Bruce Tai

From your observations you brought up many good points, Don Noon. First, minerals don't easily penetrate the wood as you said. We soaked think shavings of maple in 1% alum solution for 3 days and after rinsing only detected weak aluminum signals in the wood. When we measured significant minerals from deep inside the Strad neck samples, I am sure that some soaking was involved. Aging matters a lot, and it is possible that the treatments mainly assist better aging. To age to into a mature sound without losing the brilliance is very unusual when things are breaking down inside. Perhaps it can be considered as chemically assisted aging. And I am not just making this up. Supportive evidence for such processes will be presented in our next paper. The real surprise is that Amati, Stradivari, and Guarneri were experimenting with the mineral formulations for over 120 years based on our available samples. This will become evidence in our next paper. If it was only for preservation, I think the Amati instruments are preserved just fine and why change? I think Stradivari and del Gesu were really trying to achieve something different when they conducted chemical experiments. I suspect that their ability to sense meaningful acoustic changes in wood is still more informative than the routine acoustic tests that we know how to perform at this point.

The original research article with full details is here, open access for all: http://www.pnas.org/content/114/1/27.full Thanks to Chimei Museum for sponsoring the open access publication fee, so that everyone can read this freely.

There is no solid evidence to suggest that multi-step treatments with different mineral solutions have an effect on wood acoustics, but these possibilities need to be considered: 1. The removal of sap residues and wood extractives. Ancient Chinese books emphasize that it is critical to remove sap residues to get clear tones in guqin (7-string zither). The purpose of soaking wood in mineral solutions may be removing sap and extractives. 2. If the treatment was alkaline (wood lye or lime water), it may promote some hemicellulose decomposition. I don't think this happened in Cremona, but ancient Chinese did recommend this for tonewood processing. 3. Having some hygroscopic salts in the wood will help retain more moisture on cold winter days and prevent excessive shrinking, which may be acoustically beneficial in the long run due to better protection. 4. Metal ions may crosslink wood fibers to compensate the breakdown of hemicellulose over time, stabilizing the ultrastructure of wood. This may be acoustically beneficial in the long run. 5. Treatments may alter the surface properties of wood. Treated wood may be easier to smooth or react differently to stain/wood sealer (as seen in Brandmair's book). Wood surface treatment may also have some effects on acoustics. 6. Some of the minerals are clearly fungicides and pesticides. All six points listed above are of speculative nature. But are we really sure that none of these things matter in the long run?

In our first round of analyses we could not measure Cl, S, and Si. We are trying to resolve these issues in upcoming experiments. Can these wood treatments affect the tone immediately? Or can these treatments improve the tone 200 years later, as hemicellulose continues to break down and fiber molecules get rearranged due to vibrations? We don't have a clear answer yet. I don't know if the Cremonese makers only cared about wood preservation when they applied these chemicals. Our data says that they cared about it enough to make the chemical formulation gradually more complex over the span of 120 years. They were tinkering with some kind of goal in mind. What I do know is that there were some processes in Cremonese violin making which we knew nothing about, until Joseph Nagyvary made some bold hypotheses and went out to prove them. Nagyvary got some of it right, and we expanded his findings. There is still a lot that we don't understand about how Stradivari made his violins, both in terms of wood treatment and varnishing. Anyway I am not trying to promote any secret, and all of our findings will be published with high academic standards, accessible to the public. In ancient Chinese guqin literature, writers from the 12th century AD were emphasizing the use of wood over 500 years old. We are the first to truly characterize how wood ages in musical instruments over several centuries. But this is only the beginning.

Frankly, we cannot tell. It is impossible to do extensive sampling to understand the spatial distribution of elements. The concentration and the consistency would suggest to us that at least some of the minerals entered by soaking the wood.

French scientist Robert Palissy (1510-1590) wrote that “salt improves the voice of all musical instruments,” and he could have been referring to chemical treatments of tonewood for acoustic improvements. But he only wrote down this one sentence around , without further explanation. In his essay on “Du Sel Commun,” the last sentence says: Il aide à la voix de toutes choses animées, voire à toutes especes de metaux, et instruments de musique. [Google translator: It helps the voice of all living things, even all species of metals and musical instruments]. Palissy wrote this when Girolamo Amati (1551-1630) became a master maker. So it is possible that such ideas have already been floating around.

