Edinburgh University Press, 1997. ISBN 0 7486 0791 9.

Review by Andy Tyrrell.

When I was a second year undergraduate, I can remember being assigned to
read and summarise Stephen Shennan's first edition of *Quantifying
Archaeology* (1989) over my summer holiday. This was, unfortunately,
because I had failed to submit my end of year project for the 'Computers in
Archaeology' course, which I had been obliged to take. At the time I was
somewhat upset (granted, mostly with self pity) that I had been given this
monstrous task. After all, the 'Computers in Archaeology' course had
consisted mostly of constructing databases and this book was all about
statistics (optional groan of sympathy). I duly headed to the book shop in
Russell Square and selected *Quantifying Archaeology* from the shelves along
with the other core texts for my final year. With two weeks of summer left,
I started reading Stephen Shennan's book and found myself actually enjoying
the style he utilised to approach a topic with which I had absolutely no
familiarity. I completed my summary, handed it in and left the book on my
shelf for the next three years.

I think I probably owe quite a lot to those two weeks, so it was with some pleasure that I came to re-read the new edition of this work. Stephen Shennan and Edinburgh University Press have been timely with this 're-release of a popular classic'. The increase in the use of computers in archaeology has probably exceeded even the most grandiose predictions of the middle 1980's. The more important consequence of this is that the amount of people using computers, especially for statistical procedures, who are entirely ignorant of what legal types might term 'due process', and who are also unaware to some extent of the potential of quantitative methods, has increased at an even steeper gradient. It is with the increasing availability of statistical packages, spreadsheets and databases that statistical techniques have become available to all. This is, I think, a 'good thing'. However, a concomitant lack of understanding of the procedures being undertaken has meant that gross errors of misapplication have become increasingly common.

The new edition of *Quantifying Archaeology* can only help to rectify this.
New topics have been covered and the main text has been substantially
rewritten. Even Monte-Carlo Markov Chain simulations are broached, although
admittedly not in any great mathematical detail. For those interested
solely in this topic, I would concur with Shennan's choice of Manly (1991),
whose account is both simple and useful. Although only treated to an
outline, the very inclusion of Markov chains is quite a feat for an
introductory guide and goes some way to illustrating how well the book is
written for people with no statistical knowledge. Other new subjects
include a much fuller treatment of multi-dimensional
methods, including Principal Components Analysis, Cluster Analysis and
Factor Analysis, and an expanded section on regression and correlation. All
are clearly covered with numerous well-worked examples. The easy
approachability of the text and its good structure allow it to be read as a
whole, unsurprising in a book almost certainly designed originally to
accompany the Southampton quantitative methods courses, or in part, as a
reference for novices requiring information on how to treat particular
statistical difficulties in archaeology.

If I could change anything, it would be to put the chapter on sampling
earlier in the book. This subject is so crucial for assessing how much
emphasis to put on quantitative methods, that it appears a pity to relegate
it to the end. I can, however, understand the reasons why this has been
done. The exhaustive reader will of course gain much more from a discussion
of sampling with the good foundation of statistical techniques gained from
previous chapters. However, I would argue that the exhaustive reader is
already likely to be appreciative of sampling problems, while it is the
'pic 'n' mix' reader who needs to be made more aware of biases in the raw
data. My only other gripe is that it might have been useful to have
included a table to help illustrate which type of problem is usefully
investigated by which relevant statistical technique. Of course, full
representation of this sort would fill a book in itself, but perhaps
something along the lines of the cover tables in *Biometry* might have been
useful.

Finally, the big question is whether one should buy the new edition if one already has the original. I think the answer is a tentative yes. However, if you are a total novice to archaeological statistics, then you should definitely get hold of a copy of this book and read at least the first ten chapters and probably the last as well. Even if you have no apparent need for them in your own research, you will need to understand statistics, if only to interpret better the arguments of others. If you 'blindly' use computer statistical packages, then this book is also for you; it will help to prevent you making the same mistakes as others have made before you. If you are already accomplished in the use of statistical methods, then you probably have no need to buy this book, save only to admire the excellence with which someone has managed to communicate a subject which most people would happily confine to the 'black box'.

*Andy Tyrrell is a Ph.D. student studying Anglo-Saxon population dynamics at the Department of Archaeology and Prehistory, University of Sheffield.
*

© **Andy Tyrrell** 1997

Go on, e-mail **assemblage** today!

© **assemblage** 1997