The Proceedings of the Eighth International Conference on Creationism (2018)

self-imposed lower limit for inclusion, in Table 2, was r≥0.3, some of the ensembles have a minimum matrix pair-crossmatch at r≥0.4 or even 0.5. REVERSE-ENGINEERING THE PRE-1000 BC PART OF THE SCANDINAVIAN LONG CHRONOLOGIES The conventional presentation of a long chronology, as a fait accompli, can create the mistaken impression that the evidence behind it is equally sound in all parts of it. It is not. This section deconstructs the TRN and FIN. It then necessarily-tentatively relates this to the “clusters” and “bridges” of the Disturbance- Clustering hypothesis. 1. A Re-Derivation of the Entire Finnish Long Chronology Larsson and Larsson (2018) claim that the entire FIN can be reconstructed almost entirely by the P2YrsL t≥7.0 P2Aut alone. That is, submaster chronologies are created, overlapped, and then merged. This creates three large submaster chronologies that cover the youngest 6,500 (that is, all but the first few centuries) of FIN. Only some manual adding of series, at P2YrsL<7.0, is then needed to “suture” all these together. Owing to the fact that Larsson and Larsson (2018) did not publish any details on this, I have attempted my own P2Aut t≥7.0 reconstruction of the entire FIN, and have evidently obtained at least roughly comparable results (Figure 6). The PAut process generated 10 submasters of varying length, having “picked up” a given series only once. Some of the submasters edge-crossmatch at P2YrL ≥7.0, while satisfying all other criteria, and do so at their expected points. These can straightforwardly be merged together. Others have crossmatches at the expected points, but the crossmatch cannot be accepted because one or more of the gateway statistics is too low. Finally, a few of the junctures between submasters have too short an OVL to even attempt a crossmatching. In summary, most of the nine junctures need manually-added-on series, at increasingly lower-standards, before the crossmatches can be sufficiently high to justify a merging of all the subchronologies into one grand near-8,000-year long master chronology. Now consider what happens when this entire exercise is repeated, this time allowing TRN as well as FIN series to be “available” to the same “seeds” as formed the subchronologies shown in Figure 6. Moreover, this time I allow for each series to remain eligible for “picking up” by more than one “seed”. The subchronologies become noticeably longer, eliminating the junctures of insufficient OVL for crossmatching, except for the 5226 BC 5204 BC junction (again, possibly to be filled with a new collection). Although the ability of P2Aut to re-create nearly the entire FIN at first seems impressive, it actually encompasses a good deal of “hollowness” in TRN and FIN. This “hollowness” is identified in the remainder of this chapter. Again, this does not call into question the validity of the long chronologies, but it does “lower the bar” for the Disturbance-Clustering hypothesis to account for them. Let us first focus on P2aut itself. Larsson (2003-2018) states that, as long as t≥7.0, autoadd never “picks up” a wrong TRN or FIN series to attach to the emerging master chronology. I have made many “runs” of P2Aut using different “seeds”, and, based on my own experience, tentatively agree. However, such an outcome is far from absolute. As elaborated earlier regarding Table 1, perturbed series readily get “picked up” at P2Aut t≥7.0, moreover regardless of which disturbed series serves as the “seed”. Moreover, as it turns out, this “false recruitment”, at t≥7.0, is not limited to known- disturbed series. Let us temporarily consider the entries in Table 2 not as prospective “bridges” but as embryonic master chronologies in their own right. I have found that, in the last entry of Table 2, all eight remaining series are “picked up” by the autoadd process, of P2YrsL≥7.0 and OVL≥70 years (P2Aut), provided that “interloper” TRN series 0022027A is allowed to serve as the “seed”. In the fourth listed series, both remaining series will be “picked up” if the “seed” is either KOM5986 or KOM6750. Finally, in the next- to-last entry, both remaining series are “picked up” regardless of which of the three series is chosen as the “seed”. Next, we must ask how P2Aut relates to the pair-crossmatching capabilities of the “picked up” series. It has been suggested that it does not matter if many of the series in the matrix pair-crossmatch at low levels (P2YrsL t<6.0), as is the case with TRN and FIN, as long as the assembly found in the matrix had arisen entirely from a P2Aut at t≥7.0. However, apart from dubiously treating P2Aut as an absolute (which, as we have just seen, is not so), such an approach glosses over the poor pairwise-crossmatching quality of many “picked up” series, as is obvious by looking at the TRN and FIN matrices. Note that, in a number of modern large Scots pine collections I have surveyed (those of Baloos eastern Russia, Mongolia, Georgia, etc.), most of the series, “picked up” by P2Aut t≥7.0, pair-crossmatch in the matrix at P2YrsL t>6.0. However, this is not the case for the series in the older parts of TRN and FIN. That is, for the 790 TRN/FIN (L*≥100 years) pre1000 BC series, the P2YrsL t>6.0 matrix crossmatches (those at OVL≥50 years), pertaining respectively to TRN, FIN, and combined TRN/FIN, are relatively few, and as follows: 26%, 46%, and 32%. For P2YrsL t>7.0, the respective figures are only: 13%, 30%, and 19%. Thus far, I have considered TRN and FIN, in terms of relative numbers of “few6matchers*” and “many6matchers*” only in an overall sense. However, it turns out that that the same liability holds for the “strong” subchronologies that are created solely by the P2Aut t≥7.0 process. To make the strong-subchronology exercise even more challenging, I focused on just the most robust part of the pre-1000BC FIN/TRN chronology—that is, the interval from about 2000 BC to about 4000 BC (identified as such by Larsson and Larsson 2018, and also so identified by my own experimentation, which is later discussed in conjunction with Figure 7). I chose as the “seed” FIL6741 (-3503 -3204), and performed P2Aut, under the restrictive chronology-starting condition of each candidate series crossmatching with the emerging master chronology at P2YrsL t≥7.0, and OVL≥90 and then ≥70. This CDendro procedure “picked up” and averaged 305 series, into a master chronology, before running out of qualifying series. It thus constructed a 2,142-year- long submaster chronology (-3854 to -1713), which is shown in Figure 6 as Subchron 5. Let us consider the matrix with its individual-pair crossmatches of the 305 “picked up” series. I had CDendro rank them (left to right, in descending order, recognizing only OVL≥50 years) as follows: The best “many6matcher” (with 52 of them) down to the worst “few6matcher” (with 0). The descending order showed an exponential decline not only in the numbers of “many6matchers”, Woodmorappe ◀ Tree-ring chronology shortening via disturbances ▶ 2018 ICC 665