Well, we certainly need more than a basic quantification method to fully
exploit the potential of our data - and I do think that we all agree on that
precise point (by the way, I hope this MNI issue won't become a way to draw
a line in the community between the ones who "analyse" and the ones who
"think"... the views recently expressed sound a bit dangerous in this
respect). But we should all be aware of the fact that MNI - partly because
it is a processed and not an immediate data, but because of other problems
as well, as the non-linear character of the MNI/NISP relationship, has to be
used with caution especially when the NISP manipulated are in very different
orders (in the case of comparison between samples, though the same problem
arises with the representation of different taxa in a same sample). The main
problem of MNI (whichever be the formula adopted to evaluate it) is that we
control very badly its relationship to the original number of individuals
which have contributed to the formation of the sample, and we control a bit
less badly the relationship between NISP and this original number : this
relationship is mostly shaped by two factors : the fragmentation/destruction
of faunal elements and the number of anatomical units naturally present in
one individual, the later varying of course according to the species under
consideration. For a given species and a given state of preservation, the
relationship between the original number of individuals and the NIPS should
be linear. This is not the case with MNI which is tightly linked to the
probability of finding a certain number of times the same element (if you
use frequency to elaborate NMI) in a sample of a certain size (this
relationship is not linear). On top of this, and dispite what is comonly
said, MNI DOES NOT suppress the bias created by fragmentation - in some
respects, this bias only has opposite effects on MNI than on NISP.
In short, NISP is wrong, but we know how far it is wrong, and -more
important- the error is constant for a given species and a given state of
fragmentation (a major point for comparisons - sorry, I'm a bit stucturalist
here - French background!), MNI is perhaps more attractive, but less
accurate. This is not a matter of belief or opinion, but of calculation. For
the rest, all has probably been said by the many authors who have tackled
the subject during the past 20 years : Ducos, Poplin, Grayson, Reichstein...
Dr Anne TRESSET
ESA 8045 du CNRS "Archéozoologie et Histoire des Sociétés"
Laboratoire d'Anatomie Comparée
Muséum National d'Histoire Naturelle
55 rue Buffon - 75005 Paris - FRANCE
Tel : +33 (0)1 40 79 33 30 Fax : +33 (0)1 40 79 33 14
-----Message d'origine-----
De : Alan K. Outram <[log in to unmask]>
À : [log in to unmask] <[log in to unmask]>
Cc : Archaeozoology mail list <[log in to unmask]>; Jaco Weinstock
<[log in to unmask]>; tpoc1 <[log in to unmask]>
Date : mercredi 14 juin 2000 17:11
Objet : Re: quantification, again
>
>I'm definitely feeling the need to be provocative now!
>
>I absolutely agree that there is a need for good
>standardized descriptive quantification methods. Some
>basic issues can be resolved with those methods alone.
>Other questions can be answered from qualitative data.
>However, if one wishes to seriously attempt to reconstruct
>past economies and animal husbandry etc. (i.e. answer
>interesting questions) then some form of Minimum Numbers
>method and some indices are essential.
>
>I get the feeling that there are two major camps in
>zooarchaeology, and I would suggest that those who refuse
>to accept the need for interpretative methods as well as
>descriptive methods should really call themselves "faunal
>analysts" or something like that. Somebody has to
>interpret the data at some point. It is no use just
>describing the assemblage hoping that a general
>archaeologist is going to be able to do the interpretation.
>It is much better if the specialists, who know the problems
>better, apply the intreprative level of quantification.
>
>I would argue that zooarchaeologists (archaeologists that
>study faunal remains) have a duty to provide raw data (in a
>usable form) so that re-interpretations can be made, then
>use the best interpretative quantification methods for the
>job in hand and then form an interepretation having
>integrated other archaeological evidence (and discussed the
>key research questions with the excavator).
>
>Be just descriptive, by all means, but what's the point?
>Its the equivalent to the excavator just recording the site
>and walking off. Some body has to interpret.
>
>
>
>----------------------
>Alan K. Outram
>University of Exeter
>
>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|