VIOLIN ADVISOR
Stewart Pollens, Director
"Violin Advisor assists in the acquisition of investment-grade violins.
Enjoy solid growth plus the privileges of arts patronage." - Financial Times
Published in Diagnostic and Imaging of Musical Instruments: Selected Proceedings of the International Workshop on Diagnostics and Preservation of Musical Instruments, DISMEC, Ravenna, 2010 (Nardini, 2016): 91-100.
Dendrochronology: Tool of Truth or Deception
by Stewart Pollens
Dendrochronology, which may be defined as the dating of the year-rings of wood, has only recently been employed in the dating of violins. In 1953, Lottermoser and Meyer attempted a relative dating of Italian stringed instruments by comparing the year-ring patterns of three violins, though actual dating was not achieved until the 1980s by Corona, Schweingruber, and Klein [1]. In 1994, when I curated The Violin Masterpieces of Guarneri del Gesù exhibition at the Metropolitan Museum of Art in New York, I invited one of the world’s leading dendrochronologists, Dr. Peter Klein of the Ordinariat für Holzbiologie of the University of Hamburg, to date the spruce tops of the twenty-five violins assembled for that exhibition. Among the conclusions gleaned from Klein’s findings are that Giuseppe Guarneri apparently did not use well-aged wood in making his instruments (the last datable ring is often just a few years earlier than the date on the violin’s label) and that he occasionally used mismatched pieces for the tops (whether this was by design, disinterest, or carelessness remains a matter of conjecture).
I met Dr. Klein many years before the Guarneri exhibition, for in the 1980s and 1990s, he was often invited to the Metropolitan Museum of Art to date panel paintings and other wooden objects. In dating musical instruments for the Department of Musical Instruments (including violins, viols, lutes, guitars, and harpsichords), I never had any reason to question his findings, which were always in agreement with my assessments regarding age and attribution. A few years after the Guarneri exhibition, I employed his services in dating the wood used to make the “Messiah” violin. Though he initially determined that the last datable year-ring of that instrument was 1738 (which postdated the violin's putative date of 1716 as well as Stradivari’s death in 1737), members of the violin trade maneuvered him into temporarily retracting his findings in exchange for an opportunity to re-measure the “Messiah’s” year-rings at the Ashmolean Museum. However, because of a dispute that developed during that session, he was unable to leave the Ashmolean with his measurements. Klein now declines to date violins, which is a great loss to those of us who formerly relied upon his expertise and objectivity. (Prior to publishing Klein’s results [2], I submitted the raw data to Dr. Peter Ian Kuniholm, a dendrochronologist on the faculty of Cornell University, who confirmed Dr. Klein’s dating.)
Following Klein’s controversial dating of the “Messiah,” Mr. John Topham, and subsequently, Dr. Henri Grissino-Mayer entered the fray and claimed to have found an earlier date that re-established the “Messiah’s” authenticity. Lacking Klein’s proprietary master chronology, Topham and Grissino-Mayer employed what they termed a “floating chronology,” which was derived from year-ring measurements taken from the “Archinto” and “Kux-Castelbarco” Stradivari violas. Unfortunately, their floating chronology only extended to 1685, and as dendrochronologists Angelo Mondino and Matteo Avalle later pointed out, this did not permit Topham and Grissino-Mayer to discover a more compelling statistical match further along in time [3]. In 2005 Mondino and Avalle asserted that the last year-ring of the “Messiah” grew in 1788, fifty-one years after the death of Antonio Stradivari, though in a more recent study, they claim to have discovered an even later date, 1844 [4].
Disturbed by the multitude of conflicting dates, I decided to delve more deeply and directly into this dating technique in the hope of learning more about it and, perhaps, determine why dendrochronologists have arrived at so many different dates for the “Messiah.” In 2007 I purchased the Synchro Search computer program and the accompanying Manual of Dendrochronology Applied to the Dating of Musical Instruments written by Angelo Mondino and Matteo Avalle (Cremona Books, 2005). Synchro Search is a graphing and statistical program for comparing year-ring data with established master spruce chronologies. The program disk includes a number of master chronologies that are in the public domain, including the Ötztal-Obergurgler, Siebenlist-Kerner Ötztaler, Obersaxon-Meierhof, Schweingruber, and Malcolm chronologies, which are among the chronologies that have been used in dating violins. In using the Synchro Search program, the graphed year-ring dimensions of the violin in question can be advanced with the click of a computer key against graphs of any of the master chronologies until a match is found. The program automatically computes statistical data, including Gleichläufigkeit (similarity or G index), Pearson’s r correlation coefficient (see below), Student’s t, mean sensitivity, serial correlation coefficient, overlapping index, and linear regression. The program can also be set so that the computer will locate the best matches, which is very convenient when master chronologies are hundreds of years in length.
