Dendrochronology Testing

[Ratcliffiddles]

Hi Dean

That's just about it!

Peter

[andrew weinstein]

Having a viewed a couple dendro reports at the last Christie's sale in NY, I was struck by how little they seemed to say with confidence, not at all what you talk about here with Dean. Don't know if that's how they all are, perhaps for legal safety, or if those tests were just not successful.

[Ratcliffiddles]

quote:

Originally posted by: andrew weinstein
Having a viewed a couple dendro reports at the last Christie's sale in NY, I was struck by how little they seemed to say with confidence, not at all what you talk about here with Dean. Don't know if that's how they all are, perhaps for legal safety, or if those tests were just not successful.

 

Hello Andrew,
In my experience, I will only issue a written report if I am fully confident about the date I am putting forward. A written report will normally contain, as well as a "most significantly matching" date, information on significantly matching reference chronologies and/or significantly matching instruments.
Occasionally, it will reveal highly significant matches, strongly suggesting same tree origin. You may think that this is far fetched, but in a particular period for an important Italian maker, that is precisely what we have observed.
Was one of the reports you mentioned for a violin attributed to Sanctus Serafin?, as if it is, I know that report(and that violin) well. What in particular do you find wasn't mentioned with confidence?

The dates from a dendro report are arrived at after careful evaluation of the statistical and graphical data obtained after analysis. If there are any doubts as to the accuracy of those dates, no report will be issued.
Peter

[andrew weinstein]

Hi Peter,

I'm not sure which two violins, I want to say the Cappa and the Strad. It may well be that I'm just not that familiar with the terms commonly used in these reports, but highly significant match sounds alot more sure than what I recall. My laymans interpretation of what I thought I was reading was " it looks a bit like this but could be alot of other things as well"

[Ratcliffiddles]

quote:

Originally posted by: andrew weinstein
Having a viewed a couple dendro reports at the last Christie's sale in NY, I was struck by how little they seemed to say with confidence, not at all what you talk about here with Dean. Don't know if that's how they all are, perhaps for legal safety, or if those tests were just not successful.

 

The legal question is also interesting.
I'd be interested in views on the scenario:
Somebody comes in with his beautiful Da Salo viola, and purely as a matter of interest and for the sake of contributing to my database, asks me to perform a dendro analysis, which, of course, I do...but there is a problem, let's say my results are inequivocally 1812, matching with loads of reference. Which means of course, Da Salo could not have made it. Compounding the problem, the instrument also carries a good certificate, not great, but good, dated 1920. Would you want to know? What would you do if you owned the viola? just ignore the findings? have the test performed by somebody else? ( who, for the sake of argument finds the same date). or maybe say that the tests results must be wrong.

In the same way as when an expert has to tell somebody that in fact, their precious instrument ain't quite what they thought, we can find ourselves in slightly awkward situations.
Peter

[Wolfjk]

Hi,

There is an interesting articl about dendrochronology in the July issue of the Strad by John Topham. It displays two Strads: Le Brun of 1712 and the Park of 1711. It is obvious from the photos that they are from the same tree.

There are also 5 violins from 1711 to 1713 that show very similar growth pattern. It looks as if they are from the same tree. However I cant see any similarities to the reference curve of any violins.

How do we know for sure that the growth pattern of 17th century violin can't be similar to a reference sequence from the 19th century?

Cheers Wolfjk

[Lyndon Taylor]

As I understand it that would require 50 yrs of the same weather, a

statistically very unlikely event, I think you would agree, as

Dendrochronology is based on recognizing dry and very wet winters

and the differences in growth etc, sincerely lyndon j  Taylor

[tonewoods]

"Would you want to know? What would you do if you owned the viola? just ignore the findings? have the test performed by somebody else?"

You could donate it to the Ashmolean to be displayed under poor lighting, and take the tax benefits....

[Ratcliffiddles]

quote:

---

Originally posted by:

Wolfjk

How do we know for sure that the growth pattern of 17th century violin can't be similar to a reference sequence from the 19th century?