We have now analyzed many more samples (maple and spruce) than our initial results published in PNAS. We have often observed increases in these elements in Cremonese instruments, but each instrument shows different combinations: Boron - probably from borax used as fungicide Copper - probably from copper sulfate used as fungicide Zinc - probably from zinc sulfate used as fungicide Aluminum - probably from alum, we suspect that it was ammonium aluminum sulfate Sodium - probably from table salt Potassium - from wood lye? Calcium - from lime water or wood lye? Judging from the patterns in our data, we are now pretty sure that the violin makers bought normal wood and performed chemical treatments by themselves. In other words, Stradivari, Guarneri, and Amati all tried some experimentation on their own. Our oldest sample from Girolamo Amati is already chemically treated. I suspect that Andrea Amati already started this practice. We cannot say for sure what effects these chemical treatments bring, but they are significant enough for Cremonese makers to keep tinkering for over 100 years. Perhaps it was not a secret in Cremona that they played with wood treatment chemistry, but it certainly is a secret for the rest of us who were not in Cremona during its golden period.

I recently wrote an article for Strad magazine about our recent research on Stradivari's wood: https://www.thestrad.com/stradivaris-wood-investigating-the-cheminal-composition-of-the-masters-materials/6969.article It discusses the research paper (free download) we recently published in Proceedings of the National Academy of Sciences, USA, but in a less technical way, easier for non-scientists to understand. The relevance of our research to violin making is also discussed. All my violin publications can be downloaded from my lab web site: http://ntutailab.wixsite.com/home/strad (look for "Secrets in the Wood") Next year we will publish a paper comparing the chemical formulations used by Amati, Stradivari, and Guarneri. From our preliminary data, it is quite apparent that all three families have been treating their wood chemically, with some variations in their methodology. But why?

We did test for mercury and it was negative. Mercury chloride was a popular pesticide back then and it had been applied to wood, but not in this case.

Look at the numbers in Suppl. Table S10. Between the 1725 Strad neck and 1731 Strad cello, the increases in Ca, Na, and K are very large and very consistent. If these were introduced by surface treatments, we should have great variations in ion concentrations depending on depth and the presence of water-conducting vessels. This should be highly variable for very small pieces of wood. We took random small samples of only 10 mg for elemental analysis, and yet they showed such consistencies between different instruments. We did not even know if the sampled spot was sapwood or heartwood. I expected the mineral content to differ wildly from one Strad sample to the next. This is why I believe that the homogeneity was a result of soaking or sap displacement. I even speculate that the 1725 and 1731 instruments were built from the same batch of maple boards. The Al, B, Zn, and Cu may have been surface treatments, but who knows? The 1731 sample came from Rene Morel's repair, and the 1725 neck was dug out of an intact neck by us. They had entirely different histories over the 200 years. In our unpublished data for the next paper, the 1717 Strad violin is still similar to Strad neck and cello in terms of mineral content; the Amati and the Guarneri appear to be related to each other but quite different from the 3 Strads. Our sweat contains little Ca and K, and hence it could not just be sweat contamination. We really need to analyze chloride ions to see if the sodium was NaCl or natron. This can be done in our next paper. There is no chance to detect the carbonate in natron, or the any form of nitrate. Since Martin Swan knows so much about wood, I believe his point that many minerals could have been introduced into dried wood, by either soaking or surface application. He also made a great point about not adding potassium nitrates as it invites fungus. Nitrogen source is very important for life. Hence washing out the proteins and nitrogen-containing extractives is important for fungal and worm protection. The forest worker, the wood merchant, and the violin maker all had their chances to chemically manipulate the maple. Next, we really want to investigate the spruce and the pernambuco in famous French bows.

