Middle-Range Theory in Historical

Archaeology
Peter Kosso*
1. Introduction: Conceptual Background
EVIDENCE in archaeology,
since it is an informational
link between the unobservable past and observable
data in the present, must be accountable
to
justification
that the link is secure and accurate. The same accountability
is
true of evidence in the natural sciences where epistemic responsibility
requires
an understanding
of the connection
between the manifest data (streaks in a
bubble chamber, for example) and the theorized entities (alpha particles) of
which the data are evidence. In the archaeological
case, Lewis Binford has
suggested that the informational
link is understood,
and hence the evidence is
made meaningful
and reliable, through middle-range
theories which describe
the formation process of the data. Applying this approach to the more specific
case of historical
archaeology,
M. Leone and P. Potter suggest that the
documentary
record of the past and the material record could be used as
independent
sources of middle-range
theories for each other.* They indicate
though, that this epistemic opportunity
in historical
archaeology
is rarely
exploited.
The purpose of this paper is to present an analysis of evidence in a particular
case of historical archaeology to show the middle-range
theories in action. By
focusing on how the evidential link is made between observational
data and
the objects and events of theoretical interest it will be shown that middle-range
theories are almost always implicit in the evidence, and that they can be
articulated
and evaluated
for their independence
from the theories their
evidence is used to test. It is this independence
that allows the theoryinfluenced
evidence to be objective evidence nonetheless.
Furthermore,
the
implicit status of the middle-range
theories in the case at hand motivates the
suggestion that documentary
and material information
participate more often

*Department of Philosophy, Northern Arizona

6011, U.S.A.
Received 19 August 1992;

University,

PO Box 6011, Flagstaff,

AZ 8601 l-

in final form 21 October 1992.

L. Binford, General Introduction, in L. Binford (ed.), For Theory Building in Archaeology

(New York: Academic Press, 1977). p. 7.
2M. Leone and P. Potter, Issues in Historical Archaeology,
in M. Leone and P. Potter (eds),
The Recovery of Meaning (Washington:
Smithsonian
Institute Press, 1988).

Stud. Hist. Phil. Sci. Vol. 24, No. 2, pp. 163-184,

Printed in Great Britain

163

1993.
0

003%3681/93
S6.00 + 0.00
1993. Pergamon Press Ltd

164

Studies in History and Philosophy of Science

in such a reciprocal relation than Leone and Potter indicate. The middle-range
theorizing simply goes without saying.
This is not to be a discussion of the structure of confirmation. It is not about
the nature of the connection between the evidence, once youve got it, and a
hypothesis. It is rather about the nature of the evidence itself and its relation to
theoretical claims invoked to enhance the informational value of data.
As a case study, this is an empirical contribution to a description of
archaeological and scientific evidence, a description that requires a component
of conceptual analysis as well. Like science itself, philosophy of science must
offer both general theorizing and specific evidence. The specifics of this case
will be more meaningful if presented in their theoretical context, that is, with a
preliminary sketch of the conceptual analysis it is intended to complement. To
this end, I now briefly outline the theorizing which will help structure the
subsequent evidence.3
The accomplishment of theorizing, in science as in our everyday knowledge
about the world around us, is to make expansive claims about what is
happening or has happened behind the scenes, beyond what is manifest in
experience. Theories make claims about unobservable objects and events of all
sorts, the tiny, the distant, the cumulative, the past. Evidential claims which
function to test a theory must be relevant to these objects and events
mentioned by the theory, and to the accessible, particular objects and events
which are mid-sized, here and now. A complete picture of evidence then, must
include claims, theoretical claims, which describe the link between accessible
objects and the phenomena which make contact with theory. It is an appeal to
theory, to revive a previous example, which allows the physicist to call these
streaks in the bubble chamber indicators of alpha decay. These streaks mean
alpha decay.
The claims which add this kind of meaning and relevance to the data are
often descriptions of how the data are formed. In describing the causal process
which led from phenomenon of evidential interest (alpha decay) to the final
observable event (streaks), the data are shown to be an image of the phenomenon. Elsewhere4 I have referred to these claims which reveal the evidentially
relevant information in observations as accounting theories, indicating that
scientific evidence is not simply an episode of observation but an observation
that is accountable in the sense that the scientist, or at least the scientific
community at large, must know the circumstances of formation of the observation to know both what it means - that is, in what sense it is relevant to the
particular theory for which it is evidence - and to know that it is reliably
The full treatment
of this conceptual
analysis can be found in P. Kosso, Science and
Objectivity,
The Journal of Philosophy 86 (1989), 245-257, and P. Kosso,
Method
in
Archaeology:
Middle-Range
Theory as Hermeneutics,
American Antiquity 56 621-627.
KOSSO, Science and Objectivity, ibid.

