Dendochronology and the Messiah Stradivarius

At the recently concluded 2001 meeting of the Violin Society of America, Professor Henri D. Grissino-Mayer presented the results of his dendrochronological analysis of the Messiah Strad. For those who could not attend he has kindly given us this summation of his science and his findings. Professor Grissino-Mayer introduces himself:

I’m an Assistant Professor in the Department of Geography at the University of Tennessee. My research interests lie in two general areas, both of which include the use of dendrochronological (tree-ring dating) methods. I am interested in the reconstruction of past climate from tree-ring data, especially in the Southwestern and Southeastern United States. I’ve led the development of numerous millennial-length tree-ring chronologies throughout Arizona, New Mexico, and Colorado that were used to analyze trends in both short-term (decadal) and long-term (century-scale) climate during the past 2000 years. Secondly, I use tree-ring based fire history information from the fire-scar record to evaluate the record of past fires in an area. I spent seven years working under Dr. Thomas W. Swetnam and his colleagues at the Laboratory of Tree-Ring Research (The University of Arizona) analyzing fire history in numerous locations throughout the Southwest. During this time, I developed FHX2, software that allowed the processing of large volumes of tree-ring based fire history data.

I have given over 80 presentations and invited talks at professional regional, national, and international meetings, as well as to numerous civic groups and schools. I have also appeared on television, radio, and documentary videos numerous times. I belong to the following professional organizations: Association of American Geographers, Association of Pacific Coast Geographers, International Association of Wildland Fire, Southwestern Association of Naturalists, The Tree-Ring Society, Georgia Academy of Science, Arizona-Nevada Academy of Science. I have refereed numerous articles for such journals as The American Midland Naturalist, Arctic, Antarctic, and Alpine Research, Canadian Journal of Forest Research, Climatic Change, Ecological Applications, Ecology, The Holocene, International Journal of Wildland Fire, Journal of Vegetation Science, Northwest Science, Palaios, Radiocarbon, Wetlands, The Holocene, Plant Ecology, Quaternary Research, Dendrochronologia, and Tree-Ring Bulletin. I have also refereed proposals for such agencies as the National Geographic Society, the National Oceanic and Atmospheric Administration, the National Science Foundation, and the National Park Service.

I have received the following awards: Alton A. Lindsey Award for Research and Resource Stewardship, Who’s Who in the World, Who’s Who in the West, USDA Forest Service Certificate of Appreciation, Discover Magazine Top 50 Science Stories, Phi Kappa Phi Honor Society, Gamma Theta Upsilon Honor Society.

Describe in general terms the process of dating antiquities by dendrochronological means…

Any artifact made from a type of wood that originated from the temperate latitudes has the possibility of being dated provided it has enough tree rings, the species is known to produce annual rings, and the rings can be clearly seen on the object.

If these criteria are met, we should be able to obtain enough measurements of the tree-ring widths to be able to date the tree rings on objects manufactured of wood, even those whose date of manufacture is unknown. Why is this? Because tree growth is variable from year to year, i.e. climate is different in subsequent years. This variability in climate (and therefore in tree growth) causes rings to have “patterns” of wide and narrow rings that are unique over time, much like a “fingerprint”. If we can obtain a reference chronology from very long-lived, still living trees, then we stand a chance at dating wooden objects by matching the overlapping patterns of wide and narrow rings. It’s simply a matter of sliding the graph of the rings from the unknown object against the graph of the rings of a reference chronology. We can even extend our living tree reference chronology using wood from houses, pueblos, cathedrals, and many other types of structures where old wood can be found.

I should point out that dendrochronology is a very visual science. In other words, we can do all we want with the statistics of tree rings, but the truth is in the graphical dating of these patterns. Nothing should be considered dated by dendrochronology unless both the statistics and the graphics indicate without a doubt that the tree rings of the object do indeed date against the reference chronology.

How did you go about creating a chronological sequence by which to date fine string instruments as measured against climatic events visible in alpine softwoods?

