YOUNG ANTHONY VAN DYCK REVISITED: A MULTIDISCIPLINARY APPROACH TO A PORTRAIT ONCE ATTRIBUTED TO PETER PAUL RUBENS A 'Rubens' for the Rubenshuis

Abstract

Part of the collection of the Rubens House in Antwerp is a portrait of young Anthony van Dyck, alternatively attributed to Peter Paul Rubens and his pupil Anthony van Dyck. In order to reconstruct the genesis of the portrait in a manner that improves upon past investigations, a number of high-end technological methods, such as X-radiography, X-ray computer tomography, mammographic tomosynthesis and macroscopic X-ray fluorescence, have been employed to render the overpainted layers visible again. The results of the interdisciplinary examinations of the portrait of the youthful Van Dyck are impressive. The combined results allow the later additions to be peeled away until the original composition can be reached. Several pentimenti are easily discernible and refer to a rather immature hand that makes the authorship of Peter Paul Rubens very unlikely. What emerges is a portrait of an ambitious young man with a luxuriant head of hair and a slightly turned-up collar. The hat and cape were added later. The facial features are more recognisable and the execution of the bold curls points irrefutably in the direction of Anthony van Dyck as the author of his own portrait.

Full text

21 ArtMatters, 6, 2014
YOUNG ANTHONY VAN DYCK
REVISITED: A MULTIDISCIPLINARY
APPROACH TO A PORTRAIT ONCE
ATTRIBUTED TO PETER PAUL RUBENS
Katlijne Van der Stighelen, Koen Janssens, Geert Van der
Snickt, Matthias Alfeld, Ben Van Beneden, Bert Demarsin,
Marc Proesmans, Guy Marchal and Joris Dik
ABSTRACT Part of the collection of the Rubens House in Antwerp is a portrait of young Anthony van Dyck, alternatively
attributed to Peter Paul Rubens and his pupil Anthony van Dyck. In order to reconstruct the genesis of the portrait in a
manner that improves upon past investigations, a number of high-end technological methods, such as X-radiography,
X-ray computer tomography, mammographic tomosynthesis and macroscopic X-ray uorescence, have been employed
to render the overpainted layers visible again. The results of the interdisciplinary examinations of the portrait of the
youthful Van Dyck are impressive. The combined results allow the later additions to be peeled away until the original
composition can be reached. Several pentimenti are easily discernible and refer to a rather immature hand that makes
the authorship of Peter Paul Rubens very unlikely. What emerges is a portrait of an ambitious young man with a luxuriant
head of hair and a slightly turned-up collar. The hat and cape were added later. The facial features are more recognisable
and the execution of the bold curls points irrefutably in the direction of Anthony van Dyck as the author of his own
portrait.
A ‘Rubens’ for the Rubenshuis
Thanks to the strenuous efforts and the condence of the
Friends of the Rubenshuis, an impressive portrait of the
young Anthony van Dyck was added to the collection of the
Rubenshuis in Antwerp in the spring of 1995 (oil on panel,
36.5 × 25.8 cm). The reconstructed interior of the house in
which Rubens once lived and worked would be the ideal
place in which to hang the master’s portrait of his pupil.
This is how the portrait of ‘Young Anthony’ made its way
back from the Kimbell Art Museum in Fort Worth, Texas, to
the city where it had been painted, almost 400 years before
(Figure 1). No one questioned the painting’s authenticity.
Every expert who had seen it concluded that it was an early
seventeenth-century portrait. The then honorary curator,
Frans Baudouin, was unequivocal in attributing the work
to Peter Paul Rubens. Here he found himself in agree-
ment with Hans Vlieghe, endorsing the view expressed by
Ludwig Burchard.1 The relevant oeuvre catalogue identies
the portrait as that of Anthony van Dyck, and attributes it
to Peter Paul Rubens: ‘Critics other than Burchard have
generally attributed this portrait to Van Dyck himself …
There is also a remarkable similarity of style with Jan
Brueghel’s little son in the somewhat earlier family por-
trait in the Princes Gate Collection’ (Figure 2). The work
displays striking formal similarities to the portrait of Pieter
Brueghel the Younger (born in 1607/1608). Since the boy
appears to be about 6 or 7 years old, the family portrait is
generally dated to about 1612/13.
A self-portrait by the young Van Dyck?
In the same catalogue note, we read that before Burchard
attributed the portrait to Rubens, the work was universally
believed to be by Anthony van Dyck and hence to be a self-
portrait of the young painter. Otto Benesch was the rst to
attribute the portrait to Van Dyck, in 1954, on the basis of

KATLIJNE VAN DER STIGHELEN, KOEN JANSSENS ET AL
YOUNG ANTHONY VAN DYCK REVISITED
22 ArtMatters, 6, 2014
photographs shown to him while it was still in the collection
of Frederick W. Mont in New York.2 Michael Jaffé (1966),3
Gregory Martin (1970),4 Götz Eckhardt (1971)5 and Alan
McNairn (1980)6 all concurred with this opinion. The 1972
and 1981 catalogues published by the Kimbell Art Museum
in 1972 and 1981 also included the work as a self-portrait by
Van Dyck.7 Nora De Poorter does not include the work in her
critical catalogue of Van Dyck’s oeuvre. She does, however,
express parenthetical reservations concerning its attribu-
tion to Rubens in her comment: ‘assuming that it is by P.P.
Rubens’. A completely divergent opinion has been advanced
by Eric Larsen, who published the portrait as a work by the
Scottish portrait painter George Jamesone (1587–1644), a
view that no other researcher has endorsed.8
The attribution of the portrait of the young man to Van
Dyck was reinforced by Otto Benesch’s discovery in the
Albertina, Vienna, of a drawing reproducing every detail
of the composition of the little painting in a smaller size
(paper, 139 × 121 mm, pen and ink with brown and grey
washes, and sanguine over a black chalk underdrawing):
‘The examination is based on a meticulous but sensitive
little drawing in ink and chalks which has lain virtually
unremarked in the Albertina’ (Figure 3).9 Benesch initially
took the view that the drawing was a copy of the portrait
in the Mont collection, but he gradually became convinced
that it had been executed by Anthony van Dyck himself.
There are three inscriptions on this sheet, of which the
oldest is probably the one in the lower right corner, now
barely legible: ‘van Dijck[?]ipse fecit’. Next to that, in capi-
tals, appears the name ‘ANTHONY VDIJCK’ (V and D in
ligature), which is repeated as ‘Wandick’. Michael Jaffé
also saw this sheet as an authentic drawing by Anthony
van Dyck.10 This view soon attracted criticism, and very
few experts today regard the drawing in Vienna as a
work by Van Dyck’s own hand. Its style is descriptive and
reects neutral registration rather than creative design.
Figure 1. (a) Anthony van Dyck, Self-portrait, c. 1616/17, panel, 36.5 × 25.8
cm. Formerly said to be a portrait of Anthony van Dyck by Peter Paul Rubens.
Antwerp, Rubenshuis. (b) Detail.
a
bFigure 2. Peter Paul Rubens, Portrait of Jan Brueghel and his Family,
c. 1612/13, panel, 124.5 × 94.6 cm. London, Courtauld Institute Galleries,
Princes Gate Collection.