Middle-Range Theory in Historical Archaeology

165

linked to the phenomenon it describes. Evidential knowledge, like theoretical

knowledge, must be accountable. The accounting claims could alternatively be
thought of as imaging theories in the case of using an instrument such as a
microscope to link a specimen of interest to an observable image. In the
context of archaeology, Lewis Binford identifies the claims which give meaning
to the static archaeological record as middle-range theories. As Binford puts it,
middle-range theories describe the formation process of the archaeological
record.5 This informational link between present and past is a necessary
component of archaeological evidence since only through an accurate understanding of such processes can we reliably give meaning to the facts that
appear, from the past, in the contemporary era.6
Regardless of what you call them, the point here is that any accountable
claims which go beyond the episodic data will require something like middlerange theories, accounting theories. These theories will not always, in fact not
usually, be explicit in the presentation of evidence, but they must always be
available and amenable to articulation. This is to say that while it may be too
strict to insist that one must in fact give justification for all scientific claims, it
is a basic responsibility that the scientific community must at least have
justification available, even for evidential claims.
This conclusion is based on principles of what theories do and what is
required to make observations relevant to theory. We need to see now that in
fact, not only in principle, this is the case and that middle-range theories can be
seen in action and can be articulated. Since the conclusion in principle is very
general, dealing with any activity of theorizing beyond the apparent data,
accounting theories or middle-range theories (I use the terms interchangeably)
should be discoverable in cases of evidence in natural sciences, social sciences,
and even our day-to-day efforts to understand the world around us. The
exposure of middle-range theories in the context of historical archaeology is
the primary task of this paper, but it is done with the belief that similar
analysis will be as fruitful in other archaeological and scientific cases.
There is a secondary point I hope to make through the case study, a point
that will help to fit the analysis of evidence into a bigger picture of the
justification of scientific knowledge. Middle-range theories, the case will show,
are not special kinds of theories with a special kind of content. They are not
necessarily of middle-generality or of middling confirmation (nor particularly
solid confirmation). They are just regular theories that are, in the particular
circumstances that we find them here, used in a particular way.
This fits into a coherence model of justification of scientific knowledge.
Evidential claims require theories for justification, as theories require evidence
Binford, op. cit., note 1, p. I.

Ibid.

166

Studies in History and Philosophy of Science

for justification,
and the theories which justify evidence do not bring any
epistemic privileges into the account. They must pass their own evidential and
conceptual tests. There is a kind of hermeneutic circularity in all of this, in that
the specific, particular
reports of observation
are guided by more general
understanding
of theories, while the theoretical claims are themselves shaped
by the specifics of evidence. But it need not be a problematic
circle insofar as
we can insist on an independence
between the theories which influence a
particular
observation
and the theory for which that observation
is evidence.
All evidence is guided by theory, but not necessarily by the same theory it is
used to test.
But this is a big, conceptual
issue, and the programme
here is to consider
some particular evidence. The bulk of the exposition is to be done through an
example in historical Greek archaeology,
where evidence is sought about the
colonization
by Classical Athens.
Evidence and theory in archaeology do not make for a typical case study in
the philosophy
of science, but with the reciprocating
relation between theory
and evidence as will be shown here, it is wise to look for similarities between
the status of evidence in archaeology and in natural science. Similarities in the
nature of observation
in archaeology and natural science were pointed out by
John Fritz even before Binford introduced the concept of middle-range
theory.
Fritz classified prehistoric events and objects as being indirectly observed in the
same sense as distant galaxies or atoms are indirectly observed. Much of what
follows here will be built on Fritzs foundation.