Climate imparts a certain influence in tree growth that results in patterns of wide and narrow rings (see above). If a tree, obtained from a particular climate zone, is used to make an object, such as a violin, then we should be able to use a reference chronology from that climate zone to date the violin. The problem is that we’re not exactly sure where the trees used to make the Cremona instruments grew. We have no reference chronology for the “Forest of the Violins” in Trentino, Italy, where the wood likely originated. So we use the nearest reference chronologies from the high Alps, with many of these trees representing treeline conditions where temperature is especially sensitive for tree growth.

To produce a reference chronology many trees are collected from various individual sites, and their rings are measured to the .001 mm. Indices of tree growth are developed to remove the age-related trend in tree growth (a simple mathematical transformation), and these indices are then averaged together, across all trees, to create a site reference chronology, like those used by John Topham. However, we also found (as did John,) that the Messiah did not date against the two reference chronologies very well, so we developed what we feel is a superior, master reference chronology representing 16 sites in the high Alps across five countries.

What are your findings about the date of the wood used for the top of the Messie violin?

First, we dated the Archinto (label date of 1696) against our newly developed master reference chronology for the high Alps, then we successfully dated the Kux/Castlebarco (label date of ca. 1720), also using the reference chronology, as well as that from the Archinto. The Archinto dated from 1526-1686 while the Kux/Castlebarco dated to 1557-1684. We used these two chronologies to date the Messiah. The dating was conclusive both graphically and statistically, against both instruments. The Messiah has tree rings that date from 1577-1687, dates contemporary with Stradivari. I point out, though, that I can only assign dates to tree rings; I do not assign makers to violins. Sadly, some people refuse to believe the dendrochronological dating of the instrument because, as they are correct in stating, some statistical error does exist. However, this statistical error is extremely minute and inconsequential; i.e., the level of confidence (the inverse of margin of error) is sky high that these are the dates for the Messiah.

Do localized micro-climates affect overall dendrochronology data?

What is your determination of a source for the spruce employed by the maker of the Messie violin?

Have you been able to correlate this source with any other instruments?

I can answer all these questions together. Yes, microclimates vary considerably in the Alps, which is why we could not date the Messiah directly against any master reference chronology. I hypothesize that the wood for the Messiah did not come from anywhere in the high Alps in close proximity to the 16 sites used in the master chronology. Otherwise, the Messiah would have dated without a problem. However, in my opinion, the wood for the Messiah came from a location that was lower in elevation, perhaps on the lower slopes of the Alps, where a different climate regime exists. What we needed was a reference chronology for between the high Alps and the location from where the wood for the Messiah came. The trees used for the Archinto and Kux/Castlebarco, I feel, occupied such a central location, and recorded some properties of the climate regime at high elevations, as well as some properties of the climate regime at lower locations from where the Messiah’s wood came. This was a critical finding of our study. To help future studies, we really need a chronology from lower elevations in the Italian Alps, preferably from the Forest of the Violins itself!

Have you dated any other string instruments using the dendrochronological means used for the Messie violin?

In addition to those three: the Messiah, the Archinto, the Kux/Castlebarco, we also dated the ex-Cipriani Potter, a 1683 Strad at the Ashmolean Museum. We found the rings dated to 1594-1673, which is consistent with the label date of 1683.

What range of degree of statistical certainty did you achieve with your data and measurements?

Very high. The chances of obtaining some of the t-values we obtained simply by chance are about 1 in 10,000, i.e. we have a level of confidence of 99.99% with a margin of statistical error of 0.01%.

What is the commonly accepted range for statistical margin for error in dendrochronology?

As in most studies, we become “interested” in a date if it has a level of confidence of 95%, or a margin of error of 5%. This would equal a t-value of 2.0. (Note, for comparison: most medical studies have a margin of error of 10%; i.e., the medicine you may be taking was based on a study with a margin of error of 10%.) However, we raise the bar here: we really don’t become “very interested” until a t-value reaches 3.5, which has a margin of error of about .01% as mentioned above. All our reported t-values were above 3.5. We even had one t-value of 7.0 and the level of confidence here is astronomical.

Do you have doubt about your findings on the Messie violin?

None whatsoever.