Nathan,

I'm behind the curve on ring curvature radius. What is the outcome you are looking for, size of tree based on archaeologically recovered charcoal? Please explain.

Hi James

Yes and no for the size of tree. More the size of the part of tree the fragment came from. Essentially I have an enormous amount of Quercus  coming out of a site in central Anatolia. Quercus use seems to peak in the Late Bronze Age and then decline a little although in all periods it is either the dominant taxon or close to it. What I wanted to investigate is whether there are any indications of woodland management such as coppicing and pollarding or are they just ripping out oak trees left and right! So, in that vein I started analysing presence of pith, bark, pith+bark, hyphae, tylose percentage, and ring curvature (would love to do ring width but far to many of my fragments are too small) to see any patterns that may be occurring. The results at this stage show peaks in larger rings and more tylose during the peak in oak use. Later on moderately curved rings and pith peak. So yes, I would like to know whether there in a statistically significant change in ring curvature size indicating that smaller or bigger bits of wood were used more or less.

What I really wanted to know is what range of ring curvature size can be grouped together (binned) (0 mm to 50 mm = small or strongly curved for example) for meaningful analysis. If I keep all my measurements as raw measures almost all are unique and therefore are not as meaningful to demonstrate patterns unless plotted in a very messy and enormous curve!

Kind Regards

Nathan

I have a idea, though not sure it will help and it needs a sample of the species to get a baseline. If one recorded the ring number and curvature in a large enough sample of Quercus cerris, you could establish a mean average of size versus ring curvature. I think one could then come up with a logarithmic formula. It would then be grouping the resulting values into usable groups. Also, have you thought about the relationship of ray spacing to age/size?

Great idea James and one that may well be possible. I will talk to the Forestry ministry in Turkey to see whether they already have this and if not whether I can obtain this data in the field when I am next there. I suspect that a number of samples from different environments and locations would be needed.

As far as the ray spacing is concerned. I had not thought of this. I think its a good idea for the multiseriate rays but am unsure about the uniseriate rays as these are sometimes incomplete. Hmm, more investigation needed here and possibly the makings of another paper here sometime in the future if you are interested!

Nathan,

Let me correct my suggestion. While it would be good to collect the ring number while you're at it, what needs to be formulated is the distance from center versus the ring curvature. For the same species living in the same environment this distance to curvature relationship should be a constant, but there will be some variability. Thus a tight interval is necessary in order to contain the data. Additionally, tree growth is determinate as to environment. Thus an oak in a hot, dry environment will have less annular cellular growth between the rings than a tree growing in a mild, wet environment. Thus it would be important to measure the thickness of the cellular growth in the different climates in order to be able to devise a correcting factor.

Hi Nathan,

in our lab, we group our measurements into diameter classes whose interpretation is archaeologically meaningfull. For example, we use 0-20 mm for twigs, 20-50 and 50-100 mm and 100-200 for medium wood, and up to 200 mm for big wood as we cannot measure precisely the radius of curvature for wood bigger than 20 cm diameter. Theses classes were choosen according to our archaeological questions and the physionomie of the species (for example, for Pinus we subdivised the classe 100-200 mm into 100-140 and 140-200 mm).  These classes also take into account the margin of errors of the measurements (all the ligneous rays do not converge toward the pith, especially the last ones which were formed.

Since we tested tools based on tree-ring curvature, especially on oak, we do not use anymore this parameter.

Tell me if my english is not intelligible ;-)

Best,

Hi Alexa

Thanks a lot for this, just what I was looking for!

I actually broke my diameter classes up into the following:

0-50 mm for small and twig wood;

50-180 mm for medium wood and;

180 mm and over as large/big wood (noting that beyond 200 mm it is virtually impossible to calculate radius on archaeological fragments).

This is fairly similar to what you have with the small and twig class slightly bigger and the large class slightly smaller. Do you think that the classes I have already calculated are appropriate for trying to ascertain whether there are changes occurring in the calibre of Quercus wood collected or should I change it to the classes you use? Also, do you have a citation or reference for these classes that I could use or is pers. comm. OK in this case? 

My main concern is making my data set comparable to others so adhering a standard of some sort would be great for future meta-analysis.

Thank you again, and your English is perfect. I wish I could learn French though to read all the publications!

Kind Regards

Nathan 

Hi Nathan,

There is a mistake in the second and third diameter classes I send you :

0-20mm; 20-40mm; 40-70 mm 70-100 mm, etc.

 you can use it without problem if it is adequate for your.  A paper is in preparation.

To establish the diameter classes, I work in a first time on raw data. I work now for a long time on the subject andfor me  these diameter classes are a good compromise. Your diameter classes are appropriate to your question . I just wonder in what extent  the low number of classes (3) can bias or too much simplify the interpretation?