KATLIJNE VAN DER STIGHELEN, KOEN JANSSENS ET AL
YOUNG ANTHONY VAN DYCK REVISITED
23 ArtMatters, 6, 2014
Alan McNairn dated the drawing to the eighteenth cen-
tury, while Hans Vlieghe described only the inscription in
capitals as having been made by ‘an eighteenth-century
hand’.11 Although many questions remain unanswered, all
the authors quoted accept that the portrait painting depicts
Anthony van Dyck. Comparison with the oldest known
self-portrait of the young Van Dyck, which is preserved in
the Akademie der bildenden Künste in Vienna (Figure 4),
brings out the striking resemblance.12
Regardless of whether the (presumably) oldest of the
inscriptions on the drawing in the Albertina dates from
the seventeenth or eighteenth century, it remains highly
signicant that the draughtsman possessed information
enabling him to label the likeness a self-portrait: ‘ipse fecit’.
Furthermore, all the art historians who have subjected
the little portrait to scrupulous analysis are more or less
in agreement about its date. Jaffé dated the self-portrait
that is now in Vienna to around 1612/13, since he remarks:
‘Precocity is evident in his well-known Self-Portrait;
painted, as the appearance suggests, when van Dyck was
about thirteen, a youth of highly nervous sensibility and
exceptional promise.’13 In comparison to the little painting
from the Akademie in Vienna, he describes the sitter in the
portrait from the Kimbell Art Museum, now preserved in
the Rubenshuis, as ‘a year or so older’.14 This conclusion
implies that the self-portrait in the Rubenshuis was made
around 1613/14. Hans Vlieghe proposes a somewhat later
date, c. 1615/16.15 In short, according to the current state of
research, the painting in Antwerp is a portrait of the young
Anthony van Dyck and was painted one or two years later
than the early self-portrait in Vienna’s Akademie.
The attribution to Peter Paul Rubens
In our view, the attribution to Rubens was based entirely
on stylistic grounds. The arguments supposedly corrobo-
rating this attribution are advanced without any mention
of the specic context in which the portrait was made.
What is more, it is impossible to prove that Rubens and
Van Dyck – who was 22 years his junior – were already
acquainted in the period 1613–1616. In 1610, the young
Anthony had been admitted as a pupil to the studio of the
prominent Antwerp painter Hendrik van Balen. How long
he stayed there is unclear, but it is generally accepted that
such apprenticeships lasted for at least four or ve years.
This makes it extremely likely that the portrait in the
Rubenshuis in Antwerp dates from the time in which Van
Dyck was trying to build up a career of his own as a ‘master
pupil’. He did not join the St Luke’s Guild until 1618.
The young painter’s family circumstances were difcult,
to say the least. His mother, Maria Cuypers, had died in
1607. His two elder sisters, Catharina and Maria, had both
married, in 1610 and 1614, respectively: their husbands
were the notary Adriaen Diericx and the merchant Lancelot
Lancelots. These two sons-in-law had been appointed
guardians of the minor children of Franchois van Dyck
senior and the late Maria Cuypers, and sought to protect the
inheritance deriving from their wards’ grandmother and
mother. In 1615 they asked the city magistracy to give them
Figure 3. Anonymous drawing after a Self-Portrait by Anthony van Dyck, late
seventeenth or eighteenth century, paper, 139 × 121 mm. Vienna, Albertina.
Figure 4. Anthony van Dyck, Self-portrait, c. 1615, panel, 22.5 × 19.5 cm.
Vienna, Akademie der Bildenden Künste.

