Content uploaded by Koen Janssens
Author content
All content in this area was uploaded by Koen Janssens on Oct 08, 2014
Content may be subject to copyright.
Appl Phys A (2013) 111:157–164
DOI 10.1007/s00339-012-7490-5
Visualizing the 17th century underpainting in Portrait of an Old
Man by Rembrandt van Rijn using synchrotron-based scanning
macro-XRF
Matthias Alfeld · D. Peter Siddons · Koen Janssens ·
Joris Dik · Arthur Woll · Robin Kirkham ·
Ernst van de Wetering
Received: 21 July 2012 / Accepted: 30 November 2012 / Published online: 15 December 2012
© Springer-Verlag Berlin Heidelberg 2012
Abstract In 17th century Old Master Paintings, the under-
painting generally refers to the first sketch of a compo-
sition. The underpainting is applied to a prepared ground
using a monochrome, brown oil paint to roughly indicate
light, shade and contours. So far, methods to visualize
the underpainting—other than in localized cross-sections—
have been very limited. Neither infrared reflectography nor
neutron induced autoradiography have proven to be practi-
cal, adequate visualization tools. Thus, although of funda-
mental interest in the understanding of a painting’s genesis,
M. Alfeld (
B
) · K. Janssens
Department of Chemistry, University of Antwerp,
Groenenborgerlaan 171, Antwerpen 2020, Belgium
e-mail: matthias.alfeld@ua.ac.be
Fax: +32-0-32653233
K. Janssens
e-mail: koen.janssens@ua.ac.be
D.P. Siddons
National Synchrotron Light Source, Brookhaven National
Laboratory, 75 Brookhaven Avenue, Upton, NY 11973-5000,
USA
J. Dik
Delft University of Technology, Department of Materials Science,
Mekelweg 2, 2628 CD Delft, The Netherlands
A. Woll
Cornell High Energy Synchrotron Source, Cornell University,
Ithaca, NY 14853, USA
R. Kirkham
CSIRO, Materials Science and Engineering, Bayview Avenue,
Clayton, VIC 3168, Australia
E. van de Wetering
Rembrandt Research Project, c/o Kunsthistorisch Instituut,
Herengracht 286, 1016 BX Amsterdam, The Netherlands
the underpainting has virtually escaped all imaging efforts.
In this contribution we will show that 17th century under-
painting may consist of a highly heterogeneous mixture of
pigments, including copper pigments. We suggest that this
brown pigment mixture is actually the recycled left-over of
a palette scraping. With copper as the heaviest exclusive el-
emental component, we will hence show in a case study on
a Portrait of an Old Man attributed to Rembrandt van Rijn
how scanning macro-XRF can be used to efficiently visu-
alize the underpainting below the surface painting and how
this information can contribute to the discussion of the paint-
ing’s authenticity.
1 Introduction
The creation of a painting conventionally starts with a sketch
that transfers the artist’s initial idea on the prepared can-
vas. In 16th century painting the first sketch was executed
as an underdrawing, in that the contours of figures and ob-
jects were drawn, often with charcoal or water-based black
paint, on a white preparation layer. This practice was aban-
doned towards the end of the 16th century. Artists started
to use coloured grounds, on which the initial modeling was
carried out in a monochrome brown paint, the so-called un-
derpainting, by dead coloring areas on the canvas which de-
fined the main forms and principal shadows of the artwork to
be created. Contrary to the underdrawing, which was com-
monly meant to be completely covered by superimposing
paint layers, the underpainting can contribute to the final vi-
sual appearance of a painting. This notably applies to the
midtones between light and dark, where the underpainting
is intentionally allowed to shine through the surface paint
layer [1].
Author's
personal
copy
158 M. Alfeld et al.
The study of underdrawing and underpainting is of great
interest in art history, as it allows one to obtain a view over
the shoulder of the artist during the painting’s creation. Be-
sides revealing the modus operandi of the artist, such infor-
mation can contribute to discussions of authenticity or po-
tential conservation treatment. Conventional imaging tech-
niques used to study a painting’s substructure are X-ray
radiography (XRR) [2] and infrared reflectography (IRR)
[3]. In XRR, the intensity of X-rays transmitted through the
painting is measured; hence, thick paint layers and layers
containing heavy, strongly absorbing elements such as Pb or
Hg dominate the contrast visible in a radiograph. In IRR the
painting is exposed to infrared radiation and the intensity of
the reflected radiation, i.e. the light not absorbed in the paint-
ing, is recorded. IRR yields best results if a highly reflect-
ing ground layer mainly consisting of chalk or gypsum is
present and the underdrawing/underpainting was executed
with pigments containing carbon black [4].
