The Open Border 235
from a neighboring tribe, who produced this functionally equivalent but stylistically dif-
ferent artifact. Additional information about the goals, environment, and way of life of P
and Q may partly resolve this problem, but in the absence of such information the problem
seems insurmountable: even the distinction between functional and stylistic features,
which is a cornerstone of EA approaches, can hardly be made on the basis of
evidence.^16
Advocates of EA have recently and tentatively started to seek solutions. One type
of solution is to use cladistic methods. Some researchers have noted that there is a similar-
ity, or even an isomorphism, between the frequency diagrams used by some archaeologists
to monitor the diversity within a class of tools and the clade-diversity diagrams used
by some biologists and paleontologists to monitor the diversity within a class of organisms
(Lyman and O’Brien 2000: 48–50). Furthermore these researchers (e.g., Lyman 2001: 77)
have noted that the problem of distinguishing functional and stylistic features—natural
selection and cultural transmission—bears a strong resemblance to the biological problem
of distinguishing analogies from homologies—characteristics that are morphologically
similar because of convergent evolution or because of common ancestry, respectively.^17
They note that biologists are able, at least in principle, to make this distinction by using
cladistic methods, in which items are classifi ed in a nested hierarchy of branches, that is,
a lineage is constructed.^18 Based on these similarities, researchers have attempted to
transfer these methods from biology to archaeology. Some preliminary results have
been obtained, for example, cladograms of projectile points found in the southeastern
United States and of ceramics from the lower Mississippi Valley (Lyman and O’Brien
2000; O’Brien, Darwent, and Lyman 2001; O’Brien et al. 2002; O’Brien and Lyman
2002).
It is unclear whether these results are more than a happy coincidence. What is more
certain is that cladistics cannot be the panacea of EA: even if there are no conceptual
obstacles, cladograms are unstable if the set of data and characteristics used to construct
them is small (a typical situation in archaeology). And the problems do not end here,
for constructing a cladogram is only a fi rst step for evolutionary archaeologists.^19
Suppose that, after entering a set of artifact characteristics, the data-processing software
comes up with a single conjecture of their phylogenetic relations—what O’Brien and
Lyman claim to be the case for the projectile points. And suppose that we establish
that all artifacts are functionally equivalent—as seems to be assumed by calling all the
classifi ed items “projectile points.” Even on these assumptions, the underdetermination
problem described earlier remains unsolved. The reason is that the phylogenetic relation
is still compatible with a large number of very different cultural-transmission processes:
faulty copying, faithful transmission with a purely stylistic variation, faithful transmission
with adaptation to a perceived change in the environment, protoindustrial espionage,
and so forth. This problem becomes manifest when Lyman and O’Brien discuss an
evolutionary model for vacuum-tube radios in the early twentieth century, following