The validity of the European chronology

The validity of the European chronology - Summary and Conclusions

August 2011 updates, see Three years later - a validation of the Hollstein chronologies

Retrieved data

In this project I have retrieved data published by Ernst Hollstein and Bernd Becker and made that data available for computer analysis.
The main part of that data is based on 174 mean value ring width curves published by Hollstein, covering the time BC724-AD1974.

Becker's data is a mean value curve covering BC370 - AD1950.
The Becker data also contains data showing how he anchored his Early Middle Ages chronology into an established late time chronology.
There are also a number of short Roman time curves with latest year at AD181 included in the Becker data.

The Hollstein data looks ambiguous

The best crossdating Hollstein curves have been divided into two groups, Roman time and Late time.
The Roman time group covers the years BC340 - AD336. The Late time group covers AD410 - AD1974.
The late end of the Roman time group has a reasonably good match towards the early end of the Late time group.

If that match is a correct crossdate - as it actually looks - then it implies the removing of 207 invented years from our European calender.
Though we cannot disregard the possibility that there are some severe errors within some of the Hollstein mean value curves used for this strange crossdating!
Without more data and without original measurements we cannot make a thorough analysis that would dismiss or agree to the phantom time theory!
Which institution will be the first to publish data that will make this question come to an end?

The Becker data questionable

The late 80 years of Beckers Frühmittelalterliche Eichen Maintal chronology, presumed to cover the years 779-859, seems to be copied from the ring widht pattern of year 996-1076.
By seemingly only looking for a high Gleichläufigkeit value (0.65) he crossdates his Maintal chronology to the year 859 with a correlation coefficient of 0.28-0.33 and a T-value of 3.1-3.7 - values normally considered not to indicate a proper crossdate.

The methods used stand out as not reliable for tree ring chronology building.

Validation of published chronologies

The matter of replication. When tree ring chronologies are established they should be tested for conformance with chronologies from nearby areas. In this case I had the Danish oak chronology available. But I know nothing about its quality! Is that chronology built with modern methods on enough data to be correct and independent of other chronologies also in its very old end? It is very frustrating not to be able to check the validity of data used.

Availability of data. This leads to a comment on the availability of data...
Still today many tree ring chronologies cannot be checked by independant scrutiny! Though methods and tools are now readily available to analyze crossdatings, many institutions prefer to keep their data secret - allegedly to avoid unfair competition. Unfair in the meaning: "We pay for developing data - you copy the data and make money from it." But that is what we have public research institutions for! - they should produce knowledge that can be used by everybody! It is tragical that we have failed in organizing and funding our research institutions in a way that does not make them more willing to distribute knowledge!

I cannot escape a feeling that sometimes institutions and researchers follow a very comfortable way with their research: Best way to avoid criticism is not to show the details in what you pretend to have created! (No professor would admit that...) Why publish data, which - if found erroneous - would result in badwill for the researcher and his/her institution? Though as publishing is successively made easier through the internet, errors could hopefully be detected at a less prestigious level than when published in a scientific journal. Also EU financed projects seem to result in publishing of data and eventually also the Aarhus convention may be applied to dendro reports.

Our attitudes related to knowledge transfer as well as to copyright and patents is a great problem and challenge for our society. (E.g. look at the problems caused by software patents!)

Conclusions on the Phantom time theory

The problem whether a phantom time exists or not cannot be concluded from this study!

Without more data and without original measurements we cannot make a thorough analysis that would dismiss or agree to the phantom time theory!

What's next?

Before more data is available it is difficult to further this project. I have contacted both the Trier Landesmuseum (where Hollsteins data was archived) and the Hohenheim university and politely asked for more data. I have still not got an answer.

I can only hope and pray that most European dendrochronologists will eventually abandon their policy of keeping dendrochronology data secret. It is really a shame for our society that available knowledge can be kept secret at public research institutions.

It could be anticipated that the dendrochronological community would consider it a point of honour to have the theory of "a phantom time" removed from the agenda
through a thorough analysis including publishing of data that could be analyzed by anybody interested!
Only ignoring that theory is not a good strategy when it is so easy - as I have shown - to demonstrate inconsistencies and obvious errors in those unusual dendrochronological reports which also include data so that crossdating can be validated!

August 22 2008. Lars-Åke Larsson

Three years later - a validation of the Hollstein chronologies

Queen's University Belfast was first to publish all their dendro data. Then we got a lot of data held by Sheffield University/English Heritage. After reworking a great amount of Irish/English dendrochronology we now know that there is a deep gap (a strong timber shortage) between recent times and Roman times also on the British Isles (see "Closing the gaps" in the Belfast section).

During our work with the material we noticed that chronologies from southern England, especially London, match very well towards the Hollstein chronologies. In fact so good, that it was possible to validate some of Hollstein's single site collections. These site collections form our new collection "HollsteinValidated", AD 410 - 1242. This collection finally confirms that Hollstein bridged the Carolingian gap (the timber shortage during the eighth century AD) correctly.

