Statistical analysis of tree rings

In Message 335 in a now-closed thread, Daniel4140 wrote:

quote:

---

quote:

---

ETA; Whoops, the bristlecone measurements aren't labeled as such in the file. They are at North America Paleo Tree Ring Measurements.

​

You might want to read Biblical Chronology and the 8,000-Year-Long Bristlecone Pine Tree-Ring Chronology

---

​

Good link. So there is some data. I looked at four matches from ca. 1700 B.C., and did a few tests. First for the 'matching' sections I subtracted the ring withs to produce a difference table. I then found the average difference, and then the mean. Here is the start of the data for MWK001 and MWK002 cross match.

​

-1754 17.69%
-1753 Median 13 Avdif 2.3
-1752 MWK001 MWK002
-1751
-1750 15 21
-1749 13 17
-1748 14 22
-1747 12 18
-1746 16 27
-1745 16 21
-1744 14 13
-1743 15 15
-1742 20 22

​

This just a back of the envelope calculation. I did the same for a match in the last 500 years with similar results. Here is part of the data:

​

1616 26.25%
1617 Median 16
1618 Avdif 4.2
1619 MWK832
1620 18
1621 11
1622 0
1623 22
1624 13
1625 21
1626 4

​

Then I took this same section and deliberately missmatched it against another section. Here is the result:

​

Random Control Test #1
18 8.98 49.89%
22 3
14 18
14 5
19 1
11 1
22 3
2 27

​

The average deviation from the median ring width goes up to 49% of the median. I then performed the same tests on linkages between 4000 B.C. and 2000 B.C.

​

Test 1: 61%
Test 2: 79%
Test 3: 99%
Test 4: 47%
Test 5: 51%
Test 6: 49%
Test 7: 48%
Test 8: 45%

​

This was good enough to convince me that the whole statistical method of cross matching is highly subjective. The critical matches before 2000 B.C. seem to be no more than white noise.

​

...

​

I have read the AIG Woodmorerappe article. I disagree with John's quick rubber stamping of the statistical results on the basis that I have been unable to find a control test on a non-matching sample, plus have read allegations of others who have done just that with non-confirming results.

---

And, in Message 352:

quote:

---

You are the one misrepresenting the data. Why didn't you point out this discrepancy in the first place? Why did I have to dig it up?
And if you don't cite all the discrepant data, then why should I assume that those researchers didn't dump a lot of discrepant dates before they got some that agreed with their theory? Your "science" is just divination.

---

1. It's unclear what the results of your tests are. You write "The average deviation from the median ring width goes up to {emphasis added} 49% of the median" and then provide a table of other claculations in which the percentages are about equal to or larger than 49%. These are much larger than the percentages listed above for the "correct" matches, and suggest that your chosen test is indeed indicating that mismatches are mismatches. If this is not so, plese explain in more detail.
2. You write "This was good enough to convince me that the whole statistical method of cross matching is highly subjective." Since it appears that you are not using the statistical methods used by professionals, I don't see why you come to this conclusion. The results of your method provides no information about the objectivity or subjectivity of other methods.
3. Hearing of allegations is not evidence. If you have evidence of "others who have done just that with non-confirming results" trot it out. If you do not have evidence, drop the subject.
4. As far as I can tell, there is no discrepancy to point out. You have made a claim but have a long way to go in supporting that claim.
5. I'm sure RAZD will start another correlation topic soon.

Thread moved here from the Proposed New Topics forum.

If the tree rings were all the same width then there is nothing against which to measure any correlation.
The more measurable differences exist between the ring widths, the more possible a correlation becomes.
The less any pattern repeats the better, etc. Then the tree ring patterns become analogous to fingerprints.So differences between individual rings from the average is not the issue except that the differences exist in the first place. The issue is the pattern of differences across the specimen and some way to characterize its likelihood of being duplicated for a different time period.It would seem that the amount of fluctuation and variation allowed would be known from studying rings from living coexistent trees from the same area. Comparing this variation to mismatched samples would yield a control for purposes of correlation.Now the correlation with C14 signatures is significant since if the trees came from the same area and the rings are indeed from the same time snapshot they will have the same C14 signature however C14 might vary with time. The only way correlation with C14 would be insignificant would be if there was no variation of C14 with age at all, analogous to tree rings with no variation of width at all.So now for two ring patterns to accidentally repeat as well as the C14 signature to accidentally repeat becomes highly improbable. When the correlation of both of these patterns with varve layers is added we see that the correlation of agreement is what destroys the YEC viewpoint.

