Chemical distinctions between Stradivari’s maple and modern tonewood

[GeorgeH]

@jezzupe:

 

The problem is that there is no objective way to measure sound quality of a violin. And then there is no way to tell if a violin sound improves or degrades simply with age and/or playing. And no two individual violins sound identical.

 

What we do know is that some modern violins compare and compete favorably with some Strads in blind tests. I suspect that Strads sounded their best when they were new, and that great violins made since Strad's time will sound great for centuries as well.

[tamiya]

btw what else was spruce & maple woods used for 300yrs ago in Italy?

Was it just a common building material of the day?

(Sorry I know nothing about European carpentry. What woodwork I've done is always been with local timber, mine so far being rubberwood & pine.)

What we do know is that some modern violins compare and compete favorably with some Strads in blind tests. I suspect that Strads sounded their best when they were new, and that great violins made since Strad's time will sound great for centuries as well.

Will they? Or are they more like wine (another passing hobby) vs cellar age.

Wines made to "drink now" taste fine just 2-5yrs out of the vineyard whereas the old masters who make wine to be cellared for 20-50yrs, uncork one of those too young and it's rough enough to rip your palette. But lay it down for a few decades and it becomes sublime... whereas a drink-now red forgotten for 20yrs will just become tasteless.

[Bruce Tai]

Blind faith in poorly-designed blind tests will make one blind to facts and truths.

Sorry for the self-promotion, but we have already published our view on this topic, also in PNAS. 

 

http://www.pnas.org/content/111/27/E2778.extract

 

Full text:

Fritz et al. report the inability of violinists to distinguish Stradivari violins from modern ones with regard to timbre (1), but we disagree with their data interpretation.

Fritz et al. fail to grasp a fundamental concept in psychoacoustics research, which is the role of memory in perception. To briefly explain this concept, let us imagine two identical monitors adjusted to different color temperatures (subtle but noticeable) while showing the same picture. If they are shown side by side, it would be easy to tell the color difference. If two monitors are sequentially presented with a 30-min interval, it would be extremely difficult to make the distinction simply because the visual memory has decayed. If the subject fails to distinguish the color temperature of two monitors sequentially presented, it would be erroneous to conclude that the color difference cannot be perceived. It would only prove that visual memory decays too fast for meaningful perceptual comparisons to be made, and a better testing procedure should be devised instead.

During blind playing tests, the violinist has to rely on short-term auditory memory to make perceptual comparisons. Numerous studies have shown that short-term memory for timbre only lasts for seconds, and the memory buffer is probably updated every 10–20 s (2). However, 20 s is too short for the player to switch violins and play the same passage, and therefore it is unreasonable to expect the player to make meaningful timbre comparisons between different instruments.

Over the last two centuries, many soloists and collectors believe, correctly or incorrectly, that the most distinctive feature of Stradivari violins is timbre quality. To investigate this possibility, it would be more reasonable to record various violins and let subjects perform blind listening tests during which fast A-B switching is allowed. An alternative is to analyze recordings objectively by computer software to look for statistically significant differences. In fact, we have already done the latter, and found that Stradivari violins produced significantly higher formant frequencies compared with a selection of Old Italian violins and new violins of professional quality (3). The magnitude of the observed variation in violins was similar to the formant variation between speakers of different sex.

Based on the reported decay kinetics of timbre memory (<30 s), the blind test adopted by Fritz et al. is expected to consistently produce null results between different violin groups, if their timbre differences are not particularly striking. Unless future research can demonstrate the ability of players or listeners to retain timbre memory long enough to make meaningful comparisons, blind tests that do not allow for fast switching would appear to be inadequate for assessing timbre qualities associated with Stradivari violins. The fact that one or two instruments seemed to be frequently preferred in blind playing tests may reflect other acoustic properties (loudness, number of wolf notes, dynamic response, etc.) or playability factors (ease of vibrato, transient response, bridge/neck angle, etc.), some of which may induce longer lasting memory than others, and different memory traces may even confound one another.

[GeorgeH]

@Bruce Tai #46,

 

It is actually shocking to me how full this post is of unsupportable statements, speculation, myths, exaggerations, and outright falsehoods. 

