer through the overlying layers, but they can
207
Chapter 5
become even more exposed as a result of volume changes in the affected layer that causes
the overlying surface paint in these areas to flake off.
In this chapter, case studies examined (2002-2007) with regard to metal soaprelated transparency changes in seventeenth-century oil paintings are presented.2 The
paintings are listed below (Table 5.1). The metal-soap related degradation includes A:
increased transparency of lead-tin yellow paints, B: darkening of medium brown and
gray tones, C: selective darkening of ground layers associated with the wood grain and
D: darkening/browning of blue paints containing lead white and smalt. Cross-sections
were studied with analytical microscopic techniques, including scanning electron
microscopy coupled with energy dispersive X-ray elemental analyses (SEM-EDX) and
specular reflection Fourier transform infrared imaging (FTIR imaging), to demonstrate
the conversion to lead soaps in the paint layers. Isolated samples of the ground and
paint were also analyzed with direct temperature resolved mass spectrometry (DTMS)
to characterize the binding media. In this chapter, the lower ground is referred to as
the chalk ground and the dull-colored oil bound second ground, the imprimatura. This
build up of preparatory layers, consisting of a single oil-bound imprimatura of lead white
colored with small amounts of other pigments, applied on top of a lower chalk/glue
ground, is typical for Dutch panel paintings of the first half of the seventeenth century,
though pure oil-bound layers also occur.3
Table 5.1 List of paintings examined
2
3
o
Michael Sweerts, Draughts Players, (Mauritshuis, inv. no. 1121), 1652, oil on canvas, 48
x 38 cm
o
Roelandt Savery, Orpheus enchanting the Animals with his Music, (Mauritshuis, inv. no.
157), 1627, oil on oak panel, 62 x 131.5 cm
o
Aert van der Neer, River Landscape, (Mauritshuis, inv. nr. 912), mid-1650s, oil on oak
panel, 44.8 x 63 cm
o
Jan van Goyen, Dilapidated Farmhouse with Peasants, (Mauritshuis, inv. no. 1081),
1631, oil on oak panel, 40 x 45 cm
o
Jan van Goyen, River View, c. 1644-48, (Mauritshuis, inv. no. 759), oil on oak panel, 37
x 64 cm
Technical investigations of the paintings were carried out together with the conservators in the Mauritshuis
2002-2007, and with the conservators involved in the restoration of the Trippenhuis ceilings 2006-2007.
This build-up of ground layers is based on cross-sectional and elemental analyses of many panel paintings from
this period in the collection of the Mauritshuis. This is proving to be in accordance with recipes for grounds from
seventeenth-century documentary sources [Witlox and Carlyle 2005].
208
Da rk en in g a s a re s u l t o f i n c re as e d tr an s p are n c y i n 1 7 th c e n t u r y o i l p ai n t i n g s
o
Jan Steen, Dancing Peasants at an Inn, (Mauritshuis, inv. no. 553), c.1644-48, oil on oak
panel, 40 x 58 cm
o
Unknown artist(s), Painted ceilings with hunting scenes with skies and birds,
(Trippenhuis, Amsterdam, corridors and stairwells), 1660-62, oil on oak
e x p e r i m e n ta l
sample prepa rati on
Samples were selected with the stereo-microscope and embedded in Technovit 2000 LC
mounting resin, a one-component methacrylate that polymerizes under visible blue light
(Heraeus Kulzer, Germany). Cross-sections prepared in the Conservation Studio of the
Mauritshuis were made using Easy Sections (VWFecit, England) using Poly-pol PS230
polyester mounting resin with M.E.K.-peroxide harder (Poly-Service, Amsterdam).4
After initial grinding using silicone carbide (SiC) paper, the surface was dry polished
using Micro-Mesh¡ Sheets (Regular type, grades 1500 thru 12000, Micro-Surface
finishing products inc., Wilton, Iowa, USA).
instrumentati on
A Leica DMRX microscope was used to study the layer structure of the cross-sections.
Specular reflection FTIR imaging of the cross-sections was done using the Biorad
FTS Stingray 6000 system, with a 64x64 pixels MCT Focal Plane Array detector.
Analytical results of the FTIR imaging measurements are displayed as infrared spectra
and false color or gray-scale distribution images of selected absorption bands. The
cross-sections were then carbon coated before examination in a FEI XL30 SFEG high
vacuum scanning electron microscope (SEM). Elemental analyses (EDX) of the samples
were carried out with the energy dispersive x-ray detector from EDAX attached to the
SEM. Backscatter electron images of the cross-sections and information on elemental
composition were obtained. Furthermore, isolated samples were analyzed using DTMS
and/or single-point transmission FTIR to characterize the binding media. DTMS
analyses were carried out using a JEOL JMS-SX/SX102A 4-sector double-focusing mass
spectrometer. Selected sample material was applied onto a Graseby Specac P/N diamond
cell to perform single-point transmission FTIR measurements using a Biorad FTS-6000
FTIR spectrometer extended with a Biorad UMA-500 IR Microscope. See more
technical and instrumental details in the experimental section in Chapter 1.
4
Sample preparation was carried out by the Conservation Studio Mauritshuis.
209
Chapter 5
Fig. 5.1 Michael Sweerts, Draughts Players, (Mauritshuis, inv. no. 1121), 1652. Overall, and detail from the
yellow stocking of the right figure showing patchy translucent areas as a result of saponification of the leadtin yellow pigment.
r e s u lt s
a . i n c re a s e d t r a n s pa re n c y a n d d a r k e n i n g o f
l e a d - t i n y e l l ow pa i n ts
m ic hael sweerts ‘draug hts pl aye r s’     ( mh     )
In Sweerts’ Draughts Players, the yellow stocking of the right figure demonstrates an
unusual patchiness owing to increased transparency of the paint (Fig. 5.1).5 The stocking
is painted over a dark, almost black underlayer that is now shimmering through the
partly transparent surface paint. A cross-section of a paint sample from a similar area
painted with the same pigment, the yellow tassel of the stool on the left, shows the
upper paint layer to contain lead-tin yellow Pb2SnO4 mixed with a little ultramarine and
orange-colored earth pigment (Table 5.2; Fig. 5.2). In this sample, the reddish-brown
earth ground, demonstrated in other cross-sections from the painting, are missing.
Light microscopic and SEM studies of the yellow layer demonstrate that the original
lead-tin yellow paint consists of various heterogeneous phases. A large cluster of lead-tin
yellow particles, visible at the left in the cross-section, as well as a few particles here and
there throughout the layer (Fig. 5.2), exhibit the characteristic ochreish color in UV
5
This was noted during restoration of the painting carried out by Petria Noble in 2001-2002.
210
Da rk en in g a s a re s u l t o f i n c re as e d tr an s p are n c y i n 1 7 th c e n t u r y o i l p ai n t i n g s
BS E
20 μm
l e ad
tin
Fig. 5.2 Cross-sectional analyses: light microscopic image showing translucent areas in the yellow paint (upper left); FTIR spectrum showing lead soap features (upper right); SEM backscattered electron image (center
left) with detail showing less scattering, amorphous areas (center right); EDX elements maps of lead (lower
left) and tin (lower left).
and strong scattering in the backscattered electron images. Accordingly the EDX maps
of these areas demonstrate the elements lead and tin (Fig. 5.2).6 For the most part, the
layer, however, looks far from the normal homogenous paint layer consisting of opaque
yellow particles. Instead the light microscopic image reveals many transparent regions
that correspond to gray amorphous areas in the SEM backscatter mode having a lower
lead density compared to the particles of intact lead-tin yellow having a high lead density
6
Lead-tin yellow type I was established in a sample from the yellow stocking using X-ray powder diffraction. Analysis
carried out by Arie Wallert, Rijksmuseum Amsterdam 2002 at the ICN [Wallert and De Ridder 2002: 45, note 32].
211
Chapter 5
(Fig. 5.2). The affected layer also appears to have expanded. It is notable that no tin is
detected in these gray areas, but rather only lead and large amounts of carbon (Fig. 5.2).