While IRR is an adequate technique for revealing the un-
derdrawing, neither of these techniques is suited to study
the underpainting. As the underpainting is often applied in
thin layers and does not contain large amounts of heavy ele-
ments, the contrast yielded by them in XRR is insignificant.
A visualization of a typical 17th century underpainting exe-
cuted in earth pigments that are much less IR absorbent than
carbon black, is also very difficult by IRR.
Neutron activation autoradiography (NAAR) has proven
to be a successful method for visualization of the under-
painting. Manganese is an important elemental constituent
of earth pigments in the underpainting and can be success-
fully charted by NAAR [5]. In this technique a small fraction
of the atoms in the painting is activated to emit gamma ra-
diation by thermal neutrons. However, the transport of the
painting to a specialized research reactor and its stay there
for periods up to 3 months make this technique rather cum-
bersome [6].
Alternatively, over the past few years the technique of
scanning macro-X-ray fluorescence analysis (MA-XRF) has
been established as an imaging tool for the study of his-
torical paintings. In this method the painting is scanned
with a focused or collimated X-ray beam and for each
point an energy-dispersive XRF spectrum is recorded. From
these spectra elemental distribution images are calculated
that show the relative intensity of the recorded fluorescence
lines. Signals from hidden layers contribute to the image al-
beit their intensity is attenuated by absorption in covering
layers. Due to this effect, the acquired image is dominated
by the surface distribution of an element if it is present at
comparable concentration levels in surface and hidden lay-
ers. The origin of a signal is attributed to a superficial or hid-
den layer based on a comparison of the acquired elemental
distribution images and the visual impression of the paint-
ing.
So far, the method has been applied successfully for re-
vealing overpainted representations. Artists frequently re-
used their canvas, painting on top of existing compositions.
The first successful experiments in revealing such hidden
paintings with MA-XRF were carried out by Dik et al. [7],
visualizing the study of a female head under an impression-
istic flower piece by Vincent van Gogh. Since then, paintings
attributed to Philipp Otto Runge [8], Rembrandt van Rijn
[9], Arthur Streeton [10] and another painting by Vincent
van Gogh [11] were investigated at synchrotron sources. In
parallel, the development of mobile instruments [12] has al-
lowed the in-situ imaging of large canvasses such as Rem-
brandt van Rijn’s Saul and David [13] in the Mauritshuis
Museum in Den Haag, The Netherlands and Goya’s Portrait
of Ramón Satué from the collection of the Rijksmuseum in
Amsterdam, The Netherlands [14]. Further, the possibilities
of preparing the MA-XRF investigations of a painting with
a full-scale mock-up were investigated [15].
However, MA-XRF has not been applied yet to the study
of 17th century underpainting. From an analytical point of
view the challenge is threefold. First, underpaintings consist
of low-Z elements, mostly consisting of transition-metal el-
ements such as Fe and Mn. Second, the layer thickness of the
underpainting is limited, down to less than 10 micrometers
in some cases. Third, surface paint will contain lead white,
which will easily obscure the emission of transition-metal
pigments present in lower layers. All these factors will make
the analytical footprint of the underpainting in the XRF do-
main difficult to detect. Yet, the principal possibility to re-
veal the underpainting would be of great generic interest to
the art historical and conservation community.
In the present paper we report the results acquired by
MA-XRF and scanning electron microscopy combined with
energy-dispersive X-ray analysis (SEM-EDX) on the under-
painting present in Portrait of an Old Man
(see Fig. 1).
At the moment of our analysis (summer 2009) the paint-
ing was attributed to Rembrandt van Rijn’s workshop. It
was considered to be part of a larger group of paintings
showing an identical bearded old man. All these paintings
were considered copies after an original that was thought
to be lost. However, during the conservation treatment of
the present painting, the authoritative Rembrandt Research
Project started to reconsider the painting’s former classifi-
cation as studio work. The arguments were of art historical,
stylistical and technical nature. In this paper we focus on the
visualization of the underpainting, which proved to be an
important piece of evidence in the painting’s attribution.