KATLIJNE VAN DER STIGHELEN, KOEN JANSSENS ET AL
YOUNG ANTHONY VAN DYCK REVISITED
24 ArtMatters, 6, 2014
permission at the earliest opportunity to sell ‘nine of the
best paintings bequeathed by the children’s grandmother’
on account of ‘the disgrace into which their father-in-law,
Franchoys van Dyck, had fallen’.16 On 3 December that
year, Anthony van Dyck instituted legal proceedings. He
alleged that his brothers-in-law had unlawfully taken over
the administration of the family inheritance and demanded
that they be called to account before a commissioner, him-
self and ‘one of his close friends’.
Less than one year later, on 13 September 1617, Anthony
van Dyck asked Antwerp’s city magistracy for permission to
protect not only his own goods but also those of his younger
sisters and brother from the guardians’ actions. His family
situation in the period 1613–1616 can be described, without
exaggeration, as turbulent.
He was certainly active as a gifted trainee painter but
the key question is – where? There is nothing whatsoever
to indicate that he was still working in Van Balen’s studio
after 1613/14. What is more, several sources suggest that
Anthony van Dyck was running his own studio between
about 1615 and 1618 in the house known as ‘Den Dom
van Ceulen’ on what is now Mutsaertstraat in Antwerp.
Herman Servaes and Justus van Egmont wrote that they
were working there as his pupils. The theory of a ‘studio of
teenagers’ has often been attacked, but no viable alternative
explanation has ever been advanced.17 That Van Dyck was
active – and valued − as a painter is clear from the fact that
in 1617, before he was registered as a master, he executed
an important commission for Antwerp’s Dominican order.
This was a monumental Christ Carrying the Cross (panel,
211 × 161.5 cm) (Figure 5), which he painted as part of a cycle
of 15 paintings depicting the Mysteries of the Rosary for the
church of St Paul (Sint-Pauluskerk) in Antwerp, where they
still hang today. Rubens’s contribution, The Flagellation,
is inscribed 1617, a date that other sources conrm. It must
have been a great honour for the young painter to be asked
to work alongside the best Antwerp masters of the day. The
paintings are all more or less the same dimensions, but
the prices paid for them varied considerably. Hendrik van
Balen, Anthony’s former teacher, received 216 guilders
while Peter Paul Rubens, Jacob Jordaens and the aspir-
ing master Van Dyck each received 150 guilders. Does it
not make sense to conclude that Van Balen only received
more because he was responsible for contracting out the
entire series? Was Van Dyck selected thanks to his former
teacher?18 It is hard to believe that Anthony did not benet
from a personal recommendation – Cornelis de Vos, an
established portrait and history painter and a guild mem-
ber since 1608, received only 138 guilders for his Nativity.19
Jan van den Broeck, named as the man who commis-
sioned Van Dyck’s Christ Carrying the Cross in St Paul’s
Church (Sint-Pauluskerk), may also have been instrumen-
tal, for instance, in encouraging his ‘good friend’ Jan Charles
de Cordes (Figure 6) and his second wife, Jacqueline van
Caestre, to commission their portraits from the young
Anthony van Dyck.20
That Peter Paul Rubens and the young Anthony van
Dyck met in the years before Van Dyck became a master
is indisputable. They collaborated from 1617 onwards, but
it cannot be ruled out that this collaboration was arranged
through the good ofces of Hendrik van Balen, Anthony
van Dyck’s former teacher. In addition, one should not
forget how Rubens himself dealt with portraits in this
period. By about 1615, Rubens had already painted three
self-portraits, depicting himself together with others. In
the Self-portrait with Friends in Mantua (c. 1605), the
Honeysuckle Bower with his wife Isabella Brandt (1608)
and the Four Philosophers, he poses next to or among
kindred spirits. According to a number of leading authori-
ties, the oldest individual self-portrait of Rubens also dates
from around 1615: this is the self-portrait in the Ufzi in
Florence (oil on panel, 78 × 61 cm; enlarged) (Figure 7).
It is generally believed that Rubens only painted the face
Figure 5. (a) Anthony van Dyck, Christ Carrying the Cross, 1617, panel, 211
× 161.5 cm. Antwerp, St Paul’s Church. (b) Detail.
a
b

KATLIJNE VAN DER STIGHELEN, KOEN JANSSENS ET AL
YOUNG ANTHONY VAN DYCK REVISITED
25 ArtMatters, 6, 2014
and that the other parts were added at the beginning
of the eighteenth century when the painting was prob-
ably presented as a gift from Johann-Wilhelm, Duke of
Neuburg and Elector Palatine, to Cosimo III, Grand Duke
of Tuscany, for his gallery of self-portraits.21 The earliest
version of this self-portrait is very similar to the little por-
trait of the ‘Young Anthony’. If one focuses solely on the
head and shoulders, the two compositions are virtually
identical. Both painters are depicted in three-quarters
prole, inclined to the left, and meet the viewer’s eyes
squarely. In each of these portraits, the upper body is
shown more in prole than the face, which enhances the
liveliness of the pose. Rubens’s face is more strongly illu-
minated than that of the young Van Dyck. Rubens posed
without a hat; in the version of the portrait of the youthful
Van Dyck being discussed here, his impressive headgear
casts a broad shadow across his forehead. If the master
and the talented young man were indeed acquainted at
this time, it makes sense to conclude that Van Dyck was
inspired by Rubens’s example. We cannot entirely dismiss
the idea that Rubens himself may have derived inspira-
tion from an older self-portrait of a young painter, most
notably Willem Key (Berlin, Staatliche Museen zu Berlin,
oil on paper on panel, 44.5 × 34.5 cm, made shortly after
1542). Recently, the execution of Rubens’s copy after the
self-portrait of the young painter with the cap (Munich,
Bayerische Staatsgemäldesammlungen, panel, 43.8 × 35.3
cm) was likewise dated to the years 1615–1618 (Figure 8).22
Although Willem Key’s diagonal pose, with his head held
slightly to one side and the slightly affected gesture of the
vir elegans deviates from the approach taken by Rubens
and Van Dyck, there are nonetheless striking parallels.23 If
Rubens considered making a portrait of his pupil, it seems
improbable that he would have presented it ‘as if it were a
self-portrait’. Such an assumption is baseless.
Figure 6. Anthony van Dyck, Portrait of Jan Karel de Cordes, c. 1617/18,
panel, 72.2 × 57 cm. Brussels, Museum of Fine Arts.
Figure 8. Peter Paul Rubens after Willem Key, Self-portrait, c. 1615/18, panel,
43.8 × 35.3 cm. Munich, Bayerische Staatsgemäldesammlungen.
Figure 7. Peter Paul Rubens, Self-portrait, c. 1615, panel (enlarged), 78 ×
61 cm. Florence, Ufzi.