The FTIR imaging data demonstrate dispersed lead soaps throughout the paint layer
by the prominent absorption bands at c.1509 cm-1 and c.1403 cm-1 attributable to the
lead carboxylate group, Na(COO) and Ns(COO) (Fig. 5.2). Absorption peaks for the
C-H bonds, typical of the aliphatic chain of the carboxylate, at c.2941 cm-1 Na(CH2) and
c.2854 cm-1 Ns(CH2) are also demonstrated. In earlier published studies of lead soap,
protrusions/aggregates in lead-tin yellow paints, it was postulated that the excess lead
oxide in the lead-tin yellow pigment, a lead-tin oxide (lead ortho-stannate), reacts with
fatty acids released during drying/ageing of the oil medium to form lead carboxylates
[Boon et al. 2004; see also section on lead-tin yellow in Chapter 2]. The selective
reaction of lead with the free fatty acids can be explained by the fact that the tin(IV)
oxide component is very stable and does not form carboxylates. The spongy, porous
structure of the lead-tin yellow pigment that makes this pigment highly oil absorbent7,
explains the source of fatty acids available for saponification.
Conclusion
It can be concluded that the original lead-tin yellow paint is converted to a partly
transparent metal soap-rich layer leaving some residual pigment. The reduced amount of
lead-tin yellow particles explains the loss of scattering in the upper paint layer that allows
for the dark underpaint of the stocking to shine through. The partial saponification
of the paint also allows the other colored pigments present in the yellow paint layer,
the blue (ultramarine) and the orange (earth pigment), to play a more role in the
color perception of the transparent areas. The loss of scattering and loss of the yellow
component in the paint have led to a darkened appearance of the paint.
d a r k e n i n g o f m e d i u m b r o w n a n d g r ay t o n e s
ro e l a ndt savery ‘orpheus e n ch a nting th e a nima l s with h is mus ic’
 (mh    )
The extreme darkening of the medium brown paint layers in a landscape by Roelandt
Savery -Orpheus Enchanting the Animals with his Music-,8 that cover a large portion of the
picture, has resulted in strong visual distortion of the original light and dark contrasts
and severe loss of detail (Fig. 5.3). The animals, birds and tiny details depicted on top of
this layer are now barely visible to the naked eye. A cross-section of a paint sample from
the distant landscape in the upper left shows the brown layer as intermediate layer below
the green top layer of the landscape (Table 5.2; Fig. 5.4). The brown paint is composed
of lead white, fine carbon black and red and yellow ochres. The translucent areas in this
7
8
During one of the pigment meetings in 2003 held at FOM-AMOLF, Leslie Carlyle pointed out that the oil
absorption by lead-tin yellow is 25%, compared to 5% by lead white.
Treated by Sabrina Meloni during internship in the Mauritshuis (SRAL intern 2002-2003) and documented in her
final thesis [Meloni 2003].
212
Da rk en in g a s a re s u l t o f i n c re as e d tr an s p are n c y i n 1 7 th c e n t u r y o i l p ai n t i n g s
layer exhibit a strong fluorescence in UV. The brown layer is applied directly on top of
the priming layers, a chalk ground followed by a thin gray imprimatura consisting of
lead white, fine lamp black and a few red (earth?) particles. The backscattered electron
images reveal gray amorphous areas in much of the brown layer, apart from some
(residual) lead white pigment (Fig. 5.4). In these areas, only lead and a high carbon
content are detected. The EDX map of lead shows dispersed lead throughout the whole
brown intermediate layer (Fig. 5.4). It seems that most of the original lead white in this
layer has dissolved and converted to lead soaps. In addition, several small protrusion-like
masses are situated in the brown layer (in the right part). These occluded masses show
fine horizontal striations correlating with higher lead densities that can be interpreted
as mineralized lead soaps. The presence of lead soaps is supported by the FTIR imaging
data. The FTIR image of the 1510 cm-1 absorption band, typical of lead carboxylate
Na(COO), demonstrates a high intensity band corresponding to the middle, dark brown
layer (Fig. 5.4).
Another cross-section from the dark background tells a similar story (Table 5.2).
Here, the dark paint layer is the upper, exposed layer. The priming layers are missing
in this sample. The UV image shows many fluorescent zones in the dark paint layer.
The backscattered electron images, again, demonstrate a severely saponified layer. A few
large, coarse particles of lead white are still intact, but the main part of the layer has a
(grayer) less dense and more amorphous appearance. Here it is postulated that the small
particles of lead white are more reactive towards the oil medium and therefore dissolve
more easily and quickly than the larger particles. In addition, a thin lead-containing
film (max. 2-3 Mm) has deposited at the surface of the layer that probably accounts for
the blanched/whitened appearance of this dark paint area noticed after varnish removal.
Diffusion of lead soaps from the brown layer towards the surface, comparable to some
of the case studies discussed in the previous chapter (Chapter 4), seems to have taken
place. The presence of a minor trace of sulfur in the surface crust, in addition to lead and
carbon, suggests that some lead sulfate has formed, presumably as a result of interaction
of the lead carboxylates with sulfur compounds from the atmosphere. This would also
explain the peak present at m/z 64 (SO2) and small peak at m/z 48 (SO) in the DTMS
spectra of the brown surface paint, indicative of sulfur dioxide.
Binding medium analysis of isolated samples of the brown surface layer using
DTMS demonstrates high amounts of oil components. Mass peaks detected are m/z
256 and 284 typical of palmitic and stearic acid, and m/z 152 from a fragmentation
ion of azelaic acid (C9 diacid).9 Their occurrence in the higher temperature regions in
the chromato/thermogram (scans 78-88) indicates that part of the fatty acid fraction
is metal-bound and not in free form [Keune 2005b], which is in accordance with the
SEM and FTIR data. The inorganic fraction (scans 81-130) shows the isotope pattern
for lead at m/z 206-208, and some sulfur at m/z 64 (see discussion above). DTMS also
provides information about the presence of the cross-linked oil network fraction of the
paint sample. In the higher temperature region (scans 78-88), the typical peak pattern
9
In addition, the spectra contain longer chain fatty acids, C20FA-C26FA, at m/z 312-396.
213
Chapter 5
Fig. 5.3 Roelandt Savery, Orpheus enchanting the Animals with his Music, (Mauritshuis, inv. no. 157), 1627,
oil on oak panel. Overall, and details of birds from center foreground.
of the pyrolysis products of the oil paint network is discernible with mainly aromatic
components (e.g. m/z 91, 105).10
Conclusion
The analyses reveal an oil medium-rich, saponified layer. Nearly all the lead white in the
brown layer has reacted away into soaps: the lead soaps are dispersed throughout the
layer. Brown- and gray-toned paints typically contain black carbonized products and
earth pigments that are known to be highly oil-absorbent. It is thought that they have
supplied the fatty acids required for the conversion of the lead white pigment in the layer
into lead soaps. Originally, a lighter and cooler tone must have been meant. The loss of
the white component in this layer makes the brown/gray paint appear much darker and
warmer than originally intended. As a result of the darkening effect, the animals and
other details painted on top of this layer in the foreground are now only barely visible.
Moreover, the saponification/ loss of opacity of the lead white has strongly influenced
10 Furthermore, traces of pine resin (DHA at m/z 239, 287, 300 and OXO-DHA at m/z 253, 314, 315, 330) and
mastic (at m/z 163 and 400-450 region) were detected, that are likely to be residues of old varnish.
214
Da rk en in g a s a re s u l t o f i n c re as e d tr an s p are n c y i n 1 7 th c e n t u r y o i l p ai n t i n g s
c a lcium
lead
BS E
BSE
10 μm
Fig. 5.4 Light microscopic images of cross-section from distant landscape (upper row), with corresponding FTIR images, EDX maps and SEM backscattered electron images. The saponified brown layer is the
intermediate layer below the (green) top layer of the landscape. The white bands in the FTIR maps represent
areas of high intensity.
215
Chapter 5
the color balance of the whole picture: the dark foreground now strongly contrasts
with the light, bright blue sky, unintentionally creating a kind of dramatic chiaroscuro
effect.11
c . s e l e c t i v e d a r k e n i n g o f g ro u n d a n d pa i n t
l ay e r s o n t h e w o o d g r a i n
ae rt va n der neer ‘ri ver l an d s ca pe’ mid -     s ( mh    )
Loss of opacity is also observed associated with the wood grain on the radial surface
of panel paintings creating disturbingly dark stippled lines that have often been
interpreted as abrasion. In a panel painting by Aert van der Neer, dated mid-1650s River Landscape-, many areas in the sky were found to be disturbed by dark lines that
are associated with the wood grain in some way.12 Although they were also present in
the darker areas of the picture, they were most disfiguring in the bright peach-colored
sky. These streaks, which were found to correspond to the porous early wood on the
radial surface, have become selectively more transparent and darker in tone, in contrast
with areas of intact opaque paint on the late wood (Fig. 5.5). Examination with the
stereomicroscope revealed that the streaks are formed in the underlying lead whitecontaining pinkish brown imprimatura that shimmers through the very thin sky paint or
has become exposed as a result of loss of the sky layer.