During the conservation treatment of the painting in Am-
sterdam, hidden paint layers, present below the surface, be-
came apparent. To study their distribution, IRR and XRR
investigations were performed (see Fig. 2). Both IRR and
XRR revealed isolated compositional changes, so-called
pentimenti, mostly to the man’s headgear, the position of his
Author's
personal
copy
Visualizing the 17th century underpainting in Portrait of an Old Man by Rembrandt van Rijn using 159
shoulder as well as alterations in the figure’s collar. How-
ever, neither XRR nor IRR gave easily/straightforwardly in-
terpretable information that provided insight into the making
of this painting, let alone its original appearance. It was ex-
pected that a visualization of the underpainting could reveal
the first compositional sketch and therefore provide insight
into the object’s genesis and, hence, its attribution.
To confirm that the underpainting might be visualized
by XRF imaging, the painting was investigated by means
of a hand-held XRF device, which revealed a variation of
Fig. 1 Portrait of an Old Man, Rembrandt van Rijn, ca. 1630, oil on
wood, private collection. The red ‘+’ indicates the location from where
a paint sample was taken. Grey circles mark the spots investigated by
portable XRF (©Rene Gerritsen)
the abundance of the elements Mn, Fe and Cu in the back-
ground of the painting independent of the surface image.
This suggested a contrast that might be exploited in XRF
imaging. After that, the painting was brought to the beam-
line X7A of the National Synchrotron Light Source (NSLS)
at the Brookhaven National Laboratory (BNL) in Upton,
New York and investigated by MA-XRF. After the visual-
ization of the underpainting hidden below the surface, the
pigments present in this hidden layer were identified by the
microscopic investigation of a polished cross section taken
from the painting. This was done by SEM-EDX.
2 Experimental
Before the XRF experiments IRR and XRR images were
recorded. The IRR was acquired with a vidicon tube cam-
era (Hamamatsu C1000) with a sensitivity peaking around
2100 nm employing a Kodak Wratten Filter 87A absorbing
light below 880 nm, behind the lens. Molynx IR550 contin-
uous infrared lights were used for illumination.
For the XRR an Andrex BW155 X-ray tube set to 20 kV
and 2 mA was employed. An Agfa D7 film in daylight wrap-
ping placed behind the painting was exposed for two min-
utes to the radiation.
The preliminary investigations by portable XRF were
done with a Tracer III-V (Keymaster Technologies, cur-
rently Bruker AXS, Karlsruhe, Germany) with a Rh anode,
operated at 35 kV and 1 µA. Nineteen spots on the surface
of the painting were analyzed for 200 s. The location of the
five spots discussed below is given in Fig. 1.
For the acquisition of elemental distribution images, the
painting was brought to beamline X7A of the NSLS (see
Fig. 2 Portrait of an Old Man,
Rembrandt van Rijn, ca. 1630,
oil on wood, private collection,
as X-ray radiograph and infrared
reflectograph (©Rene Gerritsen)
Author's
personal
copy
160 M. Alfeld et al.
Fig. 3 Experimental setup at beamline X7A of the NSLS
Fig. 3). In order to prevent the lower energy photons in the
broad-spectrum radiation emitted from the bending magnet
source from damaging the painting, the primary beam was
attenuated with a 5-mm-thick Al filter. This filter effectively
absorbed all radiation below 20 keV.
The painting was mounted on a motor stage and moved
for scanning continuously through the primary beam that
was collimated to 100 × 100 µm
2
by a slit system. The el-
emental distribution images shown below have a pixel size
of 100 µm and were acquired with a dwell time of 8 ms per
pixel for a total acquisition time of 8 h.
The fluorescence radiation emitted from the painting was
recorded by (an early prototype of) the Maia detector sys-
tem, comprising a monolithic array of 96 detector pads, each
with 1 mm
2
active area and 400 µm thickness. With an ac-
tive area of approximately 100 mm
2
, positioned at approx.
3.5 cm from the impact point of the beam on the painting,
the detector system covers a solid angle of 0.08 sr. This
is a multiple of the solid angle achieved by conventional
energy-dispersive detectors in synchrotron-based setups as
these have to be collimated in order to prevent an oversatu-
ration. This provides a significant increase in sensitivity and
with it scanning speed compared to conventional detectors
[16, 17].
To estimate the sensitivity of the system, an XRF spec-
trum from Standard Reference Material (SRM) 611 ‘Trace
elements in glass’ from the National Institute of Standards
and Technology (NIST) was recorded.