If you need some translation of french paper on the subject, , tell me I cantranslate the information you need.

Best,

Alexa

Hi Alexa

I was wondering the same thing in regards to the low number of classes I have been using and whether that causes bias or oversimplification.  Can I clarify with you the class sizes you use?

0-20 mm for twig/small?

20-40, 40-70, and 70-100 for medium/moderate?

100-200, and 200 mm onward for large?

I just want to make sure I use the same as your lab for comparability. My other concern is that the more classes you have the lower the n for each class which reduces the statistical meaningfulness of the data, this is especially the case for my samples which are dominated by 2-4 mm specimens.

Also, did you get my message in regards to tylose percentages? When I first starting analysing the data back in 2011 I used >50% and 85% and

the diameter classes and their interpetation (small, medium et large)  are right .

concerning the number of fragments, we analyse at least 80 fragments, the best is about 100 . With this number, the absence of a class or the over-representation is representative of your sample. 

About tyloses, the paper will be published in Quaternary International. The values of the thresholds (>85% and

Nathan,

I am not too familiar with coppicing and pollarding. What I see is farming, where fast growth species such as popular (Populus sp.) or maple (Acer sp.) are grown to ~25 cm and then harvested for firewood. With staggered planting, the supply is constant. I suppose with coppicing and pollarding you would not necessarily have to worry about distinguishing between young growth trees and limbs, since young trees would be uncommon?

Hi Alexa

OK thanks for that. I re-binned my data with the classes you used and it barely changed the pattern the only difference was that I moved 20-40 mm into my category for small. I have aggregated my data by phase and in each phase I have over 1000 identifiable Quercus fragments. Usually of these about 30% (sometimes less) were large enough to observe ring curvature. So by phase I have plenty of data for it to be considered robust I think. By individual sample within each phase that is different of course as some samples only contain 50-60 oak fragments in total. Hense why the data is aggregated by phase. In total over 4000 measurements were taken. 

Sorry for the typo on the 35 vessels, I meant 15 per ring. Is it possible for you to provide me with a reference for this? It would then be appropriately cited as either "In Press" or "in this issue". The same applies for the ring curvature classes if possible. I want to ensure your are properly cited in my submission to Quaternary International. 

Thanks once again, it is always a pleasure to chat with you and you have been most helpful. Lets push more more knowledge sharing as it's the best way to strengthen the discipline!

Kind regards Nathan

Hi James

I get where you are coming from with the "farming" of the fast growing trees. Interestingly we see a lot of coppiced Populus. Quercus, Fraxinus, Fagus, Acer etc. in the European and Turkish contexts but we also see some staggered planting. In the Anatolian context we see a lot of coppiced Quercus and planted Pinus. The Quercus lends itself to charcoal production (useful in the Bronze and Iron Age for metallurgy but used more now for soil replenishment) but also provides robust timbers for construction. Where we have little data is the antiquity of coppicing and pollarding (probably the fruit trees to prevent over grazing) in the Anatolian region. I have a data spike where I see a lot of "medium" ring curvature so I am interested to see if this might indicate coppicing.

Cheers Nathan

I have tackled this problem more generally. See an article  published in 2012 (Context to economy...) available by download from my academia site, and other articles there.  I recommend using three or at most 5 categories for strongly, moderately and weakly/flat rings. Another authority is the WODAN site (Irish charcoal database), which pictures the way to do this. You need ca. 5mm cm2 of surface area to see this properly, (use a ceramics template to help you).  Simpler is better, as this is a subjective measurement.  Size categories I also deal with in my papers.  Other scholars besides those you mention, for example  M Robinson, and R Gale, take >5mm annual growth as inidicative of coppicing.  Agreement on this topic is still not complete.

Hi Robyn

Thanks for your help here, most appreciated. Among my thousands of samples approximately 30% were large enough to calculate the ring curvature (I used trig. in an isosceles triangle and recorded all the individual ring sizes. As you can see the issue I have been having is defining size categories for measurements I have taken so your information is of great help here. Luckily my partner speaks fluent Turkish so we have been able to make contact with the Forestry Ministry there and they are saying they have statistical data on Quercus growth and trunk sizes. This may be of use but remains to be seen.

Cheers Nathan

If you are interested in "standardization" of curvature measurements, I suggest you investigate computer assisted morphometry.  Some programs, e.g. Fovea Pro (Reindeer Games, Inc.) include algorithms for measuring curves.  The resulting data is likely to be "standardized" which does not necessarily guarantee that it will be "significant."  But, you don't know until you try.  I recommend you look at any of the texts and manuals on Computer-assisted morphometry by John C. Russ for discussion of various methods for measuring curves.  There are several other authors of excellent texts.  I find that Russ' explanations are well written and easier to grasp, especially if one is a novice.