In several cross-sections from the sky, three layers are present (Table 5.2): a
chalk ground of variable thickness, a pinkish brown imprimatura (10-20 Mm) and a
single thin pinkish-grey sky layer (approx. 10 Mm), consisting of lead white with the
addition of a little vermilion, umber and glass (also carbon black in the grayer areas
of the sky). In two cross-sections from the darkened lines, the imprimatura, which is
essentially composed of large particles of lead white and a few large rounded particles
of a relatively transparent red-brown umber (EDX: Fe, Mn, Si, Al, (Mg, K)), appears
unusually transparent. Furthermore, small reddish umber particles are observed in these
transparent areas (more Fe-rich than the browner umber particles) in close proximity to
the brown umber particles. It is notable that in the samples from darkened areas a thick
chalk ground is present (up to 60 Mm thick), while in unaffected areas the chalk layer is
thin or barely present. This difference in thickness of the ground is also apparent on the
end grain (transverse sections) of the panels when viewed with the stereomicroscope. The
ground fills the large pores of the vertical parenchyma cells of the early wood, which are
huge in comparison with the smaller pores in the dense late wood (Fig. 5.6). Hence, we
conclude that where the wood is porous (the early wood), the ground is thick because
11 From their composition, these middle brown- and gray-toned oil paints are very prone to darkening. A similar
darkening process due to saponification of the lead white component may be responsible for the now dark areas in
Raphael’s Transfiguration of Christ [Van Eikema Hommes 2004b], as well as in many other oil paintings.
12 The localized darkening in this picture was investigated together with Petria Noble 2002-2006. Treatment of the
picture was carried out by Alice Mohan (Maîtrise de Sciences et Techniques (M.S.T.), University Paris I PanthéonSorbonne) during her three-month internship in the Mauritshuis 2002. The same phenomenon, though less visible,
is present in Aert van der Neer’s River Landscape at Sunset, c.1650, (Mauritshuis, inv. no. 913).
216
Da rk en in g a s a re s u l t o f i n c re as e d tr an s p are n c y i n 1 7 th c e n t u r y o i l p ai n t i n g s
it penetrated the cells opened by planning and sanding of the panel; where the wood is
dense (the late wood), hardly any ground is present. The deep channels of the early wood
can be clearly discerned on the back of the panel (Fig. 5.6). It is primarily these areas
that become filled with ground during preparation of the panel. Dendrochronology
concluded that the oak panel is of Baltic origin with an earliest felling date of about
1642.13
Imaging FTIR analysis of the cross-sections confirmed metal soap formation
in the transparent regions of the imprimatura by strong absorption in the 1510-20
cm-1 region that is characteristic for lead carboxylates. Detailed information on the
morphology of the affected layer was demonstrated with SEM. In Figs. 5.5 and 5.7, the
SEM backscattered electron images of cross-sections from the intact and darkened areas
in the sky are displayed. The difference in the morphology and distribution of the lead
white is striking. The imprimatura in the intact sample shows a lead white-rich layer
with a broad distribution of both fine and coarse particles typical of seventeenth-century
Dutch stack-processed lead white. In contrast, in samples from darkened areas, the
fine lead white particles appear to have reacted away forming large transparent regions
throughout the layer leaving only the largest particles intact. This corresponds to the
observation of large lead white particles in a transparent brownish matrix in these areas
on the surface of the painting. Compared to normal, dense, intact lead white particles,
the saponified areas appear less dense, grayer and amorphous in the backscatter images
due to their higher organic content. Some intact lead white particles, however, do
demonstrate grayer peripheries, corresponding to transparent perimeters seen in normal
light microscope images, that indicate a progressive transformation into lead soaps. It
is notable that the altered imprimatura is much thicker than non-affected imprimatura
due to the larger volume of the saponified lead white particles and that the sky layer
is partially missing in samples from darkened areas. EDX analysis confirmed the lead
involved in soap formation.
It is also noted that the chalk ground appears particularly porous. Binding
medium analysis carried out with DTMS on an isolated sample of the chalk ground
-from an area in the sky at centre right- suggests that the highly porous glue-bound
ground has absorbed oil from the layer(s) above. In the mass spectra the signals for
animal glue are almost obscured by the large amounts of oil derived components,
the monocarboxylic and dicarboxylic fatty acids.14 The main mass peaks detected are
m/z 256 and 284 from the molecular ions of palmitic and stearic acid, and m/z 152 a
characteristic fragmentation ion of azelaic acid (C9 diacid). Only a small part of the fatty
acids seems to be bound to the oil network, since the main fraction is released from the
probe at low temperature (scans 25-40) indicating that they must be present in the free
form [Keune 2005b]. The DTMS spectra also give evidence for the presence of chalk
because of the release of CO2 (m/z 44) detected at high scan numbers (scans 92-107)
that is associated with the carbonate group from the chalk.
13 Dendrochronology by Peter Klein, University of Hamburg, 2002.
14 Resin components, pine resin and dammar, identified with DTMS in the chalk ground, were probably introduced
during past cleaning and varnishing. This indicates how absorbent/porous the chalk ground is.
217
Chapter 5
2
1
3
2
2
1
20 μm
1
20 μm
Fig. 5.5 Aert van der Neer, River Landscape, c.1650 (Mauritshuis, inv. no. 912). Overall (inset) and detail
of dark streaks in centre right sky (upper row); microscope detail (center left); light micrroscopic image of
cross-section from dark streak (center right); SEM backscatter image of cross-section (912x10) from unaffected grey sky at upper right (lower left) and SEM backscatter image of cross-section (912x11) from dark
streak immediately next to intact sample (sky layer is missing from this sample) (lower right).
218
Da rk en in g a s a re s u l t o f i n c re as e d tr an s p are n c y i n 1 7 th c e n t u r y o i l p ai n t i n g s
Fig. 5.6 Microscope detail of reverse of Aert van der Neer showing deep channels of the oak panel (left).
Detail of the transverse section (end grain) showing chalk ground that preferentially fills the large cells of
early wood (right).
Fig. 5.7 SEM backscattered electron image of cross-section of dark streak (912x11) in Fig. 5.5 enlarged showing dissolution and saponification of lead white pigment in imprimatura.
219
Chapter 5
Conclusion
In this case, the darkened imprimatura either shines through the thin translucent
upper layer or is exposed where it has expanded causing the upper layer to flake off
(expansion is a characteristic of lead soaps borne out of previous investigations of metal
soap aggregates). Not only is the selective darkening in the imprimatura related to the
high degree of saponification of the lead white, but it is also related to the loss/lack of
reflectivity of the chalk ground by absorption of oil from the paint layers. The presence
of oil in the chalk ground, as detected with DTMS, may explain the source of free fatty
acids required for the transformation of lead white into lead soaps.
During the recent treatment, careful retouching was necessary to hide the lines
in order that the picture could be displayed in the galleries; after retouching the painting
is closer to its original appearance.
j an van g oyen ‘di l api dated fa r mh o us e with pe a s a n ts’    
( m h     ) and ‘ri ver vi ew’ c.     -   ( mh    )
Investigations of two similarly affected panel paintings by Jan van Goyen -Dilapidated
Farmhouse with Peasants, dated 1631, and River View from c.1644-1648, show consistent
results.15 In these essentially tonal paintings, not only are disturbing lines visible over the
entire surface, but the loss of a reflective surface layer permits the shining through of the
darkened underlying layers (and the panel) resulting in a loss of detail and distortion of
color, particularly in the thinly painted and damaged areas such as the skies (Figs. 5.8
and 5.9). As a result, the careful attention to the creation of the atmospheric quality
of light is now largely lost. While these alterations are clearly visible with the naked
eye, the cause of the darkening effect is much more difficult to distinguish under the
stereomicroscope. In both cases, microscopic examination revealed selective flaking
and cracking of the surface paint exposing the underlying imprimatura along the areas
associated with the early wood of the wooden panel.
It is in the study of paint cross-sections from affected and non-affected paint
layers that the cause of the darkening phenomenon is found (Table 5.2). In the sky areas
from both paintings three layers are present: a chalk ground (up to 60 Mm), a light lead
white-containing imprimatura (10-20 Mm) with the addition of a little carbon black
or some umber as the coloring component, and a single very thin sky layer consisting
mainly of lead white and a little smalt (10-15 Mm). Comparable with the Aert van
der Neer, in both samples from affected areas a thick chalk ground is present while in
unaffected areas, the chalk ground is barely present. Here again, the difference in the
thickness of the ground is found to be related to the uneven structure of the wood grain:
in the denser late wood there is hardly any ground present whereas in the porous early
wood the ground is thick filling the cells opened by the planing of the wood panel.