KATLIJNE VAN DER STIGHELEN, KOEN JANSSENS ET AL
YOUNG ANTHONY VAN DYCK REVISITED
26 ArtMatters, 6, 2014
The portrait of the young Van Dyck in the Rubenshuis,
it may be added, bears an extraordinarily strong resem-
blance to the head in Van Dyck’s painting Daedalus and
Icarus in the Art Gallery of Ontario, Toronto (canvas, 115.3
× 86.4 cm) (Figure 9). Although J. Douglas Stewart attrib-
uted the composition to Pieter Thys in 1997 and dated it to
c. 1655–1660, there appears to be little reason to hold rm
to this hypothesis.24 Nonetheless, the author quite rightly
notes that the painting does not entirely t in the early
initial Antwerp period. It is a thoroughly ‘classical baroque’
picture, in handling, composition and sentiment’.25 The
dating of the work has also given rise to lively debate in
the past, with opinions divided between the rst Antwerp
period (around 1620) and the second. In the exhibition The
Young Van Dyck, the painting was described as ‘one of the
most puzzling pictures in the exhibition’. Alan McNairn
observed: ‘The Italianate quality of the composition,
colour and technique of painting has caused some doubt
that these paintings belong to Van Dyck’s First Antwerp
period. However, in the case of Daedalus and Icarus, the
facts that the partially draped youthful Icarus is a modied
self-portrait of the artist, and that the head of Daedalus is
close to that of St Judas Thaddeus, suggest that this is an
early work.’26 The same author is impressed by the way-
ward hairstyle that the painter gave himself.27 In any case,
the painting in Toronto was certainly produced later than
the self-portrait in Antwerp’s Rubenshuis. The relatively
mature workmanship in Daedalus and Icarus compels us
to allow for the possibility that the small portrait of ‘Young
Anthony’, too, may date from somewhat later than 1615.
Alan McNairn’s comments on the hairstyle are interest-
ing. Van Dyck’s early works are particularly noteworthy
for the unorthodox hairstyle of his models, regardless of
whether these are apostles or minor secular gures. His
Figure 9. Anthony van Dyck, Daedalus and Icarus, c. 1618–20, canvas, 115.3 × 86.4 cm. Toronto, Art Gallery
of Ontario.

KATLIJNE VAN DER STIGHELEN, KOEN JANSSENS ET AL
YOUNG ANTHONY VAN DYCK REVISITED
27 ArtMatters, 6, 2014
earliest self-portrait displays a characteristically tousled
mass of varicoloured curls around the pale face, and in his
early Apostle series too, besides varying the saints’ poses,
Van Dyck modied the texture of their hairstyles to reect
their emotional state. The portrait in the Rubenshuis is an
exception in this respect − at rst sight at least.
What conclusions can we draw from comparative
research and the scant information on the work’s prov-
enance? How could art historical research benet from an
integrated, interdisciplinary approach?
Cross-disciplinary research
In the past, problems of attribution were often resolved
by technical analysis in a laboratory. The development of
the Rembrandt Research Project reects the optimism sur-
rounding this approach. Now that the catalogue raisonné
has been completed, it appears that many problems remain
unresolved. Recently, Anna Tummers wrote a fascinating
book on aspects of the paradox of seventeenth-century con-
noisseurship.28 Relatively little interdisciplinary analysis has
been performed on the work of Rubens.
The traditional approach relies primarily on
X-radiography. For instance, it is clear from Alan McNairn’s
catalogue that the self-portrait of the young Van Dyck was
examined very thoroughly when it belonged to the collec-
tion of the Kimbell Art Museum in Fort Worth, Texas. The
author describes the history of the painting as follows:
Although the panel retains the original compo-
sition, it has been repainted considerably. The
wide-brimmed hat is modern, though the shadow
on the forehead is original. At some point the origi-
nal hat was removed by a restorer who found a full
head of hair beneath it. It was then discovered that
the shadow on the forehead was original, and there-
fore the hat was recreated. In 1971 the hat was again
removed, revealing a high forehead and a consider-
able amount of hair. The hat was again repainted
for the reason that the shadow made no sense with-
out it. It is quite likely that van Dyck painted this
Self-Portrait without a hat. He then modied the
composition by painting the hat over the hair.29
Scientic examination
In order to reconstruct the genesis of this much-debated
portrait in a manner that improves upon past investiga-
tions, a number of high-end technological methods have
been employed to render the overpainted layers visible
again. Radiography, which makes use of a wide beam of hard
X-ray radiation that is highly penetrative, and mammo-
graphy, which uses softer X-rays to maximise the contrast
between soft tissues, were employed. In modern medical
practice, by recording a series of digital radiographs or
mammographs under various angles of the X-ray beam rela-
tive to the patient and appropriate mathematical processing
of the resulting series of images, three-dimensional inform-
ation can also be obtained. These types of measurements
are referred to as X-ray computer tomography (CT) and
mammographic tomosynthesis respectively; for this pur-
pose, the painting was transported to the Academic Hospital
of the KU Leuven for a short duration.30 While conventional
two-dimensional imaging methods such as X-radiography
or mammography permit the recording of a single pro-
jected (and thus two-dimensional) view of all layers that
are present beneath the surface, the more advanced three-
dimensional equivalents of these imaging methods allow us
to inspect individual, millimetre-thick layers of the paint-
ing separately thus enabling us to study the way they are
structured. For example, the disturbing contribution of par-
queting may be eliminated in this manner. Another method
that has recently been employed successfully for subsurface
imaging of paintings31 and that may be performed in situ is
macroscopic X-ray uorescence (MA-XRF). This method
uses a ne pencil beam of X-rays to scan the surface of a
painting and yields one or more distribution images of the
chemical elements, such as lead, mercury, copper and iron
present in the paint.
Two-dimensional X-ray radiography and
mammography
In Figure 10, a detail of a digital X-ray radiograph and a
digital mammograph of the painting are juxtaposed. The
digital radiograph was obtained by means of an Agfa Gevaert
DX-G instrument, equipped with a needle cassette that
can be read out at the 100 µm level. A Siemens Mamomat
Inspiration with a pixel size of 85 µm was employed to
record the mammographic image. In principle, both instru-
ments therefore allow details to be visualised down to the
level of 0.1 mm. Because the major component absorbing
the X-rays is lead, the lighter areas of the representation in
particular, where lead white pigment is abundantly present,
show up in these images. While the parqueting beams at the
back of the panel to some extent obscure the images, it can
already be concluded that below the surface a more rotund
face with higher forehead and an abundant head of hair is
present. From the overview picture it is also apparent that
the person depicted originally wore a form-tting garment
instead of the loosely wrapped cape that is now visible on
the surface.
X-ray tomography and mammographic tomosynthesis
In Figure 11, the improvements in image clarity that may
be gained using X-ray computer tomography or mammo-
graphic tomosynthesis are shown. To record the former, a
Siemens Somatom Denition Flash CT scanner was used in

