CONCERNING THE AGE OF THINGS 189
I hope it is clear, from what I have just said, that material
which contained radio-carbon 5,700 years ago and has been
uncontaminated with fresh radio-carbon since then, will now
have only half the amount of radio-carbon originally present,
and should give an "electron count" of about 7 to 8 elec-
trons per minute per gram of carbon present. In another
5,700 years the count from the same sample will be about
4 electrons per minute per gram, and so on.
Here, then, is an internal clock which seems in theory
completely reliable. It is true, it is a clock which does not
tick at a constant rate, but we know exactly how its rate
varies with time—it is an exponential clock, as the
mathematicians might say—and if we can read it, we can
tell the age of the specimen with accuracy. Why should we
not always be able to read it? What difficulties arise in
practice ?
Firstly, while charcoal, wood, shell are all suitable (char-
coal is best), bone is almost useless, because it contains little
carbon, and pottery and glass contain substantially none
—yet these latter three form the bulk of archaeological
material. Flint is useless, because it contains no carbon,
and. moreover, as I remarked at the outset, archaeologists are
not interested in the geological age of the flint itself.
Secondly, it is of the utmost importance that recently-
formed (and therefore more active) radio-carbon shall not
have had access to the specimen since its "death'',
otherwise we shall have false counts, and this is the more
important the older the specimen and the smaller, therefore,
the content of original radio-carbon characteristic of its age.
A well-intentioned wrapping of cotton-wool may halve the
apparent age of a 20,000-year-old sample of charcoal when
dated by this method.
Thirdly, there is, everywhere about us, a "background"
of radioactivity or irradiation by cosmic rays or particles,
which, although at a relatively low level of intensity, will,
nevertheless, register a few particle counts irregularly each
minute on a Geiger counter. That is to say, that while, to
the 15 or 10 counts per minute from a gram of carbon in a
recently-dead specimen there will be added a few more
counts from the background, after 20,000 years the con-