The imprimatura in both Van Goyens differs slightly in composition from that in the
Aert van der Neer painting. Here, chalk has been added to the lead white, suggesting
15 The localized darkening in these pictures was investigated together with Petria Noble 2002-2006, as part of the Open
Laboratory agreement between Mauritshuis and FOM-AMOLF.
220
Da rk en in g a s a re s u l t o f i n c re as e d tr an s p are n c y i n 1 7 th c e n t u r y o i l p ai n t i n g s
a slightly translucent effect may have been aimed for, or alternatively the use of the
cheaper variety of lootwit, rather than the pure form of schulpwit [Van de Graaf 1958,
pp. 34-35]. The sky layer manifests itself as a very compact layer containing fine and
some coarser particles of lead white. Here, the sparsely distributed smalt is mostly
intact and appears grayish-blue in cross-section. In comparison, cross-sections from the
affected areas demonstrate only partial remains of the sky layer (Fig. 5.8), confirming the
surface observation that the sky paint in these areas has largely flaked off revealing the
imprimatura that has become selectively transparent and darker in appearance.
In SEM backscatter images of the cross-sections from the affected areas in both
paintings, it is remarkable that the majority of original lead white in the imprimatura
layers has converted to lead soaps (Figs. 5.8 and 5.9). The intact particulate lead
white pigment particles show strong scattering and appear white, while the less dense
(grayer) amorphous areas point to saponified areas. In the gray amorphous areas a lower
concentration of lead and higher carbon content were detected as compared to the
intact lead white where a higher lead and lower carbon content was demonstrated. These
concentration differences are also illustrated in the EDX map of lead. The low-scattering
(dark) rounded areas are the chalk particles, which are also demonstrated in the calcium
map. This is in contrast to cross-sections from unaffected areas, where much more
Fig. 5.10 DTMS analysis of chalk ground from Jan van Goyen, River View (Mauritshuis, inv. no. 759).
A) total ion current (TIC) diagram; B) summed mass spectra of low temperature event (scans 10-50) showing free fatty acids, mainly m/z 256 palmitic acid and m/z 284 stearic acid, and C) mass chromatograms of
ions from azelaic acid (m/z 152), general marker for saturated fatty acids (m/z 129), palmitic acid (m/z 256),
polymeric network (m/z 91), animal glue (m/z 117, 154) and chalk (m/z 44).
221
Chapter 5
3
2
1
Fig. 5.8 Jan van Goyen, River View, c. 1644-1648 (Mauritshuis, inv. no. 759). Overall and detail of dark
lines in upper left sky (center left); microscope detail (center right); SEM backscatter image of cross-section
(759x01) from darkened sky showing the sky layer has largely flaked off (lower left) and SEM backscatter
image of cross-section (759x02) from darkened line in sail of boat (lower right).
222
Da rk en in g a s a re s u l t o f i n c re as e d tr an s p are n c y i n 1 7 th c e n t u r y o i l p ai n t i n g s
3
3
2
2
1
Fig. 5.9 Jan van Goyen, Dilapidated Farmhouse with Peasants, 1631 (Mauritshuis, inv. no. 1081). Overall
and detail from upper left sky showing disturbing horizontal lines (center left); microscope detail (center
right); SEM backscatter image of cross-section (1081x03) from unaffected grey sky (lower left) and SEM
backscatter of cross-section (1081x01) from darkened area in grey sky (lower right) (same scaling).
223
Chapter 5
intact lead white surrounds the chalk particles, and corresponds to the light microscopic
images, where this layer also appears more opaque in the samples from the intact areas.
Binding medium analysis was also carried out on isolated samples of the ground
from each painting taken at the edges of the panels. The organic components, detected
using DTMS analysis, are similar to the chalk ground in the Aert van der Neer. In both
Van Goyens, markers for animal glue (m/z 70 and 154) and chalk (CO2 event at high
scan numbers) are present, as well as free monocarboxylic and dicarboxylic fatty acids
associated with oil (Fig. 5.10).16 The DTMS results confirm the presence of a fatty acid
source that is needed for the conversion of lead white into soaps.
Conclusion
Here again, the thickness of the ground in relation to the panel’s wood structure
corresponds with different degrees of lead soap formation in the imprimatura. It can
be concluded that the striped pattern of darkened lines created in the affected areas is a
result of the surface paint flaking off in areas corresponding to the early wood of the oak
panel whereby the darkened imprimatura (and support) has become exposed. For both
Van Goyens, the dark lines are so disturbing that the pictures can no longer be displayed
in the galleries. Given that the stripiness covers most of the surface of the paintings
introduces aesthetic and ethical issues. It remains questionable whether one should
consider retouching or overpainting such large areas.
j an s teen ‘danci ng peasants at a n inn’ c.     -   ( mh    )
In this early panel painting by Jan Steen, darkened lines associated with the wood grain/
uneven wood structure are visible over the entire surface, but are most disturbing in the
sky areas, and are also notable in the foreground and in the figures (Fig. 5.11). Prior to
the picture’s recent treatment, numerous darkened lines were found to be covered with
old, now discolored, retouching.
Cross-sections were taken from a darkened line in the sky and from an intact
(blue) area for comparison. The cross-section from the darkened area also includes two
layers of blue overpaint that comprises modern cobalt blue pigment (a cobalt aluminum
oxide, CoO.Al2O3). Unfortunately, the ground and imprimatura layers are missing in
this sample, since they crumbled during the sampling process. Like the other paintings
under examination, this darkened area clearly correlates to the porous channels of the
early wood that is filled with a thick layer of ground material. This is visible on the
surface, in other darkened areas, where the surface paint is lost and the chalk ground
that appears light brownish in color has become exposed.17 The cross-section from the
intact blue area (see Table 5.2) shows the complete paint layer build-up: a chalk ground
(up to 40 Mm), a very thin light gray imprimatura (c.5 Mm) and a single gray-blue sky
16 Ground samples in these paintings were also found to be contaminated with resin components associated with the
varnish layers: dammar, pine resin and a modern synthetic methyl ketone resin (AW2 BASF Methylcyclohexanone).
17 Analysis by FTIR of dark material collected from the deep porous channels of the early wood mainly showed peaks
for cellulose from the wood, in addition to a little chalk. This implies that local shimmering of the panel through the
upper paint layers that have become increasingly transparent also plays a role in the selective darkening.
224
Da rk en in g a s a re s u l t o f i n c re as e d tr an s p are n c y i n 1 7 th c e n t u r y o i l p ai n t i n g s
layer (c.15 Mm) consisting of lead white and smalt and a few black and red particles.
The imprimatura is composed of rounded particles of chalk, fine and coarser lead white
and a little carbon black and appears rather opaque in this cross-section. The difference
in the appearance of the sky layer in the light microscopic images between the affected
and unaffected is striking, the darkened layer consisting only of a few large lead white
particles in a transparent brown matrix. This was also noted on the surface of the
painting in the darkened areas. The unaffected sky layer, in contrast, demonstrates an
opaque layer, in which both large and small lead white and grayish angular particles
of smalt are discernable. This difference between the affected and unaffected is also
reflected in the SEM backscatter images of these layers. In the image of the unaffected
area, a broad distribution of particularly fine lead white with some coarse particles is
demonstrated, whereas in the cross-sections from the darkened area all the fine lead
white particles have reacted away leaving only a few coarse particles (Fig. 5.11). The
less dense, (gray) amorphous areas in the backscatter image, prominent throughout the
whole sky layer, indicate the transformation to lead soaps. In both cross-sections, the
low potassium concentrations inside the smalt particles (<1 wt %, measured by EDX)
indicate that the smalt is almost completely discolored. Here, the paint matrix clearly has
a strong influence on the color perception. When the discolored smalt is situated in a
highly reflecting matrix, such as the fine lead white, it still appears gray-blue on the paint
surface. In the transparent brown matrix, however, the pale blue color is completely lost.