KATLIJNE VAN DER STIGHELEN, KOEN JANSSENS ET AL
YOUNG ANTHONY VAN DYCK REVISITED
28 ArtMatters, 6, 2014
which the X-ray tube and image detector rotate over 360o
around the painting. The tomosynthesis procedure involved
the recording of a series of 25 mammographs while the X-ray
tube rotated over a range of 50° relative to the painting. In
both cases, the shape of the face is now more clearly visible
while especially in the tomosynthesis image, more topological
information appears to be present, suggesting for example
that the placement of the eyes was changed: the person’s left
eye may originally have been positioned somewhat lower
while the right one was not as close to the nose as it is now.
MA-XRF imaging
The MA-XRF allows us to study the distribution of various
pigments on or beneath the surface of the painting and
is sometimes referred to as ‘colour X-ray radiography’.32
Since the signals of the various elements do not all emerge
from the same depth below the surface, combining different
elemental maps also provides information on the hidden
paint layers. The painting of ‘Young Anthony’ was investi-
gated by MA-XRF in the Rubenshuis conservation studio.
The complete portrait was scanned over a period of 24 hours
and using a lateral resolution dened the step size during the
scan (1 mm); the dwell time per point was 1 second.
In Figure 12, the iron, calcium and lead distributions are
compared to a black and white photograph of the painting.
The iron image most resembles the visible photograph:
both the hat and the mantle of the gure and the position-
ing of the eyes are similar, however, the adjustments made
to render the face less rotund are very clear in this image.
Surprisingly, the calcium image, while still illustrating the
later version of the mantle and lace collar, only shows a
slight shading at the level of the hat and the shadow on the
forehead. In this image, both the full head of hair and the
more rotund aspect of the face are evident thereby revealing
the rst sketch. The presence of this ground layer undoubt-
edly refers to the young painter who, experimenting with
new schedules and styles, used his own head as a test case.
Also the displacement of the right eye of the gure, close
to the nose, is clearly visible. Finally, as expected, the lead
image most resembles the X-ray tomography/tomosynthe-
sis images of Figure 11: the original shape of the mantle,
the smaller, more modest collar and the original position-
ing of both eyes are all clearly visible. Since the hat and
Figure 10. Anthony van Dyck, Self-portrait, c. 1616/17, panel, 36.5 × 25.8 cm. Antwerp, Rubenshuis. (a) Overview
radiograph; (b) detail of left panel; (c) mammograph of the same area shown in the middle panel.
Figure 11. Anthony van Dyck, Self-portrait, Antwerp, Rubenshuis: (a) X-ray computed tomogram and (b)
mammographic tomosynthesis image obtained from the facial area of the painting.

KATLIJNE VAN DER STIGHELEN, KOEN JANSSENS ET AL
YOUNG ANTHONY VAN DYCK REVISITED
29 ArtMatters, 6, 2014
the shade on the forehead no longer dominate the image,
an undisturbed view of the original portrait is now pos-
sible. To the left of the gure, an arc-shaped structure in
the background is also visible in this distribution. In both
the calcium and lead images, a letter-like inscription can
be seen which, however, does not appear to be composed
of recognisable characters.
Some additional elemental maps are shown in Figure 13
and compared to a colour photograph of ‘Young Anthony’.
The mercury images (present in the red pigment cinnabar/
vermilion) show how in a later stage, red and brown-toned
paint was used to create the sweeping mantle that is now
visible on the surface but also that the original hair had
a brown/red aspect. In the copper image (corresponding
to the use of the blue pigment azurite and/or to a copper
green), the original form of the mantle, already visible in
the lead distribution, can be seen again. However, behind
the gure, a wing-like blue/green structure is also visible. In
shape and position, this angel’s wing resembles that visible
in Daedalus and Icarus (Figure 9). Finally, in the strontium
image, although noisy, the original shape of the garment
can be discerned.
Discussion
The results of the abovementioned examinations of this
portrait of the youthful Anthony van Dyck are impressive.
The combined results allow the later additions to be peeled
away, one by one, until we reach the original composition.
What then emerges is a portrait of a self-assured young man
with a luxuriant head of hair and a slightly turned-up col-
lar. The hat and cape were added later. The facial features
are more recognisable and the execution of the bold curls
Figure 12. Anthony van Dyck, Self-portrait, Antwerp, Rubenshuis: (a) black and white photograph of ‘Young Anthony’
and corresponding distributions of (b) calcium (inverted), (c) iron and (d) lead obtained by MA-XRF.

KATLIJNE VAN DER STIGHELEN, KOEN JANSSENS ET AL
YOUNG ANTHONY VAN DYCK REVISITED
30 ArtMatters, 6, 2014
points irrefutably in the direction of Anthony van Dyck as the
author of this portrait. It remains unclear when the drastic
pentimenti were added – only the analysis of paint samples
can resolve this mystery. The hat section is of such dubious
quality that it seems very unlikely to have been executed by
Van Dyck himself. The question therefore arises of whether
the shadow was part of the original composition. It should be
added that the shape of the hat in the painted portrait differs
from that as it appears in the later drawing at the Albertina
(Figure 3) – in the drawing, the hat has a strikingly wider rim
and projects much further forward. If a similar comparison
is made between the face of the young painter in the Vienna
self-portrait (Figure 4) and the portrait in Antwerp (Figure 1),
there appears to be a striking difference in age. The face has
become narrower and more elongated. It also appears that
in subsequently shifting the pupils towards the corners of the
eyes, Van Dyck was modifying the image to correspond to the
earliest version of his self-portrait. This shift accentuates the
rotation of the face, thus heightening the portrait’s dramatic
expression. It is also striking, however, that Van Dyck has
depicted the head more vertically in the painted surface in
comparison to the early self-portrait in Vienna. The vertical
axis that passes through the face runs parallel to the sides
of the panel, as in the case of the ‘hidden self-portrait’ in the
painting with Icarus in Toronto (Figure 9). In the portrait in
Vienna, a diagonal line runs through the face. It is interesting
to see that in his later self-portraits his head leans forward a
little more, bringing out the bracket-like shape of his mouth
more distinctly. This applies, for instance, to the Self-portrait
in the Metropolitan Museum of Art in New York, which can
be dated to around 1622 (Figure 14), but in this portrait also
the position of the eyes is very similar. Not one self-portrait
provides a frontal view of the painter; he preferred to depict
himself from a three-quarters angle.
Figure 13. Anthony van Dyck, Self-portrait, Antwerp, Rubenshuis: (a) colour photograph of ‘Young Anthony’ and
corresponding distributions of (b) mercury, (c) copper and (d) lead obtained by MA-XRF.