Good topic Jon, This of course is the crux of the matter. Is Daniel4140 using the same as dendrochronologists, or has he made up a process he thinks is valid. There is one all ready to go at Age Correlations and An Old Earth, Version 2 No 1 that has some updated information and a slightly revised format, including more specific reference citations.Enjoy.

---

we are limited in our ability to understand
by our ability to understand
Rebel American Zen Deist
... to learn ... to think ... to live ... to laugh ...
to share.

---

• • • Join the effort to solve medical problems, AIDS/HIV, Cancer and more with Team EvC! (click) • • •

I read Woody's report on his field work. He says he did not see any old samples to match when doing his work (he states this in a footnote). I think Woody was too hasty in the AIG article to say that the old matches are just as valid as the more recent.
I notice a lot of ring width in the data set at "0". It appears to me that these are rings circumstantially inferred by the researchers and do not actually represent physical evidence of a ring.
What tests if any have been done to directly confirm that double rings have not grown in brislecones in the last century? I saw a photo of at least one example of two extra rings appearing around the radius of a tree between two rings. It was like one highway split into two.O.K. Where is the link to a simple table of C14 data mathced with tree samples. For that matter where is it in the published literature?

---

Creation 4140 B.C. Flood 2484 B.C
Exodus 1632 B.C. Online Chronology book: The Scroll of Biblical Chronology

I already linked to plenty of sources of data linking 14C and all kinds of other measurments, at Seite nicht gefunden – MONREPOS, and Seite nicht gefunden – MONREPOS. You don't pay attention much, do you?But that's off topic. Here we're talking about statistical analysis of tree rings.You don't need to see any of Woody's work to explain your methodology and answer my questions. Or are you admitting your methodology is invalid and your "conclusions" about other's methodology are unsupported?ETA: Oh, and it's well known that bristlecones grow few if any extra rings. Certainly no tree grows enough to save YEC "chronology". You want to argue that they grow significant numbers of extra rings, dig up and present the data.Edited by JonF, : No reason given.

You ignored this post in the correlations thread:

quote:

---

It appears Daniel, that you are saying IF the published information is correct then the Earth is actually old.

​

The reason you think it is incorrect seems to be only one of two choices:
1) the researchers are incompetent
2) the researchers are lying or deluding themselves

​

Is this correct? Is this all you have to offer or is there more?

​

Since the information is, in fact, available to young earthers why haven't they shown, in detail, at a publishable level of quality how the research is wrong. We've read what they have published and it doesn't begin to tackle all the issues; not even half of them.

​

Why is that? Is it perhaps because they are the ones who are 1) or 2) from above?

---

Why didn't you answer this before.Why do you think you have to do all the research now? If it was possible to do what you are pretending to do then why hasn't ICR done it all already and published it?

I'm not interested in disucssing statistical analysys until it can be proven that the creationist community has access to the data source on the 14C dates for the Ferguson Chronology, among other things. Also, if cannot be established where rings were subjectively added by researchers, then there is no point in talking about statistics. Real science requires original data.
As it is a huge majority of the "matches" involve rings of "0" width. So if you put enough "0"'s into the ring width list at the right places, then you can created whatever signal you want to match with whatever. My current hypothesis is that the "0"'s were fraudlently placed. So provide the data and the answers first.The problems revealed by the back of the envelope tests are proving valid as I cut apart the crn file. For one thing when you whiten the file to get rid of the noise, pretty much all of the signal disappears also!Also there is no consistency with the expected percentage of mean growth in the master file and the matched trees.Column 1: whitened signal 40% threshold, "88888" means signal=true
Column 2: Master Chronolgoy, expected % of mean width
Column 3: an example tree supposed to match, actual % of mean width
Column 4: actual ring width in 100th mm.