In my investigation of this dating procedure, I did not wish to muddy the “Messiah” waters any further, so I decided to use Synchro Search in an attempt to establish the age of a large (17½”; 444 mm) viola, apparently Brescian, that the late Jacques Francais attributed to Gasparo da Salò (1540-1609). Making year-ring measurements of an old, heavily restored violin or viola top can be difficult. There may be “patches coming through” made with transplanted or later wood, painted year-rings, as well as dents, scratches, and heavily discolored or retouched varnish that obscure the grain. One must be aware of seams, wings, and grain direction. If even a single year-ring is missed, this can prevent the discovery of a match with the master chronology, so one must be very careful in measuring. When visiting museums to measure violins and other musical instruments, Klein used a little 7X loupe fitted with a metric scale that came in contact with the surface of the instrument. The advantage of this device is that it can be carried around in a pocket; the disadvantage is that it must be pushed along the surface of the instrument as one moves from year-ring to year-ring. When using a loupe, it is necessary to call out the measurements to a recording secretary, and this is another inconvenience. I devised a piece of equipment that is more accurate and convenient to use, though unfortunately, it is cumbersome. It consists of a low-power stereo microscope with one eyepiece fitted with a cross-hair reticule. Below the microscope, a violin/viola cradle is attached to a 250 mm rack-and-pinion linear translation stage that is linked via a brass angle bracket to a 12”-300 mm linear scale with SPC output. To use this device, I clamp the instrument (with its strings, bridge, and tailpiece removed) in the cradle and start by aligning the earliest year-ring (which is typically the ring at the edge of the lower bout of the instrument—though one should carefully check the year-ring shading to make certain of the orientation) with the cross hair in the microscope’s eyepiece. Using the rack-and-pinion of the translation stage, I then carefully align the next year-ring against the cross hair and press the button on the linear scale’s output device. This enters the measurement into a Microsoft Excel spreadsheet, which I have programmed to subtract the previous measurement from the last one entered, so that I wind up with a column of year-ring widths. Most violins have tops made with book-matched slabs of spruce, so one must stop at the center joint. These tabulated widths should then be multiplied by a factor of 100 so that the graphed data can be more readily compared to the master chronologies (dendrochronologists work in units of 1/100 mm, so 1 mm is expressed as 100) [5]. The column of figures must then be copied out of Excel as a text file (.txt file extension) so that the data can be compared against any of the master chronologies that are supplied with the Synchro Search program.
Returning to the Brescian viola: I compared my 74 measurements of year-rings from the treble side of the instrument to the following master chronologies: PCAB Giertz Obergurgler 1276-1974, LADE Siebenlist Obergurgler 1511-1974, PCAB Schweingruber Obersaxen 1537-1995, and Media 46° 1276-1994. I employed a statistical test termed Gleichläufigkeit (G index or similarity), initially with 60% agreement (which represents a degree of probability of 95%) between the year-rings of the viola and a master chronology, because this percentage of agreement is frequently considered a match in dendrochronological reports and literature[6]. The reader may be surprised by the number of “matches” that were found:
PCAB Giertz Obergurgler
Final Year Gleichläufigkeit
1350 62.3%
1417 61.6%
1445 65.7%
1463 65.7%
1500 61.6%
1522 62.3%
1540 60.2%
1554 64.3%
1572 65.7%
1583 63.0%
1600 62.3%
1618 61.6%
1732 63.6%
1756 62.3%
1758 65.7%
1839 63.6%
1863 62.3%
1901 65.0%
LADE Siebenlist Obergurgler
1511 62.3%
1665 65.0%
1680 62.3%
1704 63.0%
1732 63.0%
1750 60.9%
1792 60.2%
1810 60.9%
1826 63.0%
1826 65.7%
1834 65.7%
1949 65.0%
PCAB Schweingruber Obersaxen
1690 61.6%
1756 65.0%
1758 62.3%
1784 64.3%
1810 65.0%
1895 63.0%
1925 60.9%
1953 67.8%
1959 60.9%
Media 46°
1522 62.3%
1551 63.6%
1572 63.0%
1611 60.2%
1625 60.2%
1657 70.5%
1665 65.0%
1696 60.9%
1710 60.9%
1712 62.3%
1732 60.9%
1754 62.3%
1756 63.6%
Fifty-two matches were found at 60% Gleichläufigkeit or higher, and 13 matches were found at 65% Gleichläufigkeit or higher (representing a degree of probability of around 99% or better). It is also disconcerting that the PCAB Giertz Obergurgler, LADE Siebenlist Obergurgler, and PCAB Schweingruber Obersaxen master chronologies produced very few dates in common (1732 and 1756); the Media 46° chronology exhibited several concordances (1522, 1572, 1665, 1732, and 1756), but this would be expected as it is a composite derived from over thirty chronologies, including the other three that I used. The 1657 date (which has the highest Gleichläufigkeit of 70.5% and a probability of 99.9%) is of interest because violas of this large size were still being constructed at this time, though if correct, it would rule out Gasparo da Salo as the maker, as he died in 1609. However, I should point out that I subsequently discovered a date of 1805 with a considerably higher Gleichläufigkeit of 75.3% using the CEBR Schweingruber Ceader Valley (Cipro) chronology. I must admit that I did not initially seek a match with this master chronology because its synchronization dates extend only from 1675 through 1981, and my inclination was to search for dates using master chronologies that had earlier starting points. This only demonstrates that one may inadvertently overlook a master chronology because its inclusive years would appear to fall outside the date that one hopes or expects to find.
