TREE-RING DATING OF OAK PANELS FROM THE PAINTING “GRAY-BEARDED MAN HOLDING SHELL” Henri D. Grissino-Mayer Laboratory of Tree-Ring Science Department of Geography University of Tennessee Knoxville, Tennessee 37996 Henri D. Grissino-Mayer, Principal Investigator Final Report Submitted 7 July 2006 Tree-Ring Dating of Oak Panels from the Painting “Gray-Bearded Man Holding Shell” Table of Contents Executive Summary .................................................................................................................. 1 How Tree-Ring Dating Works ................................................................................................ 2 Assigning Calendar Years to Tree Rings by Crossdating...................................................... 3 Tree-Ring Dating of Panel Paintings: A Short History.......................................................... 4 Background to this Research Project....................................................................................... 9 Methods ...................................................................................................................................... 9 Surface Preparation .............................................................................................................. 9 Obtaining Tree-Ring Measurements...................................................................................... 10 The Baltic 1 Reference Chronology....................................................................................... 11 Graphical Crossdating .......................................................................................................... 12 Statistical Crossdating .......................................................................................................... 15 Results ........................................................................................................................................ 16 Original Position of the Panels.............................................................................................. 16 Graphical Crossdating .......................................................................................................... 17 Statistical Crossdating ........................................................................................................... 20 Conclusions ................................................................................................................................ 23 References Cited in this Study ................................................................................................. 25 List of Figures Figure 1. Sequence of tree rings from a red oak .................................................................. 30 Figure 2A. Crossdating a sequence of undated tree rings .................................................... 31 Figure 2B. The undated tree rings now dated ...................................................................... 31 Figure 3. A triptych by Rogier van der Weyden................................................................... 32 Figure 4. The painting that was analyzed ............................................................................ 33 Figure 5. Polishing the panel surfaces .................................................................................. 34 Figure 6. Securing the painting in a custom-designed brace ................................................ 35 Figure 7. Ensuring the vertical alignment of the painting..................................................... 36 Figure 8. Example of a 2 X 2 contingency table used in chi-square analysis ....................... 37 Figure 9-1. The tree rings seen on Panel 1............................................................................ 38 Figure 9-2. The tree rings seen on Panel 2............................................................................ 38 Figure 9-3. The tree rings seen on Panel 3............................................................................ 38 Figure 10. Locations in a log from where panels can be cut ............................................... 39 Figure 11. Skeleton plot comparing Panel 2 against the Baltic 1 chronology ...................... 40 Figure 12A. Line graph comparing Panel 2 against the Baltic 1 chronology....................... 41 Figure 12B. Line graph comparing Panel 2 against the Netherlands Baltic chronology ..... 41 Figure 13. Final dated positions for the three panels ............................................................ 42 List of Appendices Appendix 1: Complete listing of published articles that used tree-ring dating on panel paintings .................................................................................. 43 Appendix 2A: The “Baltic 1” reference tree-ring chronology supplied by the University of Sheffield .......................................................................... 50 Appendix 2B: The “Baltic 2” reference tree-ring chronology supplied by the University of Sheffield .......................................................................... 51 Appendix 3A: Original measurements for Panel 2 .............................................................. 52 Appendix 3B: Original measurements for Panel 1 .............................................................. 53 Appendix 3C: Original measurements for Panel 3 .............................................................. 54 Appendix 4A: COFECHA output that compares the tree rings from Panel 2 with the Baltic 1 reference chronology ..................................................... 55 Appendix 4B: COFECHA output that compares the tree rings from Panel 2 with the Baltic 1 reference chronology (25-year lag) ............................... 56 Appendix 5A: COFECHA output that compares the tree rings from Panel 1 with the Baltic 1 reference chronology ..................................................... 57 Appendix 5B: COFECHA output that compares the tree rings from Panel 1 with the Baltic 1 reference chronology (25-year lag) .............................. 58 Appendix 6A: COFECHA output that compares the tree rings from Panel 3 with the Baltic 1 reference chronology ..................................................... 59 Appendix 6B: COFECHA output that compares the tree rings from Panel 3 with Panel 1 ............................................................................................... 60 Tree-Ring Dating of a Panel Painting Page 1 EXECUTIVE SUMMARY In 2004, the Laboratory of Tree-Ring Science at the University of Tennessee was presented a panel painting which the owner attributed to the Dutch master Rembrandt van Rijn. Our goal was to use dendrochronological techniques to assign calendar dates to the tree rings on the three panels that make up the painting in an attempt to help experts assign a painter to this remarkable work of art. After polishing the panel surfaces, we measured the widths of all tree rings in the nearest 0.01 mm using a Velmex moving stage micrometer on which was mounted a specially-designed brace system for securing the painting. We used both graphical and statistical techniques to assign absolute calendar years to the tree rings (“crossdating”) on the three panels. We used primarily the “Baltic 1” reference tree-ring chronology developed at the University of Sheffield, but also found very convincing agreement with tree rings from the Netherlands Baltic reference chronology. Panel 2 had tree-ring patterns that were by far the most convincing both graphically and statistically. Tree rings on Panel 2 date from A.D. 1369 to 1585. The tree rings from Panel 1 crossdated from A.D. 1338 to 1543. Panel 3 crossdated convincingly with tree rings from Panel 1 and date from A.D. 1393 to 1589. Curiously, the tree rings from Panel 2 dated more strongly against the Netherlands Baltic reference chronology, suggesting the tree rings from this painting have more in common with works of art housed specifically in museums in the Netherlands. Using published estimates for sapwood rings and allowing for seasoning time, the year in which the tree(s) was harvested to make these panels could range between A.D. 1601 to 1613, but the actual harvest year could be much later. We can safely conclude that whoever painted this masterpiece must have lived and painted in the early to mid-1600s. The analyses performed in our laboratory support its possible attribution to Rembrandt van Rijn. Tree-Ring Dating of a Panel Painting Page 2 HOW TREE-RING DATING WORKS Trees are intimately bound to the environment, making them one of the most consistent and dependable recorders of processes and events that occur in nature. Trees are sensitive to both natural (precipitation and temperature patterns) and human-related (air and water pollution) events that trigger certain responses in the vigor of the tree as seen in its growth rate. In most geographic regions, climate patterns in any year cause a response by trees in the volume of wood the tree produces, and often leave indelible “fingerprints” in certain physical and chemical properties of the wood. These “fingerprints” can be seen in the varying widths of tree rings from year to year. In some years, climate conditions may be generally unfavorable for tree growth, causing a reduction in the volume of wood produced. On a cross section of wood, this reduced growth will appear as tree rings that are narrower than normal (Figure 1, ring A compared to the ring B). In other years, environmental conditions may be favorable for tree growth, allowing trees to produce greater volumes of wood. When viewed as a cross section through the trunk of the tree, the greater volume of wood will appear as wider than normal tree rings (Figure 1, ring C compared to ring B). These patterns of narrow and wide tree rings that accumulate over many years during the lifetime of the tree make tree-ring dating possible. A tree grows by forming a new sheath or layer of woody tissue each year just underneath the bark, much like a stack of cones one on top of the other. This growth occurs in a thin layer of cells that completely shrouds the tree just inside the bark called the vascular cambium. Here, cambial cells divide, with outer cells contributing to the formation of phloem, and inner cells contributing to the formation of woody tissue. In temperate and subpolar regions, most trees break from winter dormancy and begin growth by forming new cells of wood using nutrient reserves stored from the previous growing season. In conifer trees, these cells are large, less Tree-Ring Dating of a Panel Painting Page 3 dense, and have thin walls, producing light-colored wood called earlywood (Figure 1). Toward the end of the growing season, before the tree begins its dormancy period, smaller, denser, thickwalled cells are formed that are darker in color and called latewood (Figure 1). Taken together, these two bands of wood form an annual tree ring. In certain hardwoods, such as oaks, large cells called vessels are formed in the earlywood, while the latewood often lacks such vessels. ASSIGNING CALENDAR YEARS TO TREE RINGS BY CROSSDATING Crossdating is the primary guiding principle and technique in dendrochronology, and concerns the matching of patterns of ring widths from one tree with corresponding patterns for the same years from another tree. Eventually, sequences from individual trees are combined into a reference chronology for any particular location, so that any new tree-ring samples collected can immediately be crossdated against the reference chronology. Crossdating is possible because climate is largely a regional phenomenon, affecting trees within a geographic region in a like manner, so that similar patterns of ring widths are produced among many trees. A unique pattern of wide and narrow rings formed during a 50-year period is unlikely to be formed during any other 50-year period because climate varies from year to year. Furthermore, crossdating helps identify problematic rings, such as false rings (a dark band of cells within the ring that is not a true ring), and helps identify locally absent or discontinuous rings (produced when growth regulators do not reach certain points in the trunk of the tree). Crossdating is initially accomplished with graphical comparisons that accentuate the importance of patterns of wide and narrow rings. The pattern of rings is matched, or crossdated, against a reference chronology until a firm match is found (Figures 2A and 2B). The tree-ring widths are then measured, and the crossdating accuracy is statistically confirmed using computer Tree-Ring Dating of a Panel Painting Page 4 software. To ensure a high level of confidence in the dates assigned to tree rings, dendrochronologists look for high levels of statistical significance, higher than those normally used in statistics to identify a probable match. Only after a tree-ring series has been crossdated precisely, both graphically and statistically, can dates for the tree rings in a wood sample be assigned. TREE-RING DATING OF PANEL PAINTINGS: A SHORT HISTORY Tree-ring dating techniques have been used since the early years of the 20th century to determine the years of construction of historic sites throughout the United States and Europe. Andrew Ellicott Douglass, the founder of dendrochronology, first used tree-ring dating to help solve one of prehistoric archaeology’s most vexing questions: When where the pueblos of the American Southwest built? Prior to tree-ring dating, such historic structures were dated by ceramic styles but no firm technique had been developed that placed these structures in exact historical time. Douglass worked for over 15 years between 1914 and 1929 to solve this mystery and finally was able to determine the exact years of construction for 25 famous archaeological sites throughout the Southwest (Douglass 1921, 1929). Interestingly, most of the famous pueblo sites in the American Southwest were actually several hundred years younger than archaeologists had originally thought. It wasn’t long before scientists recognized the potential of tree-ring dating on other objects and structures made of wood. Several conditions