Dendrochronology is a discipline within the biological sciences that enables us to determine the age of wooden objects. At the very least, this method of dating allows us to arrive at a terminus post quem for a panel painting – that is to say, the earliest date at which the painting could have been executed – by establishing a plausible felling date for the tree from which the panel was cut.
Each concentric ring visible on the exposed surfaces of a felled tree corresponds to a year’s spring and summer growth. Dendrochronology involves measuring the width of these annual rings as they occur on a panel. A growth-ring curve is then created on the basis of these measurements, and this curve is compared with master chronologies derived from panels for which felling dates have already been established. A relatively precise dating, based on the specific characteristics of the growth-ring curve (considered in conjunction with the geographic origin of the wood itself),1 Growth-ring width varies according to local conditions, but for trees subject to similar geographic conditions growth-ring patterns will match – even for trees planted at different times. can thus be determined for a panel.2 For more detailed discussions of dendrochronology, see J. Bauch, D. Eckstein & G. Brauner, ‘Dendrochronologische Untersuchungen an Eichenholztafeln von Rubens-Gemälden’, Jahrbuch der Berliner Museen, vol. 20, 1978, pp. 209–21; M. G. L. Baillie, Tree-Ring Dating and Archaeology, Chicago, 1982; D. Eckstein, T. Wazny, J. Bauch & P. Klein, ‘New Evidence for the Dendrochronological Dating of Netherlandish Paintings’, Nature, no. 320, 1986, pp. 465–6; F. H. Schweingruber, Tree Rings Basics and Applications of Dendrochronology’, Dordrecht, 1988; P. Klein, ‘The Differentiation of Originals and Copies of Netherlandish Panel Paintings by Dendrochronology’, in Le Dessin sous-jacent dans la peinture: Colloque VIII, 1989, eds R. van Schoute & H. Verougstraete-Marcq, Louvain-la-Neuve, 1991, pp. 29–42; C. Lavier & G. Lambert, ‘Dendrochronology and Works of Art’, in Tree Rings: Environment and Humanity, eds I. S. Dean, J. M. Meko & T. W, Swetnam, Tucson, Arizona, 1996, pp. 543–56; P. Klein, ‘Dendrochronological Analyses of Panels of Hans Memling and His Contemporaries’, in Memling Studies, eds H. Verougstraete, R. van Schoute & M. Smeyers, Leuven, 1997, pp. 287–95; P. Klein, ‘Dendrochronological Analyses of Panel Paintings’, in Proceedings of a Symposium at the J. Paul Getty Museum, 24–28 April 1995: The Structural Conservation of Panel Paintings, eds K. Dardes & A. Rothe, Los Angeles, 1998, pp. 39–54.
This article presents the results of the dendrochronological testing of ten panel paintings at the National Gallery of Victoria.
Issues associated with dendrochronological analysis
In preparing oak panels for painting, panel makers working in northern Europe from the fifteenth to the seventeenth century usually cut planks with a radial orientation to the centre of the tree (fig. 1). The bark, and the light and perishable sapwood just beneath it, were then cut away. With the removal of the sapwood, evidence of the most recent growth rings was eliminated, making it impossible for dendrochronological analysis to determine the exact felling date of the tree from which a panel was cut: the latest date that can be definitively established is that of the most recent growth ring found on the panel.
The number of sapwood rings cut away can be estimated by statistical analysis, for which the provenance of the tree is significant. In Europe, the number of sapwood rings in oak trees varies: a range of from seven to fifty rings can be found in oaks from western Europe, a range of from nine to only thirty-six rings in trees from eastern Europe (fig. 2). This latter information is especially important, as the wood used for Netherlandish panels generally originated in the Baltic region. The metrical analysis of sapwood rings found in a sample of oaks from northern Poland yielded a median of fifteen: the minimum number of rings was nine, the maximum thirty-six; 50 per cent of the trees examined had thirteen to nineteen rings. For wood originating in Germany or the Netherlands, the median number of sapwood rings was seventeen, with 50 per cent of the trees having between thirteen and twenty-three rings; the minimum was seven.3 The analyses referred to here were conducted at the University of Hamburg (see Klein, ‘Dendrochronological Analyses of Panel Paintings’). The number of sapwood rings in wood from any given tree will also depend on the age of the tree: in general, a tree three hundred years old has more sapwood rings than a tree one hundred years old (see, for example, Eckstein et al., p. 465).