The imaging data was processed using Dynamic Analysis
[18] integrated in the GeoPIXE software package [19] while
the spectrum of the NIST SRM 611 was processed using
PyMCA [20].
A microscopic paint sample was taken from the painting
and embedded in polyacrylate resin. Further, it was polished
to expose a cross section showing the paint layer stratig-
raphy. During the embedding procedure, the surface paint
Fig. 4 Spectrum of NIST SRM 611 and sensitivities derived from it.
Only K
α
and L
α
lines are indicated. The spectrum was acquired in
120 s
layer with the varnish was lost, yet the underpainting layer
remained intact.
3 Results and discussion
3.1 Characterization of the MA-XRF scanner
In the NIST SRM 611 spectrum, shown in Fig. 4, signals
of a broad range of elements are visible. The elements Ca
(Z = 20) to Te (Z = 52) can be detected by means of the
K-level radiation. Furthermore, the L-level radiation of Pb,
Bi, Th and U contributes considerably to the spectrum. The
seemingly enhanced Cu signal is the result of scattered radi-
ation exciting the sample holder and is ignored for the cal-
culation of the sensitivity.
Apart from Ca, which is a matrix element of NIST SRM
611 (with a concentration of approx. 8.5 wt%), all elements
have a nominal concentration level of 500 ppm [21]; the in-
tensity of the recorded K-level fluorescence varies consider-
ably, however. This is due to two effects that influence the
sensitivity of the instrument. The 5-mm Al absorber in the
primary beam effectively absorbs all radiation below 20 keV
so that elements with lower absorption edges, such as Fe
(7.1 keV), are only weakly excited in comparison to heavier
elements such as Sr (K-absorption edge: 16.1 keV). Heavy
elements such as Ag (K-absorption edge: 25.5 keV, Ag-K
α
:
22.2 keV) are also well excited but less well detected, since
the quantum efficiency of the 400 µm-thick detector crys-
tals in this energy region is below 30 % as opposed to more
than 60 % for Rb-K
α
(13.4 keV) and nearly 100 % for Fe-
K
α
(6.4 keV). The sensitivity varies between 175 cps/(mg/g)
for Fe and over 12000 cps/(mg/g) for Mo.
As the elements of interest such as Fe, Cu, Hg and Pb are
expected to be present in a concentration range of several
weight percent, this sensitivity is sufficiently high to scan the
painting with a dwell time of a few milliseconds per point.
However, Ca (with a sensitivity of only 9 cps/(mg/g)) could
not be imaged at this pace.
Author's
personal
copy
Visualizing the 17th century underpainting in Portrait of an Old Man by Rembrandt van Rijn using 161
Fig. 5 (a) Local sum spectrum of 8 ×14 pixels from an area of high
Hg abundance in the old man’s face (indicated by blue cross). The Hg
lines are only visible as shoulders on the Pb lines. (b) Correction of Hg
distribution image from interference with Pb lines
The energy resolution of the detector was found to be
approx. 300 eV at the Mn-K
α
line (5.9 keV), consider-
ably poorer than modern silicon-drift detectors that achieve
better than half of this value. However, the energy resolu-
tion is sufficient to resolve peaks of neighboring elements
present at comparable concentration levels, see for example
Fe (Z = 26, E(Fe-K
α
) = 6.4 keV) and Co (Z = 27, E(Co-
K
α
) =
6.9 keV) in Fig. 4.
This is no longer true if the concentrations of both
elements differ considerably, as is the case for Hg and
Pb in the painting, where Hg is present in the red pig-
ment vermilion/cinnabar (HgS) that constitutes a minor
component compared to the major component lead white
((PbCO
3
)
2
·Pb(OH)
2
) (see Fig. 5a). Lead white is the main
white pigment used in the 17th century. As can be seen in
Fig. 5b, the interference between Hg and Pb cannot be fully
resolved under the chosen experimental conditions. This is
a result of several factors: (a) the large intensity of the lead
fluorescence compared to that of mercury, (b) the lack of a
full GeoPIXE model for white-beam excitation and (c) the
less than ideal energy resolution of the detector. However,
this interference can be largely corrected for by employing
Eq. (1).