2. The Form of the Analysis

Since the goal of this case study is to demonstrate
the work done by middlerange theories and to assess the relationship
between these and the object
theory, the key to the analysis is to identify and articulate the participating
claims in each presentation
of scientific evidence. The example will begin with a
statement of its hypothesis, the theoretical claim being tested by the evidence.
A bit of background
to the hypothesis, its importance
to the science and the
motivation
for its proposal, will be provided to locate the example in its larger
intellectual
setting. Then the evidential claims will be presented. We will not
uphold any significant epistemic distinction
between particular claims which
function as antecedent motivation
for the hypothesis and those in the role of
subsequent
evidence. Whether an observation
takes place before or after the
actual proposal of a hypothesis, the observation
can still function as evidential
support.
J. Fritz, Archaeological
Systems for the Indirect Observation
of the Past, in M. Leone (ed),
Contemporary
Archaeology (Carbondale:
Southern Illinois University Press, 1972), p. 136.

Middle-Range Theory in Historical Archaeology

167

The example analysed here is an on-going project. It is not a closed case, and
so the evidence will be both actual, observations that have already been made,
and potential, relevant observations proposed but yet to be carried out. In
either case it is important to know of the theoretical influence which makes the
observation relevant to the hypothesis. Of these evidential claims, care will be
taken to identify two kinds: those which describe the manifest state of affairs,
the data, and those which describe the phenomena which are not manifest.
The final set of claims to be identified in the case will be those theoretical
claims used to link the manifest data to the relevant phenomena. These will be
the background information employed to tell us what the observations mean
-that
is, what they indicate about the hypothesis. These are the middle-range
theories, or the accounting theories, which attest to the relevance and accuracy
of the observations. These are what make observations into evidence.
Once all these components of the evidence and testing have been identified,
the analysis will continue by assessing the relation between the hypothesis, that
is, the theory being tested, and the accounting theories. In particular, we will
evaluate the independence between these two components to check for any
covert circularity in the presentation of evidence. Finally, each case will end
with a discussion of the negotiation between theory and evidence. When things
go wrong and evidence does not match hypothetical expectation, something
has to give. With the large cast of participants in each evidential claim there is
apparently much work to be done in deciding which to revise or reject.
Initially, none are immune to revision and the ensuing give-and-take between
evidence and theory is a necessary concluding step in each case.

3. The Cleruchy Hypothesis

This is an example of using evidence to test a theory in the domain of
historical archaeology. Historical archaeology, in contrast to prehistoric
archaeology, is a study of places and people in times past for which there is a
written record. Like all archaeology though, the technique of historical
archaeology is to use the material remains, artifacts and lasting impact on the
environment, as evidence for claims about past behaviour.
The particular example to be analysed here deals with Classical Athens and,
more specifically, with Athenian settlements outside of Attica. As understood
from ancient inscriptions and from ancient authors, the traditional view
classifies Athenian foreign settlements into three broad and somewhat flexible
categories. The distinction is based on two things: the citizenship status
(including military obligation) of the people in the settlement, and the political
Much of what follows about
spondence with Mac Wallace.

these settlements

I have

learned

with much

help from

corre-

168

Studies in History and Philosophy of Science

autonomy of the settlers (including rights of ownership of land). Settlements

differed, in other words, in terms of the occupants obligations and rights with
respect to the homeland.
One type of settlement was the garrison, essentially a military outpost.
Residents of a garrison were given no land, either to work or to own. They had
virtually no rights of self-rule and they were fully obliged of military service in
the interests of Athens.
At the other extreme was the Athenian colony. Colonists generally owned
their own land in the settlement and were free to empower their own governors. They did not owe military service to Athens, but interpretations differ as
to whether they retained Athenian citizenship.
The third sort of settlement, the cleruchy, was a kind of hybrid between
garrison and colony. Cleruchs were given land to farm, though the land was
not freely disposable. They retained their Athenian citizenship and were taxed
accordingly. They were also obliged to Athens for military service. Cleruchies
seem to have been a sort of paramilitary community, an Athenian presence on
foreign soil, there to keep an eye on unwilling allies, a residential extinguisher
of brushfires of discontent. It has been suggested9 that cleruchs were chosen
from the lower classes of Athenian citizens. Service as a cleruch may have been
marketed as an assistance to the poor - the use of land in exchange for
minimal military service - but deployed in fact as a way to decrease the
population of tiff-raff at home and to control unruly neighbours at the same
time.
Many of the details about cleruchies are unknown. There are questions
about the specifics such as exactly where they were established and when. And
there are more general questions such as whether the cleruchies were generally
situated on good farm land or poor, and whether the group of cleruchs were
grouped in a defensible, perhaps even fortified, community, or spread across
the land of their hosts. The written record has so far been uninformative on
these points.
The issue of the Athenian cleruchy is relevant to studies by the Southern
Euboea Exploration Project (SEEP), a group of archaeologists at work on a
regional study of, not surprisingly, the southern third of the island of Euboea.
(Map 1 shows the island of Euboea relative to the mainland and other Greek
islands. Map 2 below shows southern Euboea, the area of study.) Several
seasons of work have been focused on the cleruchy hypothesis which is, in its
most basic presentation, the suggestion that there was a cleruchy on southern

%ee, for example, R. Meiggs, The Athenian Empire (Oxford: Oxford University Press, 1972).
p. 121.