are required for tree-ring dating to work successfully, however. First, the wood used to make the object must have distinct rings. Some species of trees do not form distinct rings while others form erratic ring patterns that can not be dated. Second, the object must have enough rings to provide a conclusive match with other Tree-Ring Dating of a Panel Painting Page 5 samples. In general, dendrochronologists like to have a strong statistical match between two chronologies that covers at least 70 to 80 years, although actual matches often cover many more years (greater than 100 or even 200 years). Shorter sequences of tree rings often are problematic to date. Third, a reference chronology must be available for the region from which the tree grew that was eventually crafted into an object or made into a structure. Without a reference chronology, the object can not be dated. In 1970, tree-ring scientists from the University of Hamburg first applied tree-ring dating techniques to verify the year of attribution for paintings on panels. Bauch and Eckstein (1970) analyzed the panels on 17th century paintings made by several Dutch masters: Philips Wouverman (1619–1668), Klaes Molenaer (1630–1676), and Cornelius Dusart (1660–1704). Bauch and Eckstein proved that tree-ring patterns matched significantly between different panels from the same painting as well as between different paintings, firmly establishing the possibility of developing reference chronologies to date other panel paintings. Second, Bauch and Eckstein were able to show that the dates for the outermost tree rings on these panels supported the attributed years the paintings were made. In other words, the outermost tree-ring dates had to pre-date the year the painting was supposedly created. This demonstrated the possibility that tree-ring dating could be used to help authenticate the year in which a painting was made, although tree-ring dating can not provide the actual year the painting was made. The technique only provides the year after which the painting could have been made. More importantly, this initial study showed how tree-ring techniques could potentially isolate fakes and copies of famous works, although none were found in this first study. Should the year for the outermost ring post-date the year the painting was made, then the painting is likely a fake or copy. Tree-Ring Dating of a Panel Painting Page 6 Bauch and Eckstein (1981) later analyzed 150 panels from paintings accepted or not accepted as authentic Rembrandts and Rembrandt associates, provided by the Rembrandt Research Group in Amsterdam. In all cases, the outermost ring on these panels pre-dated the attributed date for all paintings, even after the number of additional rings were estimated and placed on the outside of the outermost ring. This procedure of estimating additional rings on the outside is necessary because wood craftsmen would remove some or all of the sapwood from the oak used to make the panels when squaring the wood. Bauch and Eckstein cautioned, however, that tree-ring dating limit itself to the determination of the outermost (youngest) tree ring without any further interpretations, “which should be reserved to the art-historian.” Nearly simultaneous with the researchers from the University of Hamburg, a scientist at Oxford University in England began a systematic analysis of the tree-ring dates from panel paintings to again verify the reported attribution dates for famous paintings, but also wished to determine the source region of the wood used to make the panels. Fletcher (1973, 1976, 1977) developed some the longest tree-ring reference chronologies from archaeological art-historical objects ever, nearly all of which are in use to this day. Fletcher was able to show how the tree rings on panels in many paintings could actually be traced to a certain geographic region of Europe. This would greatly aid the crossdating process by creating a smaller subset of reference chronology to which the tree rings being tested would be crossdated. Approximately 80 studies have since been published that used tree-ring dating techniques to analyze the rings on panels from famous paintings (see Appendix1, which lists all 80 studies). These include the many studies by Dr. Peter Klein, also of the University of Hamburg, who elevated the science by analyzing the panels on many famous paintings, such as those made by Rogier van der Weyden (Klein 1981), Jean Fouquet (Klein 1984), Lucas Cranach the Elder Tree-Ring Dating of a Panel Painting Page 7 (Klein 1986), Raphael (Klein and Bauch 1990), Petrus Christus (Klein 1994a), Lucas Cranach (Klein 1994b), Hieronymus Bosch (Klein 1996a), and Tom Ring (Klein 1996b). Klein was instrumental in helping art historians revise their interpretations of famous works of art, such as the triptych by Rogier van der Weyden (The Nativity, The Pietà and Christ Appearing to His Mother) (Figure 3). Art historians had previously thought that the original was displayed in Capilla Real, Granada (left and middle panels) and the Metropolitan Museum of Art in New York (right panel), while a copy was displayed in the Staatliche Gemaldegalerie Berlin-Dahlem. (Klein 1986) showed instead, using tree-ring dating, that the Berlin-Dahlem has the early original painting while New York and Granada have a “posthumous copy” (Kuniholm 2000) In addition to the concentration of such studies in Germany and England, tree-ring research on panels has also been conducted in Italy by Professor Elio Corona (Corona 1984, 1994), in Denmark by Dr. Niels Bonde (Bonde et al. 1993), in France by Dr. Catherine Lavier and Dr. Georges Lambert (Lavier and Lambert 1996), in the Netherlands by Dr. Esther Jansma (Jansma et al. 2004), and in Belgium by Dr. Patrick Hoffsummer (Hoffsummer 1989). In the United States, Dr. Peter I. Kuniholm from Cornell University, has been involved in the tree-ring dating of panels from paintings (Kuniholm 2000). Key among these latter studies (and very relevant to this study) is that study conducted by Dr. Jennifer Hillam and Dr. Ian Tyers of Sheffield University and the Museum of London, respectively (Hillam and Tyers 1995). These researchers first conducted independent crossdating analyses on 177 sets of tree-ring measurements first developed by Dr. John Fletcher at Oxford University obtained from panel paintings to test the veracity of the original datings. Hillam and Tyers concluded that their analyses revealed “a high level of agreement between the [two] laboratories,” suggesting correct temporal placements of the 177 series. However, the authors Tree-Ring Dating of a Panel Painting Page 8 also concluded that one of the original temporal placements obtained by Fletcher was incorrect. The painting Virgin and Child by Jan Mabuse (ca. 1478–1532) that Fletcher had classified a fake was in fact probably genuine. In addition, 49 panels that Fletcher had dated previously could not be dated by either of the two laboratories. This conclusion concerning the veracity of some of Fletcher’s earlier datings can be attributed to new technologies used in tree-ring dating in the 1980s and 1990s, particularly computer-assisted statistical crossdating (see Holmes 1983 and Grissino-Mayer 2001) which revolutionized tree-ring dating. Perhaps the most important outcome of this study was the recognition of two major source regions for the oak timbers used in these 177 panels. Based on the strength of the crossdating among the tree-ring series, Hillam and Tyers were able to develop two reference chronologies from this larger data set: “Baltic 1” consists of tree-ring sequences from 67 panels and spans AD 1156 to 1597, while “Baltic 2” consists of tree-ring sequences from 40 panels and spans AD 1257 to 1615. The differences between the two chronologies could be attributed to changes in source regions for the oak timbers, which had to be imported from the region surrounding the eastern Baltic Sea region. This change in the source region could reflect fluctuations in exports from different ports (Hillam and Tyers 1995) caused by changing numbers of trees available for harvesting. Our study will rely extensively on the use of the “Baltic 1” reference chronology, which was kindly supplied by Dr. Ian Tyers, now with the University of Sheffield. Oak trees were often imported from this region to supply wood for art objects and artists working in western Europe (Baillie 1984; Wazny 1992, 2002; Klein 1998; Jansma et al. 2004), while wood from Germany was used mostly to make furniture (Jansma et al. 2004). The “Baltic 1” reference chronology consists of 98 dated series and spans from A.D. 1167 to 1597 but was recently extended by Tree-Ring Dating of a Panel Painting Page 9 Dutch researchers out to A.D. 1637 (Jansma et al. 2004). Artists working in the Netherlands in this time frame likely would have used wood imported from the Baltic region. BACKGROUND TO THIS RESEARCH PROJECT In 2004, the Laboratory of Tree-Ring Science at the University of Tennessee was contacted by an individual to evaluate the potential of our using tree-ring dating techniques on a panel painting that was believed by the owner to have to have been painted by Rembrandt van Rijn (1606–1669) (Figure 4). We had not previously been involved in the dating of panel paintings, but we had considerable experience dating the tree rings on wood from musical instruments and on logs from historical structures. Our primary objective was to determine the exact range of years for the tree rings found in the panels of this painting. The process consisted of first measuring the widths of all tree rings found in the panels, then comparing the pattern of wide and narrow rings against a reference tree-ring chronology created from many tree-ring sequences found in other panels on paintings made by well-known painters between the 15th and 17th centuries. The match between the panels being tested from this painting (Figure 4) and the reference tree-ring chronology had to be convincing both graphically and statistically. METHODS Surface Preparation Both the upper and lower surfaces of each of the three panels were surfaced using progressively finer polishing film to protect the delicate surfaces of the panels (Orvis and Grissino-Mayer 2002). Because the painting had to be examined intact, the sanding of the panels was done solely by hand-sanding (Figure 5). Normally, the sanding process results in a wood Tree-Ring Dating of a Panel Painting Page 10 surface with cellular features clearly defined under 10x magnification for clear ring identification. We noticed, however, that only the middle panel of the three panels had rings that could be clearly defined along the entire surface from one end to the other. Other techniques, such as planing the panel surfaces flat followed by calcium carbonate pressure infusion, were considered too invasive and potentially damaging, and were not used in this project. Once sanded, we used a jeweler’s loop to count all rings and marked with a lead pencil each decade ring according to established convention: one dot for each decade ring (e.g., rings 10, 20, and 30), two dots for each 50th ring (e.g., rings 50 and 150), and three dots for each 100th ring (e.g., rings 100 and 200) (Stokes and Smiley 1996). Special care was given to ray features common in oak wood (and other ring-porous hardwood tree species) that often can divide an individual ring in two and appear to offset on either side of the ray (Baillie 1982), which would result in inaccurate ring counts and errors in assigning absolute calendar years to all tree rings. Obtaining Tree-Ring Measurements We then measured the widths of all tree rings on all panels surfaces to 0.01 mm accuracy with a Velmex measuring stage coupled with MEASURE J2X software running on a Pentium personal computer. A special adaptation had to be added to the measuring stage to ensure measurement of the transverse surface of the rings. We constructed a vise-like device that would firmly hold the entire painting upright from the bottom (Figure 6). A stereozoom boom-arm microscope was then placed over the painting, on which was attached a color CCD camera through the ocular of the microscope (Figure 7). This arrangement allowed measurements to be captured without having to peer through the microscope at an awkward height. The image of the Tree-Ring Dating of a Panel Painting Page 11 panel surface with the tree rings was then projected onto a 21 inch color monitor. The widths of the tree rings were measured by turning the movable stage (on which the newly-devised apparatus firmly held the painting) from the beginning of one tree ring to the beginning of the next tree ring. Because the Velmex stage sat on the floor, one person slowly turned the stage (to avoid unnecessary movement of the painting) while another person instructed when to start and stop the movement of the stage. Because the dates of the tree rings are unknown, the ring series are considered “floating” in time (Bannister 1962) and therefore we assigned the year “1” to the first complete tree ring in each series and measured subsequent rings sequentially. To ensure all tree rings were identified, we measured the widths of the tree rings on the panels twice. All tree-ring measurements were captured and archived in internationally-accepted Decadal format as recognized by the International Tree-Ring Data Bank, part of the National Climatic Data Center and World Data Center for Paleoclimatology in Asheville, North Carolina. The accuracy of our measurements (0.01 mm) would be considered exceptional by art history standards because many previous studies simply used a hand lens or 10X eye piece to inspect the widths of tree rings on famous panel paintings and other art historical objects. Some studies assigned an “index” of ring width to individual rings (Jansma et al. 2004) while other studies measured the ring width to the nearest 0.1 mm (Fletcher 1977) rather than obtaining more accurate absolute measurements as we accomplished in our study. The Baltic 1 Reference Chronology Absolute dating was accomplished by comparing the pattern of wide and narrow rings from our floating tree-ring