Cheers Wolfjk

---

Hi Woljk,

If that were the case, most 17th instruments, when measured, would flash up with a 19th century date as well as the correct one. This occurrence doesn't happen. What we are looking for when we analise the findings, is a repeating date, that is to say several matches with different chronologies, from different source, ie reference and instruments. It is possible to get a good statistical match for two separate sets of data with a spurious date, but the likelyhood of that date repeating, is so remote that it just doesn't happen. Furthermore, a graphical comparison will almost certainly reveal that in fact, the sequence are not similar enough to have a temporal relationship. The "accepted" number of rings that have to overlap, in order to avoid possible spurious matches is 50, and most violins diplay more than that. A Klotz, with a one piece front, displayed 314 growth rings!, so no spurious date there!! and the earliest ring present on that violin was growing in the year 1424, that' a loooong time ago...

Peter

[tonewoods]

"A Klotz, with a one piece front, displayed 314 growth rings!"

38 GPI is not all that rare in nature...

I see 50 GPI across an 8.25" run of spruce quite often, in all species....

I think what is rare is finding a maker--then or now--who likes to build with stuff that is that tightly grained....

I'd like to learn more about dendo, Peter.

Is there a good book you could recommend to get me started??

I'm ambivalent about dendro, to be honest, but that attitude is based on personal, rather than scientific, experience....

For instance, many years ago I took xeroxes of tops from a Strad iconography and was able to line up grainings so that they matched up perfectly, thus denoting the use of the same tree...

It was quite amazing really. I moved one xerox slightly sideways in relationship to the other xerox until bingo!...they matched up perfectly...

But I've also walked many-a-clearcut and have seen trees with 8 GPI growing 10 feet away from trees with a 50 GPI grain count...

This is the norm, not some unusual occurance...

Yeah, I know that dendro charts the relationship of graining in spruce samples, but it just seems that there are so many variables other than weather that go into creating that relationship...

So-ooo, love to learn more....

[Ratcliffiddles]

quote:

---

Originally posted by:

tonewoods

I see 50 GPI across an 8.25" run of spruce quite often, in all species....

---

Hi Tonewoods

If my calculations are correct 50GPI across 8.25" makes 412 rings, and you say you see that quite often in pruce? Next time you do, let me know, because I haven't come across that many rings in spruce, nor that 50 GPI so far. Maybe you are talking about right across a board full quarter sawn board, which would include the central part of the tree? Or maybe you are talking about a different type of spruce than Picea Abies, which is normally used in violins.

Im my experience, I have not seen anything like the differences of GPI from nearby spruce trees that you have. You are correct in saying that there are many variables but the data collected throughout the Alps, in Austria, Italy, & Switzerland, from sites separated, in some instances, by two or more hundred miles, correlates beautifully, especially at higher altitude. Of course, when the data is collected at the bottom of a valley, where microclimates, exposure etc are different to higher plains, it's a different story.

The ultimate tree ring page, website created by Henry Grissino Mayer is a very informative site, if you are interested.

Peter

[Wolfjk]

quote:

---

Originally posted by:

Ratcliffiddles

quote:

---

Originally posted by:

Wolfjk

How do we know for sure that the growth pattern of 17th century violin can't be similar to a reference sequence from the 19th century?

Cheers Wolfjk

---

Hi Woljk,

If that were the case, most 17th instruments, when measured, would flash up with a 19th century date as well as the correct one. This occurrence doesn't happen. What we are looking for when we analise the findings, is a repeating date, that is to say several matches with different chronologies, from different source, ie reference and instruments. It is possible to get a good statistical match for two separate sets of data with a spurious date, but the likelyhood of that date repeating, is so remote that it just doesn't happen. Furthermore, a graphical comparison will almost certainly reveal that in fact, the sequence are not similar enough to have a temporal relationship. The "accepted" number of rings that have to overlap, in order to avoid possible spurious matches is 50, and most violins diplay more than that. A Klotz, with a one piece front, displayed 314 growth rings!, so no spurious date there!! and the earliest ring present on that violin was growing in the year 1424, that' a loooong time ago...

Peter

---

Hi Peter,

Thanks for the explanation.

Is there a requirement for the 50 matches to come in sequences of more than three or four, as single years matches could easyly occur by chance?

Cheers Wolfjk

[Ratcliffiddles]

Hi Wolfjk,

I am not sure I understand your question, the fifty years is the minimum number of years that two sequences have to overlap, but of course very often the overlap is far greater than that, often covering the whole length of he series, in which case a good correlation is more significant.