Middle- Range Theory in Historical

Archaeology

169

Map I.

Euboea. It is a determinate claim about specific events in the past in the sense
that the concept of a cleruchy is reasonably well distinguished and the region
of southern Euboea is a distinctive region due to its isolation by mountains
from the rest of the island. It is a theoretical claim in the sense that it describes
unobservable objects and events.
The motivation for proposing the cleruchy hypothesis and for spending time
and money to test it is from two sources, written evidence (history) and
material evidence (archaeology).
Consider first the historical record.
Herodotus and Thucydides both mention cleruchies but make only one
reference each, and neither is in southern Euboea. Herodotus writes of a
cleruchy at Chalkis in central Euboea, while Thucydides tells of a cleruchy at
Mytilini on the island of Lesbos, near the coast of Asia Minor. Later historians
add more accounts of cleruchies. Diodorus Siculus, a Sicilian writing four
centuries after the fact, writes that Athens dispatched 1000 cleruchs to be
divided into three cleruchy communities, one on Naxos, another on Andros,

Much of this work was conducted under the title of the Canadian Karystia Project, work that
was staffed and directed by SEEP personnel. Descriptions
of the work and the cleruchy hypothesis
can be found in D. Keller and M. Wallace, The Canadian
Karystia Project, Ectros Du Mona?
Classique/Classicaf
Views 30 (1986). 155-I 59; D. Keller and M. Wallace, The Canadian Karystia
Project, 1986, Ethos Du Monde Classique/Classical
Views 31 (1987), 225227; D. Keller and
M. Wallace, How to Catch a Cleruch, Canadian Archaeological Bulletin, Fall (1987), 6-7; and
D. Keller and M. Wallace, The Canadian Karystia Project: Two Classical Farmsteads,
Ethos Du
Monae Classique/Classical
Views 32 (1988), 151-l 57.
Epistemological
questions about the historical record, questions of accuracy and trustworthiness of the accounts, will be taken up later. I do not intend for any of these historical reports to be
believed unequivocally,
but it is helpful at this point to present who said what about cleruchies.

170

Studies in History and Philosophy of Science

and a third on Euboea. Plutarch, in the Life of Pericles, mentions the sending
of 500 cleruchs to Naxos and 250 to Andros. Though no ancient author says
explicitly
how many cleruchs went to Euboea,
we can put together the
information
from Diodorus and from Plutarch to conclude that the remaining
250 cleruchs of the 1000 reported by Diodorus went to Euboea. Furthermore,
while the northern
two-thirds
of Euboea were culturally
and economically
most closely associated with the mainland of Greece, the southern third of the
island participated
more with the Cycladic islands, like Andros and Naxos.
This is reason to think that a dispatch of cleruchs to Andros, Naxos. and
Euboea might well be a mission to the Cyclades, that is, to Andros, Naxos, and
southern Euboea. Thus, the inference from the historical
sources gives some
reason to think that Athens sent 250 cleruchs to southern Euboea. There is
some reason, that is, to entertain the cleruchy hypothesis.
Interestingly,
the evidence for southern
Euboeas Cycladic connection
is
largely archaeological
evidence. Much of the prehistoric
pottery found in
southern Euboea shows similarity to Cycladic pottery and can thereby be seen
as indicative of interaction
with Cycladic islands. The architecture
of modern
houses in southern Euboea also shows distinct similarities to Cycladic styles
and dissimilarity
from that of the northern
two-thirds
of Euboea.12 This is
interesting for an analysis of the nature of the evidence because it is a case of
archaeology
helping to make sense of the historical evidence. Through an
independent
analysis of ceramic remains, the informational
content of the
textual data is enhanced. We see more fully what the text means in light of the
background
knowledge about pottery.
These historical accounts might be thought of as direct references in that
they mention cleruchies explicitly, but it would be misleading to regard them as
direct information
about events in the past. This is information
that has been
passed along by the historian,
via his sources and with his writing, through
time. There is work to be done in securing the credibility
of these ancient
writers on this particular
subject. Diodorus
and Plutarch,
after all, were
describing events that had taken place centuries earlier and their evidence is
acceptable only insofar as we can keep track of how they would know about
these things. Diodorus,
in fact, is often inaccurate
in his tales of ancient
Greece, underscoring
the need for justification
of his particular claims about
cleruchies. Pausanias, also writing centuries after cleruchs had come and gone,
also makes explicit reference to the cleruchs sent to Euboea and Naxos, and
Pausanias
enjoys a good reputation
for accuracy. This is due largely to his
frequent description
of landscapes and monuments,
features of the land and