KATLIJNE VAN DER STIGHELEN, KOEN JANSSENS ET AL
YOUNG ANTHONY VAN DYCK REVISITED
31 ArtMatters, 6, 2014
In order to answer the question ‘How would Van Dyck’s
face have looked from the front?’, a technique called 3DMM
was employed.33 This is used for extracting the three-
dimensional shape of human faces from photographs and
is based on the use of statistical information concerning the
shape and the skin colour of human faces. These detailed
statistics have been gathered by studying a substantial data-
base of high-resolution facial 3D scans. By simulating the
way in which light is reected from the face, a computer
algorithm is capable of automatically estimating the most
probable shape of a person’s face. Thus, depending on how
well the painter was able to capture the shape and colour
of the face, a life-like representation of the real face may
be obtained via the 3DMM reconstruction. Some results
are shown in Figure 15. We believe that on the basis of
this computer model, in the future it will become easier to
compare different paintings and make the identication of
the same person in different poses and different paintings
more reliable. The results of this research show that Van
Dyck was intuitively aware of the advantages of depict-
ing his face from a three-quarters angle – it strengthens
the vitality and elegance of the pose and makes his physi-
ognomy more interesting.34
Returning to the comparison between the two oldest
known self-portraits of Van Dyck (Figures 1 and 4), the
two paintings were executed in mirrored composition.35 It
is therefore almost impossible that the version from the
Rubenshuis was painted before 1616/17, the period when
we can be certain that the master Rubens and the talented
young Van Dyck were acquainted. The age gap − 22 years –
places the two artists in different generations.36 By around
1617, Van Dyck was starting to realise that a confrontation
with Rubens’s ‘classical’ style was inevitable. The self-
portrait in Antwerp may be seen as one of the earliest products
Figure 14. Anthony van Dyck, Self-portrait, c. 1620, canvas, 119.7 × 87.9 cm. New York, Metropolitan Museum
of Art, The Jules Bache Collection.

KATLIJNE VAN DER STIGHELEN, KOEN JANSSENS ET AL
YOUNG ANTHONY VAN DYCK REVISITED
32 ArtMatters, 6, 2014
of this realisation. Abandoning the impasto of the painting in
Vienna, he used a far smoother, sculptural technique here,
most notably in depicting esh. When the work is examined
with the aid of the technique that is generally used in mam-
mography, the resulting image shows that the impasto of the
reddish brown hair underneath the hat is essentially the same
as the luxuriant mass of curls in the self-portrait in Vienna.
The edge of the double collar in the original Antwerp version
is executed sketchily, as is the bold white brushstroke that
serves to indicate the collar of the white shirt in the earliest
self-portrait in Vienna. Whereas the face in the latter is mod-
elled with brushstrokes, in the Antwerp portrait the painter’s
hand is scarcely discernible in the esh of the face. The shades
of colour change far more gradually and the curves are sug-
gested with real chiaroscuro accents under the chin and near
the ears. The lips are full, deep red and turned up, exhibiting
far more tension than in the earlier work. The inuence of
Rubens is integrated perfectly here.
Once Van Dyck could produce a self-portrait of this
kind ‘in the style of Rubens’, he was ready to compete at
the highest level. Not much later Van Dyck produced two
versions of ‘St Martin’, of which the rst is still preserved
in St Martin’s Church (Sint-Martinuskerk) in Zaventem
(Figure 16), while the second is in the UK’s Royal Collection.
The Zaventem version is undoubtedly the oldest, and was
probably executed around 1620. Van Dyck based the com-
position on an older oil sketch by Rubens that is dated to the
period 1609–1613.37 Rubens’s prototype served as the point
of departure, to which Van Dyck made minor changes. Every
aspect of the large painting’s execution reects Rubens’s con-
cept of style at that time. A similar situation arose in the
Antwerp self-portrait, in which the young pupil was trying to
master the idiom of the leading master of his day. Van Dyck
had trained so rapidly that his hand was soon barely distin-
guishable from that of Rubens. On 6 July 2011, the Portrait
of a Carmelite Monk was sold at Sotheby’s in London. The
portrait is of a sublime quality and was frequently linked to
Rubens (Figure 17). However, when the work was subjected
to analysis by several specialists familiar with the oeuvre
of the young Van Dyck, it was immediately clear that Van
Dyck, not Rubens, was the author of this fascinating por-
trait. The new attribution is corroborated by research on the
oldest sources on this work’s provenance.38 The date of the
cleric’s portrait is uncertain, but most experts believe that
it was painted around 1618, when Van Dyck was working
as Rubens’s assistant. A step-by-step analysis of the self-
portrait from the Rubenshuis, the cleric’s portrait, and the
execution of the face of St Martin in the Zaventem work
reveals how the young Van Dyck absorbed and adapted the
stylistic inuence of Rubens in the period from 1617 to 1620.
Figure 15. Anthony van Dyck, Self-portrait, Antwerp, Rubenshuis: 3DMM reconstruction of the frontal face of Anthony van Dyck.
Figure 16. Anthony van Dyck, St Martin (detail), c. 1618–20, panel, 171.6 ×
158 cm. Zaventem, Sint-Martinuskerk.