Peter

[Wolfjk]

quote:

---

Originally posted by:

Ratcliffiddles

Hi Wolfjk,

I am not sure I understand your question, the fifty years is the minimum number of years that two sequences have to overlap, but of course very often the overlap is far greater than that, often covering the whole length of he series, in which case a good correlation is more significant.

Peter

---

Hi Peter,

Sorry I was not clear in my post.

I was trying to say that in the minimum of 50 years overlap with the reference curve there should be squences more than three or four years of perfect matches. One or two years of growth rings matching can occur anywhere, from any time.

If you take the strad there are very good examles. On page 25, graph 1, there are two violins matched to a reference curve. I can't see any similarities between the violins and the reference curve, the computer of JT spots some similarities - especially between curve 1 and the reference curve.

On page 29 graph two, there are 5 violins represented and one can see clearly that their curves match very well, they are probably from the same tree.

On page 27, there are photos of two violin fronts. Even from the colour of the photos one can see they came from the same tree. Perhaps colour matching rings and areas would be a better method of identification.

My experience is - working with alarge volume of timber - that it would be very difficult to match two different trees from the same forrest.

Cheers Wolfjk

[Ratcliffiddles]

quote:

Originally posted by: Wolfjk

I was trying to say that in the minimum of 50 years overlap with the reference curve there should be squences more than three or four years of perfect matches. One or two years of growth rings matching can occur anywhere, from any time.

 

Hi Wolfjk,
There is no such thing as a pefect match in dendro, perfect matching would entail exactly matching growth ring widths over a given period. A series of measurements taken from a sample at a given axis, and repeated one inch below, will give different measurements(not different dates) simply because of the nature of cell growth in the wood not being uniform. Apart from the (t) value, and graphical comparison, there are other mathematical formulae applied to the measurements, giving extra information as to the relationship between two or more samples, f.e. the percentage of growth rings increasing in size, decreasing in size or remaining static, expressed by another letter (G).
Regarding John's(for whom I have the greatest respect) article in the Strad, I will have a look, but it is likely that editing the graphs to fit comfortably on a page may not have given the desired effect. If the graphs were superimposed and maybe the scale of the y axis increased a bit, the relationship would be more apparent. But of course I left my copy in the workshop...
Peter

[Ratcliffiddles]

A

[Ratcliffiddles]

Something weird happened earlier today.

A member of this forum sent me a couple of pictures of one of his violins, just showing the widest parts of each lower bouts, wihout any distinguishing features such as purfling / sound holes/ shape etc.. to see if a date could be found for the wood. I measured the growth rings and analysed the data. Extraordinarily, I got extremely strong correlation with a couple of my files, normally only achieved with wood from the same tree. I immediately did a graph comparison and, without the shadow of a doubt, concluded that in fact, this was the same violin!

I saw this instrument about four years ago and remembered some of the features which would identify it and checked with the forum member, who confirmed that they were present on his instrument and also that the instrument was acquired fairly recently.

What was likelyhood of something like that happening???( not expressed in(t)value, please...)

Peter

[Ratcliffiddles]

quote:

---

Originally posted by:

Wolfjk

If you take the strad there are very good examles. On page 25, graph 1, there are two violins matched to a reference curve. I can't see any similarities between the violins and the reference curve, the computer of JT spots some similarities - especially between curve 1 and the reference curve.

My experience is - working with a large volume of timber - that it would be very difficult to match two different trees from the same forrest.

Cheers Wolfjk

---

Hi Wolfjk,

I had a look at John Topham's graphs on page 25 of the Strad, and to my eye, there are lots of similarities between curve 1 and ref., and as mentioned in a previous post, those similarities would be much more apparent were the two graphs superimposed, but for the purposes of illustration, to have several graphs superimposed on each other can get very messy. With curve 2 and ref, the similarities are much less obvious graphically, and the similarity of intensity of the simultaneous increases or decreases of the ring widths of the two sequence is much less pronounced.

The subsequent results of the two analysies are also substantially different in significance, but still correctly indicate the relative position of each curve against the reference.