%x,
for example,
D. Keller, Archaeological
Survey in Southern
Euboea,
Greece: A
Reconstruction
of Human Activity from Neolithic Times through the Byzantine Period, (unpublished Ph.D. Dissertation,
Indiana University,
1983, pp. 64465.

Middle-Range Theory in Historical Archaeology

171

culture that are preserved today and can be checked against his account. Here
again, archaeology, the study of the material record, is used in accounting for
historical evidence.
There is another complication in reading the written evidence for cleruchies,
even when the writing is explicitly about that kind of settlement. The ancient
authors did not always use the same word to refer to a cleruchy. Thucydides,
for example, used the term apoikia (~UCOW~~U)
rather than cleruchy (~S~pon~os)
in his account of the settlement at Mytilini,n but scholars know from the
context what he meant. He meant to describe a cleruchy.
The point in mentioning these uncertainties in the written record is one that
is obvious to historians. The historical text is not a direct access to the events
of the past. Its not even close. Evidential claims about the past which come
from a written record are always in need of accounting in the sense of requiring
some verification of their accuracy and reliability, and of requiring an interpretation to pinpoint just what the text means, for as Kuhn points out, there
are many ways to read a text.14
Besides the ancient sources which explicitly refer to cleruchies, there is
implicit textual evidence that can be finessed from accounts of other, related
phenomena. For example, the Athenian Tribute Lists, inscriptions found in the
Athenian Agora, listing the taxesI collected from the communities in the
empire, can be used as a source of information about cleruchies. In the year
450 B.C., Andros paid 12 talents in taxes to Athens. (One talent is equal to
6000 drachmas, a drachma being a days wage.) The subsequent year, 449,
Andros paid only 6 talents. This abrupt reduction can be explained by the
establishment of a cleruchy on the island. If the cleruchs from Athens seize a
large proportion of farmland, the commercial value of the island, as realized
by the pre-settlement residents, is diminished and the locals, in all fairness, are
assessed a lower tax. In this way, establishment of a cleruchy causes an abrupt
decrease in payment of talents to Athens, allowing historians to use the
Athenian Tribute Lists as a rough indicator of cleruchies.
The Athenian Tribute Lists record for Karystos, the principle town in
southern Euboea, a payment in 450 B.C. of 7+ talents, and in 449 a payment of
5 talents. There is no corresponding drop in the assessment for Chalkis or
Eretria, the major towns in the northern two-thirds of the island. This suggests
a cleruchy in the domain of Karystos as a cleruchy would explain the reduced
tax assessment and would fit nicely with the coincident reduction on Andros.
There are no tax records for Naxos prior to 447 B.C., but the assessment in
See A. Graham,
Colony and Mother City in Ancient Greece (Manchester:
Manchester
University Press, 1983), p. 170.
l4T. Kuhn The Essential Tension (Chicago: University of Chicago Press, 1977), p. xii.
ISThe ATd actually lists the Quota, one sixteenth of the taxes paid. I am grateful to Mac Wallace
for clarification
of this point.