0	133%	114%					42
0	92%	88%					22
8888888	22%	0%					0
0	98%	106%					28
0	71%	112%					21
0	94%	116%					29
0	105%	97%					27
0	111%	99%					29
8888888	31%	12%					1
0	82%	90%					19
0	76%	111%					22

Notice that the whitened signal in column 1 only matches a "0" and a "1". For the rest of the data, it is totaly random, dancing a normal distribution around the mean.
So it is imperative that the meaning of the "0" in the data set be established. Is it an INFERED ring by the researcher, or does it represent physical evidence of say a meansurement .004 mm rounded off to 0? Until then tree ring matching is no better than Bible Code ESL sequences. Given enough white noise, you can find a signal in anything!Edited by Daniel4140, : No reason given.

---

Creation 4140 B.C. Flood 2484 B.C
Exodus 1632 B.C. Online Chronology book: The Scroll of Biblical Chronology

I'm just now coming up to speed on this topic. Here's that link to the data table, but I can't see how you know they're for bristle cone pines: At the same website is a text file describing the analysis method and the file formats: Apparently you cannot just take the raw numbers and get valid results: It doesn't appear that Daniel made any attempt to apply this correction, but looking at the raw data I confess I can't see much correlation just by looking. If I were given the raw data but without the decades in the second column I doubt I could figure out they were describing the same years.--PercyEdited by Percy, : Grammar.

And Bristlecone Pines are known for having years with no growth ring due to the extreme climate. Matching 0 widths is thus the same as matching other widths. The final correlation between the Bristlecone Pines and the two Oak Chronologies after some 8000 years of correlations is a mismatch of 37 years, with the Bristlecone Pines being "younger" - evidence of more zero growth rings than what was determined in the analysis of the data.The most recent calibration review was done in 2004, and it goes by the name of INTCAL04 (the previous, INTCAL98, was done in 1998):http://www.radiocarbon.org/IntCal04_TOC.pdf
See: INTCAL04 TERRESTRIAL RADIOCARBON AGE CALIBRATION, 0-26 CAL KYR BP
and data sets: http://www.radiocarbon.org/IntCal04.htm
from http://www.radiocarbon.org/ (Radiocarbon Journal on line archives) Volume 46 Number 3, where number 2 and 3 contain the proceedings of the conference.Notice that they did not use the bristlecone pine data, as over a period of some 8000 years covered by the tree rings in three different dendrochronologies, the data was off by 37 years to the other two chronologies:

quote:

---

The replicate measurements have a mean offset of 37 6 14C yr (n = 21) from the Tucson measurements. Applying this shift to the Tucson data results in a close fit to the wiggles of the German oak, which would not occur if there were an error in the dendrochronology of either series. Because of this offset, the IntCal working group has decided not to include the BCP record in IntCal04.
(IntCal04 Terrestrial 14C Age Calibration, 0-26 cal kyr BP page 1033)

---

That's an error of less than 0.5% ...In other words, the effects have been very closely studied and accounted for.This 0 growth ring problem also does not occur with the oak chronologies except for the "year with no summer" - where this shows up right on time.Year Without a Summer - WikipediaAll three chronologies correlate with this event - can you explain how this occurs if the method is fraught with problems? Which just proves that your method to "whiten" the file is false, that the variations you are eliminating are part of the data rather than noise.What you have shown is that in order for you to get rid of the correlations between samples, you have to "whiten" the file until there is no data left. That's pretty conclusive evidence that your method is false.Enjoy.Edited by RAZD, : added linksEdited by RAZD, : updated link to article for website revisionsEdited by RAZD, : no

---

we are limited in our ability to understand
by our ability to understand
Rebel American Zen Deist
... to learn ... to think ... to live ... to laugh ...
to share.

---

• • • Join the effort to solve medical problems, AIDS/HIV, Cancer and more with Team EvC! (click) • • •

That was my first impression. I also see his attempt to "whiten" the data as a means to remove correct correlations and make more erroneous "correlations" possible by having fewer data points to compare.Enjoy.

---

we are limited in our ability to understand
by our ability to understand
Rebel American Zen Deist
... to learn ... to think ... to live ... to laugh ...
to share.