In general, my experiences with the Alpine master chronologies supplied with the Synchro Search program (which include the PCAB Siebenlist-Kerner Ötztaler and PCAB Gierz Obergurgler chronologies that were employed by Grissino-Mayer, et. al., in fixing a date to their floating chronology when attempting to date the “Messiah”) have not yielded overwhelmingly convincing, or unique matches with a number of Italian instruments that I have attempted to date [7]. The reason for this may be due to the fact that these master chronologies were constructed from wood samples taken from Alpine regions or altitudes that do not have precisely the same climatic conditions as the locale where the Cremonese and Brescian makers obtained their wood, which I believe was likely from the Italian Alps directly north of Brescia. Many of the master chronologies in the public domain are derived from wood found in the Ötztal and Obergurgl mountain ranges, which are close to Innsbruck, Austria. It is also possible that the Synchro Search program itself is at fault--some dendrochronologists have been critical of certain of its statistical operations, including its method of calculating Pearson’s r correlation coefficient and Student’s t [8]. However, determining Gleichläufigkeit is fairly straightforward, and it is often the only measurement cited in dendrochronological reports, including those of Peter Klein. Yet another reason for the failure to discover singular matches may be due to an underlying flaw in the technique itself—that it is simply not precise enough to pin irrefutable dates on relatively short sequences of year-rings.
In conclusion, I would strongly advise violin historians, authenticators, appraisers, and purchasers to be wary of the dendrochronology dates that are being published in auction catalogs, certificates, and scholarly journals. I would also suggest that dendrochronologists produce reports that include raw data and statistical details, and if they arrive at several possible dates for an instrument, they should report all of them; moreover, they must never rule out dates that they have been prompted to believe lie outside the probable period of manufacture.
_____________________________
1. Lottermoser W. and Meyer J., “Über die Möglichkeiten einer Dendrochronologie von altitalienischer Geigen,” Instrumentenbau-Zeitschrift 12 (1958): 295-297; Corona E., Ricerche dendrocronologiche su due violini del XVIII secolo,” L’Italia For. e Mont., XXXV, 3 (1980): 112-115; Schweingruber F. H., Tree Ring Basics and Application of Dendrochronology (Dordrecht, 1983); Klein P., Mehringer H., and Bauch J., “Tree-Ring Chronology of Spruce Wood and Its Application in the Dating of Stringed Instruments,” Icom Committee for Conservation 7th Triennial Meeting Copenhagen 10-14 September 1984 Preprints (1984): 84.1.69-84.1.72.
2. Pollens S., “Le Messie,” Journal of the Violin Society of America 16.1 (1999): 75-101.
3. Mondino A. and Avalle M., Manual of Dendrochronology Applied to the Dating of Musical Instruments (Cremona, 2005): 85-107.
4. Mondino and Avalle, Manual of Dendrochonology, pp. 85-87; Mondino and Avalle, New Dendrodating Procedure Exercises (Cremona, 2009): 91-92.
5. Below are the raw measurements taken from the viola multiplied by 100:
236,189,175,165,119,135,140,173,169,150,116,176,190,169,202,174,132,205,
193,205,212,204,200,216,215,220,188,198,156,161,205,188,167,157,172,193,
151,181,189,244,196,199,173,214,177,214,183,182,192,171,166,195,181,171,
205,168,185,175,127,172,151,170,162,170,225,237,260,274,257,262,245,116,
128,175.
6. Bernabei M., Bontadi J., and Rossi Rognoni G., “A dendrochronological investigation of stringed instruments from the collection of the Cherubini Conservatory in Florence, Italy,” Journal of Archaeological Science 37 (2010): 195.
7. Mondino and Avalle, Manual of Dendrochronology: 68.
8. I would like to thank Peter Ratcliff for pointing this out. Personal communication, 2010. I should add that the Synchro Search program was unfairly, and in some cases, fallaciously criticized in an article by Henri Grissino-Mayer, et. al. entitled "Adverse implications of misdating in dendrochronology: Addressing the re-dating of the 'Messiah' violin" in Dendrochronologia (2009). It should be pointed out that non-canonical variants of statistical tests are sometimes employed by dendrochronologists.