In the case of wood from the Baltic region, therefore, in order to determine the earliest possible felling date for an oak from which a panel has been cut, at least nine sapwood rings must be accounted for: nine years must be added to the dating of the latest heartwood ring found on the panel. If some of the sapwood has been preserved, the felling date of the oak can be estimated with greater accuracy than is possible when a panel is made exclusively from heartwood. It should also be borne in mind that some heartwood rings may have been cut off during the preparation of the panel.
When panels were made of beech or spruce rather than oak, the entire tree – except for the bark – was generally used. For beech and spruce panels, therefore, the last growth ring available for measurement on the panel corresponds in many cases to the last ring formed in the living tree prior to its felling.
The next problem is to determine how much time elapsed between the felling of a tree and the painter’s use of a panel from it. This so-called seasoning time, when planks were allowed to dry in storage prior to being cut into panels and painted, can be estimated on the basis of seasoning times calculated for dated panels by subtracting the estimated felling date (determined by dendrochronological analysis) from the year of a painting’s execution. For most oak panels of the sixteenth and seventeenth centuries, this seasoning time has been determined to be approximately two to eight years.4 See Bauch, Eckstein & Brauner. Since only a small number of investigations have been carried out with dated paintings from the fifteenth century, we cannot yet make such an accurate estimate for these earlier panels,5 See Klein, ‘Differentiation of Originals and Copies of Netherlandish Panel Paintings’, p. 30. although the studies that have been conducted to date (for examples, see tables 1 & 2) indicate an average seasoning time of ten to fifteen years. These findings correspond to the results of earlier analyses of fifteenth-century panel paintings by artists of the Cologne Schoo1.6 See J. Bauch, D. Eckstein & P. Klein, ‘Dendrochronologische Untersuchungen an Gemäldetafeln des Wallraf-Richartz- Museums Köln’, in Katalog der Altkölner Malerei, ed. F. G. Zehnder, Cologne, 1990, pp. 677–83.
Similar investigations with sixteenth-century beech wood resulted in estimates of a seasoning time of two to seven years, a finding that clearly corresponds to the estimates for oak wood from the same period.7 See P. Klein & J. Bauch, ‘Aufbau einer Jahrringchronologie für Buchenholz und ihre Anwendung für die Datierung von Gemälden’, Holzforschung, vol. 37, 1983, pp. 35–9.
Dendrochronology is particularly useful when the felling date for a tree, determined on the basis of an analysis of a panel’s growth-ring curve, proves to be later than the date of execution proposed for the painting on art historical grounds. When a felling date looks to be significantly earlier than a proposed date of execution, then the plank in question was either cut from a tree’s centre (recent growth rings would thus have been eliminated) or stored for a long time prior to being painted. A further possibility, of course, is that the suggested date of execution is too recent. Dendrochronology cannot determine which of these solutions applies in any given case.
Table 3
Results of dendrochronological analyses of ten panel paintings at
the National Gallery of Victoria
(image pending)
In the course of dendrochronological analysis, a number of additional problems may be encountered. In some years, for example, conifers (such as spruce or fir) and diffuse-porous broadleaved trees (such as lime or beech) may not produce a growth ring, thus preventing accurate dating.8 See Klein, ‘Dendrochronological Analyses of Panel Paintings’, pp. 39–40.Sometimes the state of conservation of a sample, as in cases where sapwood has been destroyed by insects, bacteria or fungi, does not permit a determination of the widths of growth rings. In some cases, not even the number of rings can be determined. It is also important to remember that, despite the availability of master chronologies for several regions and time frames, climatic conditions can vary significantly in response to local influences, making the use of general chronologies problematic.9 See note 1 above.
These problems notwithstanding, dendrochronological analysis can provide extremely valuable evidence with respect to the date of a panel and can still be helpful in the resolution of some art historical questions. However, dendrochronological data should always be considered in conjunction with data obtained by other methods.
Above all, it is important to be aware that it is always more useful to analyse a group of panels, rather than a single panel, from a particular workshop. To that end, since 1968 the dendrochronological department at the University of Hamburg has assembled more than two thousand analyses of panel paintings.
We now turn to the findings derived from dendrochronological analysis of ten panel paintings in the collection of the National Gallery of Victoria (table 3).