172

Studies in History and Philosophy of Science

that year is quite minimal, an assessment that is compatible with the theory
that Naxos, Andros, and southern Euboea were all saddled with cleruchies in
the year 450/449 B.C.
Further implicit written evidence of a cleruchy in southern Euboea is in the
indication that Karystos did not participate in the Euboean revolts against
Athens of 446 and 424 B.C. It is as if the Karystians were kept under control
by resident Athenian loyalists. This, like the rest of the textual evidence
presented so far, is surely not conclusive proof of a cleruchy in southern
Euboea, but it is motivation for the hypothesis.
There is also archaeological evidence which makes the hypothesis plausible.
A recent archaeological survey by Don Keller of the Karystian Bay
watershedI revealed an apparent network of Classical-period farms on the
Paximadhi peninsula (see Map 2 below). The method of survey archaeology is
to walk over large areas of land, noting artifactual remains that are visible on
the surface. There is no excavation involved. From the surface deposits of
ceramics, worked stones, and remains of walls, information about the habitation and use of the land can be reconstructed. Survey information is a
valuable complement to excavation in that it allows for an account of patterns
of land use and of behaviour spread over a large area.
It is exactly this kind of pattern-of-habitation
information that turned up on
the Paximadhi. A network of sites emerged and the sites were dated as being of
Classical period (500 B.C.-320 B.C.) on the basis of the pottery remains found
on the surface at each site. (There will be more on how this is done and on the
uncertainties involved when we discuss the relevant middle-range theories.)
These sites were identified as farms on the basis of several features. Those with
remaining identifiable groundplans resemble the groundplan of farm sites in
Attica, the function of the Attic sites being known through inscriptional and
other types of evidence. The pottery on the putative farm sites in Euboea is
largely coarseware - the crude, functional stuff associated with domestic
chores - and the relatively small amount of fineware tends to rule out any
possible interpretation of the sites as being aristocratic domiciles. And finally,
the sites are located in proximity to heavily terraced land, olive presses, and
threshing floors, indicating a link between the sites of habitation and the
activities of farming. Interestingly, the Paximadhi is no longer farmed. Today
the land is marginal at best and the peninsula is barren. But the wealth of
terraces and abundance of habitation sites indicates agricultural activity in the
past, and the even spacing of the farm sites further indicates a planned
community rather than a spontaneous growth over time. If the land was
marginal in Classical times, this might explain why the tribute reduction for
Euboea was only 7+ to 5 as compared to the more impressive reduction of 12
lbKeller,
op. cit.,

note

12.

Middle-Range

Theory in Historical Archaeology

to 6 for Andros.

For this reason,

173

it would be interesting

to locate the Andros

cleruchy and evaluate the quality of its land.i8

The network of farmsites is exciting news for the cleruchy

hypothesis,

since

the material remains are not only compatible

with the hypothesis: they seem,
with further attention
to details, to indicate with greater precision where the
cleruchy was located. What the initial survey calls for though is more material
evidence and more precision. Since the written evidence indicates the presence
of 250 cleruchs, further survey is called for to find the full complement
of 250
farmsites, assuming that each cleruch receives his own farm, along with a
commensurate
number of graves, altars, roads, and the other accoutrements
of
Classical Athenian life. More precise dating of the sites and the complex is also
needed. The crux to establishing
the existence of a cleruchy is the sudden
appearance
of the entire complex in roughly 450 B.C. and its equally sudden
abandonment
in roughly 400 B.C. at the demise of the Athenian empire. Thus
a precision greater than Classical period would be helpful. It would also be
informative
to collect background
data from the surrounding
area. Only by
knowing the patterns of sites and the style and fabric of pottery nearby can it
be determined
that the Paximadhi is unusually
well-ordered
in its pattern of
settlement
or unusually
abrupt in the establishment
and abandonment
of a
complex of sites. Thus the initial survey work, while encouraging,
calls most
immediately
for more data.
There is a third source of motivation
for the cleruchy hypothesis: empathy
with the Athenians.
This is a risky technique, suitable at least in a context of
discovery, if not as justification.
The idea here is to think like an Athenian and
see if it makes sense, in light of the fragmentary
evidence so far, to put a
cleruchy
in southern
Euboea
on the Paximadhi
peninsula.
The land is
marginal, and this might make it appropriate
for a cleruchy for two reasons.
For one, there is no reason to anger the locals by confiscating
their best
farmland. Better to keep the peace and get by if you can with the less desired
land. And second, the Athenians can get by with the poor farmland since these
are only lower class, poorer citizens who end up as cleruchs. If a cleruchy is a
punishment,
or at least no reward, for the lower class of society, it only makes
sense to stick them with cheap land. I9 Furthermore,
the Paximadhi peninsula is
a strategic place to establish a military
population
centre of Karystos,
but the
shipping lane between Andros and the
Paximadhi
cleruchs could keep an eye

presence. Not only is it near the

peninsula
juts into the important
mainland.
From this one position,
on the local Karystians
and the

Mac Wallace points out (private communication)

that the Karystian
tribute was 12 talents in
453. The reduction to 74 in 450 is unexplained.
*This good idea came from Cynthia Kosso.
Neither of these reasons for putting a cleruchy on marginal land is a considered conclusion or
even a serious hypothesis of the SEEP archaeologists.
Both were casual suggestions in the course
of conversation.