chronology obtained from the panels with an anchored reference tree- Tree-Ring Dating of a Panel Painting Page 12 ring chronology developed by members of the University of Sheffield Dendrochronology Laboratory (see http://www.shef.ac.uk/archaeology/research/dendrochronology). On April 21, 2006, we contacted Dr. Ian Tyers of this laboratory and inquired about the availability of the “Baltic 1” and “Baltic 2” reference chronologies that were developed in 1995 (Hillam and Tyers 1995). Dr. Tyers was very enthusiastic about our using these data sets and immediately sent both reference chronologies to the Laboratory of Tree-Ring Science at the University of Tennessee. These are included in Appendix 2A (“Baltic 1”) and Appendix 2B (“Baltic 2”). The reference chronologies had recently been used in a monumental study conducted by the Netherlands Centre for Dendrochronology, a laboratory of the Netherlands’ National Service for Archaeological Heritage Management, stationed in Amersfoort, the Netherlands (Jansma et al. 2004). The Dutch researchers analyzed numerous objects currently housed in museums in the Netherlands (including the Rijksmuseum in Amsterdam), including panel paintings, sculptures, cabinets, chests, desks, and tables, among other items, which yielded 159 individual dated series. Graphical Crossdating The comparison of ring patterns between two tree-ring data sets is known as crossdating and is accomplished using both graphical and statistical crossdating techniques (Swetnam et al. 1985; Stokes and Smiley 1996; Grissino-Mayer 2001). Initial crossdating via graphical techniques is first attempted by creating a “skeleton plot” of the ring patterns from the series being dated. Such skeleton plots only emphasize the narrow rings in the series because the narrow tree rings are more diagnostic than are the wider rings, which often exhibit considerable lack of variability in their widths compared to narrow rings. A second skeleton plot is next created for the anchored reference tree-ring chronology, over which the plot for the series being Tree-Ring Dating of a Panel Painting Page 13 dated is placed. Much like fingerprint matching, the skeleton plot of the series being dated is then slid, one year at a time, down the skeleton plot created from the reference chronology until the majority of lines (indicating narrow rings) on the two plots line up to ensure correct temporal placement. The alignment of the skeleton plots was tested statistically used a non-parametric technique called a chi-square analysis (Burt and Barber 1996), used by Grissino-Mayer et al. (2004) to help date the famous “Messiah” violin. This analysis uses a 2 X 2 contingency table based on matches between wide and narrow rings between the two skeleton plots. The table consists of four cells labeled A, B, C, and D (Figure 8). “Test series” is the tree-ring series that is being dated (in this case, from the panels of the painting), while “Reference series” is the reference chronology (here, the “Baltic 1” chronology). Rings are evaluated as being wide or narrow for both series. If the two series are aligned correctly, then the wide and narrow rings should match up between the two, in the A (wide and wide on both = match) and D (narrow and narrow on both = match) cells. Mismatched rings will occupy cells B (narrow versus wide = mismatch) and C (wide versus narrow = mismatch). The “actual” number of matches or mismatches is placed in each cell, along with the “expected” number in each cell, from which the chi-square statistic is generated: X2 (fij – Fij)2 = ∑ ∑ –––––––– i=1 j=1 Fij r c where fij is the actual number of matches or mismatches in cell “ij” (row i, column j) and Fij is the expected number of matches or mismatches in cell “ij”. The calculated X 2 statistic is then compared to the critical X 2 statistic based on (r – 1)(c – 1) degrees of freedom, where r is the Tree-Ring Dating of a Panel Painting Page 14 number of rows and c is the number of columns (two each in a 2 X 2 table, or 1 degree of freedom). If the calculated X 2 value is greater than the critical value, then we reject the null hypothesis that states that no relationship exists between the two series (i.e., the two series are independent and do not date against each other at this temporal placement). We then accept the alternate hypothesis that the two series are indeed “dependent” on each other and that a significant relationship exists between the two series at this dated position. Our chi-square analysis was performed in Microsoft Excel (function “CHITEST”) and verified using the online chi-square calculator at the University of Illinois Urbana–Champaign’s Statistical Computing Instruction Laboratory (see http://www.stat.uiuc.edu/courses/stat100/java/chisquare/ChiSquareApplet.html). A second graphical technique we used in this project uses common X-Y line plots, created in Microsoft Excel, in which the series being tested is again slid down the line plot of the anchored series one year at a time until the two plots line up. At each position, a correlation coefficient (r-value) is calculated that shows the strength of the agreement between the two series (the panels being tested and the reference anchored tree-ring chronology). Correlation coefficients are easily calculated in Excel using the “CORREL” function which returns r-values between –1 and +1. A value near –1 indicates an inverse relationship between two series; in other words, as the value for any one year goes up on one series, the value for that same year goes down on the other series. A positive r-value indicates that as the value for any one year for one series goes up (or down), the value for the other tree-ring series also goes up (or down) in a similar manner. Tree-ring dating is only concerned with positive r-values. The r-value is calculated as: Tree-Ring Dating of a Panel Painting Page 15 where n is the number of years in common between the two tree-ring series, x is a value for the first series in any given year, and y is the value for the second tree-ring series in that same year. In this study, we compared the line graph created from the ring measurements of Panel 2 with the line graph created from the Baltic 1 reference chronology made available by the University of Sheffield Dendrochronology Laboratory. In addition, we compared the line graph from Panel 2 with a line graph created from the Netherlands Baltic reference chronology made available by the Netherlands Centre for Dendrochronology and listed in Jansma et al. (2004). Correlation coefficients were computed between Panel 2 and the Baltic 1 and Netherlands Baltic reference chronologies. If crossdating could be achieved solely for Panel 2, then the chances of dating Panels 1 and 3 would improve. Statistical Crossdating We confirmed the initial graphical crossdating and relative placements of the tree-ring series being tested using COFECHA, a quality-control program that uses segmented time-series correlation techniques to confirm the temporal placements of all tree rings (Holmes 1983). Because crossdating is essentially a “high-frequency” process (pattern matching of sequences of individual rings), COFECHA removes all low-frequency trends using both spline-fitting algorithms and autoregressive modeling (Cook and Peters 1981, Cook 1985, Grissino-Mayer 2001). Such trends could also arise due to natural and human disturbances that otherwise could mask the (common) climate signal desirable for accurate crossdating. COFECHA then tests Tree-Ring Dating of a Panel Painting Page 16 consecutive 50-year segments from each series with a temporary master chronology created from all other measurement series. In this analysis, we tested 50-year segments with a 10-year lag to refine our crossdating and help identify problem area in the ring sequences. In Other words, we first tested rings 1–50, then rings 11 through 60, then rings 21 through 70, until the end of the series was reached, comparing these segments with an average from the same segments created from the remaining series. Crossdating is verified when the correlation coefficient for each tested segment in each series being tested exceeds 0.37 (p < 0.01), although coefficients for segments from undated series are usually higher. The final placement made by COFECHA had to be convincing graphically (similar patterns in wide and narrow rings) and statistically (highly significant correlation coefficients). Once confirmed, we assigned absolute years to all individual rings in each measurement series being tested. RESULTS Panel 1 (left panel, facing the painting) was found to measure 8 1/16 x 35 in (20.5 x 88.9 cm) and contained 206 rings. Panel 2 (middle panel) measured 9 9/16 x 35 in (24.3 x 88.9 cm) and contained 217 rings. Panel 3 (right panel) measured 10 3/8 x 35 in (26.4 x 88.9 cm) and contained 197 rings. Each panel was approximately (0.25 in) 6 mm thick. All measurements from the three panels are provided in Appendices 3A (Panel 2), 3B (Panel 1), and 3C (panel 3). Original Position of the Panels The oak rings on the panels could easily be discerned once the surfaces were polished to a high degree (Figures 9-1, 9-2, and 9-3). Panel 1 (towards the left facing the painting) showed significant curvature in the innermost rings of the surface (Figure 9-1), which always indicates a Tree-Ring Dating of a Panel Painting Page 17 panel that was cut to one side of the center of the log, called a “tangential cut” (see Figure 10 b). Because of this curvature, special care had to be taken when measuring because the ray cells that emanate from the center of the oak tree can cause rings to become offset on either side of the ray. Panel 2 (the middle panel) had by far the clearest sequences of tree rings of the three panels. More importantly, the rings on this panel were nearly all perpendicular to the cut of the wood and all ray cells were aligned parallel to the panel, when visible. This can be clearly seen in Figure 9-2. Hence, the ray cells were not problematic in this panel. The alignment of rings and cellular features strongly suggests that this panel was cut nearly centered towards the pith (center) of the log, called a “radial cut” (see Figure 10a). Panel 3 (towards the right as one faces the painting) displayed features that suggested an intermediate location inside the log between panels 1 and 3. Nearly all the tree rings are tilted at an angle to the panel (Figure 9-3), but do not display the extreme curvature seen in Panel 1. In addition, the ray cells also cut across the surface of the panel at an angle causing many rings to become offset (see red circle areas in Figure 9-3). In general, based on the lack of curvature in the rings, Panel 2 is the highest quality panel for dating purposes for this project. Panel 1 had a good chance of being dated accurately, while the tree rings from Panel 3 posed the most problems for dating. Graphical Crossdating Because Panel 2 had the clearest ring sequences that we felt were the most reliable, we created a skeleton plot from the measurements and slid this plot down a skeleton plot created for the Baltic 1 master chronology one year at a time until a clear match occurred. We found a very good match that clearly showed ring 1 on Panel 2 as being the year AD 1369 while ring 217 on Tree-Ring Dating of a Panel Painting Page 18 Panel 2 was the year AD 1585 (Figure 11). Because trees are individualistic, based on their immediate surroundings, scientists rarely expect tree-ring sequences to be perfectly aligned. We look for matches in the narrowest of rings (for example, rings 35–36, 51–52, 50, 55, 95, 102, and 111 on the upper series from Panel 2, Figure 11) as well as matches in clusters of rings (for example, rings 124–131, 198–203, and 213-216 on the upper series from Panel 2, Figure 11). This was by far the best match between the two series. The chi-square analysis also substantiated the temporal placement of AD 1369–1585 for Panel 2 as being the most likely dated position. The analysis of the skeleton plots revealed 76 wide-wide matches and 68 narrow-narrow matches (total 144 matches) versus only 72 mismatches (28 narrow-wide and 44 wide-narrow). Expected values in each cell (indicating randomness, or no association between the two series) were: Cell A: 58, Cell B: 46, Cell C: 62, and Cell D: 50. The chi-square statistic (X 2) was 24.335 which had a probability (p-value) of less than 0.005, meaning that a X 2 of this value has less than a 5 in 10,000 chance (1 in 2,000) of occurring simply by chance. In other words, we have a confidence level greater than 0.995 (1 minus 0.005), or > 99.5% confidence, that we have found the correct temporal position for the tree rings in Panel 2, based on the chi-square analysis. The suggested temporal position of AD 1369 to 1585 was again confirmed by the alignment of the line graphs for Panel 2 versus the Baltic 1 reference chronology (Figure 12A) and versus the Netherlands Baltic reference chronology (Figure 12B). These graphs showed coherency in both the year to year variation (high frequency variation) and decadal-scale variation (low frequency variation). The common patterns in the low-frequency variation is very convincing. For example, in both the measurements for Panel 2 and the Baltic 1 reference chronology (Figure 12A), the values decline steadily between A.D. 1375 and 1425, then steadily Tree-Ring Dating of a Panel Painting Page 19 increase between 1425 and 1440. Between 1440 and 1490, both series again decline steadily followed by a gradual increase in values until A.D. 1515. Both series then decrease again until about 1530, after which the series appear to diverge. Between A.D. 1560 and 1585, the series show little similarity in the low-frequency, likely an artifact of low sample size in the Baltic 1 reference chronology. This would explain the very low values seen after A.D. 1560 in the Baltic 1 reference chronology. Surprisingly, the measurements from Panel 2 show better agreement with the Netherlands Baltic tree-ring chronology (Figure 12B). The Netherlands data set was created independently from art historical objects and furniture not analyzed originally by Dr. John