Shi Wen Long donated his life's fortune to build the Chi Mei Foundation and built the largest private museum in Asia. The building itself cost 80M USD. He has collected Western art, musical instruments, arms and armour, animal taxidermy and fossils. He collected only classical-style paintings and sculptures, insisting that old ladies should be able to walk in and appreciate the kind of art that he collects. Nothing too abstract or avant-garde. No one knows how much he spent on the collection, because he does not want to publicize it. Perhaps 200M USD? He initially insisted that the museum should be completely free of admission fee. But others told him that it is going to deplete his endowment fund so fast that he really needs to charge a small fee. Now it is 6 USD. He said that the endowment fund will eventually run out and by that point all the collection can be donated to the government. The company may last 100 years, the concrete building maybe 150 years, but he believes that art pieces should last for much longer than that. He also built a hospital and donated a lot of money into it. Shi Wen-Long is a world-class philanthropist and left very little fortune to his children, unlike most Asian tycoons. His generosity has gradually influenced other Taiwanese tycoons to become more engaged in philanthropy. He has little regard for money and fame. He thinks Taiwan needs to become more culturally enriched and he donate most of his fortunes to this simple cause.

We don't have enough instrument time to subject every modern sample to every test. As we collect more modern tonewood maples from more sources, we are starting to see what may be signs of over-baking. Perhaps it was a means of artificial aging. Some people tell us that mineral treatments have been applied to modern maples as well, although we have not seen any yet. Nagyvary found one such modern maple in his 2009 PloS One paper. As I mentioned before, I don't think treating dried wood would be the same as treating green wood in terms of removing extractives. Once wood is dried, there is little risk for insects and fungi. Intentional mineral treatment on dried wood may be more acoustic than biocidal, who knows? The wood is still full of mysteries. So what's next? To look into the spruce, of course.

Many people assume that when normality cannot be firmly established, one should use Mann-Whitney U test instead of t test. But simulation has actually shown that when the variance is unequal between the two groups, Mann-Whitney U test is not the better choice. Instead, the Welch's t-test, which assumes unequal variance, is generally better and most widely applicable. Here is the reference. The Welch's t test can be run in Excel simply by selecting the unequal variance option. According to simulations it can also work fairly well with skewed, non-normal distributions. It remains to be the statistical workhorse when we deal with complex samples with small sample size. Let's not forget that in medicine, even n=1 is sometimes extremely meaningful. When someone has a totally unique type of brain damage we can learn a lot from that person. Ever heard of patient H. M.? We learned so much about human brain function by studying him for 50 years. Of course statisticians can point out the many dangers of relying on a small sample size, but that has never stopped people from making meaningful discoveries (and mistakes) based on studies with small n.

So you are saying that t-tests are do not give correct P values with small sample numbers? Congratulations, you have just reinvented the field of statistics and deserve a Fields Medal. So what's your new statistical model that works so great? Please let us know and it will transform the entire biomedical science field. Or maybe you are calling the dozens of Nobel Laureates who had used t tests with low sample numbers liars or being clueless. Why don't you show us your great scientific reports and your statistics models? Student's t test, Welch's t test, and Mann-Whitney U test are fairly simple concepts in statistics. Recently there are some recommendations to use Welch's t-test directly when the underlying distributions are not clearly understood, which is almost always the case in complex problems. p< 0.05 is called significant, and p < 0.01 being highly significant, this is the language of biomedical scientists. The language does not matter so much because statistics is just about probabilities. For physicists who can collect many more data points they will demand much greater statistical power, in order to fit experimental data with their exact equations. We use statistics which is practical for the problem in front of us, that's all. As I posted earlier, we have done the TGA experiment without moisture control chambers, just ambient environment, and clearly Strads absorbed less moisture than modern maples. Realizing that this is a significant new discovery, we repeated it with moisture control measures and generated publishable data. Antique wood shavings are limited and so we are saving the rest for different experiments. Lots of fun things to do with these wood shavings: Raman spectroscopy, attenuated total reflection infrared spectroscopy, confocal-based two-photo spectral fluorescence imaging, second harmonic generation imaging, small angle x-ray scattering, two-dimensional 27Al nulear magnetic resonance. Not that I really understand these techniques, but I find experts who do to help out.

I suspect that one would not want to fully re-wet tonewood that has been air dried for ten years. It may take several years of additional air drying to re-establish absolute stability. But treating green wood right after cutting the tree may be quite different. Unfortunately none of us can tell forest workers to try these experiments on top quality tonewood. A book in 1840 said that putting wood in lime water is good because it removes proteins and sugars in the sap and makes the wood harder. In those days they sometimes treated green wood that way. Mineral treatment has been long obsolete in today's forestry. We can only experiment with dried wood, not green wood. This may be a big obstacle.