 

Violin makers since Stradivari's time have been making violins every bit the equal of or better than his best without "mineral-treating" their wood. Blind testing proves this. You apparently know this, but are trying to refute it because it does not fit your confirmation bias. There are no undiscovered "secrets" of Stradivari violins that makes them uniquely better than other fine violins of equal tonal quality. There is no singular "Stradivari Tone" common to all his instruments; they are all different. 

 

Your paper shows that you have "discovered" that old violins have old wood and that some may have been treated with chemicals that act as insecticides or fungicides. Nothing more. The provenance, sampling methods, and too-small number of samples are totally insufficient to draw any general scientific conclusions about Stradivari violins. The idea that your tests on a few tiny samples of questionable provenance indicate some kind of "Stradivari maple" is risible. You can't even relate the results of your tests to the quality of the tone of the individual instruments from which they were taken! 

 

It is remarkable to me that you can't see the irrelevancy of comparing old maple with new maple. Just like modern makers, Stradivari did not make his instruments from 300 year-old maple. And his instruments were highly-prized when they were new. And, as I have written before, there is no proof that violins get better as they get older. I imagine that Stradivari is laughing in his grave at the idea that the insecticides were some kind of secret to the tone of his violins.

 

Clearly, you have a self-interest in perpetuating the popular myth of the "secrets" of Stradivari. Too bad you confuse myths with facts, and speculation with science. 

 

George

 

P.S. There are no secrets to making Coke, either. It is all about the brand. Think about it.

 

P.S.S. The blind test was repeated in 2014 under different conditions. Strads lost again. 

http://www.thestrad.com/blind-tested-soloists-unable-to-tell-stradivarius-violins-from-modern-instruments/

[Don Noon]

... there is no objective way to measure sound quality of a violin...   I suspect that Strads sounded their best when they were new...

 

There is no way to objectively measure whether the sound is good, bad, best, or worst... but sound can be objectively measured.

 

And in the objective measurements I've seen of Strads, I do believe I see a general pattern that is different from what modern (and even other Cremonese) instruments will produce... perhaps similar to Bruce's conclusions in post #53.  My sample size is very small, but the Strad response curves do look different to me. I don't see this as any magical power, but the fact that no Cremonese maker made the arching, form, and graduations exactly like Strad, and modern makers attempting exact duplications do not have the 300 years of aging and the attendant changes in wood properties.  

 

I'd much rather get rid of all the hype of the "sound of Strad" and the competition to sound better, as that is pointless, and get on with looking objectively at what the differences are in the tone, and what might be contributing to those differences.   While I don't agree with all of the conclusions of the paper, I think that the issues of hemicellulose decay and EMC reduction are important pieces of information.

 

Personally, I didn't care much for the sound of 3 of the 4 Strads I have played, and one I thought was very good.  That's just my taste.  So I can see how blind testing for preferences won't show Strads always on top.  But who cares.

[Violadamore]

The problem is that there is no objective way to measure sound quality of a violin.

Naaah, we've got plenty of ways to measure it.  What we need is a baseline to work from.  I am once more going to propose that we begin work on an International Standard Violin.  Anyone wishing to fund grants for this purpose is requested to contact the Accademia Segreti di Maestronet.

[GeorgeH]

There is no way to objectively measure whether the sound is good, bad, best, or worst... but sound can be objectively measured.

 

And in the objective measurements I've seen of Strads, I do believe I see a general pattern that is different from what modern (and even other Cremonese) instruments will produce... perhaps similar to Bruce's conclusions in post #53.  My sample size is very small, but the Strad response curves do look different to me. I don't see this as any magical power, but the fact that no Cremonese maker made the arching, form, and graduations exactly like Strad, and modern makers attempting exact duplications do not have the 300 years of aging and the attendant changes in wood properties.  

 

I'd much rather get rid of all the hype of the "sound of Strad" and the competition to sound better, as that is pointless, and get on with looking objectively at what the differences are in the tone, and what might be contributing to those differences.   While I don't agree with all of the conclusions of the paper, I think that the issues of hemicellulose decay and EMC reduction are important pieces of information.