KATLIJNE VAN DER STIGHELEN, KOEN JANSSENS ET AL
YOUNG ANTHONY VAN DYCK REVISITED
33 ArtMatters, 6, 2014
Conclusion
Interdisciplinary research underscores the need for cau-
tion; it illuminates matters that cannot be determined
with the naked eye, even by leading experts. This does not
detract from the value of expertise, however, which should
ideally embrace provenance research and a thorough study
of the historical context. Nevertheless, the search for the
master’s hand will have to be completed in the near future
with the assistance of different disciplines and different sci-
entic traditions. The case of ‘Young Anthony’ illustrates
convincingly how complementary views may corroborate
(or weaken) an attribution. The technical analysis of the
different layers uncovers the genesis of the portrait and
corroborates every aspect of the attribution to Anthony
van Dyck.
Notes
1. H. Vlieghe, Rubens Portraits of Identied Sitters Painted
in Antwerp, Corpus Rubenianum Ludwig Burchard XIX.2,
London and New York, Har vey Miller Publisher s, 1987: 77–8,
cat. no. 89.
2. M. Jaffé, Van Dyck’s Antwerp Sketchbook, 1, London,
Macdonald, 1966: 105, note 89: ‘Photographs were shown to
the late Professor Dr. Otto Benesch in 1954, while it was in
the possession of Frederick W. Mont, New York. Dr Benesch
corrected the impression of Mr. Mont that his painting was
by Rubens of van Dyck, giv ing his opinion that it was a Self-
Portrait by the young van Dyck. I am grateful to Dr. Konrad
Oberhuber for bringing a note of dr. Benesch’s correspon-
dence to my attention at t he Alberti na; to Dr. Benesch himself
for showing me the photographs sent by Mr. Mont from New
York, and for telling me more of the c orrespondence.’
3. Jaf fé 1966: 47–8.
4. G. Martin, National Galle ry Catalogues: The Flemi sh School,
London, National Galler y, 1970: 33, note 2.
5. G. Eckhardt, Selbstbildnisse Niederländischer Maler des 17.
Jahrhunderts, Berlin, Henschelverlag, 1971: 70, 180, pl. 62.
6. A. McNairn, The Young van Dyck / Le jeune Van Dyck, exh.
cat. National Ga llery of Can ada/ Galerie Nationale du C anada,
1980: 36–7, cat. no. 2 (Self-Portrait about Age Fifteen).
7. Kimbell Art Mus eum: Catalogue of the Collec tion, Fort Worth,
TX, 1972, No. 64– 65; Kimbell Art Museum: Handbook of the
Collection, Fort Worth, 1981: 78.
8. E. Larsen, ‘New suggestions concerning George Jamesone’,
Gazet te des Beaux-Arts, 1979: 16, note 32.
9. Jaffé 1966: 48.
10. Jaffé 1966: 106, note 90.
11. McNairn 1980: 37; Vl ieghe 1987: 77.
12 . For an overview of r elevant literature on t his portrait a nd the
stylistic context of Van Dyck during his rst Antwerp period,
see A. Vergara and F. Lammertse (eds), The Young Anthony
Van Dyck, exh. cat. Museo Naciona l del Prado, Madrid, 2012:
94–5.
13 . Jaffé 1966: 47.
14. Jaffé 1966: 48: ‘Perhaps even more characteristic of this
perennial Narcissus in his adolescence is another painted
presentation of himself a year or so older.’
15. Vlieghe 1987: 77. Alan McNairn compares the Antwerp self-
portrait (then in Kimbell Art Museum, Fort Worth, Texas)
with the self-portrait of Van Dyck in Munich, to which he
assigns t he exceedingly early date of c. 1617/18. See McNairn
1980: 37 and g. 6.
ab
Figure 17. (a) Anthony van Dyck, Portrait of a Carmelite Monk (before restoration), c. 1618, panel, 62.3 × 48 cm. Private Collection. (b) Detail (before restoration).

