The Proceedings of the Eighth International Conference on Creationism (2018)
correlating at r<0.3281 with the master chronology). With reference to MWK, FIN, and TRN, the percentages of their respective total series with at least one flag are: 9.5, 11.0, and 20.4. (CDendro does not readily lend itself to an after-the-fact examination, of a large collection, for wrstblk characteristics, so this was not attempted). Actually, the number of identified flagged segments, based on the archived ring-width measurements I used, may be understated. Bearing in mind that very-young and very-old trees are more likely to respond poorly to the annual climatic signal, there is a tendency (personal communications) to clip off the early or late parts of a tree-width series in order to improve the overall statistical “fit”, and especially to reduce the number of poorly-matching segments, of a given series, against another series or against the master chronology. It is also possible to take a sufficiently-long series, cut out a middle-situated “misbehaving” segment, and then treat the remaining “well behaved” segments as two separate series. While this procedure does not, of course, call the long chronologies into question, it does make the archived evidence for them somewhat more favorable than that in reality. 4. How Individual Tree-Ring-Series Coalesce to Form a Master Chronology Dendrochronologists do not construct long chronologies by indiscriminately gathering thousands of subfossil samples and then determining if and how they crossmatch. They start with much smaller and locally-derived aggregations of samples, and perform the interactive procedure detailed in the ensuing few paragraphs (and executed in CDendro as P2Aut*). For example, consider the Belfast long chronology. Ten or twenty subfossil trees from a single location could make a site submaster(*) chronology that is 400-800 years length (Baillie 1995), and a single farm in Ireland could yield 50 subfossil trees that coalesce into a site submaster chronology that is 500-1,000 years long. (Baillie 2009). Eventually, such submaster chronologies became crossmatched with each other [as done, in preliminary fashion, in Figure 6], and then merged. This iterative process culminates in a single multi-millennial master chronology. The modus operandi of building submaster chronologies is as follows (and is abbreviated P2Aut): An individual series is chosen from among the best “many6matchers”*, normally at or above the previously-stated “ceiling” P2YrsL t-value of at 7.0 (Larsson 2003- 2018; Larsson and Larsson 2018). The chosen series, by virtue of the high frequency of its high-t pair-crossmatches against several other series, already contains a strong annual climatic signal, and is therefore designated the “seed”* or “kernel” of the impending submaster chronology. In CDendro, the “seed” is allowed to “choose” and “pick up” the highest-matching series to “join” it and be averaged with it (either one a time and checked by the dendrochronologist, or automatically). With the autoadd* function turned on, this averaged assembly then “chooses” and “picks up” a series that matches it well, averaging it in with the first two. And so the process “snowballs”, owing to the “master chronology effect” (defined in next paragraph). Eventually, the process runs out of Woodmorappe ◀ Tree-ring chronology shortening via disturbances ▶ 2018 ICC 658 Figure 4. Screenshot Showing the CDendro Display of the Statistics Behind the FIL6924X/FIL9042 False Crossmatch in the Finnish Long Chronology (FIN) Illustrated in Figure 3. The most important numbers for the reader to follow are: The OVL, located at relative year 35, is 149 years long; the Besancon* is at r=0.41 and t=5.5; the skel-Chi2* is 12.9; the P2YrsL* is at r=0.48 and t=6.6. The gleichlaufigkeit* is a congenial 61%. Note the (observed and desirable) sharp discontinuity to the second-place (next best crossmatching position) finisher at P2YrsL t=2.9.
Made with FlippingBook
RkJQdWJsaXNoZXIy MTM4ODY=