[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: orion 14C and falsifiability

At the risk of displaying my ignorance, let me enter the 14C debate by
offering and asking for some clarifications.

I have no particular expertise in 14C testing, but I think I understand the
physics and statistical matters in a rudimentary way. So the following,
quoted by Greg Doudna, makes perfect sense to me:

> "Little reliance should be placed on an individual 14C date to 
> provide an estimate of age for a given object, structure, feature, or 
> stratigraphic unit.  A critical judgment of the ability of 14C data 
> to infer actual age can best be made with a suite of determinations . 
> . . Concordance of values on different sample types . . . from 
> well-defined stratigraphic contexts provides one of the strongest 
> arguments for the accuracy of age assessments based on 14C values" 
> (R. E. Taylor, _Radiocarbon Dating_, 1987, p. 105).

However, it seems to me that the key words here are *well defined
stratigraphic contexts*. Although the DSS come from the same general
archaeological context, there is no way to be sure of how long a span of
time there was for either their production or deposit. So, it is *possible*
that one or more individual mss. have a date that it is quite different
from the average. And it would take only one *confirmed* later date to move
the _terminus ante quem_ for the *deposit* of the collection to a later time.

"Possible", of course, does not mean "confirmed" or even probable. So, the
substance of the argument - as I have understood it - runs as follows:

1. One side says: Don't put too much weight on a single datum far from the
others; it *could* be an outlier.

2. The other side says: Don't be so quick to dismiss any datum that doesn't
fit a certain theory; it *could* be an example of a later ms.

Now, both sides here have, in effect, agreed that there is not enough
information yet to allow any confident conclusion. However, in the heat of
the argument, both sides have argued *as though* the other side claimed
that their position was virtually certain. And I suppose some statements
made did sound a bit too confident; but in general, both sides seem to have
been arguing against something of a caricature of their opponents' arguments.

Is my paraphrase reasonably accurate so far?

Now I'd like to ask a bit of further clarification. If I understand it
correctly, the Copenhagen tests now underway, to which Fred Cryer
frequently refers, will help to resolve the questions in two ways:

a. These tests will attempt to control for possible castor oil
contamination and to determine how much such contamination might have
thrown previous 14C dates off.

b. By adding to the database of reliable 14C dates of DSS, these tests will
increase the statistical confidence of conclusions that might be drawn. For
example, if the deposit date of the collection were later than most of the
14C dates that we have, then, the larger the database, the more likely that
more than one or two of these later dates would turn up. If, on the other
hand, as the database grows larger, more later dates do not turn up, then
it is more *probable* that the isolated values are, in fact, outliers.

Is this an accurate understanding?

Paul Sodtke
London, Ontario, Canada