 

Personally, I didn't care much for the sound of 3 of the 4 Strads I have played, and one I thought was very good.  That's just my taste.  So I can see how blind testing for preferences won't show Strads always on top.  But who cares.

 

We can measure sound, for sure. But like you said, we can't "objectively measure whether the sound is good, bad, best, or worst."

 

​I think to do this we would need a machine to play each violin identically, record the sound spectra, and then ask a pool of listeners in the room with the machine to rank the violin. Then we could try to match the sound spectra with rankings, and see if we can get a statistically significant correlation between the sound spectra and listener ranking. We may also want to measure the hearing frequency range of the listeners to see how that correlates with ranking. 

 

This would need to be done with hundreds of violins.

 

The key is making the machine that can play a violin uniformly and well enough to make the ranking! Need funding!

[David Burgess]

Fritz et al. report the inability of violinists to distinguish Stradivari violins from modern ones with regard to timbre (1), but we disagree with their data interpretation.

Bruce, I think it's important to point out that the Fritz et al paper does not report any inability to discern timbre differences between instruments, including differences between Strads and moderns. Rather, what they found was a failure, overall, to correctly categorize the instruments as either new or old, at better than chance levels. That doesn't mean they couldn't or didn't hear differences. Nor was there any claim that there were no differences. In fact large differences between some Strads and moderns were quite conspicuous in the scoring.

[Violadamore]

We can measure sound, for sure. But like you said, we can't "objectively measure whether the sound is good, bad, best, or worst."

The key is making the machine that can play a violin uniformly and well enough to make the ranking! Need funding!

No, first you need the uniform violin for calibration.  Fund us first!

[David Burgess]

 

The key is making the machine that can play a violin uniformly and well enough to make the ranking! Need funding!

George, one important component which would be missing from that scenario is the way the violin reacts and provides tactile and sonic feedback to the player. One big criticism of "listening only" tests, which often found that moderns were competitive with Strads, was that this element was missing. So the latest tests have included player impressions. As it turned out, this didn't change the outcomes much, but at least the change addressed one major area of criticism.

 

The same goes for Bruce Tai's recommendation that violins be evaluated by listening to recordings. Critics would have that picked to pieces in a heartbeat.

[Violadamore]

George, one important component which would be missing from that scenario is the way the violin reacts and provides tactile and sonic feedback to the player. One big criticism of "listening only" tests, which often found that moderns were competitive with Strads, was that this element was missing. So the latest tests have included player impressions. As it turned out, this didn't change the outcomes much, but at least the change addressed one major area of criticism.

Oh, my, maybe we need a standard player as well!  Maybe listeners too. Genetic engineering and all that good stuff.  This could be the first great research boondoggle of the 21st. Century, it has everything!  Silicon nitride and iridium violins, robotics, mutant test animals, Hell, I'll even bet there's a frightening sci-fi bestseller in there somewhere complete with a series of appalling movies............

[jezzupe]

George, one important component which would be missing from that scenario is the way the violin reacts and provides tactile and sonic feedback to the player. One big criticism of "listening only" tests, which often found that moderns were competitive with Strads, was that this element was missing. So the latest tests have included player impressions. As it turned out, this didn't change the outcomes much, but at least the change addressed one major area of criticism.

yup, play-ability is hugeginormugus when judging violins. You can have the fastest car on the track, but it don't mean diddly squat if you can't drive it 

[jezzupe]

Oh, my, maybe we need a standard player as well!  Maybe listeners too. Genetic engineering and all that good stuff.  This could be the first great research boondoggle of the 21st. Century, it has everything!  Silicon nitride and iridium violins, robotics, mutant test animals, Hell, I'll even bet there's a frightening sci-fi bestseller in there somewhere complete with a series of appalling movies............

Robotavarius strikes back! part deux!

[Violadamore]

Robotavarius strikes back! part deux!

Naaaah.  The Intonator..........

[Conor Russell]

I'm afraid that I haven't had time to read the paper fully (nor the thread), so forgive me if my questions have already been gone over.