A darkened line in the brown foreground, next to the tree trunk in the lower
foreground, was also sampled. Here the cross-section reveals an incomplete chalk
ground (c.10 Mm), since again part of it crumbled during sampling. In this sample, the
imprimatura is barely present. The semi-transparent dark brown paint layer (c.15 Mm)
comprises azurite, smalt, chalk, dispersed lead and red earth. In comparison, a sample
from a non-darkened area, just next to the darkened line, reveals a light-gray underlying
imprimatura (c.10 Mm) containing some rather coarse lead white, followed by the same
dark brown upper layer (c.20 Mm). Here, the light imprimatura reflects some of the
light and shimmers through the upper layer, creating a lighter appearance on the surface
compared to the darkened line where the imprimatura is barely present.
Conclusion
The darkened lines in the painting can be explained by saponification of the lead white
particles in the surface layers and very thinly applied imprimatura. The resulting loss of
reflectivity of the paint and imprimatura, combined with the darkened chalk ground
in these areas (from absorption of the oil from the imprimatura), permits the shining
through of the panel in these areas. The results show that saponification of lead white is
not restricted to the imprimatura, but can also affect lead white-containing surface layers,
in this case, selective areas of the sky, as determined by the chalk ground that fills the
uneven oak wood structure.
225
Chapter 5
3
3
2
1
Fig. 5.11 Jan Steen, Dancing Peasants at an Inn, c. 1646-1648 (Mauritshuis, inv. no. 553). Overall and
detail of dark streaks upper left sky (center left); microscope detail from figure in foreground (center right);
SEM backscatter image of cross-section (553x04) from unaffected blue sky (lower left) and SEM backscatter
image of cross-section (553x01) from overpainted dark line in sky (lower right) (same scaling).
226
Da rk en in g a s a re s u l t o f i n c re as e d tr an s p are n c y i n 1 7 th c e n t u r y o i l p ai n t i n g s
Fig. 5.12 Trippenhuis, painted ceiling N1, 1660-62 with detail of the discolored sky; detail of backscattered electron image showing saponification of the lead white, as seen by the gray halos (center right); light
microscopic image of cross-section from discolored sky N1 (lower left) with corresponding SEM backscatter
image (lower right). The browish areas in the cross-section correspond with low scattering, saponified areas in
the SEM.
227
Chapter 5
d . i n c re a s e d t r a n s pa re n c y a n d d a r k e n i n g
( b r o w n i n g ) o f b l u e p a i n t s c o n ta i n i n g s m a lt a n d
lead white
t rippenhui s ‘pai nted cei li n gs with h unting s ce ne s with s k ie s a n d
b ird s’     - 
The ceilings in the corridors and stairwells of the Trippenhuis, a palatial house (actually
a double house) on the Kloveniersburgwal in Amsterdam built for the two brothers
Louys and Hendrick Trip by the architect Justus Vingboons between 1660 and 1662, are
richly decorated depicting hunting scenes with painted skies with birds. For centuries,
the painted ceilings had been covered by many layers of monochrome overpainting,
but these were removed in two phases, in the southern house in the 1980s and in the
northern house in the early 1990s, to expose the original decoration schemes. Large
areas of the originally blue skies now appear discolored and have a brownish, patchy
appearance (Fig. 5.12). We examined about 15 cross-sections from the skies, from
discolored areas as well as from areas that still appear blue, using light microscopy and
SEM-EDX.18
The cross-sections show great similarities in the layer structures: a chalk ground,
followed by a light lead white-containing preparation layer, and one or two applications
of blue paint mixtures of lead white and smalt (Table 5.2). It seems that the brownish
appearance of the paint is only partly due to degradation of the blue smalt particles.
Although they are clearly in the process of discoloration, there are still many blue or
bluish particles (with high K levels) discernable in cross-sections from both blue and
discolored/browned areas. We infer from the cross-sections that saponification of the
lead white pigment is mainly responsible for the observed color differences on the paint
surface. In general, the cross-sections from the blue areas reveal higher densities of
reflecting lead white particles, while the smalt layers from brownish areas have become
transparent in places throughout the layer. The transparent areas in the light microscopic
images correspond to less dense (grayer) areas in the SEM backscatter images showing
amorphous, cloudy morphologies that point to saponified areas. The EDX maps of lead
confirm the presence of lead in these areas where it is not associated with distinct lead
white pigment particles. It is noted that the smalt particles that are situated in these
brownish transparent areas are hardly perceptible in the light microscopic images and
appear almost transparent (Fig. 5.12). Dissolution of the lead white reduces the number
of internal light reflections in the paint decreasing the hiding power. The light penetrates
deeper. This also implies that less blue light is bounced back by the smalt particles, but
instead is absorbed by the brownish matrix. A similar optical effect was noticed in the
cross-sections from the sky in the Steen (see previous example). The visual impact of the
darkening effect is also found stronger in the thinly painted areas, like the interstices of
the brushwork as compared to the raised areas.
18 The ceilings are currently being restored by Edwin Verweij | Verweij Office for Architectural Paint Research and
Conservation Amsterdam and Ruth Jongsma | Bureau voor Kleuronderzoek en Restauratie Amsterdam.
228
Da rk en in g a s a re s u l t o f i n c re as e d tr an s p are n c y i n 1 7 th c e n t u r y o i l p ai n t i n g s
Conclusion
Saponification of the lead white is considered the main cause for the increased
transparency and consequent darkening/browning of the blue sky paints (composed of
smalt mixed with lead white) that affect large areas of the Trippenhuis ceilings. The loss
of scattering material in the blue paint not only makes the paint more transparent but
also reduces the blue tone/hiding power of the smalt. Unlike the Steen, the darkening
of the sky paints in the Trippenhuis cannot be associated with the wood structure/
grain. Why some areas in the skies are more affected by degradation than others
largely seems to depend on small local differences in climate conditions (moisture,
pH and temperature) within the paints. Furthermore, such handmade paints always
show a certain degree of inhomogeneity. Even though the sky paints show very similar
compositions and layer structures, there are always small differences in e.g. particle size,
pigment volume concentration, support, and thickness of the preparation layers that
may further explain the differences in degree/level of degradation.
discussion
c h emi cal and opti ca l cha nge s
The reduction in opacity (i.e. increase in transparency) causing the darkening observed
on the surface of the paintings presented can be explained by (partial) conversion of
the lead white or lead-tin yellow pigment to metal soaps. From observation of the paint
surface and the cross-sections, and as previously noted, it would seem that the smaller
lead white particles first react away, leaving only the larger particles that are also clearly
in the process of transforming into lead soaps. The opaqueness of lead white in oil is
directly correlated to its light scattering ability. The transformation of lead white pigment
particles to amorphous lead soaps will lower the light scattering ability and, instead,
causes the light to penetrate deeper into the paint resulting in a darker appearance (Fig.
5.13). While a RI value for lead soaps has been assumed between that of the oil and the
pigment [Robinet and Corbeil 2003: 24], recent measurement indicates a value close
to that of the oil, 1.50-1.55.19 The gradual transformation of the lead white (or leadtin yellow) particles to lead soaps significantly lowers the difference in refractive index
between the binding medium and the pigment, leading to an increase in transparency.
Moreover, any partially converted particles of lead white effectively become embedded in
a high RI medium (the lead soaps) and thus appear transparent.
The optical effects caused by saponification of the lead white or lead-tin yellow
can be significant, especially in the case of pigment mixtures or the presence of a dark
underlayer. As with the yellow stockings in the Sweerts, the darkening effect observed
on the surface is enhanced by the dark, almost black underlayer that shimmers through
the yellow layer that has become increasingly transparent. In the case of increased
transparency and darkening of ground layers, as with the phenomenon of localized
19 Measurement carried out on synthesized lead palmitate at room temperature. Personal communication, Yoshiko
Shimadzu, ICN, Amsterdam, October 2007.
229
Chapter 5
Table 5.2 Increased transparency and darkening: summary of samples and analysis
Artist, title,
date, support
Sa m p l e
location
Pa i n t b u i l d - u p
Pa i n t d e f e c t
Analysis
Saponification of lead-tin yellow in
yellowish top layer resulting in a patchy,
darkened appearance of the paint surface
LM
FTIRima
SEM
4. green landscape layer: copper
pigment, chalk, silicates?
