TREE RING DATA FOR GERMANIA

            As discussed in an earlier chapter, Ernst Hollstein released the only published data from the area of Germania in 1980, but it contained errors in the Roman Era (Hollstein, 1980, Friedrich and Greiner, 2001) .  This chronology was corrected and updated in 2001.  To illustrate this change a short series of computer Boolean logic tests on the data were performed.  This Boolean logic test allowed for systematic approach to identifying changes in the long chronology where the numerical data and line graphs can become visually confusing.  The 1980 and 2001 data were evaluated using their standardized, 21-year running mean format for significant changes in values.  The first Boolean function identified the years in which the chronology differed in the change of tree ring size.  If the 1980 and the 2001 chronology did not move in the same direction (both chronologies showing either a smaller or larger ring than the previous year), then the Boolean function would identify a “false” result (moving in opposite directions) and record the Absolute Value difference between the values in that year.  These results were subjected to a second Boolean logic test to identify any years that had values that changed sides of the mean value from the 1980 data to the 2001 data.  If the tree ring values were on different sides of the mean from the 1980 data to the 2001 data the ring widths would be getting either larger not smaller or smaller not large.  This change in the chronology would be a significant change because it could change the interpretation of the climate signal in the record. 

According to the results of these Boolean logic tests, the error in Hollstein’s data changed the 1980 published chronology roughly between the years AD 302 and 436 (Figure 4.13).  It is easy to identify the changes in the updated data set in the figure by adding a 5-year running average to help emphasize these changes (shown by the peaks in Figure 4.14).  The other errors that fall outside of this section, AD 302-436, are due to several smaller errors identified in the re-alignment of the 27-year error and several other smaller corrections.  Even though the change did not effect the interpretation of the droughts of the 3^rd century AD, the new values in the corrected 2001 version will affect the interpretation of the Late Empire Period. 

This significant change in the ring width values of the 2001 chronology was due to the misplacement of the wooden sample from Rainer-Buch well #13, but the effects were deepened because this section of the chronology had such a low sample depth.  The sample depth for a chronology is the number of tree ring samples that cross any given year in the chronology (Figure 4.15).  Ideally, there would be well over 100 samples per year.  This does not mean that a chronology with a low sample depth is incorrect, just that the potential for recording local

 and physical phenomenon in lieu of a climatic signal is increased with fewer samples.  The Hollstein data set for the 1980 chronology was problematic because the sample depth for much of the 4^th century AD was less than 10 with only three tree ring samples for several of the years (Figure 4.16).

The master tree ring chronologies used in this study are the updated Hollstein chronology compiled from nine river crossings in central Europe (Hollstein, 1980, Friedrich and Greiner, 2001) (Figure 4.17), the Becker Roman archaeological chronologies compiled from several Roman sites including the wooden foundation of the Roman road through the Alps (Becker, 1981) (Figure 4.18), the Main River Chronology compiled from samples buried in the river’s gravel bed (Spurk et al., 1998) (Figure 4.19), and the Danube River Chronology compiled from samples recovered from the gravel mines that are in southern modern Germany and Austria (Spurk et al., 1998, Becker, 1993) (Figure 4.20).  These four chronologies are all crossdatable, meaning that individually the signal contained in each is similar enough to produce similar climate interpretations, so they were combined into one Master Chronology to represent the southwestern portion of Germania (Fritts, 1976, Glock and Pearson, 1937) (Figure 4.21). 

            Looking at the Master Chronology, it reveals a distinctive trough in the growth index in the third century that stands out from the rest of the tree-ring record (Figure 4.22). The four chronologies used in this paper have a combined minimum sample depth of about 100 (Figure 4.16). The third century, already identified as a “turbulent century”, is also identifiable as possessing a significantly long drought period reflected in the tree ring chronology of the river  valleys of Germania.  The period from AD 216 to 232, with the brief exception of 231, all display a below average growth trend in the Master Chronology.

There are other troughs in the Master Chronology in the end of the 1^st, 3^rd, and 5^th centuries AD, but none as long or severe as the early 3^rd century AD.  If the data are evaluated to identify drought and non-drought periods in the tree ring chronologies, significant periods of both categories are apparent.  This is expected within the normal parameters of the oscillations of climate on the Earth.  The simplified definition of “drought” and “non-drought” is any time there are four or more consecutive years above or below the 21-year mean.  Above the mean is a “non-drought” period and below the mean is a “drought” period.  The ends of the drought or non-drought period are defined when the chronology has two consecutive growth years in the opposite direction.  This allows for single-event years that are inside the longer drought or non-drought years to still be included in the overall event.  This can sometimes lead to a slight overlapping of drought and non-drought years, but does allow for an overview of the period as a whole.

            Changes of this magnitude in the growth index are usually indicative of a climate change (Phipps et al., 1994, Douglass and Glock, 1934, Glock, 1950, Fritts, 1969) .  With trees from the river valley, this is also indicative of significant changes in soil moisture availability (Grimm, 1962, Schweingruber, 1966, Wimmer and Vetter, 1999) .  By examining trees from the river valley, the sampled trees are in close proximity to the human inhabitants, especially evident in the Roman felled trees in the Danube and Becker chronologies (Figure 2.8).  However, trees could be carrying a non-climate signal, and this may be true of the trees of the river valleys that could respond to floods though this is unlikely due to the number of samples in the chronologies.

This analysis suggests that the years AD 178-182, 198-211, 214-232, 264-273, 277-283, 294-301, 316-321, 334-353, 364-382, 399-405 could be drought periods.  Conversely, the years AD 167-177, 183-190, 194-199, 239-249, 253-263, 300-317, 322-333, 354-365, 381-390, and 406-419 could be non-drought periods.  The 3^rd century consists of 4 long droughts totaling 42 years and 2 long non-droughts totaling 28 years.  The remaining 40 years are classified as “neutral” years because none of the above criteria have been met (Figure 4.23).