Studies in History and Philosophy of Science

174

Paximadhi

Peninsula
Map 2.

important
shipments of grain via Andros to Athens. Many of the Paximadhi
farmsites have an excellent view of both the sea lands and the good Karystian
farmlands.
None of this proves that there was an Athenian cleruchy on the Paximadhi
peninsula.
But altogether the written and material remains make it reasonable
to pursue the cleruchy hypothesis in its now more precise formulation.
The
Paximadhi supported at least part of a settlement of 250 cleruchs from the year
450/449 B.C. until roughly 405 B.C. when the Athenian empire dissolved. This
is the hypothesis to be tested.
4. The Evidence
The evidence for the testing was generated by further archaeological
work as
suggested
by the initial
survey and the predictions
of the hypothesis.
Confirmation
of the hypothesis would be furthered by a linked community
of
roughly 250 farmsteads with coincident
periods of occupation
from roughly
450 to 405 B.C., and perhaps with evidence of unusually close ties to Athens.
To this end, other areas of southern Euboea, namely the vicinities of Geraistos,
Marmari, and Kallianou,
were surveyed in 1984, revealing that the pattern of
finds in the Paximadhi was distinctive. Then the remainder of the Paximadhi
peninsula
was surveyed in 1986. (Only the eastern side, the Karystian
bay
watershed, was done in the initial survey by Keller.) The 1986 survey showed a
continuation
in the regular settlement pattern, finding more sites of roughly
Classical period and apparently for use as farmsteads. The surface materials, in
other words, seemed to bear out the prediction of an extensive community
of

Middle-Range Theory in Historical Archaeology

coincident
occupation
cleruchy.

175

farmsteads.
But the surface remains could not fix the dates of
of the sites with sufficient precision to count as clear evidence of a
This precise

dating

could

only come by excavation.

By excavating

one or two of the sites, hopefully determining

the dates of the materials and
looking for signs of Athenian
material and stylistic influence, information
would become available about the entire complex of sites. Survey shows that
the sites are related; excavation
refines the information
on one or two.
Together then, there is more precise information
on the lot of them.
This is something
of an oversimplification
about the ease and efficacy of
survey archaeology.
Nearly every concentration
of artifacts, every findspot or
site, shows material from many time periods. A typical survey report will list
pottery finds of perhaps Archaic to Roman, or Classical to Byzantine.
The
demonstration
that several sites are associated and were occupied at the same
time is a complicated
one which must cite relative numbers of potsherds and it
is not without ambiguity.
The ambiguity
though is not caused by an inadequacy in the middle-range
theories. It is that the specimen itself is fuzzy, not
that the image is out of focus. Excavation
of a site can bring the information
into even sharper focus, being in a sense a higher power microscope
than
survey, and with a narrower field of view, but it will still show a fuzzy image of
a fuzzy specimen, fuzzy in the sense of having no sharp, distinct period of
occupation.
In any case, the survey data of surface material plays the role of
finding evidence of community
and of roughly dating the community.
Excavation is used for more careful dating as well as to reveal groundplans
of structures and to collect ceramic remains that might be diagnostic as to the
origin of the clay or vessel. Two sites were excavated by SEEP in pursuit of
evidence relevant to the cleruchy hypothesis. One was an abandoned
cistern on
cape Mnima (see Map 2), found during the survey to be filled with debris. The
cistern is located within the network of farmsteads
and is in a fortified and
strategic location overlooking
sheltered harbours. All indications
are that the
cistern was an important
feature of the putative cleruchy, and dates of use of
the cistern would be dates of use of the community.
The idea was simply to
excavate the cistern and determine the date of the oldest material inside. This
would be the first debris that fell in or were thrown in and not cleared out, and
would mark the end of use of the cistern as a cistern rather than a garbage pit
or a shelter from the wind. Material in the cistern, in other words, would be
evidence of the time of abandonment
of an important
site in the community.
Confirmation
of the cleruchy hypothesis would be a date of around 405 B.C.
The other site of excavation
was a putative farmsite at Palio Pithari (see
Map 2). Excavation
of this very typical site would provide more pottery, and
importantly,
more diagnostic pottery found in informative context with respect
to the building structure and the strata. It would provide, that is, much more
information
on dating this characteristic
site. It would also clarify the ground-

176

Studies in History and Philosophy of Science

plan of the building and allow a comparison to excavated farmsteads in Attica.