Panel paintings at the National Gallery of Victoria
Oak
From the fifteenth to the seventeenth century, artists from the northern regions of central Europe painted almost exclusively on oak panels. From the fifteenth century to the end of the sixteenth century, the oak on which these artists worked came from the Baltic region; in the seventeenth century, wood from the Netherlands and western Germany was also used.
In the case of the Melbourne Triptych with the Miracles of Christ, the centre panel of which is attributed to the Master of the Legend of St Catherine, each of the wings has been constructed from two planks, and the centre panel from four. The right-hand planks of both the left and the right wing are from the same tree; the left-hand planks of both wings were cut from another tree. In the centre panel, planks I and III, and planks II and IV, originate from the same trees. The most recent growth ring dates to the year 1466 (plank I, centre panel); the earliest felling date for the tree from which plank I was cut (if we assume the minimum of nine sapwood rings) must therefore be 1475. If we assume the median of fifteen sapwood rings (which gives us a felling date of 1481) and a seasoning time of ten years, it is plausible to suggest a painting date from 1491 onward for the centre panel.10 When a panel consists of more than one plank, as in the case of the centre panel of the Melbourne triptych, planks from the tree most recently felled must be used as the basis for estimates of a plausible painting date.
An examination of the Virgin and Child by Simon Marmion (c.1425–1489) reveals that the panel was at some stage thinned to the paint layer and then glued to a new board. Therefore, dating the painting by means of dendrochronological analysis was not possible (the new board was made in the nineteenth century at the earliest).
Sometimes dendrochronological analysis can be used to differentiate between an original and a copy or a fake. For example, dendrochronological investigations of attributions to Jan van Eyck have proved that the painting Man with Pinks (Gemäldegalerie, Berlin) must have been executed by a follower of van Eyck, rather than by the master himself. Similarly, the wood planks used for another panel, Christ Bearing the Cross (Szépmüvészeti Múzeum, Budapest), give evidence of an earliest felling date of 1500, suggesting that this painting was created at the beginning of the sixteenth century, again by a follower of van Eyck. A dendrochronological solution has not been possible, however, for the National Gallery of Victoria’s The Madonna and the Child (currently attributed to a follower of van Eyck). The 113 growth rings that could be measured on the single oak plank could not be correlated with growth-ring curves in the existing master chronologies, and thus could not be used to determine a felling date. Therefore, it is not possible to say, at least on the basis of dendrochronological data, whether the painting is an original or a later copy.
Three of the Gallery’s paintings from the seventeenth century underwent dendrochronological analysis some years ago: Rembrandt’s Two Old Men Disputing, Rubens’s Louis XIII of France, and Hercules and Antaeus, a painting that may or may not show the hand of Rubens.11 The Rembrandt and Rubens panels were studied in 1974 by Dr J. Bauch, and were further evaluated by the present author.
In the case of the Rembrandt painting – which carries the date 1628, although the inscription is no longer discernible – the most recent growth ring on the single plank can be dated to the year 1603. This dating implies an earliest possible felling date of 1612 (if we assume the minimum of nine sapwood rings) and a plausible date of execution from 1620 onward (if we assume the median of fifteen sapwood rings and a seasoning time of two years). This date is well in keeping with the date on the panel itself.
The panel that provides a secondary support for the Louis XIII, a work executed on paper, shows a youngest growth ring dating to 1615; the most recent growth ring for the Hercules and Antaeus dates to 1609. The wood used for the first painting originated in western Germany or the Netherlands and is of a type used for various paintings by Rubens; the wood used for the second panel is from the Baltic region.
Beech
In central Europe, other hardwoods – such as beech, lime and poplar – and conifers (softwoods) were also employed in the making of art objects. Following on from the studies of oak panels, panels of lime and beech, by early German painters, have also been studied; dendrochronological dating has been found to be successful with beech panels, though not with lime.
Beech panels were used (with some exceptions) only in the workshop of Lucas Cranach and only for a short time (1520–35). Analyses of Cranach’s signed and dated panels clearly indicate that there were only a few years between each panel’s youngest growth ring and the date of signing. Dendrochronological analysis reveals that the most recent growth ring on Melbourne’s Philipp Melanchthon is from the year 1510, a date that fits in with the dates for the latest growth rings on panels used for various other Melanchthon pictures by Cranach or his workshop.
Peter Klein, Ordinariat für Holzbiologie, Universität Hamburg (in 1999).