With "HollsteinValidated" as a seed collection we then tried to fill up with additional Hollstein material matching without doubt and with a TTest-value not less than 9.0. In this way a complete Hollstein master could be assembled, spanning AD 410 - 1974. See "HollsteinValidatedExtended" in the download section below. It was not possible to extend that Hollstein master towards earlier times with this method.

The same strategy could be applied on Roman time data. We built the collection "HollsteinValidatedRoman", BC 210 - AD 203, against SouthEnglishRoman. The extended master "HollsteinValidatedRomanExtended" is spanning BC 547 - AD 336.

Validation against the Becker-chronology. The two validated Hollstein masters also fit absolutely convincing against Bernd Becker's master chronology (and even better against A100). The overall correlation of the younger master "HollsteinValidatedExtended" is as high as 0.70 towards the Becker chronology, which probably means that large parts of the historical material which Hollstein used also are included in the Becker chronology. The Roman master fits Becker's chronology with corr. 0.48 which is interesting as the Becker chronology in its older end mainly consists of subfossile oaks, in contrast to Hollstein's historically anchored material.

This study shows 1) that it is possible and meaningful to retrieve the Hollstein data from the drawn curves in his book, and 2) that the quality of the data and its synchronisation is very high compared with the currently valid master A100. Both points have been heavily questioned in contacts with well-known German dendrochronologists. (The retrieved Hollstein curves show corr. 0.98 and better against some original Hollstein site collections we got hold of.)

What is wrong with the "leftovers" of the Hollstein material? Not much at all, most of the collections fit "as dated" towards the new validated masters. They just do not fulfill the sharp criteria for addition to the "extended" collections, perhaps due to regional diversity (see e.g. ZOBB (Zöbingen in Baden Württemberg) which would extend "HollsteinValidatedExtended" down to AD 383). The only controversial sample is BDN1 (Bad Nauheim, Siebel (carolingian), dated AD 653 - 792) which fits much better at AD 1656 (!). But this sample is a single, deteriorated stem of unknown provenience and can easily be excluded.

Why is it impossible to connect the two validated masters and fill the Migration gap with Hollstein material? To reach beyond AD 400, Hollstein only had an extremely slowgrown sample from Cologne (KKNA, AD 273 - 511) and a collection from Beerlegem in eastern Flanders (BERL, AD 284 - 550). This means he shiftet region from highland to lowland, with low correlation values as a result. While KKNA fits "so so" against "HollsteinValidatedExtended" (P2Yrs: corr 0.36, TTest 3.8; Baillie/Pilcher: only corr 0.24, TTest 2.5 at 101 years overlap), BERL does virtually not fit at all (corr 0.11, TTest 1.3 at 140 years overlap). However, it is most probable that the dating of BERL is correct (this can be demonstrated against the "EnglishMaster" via two collections from northern France, will be discussed later), while a replication of the KKNA-dating is not possible. So we actually wonder how Hollstein could date BERL correctly without access to external chronologies, and why he included that collection at all. Maybe because he needed a third collection from Broichweiden in the Aachen area (BROI, AD 129 - 365) to make his (the conventional) bridge to Roman time more trustworthy. BROI has indeed a short match against BERL (corr 0.41, TTest 4.0 at 81 years overlap), but instead this collection does not really match any other Hollstein collection "as dated" (with the exception of KORB, but the KORB collection has to be truncated by 32 years at its younger end to give an acceptable match). It actually seems that we are dangerously near reading the coffee grounds.

What about the annoying "207-years-match", does it still show up? Yes, this match is alive and well! "HollsteinValidatedRomanExtended" fits "HollsteinValidatedExtended" at AD 543 with a resounding P2Yrs: corr 0.50, TTest 6.7; Baillie/Pilcher: corr 0.42, TTest 5.3 at 133 years overlap. And this match is not only depending on just one sample, it is present in several of Hollstein's regional mean curves which were retrieved from the green book and given to us by Kristof Haneca. We can't just ignore this, especially not as the conventional bridge is that weak and doubtful. Conclusion: the Roman time chronology is still floating and even more data is needed to show how long the bridge over the Migration gap is.

Petra Ossowski Larsson, Lars-Åke Larsson. August 13 2011.

Files to download:

These collections contain synchronized Hollstein mean value files.

HollsteinValidatedRoman, BC 210 - AD 203 validated against South English data.
HollsteinValidatedRomanExtended, BC 547 - AD 336. Same as HollsteinValidatedRoman but extended with other well matching Hollstein data.
HollsteinValidated, AD 410 - 1242 validated against South English data.
HollsteinValidatedExtended, AD 410 - 1974. Same as HollsteinValidated but extended with other well matching Hollstein data.

All the files are packed together into HollsteinValidated.zip

Continue with the next section/More information!