3. brown background layer: lead white,
carbon black, earth pigments
2. gray imprimatura: lead white, fine
lamp black, little red
1. chalk ground
Saponification of lead white in medium
brown background layer that has led to
darkening of the layer
LM
FTIRima
SEM
DTMS
1. brown upper layer: lead white, carbon
black, earth pigments
(priming layers missing)
idem
idem
3. pinkish gray sky layer: lead white,
little vermilion, umber, glass +
carbon black in the grayer areas of
the sky (c.10 μm)
2. pinkish brown imprimatura: lead
white, umber (10-20 μm)
1. chalk ground (up to 60 μm thick)
Darkening due to saponification of lead
white in imprimatura
LM
FTIR
SEM
DTMS
(ground)
3. gray-blue sky layer: lead white,
(intact) smalt (10-15 μm)
2. light-gray imprimatura: lead white,
chalk, little carbon black (10-20 μm)
1. chalk ground (up to 60 μm)
Darkening due to saponification of lead
white in imprimatura
LM
SEM
DTMS
(ground)
FTIR
(ground)
3. gray-blue sky layer: lead white,
(intact) smalt (10-15 μm)
2. light-beige imprimatura: lead white,
chalk, little umber (10-20 μm)
1. chalk ground (up to 60 μm)
Darkening due to saponification of lead
white in imprimatura
LM
SEM
DTMS
(ground)
In c r e a s e d t r a n s p a r e n c y i n l e a d - t i n y e l l ow p a i n t s
Michiel Sweerts
Draughts Players
(MH 1121)
1652 canvas
Yellow tassel of
stool, at left
1. yellowish upper layer: lead-tin yellow,
orange earth pigment, ultramarine
(ground missing)
1121x18
D a r k e n i n g o f m e d i u m b r ow n a n d g r a y t o n e s
Roelandt Savery
Orpheus Enchanting
the Animals with his
Music (MH157)
1627 oak panel
Green
landscape
over brown
background
157x36
Dark
background
157x14
Se l e c t i v e d a r k e n i n g o n t h e w o o d g r a i n
Aert van der Neer
River Landscape
(MH 912)
mid-1650 oak panel
Pink sky
Jan van Goyen
Dilapidated
Farmhouse with
Peasants
(MH 1081)
1631 oak panel
Gray-blue sky
Jan van Goyen
River View
(MH 759)
c.1644-48 oak panel
Gray-blue sky
230
912x05
(darkened)
912x10
(intact)
912x11
(darkened)
1081x01
(darkened)
1081x02
(darkened)
1081x03
(intact)
759x01
(darkened)
759x02
(darkened)
759x03
(darkened)
Da rk en in g a s a re s u l t o f i n c re as e d tr an s p are n c y i n 1 7 th c e n t u r y o i l p ai n t i n g s
Jan Steen
Dancing Peasants at
an Inn
(MH 553)
c.1644-48 oak panel
Gray-blue sky
553x01
(darkened)
553x04
(intact)
Dark-brown
background
553x13
(darkened)
553x14
(intact)
3. gray-blue sky layer: lead white,
(discolored) smalt (c.15 μm)
2. light-gray imprimatura: lead white,
chalk, little carbon black (c.5 μm)
1. chalk ground (up to 40 μm)
Darkening due to saponification of lead
white in sky layer
LM
SEM
3. dark paint layer: azurite, smalt, chalk,
dispersed lead, red earth (15 μm)
2. light-gray imprimatura: lead white,
chalk, little carbon black (barely)
1. chalk ground (partial)
D a r k e n i n g / b r ow n i n g o f b l u e p a i n t s c o n t a i n i n g s m a l t a n d l e a d w h i t e
Unknown artist(s)
Trippenhuis
Amsterdam
Painted ceilings with
hunting scenes with
skies and birds
1660-62 oak panel
Sky
TKA06 A to S
(discolored and
intact)
3. blue sky layer: lead white, smalt
2. light imprimatura: lead white
1. chalk ground
Saponification of lead white and partly
discolored smalt in blue layer that has led
to darkening/browning of the layer
LM
SEM
darkening on the wood grain, the thinner or translucent areas in the paintings, usually
the sky areas, are affected worst since they allow the darkened underlayers to shine
through.20 The loss of the white pigment in medium to dark paint layers as a result
of saponification may also lead to color changes that can have a great visual impact
on paintings, the dark areas of the painting becoming much darker than originally
intended. This was seen for example in the landscape by Savery, where darkening
of the brown and gray medium tones in the foreground has resulted in a noticeable
distortion of the original light/dark contrasts: the animals disappear in the foreground,
and the dark foreground stands out too strongly against the light, blue sky. Particularly
interesting is also the effect of increased transparency of the surrounding medium on the
hiding power and color intensity of other pigments in the layer, such as the smalt in the
Trippenhuis samples and in the Steen. The smalt particles also appear more transparent
in a transparent brown matrix where all the lead white has transformed into soaps than
when embedded in reflecting lead white particles. This also has to do with differences in
RI, the difference between lead white and smalt (RI: 1.47-1.55) is much larger than that
between smalt and lead soaps or the oil medium.
These sorts of alterations involving the increased transparency of ground and
paint layers as a result of saponification processes may have a larger impact on the
20 In the Aert van der Neer painting, a color change may also take place due to interaction of the lead soaps with
the partially dissolved umber particles. Partial dissolution of umber facilitated by the basic conditions is also
held responsible for the differences in tone in the umber particles that range from red to brown. The De Mayerne
manuscript warns that umber-containing paints can darken and become dull. Mention is made of a process known
as bleeding [Van Eikema Hommes 1998, pp. 114-115, 129]. According to Hess [Hess 1979] “bleeding” refers to the
diffusion of a coloring matter […] and also to the discoloration resulting from such diffusion.
231
Chapter 5
changing appearances of oil paintings than is often considered. However, to what extent
this has occurred is difficult to assess, and is perceptible usually only in the most extreme
cases. We hope that the results presented in this paper will be a starting point for more
systematic and quantitative studies. In the next sub-section, we will further discuss the
phenomenon of the localized darkening on the wood grain.
loc ali zed da rk eni ng a ssoc iate d with th e wo o d gr a in
The analyses have shown that differences in the thickness of the ground in relation to the
panel’s wood structure correspond with different degrees of lead soap formation in the
layer. We propose that the fatty acids required for the soap formation are partly supplied
by the thick areas of chalk ground that selectively fills the early wood channels.21 This
explains the relationship between the thickness of the ground and the degree of lead
soap formation. Moreover, a large amount of oil (mostly as free fatty acids), required for
the transformation of lead white into lead soaps, was detected with DTMS in the chalk
ground from all four paintings.
Since there is no evidence in the documentary sources to suggest oil was added
expressly to chalk grounds at this time22, we therefore have to conclude that the oil (and
resin) components identified in the chalk/glue ground was most likely absorbed from
the oil paint (and varnish) layers above. This was borne out of recent DTMS analyses
of ground reconstructions comprising animal glue-bound chalk grounds covered with
different types of oil-based imprimaturas. Here DTMS confirmed the absorption of oil
components in the chalk grounds.23 The presence of oil in the chalk/glue ground also
explains its lack of reflectivity and role in the localized darkening. Light can penetrate
even deeper allowing the ground, and possibly also the panel, to shine through the
transparent upper layer(s). In the four case studies presented here, the chalk ground has
been deliberately applied very thinly, only filling the wood grain. In the darkened areas
in the Van Goyen (River View) and Jan Steen, the lead white-containing imprimatura
in the painting is also very thin, and seems to have been partly absorbed by the chalk
ground thus explaining the loss of light reflectivity of the grounds in these paintings. Of
course, it is difficult to know how reflective/absorbent the chalk grounds originally were,
and indeed some chalk grounds appear more transparent and browner than others. In
this sense, the thickness/porosity of the grounds is considered to be a decisive factor in
the final appearance of paintings.
As to what may have triggered the saponification process resulting in selective
darkening, fluctuations in temperature and relative humidity are considered to play an
important role. Since the porous early wood with the thickest areas of chalk ground is
expected to absorb more moisture, as compared to the denser latewood, the influence
of moisture will be greater for the paint on top of the early wood. Furthermore, a
chalk ground is hygroscopic and is likely to respond to humidity changes. Swelling
21 Acidic emissions or extracts from the wood do not seem to play a significant role in the darkening process as initially
thought, since the thick chalk ground in the darkened areas would act as a buffer to neutralize acidic compounds.
22 Personal communication Maartje Witlox 2007.
23 Ground reconstructions were provided by Maartje Witlox and Tiarna Doherty 2007.
232
Da rk en in g a s a re s u l t o f i n c re as e d tr an s p are n c y i n 1 7 th c e n t u r y o i l p ai n t i n g s
1
2
3
4
5
Cross-section
Surface
Fig. 5.13 Schematic diagram showing the gradual dissolution and saponification of lead white in the upper
paint layer and explaining the effect of saponification on the reflectance of light and on the surface appearance. Stages 1-5: (1) intact paint, light is scattered (2) early stage, first the small lead white particles react
away, (3) mature stage, the small particles have reacted away, edges of the large particles are also dissolving,
(4) late stage, the large particles are almost dissolved, (5) final stage, all lead white is saponified, light is
absorbed.