Was there any comparison made with wood samples from violins of a similar age but from different makers or even regions?

It would seem to me that a piece of maple made into a violin 250 years ago has had a very eventful life. It's been cut, carved, polished, varnishes, played, sweated on, breathed on, repaired, and on and on. It might be important to compare it with a similarly experienced sample before any conclusions could be drawn.

[David Burgess]

 

There was one reference cited (31) claiming reduction in damping due to vibration for short periods; I don't have that one, and it goes against what I have found to be true... but I don't have the details of what wood, what vibration levels, etc.

 

 

Don, this may be the referenced item at the top of this search page:

https://scholar.google.com/scholar?q=Sobue+Okayasu+1992&as_sdt=0&lr=&hl=en

 

If so, it offers very little in the way of detail, but includes some interesting observations, such as the changes from vibration reversing, or going back in the direction of un-vibrated wood, if the wood is exposed to high moisture levels.

 

Their hypothetical explanation for the change from vibration is that when the wood initially dries, some components contract with the moisture loss, and others (crystaline cellulose) do not, leaving internal stresses on a cellular level, which can be relaxed with vibration.

 

If you have a chance to take a  look at the paper, and can offer an opinion on whether or not it's significant, and whether the changes they found are significant, that would be great. My initial reaction was that the paper was a lot less than I had hoped for.

 

Edit: On second thought, that may not be the right paper.

Maybe this?

https://www.jstage.jst.go.jp/article/jsms1963/41/461/41_461_164/_pdf

[GeorgeH]

George, one important component which would be missing from that scenario is the way the violin reacts and provides tactile and sonic feedback to the player. One big criticism of "listening only" tests, which often found that moderns were competitive with Strads, was that this element was missing. So the latest tests have included player impressions. As it turned out, this didn't change the outcomes much, but at least the change addressed one major area of criticism.

I think that some of this information might be available from our violin playing robot thought experiment ("The Intonator" thanks, violadamore!). The robot would need to have the capability to be programmed to use the bow reproducibly in the ways a violinist does so that many different bow strokes could be tested. You could vary attack, pressure, speed, angle to the string, contact point, etc.  simultaneously, both for single bow strokes and within a single bow stroke. Ideally you would also be able to measure the drag of the bow at a given speed. 

 

The sonic fingerprint for each violin would then show how the violin responds to different bow strokes. For example, perhaps some violins would show volume building quickly versus a more slower developing volume on a different violin for the identical bow stroke. This might give us some insight into responsiveness. 

 

But this thought experiment would be primarily concerned with tone output, not tactile feedback. 

 

I think that it would also be important to check the hearing of the listener evaluators. We know that hearing deteriorates with age, and it deteriorates differently for males and females. So older people might prefer a different tone than a younger person.

 

I also think that it would be interesting to first study violin players using the same types of motion-tracking technology used to study athletes and that actors use to generate 3D animations. Data from such a study would be useful to understand how bows and violins are used in realtime by virtuosos, and could then be modeled on the machine. I should note that most violin players use their whole bodys to play, and that aspect would be very difficult to mimic.

 

Finally, the data would need to be analyzed using data mining techniques that can discover statistically-significant correlations between multiple simultaneous outputs and inputs from massive amounts of seemingly random data.

[Don Noon]

Don, this may be the referenced item at the top of this search page:

https://scholar.google.com/scholar?q=Sobue+Okayasu+1992&as_sdt=0&lr=&hl=en

 

Edit: On second thought, that may not be the right paper.

Maybe this?

https://www.jstage.jst.go.jp/article/jsms1963/41/461/41_461_164/_pdf

 

In the first paper, the change in damping is quite small, only 2% and well below what I can reliably test with my simple methods.

In the second paper (looking only at the spruce sample), the damping decreased by 5% or so... still a relatively small change, as I have seen damping change by far larger amounts just due to humidity changes.  I doubt that a damping change of that small magnitude would be detectable by a player, much less by a listener.

 

On thing that wasn't mentioned (or maybe it was mentioned in the Japanese text, but I can't read it) is that if you're continuously vibrating something, and there's damping, that means that some energy is being transferred to the test sample.  That energy has to be turned into heat, and perhaps such heating, even though small, could change things.

