[meteorite-list] RePost: Hunting (and radiometric dating) shortline

From: Kelly Webb <kelly_at_meteoritecentral.com>
Date: Thu Apr 22 09:44:43 2004
Message-ID: <3ABEFE80.798223CB_at_bhil.com>

{If the original of this didn't wordwrap, here's one that should. If the
original did wordwrap, delete.)

Hi, Steve,
    The usual method of dating the terrestial age of a meteorite depends
on comparing the amounts of various unstable isotopes produced by cosmic
ray exposure. When the stone is in space, cosmic radiation produces a
number of unstable isotopes, which are both continuously decaying
according to the half life of each and being continuously produced. When
the stone lands on earth, the cosmic radiation is shut off and with it,
isotope production, but decay continues. By comparing abundances of
short-lifed with long-lifed isotopes in a given stone with other similar
meteorites of recent arrival, the time-on-earth can be estimated from
the shortfalls in short-lifed isotopes.
    Presumably, the comparison stone for Lafayette is Nakhla, about
whose fall date there is no dispute and whose estimated CRE age is
nearly identical (11,600,000 years vs. 11,400,000 years). Presumably,
there are adjustments that were made for the differences in bulk
composition of Nakhla and Lafayette, which, while they are similar, are
not identical. For the best results, the levels of as many different
decaying isotopes as possible should be measured. The procedure works
best for the oldest (in time-on-earth) stones. For a stone to be dated
at 13,000 years time-on-earth, we are probably talking about shortfalls
in most isotopes of at most only 1% or 2%.
    Then, there are relevant factors which cannot be known. The
penetration of even very energetic cosmic radiation falls off sharply in
distances of less than a meter of stone. How do we know where in the
original unablated meteroid the portion that became Lafayette (or Nakhla
either for that matter) came from. Answer: we don't. The geometry
(shape) of the original meteoroid is also a major consideration. These
factors alone could probably account for small differences in
     Another annoying item is that so many references (like the
Catalogue) usually give these dates without the requisite +/- of error
range. This can be very misleading, as it suggests a precision that is
illusory. Got to see those error bars. Even with the precision noted,
however, there's still another problem.
    That other part of the problem is the confusion between precision
and accuracy. Atomic assay can be both very precise and of poor accuracy
at the same time. No conflict at all. This is not well understood. While
the time-on-earth method uses short lifed isotopes, in general, the most
used isotopes for cosmic dating are long lifed (it's an old universe)
like Rb/Sr decay.
    I've seen very precise determinations of Rb/Sr isotopes that yielded
formation ages for a sample that were negative, that is, the stone
wouldn't form until a few million years from now. Not accurate, but the
results were very precise, fixing its future formation time to within
50,000 years! Of course, what we're actually dating is not the sample
itself, but some event (unknown) which produced a fractionation of a
volatile (the Rb) that left behind a refractory (the Sr). When you get a
really anomalous date, like a negative date, you throw it out because
obviously there were other events (equally unknown) in the life of the
sample, so you can no longer be certain you are dating a single unique
event anymore.
    However, this also means that there may well be undetected
fractionation events in seemingly "good" samples that have pushed the
values up and down and back and forth in undetermined and unforeseen
ways, in effect smearing out the values, blurring the data. This would
show up as variations in the values found in supposedly "identical"
samples, which often happens.
    Say, for example, that you had ten samples of objects you knew were
equal aged, like stones from the same strewn field. You might get
Sr87/Sr86 ratios whose precision was +/- 0.00005 for each individual
item (very precise), but the value of the ratio for each item might
differ from other items in the same assemblage by +/- 0.001.
    So, while your precision is one part in 20,000, your accuracy is
only equal to the total spread in sample values times the half-life of
the decay. Since Rb87/Sr87 decay has a half life of 48,800,000,000 years
(yeah, that's 50 billion years), your accuracy in this imaginary case is
+/- 50 million years (+/- 0.001 x 50,000,000,000), which is a much
larger margin of error than what the precision suggests.
    Pinning down an age somewhere in a 100 million year span is accurate
or not depending on the span being dated. If you're talking about the
formation age of the solar system, that's moderately accurate. On the
other hand, if the sample were relatively young, like Australian
tektites, this level of "accuracy" would not cast much light into the
    Even worse is the fact that we cannot be sure that variations in
"identical" samples are due to a variety of minor fractionation events
in the life history of the sample. The variations may have been there
from the beginning and instead of having been altered, the sample may
have been protected from minor events that would have obliterated its
initial variation! There's simply no way to tell.
    Next, of course, the data is most often presented in a notation
(like sigma's) or a format (log graphs) that is exponential, because
nothing makes unruly data lie down flat and neat like reducing it to
order of magnitude values. Like if I owed you $99.00 but I paid you $10
because I was doing my accounting by order of magnitude.
    At any rate, if you wanted to ignore Lafayette's supposed 13,000
year date as a practical matter, you'd probably be on reasonably good
    Radiometric dating is not a flawed model. The model (the random
decay clock) is perfect. But it's a sloppy universe, very messy, and
time is long and full of disasters. Radiometric dating has to be
evaluated within its limitations and constraints, compared with data
from as many other dating techniques as possible, and, most important of
all, be considered in context, and contexts are often best established
by other means.

Sterling K. Webb
Received on Mon 26 Mar 2001 03:32:00 AM PST

Help support this free mailing list:

Yahoo MyWeb