---

• • • Join the effort to solve medical problems, AIDS/HIV, Cancer and more with Team EvC! (click) • • •

How, exactly, does any of that answer my questions? I asked why the ICR folks haven't done this. Did you read it slowly enough.It doesn't need bristle cone pine or any specific trees so arguing about what is available and not is an utter red herring (and just as smelly). They can sample 3 or 4 different species of trees in different locations over just a 1,000 or 2,000 years. If the method is so full of error they will be able to easily demonstrate it.However, you avoid commenting on where the results of the ICR (or others) work is. The reason you do this is because they haven't done it. And the reason they haven't done it is because they know very well what the outcome will be if they are honest when they do the work.Now, instead of flapping around all over the place pretending to do maths just answer the question actually raised could you?

One of the places is the IntCal databases, where they have accumulated 14C matches to many other sources of measuring ages, including the tree rings. RADIOCARBON AGE CALIBRATION, 24,000-0 cal BP, RADIOCARBON, VOL. 40, No. 3, 1998, P.1041-1083, has a discussion of the methodology used and of the results for the 1998 calibration. INTCAL04 TERRESTRIAL RADIOCARBON AGE CALIBRATION, 0—26 CAL KYR BP, RADIOCARBON, Vol 46, Nr 3, 2004, p 1029—1058, has a discussion of the updated 2004 calibration and the changes made since 1998. Their data sets are at http://www.radiocarbon.org/IntCal04.htmThe problem you have is that you are fighting the correlation between a linear age counting system - the number of tree rings - against an exponential decay level of 14C inside those rings, and this presents you with an unusual problem to explain -- the different levels of 14C at different ages of the tree rings AND the similar levels of 14C for the same ages of tree rings in the three different tree ring chronologies.Carbon-14 has a half-life of 5730 years and this can be used to calculate an apparent "C-14 age" from the proportion of C-14 to C-12 in an organic sample (that derives its carbon from the atmosphere) and this "date" can be checked against known dates to determine the amount of C-14 that was in the atmosphere:

(Image based on calibration curvefrom Wikipedia(2) - Both images are in the public domain.)Note that the "C-14 age" is really a measurement of the actual ratio of C-14 to C-12 isotopes in the sample, and a comparison of that to modern day proportions.How Carbon-14 Dating Works | HowStuffWorks (5)

quote:

---

A formula to calculate how old a sample is by carbon-14 dating is:

​

t = {ln (Nf/No)/ln (1/2)} x t_1/2
where t is the "C-14 age", ln is the natural logarithm, Nf/No is the percent of carbon-14 in the sample compared to the amount in living tissue, and t_1/2 is the half-life of carbon-14.

---

These calibration curves have been extended now to the limits of Carbon-14 dating, but it is also of interest to look at just the Carbon-14 calibration curve for dendrochronology - the results of matching tree-rings to Carbon-14 levels and their implied "C-14 age":404 Page not found (9)

quote:

---

---

This means we can look at the "C-14 age" as a measurement of the Carbon-14 actually remaining in the samples from what was absorbed from the atmosphere at the time that the tree-rings were formed and note the following:

• If there were numerous errors in the tree-ring data caused by false rings (as you and others have proposed), then this would show up as a steep rising "C-14 age" that would be much younger than the recorded tree-ring age. This is not the case.
• The false rings would also have to be perfectly matched for each of the species used for the overall dendrochronology ages or the "C-14 age" for each one would be different and the line of calibration would be extremely blurred. This is not the case.
• The age derived from Carbon-14 analysis is consistently younger than the actual age measured by the numerous tree-ring chronologies in pre-historical times, meaning that C-14 dating underestimates the ages of objects.