 

For comparison:  On sample of processed wood that I brought with me to VSA this year had 22% lower damping after processing (allowing sufficient time for stabilizing, measured in 50 - 60% RH conditions).  That same sample, measured at VSA in the much dryer conditions, lost ANOTHER 25% in damping, or -41% total compared to its natural state.  By the way, this wood was air-dried naturally for over 25 years before I did anything to it.

 

So my overall take is that the papers show very small changes, and uncertain sources of the change.  Nothing significant in my view.

[Carl Stross]

They would sound "modern".  If you look around, you can find some Strads that come up short (whatever that means).  He was admittedly a good maker, but no supernatural powers.

 

 

Oh my gawd, hope you've got the weapons in place to protect your home and family from rabid retributive Stradivarites!

 

We'll let this one slide...

[David Burgess]

 

On thing that wasn't mentioned (or maybe it was mentioned in the Japanese text, but I can't read it) is that if you're continuously vibrating something, and there's damping, that means that some energy is being transferred to the test sample.  That energy has to be turned into heat, and perhaps such heating, even though small, could change things.

 

For comparison:  On sample of processed wood that I brought with me to VSA this year had 22% lower damping after processing (allowing sufficient time for stabilizing, measured in 50 - 60% RH conditions).  That same sample, measured at VSA in the much dryer conditions, lost ANOTHER 25% in damping, or -41% total compared to its natural state.  By the way, this wood was air-dried naturally for over 25 years before I did anything to it.

 

So my overall take is that the papers show very small changes, and uncertain sources of the change.  Nothing significant in my view.

Interesting, thanks. So I suppose it would be possible that a slight temperature increase of the wood, during the period of  vibration, lowered the water content of the wood, reducing damping, at least temporarily. I guess I don't see anything in the papers to indicate that they checked for this by looking for any changes in moisture content, (weighing the wood before and after?) or ruled it out as an explanation..

[Carl Stross]

 

There is no objective "Stradivari sound." Every violin has a unique sound; no two individual violins sound identical, including Strads, Guarneris, and the lowliest factory fiddle. Trying to extrapolate these analytical results from (maybe) a few Strads and Guarneris to explain a nonexistent quality of all Strads and Guarneris is absurd.

 

 

 

How do YOU know that ?

Empirical evidence has shown that played violins tend to loosen up and mature in an audible manner over time.

 

 

Prove it. 

 

He doesn't need to. You need to prove it doesn't happen.

 

But there is much more to argue about in their conclusions that this data has any bearing on the sound quality of Stradivari’s violins or that it has anything to offer to modern violin makers who continue to build violins with tone quality equal to Stradivari's best using contemporary wood. Just like Stradivarius himself used when he was a “modern maker.”

 

 

Sure. Then, argue THAT.

 

 

1. The problem is that there is no objective way to measure sound quality of a violin. And then there is no way to tell if a violin sound improves or degrades simply with age and/or playing. And no two individual violins sound identical.

 

2. What we do know is that some modern violins compare and compete favorably with some Strads in blind tests. 

 

1. What is plain clear is that you don't know of any and can't tell.

2. Yup. But that doesn't matter. Never did, never will. That's a "feature" built in the art, craft and science of TOP violin playing. Art is packed with similar "workings" which outsiders are unable to decode. On the one hand the perspective is wrong and on the other, qualifying violins at that level is an inside job. It's for the Elect and the Enlightened. I'm being serious.

[skelly]

Hello Mr. Stross. Are you one of the Elect and Enlightened? You sound like Pylorius.

Take care.

[Violadamore]

How do YOU know that ?

 

 

 

He doesn't need to. You need to prove it doesn't happen.

 

 

Sure. Then, argue THAT.

 

 

1. What is plain clear is that you don't know of any and can't tell.

2. Yup. But that doesn't matter. Never did, never will. That's a "feature" built in the art, craft and science of TOP violin playing. Art is packed with similar "workings" which outsiders are unable to decode. On the one hand the perspective is wrong and on the other, qualifying violins at that level is an inside job. It's for the Elect and the Enlightened. I'm being serious.

