In search for a correct and continuous Middle European Oak chronology
- an abstract
"Until somebody demonstrates a better alternative, we consider these results as a proof that there are invented years in our time line..."
This project started in 2002 at using dendrochronology to prove, whether Heribert Illig's theory of invented years during the Early Middle Ages is nonsense or in some way right.
After retrieval of Ernst Hollstein's German data from his hand drawn curves of site chronologies, we could clearly show that a 142 years long segment from Merovingian time (AD 401-543) ALSO (!) matches towards late West Roman time data (at AD 195-336) with high correlation values. Such a multiple match towards a reference curve implies a very big error. With that big error we understood that Hollstein could reasonably not have been able to create a correct and continuous tree ring chronology from now and back into Roman time. These results were first published here in August 2008 and later presented at a round table dendro meeting in France in May 2010.
The Belfast data. At the end of April 2010, Queen's University Belfast (QUB) made all their dendrochronological raw data available as some 9500 unsynchronized files identified by numbers and site names. After we had synchronized a major part of it, we found that the Irish oak chronology fell apart into three segments: BelfastLong, LateBC and BelfastAD with gaps in between them at 948 BC and around AD 1.
English data. In June 2010 we got unsynchronized English raw data of archaeological origin from Roman time and the Middle Ages from Sheffield University/English Heritage. With that data we were able to reconstruct QUB's proposed linkage of the Irish oak chronology around AD1, though we do not regard it as really convincing.
The error at 948 BC. The gap between BelfastLong and LateBC was originally bridged with a bog oak collection from England, indicating that there was a real gap of one single missing ring at 948 BC in the Irish data. In contrast to that, our crossdating of the series showed that the BelfastLong and LateBC collections actually overlap each other by 110 years when dated conventionally, but with a completely unacceptable low correlation though the samples are from the same bog. Later we found that the overlap is caused by a new stem not measured at QUB before 2009 - so this new stem overturns the original QUB solution. Our investigation indicated that the "one year gap" at 948 BC has to be widened by possibly more than 200 years! These results were published on this site in February 2012 and then also presented to QUB, who asked for the results to be published in a peer reviewed scientific journal. Such an article has been in review for six months and was then rejected.
Inserting many years at 948 BC would either push LateBC (with its linkage to Roman time) towards our time, implying that Roman time is incorrectly dated by some 200 years - or would push the BelfastLong collection towards older times. Both alternatives seem to be impossible as the 14C calibration curve - based on exactly the same dendro data in which we found the error - is said to heavily support the current solution because of its conformity with a historical American Bristlecone pine calibration. But this argument is circular reasoning. First you create a calibration curve with allegedly correctly dated wood, then you use that calibration curve to prove that the wood is correctly dated. Even if the parts of the Irish oak chronology may have been roughly placed in time by wiggle matching against the Suess calibration, there is no Bristlecone pine data in the IntCal09 calibration curve! However, the wiggle match of BelfastLong against Bristlecone pine looks about correct, while the period corresponding to LateBC (1st millennium BC) shows major irregularities.
Crossdating European oak towards North Scandinavian pine. In February 2013 we came to discuss how and why an absolutely dated, 1800 years long Danish oak curve matched supra-long absolute pine masters from Northern Scandinavia. The correlation coefficient is not high (0.14), but because of the long overlap the T-value (together with the high skeleton-Chi2 value) clearly discriminates the match (t>6). Also absolutely dated oak from other north-western central European sites showed significant correlation. So we considered it worth trying to combine all our Roman time oak mean value curves in one big collection and match it towards the more than 7600 years long Scandinavian masters.
The result was overwhelming! We did not see a trace of the "correct" conventional dating, but instead a well discriminated match 218 years later than expected with a T-value between 5.0 and 6.5 (depending on the normalization method used), and with a high skeleton chi2-value. We then also confirmed a 218 years offset with reasonable correlation between Roman time and recent time local oak masters.
Conclusions. Until somebody demonstrates a better alternative, we consider these results as a proof that there are invented years in our time line, possibly 232 years as proposed in our astronomical excursions.
Matching Danish oak - absolutely dated to AD 1986 - towards North Scandinavian pine
Matching Roman time oak of NW Europe - conventionally dated to AD 328 - towards North Scandinavian pine
The diagram shows T-values (T) for all the 5850 possible matches between the Danish oak curve and the North Scandinavian pine curve. The correct (true) match at 1986 corresponds to the highest peak (T=6.1). The other matches - which are all false - have significantly lower T-values (the "best" at T=4). - It should be noted, that if we insert or remove one single ring width in the inner parts of the oak curve, the highest T-value immediatly drops to 4 - i.e. then the match disappears.
The diagram shows all possible matches between a NW European Roman time oak curve, conventionally dated to AD 328, and the North Scandinavian pine curve.
The best matching point is at AD 546, i.e. 218 years later than expected! Please note, that there is no competing match at the expected year AD 328!
(The high peak at AD 1721, with a T-value near 5 is outside the region of reasonable matching points.
A separate analysis shows that it is accompanied with a very low Skeleton-Chi2-value which clearly indicates that this is a false match.
However, the best match in each of the two diagrams above has not only a high T-value, it also has a high Skeleton-Chi2-value.)
Please use the links in the Contents (to the left) to view our detailed reports and download sections.
April 6 2013. Lars-Åke Larsson & Petra Ossowski Larsson (updated July 3 2013)