233
Chapter 5
and shrinking of the thick chalk ground may have played a role in causing the surface
paint to flake off in these areas exposing the darkened imprimatura. However, as already
pointed out, a saponified lead white layer takes up a larger volume than it originally
had;24 in this sense the physical expansion of the affected layer explains the paint loss
associated with the early wood.25
The fact that this particular phenomenon is often observed in early to midseventeenth-century panel paintings has to do with a number of factors: not only the use
of coarsely grained oak panels,26 but also a change in painting technique involving the
use of thin ground and paint layers. The stylistic changes that had come about as a result
of this faster and more economical approach resulted in open brush work, the use of thin
and often translucent oil-rich paint layers, and a simple paint layer build up, consisting
of one or at most two paint layers. In this sense, this type of degradation is technique
dependent.
Alterations, such as darkened lines, changes in tone and contrast, and the
associated color change and loss of detail, naturally can have profound consequences
for the interpretation of the pictorial, illusionistic and aesthetic intentions of the artist.
However, with some artists, caution has to be exercised because the use of transparent
ground layers may be part of the artist’s technique that allowed the color and structure
of the panel to play a role. This is the case with Jan van Goyen, particularly in his mature
works where even the lines of the wood grain are sometimes incorporated as a pictorial
element, for instance as ripples in the water [Gifford 1996: 74-77]. However, one has
to be careful not to misinterpret lines, such as we see in River View, as deliberate effects.
Here, the surface paint is thin or has flaked off exposing the darkened imprimatura and
thus taking on an unintended pictorial role. Moreover, in the thinly painted sky the
horizontal lines produced by exposure of the darkened imprimatura due to flaking and
cracking of the surface paint have formed an additional pattern that was never intended
by the artist.
To achieve his artistic goals Van Goyen applied an extremely thin chalk/glue
ground, just filling the wood grain. The use of such thin grounds that scarcely fills the
wood grain seems to represent a change in painting technique, since most recipes for
chalk/glue grounds from the fifteenth to the seventeenth centuries state that the ground
should be applied in several layers. In 1604, Van Mander mentions that earlier artists
24 Conversion of lead white into lead soaps, calculated on a molar basis, leads to a 26x increase in volume compared
to the original volume of 1M of lead white, assuming that the fatty acids come from elsewhere and do not leave a
negative volume (Frank Hoogland, FOM-AMOLF 2007). Therefore, the volume change can be considerable in
a complex lead white paint layer where fatty acids possibly migrate from the pore space of another layer. Volume
increase by a factor of two or three would already have a large effect on the physical strength of such a paint layer
complex.
25 Shrinkage across the grain may also play a role in the delamination of the paint, since both the underlying wood and
chalk ground are hygroscopic and undergo tangential expansion and compression. Two early panel paintings from
1480/90 and 1546, one on red beech with a chalk ground and the other on oak with a red lead ground, demonstrate
similar localized damage and paint loss [Bünsche 1993].
26 By nature, oak is a ring-porous wood, its coarse structure easily recognized because of the formation of bands of large
early wood vessels to conduct water, followed by the formation of more compact late wood with smaller vessels [Klein
1998: 39, 40].
234
Da rk en in g a s a re s u l t o f i n c re as e d tr an s p are n c y i n 1 7 th c e n t u r y o i l p ai n t i n g s
used to ground their panels more thickly [Miedema 1973], while Willem Beurs in 1692,
says that the chalk ground only needs to fill the pores of the wood [Beurs 1692]:
…Our modern ancestors before used to whiten
their panels more thickly / and scraped them as
smooth as they could / also used cartons / which
they applied to this even beautiful white / and
sat down to trace them with some material that
rubbed off / applied to the back / then they drew
on neatly with black chalks or pencils…
…Ons moderne Voorders voor henen plachten
Hun penneelen dicker als wy te witten /
schaefden s’alsoo glat als sy wel mochten /
Ghebruyckten oock cartoenen / die sy brochten
op dit effen schoon wit / en ginghen sitten Dit
doortrecken soo met eenich besmitten / Van
achter ghewreven / en trockent moykens daer nae
met swarte krijkens oft potlykens…
[translation by M. Witlox].
[Van Mander 1604: 256, 257, fol. 47v]
…One first lays on the panel a ground of weak
glue mixed with white chalk to cover the grain
of the wood; after this one must scrape it off
and make the panel smooth and even, taking
care that the grain remains filled. After this one
grinds umber and lead white very thickly with
oil and applies it to the panel first with a knife,
after which one smooths it with the hand; [do
this] 3 to 4 times depending on how smooth one
wants it to be, and thus is it ready for a painter
of images, but for a landscape painter, one takes
black mixed with lead white…
…Op ‘t panneel legtmen eerst een grond met een
flaauw lijmverwtje met krijtwit gemengt; om
de nerf van ‘t hout te dekken, ‘t welk volgragt
zijnde, moet men ‘t krijt wederom schoon
afschrabben en het panneel effen en glad maken,
sorge nogtans dragende, dat de nerf wel gevult
blijve. Hier na vrijftmen omber met loodwit heel
dik in oly, en men doet dat met een mes voor
d’eerstemaal op ‘t panneel, dan strijkt men ‘t
glad met de hand tot 3 a 4 maalen, na datmen
‘t glad gebeert, en dus is ‘t bequaam voor een
beeldschilder: maar voor een landschap-schilder
neemtmen swart met loodwit gemengt…
[translation by P. Noble and M. Witlox].
[Beurs 1692: 19-20]
Because of its great impact on the tonality and the long-term effects on the
appearance of paintings, the nature of the ground attracts considerable attention in
historical sources. In 1839 Henri Fielding refers to Van Goyen’s (and others) use of
transparent grounds as being ‘the destruction of many fine pictures’ stating that ‘…Van
Goyen and others have prepared or filled up the pores of the wood with their transparent vehicle, which
also enabled them to shew [original spelling] the grain of the wood through the shades of the subject’.
[Fielding 1839: 81-82]. Even though Van Goyen made deliberate use of transparent
grounds, it would appear that two centuries later, Fielding is referring to an increase in
235
Chapter 5
the transparency of Van Goyen’s paintings due to the loss of hiding power in the priming
and paint layers as has been described in this paper.27
conclusion
Due to the insights gained about metal soap formation, the conversion of particulate
mineral lead white and lead-tin yellow to organic complexes of metal soaps is considered
to play a major role in increasing transparency, and the consequent darkening of oil
paints containing lead white or lead-tin yellow pigment. SEM backscatter images of
paint cross-sections from affected paintings reveal the gradual transformation of highly
scattering well-defined pigment particles into less dense (grayer) amorphous lead soap
areas. This chemical change, which results in the decreased ability of lead white and
lead-tin yellow to scatter light, explains the reduction in hiding power. The visual impact
of this phenomenon on the appearance of paintings can be significant, as seen in the
seven paintings presented in this study. Not only the degree of saponification, but also
the other coloring components in the affected paints and the presence of a dark(ened)
underlayer are decisive factors in the final appearance of the paint at the surface.
Loss of opacity has also been found to occur selectively. The investigations of
the panel paintings by Aert van der Neer, Jan van Goyen and Jan Steen revealed that the
paint on top of the porous early wood on the radial surface appeared darker in relation
to the nearby intact paint on the late wood. These darkened areas were found to be
associated to localized lead soap formation in the lead white-containing priming/paint
layers as a function of the thickness of the chalk grounds that fills the uneven structure
of the wood grain. As a consequence, the darkened imprimatura either shimmered
through the thin translucent sky paint or was exposed as a result of loss of the sky layer
due to swelling of the underlying saponified priming layer creating disturbingly dark
stippled lines.