Hiyas, Carl.  Was wondering when you'd turn up.  Keep up the good work.

 

 

Hello Mr. Stross. Are you one of the Elect and Enlightened? You sound like Pylorius.

Take care.

Welcome to our little discussion.  How long have you been suicidal?

 

 

So my overall take is that the papers show very small changes, and uncertain sources of the change.  Nothing significant in my view.

One problem I perceive about evaluating this stuff is that 2% here and 5% there, and so on, and you're talking serious changes before you know it, particularly when the entire effect under discussion probably amounts to (much, but let's be conservative) less than 10% performance difference overall between a good Strad in the hands of a superb player, and anything else well made in the hands of the same player.  While I would not group myself in the Elect and Enlightened in this area, even I can tell the difference between my various violins and can compensate for their vagaries while playing them.  Very tiny changes can be noticed by the player.

[Don Noon]

You need to prove it doesn't happen.

Oh, come on... you know that's impossible, like proving there are no ghosts.  

 

One problem I perceive about evaluating this stuff is that 2% here and 5% there, and so on, and you're talking serious changes before you know it, particularly when the entire effect under discussion probably amounts to (much, but let's be conservative) less than 10% performance difference overall between a good Strad in the hands of a superb player, and anything else well made in the hands of the same player.

 

I view the problem a little differently.  

There is ALWAYS a huge difference between any two violins, just due to the nature of the large number of resonances, their frequencies, amplitudes, spacing, dips between them, etc.  For example, here is the difference in the impact spectra between two of my recent violins:

 

The things that most strongly determine these features are in the form, arching, adjustments, wood properties, and graduations.  However, the effects of each variable are different.  Arching I believe has the strongest effect on what the instrument does in terms of where all the bumps and dips show up, i.e. tone.  Wood properties, on the other hand, are more of an overall effect, moving everything slightly up or down.

 

Look again at the chart:  differences of 3 dB and more are all over the place, and quite a few 6 dB and more.  This is not unusual.  A difference of even 10% in damping would have at most a theoretical overall effect of a tiny fraction of a dB.  And differences of far, far more than that happen all the time in the same instrument with changes in humidity and temperature.  So, although I wouldn't want to give up any advantage I think there might be, no matter how small, this is why I don't think the papers on damping show anything significant.

 

It is also why Bruce's paper is more interesting, showing ~40% EMC reduction which in turn implies (to me, anyway) an equivalent reduction in damping.  That might show up on the radar, but still not up to a dB sized effect.  And that, in turn, is why I'm really focused on arching lately.

[GeorgeH]

It is also why Bruce's paper is more interesting, showing ~40% EMC reduction which in turn implies (to me, anyway) an equivalent reduction in damping.  That might show up on the radar, but still not up to a dB sized effect.  And that, in turn, is why I'm really focused on arching lately.

 

Don, The tests were performed on two sample sets: one with an n=4 (modern maple) and the other with an n=5 (old maple). These n's aren't nearly large enough to have statistical power to make any valid claims about the differences between the two populations, and assigning a p<0.01and claiming this is "highly significant" is crazy. Also interesting to note that sample M4 was not included in this data set, and the reason why is not discussed. Either M4 should have been included or the reason it is not included should have been discussed, otherwise it is suspicious.

 

Also, the paper states that the the samples were "Equilibrated for 48 hours in a sealed chamber with saturated NaBr solution. The room controlled between 24−27 oC, which corresponded to 57−58% relative humidity inside the chamber," but no data is offered to show that the samples had actually reached equilibrium saturation (constant weight) before being tested.

 

The paper further states that "Reduced hygroscopicity in historical maples may be attributed to the decomposition of hemicellulose, which is more hygroscopic than cellulose and lignin." Besides the fact that the assertion that "reduced hygroscopicity in historical maples" is insupportable due to tiny sample sizes, if the difference WERE real, it could also be due to differences in the surface areas of the samples or surface contaminates which could cause wide variances in the rates of absorption, and the samples reaching equilibrium saturation at different rates. 

 

So these tests really tell us nothing.