Fletcher at Oxford University, nor analyzed by Dr. Jennifer Hillam and Dr. Ian Tyers of Sheffield University. In other words, the Netherlands Baltic reference chronology may consist of series of tree-ring measurements that have much more in common with the tree rings from the oak panels of the painting being investigated in this study. The correlation coefficients calculated between Panel 2 and both the Baltic 1 and Netherlands Baltic reference chronologies further substantiate the temporal placement of A.D. 1369 to 1585 for Panel 2. An r-value of 0.25 (n = 217 years overlap, t-value = 3.79, p < 0.0001) was found in the comparison with the Baltic 1 reference chronology, while a higher r-value of 0.38 (n = 217 years overlap, t-value = 6.02, p < 0.0001) was found when comparing with the Netherlands Baltic chronology. These r-values are highly significant with a probability of less than 1 in 10,000 (p < 0.0001) that these r-values occurred simply by chance. Tree-Ring Dating of a Panel Painting Page 20 Statistical Crossdating We input the measurement series from Panel 2 (found in Appendix 3A) into the computer program COFECHA as an undated series, and compared it with the dated Baltic 1 reference chronology. The results from this crossdating attempt would be considered definitive by dendrochronological standards. COFECHA suggested that 12 of the 18 50-year segments tested from Panel 2 be adjusted by +1368 (see details in Appendix 4A); in other words, the value 1368 had to be added to all the rings in Panel 2 (with rings numbered 1 through 217) and it would then be dated against the Baltic 1 reference chronology. This suggests that the rings for panel 2 date to 1369 to 1585, similar to the results obtained in the skeleton plot, line graph, chi-square, and correlation analyses. The average r-value for the 12 segments was 0.44 at the +1368 adjustment, which was significantly better than the second best dating adjustment recommended by COFECHA: 5 segments with a dating adjustment of +1107, and an average r-value of 0.31. Significant in this analysis is that 7 of the 12 segments that showed this one dating adjustment dated at the number 1 position, while two other segments dated at the number 2 position (see Appendix 4A). To test the efficacy of this dating adjustment, we re-ran COFECHA but this time selected to test 50-year segments with a 25-year lag rather than a 10-year lag. This is a more rigorous analysis because fewer segments are now tested. We found that five of the eight 50-year segments tested again dated at the +1368 dating adjustment (Appendix 4B), similar to the previous analysis. Three of the segments with the +1368 adjustment dated at the number 1 position while another dated at the number 2 position. This additional analysis confirms that the tree rings in Panel 2 date to A.D. 1369 to 1585. Tree-Ring Dating of a Panel Painting Page 21 We next input the ring measurements from Panel 1 (found in Appendix 3B) in COFECHA as an undated series and tested 50-year long segments (lagged 10 years) against the Baltic 1 reference chronology. We found 12 of the 17 50-year segments tested all showed a common dating adjustment of +1337 (Appendix 5A). The strength of this crossdating is further shown by the five consecutive segments for rings 91 through 180 that dated in the number 1 position, with another 2 segments dating in the number 2 position. These 12 segments had an average r-value of 0.39 which was higher than the average r-value (0.35) for the second best dating adjustment of +1396 based on 7 segments. These results show that the tree rings on Panel 1 date from A.D. 1338 to 1543. The earlier innermost ring on this panel relative to Panel 2 (A.D. 1369) is substantiated by the curvature of the rings on this panel which indicate the panel was cut close to the pith of the log. We again ran COFECHA but this time we tested 50-year segments lagged 25 years rather than 10 years, and found four of seven 50-year segments dated at the +1337 adjustment (Appendix 5B). We next input the measurements from Panel 3 into COFECHA as an undated series and compared it to the Baltic 1 reference chronology. Panel 3 was our most problematic panel to measure because of the propensity of offset rings caused by ray cells that were angular to the panel surface (see Figure 9-3). Although we feel that we accounted for all the rings despite the offsets, the widths of our measurements may not be as accurate as could be because of the angle of the rings near the offsets caused by the ray cells. The top three dating adjustments suggested by COFECHA were: (1) +1205, 7 segments, average r-value = 0.35, (2) +1223, 6 segments, average r-value = 0.40, and (3) +1392, 6 segments, average r-value = 0.45 (Appendix 6A). The +1205 dating adjustment is not convincing because it is not found in the number one suggested position, and only shows up once in the number two position. The +1223 adjustment shows up Tree-Ring Dating of a Panel Painting Page 22 once in the number one position and 4 times in the number 2 position, all consecutive, which is only somewhat convincing. Of these, the third dating adjustment is the most convincing because of the very high average r-value of 0.45 based on six segments tested out of 16 total segments. In addition, three segments dated at the number one position, and another at the number 2 position, all consecutive. These findings suggest that this panel may have tree rings that date from A.D. 1393 to 1589, but one additional crossdating attempt was warranted. We then tested the tree rings of this panel against the tree-ring patterns on the other two dated panels, again using COFECHA. This is a common strategy when attempting to date floating tree-ring series. If the reference tree-ring chronology yields inconclusive results (which occasionally happens), then dendrochronologists try dating the floating series against individual dated measurement series. The assumption here is that the individual dated series will likely represent a tree from the same vicinity as the tree for the floating series. In this analysis, we input the dated measurements from Panel 1 and then Panel 2 into COFECHA as the dated series. The most significant result for Panel 3 was found in its comparison with the tree rings from Panel 1 (Appendix 6B). In fact, this result can be considered conclusive. We found that 11 of the 16 50year segments tested showed a similar dating adjustment of +1392 with an average r-value of 0.45. More importantly, seven of these segments dated at the number 1 position in consecutive order, representing rings 21 through 130, indicating a strong match. The second best dating adjustment was +1320, based on six segments, with an average r-value of only 0.26. These results strongly show that the tree rings for Panel 3 date from A.D. 1393 to 1589. Tree-Ring Dating of a Panel Painting Page 23 CONCLUSIONS This project was conducted on oak panels from a painting possibly made by a Dutch master, although the tree-ring aspect of this project certainly can not tell who exactly the painter was. This situation, however, is common for virtually all objects dated by tree-ring analyses – dendrochronologists can date the tree rings and provide precise dates for the range of years represented in a wooden object, but dendrochronologists can never prove who the painter or creator of the wooden object was. Still, dendrochronology and archaeology together can help provide information useful to experts who wish to assign a painter or maker to a wooden object. For example, when was the tree (or trees) cut that was used to make the panels used in the painting? First, we have an outermost date on any of the three panels of A.D. 1589 (Panel 3). Second, no sapwood was present on any of the three panels, as is expected because sapwood is less dense and decays more rapidly than the denser heartwood found on oak. We can assume for the time being that the outermost ring was near the sapwood/heartwood boundary on Panel 3 (see Figure 10d). Klein (1993) estimated that 15 years +4/–2 years should be added to the outermost ring found on Baltic oaks used in wooden objects. This would place the outermost ring on the living tree when it was harvested at A.D. 1601 (range A.D. 1599 to 1605). Finally, the estimate must again be modified to account for seasoning of the wood. Freshly cut trees contain a large amount of moisture in their wood, and paint does not adhere to wet wood (Kuniholm 2000). Therefore, the wood had to be left to air-dry (season) for a number of years before it could be used as a backing for a painting. Bauch and Eckstein (1970) observed that the difference between dated paintings and their outermost rings were 5 years ±3 years. Using the previous sapwood estimates, this would place the outermost rings at A.D. 1604 to 1610, with a range from A.D. 1601 to 1613. This range of years represents the earliest years of Tree-Ring Dating of a Panel Painting Page 24 tree harvest that may be represented on the three panels based on the outermost ring on Panel 3. However, because we observed no sapwood/heartwood contact zone (see Figure 10a), we can also assume that some heartwood rings may have been trimmed off during the shaping process (Kuniholm 2000), pushing the date for tree harvest even later. No means exists, though, to estimate the number of heartwood rings that may have been trimmed off. Whoever painted this masterpiece lived and painted in the early to mid-1600s. The analyses performed in our laboratory support its possible attribution to Rembrandt van Rijn. Tree-Ring Dating of a Panel Painting Page 25 REFERENCES CITED IN THIS STUDY Baillie, M.G.L. 1982. Tree-Ring Dating and Archaeology. University of Chicago Press, Chicago. 274 pp. Baillie, M.G.L. 1984. Some thoughts on art-historical dendrochronology. Journal of Archaeological Science 11: 371–393. Bauch, J., and D. Eckstein. 1970. Dendrochronological dating of oak panels of Dutch seventeenth-century paintings. Studies in Conservation 15: 45–50. Bauch, J., and D. Eckstein. 1981. Wood biological investigations on panels of Rembrandt paintings. Wood Science and Technology 15: 251–263. Bannister, B. 1962. The interpretation of tree-ring dates. American Antiquity 27: 508–514. Bonde, N., T. Bartholin, K. Christensen, A. Daly, and O.H. Eriksen. 1993. Dendrokronologiske dateringsundersogelser pa Nationalmuseet 1992. Arkæologiske udgravninger i Danmark: 305–321. Burt, J.E. and G.M. Barber. 1996. Elementary Statistics for Geographers, 2nd edition. Guilford Press, New York. 640 pp. Cook, E.R. 1985. A time series analysis approach to tree ring standardization. Ph.D. dissertation, University of Arizona, Tucson, Tucson, AZ. 171 pp. Cook, E.R., and K. Peters. 1981. The smoothing spline: a new approach to standardizing forest interior tree-ring width series for dendroclimatic studies. Tree-Ring Bulletin 41: 45–53. Corona, E. 1984. Il supporto ligneo della Pala di Monteluce. Bollettino dei musei e gallerie pontificie 5: 137–151. Corona, E. 1994. Sulla caratterizzazione dei manufatti lignei: opere pittoriche. Italia Forestale e Montana 49(4): 401–412. Tree-Ring Dating of a Panel Painting Page 26 Douglass, A.E. 1921. Dating our prehistoric ruins: how growth rings in trees aid in establishing the relative ages of the ruined pueblos of the Southwest. Natural History 21(1): 27–30. Douglass, A.E. 1929. The secret of the Southwest solved by talkative tree rings. National Geographic Magazine 56(6): 736–770. Fletcher, J.M. 1973. European dendrochronology and C-14 dating of timber. Pages 9–27 in T. Watkins, ed., Radiocarbon: Calibration and Prehistory. Edinburgh University Press, Edinburgh. Fletcher, J.M. 1976. A group of English royal portraits painted soon after 1513. A dendrochronological study. Studies in Conservation 21: 171–178. Fletcher, J. 1977. What wooden panels can reveal. Conservator 1: 12–16. Grissino-Mayer, H.D. 2001. Evaluating crossdating accuracy: A manual and tutorial for the computer program COFECHA. Tree-Ring Research 57(2): 205–221. Grissino-Mayer, H.D., P.R. Sheppard, and M.K. Cleaveland. 2004. A dendroarchaeological reexamination of the “Messiah” violin and other instruments attributed to Antonio Stradivari. Journal of Archaeological Science 31(2): 167–174. Hillam, J., and I. Tyers. 1995. Reliability and repeatability in dendrochronological analysis: tests using the Fletcher archive of panel-painting data. Archaeometry 37(2): 395–405. Hoffsummer, P. 1989. La dendrochronologie des panneaux peints. L’exemple d’un plafond peint par Damery à Alden Biesen. Bulletin de l'Institut archéologique liégois 101: 33–47. Jansma, E., E. Hanraets, and T. Vernimmen. 2004. Tree-ring research on Dutch and Flemish art and furniture. Pages 139–146 in E. Jansma, A. Bräuning, H. Gärtner, and G. Schleser, eds., Tree Rings in Archaeology, Climatology and Ecology, Volume 2. Proceedings of the Dendrosymposium 2003. Schriften des Forschungszentrum Jülich, Reihe Umwelt 44. Tree-Ring Dating of a Panel Painting Page 27 Klein, P. 1981. Dendrochronologische Untersuchungen an Eichenholztafeln von Rogier Van der Weyden. Jahrbuch der Berliner Museen 23: 113–123. Klein, P. 1984. Dendrochronological studies on panels by Jean Fouquet (1415/20–1477/81). Pages 25–26 in ICOM, Committee for Conservation, Seventh Triennial Meeting, September 1984, Copenhagen. Klein, P. 1986. Age determinations based on dendrochronology. Pages 225–237 in R. van Shoute and H. Verougstraete-Marcq, eds., Scientific Examination of Easel Paintings. Journal of the European Study Group on Physical, Chemical, Biological and Mathematical Techniques Applied to Archaeology (PACT) 13. Klein, P. 1993. Dendrochronological analysis of German panels in the National Gallery of Art. Pages 195–197 in J.O. Hand and S.E. Mansfield, eds., German Paintings of the Fifteenth through Seventeenth Centuries. National Gallery of Art, Washington, D.C., and Cambridge University Press, Cambridge. Klein, P. 1994a. Dendrochronological analysis of panels attributed to Petrus Christus. Pages 213–215 in Petrus Christus, Renaissance Master of Bruges. Metropolitan Museum of Art, New York. Klein, P. 1994b. Lucas Cranach und seine Werkstatt. Holzarten und dendrochronologische Analyse. Pages 194–200 in C. Grimm, J. Erichsen, and E. Brockhoff, eds., Lucas Cranach. Ein Maler-Unternehmer aus Franken. Veröffenlichungen zur Bayerischen Geschichte und Kultur 26. Klein, P. 1996a. Dendrochronolgicky pruzkum deskovych obrazu Hieronyma Bosche a jeho nasledovniku. Pages 54–59 in O. Kotkova, ed., Hieronymus Bosch – nasledovnik: Tree-Ring Dating of a Panel Painting Page 28 Dvanactilety Jezi v chramu. Obraz po zrestaurovani. Nationalgalerie, Prague, Czech Republic. Klein, P. 1996b. Dendrochronologische Untersuchungen an Gemäldetafeln der Malerfamilie Tom Ring. Pages 212–215 in A. Lorenz, ed., Die Maler Tom Ring. Westfälisches Landesmuseum für Kunst und Kulturgeschichte, Münster, Germany. Klein, P. 1998. Dendrochronological analyses of panel paintings. In: K. Dardes, ed., The Structural Conservation of Panel Paintings: Proceedings of a Symposium at the J. Paul Getty Museum. Oxford University Press, Oxford: 39–54. Klein, P., and J. Bauch. 