It is part of the restorer’s job to bring back the balance in the picture and
improve the readability. However, the question remains how far one wants to go with
retouching, especially when the darkening involves large areas of the painting. In the
Aert van der Neer, careful retouching was necessary to hide the lines in order that the
picture could be displayed in the galleries; after retouching the painting is closer to its
original appearance. This will also be the case with the Jan Steen painting. As for the Van
Goyens, they are no longer displayed in the galleries. Given Van Goyen’s predilection
for transparent paint layers, the striped pattern of lines covers most of the surface and
despite the fact that the aesthetic, pictorial and illusionistic intentions of the artist are
compromised, it remains questionable as to whether one should consider retouching or
overpainting such a large area. Likewise, in the Savery and the Trippenhuis ceilings, the
darkening involves large areas of the painting. Some restorers employ selective cleaning
as a solution, or apply thin tinted glazes on light or bright areas in the painting that
27 Maartje Witlox is thanked for pointing out the relevant historical sources from the HART database.
236
Da rk en in g a s a re s u l t o f i n c re as e d tr an s p are n c y i n 1 7 th c e n t u r y o i l p ai n t i n g s
contrast too strongly with the darkened areas in order to repair/restore the unity in a
picture.
ac k n ow l e d g e m e n ts
The close collaboration with the Conservation studio of the Mauritshuis, particularly
with Petria Noble and Sabrina Meloni, as well as the architectural paint researchers of
the Trippenhuis ceilings, Edwin Verweij and Ruth Jongsma, is gratefully acknowledged.
Thanks are also extended to Maartje Witlox for sharing data from the HART ground
recipe-database (De Mayerne Research Program 2004-2006), as well as Jerre van der
Horst, FOM Institute AMOLF, for his kind assistance with the DTMS analyses.
references
Beurs, W., 1692, De groote Waereld in het Kleen Geschildert, Johannes en Gillis Ansonius van
Waesberge, Amsterdam.
Boon, J. J., van der Weerd, J., Keune, K., Noble, P., Wadum, J., 2002, ‘Mechanical and chemical
changes in old master paintings: dissolution, metal soap formation and remineralization
processes in lead pigmented ground/intermediate paint layers of 17th century paintings’. In
ICOM Committee for Conservation, 13th Triennial Meeting, Rio de Janeiro, Preprints, ed. R.
Vontobel, James and James, London, pp. 401- 406.
Boon, J. J. Gore, E., Keune, K., Burnstock, A., 2004, ‘Image analytical studies of lead soap
aggregates and their relationship to lead and tin in 15th century lead-tin yellow paints from
the Sherbourne Triptych’, in Conference proceedings IRUG 6 Sixth Infrared and Raman Users
Group Meeting, Florence, Il Prato, 29 March – 1 April 2004, ed. M. Picollo, pp. 66-74.
Bünsche, B., 1993, ‘Schäden der Grundier- und Farbschicht an Holztafelbilderin’, Restauro , pp.
246-249.
Carlyle, L., 2001, The artist’s assistant: oil painting instruction manuals and handbooks in Britain
1800-1900, with reference to selected eighteenth century sources, Archetype Publications,
London.
Eibner, A., 1909, Malmaterialenkunde-als Grundlage der Maltechnik, Verlag von Julius Springer,
Berlin.
Van Eikema Hommes, M., 1998, ‘Painters’ methods to prevent color changes’, in Looking through
paintings, Leids Kunsthistorisch Jaarboek XI, eds. E. Hermens, A. Ouwerkerk, N. Costaras,
Archetype Publications, London, pp. 91-131.
Van Eikema Hommes, M., 2004a, ‘Methods used by painters to prevent color changes described
in 16th to early 18th century sources on oil painting techniques’, in Changing pictures,
discoloration in 15th-17th century oil paintings, Archetype Publications, London, pp. 17-50.
Van Eikema Hommes, M., 2004b, ‘Discoloration of Chiaroscuro? An interpretation of the dark
areas in Raphael’s Transfiguration of Christ’, in Changing pictures, discoloration in 15th-17th
century oil paintings, Archetype Publications, London, pp. 171-213.
237
Chapter 5
Fielding, T. Henri, 1839, On painting in oil and water colors, for landscape and portraits; Published
for the author, by Ackermann & Co., Strand, London.
Gettens, R.J., Stout, G.L., 1966, Painting materials, a short encyclopaedia, Dover Publications,
New York.
Gifford, M. E. 1996. ‘Jan van Goyen en de techniek van het naturalistische landschap’ in Jan
van Goyen, Exh. Cat. Stedelijk Museum De Lakenhal, ed. C. Vogelaar, Waanders Publishers,
Zwolle, pp. 70-80.
Van de Graaf, J. A., 1958, Het Mayerne manuscript als bron voor de schildertechniek van de barok,
PhD dissertation, Mijdrecht.
Hess, M., 1979, Paint film defects, their causes and cure (3rd edition), eds. H.R. Hamburg, H.R.,
W.M. Morgans, Chapman and Hall, London, pp. 89.
Keune, K., 2005a, ‘Metal soap aggregates in oil paintings from the 15th-20th century’, in Binding
medium, pigments and metal soaps characterized and localized in paint cross-sections, PhD
dissertation, University of Amsterdam, MOLART Series (11), AMOLF, Amsterdam, pp.
128-132.
Keune, K., 2005b, ‘Secondary ion mass spectrometry characterisation of traditional oil paint:
comparative studies with DTMS and GC/MS’, in Binding medium, pigments and metal soaps
characterized and localized in paint cross-sections, PhD dissertation, University of Amsterdam,
MOLART Series (11), AMOLF, Amsterdam, pp. 44-66.
Laurie, A.P., 1926, ‘On the change of refractive index of linseed oil in the process of drying and
its effect on the deterioration of oil paintings’, Proceedings of the Royal Society of London, no
A760, vol. 112, pp. 176-181.
Laurie, A.P., 1967, The painter’s methods and materials, Dover Publications, New York.
Van Loon, A., Keune, K., 2007, Analytical report of the smalt paints in the Trippenhuis,
Kloveniersburgwal Amsterdam. Unpublished.
Meloni, S., 2003, Orpheus betovert dieren van Roelandt Savery, unpublished final thesis, Limburg
Conservation Institute (SRAL) Maastricht.
Miedema, H., 1973, Karel van Mander Den grondt der Edel vry Schilderkunst, Vol 1, Haentjens
Dekker & Gumbert, Utrecht.
Noble, P., van Loon, A., Boon, J.J., 2005, ‘Chemical changes in Old Master paintings II:
darkening due to increased transparency as a result of metal soap formation’, in ICOM
Committee for Conservation 14th Triennial Meeting, The Hague, 12-16 September 2005,
Preprints, ed. I. Verger, James and James, London, Vol. I, pp. 496-503.
Noble, P., van Loon, A., Boon, J.J., 2007, ‘Selective darkening of ground layers associated with
the wood grain in 17th-century panel paintings, in ICOM Committee for Conservation Interim
meeting, London, 31 May- 1 June 2007, Postprints. In preparation.
Rees Jones, S., 1991, ‘The changed appearance of oil paintings due to increased transparency’,
Studies in Conservation 36, pp. 151-154.
Robinet, L., Corbeil, M.C., 2003, ‘The characterization of metal soaps’, Studies in Conservation
48, pp. 23-40.
Shimadzu, Y., van den Berg, K.J., 2006, ‘On metal soap related color and transparency changes
in a 19thC painting by Millais, in Reporting Highlights of the De Mayerne Programme, eds. J.J.
Boon, E.S.B. Ferreira, NWO, The Hague, pp. 43-52.
238
Da rk en in g a s a re s u l t o f i n c re as e d tr an s p are n c y i n 1 7 th c e n t u r y o i l p ai n t i n g s
Townsend, J., 1993, ‘The refractive index of 19th-century paint media: a preliminary study’, in
ICOM Committee for Conservation, 10th Triennial Meeting, Preprints, ed. J. Bridgland, James
and James, London, pp. 586-592.
Wallert, A. (ed.), 1999, Still lifes: techniques and style, Rijksmuseum Amsterdam, Waanders
Publishers, Zwolle.
Wallert, A., de Ridder, W., 2002, ‘The materials and methods of Sweerts’s paintings’, Michael
Sweerts 1618-1664, Exh. Cat. Rijksmuseum Amsterdam, ed. G. Jansen, P.C. Sutton,
Waanders Publishers Zwolle, pp. 37-47.
Witlox, M., Carlyle, L., 2005, ‘ ’A perfect ground is the very soul of the art’: ground recipes for
oil painting 1600-1900’, in ICOM Committee for Conservation 14th Triennial Meeting, The
Hague, 12-16 September 2005, Preprints, ed. I. Verger, James and James, London, Vol. I, pp.
519-528.
Zucker, J., Boon, J. 2007, ‘Opaque to transparent: paint film defects in the work of Frederic
Church and the Hudson River School’, in AIC Paintings Specialty Group, Annual Meeting in
Providence, Rhode Island, June 16-19, 2006: Postprints, compiler Mar Parkin, H., Vol.19, pp.
1-9.
239
240
Download PDF
Similar pages