KATLIJNE VAN DER STIGHELEN, KOEN JANSSENS ET AL
YOUNG ANTHONY VAN DYCK REVISITED
34 ArtMatters, 6, 2014
16. K. Van der Stighelen, ‘Young Anthony: archival discoveries
relating to van Dyck’s early career’, in S.J. Barnes and A.K.
Wheelock, Jr. (eds), Van Dyck 350, Studies in the History of
Art 46, Washington, Hanover and London, National Gallery
of Art, Washington DC, 1994: 24, 37.
17. Endorsed by S. Als teens, ‘A note on the young Van Dyck’, The
Burlington Magazine, CLVI (1331), 2014: 85–90.
18. Hans Vlieghe (‘Rubens’ beginnende invloed: Arnout
Vinckenborch en het probleem van Jordaens’ vroegste teke-
ningen’, Nederlands Kunsthistorisch Jaarboek, 38, 1987:
383, 394) suggests that Rubens may have coordinated the
cycle; K. Van der Stighelen and H. Vlieghe, ‘Cornelis de Vos
(1584/5–1651) als historie- en genreschilder’, Academiae
Analecta, 54, 1994, no. 1: 7–10.
19. Van der Stighelen and Vlieg he 1994: 7.
20. N. De Poorter, cat. nos. 35–36 in Rubens. Een genie aan het
werk, exh. cat. Brussels, KMSKB, Tielt, 2008: 138–41. Since
the couple were mar ried on 3 October 1617 and J. van Caest re
died in ch ildbirth, af ter giving bir th to her son on 18 July 1618,
the portrait can be condently dated to 1617/18. Although the
attri bution to Rubens regula rly resurfaces, a c ertain consen-
sus has formed in recent times endorsing the attribution to
Anthony van Dyck. Ties that existed between Jan van den
Broeck, who m ay have been his patron, and a per sonal friend
support th is hypothesi s; K. Van der St ighelen, ‘Van Dycks eer -
ste Antw erpse periode. Pr oloog van een barok levens verhaal’,
in C. Brown and H. Vlieghe (eds), Van Dyck, 1599–1641, exh.
cat., Ant werp, Royal Museum/L ondon, Royal Academy, 1999:
40, 46, note 52; De Poorter, op cit. p. 42 (I.25); see also: E.
Duverger, Antwerp se kunstinventar issen uit de zevent iende
eeuw, Fontes Historiae Artis Neerlandicae, I.6. 1649–1653,
Brussels, Koninklijke Vlaamse Academie van België, 1992:
119–27: ‘voor eene memor ie van zynen v rient ... zyne antic que
silvere med aliën van de Keyse rs wesende tsamen in get ale van
vierhondert e nde sesthien meda liën ... zynen goeden v rindt ...
oick syns te stateurs goeden vr iendt ... zyns testateu rs leesbo-
ecken, caertboecken ende alle andere gedrueckte boecken ...
die eenichssints verboden syn ... consent van heuren biecht-
vader om die te moghen gebr uycken’.
21 . Vlieg he 1987 (cited in note 1): 151–3, no. 134. A nother question
that arises concerns the relationship bet ween the portrait in
the Ufzi and the Rubens self-portrait that Michael Jaffé pub-
lished as an original and that was described by Christopher
Brown as a c opy af ter an original t hat had been lost. The por-
trait (Antwerp, Royal Museum) is currently on loan to the
Ru ben shu is in An twer p. See Br own and Vl ieg he 19 9 9: 94 , g.
1.
22. R. Baumstark, K.L . Belkin et al., Vorbild und Neuerndung.
Rubens im Wettstreit mit alten Meistern, exh. cat. Munich,
Bayerisc hen Staatsgemäldesa mmlungen, 2010: cat. nos. 6–7;
K. Renger and C. D enk, Flämische Malerei des Barock in der
Alten Pinakothek Munich, Colo gne, Dumont, 2002: 280– 82,
no. 341 (c. 1615–1618); Kristin Lohse Belkin dates Rubens’s
copy to c. 1620–1625.
23. Thus, it is striking that Willem Key depicts the eyelashes of
both eyes i n meticulous detai l. While Rubens almo st entirely
omits to do so, Van Dyck adopt s a similarly deta iled approach
here.
24. J.D. Stewart, ‘Pieter Thys (1624–77): recovering a “scarcely
known” A ntwerp painter’, Apollo, February, 1997: 37–43, g.
2. An impor tant element advanced by Douglas Stewart is his
reference to a painting depicting Icarus (‘an Icarus by Pieter
Thys’) in the inventory of Maria Verschueren-Pauweijns,
which he derives from M.L. Hairs. The inventory was
published more recently by E. Duverger, Antwerpse kunst-
inventarissen eeuw, Fontes Historiae Artis Neerlandicae,
I.10. 1674–1680, Brussels, Koninklijke Vlaamse Academie
van België , 1999: 85, ‘Item noch een stuck sch ilderije wesende
het subiect eenen Icarus’; ‘Item noch een stuck schilderije
wesende het subiect van Ulijsses ende geschildert benef fens
het voorgaende van de voors. Thijssens.’ It should be added
that the fact that Pieter Thys painted a composition of this
kind as a companion piece doe s not in itself prove anyt hing.
25 . S tewart 1997: 38.
26. McNair n 1980: IX, cat. no. 67 (pp. 144–6).
27. McNairn 1980: 144.
28. A. Tummers, The Eye of the Connoisseur: Authenticating
Paintings by Rembrandt and his Contemporaries,
Amsterdam, Amsterdam University P ress, 2011: 81–94.
29. McNairn 1980: 37. Unfortunately, the X-radiographs of
the portrait of the ‘Young Anthony’ have not been reprodu-
ced, and thus are not included in the documentation of the
Ruben ianum, nor are they present in the les of the K imbell
Art Museum, Fort Worth.
30. Performed on the painting using medical apparatus at the
Academic Hospita l ‘Gasthuisberg’, in Leuven.
31. P. Noble, A. van Loon, M. Alfeld, K. Janssens and J. Dik,
‘Rembrandt and/or Studio, Saul and David, c. 1655: visua-
lising the curtain using cross-section analyses and X-ray
uorescence imaging’, Technè, Special Issue Rembrandt,
Sc ien t i c Appro ac hes an d Res tora t io ns, 35 , 2012: 36–45.
32. K. Janssens, J. Dik, M. Cotte and J. Susini, ‘Photon-based
technique s for nondestructive sub surface ana lysis of painted
cultural heritage artifacts’, Accounts of Chemical Research,
43, 2010: 814–25.
33. The 3DMM technique was developed by Michael De Smet
(ESAT, KU Leuven), as part of his doctoral research ‘Gener ic
3D models for the parametr isation of the human face’.
34. For further information in regard to the application of 3D
techniques a nd parametri sation of the van Dyck por trait, see
http://homes.esat.kuleuven.be/~mproesma/mptmp/van-
dyck/vandyck01.htm l (acc essed November 2014).
35. In his self-portraits, Van Dyck looks to the right less often
than to the left. The three-quarter prole facing right cor-
responds to the heraldic position on the right of male sitters.
Nonetheless, Van Dyc k appears in thr ee-quar ters pose facing
left in the self-portraits in Munich, New York, St Petersburg
and Stra sbourg, and in the double por trait wit h Sir Endymion
Porter in Madr id.
36. McNair n 1980: IX.
37. N. De Poorter, in S.J. Bar nes, N. De Poorter, O. Millar and H.
Vey, Van Dyck. A Complete Catalogue of the Paintings, New
Haven and London, Yale University Press, 2004: cat. nos.
I.38–I.39, 53–6.
38. Sotheby’s London, 6 July 2011, lot 21 (as by Sir Anthony van
Dyc k).
Authors’ addresses
• Katlijne Van der Stighelen, KU Leuven, Department of Art
History, Leuven, Belgium (katlijne.vanderstighelen@arts.
kuleuven.be)
• Koen Janssens, University of Antwerp, Department of
Chemistry, Antwerp, Belgium (koenjanssens@uantwerp.be)
• Geert Van der Snickt, University of Antwerp, Depar tment
of Chemistr y, Antwerp, Belgium (geert.vandersnickt@
ua ntwerp.be)
• Matthias Alfeld, University of Antwerp, Department of
Chemistry, Antwerp, Belgium (matthias.alfeld@uantwerp.
be)

KATLIJNE VAN DER STIGHELEN, KOEN JANSSENS ET AL
YOUNG ANTHONY VAN DYCK REVISITED
35 ArtMatters, 6, 2014
• Ben Van Beneden, Rubenshuis, Antwerp, Belgium (ben.
vanbeneden@antwerpen.be)
• Bert Demarsin, KU Leuven, Leuven/ Campus Brussels,
Research Unit Economic Law, Leuven: Brussels, Belgium
(ber t.demarsin @law.kuleuven.be)
• Marc Proesmans, KU Leuven, Esat, Processing Speech &
Images, Leuven, Belgium (marc.proesmans@esat.kuleuven.
be)
• Guy Marchal, KU leuven, KU Leuven, UZ Leuven,
Department of Radiology, Leuven, Belgium (guy.marchal@
uzleuven.be)
• Joris Dik, TU Delft University of Technology, Delft, The
Netherlands ( j.dik@tudelf t.nl)