This may seem off-topic to the issue of the statistical analysis of tree rings, but there is this one little factor that makes it relevant: The ages of the tree-ring data are validated by the carbon-14 levels in the samples. This a test of the methodology used by the dendrochronologists: they made their chronologies based on the tree ring patterns, and if their method were wrong there should be no correlation between age and 14C content and there should be no correlation between 14C levels for the same ages in the different chronologies.The "carbon-14 age" of a sample is really a measurement of the quantity of carbon-14 in the sample compared to the total carbon in the sample. This existing proportion of 14C/12C is just as much a part of the objective data to be matched from tree ring to tree ring as the width and density of the rings.This quantity measurement is transformed by a mathematical formula based on radioactive decay into a theoretical "age," but this "age" is really just a mathematical scale for displaying the actual amount of carbon-14 in the sample. The point here is that it does not matter what you or other creationists think about the validity of carbon-14 dating in particular, radiometric dating in general, or radioactive decay, because two samples of the same age - that lived in the same atmospheric environment and absorbed the then existing levels of atmospheric carbon-12, carbon-13 and carbon-14 (the three common isotopes) - will have the same levels of carbon-14 in the samples today.No fantastic scheme invented to change the way radioactivity works will change that simple fact, for whatever is changed in one sample is changed in all the others of the same time. Thus, when sample {A} is dated to {X} years by dendrochronology and it has level {Y} carbon-14 content, and when sample {B} is also dated to {X} years by dendrochronology and it has level {Y} carbon-14 content, the carbon-14 content validates the age - because, growing in the same environment, they could not be the same age and NOT have the same carbon-14 content.This simple fact will also refute any mismatch you propose to fit old wood to younger years.The earth is old, get used to it. You have been unable to refute the Lake Suigetsu data, your sources have been shown to provide false information on tree ring (your medical doctor's powerpoint presentation) and on the "equilibrium" of 14C in the atmosphere (see Message 159), and your method here is demonstrably questionable at best.Why do creationists need to use false information if their methods and conclusions are correct?Enjoy.

---

we are limited in our ability to understand
by our ability to understand
Rebel American Zen Deist
... to learn ... to think ... to live ... to laugh ...
to share.

---

• • • Join the effort to solve medical problems, AIDS/HIV, Cancer and more with Team EvC! (click) • • •

Chronology file name   : CA535.CRN
Measurement file name  : CA535A.RWL
Date checked           : 27MAR94
Technician's name      : HENRI D. GRISSINO-MAYER
Supervisor's name      : HENRI D. GRISSINO-MAYER
Beginning year         : -6000
Ending year            : 1989

The file is for BCP on the long white mountain chronology. The dates run from -6000 B.C. to 1980 in the second column, but they are not in order of date. The data in the correlation stats file and the crn file correspond to the data and descriptors in the rwl file you gave me.
The other file I already read.I did some stats on the "0"s which are about 5-7%. I've seen 5% reported as the number of "inferred rings". So there is a correspondence here. I couldn't guess where "inferred rings" were unless they are the "0"s.I eliminated growth changes by computing a moving average to determine the mean for a given range of years. 24 years before 24 years after, for a total of 49 years around the single growth ring to be compared to the mean.
The CRN file is given in % of mean expected exactly for the reason that actual width will vary due to the factors you cited.
What concerns me is the "0" values. Given as they are, it is clear to me that the correlation stats will look good. If one were to go through the rwl file and delete all the "0"'s and run it through the software packages, I'm guessing it will turn into white noise.
So I really need to know about the zero values and how they got into the file. A lot of them line up over multiple trees. What you've got to do is take the rwl data and copy out one tree. Then transpose the data in ms-excel to columns for each decade on that tree. Then you have to cut/paste all the decades together into one continous column for that tree. A similar procedure is necessary for the crn file. I then concentrated on trees 2000-3000 B.C. after sorting the correlation stats file. By the way, the years in that file are coded you have to subract 8000 from the numbers to get B.C./A.D. and watch out for the 0 year. I learned this by reading one of Ferguson's papers.
I know my stats are not as rigourous as the software, but I know that most people will not take the hands on approach. I am doing so, because I know that the original investgators took a hands on approach.
But I am really disappointed with Woody's lack of attention to the inferred year problem. What he said is likely correct for the good stats given the rwl file. Problem is that "0" in that file likely does not represent an actual ring value that just rounds off to zero. I need to know. If it does, then the mystery only deepens. If it does not, then I am a long way to pinning down the problem.

---

Creation 4140 B.C. Flood 2484 B.C
Exodus 1632 B.C. Online Chronology book: The Scroll of Biblical Chronology

I already ran calibration programs years ago on my computer. What I need is the specific data set for the BCP White mountain, decade measurements 14C. There are 500 of them. Either a linked file or a specific journal article containing the list will do.

---

Creation 4140 B.C. Flood 2484 B.C
Exodus 1632 B.C. Online Chronology book: The Scroll of Biblical Chronology