1990. Analyses of wood from Italian paintings with special reference to Raphael. Pages 85–91 in J. Shearman and M.B. Hall, eds., The Princeton Raphael Symposium 1983. Princeton University Press, Princeton, New Jersey. Kuniholm, P.I. 2000. Dendrochronology (tree-ring dating) of panel paintings. Pages 206–226 in W.S. Taft, Jr., and J.W. Mayer, eds., The Science of Paintings. Springer Verlag, New York. Lavier, C., and G. Lambert. 1996. Dendrochronology and works of art. Pages 543-556 in J.S. Dean, D.M. Meko, and T.W. Swetnam, eds., Tree Rings, Environment, and Humanity. Radiocarbon 1996, University of Arizona Press, Tucson. Orvis, K.H., and H.D. Grissino-Mayer. 2002. Standardizing the reporting of abrasive papers used to surface tree-ring samples. Tree-Ring Research 58(1/2): 47–50. Stokes, M.A., and T.L. Smiley. 1996. An Introduction to Tree-Ring Dating. University of Arizona Press, Tucson. 73 pp. Tree-Ring Dating of a Panel Painting Page 29 Swetnam, T.W., M.A. Thompson, and E.K. Sutherland. 1985. Using dendrochronology to measure radial growth of defoliated trees. USDA Forest Service Agricultural Handbook 639. 39 pp. Wazny, T. 1992. Historical timber trade and its implications on dendrochronological dating. In: T.S. Bartholin, B.E. Berglund, D. Eckstein, F.H. Schweingruber, and O. Eggertsson, eds., Tree Rings and Environment: Proceedings of the International Symposium, Ystad, South Sweden, 3–9 September, 1990. Lundqua Report (Department of Quaternary Geology, Lund University, Sweden) 34: 331–333. Wazny, T. 2002. Baltic timber in western Europe – an exciting dendrochronological question. Dendrochronologia 20(3): 313–320. Tree-Ring Dating of a Panel Painting Page 30 heartwood sapwood EW LW C B A bark one annual ring = EW + LW Figure 1. A sequence of tree rings from a red oak (Quercus rubra L.) showing a tree ring that is narrower (A) than a tree ring that is average in width (C), and a tree ring that is wider (B) than average. Growth of the tree is from left to right with the center of the tree (called the “pith”) shown by the large white dot to the left. EW = earlywood portion of ring, formed in the early part (spring) of the growing season, and composed of large cells called “vessels” needed for transporting water up the stem of the tree. LW = latewood portion of the ring, formed later (summer to late summer) in the growing season, and composed of very small vessels and wood fibers. Heartwood is non-functional (passive) wood that shows up as the darker portion of the tree trunk, composed of wastes and toxins transported inward by ray cells. Sapwood is functional (active) wood and consists of lighter-colored wood that serves to transport materials up the stem. Tree-Ring Dating of a Panel Painting Page 31 Panel rings 1.5 1.0 Reference chronology 0.51400 1450 1500 1550 1600 1650 Figure 2A. Crossdating a sequence of undated (“floating”) tree rings (top, labeled “panel rings”) against a reference tree-ring chronology created previously for the geographic region. The sequence of ring measurements from the panel is slid towards the left one year at a time until the sequence falls in place, much like fingerprint matching. Panel rings 1.5 1.0 Reference chronology 0.51400 1450 1500 1550 1600 1650 Figure 2B. The sequence of undated rings from the panel now matches the pattern from the reference chronology below, indicating rings on the oak panel that were formed between the years 1400 and 1540. Tree-Ring Dating of a Panel Painting Page 32 Figure 3. A triptych (work of art that consists of three painted panels hinged together) by Rogier van der Weyden. The two panels on the left (The Nativity and The Pietà) are housed in the Capilla Real, Granada, while the right panel (Christ Appearing to His Mother) is housed in the Metropolitan Museum of Art in New York. A second identical triptych is displayed in BerlinDahlem. Art historians believed the Metropolitan-Granada triptych to be the original while the Berlin triptych was believed to the a copy. However, the Berlin triptych has an outermost ring dated to 1406 while tree rings from the wood of the Metropolitan triptych was cut after 1482, almost two decades after van der Weyden’s death in 1464. Tree-ring dating techniques demonstrated that Berlin-Dahlem instead has the early original painting from Rogier van der Weyden’s lifetime while New York and Granada have a posthumous copy (adapted from Kuniholm 2000). Tree-Ring Dating of a Panel Painting Figure 4. The painting that was analyzed at the Laboratory of Tree-Ring Science, University of Tennessee, on 5 May 2004 and again on 23 June 2004. Page 33 Tree-Ring Dating of a Panel Painting Page 34 Figure 5. The painting was removed from its frame and the top and bottom of each of the three panels were carefully and meticulously polished with fine-mesh finishing film. This polishing was necessary to reveal the cellular structure of the wood so that rings could be easily discerned and measured. Tree-Ring Dating of a Panel Painting Figure 6. The painting was next carefully secured via cushioned clamps to a custom-designed brace that was attached to a Velmex movable stage micrometer that rested on the floor. Shown are Dr. Henri D. Grissino-Mayer (left) and laboratory technician Daniel Lewis (right). Page 35 Tree-Ring Dating of a Panel Painting Page 36 Figure 7. The vertical alignment of the painting was crucial to ensure that ring measurements were as precise as possible. The microscope above the painting was fitted with a SONY color CCD camera that projected the tree rings from the panels onto a color video monitor. The three panels can clearly be seen, connected with two sets of cross braces on the top and middle sections of the painting. Shown are laboratory technicians Saskia van de Gevel (left), Evan Larson (center), and Dr. Henri D. Grissino-Mayer (right). Tree-Ring Dating of a Panel Painting Page 37 Figure 8. A 2 X 2 contingency table set up for a chi-square analysis. “Test series” is the tree-ring series that is being dated, while “Reference series” is the reference chronology. Rings are evaluated as being wide or narrow for both series. If the two series are aligned correctly, then the wide and narrow rings should match up between the two, in the shaded cells A and D. Mismatched rings will occupy cells B and C. The “actual” number of matches or mismatches is placed in each cell, along with the “expected” number in each cell, from which the chi-square statistic is generated. Tree-Ring Dating of a Panel Painting Page 38 Figure 9-1. Tree rings found on Panel 1, shown here as the lighter bands that go up and down relative the panel. Growth is from the left to right. The curvature of the rings on the inside towards the left of the photograph indicates this panel was cut off-center (called a “tangential cut,” see Figure 10). The darker bands that are almost perpendicular to the rings are ray cells that emanate from the center of the tree. These can occasionally cause a ring to become offset when divided by a ray cell. Figure 9-2. Tree rings found on Panel 2, the middle panel. Growth is again from left to right. The rings are nearly perpendicular to the core, indicating this panel was cut with the center of the log (called “radial cut”). Notice also that the ray cells are almost parallel to the panel itself, again indicating a position in the log that is aligned with the center. This was the highest-quality panel of the three for tree-ring dating purposes. Figure 9-3. Tree rings found on Panel 3. Growth is from left to right. The rings on this panel were the most problematic, being unclear in some locations. In addition, the ray cells (the darker bands perpendicular to the lighter-colored rings) caused significant offsets of rings on either side of the ray, which can be clearly seen inside the red circles. Tree-Ring Dating of a Panel Painting Figure 10. Different cuts from a log that demonstrate the various locations of the panels used in paintings by noting the: (1) curvature of innermost rings, (2) presence or absence of heartwood, (3) presence or absence of bark (which provides a precise cutting date) (from Kuniholm 2000). Page 39 Tree-Ring Dating of a Panel Painting Page 40 Figure 11. The skeleton plot for the tree rings from Panel 2 (top) were overlain on top of the ring patterns from the Baltic 1 reference chronology (bottom). The match shows that the rings from Panel 2 date from A.D. 1369 to 1585. Tree-Ring Dating of a Panel Painting Page 41 Standard Deviation Units 3 2 1 0 -1 -2 -3 1350 1375 1400 1425 1450 1475 1500 1525 1550 1575 1600 Figure 12A. Line graphs showing the high level of correspondence between the measurements from Panel 2 (black line) and the tree-ring indices from the Baltic 1 reference chronology (red line) made available from the University of Sheffield Dendrochronology Laboratory. All data were converted to standard deviation units (z-scores) to give both series a common mean (0.0) and standard deviation (1.0) for ease in comparison. Note especially the strong level of correspondence in the long-term trends (low-frequency variation). Standard Deviation Units 3 2 1 0 -1 -2 -3 1350 1375 1400 1425 1450 1475 1500 1525 1550 1575 1600 Figure 12B. Line graphs showing the high level of correspondence between the measurements from Panel 2 (black line) and the tree-ring indices from the Netherlands Baltic chronology (red line) made available from the Netherlands Centre for Dendrochronology. Tree-Ring Dating of a Panel Painting Figure 13. The final dated positions for the three panels compared against the Baltic 1 reference chronology. Page 42 Tree-Ring Dating of a Panel Painting Page 43 APPENDIX 1 A Complete Listing of Published Articles that Used Tree-Ring Dating on Panel Paintings (arranged chronologically) Bauch, J., and D. Eckstein. 1970. Dendrochronological dating of oak panels of Dutch seventeenth-century paintings. Studies in Conservation 15: 45–50. Corona, E. 1970. Note dendrocronologiche sul quadro di S. Maria della Clemenza in Roma. Studi Trentini di Scienze Naturali 47(2): 133–140. Meier-Siem, M. 1970. Die Beurteilung von Gemälden unter Berücksichtigung der jahrringchronologischen Datierung der Bildtafel und anderer technologischer Methoden. Mitteilungen der Bundesforschungsanstalt für Forst und Holzwirtschaft, Reinbek bei Hamburg 77: 59–70. Bauch, J., D. Eckstein, and M. Maier-Siem. 1972. Dating the wood of panels by a dendrochronological analysis of the tree-rings. Nederlands Kunsthistorisch Jaarboek 23: 485–496. Fletcher, J.M. 1973. European dendrochronology and C-14 dating of timber. In: T. Watkins, ed., Radiocarbon: Calibration and Prehistory. Edinburgh University Press, Edinburgh, Scotland: 9–27. Bauch, J., D. Eckstein, and G. Brauner. 1974. Dendrochronologische Untersuchungen an Gemäldentafeln und Plastiken. Maltechnik Restauro 80: 32–40. Eckstein, D., and J. Bauch. 1974. Dendrochronologie und Kunstgeschichte, dargestellt an Gemälden holländischer und altdeutscher Malerei. Mitteilungen der deutschen Dendrologischen Gesellschaft 67: 234–243. Fletcher, J.M. 1974. Tree ring dates for some panel paintings in England. The Burlington Magazine 116: 250–258. Fletcher, J.M. 1974. Slices from a deep cake – dating panel paintings of St. Etheldreda from Ely. Country Life 155(4004): 728–730. Bauch, J., D. Eckstein, and G. Brauner. 1974. Dendrochronologische Untersuchungen an Gemäldetafeln und Plastiken. Maltechnik Restauro (1/74): 32–40. Grosser, D. 1974. Holzanatomische Untersuchungsverfahren an kunstgeschichtlichen, kulturgeschichtlichen und archäologischen Objekten. Maltechnik Restauro 80: 68–86. Bauch, J., and D. Eckstein. 1975. Tree-ring analysis applied to the dating of paintings. In: Bioecological fundamentals of dendrochronology. Symposium materials of Xll-international botanical congress, Leningrad, July, 1975. 13–15. Tree-Ring Dating of a Panel Painting Page 44 Grosser, D., and E. Geier. 1975. Die in der Tafelmalerei und Bildschnitzerei verwendeten Holzarten und ihre Bestimmung nach mikroskopischen Merkmalen. Teil I. Nadelhölzer. Maltechnik Restauro 3: 127–148. Fletcher, J.M. 1976. A group of English royal portraits painted soon after 1513. A dendrochronological study. Studies in Conservation 21: 171–178. Grosser, D., and E. Graessle. 1976. Die in der Tafelmalerei und Bildschnitzerei verwendeten Holzarten und ihre Bestimmung nach mikroskopischen Merkmalen. Teil II. Europäische Laubhölzer, Fortsetzung. Maltechnik Restauro 4: 232–252. Grosser, D., and E. Graessle. 1976. Die in der Tafelmalerei und Bildschnitzerei verwendeten Holzarten und ihre Bestimmung nach mikroskopischen Merkmalen. Teil II. Europäische Laubhölzer. Maltechnik Restauro 4: 40–54. Fletcher, J. 1977. What wooden panels can reveal. Conservator 1: 12–16. Fletcher, J.M. 1978. Tree-ring analysis of panel paintings and chests. In: J. Fletcher, ed., Dendrochronology in Europe. British Archaeological Reports International Series 51: 303– 306. Klein, P. 1979. Alte Gemälde auf Holz gemalt. Jahrringanalytische Untersuchungen an Gemäldetafeln. Holz Zentralblatt 105(153): 2287–2288. Fletcher, J.M. 1980. Tree-ring dating of Tudor portraits. Proceedings of the Royal Institute of Great Britain 52: 81–104. Fletcher, J.M. 1980. Dendrochronology: principles for its application to oak in Northern France. In: Actes du XX symposium international d’archéométrie, Paris 26–29 mars 1980 Volume 1. Revue d'Archeometrie 4: 1–8. Bauch, J., and D. Eckstein. 