From	Martin Munro	Date	Sun, 29 Aug 2004 11:46:03 -0700
To	itrdbfor@listserv.arizona.edu
Subject	Calibration loose ends (was Re: [ITRDBFOR] crossdating)
This an attempt to tie up the loose ends from an earlier part of the discussion, the idea that calibration of the radiocarbon timescale be considered invalid, pending a better understanding of crossdating. Some of the previous posts seem to imply that measurements of the C-14 half-life depend on the calibration; in fact it can be determined by present-day laboratory measurements without reference to any old material, simply by observing the decay rate in a known quantity of the isotope. Physicists seem happy that beta decay isn't affected by mundane external influences, so the half life should be constant. If the amount of C-14 in a sample depends only on its age and the (constant) half life, a calibration curve from a collection of samples of known true age would be a diagonal straight line; but this would imply that each sample started with the same concentration of C-14. There are many effects that could change this concentration through time: variations in cosmic ray sources, changing solar activity, changes in the upper atmosphere, atmospheric circulation, uptake and release of carbon from large sinks and sources... etc. Given enough correctly dated samples, you can recover the sum of these variations from the form of the calibration curve. In practice, the most important variation appear to be on multi-millennial scales, with smaller fluctuations (wiggles) on century/multi-decadal scales superimposed on this. Wood from crossdated tree rings provided the known-age reference material used in the calibration curves, and there were two main phases of work, the first of which roughed out the general form of the curve and hinted at the short-period structure, the second of which reconstructed the century-scale variations in detail using higher precision measurements. Contamination of old samples with C-14 of more recent origin is a widely recognized problem, addressed by physical and chemical pre-treatment protocols for the material. A couple of complicating effects that are of more interest from a tree- physiological point of view. Isotopic fractionation occurs along the entire chain of processes between carbon in the environment and its incorporation in the specific components of the wood that end up in the calibration samples. A ring forming in a particular year might continue to accumulate C-14 in subsequent years. But people who work with C-14 are well aware of various corrections for isotopic fractionation, and the migration of carbon across ring boundaries has been the subject of several empirical investigations, notably using the stepwise change in C-14 concentrations following atmospheric nuclear tests in the 1950s and 60s as a tracer. The more recent phase of calibration work was substantially complete around 15 years ago, and was covered in an extensive series of journal articles and symposia. Let's suppose we have been provided with a demonstration that crossdating is invalid: what would be the consequences for C-14 calibration? One of the most alarming would be that we would have to come up with a convincing explanation of how independent tree ring chronologies could be in error in precisely the same way---the known-age reference samples are not just from bristlecone pines, and crossdating within the network of oak chronologies is completely independent of the bristlecones. Both are completely self-supporting chains of inferences anchored in living trees and extending back into sub-fossil wood. There are published comparisons of paired calibration curves, with the absolute dates and C-14 concentrations based on oaks in one case, and on bristlecones in the other. My understanding of tree physiology is rudimentary at best, but surely when two such vastly different wood anatomies are involved there must be differences in the physiological constraints on wood formation. If potentially unidentified missing rings are supposed to be the most serious problem with the bristlecone chronologies, the oak chronologies should not be affected in any case, since they almost never include missing rings in this sense (although that's not to say they have no anatomical ambiguities that can confound crossdating). The crossdating error could not be merely a shared systematic bias; not only does the long term trend in the calibration curves derived from the two chronologies share a common non-linear trend, but the short-term fluctuations in C-14 concentration (wiggles) match between the two curves. There are small differences between calibrations derived from different geographical regions, but these have themselves formed the basis for further research and geophysical modeling. The strengths of the two sets of chronologies are complimentary. Oaks may have almost no missing rings (sensu stricto) and provide larger volumes of wood for C-14 analysis, but the individual samples are only a few hundred years long, showing significant variations in growth with increasing pith age, and (particularly in the case of the sub-fossil wood) there will be uncertainties about the environment in which the tree was growing. Bristlecone pines give a much better chance of finding wood that has grown over periods of many centuries with no marked age-related trends, and there's a compelling continuity between the living trees and the remnant wood lying on the ground nearby. An account of wood formation from a physiological perspective would undoubtedly be a beautiful thing in its own right, even if it had little to contribute to dendrochronology. Moreover one of my pet peeves is seeing people manipulate data as mere collections of numbers divorced from any underlying model---and in the case of dendrochronolgy the model has to be biological. But I'd number myself amongst those who can't see why our use of crossdating must await a reasonably complete physiological model of wood formation. By analogy, if the doctors in some traditional society are using a human physiology based on the balance or imbalance of the four humours, but they have a treatment for a particular disease that results in an 80% survival rate, as opposed to a %40 survival rate if it goes untreated, you're obviously better off slurping down their bitter potion first and working out the explanation in current Western physiological terms afterwards (if that's the only treatment option). So even if at present our understanding of crossdating is largely limited to statistical phenomenology, that may be good enough to live with until something better comes along. That's not to imply that we should be credulous, and automatically accept current practices simply because great authorities have taken the same route: astronomers were at one time expected to work as astrological consultants, casting horoscopes for rulers and interpreting signs in the sky in terms of current political affairs. There's no necessary reason to follow Douglass' crossdating methods any more than we should follow Kepler's example of casting horoscopes---unless they work. Although the seeming effectiveness of crossdating could in principle be invalid, it has been applied so widely that we would need presented with a very strong critique before abandoning it. I'm not really qualified to discuss crossdating and C-14 calibration from a point of view of someone active in current research, but was fortunate to be sitting on the sidelines of the oak calibration work in the 80s, and just the other day Tom Harlan dropped by with the oldest known absolutely dated bristlecone sample, so will offer this as a kind of correction by proxy until any of the people who've done the real work care to comment ---Martin.