I
∗
Hg,j
= I
Hg,j
− c · I
Pb,j
, (1)
Table 1 Intensities of Mn, Fe, Cu and Pb measured in selected spots
by means of a portable XRF instrument. The intensities are in counts
per second
Point Mn Fe Cu Pb Mn/Fe
1 3.7 110 26 1800 3.3 %
2 6.5 64 53 1400 10.1 %
3 7.4 71 39 1600 10.4 %
4 18 98 24 1800 18.3 %
5 12 110 65 1600 10.8 %
where I
i,j
is the measured intensity of element i in pixel j ,
I
∗
Hg,j
the corrected intensity of Hg in pixel j and c a correc-
tion factor.
It was assumed that the contribution of the Pb signal to
the measured Hg signal was linearly dependent on the Pb
signal intensity. So, as correction factor the slope of a linear
fit through a plot of measured Pb intensity against measured
Hg intensity in a reference area was used. As reference area
200 × 200 pixels in the background of the painting with a
weak measured Hg signal was chosen.
The corrected Hg distribution image gives a good rendi-
tion of areas with a high abundance of Hg but also features
artifacts in areas with no Pb present, most notably in the cor-
ners of the oval painting.
3.2 Spot measurements by portable XRF
In Table 1 the intensities recorded for the elements Mn, Fe
and Cu at the spots indicated in Fig. 1 are shown. These el-
ements are expected to be good elemental markers for the
visualization of the underpainting by means of MA-XRF.
A considerable variation of these elements could be ob-
served. While the other points show a comparable Mn/Fe
intensity ratio, point 1 features a considerably lower and
point 5 a considerably higher ratio. This suggests a lower,
respectively higher, amount of the earth pigment umber
present at these locations. Furthermore, points 3 and 5 fea-
ture stronger Cu signals than the other points. As this varia-
tion does not correlate with changes in the color of the sur-
face painting, it was decided to explore this compositional
contrast in more detail by means of XRF imaging.
3.3 Elemental distribution images
In Fig. 6 elemental distribution images of four elements are
shown that provide an insight in the hidden paint layers and
those at the surface. Fe is present in earth pigments that were
used to paint the surface composition, especially the old
man’s clothes and the shadows in his face. Other earth pig-
ments were used in an overpainted sketch of features in the
background, possibly including a horizon in the lower right
corner. Also, the hat visible in the IRR is faintly discernible
in this image. Otherwise no hidden paint layers belonging to
Author's
personal
copy
162 M. Alfeld et al.
Fig. 6 Elemental distribution
images of Fe, Cu, Hg and Pb
obtained at beamline X7A of the
NSLS
Fig. 7 Left: silhouette of an
unfinished self portrait of
Rembrandt, superimposed on
the Cu distribution image.
Right: Rembrandt van Rijn, Self
portrait, 1630, 15.5 × 12 cm
2
.
The red ‘+’ indicates the
location from where a paint
sample was taken
Author's
personal
copy
Visualizing the 17th century underpainting in Portrait of an Old Man by Rembrandt van Rijn using 163
the underpainting are visible in the Fe map, mainly due to
the high abundance of earth pigments in the surface paint-
ing.
The Hg distribution image indicates the area where ver-
milion/cinnabar (HgS) was used to set the highlights in the
flesh tones of the old man’s face. The original headgear of
the person sketched is visible like in the IRR image as a faint
shadow, suggesting that Hg might be present throughout the
painting in low abundance.
The Pb distribution image offers information comparable
to the XRR, albeit with a considerably reduced contribution
from the panel the painting is executed on.
The Cu distribution image is not correlated to the surface
painting, but visualizes the underpainting applied to sketch
a self portrait in the style of Rembrandt (Fig. 7) by filling
in the background and leaving a ‘reserve’ for the figure in
the foreground. Based on the Pb and Cu distribution images,
the main contour lines of the original portrait could be re-
constructed, which shows a strong resemblance to other self
portraits of Rembrandt, such as the one present at the Na-
tional Museum of Sweden in Stockholm.
The absence of the hidden portrait in the Fe and Hg dis-
tribution images leads us to the conclusion that the original
representation was not finished and it was overpainted with
the portrait of the old man that is now visible.
The pigments which were used in the underpainting were
unknown. Cu is present in blue (azurite, Cu
3
(CO
3
)
2
(OH)
2
)
and green (e.g. verdigris, Cu(CH
3
COO)
2
·nCu(OH)
2
). These
are both pigments used in the 17th century, yet a clear green
or blue background was not expected from comparison with
other paintings by Rembrandt. To identify the pigments a
paint sample was taken.