1981. Wood biological investigations on panels of Rembrandt paintings. Wood Science and Technology 15: 251–263. Klein, P. 1981. Dendrochronologische Untersuchungen an Eichenholztafeln von Rogier Van der Weyden. Jahrbuch der Berliner Museen 23: 113–123. Bauch, J. 1981. Dendrochronology of wood panels, demonstrated by analysis of the painting The incredulity of Thomas. In: Rembrandt and the Incredulity of Thomas: Papers on a Rediscovered Painting from the Seventeenth Century. Verlag Aliotta and Manhart, Leiden: 35–39. Klein, P. 1981. Holz und Kunst. Holzanalytische Untersuchungen am Tostedter Altar. Zeitschrift des Vereins für Geschichte, Natur- und Heimatkunde Tostedt und Umgebung 1/81: 10–15. Tree-Ring Dating of a Panel Painting Page 45 Fletcher, J. 1982. Panel examination and dendrochronology. The J. Getty Museum Journal 10: 39–44. Gaborit, J.-R. 1983. L'‚tude des bois en laboratoire. Revue de l'Art 60: 70. Klein, P. 1983. Dating of art-historical objects. In: D. Eckstein, S. Wrobel, and R.W. Aniol, eds., Dendrochronology and Archaeology in Europe. Mitteilungen der Bundesforschungsanstalt für Forst- und Holzwirtschaft, Hamburg 141: 209–222. Klein, P. 1983. L'examen dendrochronologique des panneaux peints. Revue de l'Art 60: 71–72. Baillie, M.G.L. 1984. Some thoughts on art-historical dendrochronology. Journal of Archaeological Science 11: 371–393. Fletcher, J.M., and M.C. Tapper. 1984. Medieval artefacts and structures dated by dendrochronology. Medieval Archaeology 28: 112–132. Klein, P. 1984. Dendrochronological studies on panels by Jean Fouquet (1415/20–1477/81). In: ICOM, Committee for Conservation, Seventh Triennial Meeting, September 1984, Copenhagen. 25–26. Corona, E. 1984. Il supporto ligneo della Pala di Monteluce. Bollettino dei musei e gallerie pontificie 5: 137–151. Baillie, M.G.L., J. Hilliam, K.R. Briffa, and D.M. Brown. 1985. Age-dating (Re-dating) the English art-historical tree-ring chronologies. Nature 315(6017): 317–319. Klein, P. 1985. Dendrochronologische Untersuchungen Musikinstrumenten. Dendrochronologia 3: 25–44. an Gemäldetafeln und Bunney, S. 1985. Tree-ring dating for paintings is thrown into doubt. New Scientist (24): 37. Klein, P. 1986. Age determinations based on dendrochronology. In: R. van Shoute and H. Verougstraete-Marcq, eds., Scientific Examination of Easel Paintings. PACT 13: 225–237. Klein, P. 1986. Dendrochronological analysis of early Netherlandish panels in The National Gallery. In: Early Netherlandish Paintings. National Gallery of Art, Cambridge University Press, Cambridge, United Kingdom: 259–260. Klein, P. 1987. Dendrochronologische Untersuchungen an Bildtafeln des 15. Jahrhunderts. In: Le dessin sous-jacent dans la peinture. Colloque VI, 12–14 September 1985. Universit‚ de Louvain-la-Neuve: 29–40. Klein, P., D. Eckstein, T. Wazny, and J. Bauch. 1987. New findings for the dendrochronological dating of panel paintings of the 15th to 17th century. In: Scientific Examination of Works of Art, Volume 1. International Council of Museums Working Group: 51–54. Tree-Ring Dating of a Panel Painting Page 46 Klein, P. 1989. Dendrochronological examinations on panel paintings and musical instruments. Erdeszeti es Faipari Tudomanyos Kozlemenyei 0(2): 107–124. Klein, P. 1989. Dendrochronological studies on oak panels of Rogier van der Weyden and his circle. In: Le dessin sous-jacent dans la peinture. Colloque VII, September 1985. Universit‚ Louvain-La-Neuve: 25–36. Klein, P. 1989. Zum Forschungsstand der Dendrochronologie europäischer Tafelmalerei. Restauratorenblaetter 10: 35–47. Hoffsummer, P. 1989. La dendrochronologie des panneaux peints . L'exemple d'un plafond peint par Damery. Alden Biesen. Bulletin de l'Institut archéologique lingois 101: 33–47. Klein, P. 1990. Tree-ring chronologies of conifer wood and its application to the dating of panels. In: ICOM, Committee for Conservation, Ninth Triennial Meeting, August 1990, Dresden. 38–40. Klein, P., and J. Bauch. 1990. Analyses of wood from Italian paintings with special reference to Raphael. In: J. Shearman and M.B. Hall, eds., The Princeton Raphael Symposium 1983. Princeton University Press, Princeton, NJ: 85–91. Klein, P. 1991. Dendrochronologische Untersuchungen an Gemäldetafeln der Hamburger Kunsthalle. In: T. Sello and R. Mueller, eds., Nicht nur mit Pinsel und Oel. Verlag am Galgenberg, Hamburg, Germany: 108–113. Klein, P. 1991. The differentiation of originals and copies of Netherlandish panel paintings by dendrochronology. In: Le dessin sous-jacent dans la peinture. Colloque VIII, September 1989. Universit‚ Louvain-La-Neuve: 29–42. Klein, P., and T. Wazny. 1991. Dendrochronological analyses of paintings of Gdansk painters of the 15th to the 17th century. Dendrochronologia 9: 181–191. Bonde, N., T. Bartholin, K. Christensen, A. Daly, and O.H. Eriksen. 1993. Dendrokronologiske dateringsundersogelser pa Nationalmuseet 1992. Arkeologiske udgravninger i Danmark: 305–321. Bonde, N., T. Bartholin, K. Christensen, and O.H. Ericksen. 1993. Dendrokronologiske dateringsundersogelser pa Nationalmuseet 1993. Arkeologiske udgravninger i Danmark: 294–310. Klein, P. 1993. An overview about dendrochronological analyses of panel paintings. In: Le dessin sous-jacent dans la peinture. Colloque IX, September 1991. Universit‚ de Louvain-laNeuve: 165–178. Tree-Ring Dating of a Panel Painting Page 47 Klein, P. 1993. Dendrochronologische Untersuchungen an Gemälden von Stefan Lochner und Nachfolge. In: Stefan Lochner – Meister in Koeln. Ausstellungs-Katalog, Koeln, WallrafRichartz-Museum: 187–189. Klein, P. 1993. Dendrochronologische Untersuchungen an Niederländischen Tafelbildern des 15. und der ersten Hälfte des 16. Jahrhunderts im Städel. In: J. Sander, ed., Niederlaendische Gemaelde im Staedel 1400–1500. Verlag Phillip von Zabern, Mainz am Rhein, Germany: 453–457. Klein, P. 1993. Dendrochronological analysis of German panels in the National Gallery of Art. In: J.O. Hand and S.E. Mansfield, eds., German Paintings of the Fifteenth through Seventeenth Centuries. National Gallery of Art, Washington, D.C., Cambridge University Press, Cambridge, United Kingdom: 195–197. Plahter, U. 1993. Norwegian altar fronts. A medieval panel construction suitable for dendrochronology? In: O. Storsletten and T. Thun, eds., Dendrochronology and the Investigation of Buildings, Proceedings of an International Seminar at the Academy of Science and Ketters, Oslo, 1–2 November, 1991. Riksantikvarens Rapporter 22: 62–70. Klein, P. 1994. Dendrochronological analysis of panels attributed to Petrus Christus. In: Petrus Christus, Renaissance Master of Bruges. Metropolitan Museum of Art, New York: 213–215. Klein, P. 1994. Lucas Cranach und seine Werkstatt. Holzarten und dendrochronologische Analyse. In: C. Grimm, J. Erichsen, and E. Brockhoff, eds., Lucas Cranach. Ein MalerUnternehmer aus Franken. Veröffenlichungen zur Bayerischen Geschichte und Kultur 26: 194–200. Bräker, O.U. 1994. Witterungs- und Klimamodelle mit Jahrringen, eine Fallstudie mit Holz aus Zillis, Schams. Bündner Wald 47(6): 59–67. Corona, E. 1994. Sulla caratterizzazione dei manufatti lignei: opere pittoriche. Italia Forestale e Montana 49(4): 401–412. Klein, P. 1994. Dendrochronological analysis of panels of Hans Memling. In: Essays, Hans Memling. Stedelijke Museum, Brügge: 101–103. Klein, P. 1994. Holzartenbestimmung und dendrochronologische Analyse an Gemäldetafeln von Lucas Cranach d.Ž. und seiner Werkstatt. In: Katalog des Schloßmuseums Gotha „Gotteswort und Menschenbild“, Teil 1. Schloámuseum, Gotha: 210–215. Baillie, M.G.L. 1995. A Slice Through Time: Dendrochronology and Precision Dating. B.T. Batsford, London, 176 pp. Hillam, J., and I. Tyers. 1995. Reliability and repeatability in dendrochronological analysis: tests using the Fletcher archive of panel-painting data. Archaeometry 37(2): 395–405. Tree-Ring Dating of a Panel Painting Page 48 Pousset, D. 1996. Le Chateau de Vincennes: Etude dendrochronologique des lambris du donjon. Mémoire de Diplome d’Etudes Approfondies en Méthodes et Techniques Nouvelles en Sciences Humaines, Université de Franche-Comt, Faculté des Lettres et Sciences Humaines, Besancon, France, Besancon, France. 76 pp. Klein, P. 1996. Dendrochronologische Untersuchungen an Gem„ldetafeln der Malerfamilie tom Ring. In: A. Lorenz, ed., Die Maler Tom Ring. Westfälisches Landesmuseum für Kunst und Kulturgeschichte, Münster, Germany: 212–215. Klein, P. 1996. Dendrochronolgicky pruzkum deskovych obrazu Hieronyma Bosche a jeho nasledovniku. In: O. Kotkova, ed., Hieronymus Bosch – nasledovnik: Dvanactilety Jezi v chramu. Obraz po zrestaurovani. Nationalgalerie, Prague: 54–59. Klein, P. 1996. Dendrochronological findings in panels of the Campin group. In: S. Foister, S. Nash, eds., Robert Campin. New directions in scholarship. National gallery in collaboration with Brepols Publishers, London, Turnhout (Belgium): 77–86. Stöckli, V. 1996. Sprechende Jahrringe. Terra Grischuna 2: 33–36. Lavier, C., and G. Lambert. 1996. Dendrochronology and works of art. In: J.S. Dean, D.M. Meko, and T.W. Swetnam, eds., Tree Rings, Environment, and Humanity. Radiocarbon 1996: 543–556. Klein, P. 1997. Dendrochronological analyses of the two panels of “Saint Francis Receiving the Stigmata”. In: Jan Van Eyck: Two Paintings of Saint Francis Receiving the Stigmata. Antique Collectors Club, Woodbridge, United Kingdom: 47–50. Arnold, M., and K. Richter. 1997. Holztechnologische Aspekte bei der Erhaltung der Bilderdecke. In: Ch. Bläuer Böhm, H. Rutishauser and M.A. Nay, eds., Die romanische Bilderdecke der Kirche St. Martin in Zillis. Grundlagen zur Konservierung und Pflege. Paul Haupt, Berne: 227–242. Klein, P. 1997. Dendrochronological analyses of panels of Hans Memling and his contemporaries. In: H. Verougstraete, R. Van Schoute and M. Smeyers, eds., Memling Studies. Uitgeverij Peeters, Leuven. Proceedings of the International Colloquium (Bruges, 10–12 November 1994): 287–295. Klein, P. 1998. Dendrochronological analyses of panel paintings. In: K. Dardes, ed., The Structural Conservation of Panel Paintings: Proceedings of a Symposium, J. Paul Getty Museum. Oxford University Press, Oxford: 39–54. Klein, P. 1998. Some aspects of the utilization of different wood species in certain European workshops. In: A. Roy and P. Smith, eds., Painting techniques history, materials and studio practice. The International Institute for Conservation of Historic and Artistic Works, London. Contributions to the Dublin Congress 7–11 September 1998: 112–114. Tree-Ring Dating of a Panel Painting Page 49 Miles, D.H., and M.J. Worthington. 1999. List 100: Tree-ring dates from Oxford Dendrochronology Laboratory: General list. Vernacular Architecture 30: 98–105. Kuniholm, P.I. 2000. Dendrochronology (tree-ring dating) of panel paintings. In: W.S. Taft, Jr., and J.W. Mayer, eds., The Science of Paintings. Springer Verlag, New York: 206–226. Vermet, B. 2001. Hieronymus Bosch: painter, workshop or style?. In: J. Koldeweij, P. Vandenbroeck and B. Vermet, eds., Hieronymus Bosch: The Complete Paintings and Drawings. NAI Publishers, Rotterdam, The Netherlands: 83–99. Fraiture, P. 2002. Contribution of scientific methods to the understanding of the work of the 16th century painter, Henri Bles. Dendrochronologia 20(3): 285–299. Jansma, E., E. Hanraets, and T. Vernimmen. 2004. Tree-ring research on Dutch and Flemish art and furniture. In: E. Jansma, A. Bräuning, H. Gärtner, and G. Schleser, eds., Tree Rings in Archaeology, Climatology and Ecology, Volume 2. Proceedings of the Dendrosymposium 2003. Schriften des Forschungszentrum Jülich, Reihe Umwelt 44: 139–146. Tree-Ring Dating of a Panel Painting Page 50 APPENDIX 2A The “Baltic 1” reference tree-ring chronology kindly supplied by the University of Sheffield Dendrochronology Laboratory by Dr. Ian Tyers. The first three lines are header lines and contain required metadata (species, range of years, investigators, date of chronology development). Subsequent lines represent the actual data. After the site name “BALTI,” is the year/decade year followed by the annual index (without decimals) and sample depth for the years in that decade. For example, “BALTI 1160 122 1” in the fourth line shows that the year 1160 has an index of 1.22 based on one measurement series. “BALTI 1210 124 6” in the tenth line shows that the year 1210 has an index of 1.24 based on a sample depth of 6 panels. “9990” represents no data. BAL BAL BAL BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI 1 FLETCHER PANELS BALTIC AREA 1 - 64 TIMBER MEAN 2 BALTIC 1 3 MADE JH/IT 12/11/93 [HILLAM| 1995 #237] 11569990 09990 09990 09990 09990 09990 0 62 1160 122 1 82 1 82 1 52 1 92 1 92 1 72 1170 92 2 112 2 132 2 142 2 177 2 137 2 122 1180 130 2 108 3 125 3 162 3 164 3 115 3 147 1190 162 3 131 3 106 3 110 3 122 3 144 3 131 1200 159 4 129 4 96 4 132 4 137 4 130 4 125 1210 124 6 135 6 141 6 95 6 114 6 113 6 104 1220 92 6 84 6 112 6 114 6 95 6 119 6 75 1230 144 6 144 6 135 6 104 6 136 7 119 8 112 1240 136 9 117 9 138 9 142 9 134 9 138 9 182 1250 174 9 187 9 165 9 162 9 153 10 165 10 160 1260 132 12 102 12 141 13 110 13 136 13 122 14 142 1270 129 14 151 16 122 17 157 16 148 16 136 17 133 1280 125 21 130 21 117 21 140 23 132 23 144 23 136 1290 119 25 129 25 97 25 109 26 131 26 132 26 137 1300 142 28 140 28 151 29 140 30 122 30 120 30 118 1310 123 31 138 31 154 31 143 31 134 32 141 32 176 1320 140 32 146 32 131 32 126 33 107 33 135 34 130 1330 112 36 130 37 141 38 114 38 137 39 128 39 163 1340 136 40 160 40 141 40 122 40 131 40 148 40 133 1350 152 41 147 41 149 43 150 43 137 43 113 44 119 1360 141 49 125 49 157 49 166 50 151 50 140 50 155 1370 111 50 136 50 103 50 118 50 121 50 139 50 150 1380 161 52 178 52 156 53 147 56 157 56 128 56 139 1390 162 59 142 59 150 59 149 59 144 59 109 59 141 1400 157 59 128 60 155 61 116 61 133 61 127 61 156 1410 118 60 150 60 159 60 152 60 138 60 153 59 150 1420 115 58 110 57 115 57 87 57 106 57 120 57 106 1430 114 55 105 55 115 55 107 55 