Notes
1 Growth-ring width varies according to local conditions, but for trees subject to similar geographic conditions growth-ring patterns will match – even for trees planted at different times.
2 For more detailed discussions of dendrochronology, see J. Bauch, D. Eckstein & G. Brauner, ‘Dendrochronologische Untersuchungen an Eichenholztafeln von Rubens-Gemälden’, Jahrbuch der Berliner Museen, vol. 20, 1978, pp. 209–21; M. G. L. Baillie, Tree-Ring Dating and Archaeology, Chicago, 1982; D. Eckstein, T. Wazny, J. Bauch & P. Klein, ‘New Evidence for the Dendrochronological Dating of Netherlandish Paintings’, Nature, no. 320, 1986, pp. 465–6; F. H. Schweingruber, Tree Rings Basics and Applications of Dendrochronology’, Dordrecht, 1988; P. Klein, ‘The Differentiation of Originals and Copies of Netherlandish Panel Paintings by Dendrochronology’, in Le Dessin sous-jacent dans la peinture: Colloque VIII, 1989, eds R. van Schoute & H. Verougstraete-Marcq, Louvain-la-Neuve, 1991, pp. 29–42; C. Lavier & G. Lambert, ‘Dendrochronology and Works of Art’, in Tree Rings: Environment and Humanity, eds I. S. Dean, J. M. Meko & T. W, Swetnam, Tucson, Arizona, 1996, pp. 543–56; P. Klein, ‘Dendrochronological Analyses of Panels of Hans Memling and His Contemporaries’, in Memling Studies, eds H. Verougstraete, R. van Schoute & M. Smeyers, Leuven, 1997, pp. 287–95; P. Klein, ‘Dendrochronological Analyses of Panel Paintings’, in Proceedings of a Symposium at the J. Paul Getty Museum, 24–28 April 1995: The Structural Conservation of Panel Paintings, eds K. Dardes & A. Rothe, Los Angeles, 1998, pp. 39–54.
3 The analyses referred to here were conducted at the University of Hamburg (see Klein, ‘Dendrochronological Analyses of Panel Paintings’). The number of sapwood rings in wood from any given tree will also depend on the age of the tree: in general, a tree three hundred years old has more sapwood rings than a tree one hundred years old (see, for example, Eckstein et al., p. 465).
4 See Bauch, Eckstein & Brauner.
5 See Klein, ‘Differentiation of Originals and Copies of Netherlandish Panel Paintings’, p. 30.
6 See J. Bauch, D. Eckstein & P. Klein, ‘Dendrochronologische Untersuchungen an Gemäldetafeln des Wallraf-Richartz- Museums Köln’, in Katalog der Altkölner Malerei, ed. F. G. Zehnder, Cologne, 1990, pp. 677–83.
7 See P. Klein & J. Bauch, ‘Aufbau einer Jahrringchronologie für Buchenholz und ihre Anwendung für die Datierung von Gemälden’, Holzforschung, vol. 37, 1983, pp. 35–9.
8 See Klein, ‘Dendrochronological Analyses of Panel Paintings’, pp. 39–40.
9 See note 1 above.
10 When a panel consists of more than one plank, as in the case of the centre panel of the Melbourne triptych, planks from the tree most recently felled must be used as the basis for estimates of a plausible painting date.
11 The Rembrandt and Rubens panels were studied in 1974 by Dr J. Bauch, and were further evaluated by the present author.
Notes to Table 3
The analyses of the Gallery’s Flemish pictures, and of the painting from the Cranach studio, were conducted by the author in November 1998. The Rembrandt and Rubens panels were studied in 1974 by Dr J. Bauch, and were subsequently further evaluated by the author.
Hoff = U. Hoff, European Paintings before 1800 in the National Gallery of Victoria, 4th edn, Melbourne, 1995.
Roman numerals refer to individual planks. Planks designated by the same superscripted numerals (1–5) are from the same tree.
a Data based on minimum of seven or nine sapwood rings.
b Data based on median of fifteen or seventeen sapwood rings.
c Data based on plausible felling date, plus seasoning time of two or ten years.
d Because the entire tree – with the exception of the bark – was generally used in the making of beech panels, estimates of sapwood rings are not relevant to the dating of these panels.
e In cases such as this, dating has not been possible, due to the fact that the growth-ring curve associated with the panel could not be correlated with the master chronologies.