3.4 Cross-section analysis
A sample representing all layers of the painting down to the
ground was taken from the upper right corner of the paint-
ing, as indicated in Figs. 1 and 6. During the preparation of
the sample, the surface paint layer and the varnish layer de-
tached themselves from the rest of the sample and were lost,
but the layers containing the underpainting and the ground
were preserved.
In Fig. 8, three layers can be discerned in the elemental
distribution images acquired by SEM-EDX. From the bot-
tom: the chalk-containing ground layer, followed by a Pb-
rich imprimatura layer on which the underpainting layer was
applied. EDX spectra were acquired at several spots in order
to confirm the presence of the elements indicated in the fig-
ure. These measurements also revealed the presence of Co in
the Si-rich glass particles. The underpainting layer contains
a wide range of pigments including copper green (proba-
bly verdigris), earth pigments, smalt (ground Co-rich glass),
lead white, carbon black and potentially calcite, resulting in
a brownish/greenish color.
Fig. 8 Microscopic image and secondary electron image (SEI) of the
paint sample’s cross section with elemental distribution images ac-
quired by SEM-EDX. Yellow lines indicate different layers
There is no apparent reason why an artist should have
mixed these pigments deliberately to create a brownish tone
that would have been difficult to be reproduced, and did
not choose a simpler mixture of pigments. We assume that
palette scrapings were used to execute the underpainting.
Each evening the palettes in a workshop were cleaned by
scraping off the dried paint. These (in sum brownish) scrap-
ings could be recycled by dissolving them in oil for sketch-
ing.
4 Conclusions
We succeeded in visualizing the underpainting under the
surface of a painting by Rembrandt van Rijn by means of
Author's
personal
copy
164 M. Alfeld et al.
MA-XRF, which was not possible using the conventional
imaging techniques IRR and XRR. From these results we
reconstructed a sketch of an unfinished self portrait.
In combination with a detailed study of the painting tech-
nique used to execute the painting and indirect references
to it in literature, the visualization of the sketched self por-
trait provided proof for the painting’s authenticity, as it is
highly unlikely that a copyist would have started and aborted
a copy of a self portrait in order to copy another painting
by Rembrandt. These material indications were considered
so convincing that the painting was acknowledged by the
Rembrandt Research Project and exhibited from December
2011 to October 2012 in the Rembrandthuis Museum, Am-
sterdam, as an original painting by Rembrandt van Rijn.
Furthermore, we demonstrated that the underpainting in
this panel consists of a highly heterogeneous mixture of pig-
ments, including copper-containing ones. It is not unlikely
that this brown pigment mixture is actually the recycled left-
overs of a palette scraping.
Acknowledgements This research was supported by the SSD pro-
gramme of BELSPO, Brussels (project S2-ART). The text also
presents results of GOA ‘XANES meets ELNES’ (Research Fund,
University of Antwerp, Belgium) and from FWO (Brussels, Belgium)
projects nos. G.0704.08 and G.01769.09. Further, the research leading
to these results has received funding from the European Community’s
Seventh Framework Programme (FP7/2007-2013) under grant agree-
ment no. 226716. M. Alfeld receives a Ph.D. fellowship of the Re-
search Foundation–Flanders (FWO). Use of the National Synchrotron
Light Source, Brookhaven National Laboratory, was supported by the
U.S. Department of Energy, Office of Science, Office of Basic En-
ergy Sciences, under Contract No. DE-AC02-98CH10886. We ac-
knowledge the assistance of C. Ryan, CSIRO Australia, in the prepa-
ration of the elemental maps using GeoPIXE and Rene Gerritsen
(http://www.renegerritsen.nl) in providing photographs, XRR and IRR
of the painting. We thank Sullivan Entertainment for documenting part
of this project in their TV documentary ‘Out of the shadows’.