113 55 119 54 135 1440 132 54 154 54 142 54 160 54 136 53 160 53 128 1450 117 52 132 52 127 51 110 51 135 50 128 50 124 1460 142 50 100 50 89 50 83 50 118 50 101 50 96 1470 106 47 113 45 137 43 109 43 115 43 124 42 121 1480 121 42 136 42 133 42 113 42 113 42 112 42 142 1490 140 38 102 36 89 35 98 35 100 35 122 35 116 1500 126 29 158 29 139 28 137 28 121 28 129 28 146 1510 144 26 141 26 139 26 129 25 138 22 164 22 159 1520 103 19 139 19 148 18 130 16 126 15 139 12 143 1530 113 7 110 7 88 7 114 7 106 7 132 7 113 1540 126 6 125 6 134 6 144 5 140 5 170 5 126 1550 120 4 172 4 150 4 188 3 117 2 125 2 112 1560 107 1 92 1 92 1 92 1 102 1 82 1 77 1570 82 1 77 1 94 1 87 1 67 1 96 1 82 1580 102 1 92 1 86 1 102 1 97 1 102 1 132 1590 92 1 92 1 102 1 102 1 102 1 112 1 92 QUSP 1156 1597 1 1 2 3 4 4 6 6 9 9 11 14 18 24 26 30 32 34 39 40 46 50 51 56 59 60 58 56 54 53 50 49 42 40 32 28 22 11 7 5 1 1 1 1 1 82 52 107 164 130 122 104 124 100 167 138 121 100 128 139 98 154 127 123 153 82 162 162 128 85 146 142 87 135 139 130 116 119 148 107 144 148 144 127 130 97 82 107 87 112 1 102 1 92 2 125 3 130 4 171 4 145 6 92 6 122 9 83 9 169 12 105 14 133 19 97 24 142 27 131 31 123 32 151 36 123 39 140 41 133 47 95 50 142 51 171 58 137 59 117 60 159 57 138 56 92 54 125 52 151 49 103 49 127 42 107 40 139 32 113 27 136 21 121 10 138 7 110 5 144 1 112 1 82 1 97 1 112 19990 1 102 1 72 2 105 3 107 4 153 5 111 6 140 6 128 9 123 9 162 12 142 14 130 19 115 24 146 27 138 31 129 32 130 36 106 39 150 41 135 47 145 50 126 52 154 58 156 59 139 60 140 58 100 55 116 54 130 52 127 49 114 48 123 42 93 40 141 32 112 27 137 21 125 9 112 7 114 5 140 1 87 1 82 1 102 1 102 09990 1 1 2 3 4 5 6 6 9 9 12 14 20 25 28 31 32 36 40 41 48 50 52 58 59 60 58 55 54 52 50 47 42 40 30 26 21 8 7 5 1 1 1 1 0 Tree-Ring Dating of a Panel Painting Page 51 APPENDIX 2B The “Baltic 2” reference tree-ring chronology kindly supplied by the University of Sheffield Dendrochronology Laboratory by Dr. Ian Tyers. See comments in Appendix 2A for explanation of the Index format. More information on the Index format can be obtained from the World Data Center for Paleoclimatology at http://www.ncdc.noaa.gov/paleo/treeinfo.html. BAL BAL BAL BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI BALTI 1 FLETCHER PANELS BALTIC AREA 2 - 40 TIMBER MEAN 2 BALTIC 2 3 MADE JH/IT 8/4/94 [HILLAM| 1995 #237] 12579990 09990 09990 09990 09990 09990 09990 1260 142 1 122 1 162 1 152 1 182 1 132 1 202 1270 122 1 152 1 172 1 172 1 162 1 201 2 166 1280 176 2 154 3 151 3 141 3 161 4 184 4 166 1290 144 4 139 4 129 4 139 4 184 4 181 4 196 1300 195 5 185 5 209 5 155 5 113 5 99 5 141 1310 191 5 143 5 179 5 213 6 184 6 198 6 238 1320 174 6 193 6 206 8 181 8 142 8 172 8 144 1330 167 8 141 9 122 9 115 10 184 10 175 10 191 1340 162 10 192 10 171 10 158 11 165 12 196 12 150 1350 188 14 188 14 164 14 190 14 181 14 180 14 168 1360 178 17 115 17 171 17 200 18 165 18 177 18 181 1370 177 18 152 19 141 19 134 19 155 19 170 19 142 1380 183 19 199 20 160 21 183 21 166 21 177 21 167 1390 173 20 177 22 175 22 153 23 166 24 138 24 161 1400 191 25 177 27 176 27 130 27 145 27 167 27 216 1410 161 25 180 25 174 24 173 25 176 24 164 26 180 1420 138 25 157 24 141 24 124 24 111 24 133 24 122 1430 152 24 153 24 167 24 143 24 151 24 159 24 169 1440 186 24 153 24 158 25 170 25 165 26 190 26 171 1450 176 27 166 27 150 27 137 27 188 27 149 27 171 1460 190 26 150 27 151 27 156 27 156 28 168 29 186 1470 159 30 168 30 186 30 171 30 151 30 185 30 167 1480 215 30 240 30 211 30 192 30 210 30 180 30 257 1490 201 30 172 30 166 30 195 29 187 29 218 29 192 1500 172 29 219 29 204 28 191 28 187 28 171 28 199 1510 191 28 210 27 190 26 171 26 206 26 239 26 201 1520 165 23 175 23 178 22 216 22 194 22 154 22 202 1530 158 20 157 20 121 20 172 20 178 20 153 19 143 1540 182 18 171 18 201 18 175 18 174 18 188 18 181 1550 186 17 160 17 156 17 165 16 156 14 184 12 170 1560 194 9 202 9 209 8 215 7 175 7 200 6 175 1570 226 4 176 4 179 4 156 4 181 4 221 4 184 1580 151 2 151 1 171 1 151 1 131 1 121 1 141 1590 111 1 121 1 151 1 191 1 161 1 231 1 171 1600 111 1 101 1 131 1 111 1 111 1 101 1 81 1610 131 1 91 1 191 1 131 1 41 1 51 19990 QUSP 1257 1615 0 192 1 152 2 166 4 134 4 186 5 165 6 171 8 169 10 155 14 161 16 168 18 193 19 145 21 202 24 132 27 169 25 144 24 113 24 155 26 172 27 155 30 180 30 159 30 242 29 136 28 192 24 210 22 174 19 166 18 180 12 182 5 197 4 186 1 81 1 211 1 151 09990 1 182 1 202 2 136 4 171 4 191 5 175 6 168 8 177 10 197 14 179 16 149 18 173 19 195 21 178 24 187 26 192 25 181 25 155 24 176 26 147 26 177 30 194 30 233 30 234 29 172 28 217 23 192 22 174 18 156 18 193 12 200 5 229 4 195 1 201 1 161 1 141 09990 1 162 1 142 2 161 4 149 5 201 5 197 6 166 8 165 10 179 14 148 16 179 18 170 19 168 21 188 25 185 25 172 25 154 25 159 24 129 26 168 26 169 30 192 30 209 30 248 29 169 28 182 23 160 22 156 18 194 18 235 12 164 5 239 3 178 1 191 1 121 1 141 09990 1 1 2 4 5 5 6 8 10 14 16 18 19 20 25 25 25 25 24 26 26 30 30 30 29 28 23 22 18 18 11 5 3 1 1 1 0 Tree-Ring Dating of a Panel Painting Page 52 APPENDIX 3A The original measurements for Panel 2 in Decadal Format (format recognized by the World Data Center for Paleoclimatology). Each line begins with the series ID = “panel02a” followed by the ring number of that decade (1, 10, 20, etc.). The measurements then follow, 10 per line (except on the first and last lines). Each value is recorded in 0.01 millimeters with the decimal removed. For example, “164” = “1.64 mm.” The “999” at the end of the series is an end-of-series “sentinel” used by the software. panel02a panel02a panel02a panel02a panel02a panel02a panel02a panel02a panel02a panel02a panel02a panel02a panel02a panel02a panel02a panel02a panel02a panel02a panel02a panel02a panel02a panel02a 1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 164 184 121 131 122 84 81 105 130 127 123 141 112 51 122 160 148 76 146 88 95 125 198 129 191 149 123 65 104 130 82 113 115 57 115 69 152 94 132 112 84 94 107 109 190 191 131 149 87 69 122 88 115 156 68 114 129 102 97 102 83 81 92 99 111 132 130 183 145 79 95 122 103 89 121 104 88 119 92 149 122 99 99 73 94 97 95 98 109 150 156 93 130 91 94 100 111 119 110 130 88 95 100 139 110 77 94 140 108 75 108 140 164 53 100 42 96 142 135 79 100 95 79 90 101 100 104 111 96 121 122 69 117 179 178 64 92 111 116 164 103 119 114 108 61 93 149 110 134 79 95 99 134 68 132 156 147 103 92 109 99 110 134 97 85 86 128 68 138 90 97 124 81 105 137 87 132 129 172 129 118 105 162 114 129 91 124 129 90 61 123 104 102 76 93 75 133 999 171 116 93 98 100 129 153 110 96 82 124 82 92 121 115 99 77 72 104 118 Tree-Ring Dating of a Panel Painting Page 53 APPENDIX 3B The original measurements for Panel 1 in Decadal Format (format recognized by the World Data Center for Paleoclimatology). Each line begins with the series ID = “panel02a” followed by the ring number of that decade (1, 10, 20, etc.). The measurements then follow, 10 per line (except on the first and last lines). Each value is recorded in 0.01 millimeters with the decimal removed. For example, “69” = “0.69 mm.” The “999” at the end of the series is an end-of-series “sentinel” used by the software. panel01a panel01a panel01a panel01a panel01a panel01a panel01a panel01a panel01a panel01a panel01a panel01a panel01a panel01a panel01a panel01a panel01a panel01a panel01a panel01a panel01a 1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 69 95 55 79 91 127 85 54 59 51 94 95 86 110 104 119 65 44 69 91 56 88 98 60 48 96 101 117 70 67 65 75 122 87 107 93 99 54 46 71 52 83 61 111 96 57 112 82 72 45 49 43 90 73 89 108 56 108 42 47 76 99 71 94 62 84 66 119 87 83 56 44 45 71 119 81 49 107 95 39 61 63 55 57 73 84 58 49 127 109 64 65 43 42 117 131 47 82 105 87 65 56 111 63 89 82 65 53 24 101 122 79 75 36 62 99 109 102 100 104 72 50 65 90 76 74 76 88 44 53 83 105 68 48 32 60 124 110 92 77 119 69 58 66 76 72 98 88 75 43 45 117 123 77 53 36 89 91 93 94 106 155 76 47 60 61 85 999 80 50 66 60 87 120 67 57 34 90 91 91 59 125 92 80 45 53 71 80 78 76 88 118 117 60 62 46 95 67 74 60 108 99 82 46 63 70 52 Tree-Ring Dating of a Panel Painting Page 54 APPENDIX 3C The original measurements for Panel 3 in Decadal Format (format recognized by the World Data Center for Paleoclimatology). Each line begins with the series ID = “panel02a” followed by the ring number of that decade (1, 10, 20, etc.). The measurements then follow, 10 per line (except on the first and last lines). Each value is recorded in 0.01 millimeters with the decimal removed. For example, “139” = “1.39 mm.” The “999” at the end of the series is an end-of-series “sentinel” used by the software. panel03a panel03a panel03a panel03a panel03a panel03a panel03a panel03a panel03a panel03a panel03a panel03a panel03a panel03a panel03a panel03a panel03a panel03a panel03a panel03a 1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 139 120 169 165 123 167 116 88 148 114 80 108 135 179 133 123 115 236 310 100 153 144 110 89 146 179 123 88 95 129 90 65 122 152 96 146 131 247 175 71 149 201 122 129 143 141 128 109 119 148 93 86 144 147 129 144 160 127 109 89 145 134 145 90 169 127 141 121 150 133 117 74 124 138 136 136 285 116 107 90 155 153 112 112 150 108 114 110 100 121 97 112 85 154 97 93 161 91 95 112 147 107 101 132 163 117 100 124 109 133 123 92 112 156 104 101 141 85 83 122 179 131 104 138 178 109 118 140 113 127 115 90 100 140 127 141 142 128 105 130 130 139 117 134 120 117 110 146 68 136 91 134 137 179 132 145 231 105 106 131 95 142 141 108 119 118 113 119 71 131 92 139 119 133 110 98 184 180 102 999 190 154 146 130 151 87 144 113 118 119 130 130 99 148 105 164 147 100 Tree-Ring Dating of a Panel Painting Page 55 APPENDIX 4A COFECHA output that compares the tree rings from Panel 2 with the Baltic 1 reference chronology. The highlighted values show the suggested date adjustment for the panel. The adjustment of “+1368” stands out and was found for 12 of the 18 50-year segments tested. This adjustment would put the beginning year at 1+1368 = A.D. 1369. This dating adjustment was by far the strongest of any adjustments found as shown by the “+1368 12 .44” at the bottom of the table. The next best adjustment was to +1107, with 5 segments showing this adjustment, with an average r-value of only 0.31. PART 8: ADJUSTMENTS FOR UNDATED SERIES: 10:10 Fri 30 Jun 2006 Page 0 -------------------------------------------------------------------------------------------------------------------------------Time span 1156 1597 442 years, best matches for Listed in order from highest correlation Counted Series Segment -------- --------panel02a 1 50 panel02a 11 60 panel02a 21 70 panel02a 31 80 panel02a 41 90 panel02a 51 100 panel02a 61 110 panel02a 71 120 panel02a 81 130 panel02a 91 140 panel02a 101 150 panel02a 111 160 panel02a 121 170 panel02a 131 180 panel02a 141 190 panel02a 151 200 panel02a 161 210 panel02a 168 217 18 segments - Corr Corr Add # 1 Add # 2 -------- -------1272 .56 1337 .48 1203 .47 1368 .46 1368 .49 1203 .46 1368 .53 1203 .42 1372 .48 1342 .46 1368 .46 1372 .46 1485 .41 1163 .35 1226 .45 1368 .39 1368 .52 1226 .45 1368 .57 1308 .36 1368 .44 1394 .37 1368 .45 1246 .39 1051 .37 1162 .33 1335 .35 1341 .34 1387 .36 1069 .36 1199 .39 1376 .38 1376 .45 1159 .44 1159 .47 1376 .42 - - - - - - - Corr Add # 3 -------1322 .46 1206 .45 1206 .40 1166 .38 1368 .42 1302 .43 1324 .34 1286 .33 1181 .38 1181 .36 1354 .35 1367 .37 1367 .33 1082 .33 1135 .36 1159 .36 1199 .42 1199 .37 - - - 50-year segments lagged Corr Corr Add # 4 Add # 5 -------- -------1257 .43 1202 .41 1532 .42 1272 .36 1166 .36 1477 .36 1477 .37 1398 .37 1206 .41 1252 .38 1342 .40 1398 .39 1226 .33 1439 .33 1302 .33 1439 .32 1316 .37 1152 .33 1107 .35 1409 .34 1259 .33 1246 .33 1167 .34 1218 .34 1394 .32 1293 .30 1321 .30 1395 .30 1037 .36 1376 .32 1322 .35 1037 .31 1079 .36 1004 .36 1170 .33 1263 .32 - - - - - - - Corr Add # 6 -------1362 .38 1477 .35 1428 .34 1206 .34 1411 .36 1206 .36 1111 .33 1111 .32 1286 .33 1324 .33 1402 .32 1354 .33 1199 .29 1071 .29 1105 .31 1321 .30 1170 .33 1004 .32 - - - 10 years Corr Corr Add # 7 Add # 8 -------- -------1532 .36 1243 .35 1322 .33 1158 .33 1322 .32 1411 .30 1411 .32 1372 .31 1398 .35 1477 .35 1112 .32 1411 .31 1302 .31 1389 .30 1324 .30 1152 .30 1107 .31 1439 .30 1231 .33 1226 .32 1286 .30 1231 .30 1327 .33 1381 .32 1404 .29 1107 .29 1065 .29 1417 .28 1103 .30 1071 .30 1135 .29 1221 .28 1320 .30 1235 .30 1094 .31 1089 .30 - - - - - - - Corr Add # 9 -------1368 .34 1428 .33 1274 .29 1504 .31 1186 .35 1252 .30 1107 .30 1397 .30 1324 .30 1286 .32 1429 .30 1152 .32 1340 .28 1155 .28 1341 .28 1187 .28 1037 .30 1063 .30 - - - Corr Corr Add #10 Add #11 -------- -------1350 .33 1392 .33 1147 .32 1424 .30 1342 .28 1183 .28 1302 .30 1237 .30 1302 .33 1346 .32 1477 .29 1166 .28 1143 .30 1368 .28 1107 .30 1316 .29 1262 .30 1090 .30 1354 .31 1259 .31 1100 .30 1175 .30 1362 .31 1293 .31 1308 .28 1362 .28 1170 .28 1289 .27 1024 .27 1321 .27 1306 .27 1341 .26 1039 .29 1103 .29 1157 .29 1091 .29 - - - - - - Number of segments Add No R_av Add No R_av Add No R_av Add No R_av Add No R_av Add No R_av Add No R_av Add No R_av +1368 12 .44 +1107 5 .31 +1206 5 .39 +1302 5 .34 +1477 5 .34 +1199 4 .37 +1226 4 .39 +1286 4 .32 +1322 4 .36 +1324 4 .32 +1376 4 .39 +1411 4 .32 +1321 3 .29 +1152 3 .32 +1170 3 .31 +1341 3 .29 +1342 3 .38 +1354 3 .33 +1362 3 .32 +1037 3 .32 +1372 3 .42 +1203 3 .45 +1398 3 .37 +1159 3 .43 Tree-Ring Dating of a Panel Painting Page 56 APPENDIX 4B COFECHA output that compares the tree rings from Panel 2 with the Baltic 1 reference chronology, this time testing 50year segments lagged 25 years. The highlighted values show the suggested date adjustment for the panel. The adjustment of “+1368” stands out and was found for 5 of the 8 50-year segments tested. This adjustment would put the beginning year at 1+1368 = A.D. 1369. This dating adjustment was the only one found for this more rigorous test. PART 8: ADJUSTMENTS FOR UNDATED SERIES: 12:36 Fri 30 Jun 2006 Page 0 ------------------------------------------------------------------------------------------------------------------------------Time span 1156 1597 442 years, best matches for Listed in order from highest correlation Counted Series Segment -------- --------panel02a 1 50 panel02a 26 75 panel02a 51 100 panel02a 76 125 panel02a 101 150 panel02a 126 175 panel02a 151 200 panel02a 168 217 8 segments Number of segments Add No R_av +1368 5 .43 Corr Corr Add # 1 Add # 2 -------- -------1272 .56 1337 .48 1203 .45 1368 .45 1368 .46 1372 .46 1368 .47 1226 .46 1368 .44 1394 .37 1340 .42 1289 .34 1199 .39 1376 .38 1159 .47 1376 .42 - - - - - - Add No R_av Corr Add # 3 -------1322 .46 1206 .37 1302 .43 1286 .42 1354 .35 1341 .34 1159 .36 1199 .37 - - - Add No R_av 50-year segments lagged Corr Corr Add # 4 Add # 5 -------- -------1257 .43 1202 .41 1398 .35 1477 .34 1342 .40 1398 .39 1262 .35 1111 .35 1259 .33 1246 .33 1293 .33 1362 .32 1322 .35 1037 .31 1170 .33 1263 .32 - - - - - - Add No R_av 25 years Corr Add # 6 -------1362 .38 1166 .33 1206 .36 1316 .32 1402 .32 1071 .29 1321 .30 1004 .32 - - - Add No R_av Corr Corr Add # 7 Add # 8 -------- -------1532 .36 1243 .35 1130 .32 1411 .31 1112 .32 1411 .31 1152 .31 1302 .30 1286 .30 1231 .30 1051 .29 1135 .28 1135 .29 1221 .28 1094 .31 1089 .30 - - - - - - - Add No R_av Corr Add # 9 -------1368 .34 1257 .31 1252 .30 1439 .30 1429 .30 1040 .28 1187 .28 1063 .30 - - - Add No R_av Corr Corr Add #10 Add #11 -------- -------1350 .33 1392 .33 1428 .31 1458 .31 1477 .29 1166 .28 1086 .29 1381 .29 1100 .30 1175 .30 1335 .27 1367 .27 1306 .27 1341 .26 1157 .29 1091 .29 - - - - - - Add No R_av Tree-Ring Dating of a Panel Painting Page 57 APPENDIX 5A COFECHA output that compares the tree rings from Panel 1 with the Baltic 1 reference chronology. The highlighted values show the suggested date adjustment for the panel. The adjustment of “+1337” stands out and was found for 12 of the 17 50-year segments tested. This adjustment would put the beginning year at 1+1337 = A.D. 1338. This dating adjustment was the strongest of any adjustments found as shown by the “+1337 12 .39” at the bottom of the table. The next best adjustment was +1396, with 7 segments showing this adjustment, with an average r-value of 0.35. PART 8: ADJUSTMENTS FOR UNDATED SERIES: 13:52 Fri 30 Jun 2006 Page 0 -------------------------------------------------------------------------------------------------------------------------------Time span 1156 1597 442 years, best matches for Listed in order from highest correlation 50-year segments lagged 10 years Counted Corr Corr Corr Corr Corr Corr Series Segment Add # 1 Add # 2 Add # 3 Add # 4 Add # 5 Add # 6 -------- --------- -------- -------- -------- -------- -------- -------panel01a 1 50 1223 .46 1337 .45 1539 .44 1289 .44 1392 .44 1272 .39 panel01a 11 60 1322 .45 1320 .41 1189 .41 1272 .40 1257 .40 1431 .34 panel01a 21 70 1226 .42 1322 .37 1323 .34 1154 .33 1335 .32 1257 .31 panel01a 31 80 1257 .41 1335 .41 1427 .39 1154 .39 1226 .37 1333 .36 panel01a 41 90 1333 .45 1335 .40 1271 .36 1472 .34 1325 .31 1173 .31 panel01a 51 100 1333 .44 1335 .41 1173 .38 1241 .36 1325 .34 1117 .31 panel01a 61 110 1169 .40 1333 .38 1220 .36 1341 .36 1335 .35 1241 .34 panel01a 71 120 1285 .40 1411 .39 1447 .37 1169 .37 1221 .36 1149 .34 panel01a 81 130 1183 .40 1334 .34 1263 .34 1408 .33 1205 .32 1169 .31 panel01a 91 140 1337 .38 1200 .38 1073 .37 1264 .35 1069 .35 1438 .34 panel01a 101 150 1337 .50 1396 .41 1060 .40 1195 .38 1390 .34 1255 .33 panel01a 111 160 1337 .46 1255 .39 1195 .36 1060 .35 1390 .35 1396 .35 panel01a 121 170 1337 .47 1195 .41 1396 .40 1329 .38 1096 .36 1350 .34 panel01a 131 180 1337 .46 1096 .41 1255 .37 1136 .35 1369 .34 1376 .33 panel01a 141 190 1255 .43 1337 .39 1096 .39 1261 .38 1195 .38 1150 .37 panel01a 151 200 1396 .36 1194 .34 1232 .32 1070 .32 1144 .32 1369 .32 panel01a 157 206 1130 .41 1194 .34 1070 .32 1144 .32 1329 .31 1369 .31 17 segments - - - - - - - - - - - - - - - - - - - - Number of segments Add No R_av Add No R_av Add No R_av Add No R_av Add No R_av Add +1337 12 .39 +1396 7 .35 +1149 5 .32 +1195 5 .37 +1255 5 .37 +1335 +1333 4 .41 +1232 3 .31 +1241 3 .33 +1154 3 .34 +1257 3 .37 +1200 +1169 3 .36 +1096 3 .39 +1362 3 .29 +1130 3 .34 +1376 3 .31 +1117 Corr Corr Add # 7 Add # 8 -------- -------1189 .36 1458 .33 1337 .33 1226 .33 1337 .31 1458 .30 1428 .35 1337 .33 1205 .31 1446 .31 1149 .30 1309 .30 1149 .34 1325 .33 1311 .33 1337 .32 1149 .31 1447 .31 1085 .33 1183 .33 1408 .33 1445 .32 1200 .34 1408 .33 1376 .32 1255 .31 1195 .32 1165 .32 1404 .37 1136 .35 1130 .31 1329 .29 1204 .30 1232 .30 - - - - - - - Corr Add # 9 -------1178 .31 1309 .32 1427 .30 1362 .32 1241 .30 1446 .30 1337 .32 1325 .32 1311 .30 1396 .32 1200 .30 1426 .31 1369 .30 1378 .32 1369 .33 1014 .27 1234 .29 - - - Corr Corr Add #10 Add #11 -------- -------1396 .30 1362 .28 1392 .32 1458 .31 1497 .29 1362 .28 1393 .32 1132 .32 1432 .30 1154 .29 1426 .29 1444 .29 1411 .32 1271 .31 1117 .31 1462 .31 1341 .30 1117 .29 1454 .31 1149 .31 1084 .28 1376 .28 1137 .28 1168 .27 1130 .29 1168 .29 1159 .32 1396 .32 1296 .32 1232 .32 1099 .27 1234 .27 1058 .28 1039 .28 - - - - - - No R_av Add No R_av Add No R_av 5 .38 +1369 5 .32 +1325 4 .33 3 .34 +1329 3 .33 +1226 3 .37 3 .31 +1408 3 .33 +1458 3 .31 Tree-Ring Dating of a Panel Painting Page 58 APPENDIX 5B COFECHA output that compares the tree rings from Panel 1 with the Baltic 1 reference chronology, this time testing 50year segments lagged 25 years. The highlighted values show the suggested date adjustment for the panel. The adjustment of “+1337” stands out and was found for 4 of the 7 50-year segments tested. This adjustment would put the beginning year at 1+1337 = A.D. 1338. This dating adjustment was clearly the best, ranking above the second best dating adjustment of +1255 which had an average r-value of 0.32 for 3 segments. PART 8: ADJUSTMENTS FOR UNDATED SERIES: 13:58 Fri 30 Jun 2006 Page 0 --------------------------------------------------------------------------------------------------------------------------------Time span 1156 1597 442 years, best matches for Listed in order from highest correlation 50-year segments lagged 25 years Counted Corr Corr Corr Corr Corr Corr Series Segment Add # 1 Add # 2 Add # 3 Add # 4 Add # 5 Add # 6 -------- --------- -------- -------- -------- -------- -------- -------panel01a 1 50 1223 .46 1337 .45 1539 .44 1289 .44 1392 .44 1272 .39 panel01a 26 75 1154 .44 1226 .42 1335 .40 1257 .39 1497 .39 1427 .39 panel01a 51 100 1333 .44 1335 .41 1173 .38 1241 .36 1325 .34 1117 .31 panel01a 76 125 1447 .42 1408 .40 1183 .35 1169 .34 1273 .34 1117 .34 panel01a 101 150 1337 .50 1396 .41 1060 .40 1195 .38 1390 .34 1255 .33 panel01a 126 175 1337 .52 1096 .37 1215 .37 1195 .37 1376 .36 1116 .36 panel01a 151 200 1396 .36 1194 .34 1232 .32 1070 .32 1144 .32 1369 .32 panel01a 157 206 Lag from prior segment 6 years - insufficient 7 segments - - - - - - - - - - - - - - - - - - - - Number of segments Add No R_av Add No R_av Add No R_av Add No R_av Add No R_av Add +1337 4 .44 +1255 3 .32 +1396 3 .36 Corr Add # 7 -------1189 .36 1333 .34 1149 .30 1311 .33 1408 .33 1350 .36 1130 .31 - - - No R_av Corr Add # 8 -------1458 .33 1337 .31 1309 .30 1334 .33 1445 .32 1255 .33 1329 .29 - - - Corr Add # 9 -------1178 .31 1255 .30 1446 .30 1472 .32 1200 .30 1369 .32 1014 .27 - Add No R_av - - - Corr Add #10 -------1396 .30 1362 .29 1426 .29 1205 .32 1084 .28 1159 .30 1099 .27 - - Add No R_av - Corr Add #11 -------1362 .28 1189 .28 1444 .29 1309 .32 1376 .28 1144 .30 1234 .27 - - - Tree-Ring Dating of a Panel Painting Page 59 APPENDIX 6A COFECHA output that compares the tree rings from Panel 3 with the Baltic 1 reference chronology. The highlighted values show the suggested date adjustment for the panel. No one adjustment of the top three listed stands out: (1) +1205, 7 segments, average r-value = 0.35, (2) +1223, 6 segments, average r-value = 0.40, and (3) +1392, 6 segments, average rvalue = 0.45. PART 8: ADJUSTMENTS FOR UNDATED SERIES: 15:45 Fri 30 Jun 2006 Page 0 -------------------------------------------------------------------------------------------------------------------------------Time span 1156 1597 442 years, best matches for Listed in order from highest correlation 50-year segments lagged 10 years Counted Corr Corr Corr Corr Corr Corr Series Segment Add # 1 Add # 2 Add # 3 Add # 4 Add # 5 Add # 6 -------- --------- -------- -------- -------- -------- -------- -------panel03a 1 50 1523 .46 1388 .40 1364 .38 1205 .36 1230 .34 1193 .32 panel03a 11 60 1230 .43 1205 .36 1366 .35 1372 .32 1505 .31 1425 .30 panel03a 21 70 1259 .42 1223 .38 1501 .35 1366 .33 1189 .31 1409 .29 panel03a 31 80 1259 .48 1223 .38 1207 .38 1501 .37 1366 .37 1132 .34 panel03a 41 90 1223 .46 1392 .45 1205 .39 1310 .38 1301 .38 1259 .34 panel03a 51 100 1392 .53 1223 .47 1316 .39 1205 .39 1270 .37 1310 .35 panel03a 61 110 1392 .55 1223 .42 1270 .38 1205 .37 1310 .37 1285 .33 panel03a 71 120 1392 .49 1286 .42 1310 .39 1150 .37 1270 .36 1316 .36 panel03a 81 130 1286 .41 1310 .40 1102 .40 1246 .39 1392 .35 1450 .35 panel03a 91 140 1246 .49 1102 .48 1113 .42 1233 .40 1078 .39 1154 .38 panel03a 101 150 1166 .55 1246 .42 1233 .38 1446 .38 1244 .37 1231 .36 panel03a 111 160 1166 .54 1233 .39 1350 .38 1338 .36 1231 .36 1183 .34 panel03a 121 170 1338 .42 1153 .42 1183 .38 1117 .35 1151 .34 1166 .34 panel03a 131 180 1185 .37 1183 .36 1117 .35 1305 .34 1138 .33 1119 .33 panel03a 141 190 1117 .34 1183 .33 1269 .32 1048 .32 1103 .31 1169 .30 panel03a 148 197 1103 .35 1183 .34 1319 .30 1138 .30 1169 .29 1382 .28 16 segments - - - - - - - - - - - - - - - - - - - - Number of segments Add No R_av Add No R_av Add No R_av Add No R_av Add No R_av Add +1205 7 .35 +1223 6 .40 +1392 6 .45 +1183 5 .35 +1259 5 .37 +1310 +1286 4 .35 +1316 4 .35 +1338 4 .35 +1169 3 .30 +1213 3 .31 +1117 +1113 3 .36 +1366 3 .35 +1166 3 .48 +1409 3 .31 Corr Corr Add # 7 Add # 8 -------- -------1418 .30 1372 .30 1200 .28 1277 .28 1205 .29 1207 .29 1392 .33 1302 .32 1316 .33 1270 .31 1157 .35 1409 .33 1362 .33 1191 .32 1222 .36 1242 .31 1154 .34 1081 .33 1181 .36 1081 .36 1113 .34 1259 .33 1308 .32 1113 .31 1119 .32 1273 .31 1216 .32 1213 .31 1338 .30 1213 .29 1079 .28 1237 .27 - - - - - - - Corr Add # 9 -------1499 .30 1259 .28 1286 .28 1341 .31 1115 .30 1288 .33 1316 .32 1362 .31 1078 .33 1244 .35 1338 .33 1286 .30 1269 .31 1169 .30 1189 .28 1320 .27 - - - Corr Corr Add #10 Add #11 -------- -------1402 .29 1324 .27 1171 .28 1223 .28 1460 .27 1171 .27 1409 .30 1304 .30 1363 .30 1191 .29 1191 .30 1285 .30 1157 .32 1192 .31 1284 .30 1246 .29 1205 .31 1137 .30 1350 .35 1408 .34 1213 .33 1435 .32 1153 .30 1355 .30 1216 .31 1097 .30 1151 .30 1415 .29 1305 .28 1185 .28 1292 .26 1118 .25 - - - - - - No R_av Add No R_av Add No R_av 5 .38 +1246 4 .40 +1270 4 .35 3 .34 +1191 3 .31 +1233 3 .39 Tree-Ring Dating of a Panel Painting Page 60 APPENDIX 6B COFECHA output that compares the tree rings from Panel 3 with Panel 1. The highlighted values show the suggested date adjustment for the panel. The adjustment of “+1392” clearly stands out and was found for 11 of the 16 50-year segments tested. This adjustment would put the beginning year at 1+1392 = A.D. 1393. This dating adjustment was the strongest of any adjustments found as shown by the “+1392 11 .43” at the bottom of the table. The next best adjustment was +1320, with 6 segments showing this adjustment, with an average r-value of 0.26. PART 8: ADJUSTMENTS FOR UNDATED SERIES: 15:54 Fri 30 Jun 2006 Page 0 -------------------------------------------------------------------------------------------------------------------------------Time span 1338 1543 206 years, best matches for Listed in order from highest correlation 50-year segments lagged 10 years Counted Corr Corr Corr Corr Corr Corr Series Segment Add # 1 Add # 2 Add # 3 Add # 4 Add # 5 Add # 6 -------- --------- -------- -------- -------- -------- -------- -------panel03a 1 50 1398 .45 1341 .44 1448 .35 1400 .35 1376 .33 1484 .32 panel03a 11 60 1341 .38 1376 .36 1439 .36 1368 .36 1423 .35 1392 .32 panel03a 21 70 1392 .39 1341 .34 1376 .33 1423 .33 1430 .31 1352 .31 panel03a 31 80 1392 .42 1341 .39 1375 .34 1376 .33 1439 .29 1430 .28 panel03a 41 90 1392 .55 1303 .40 1336 .35 1335 .33 1302 .28 1412 .27 panel03a 51 100 1392 .61 1301 .33 1303 .32 1319 .30 1359 .28 1302 .25 panel03a 61 110 1392 .59 1285 .39 1303 .33 1338 .28 1302 .28 1319 .27 panel03a 71 120 1392 .47 1338 .34 1359 .31 1398 .31 1322 .27 1319 .27 panel03a 81 130 1392 .47 1322 .34 1404 .33 1412 .32 1285 .31 1391 .30 panel03a 91 140 1261 .47 1322 .34 1392 .29 1250 .29 1299 .29 1346 .28 panel03a 101 150 1261 .42 1389 .30 1392 .30 1244 .29 1320 .28 1322 .28 panel03a 111 160 1231 .40 1261 .37 1320 .30 1355 .30 1265 .27 1229 .25 panel03a 121 170 1265 .44 1231 .33 1285 .32 1305 .31 1320 .30 1249 .29 panel03a 131 180 1214 .38 1265 .37 1249 .27 1231 .27 1340 .26 1216 .25 panel03a 141 190 1265 .43 1214 .39 1231 .33 1325 .32 1249 .28 1338 .27 panel03a 148 197 1255 .37 1288 .35 1214 .35 1218 .34 1325 .34 1247 .29 16 segments - - - - - - - - - - - - - - - - - - - - Number of segments Add No R_av Add No R_av Add No R_av Add No R_av Add No R_av Add +1392 11 .43 +1320 6 .26 +1338 6 .26 +1391 6 .27 +1319 5 .28 +1341 +1302 4 .26 +1265 4 .38 +1231 4 .33 +1322 4 .31 +1249 4 .27 +1355 +1301 4 .26 +1439 4 .30 +1446 4 .27 +1303 3 .35 +1285 3 .34 +1214 +1423 3 .33 +1325 3 .30 +1288 3 .27 Corr Corr Add # 7 Add # 8 -------- -------1423 .30 1392 .29 1398 .32 1400 .32 1439 .31 1391 .30 1391 .28 1446 .28 1391 .26 1401 .26 1391 .23 1335 .22 1301 .27 1428 .26 1412 .25 1302 .23 1261 .26 1298 .25 1355 .27 1389 .25 1355 .27 1354 .26 1338 .22 1244 .21 1355 .28 1281 .27 1224 .25 1338 .24 1224 .26 1255 .25 1198 .27 1197 .27 - - - - - - - Corr Add # 9 -------1452 .27 1375 .30 1319 .28 1319 .26 1444 .25 1412 .21 1359 .26 1288 .23 1359 .25 1391 .24 1338 .23 1316 .21 1301 .24 1325 .24 1228 .23 1216 .26 - - - Corr Corr Add #10 Add #11 -------- -------1446 .26 1439 .25 1446 .28 1425 .28 1467 .28 1446 .26 1352 .25 1409 .24 1362 .24 1353 .22 1320 .21 1336 .21 1412 .24 1342 .23 1301 .22 1368 .22 1345 .24 1320 .24 1354 .24 1320 .22 1305 .23 1359 .22 1298 .21 1354 .20 1248 .22 1361 .22 1361 .24 1284 .22 1341 .22 1288 .22 1249 .24 1345 .24 - - - - - - No R_av Add No R_av Add No R_av 5 .36 +1359 5 .26 +1412 5 .26 4 .28 +1376 4 .34 +1261 4 .38 3 .37 +1398 3 .36 +1354 3 .23
© Copyright 2026 Paperzz