On your other point, the data collected from various areas and forests that constitute the reference chronologies is formed out of dozens of sets of measurements combined together, and as a result it is normally easy to correlate two trees from the same forest, as long as they lived at the same time for at least an overlapping period of time, that is not to say that if two samples of wood match each other, even fairly well, that they necessarily come from the same forest.

Peter

[mcarufe]

The data base has to be enormous. Climate cycles within climate cycles. There are solar variables, earth sun eccentricities it really makes your head spin. The smaller 40 year cycles are the ones that make me wonder how reliable dendro dating will ever be for more vague samples from possible locations. (small climate cycles, questionable samples, questionable locations, how often can these variables be satisfied)If climate was an old record player the music would repeat about every forty years than some one gets frustrated bangs on the side of the record player, takes it further into the song to have it repeat again. I am not saying it is not a valid science, obviously it is, but the public perception of science and how it filters into marketable ideas and items gets driven buy market forces and...lalalalala Moreover for the scientist on board, the doubt defines the model as much as the supporting data. Failure to leave out brings us men of science in the same class as the not very credible groups that lived before us, that have all but nice names descibing them.

There can not be a "eureka moment"(for the not so science geeky, this goes back to the tale of Archemedes figuring out his law while bathing at a public bath. I believe Eureka means "I found it" in Greek) when building models, even though the model gets better with time, most people can't seperate in their mind that it is a possible model and not a fact in and of itself. Nor is it as specific and DNA testing or finger prints unless a certain set of variables are for certain. That is where it finds it's limitations.IMHO.

Mike Carufe

[Ratcliffiddles]

quote:

Originally posted by: mcarufe
The data base has to be enormous. Climate cycles within climate cycles. There are solar variables, earth sun eccentricities it really makes your head spin. The smaller 40 year cycles are the ones that make me wonder how reliable dendro dating will ever be for more vague samples from possible locations. (small climate cycles, questionable samples, questionable locations, how often can these variables be satisfied)If climate was an old record player the music would repeat about every forty years than some one gets frustrated bangs on the side of the record player, takes it further into the song to have it repeat again. I am not saying it is not a valid science, obviously it is, but the public perception of science and how it filters into marketable ideas and items gets driven buy market forces and...lalalalala Moreover for the scientist on board, the doubt defines the model as much as the supporting data. Failure to leave out brings us men of science in the same class as the not very credible groups that lived before us, that have all but nice names descibing them.

most people can't seperate in their mind that it is a possible model and not a fact in and of itself. Nor is it as specific and DNA testing or finger prints unless a certain set of variables are for certain. That is where it finds it's limitations.IMHO.

Mike Carufe

 

Hi Mcarufe
You are right, the database has to be enormous, and that is why I measure every single instrument that comes through my hands, of any nationality and period, and on the whole instruments fit into dendro
categories, with wood origin appearing to be fairly specific according to period and country, categories often merging or overlapping.
So the research carried out is an ongoing and never ending process which should, as time goes by, reveal ever more information. Of course there will always be dubious violin fronts, with wild and eneven grain, which will never reveal their origin,and trying to attribute dates to such wood may well be futile. There are, on the other hand, many instruments which are obviously related stylistically,
with a great many growth rings, which repeatedly match each other so well which dates are as far as
I am concerned 100% acurate. When you say "most people can't separate in their mind that it is
a possible model and not a fact in and of itself", I fear you are right, but this attitude applies not only in relation to dendrochronology, but in the case of violins, to expert and not so expert opinions, which have no more absolute validity than anything else, and "opinions" too, by nature, have their limitations and are based on observations, memory, an ability to recall acurately those memories, and a constant aquaintance with instruments. Yet how many experts of the past have been "proved" wrong by modern ones.
It wouldn't be fun if we knew everything!

Dendrochronology is another tool, it is not and does not pretend to be absolute. Yes, it can upset
a traditional attribution, or contradict a modern one, so I also see why it is not in everybody's interest to endorse it as an accepted method, yet it works...
Fingerprints I believe, are also analysed using statistical and probabilities software, yet is used daily as infallible evidence.
Peter

[Fellow]

We are sure that from inspecting a section of tree (tree rings) is one year to the next (one ring).

From their ring patterns is much less reliable informations. Confidence of the conclusion may

not be within useful range.