References
1. D. Bomford, J. Kirby, A. Roy, A. Rüger, R. White, Art in the
Making: Rembrandt (National Gallery Company, London, 2006),
pp. 30–34
2. A.E.J. James, S.J. Gibbs, M. Sloan, J. Erickson, J. Diggs, Am. J.
Roentgenol. 140, 215 (1983)
3. J.R.J. Van Asperen de Boer, Appl. Opt. 7, 1711 (1968)
4. K. Janssens, J. Dik, M. Cotte, J. Susini, Acc. Chem. Res. 43, 814
(2010)
5. E.V. Sayre, H.N. Lechtman, Stud. Conserv. 13, 161 (1968)
6. M.W. Ainsworth, J. Brealey, M.J. Cotter, K. Groen, E.
Haverkamp-Begemann, P. Meyers, E.V. Sayre, L. van Zelst, Art
and Autoradiography: Insight Into the Genesis of Paintings by
Rembrandt, van Dyck and Vermeer (Metropolitan Museum of Art,
New York, 1982)
7. J. Dik, K. Janssens, G. Van der Snickt, L. van der Loeff, K. Rick-
ers, M. Cotte, Anal. Chem. 80, 6436 (2008)
8. M. Alfeld, K. Janssens, K. Appel, B. Thijsse, J. Blaas, J. Dik, Z.
Kunsttechnol. Konserv. 25, 157 (2011)
9. J. Dik, K. Janssens, M. Alfeld, K. Rickers, E. van de Wetering, in
Nouveau Regard sur Rembrandt, ed. by E. van de Wetering (Local
World, Amsterdam, 2010), pp. xx–xxiv
10. D.L. Howard, M.D. de Jonge, D. Lau, D. Hay, M. Varcoe-Cocks,
C.G. Ryan, R. Kirkham, G. Moorhead, D. Paterson, D. Thurrow-
good, Anal. Chem. 84, 3278 (2012)
11. L.S. van der Loeff, M. Alfeld, T. Meedendorp, J. Dik, E. Hen-
driks, G. Van der Snickt, K. Janssens, M. Chavannes, in Van Gogh
Studies 4: New Findings, ed. by L. van Tilborgh, D. van Halsema,
J. House, G. Weisberg (WBOOKS, Zwolle, 2012), pp. 33–53
12. M. Alfeld, K. Janssens, J. Dik, W. de Nolf, G. Van der Snickt,
J. Anal. At. Spectrom. 26, 899 (2011)
13. P. Noble, A. van Loon, M. Alfeld, K. Janssens, J. Dik, Technè 35,
36 (2012)
14. D. Bull, A. Krekeler, M. Alfeld, J. Dik, K. Janssens, Burlington
Mag. 153, 668 (2011)
15. M. Alfeld, W. De Nolf, S. Cagno, K. Appel, D.P. Siddons,
A. Kuczewski, K. Janssens, J. Dik, K. Trentelman, M. Walton,
A. Sartorius, J. Anal. At. Spectr. 28, 40 (2013)
16. C.G. Ryan, D.P. Siddons, R. Kirkham, P.A. Duun, A. Kuczewski,
G. Moorhead, G. De Geronimo, D.J. Paterson, M.D. de Jonge,
R.M. Hough, M.J. Lintern, D.L. Howard, P. Kappen, J. Cleverley,
AIP Conf. Proc. 1221, 9 (2010)
17. R. Kirkham, P.A. Dunn, A.J. Kuczewski, D.P. Siddons, R. Dodan-
wela, G.F. Moorhead, C.G. Ryan, G. De Geronimo, R. Beutten-
muller, D. Pinelli, M. Pfeffer, P. Davey, M. Jensen, D.J. Paterson,
M.D. de Jonge, D.L. Howard, M. Küsel, J. McKinlay, AIP Conf.
Proc. 1234, 240 (2010)
18. C.G. Ryan, B.E. Etschmann, S. Vogt, J. Maser, C.L. Harland,
E. van Achterbergh, D. Legnini, Nucl. Instrum. Methods Phys.
Res., Sect. B, Beam Interact. Mater. Atoms 231, 183 (2005)
19. C.G. Ryan, D.R. Cousens, S.H. Sie, W.L. Griffin, G.F. Suter,
E. Clayton, Nucl. Instrum. Methods Phys. Res., Sect. B, Beam
Interact. Mater. Atoms 47, 55 (1990)
20. V.A. Sole, E. Papillon, M. Cotte, P. Walter, J. Susini, Spectrochim.
Acta, Part B, At. Spectrosc. 62, 63 (2007)
21. N.J.G. Pearce, W.T. Perkins, J.A. Westgate, M.P. Gorton, S.E.
Jackson, C.R. Neal, S.P. Chenery, Geostand. Newsl. 21, 115
(1997)
Author's
personal
copy