[fiddlecollector]

Well , im impressed by Peters ability to recognise patterns from photographs in his database. The violin in question was mine , a 17th century English violin,although i only sent him a 1" high strip across the widest areas of both lower bouts, the photos came up very quickly in his database.I suppose i was as surprised as he was because we live at opposite ends of the country. Not getting a date was a bit disappointing but i suppose theres still plenty of huge gaps in the database.

Thanks Peter for trying !

p.s when you say you record every violin that comes through your hands,is there similar people recording all those millions of factory instruments,going back 200 years or more? Its a job i wouldnt really want to take on.

[Ratcliffiddles]

quote:

---

Originally posted by:

fiddlecollector

Well , im impressed by Peters ability to recognise patterns from photographs in his database. The violin in question was mine , a 17th century English violin,although i only sent him a 1" high strip across the widest areas of both lower bouts, the photos came up very quickly in his database.I suppose i was as surprised as he was because we live at opposite ends of the country. Not getting a date was a bit disappointing but i suppose theres still plenty of huge gaps in the database.

Thanks Peter for trying !

p.s when you say you record every violin that comes through your hands,is there similar people recording all those millions of factory instruments,going back 200 years or more? Its a job i wouldnt really want to take on.

---

Indeed, there are many gaps, and of course, the very early stuff, due to the relative small number of instruments and reference available, has more gaps than later data. But if you are patient, in time....

"Factory" instrument were made generally out of the proper materials and they are useful indicators, as often factories used wood from one source during a particular period, f.e. lots of Mirecourt production from the early 20th century comes from the Jura mountains but no evidence of Jura wood used much before 1900. Similarly, Klotz workshops which can be described factories, used often one piece fronts with tight grain with loads of useful dendro information.

So it's all useful data to try to get the whole picture.

Peter

[Wolfjk]

Hi Peter Radcliff,

Thanks for your post.

I would like to learn how to co-ordinate the gaphs on page 25, I can see some odd years of similar growths but I could not say that there are certain relations.Here is a portion of the graph. I would be thankful if you could poin out where the critical similarities occur.

In one of your earlier posts you also stated that it is not the actual ring width but the similarityin growth rat that is important. I believe that every tree responds differently to external conditions and the configurations of roots and branches on individual trees, plus the genetic differences would make it difficult to do any comparissons.

Cheers Wolfjk

[Ratcliffiddles]

quote:

Originally posted by: Wolfjk
Hi Peter Radcliff,

Thanks for your post.

I would like to learn how to co-ordinate the gaphs on page 25, I can see some odd years of similar growths but I could not say that there are certain relations.Here is a portion of the graph. I would be thankful if you could poin out where the critical similarities occur.


In one of your earlier posts you also stated that it is not the actual ring width but the similarityin growth rat that is important. I believe that every tree responds differently to external conditions and the configurations of roots and branches on individual trees, plus the genetic differences would make it difficult to do any comparissons.

Cheers Wolfjk

 

Hi Woljk,
In the recent past, the graphs would be on semi transparent paper, superimposed and visually analysed and compared on large illuminated tables. With modern techniques and computer graphics, all that can be done on screen. As I haven't got the original data for the two curves and don't know what the reference John T used in the graph, I can't produce graphics to illustrate the correlation better. In his graph, I think he simply wanted to illustrate two different degrees of matching criteria. Computer programs can "squash" or expand the scale of any given data to accentuate the similarities or differences. But if the (t) value is in the region of 12(I can't remember exactly), the probability of the assesment being incorrect is completely negligeable.

There is evidence from thousands of samples taken from hundreds of locations, analysed and proven to correlate perfectly well, with varying growth ring widths. I would love to be able to show you that trees that, in your view "have responded differently to external conditions" and therefore appear different in their growth pattern, in fact over a long enough period, will match. Is your belief based on quantitative and comparative data, or visual observation?
What people seem to forget, is that the wood mostly used for violins is highly selected, it is not generally compression wood, it is normally straight grained, quarter sawn, no knots, no resin pockets. It is therefore ideal material because it is as far as nature allowed, free from defects. You may think it strange, but the tight even(apparently) growth often yealds the best dendrochronology results.
Peter