Again, if the hypothesis is true, one expects dates of occupation of the Palio
Pithari site to run from 450 B.C. to roughly 405 B.C. That would be
confirming evidence. Further confirmation would come from analysis of the
ceramic fabrics, finding at least some of the Palio Pithari pottery to be of a
fabric unlike that found at the local kiln and at sites away from the cleruchy. A
greater amount of Attic fineware or coarseware of Attic clay would indicate a
community with closer ties to Athens. In summary, the evidence for the
cleruchy hypothesis, the predictions of the hypothesis, would be that Palio
Pithari is indeed a farmsite with ties to Athens, in use during the second half of
the fifth century B.C.
The theory and predictions of the cleruchy hypothesis are clearly presented
in a SEEP report of 1986. Subsequent reports get right to the experimental
results.* The cistern was excavated in 1986, and the oldest debris were in fact
dated to around 400 B.C.** The Palio Pithari farmsite was excavated in 1987
with results that were not so unambiguously positive.23 The building was
cleared down to a bedrock floor, revealing a distinct groundplan which was
similar to several Classical farmsteads in Attica, having small rooms along the
north wall and a courtyard on the south. The pottery remains were dated into
the early fourth century, a bit later than would be expected of a cleruchs farm.
Some of the ceramic materials are still in the process of petrographic analysis
for comparison to Attic and to local Karystian clay.
This is the evidence to date for the cleruchy hypothesis and there are two
important points to note about the nature of the evidence and the role it plays
in confirmation. The first is that, in citing as evidence a Classical farmsite or
pottery from the fifth century, we are not citing things that we see. Thats not
what is on the ground. Whether in survey or in excavation it is not at all
apparent that a particular worn chip of ceramic is of late-fifth-century vessel or
that a certain jumble of stones is a wall of a farm house. What is used as
evidence is not what is seen, and that is exactly the nature of evidence that is of
concern here. The picture must be filled in with an account linking what is seen
to what functions as evidence. Thats the first point. The second involves the
question of the extent to which the evidence presented so far in fact confirms or
disconfirms the hypothesis. I will pursue these two points in order, spending
more time on the first.
What do the SEEP archaeologists see when they go looking for evidence to
Keller and Wallace, The Canadian Karystia Project, op cir., note 10.
2Keller and Wallace, The Canadian Karystia Project 1986, and The Canadian Karystia
Project: Two Classical Farmsteads, op cit., note 10.
zKeller and Wallace, How to Catch a Cleruch, op. cit., note 10.
Z3Kellerand Wallace, The Canadian Karystia Project: Two Classical Farmsteads, op cit., note
10.

Middle-Range Theory in Historical Archaeology

177

test the cleruchy hypothesis?

Mostly they find pieces of ceramic: potsherds,
fragments of roof tiles and loom weights. They find these things and carefully
record their spatial

context

whether

on the surface in survey or in layers from

excavation.
Very few of the finds are whole pots, and a very few more are
complete enough to make obvious, without any inferential reconstruction,
the
shape of the whole vessel. Mostly the data are small fragments of the body,
rim, or base of a broken vessel. Unfortunately,
these fragments
are not
stamped with their date or place of origin, nor are they or the places they are
found marked for their use (as, for example, being a farm or a farm tool). This
important
information,
the information
that is relevant to claims about a
cleruchy, must be somehow unnested from the more manifest. The data must
be linked to the evidence.
The first step in the archaeological
case is to record all of the useful data by
carefully describing the apparent physical features of the ceramics. This is the
pottery inventory and anyone, regardless of their knowledge of archaeology or
of the past, can do it. All it takes is average eyesight and maybe above-average
patience. In the process of inventory,
a sherd is described in a standardized
language. Its colour, for example, is compared under controlled
lighting to
patches of colour on the standard Munsell colour chart so that an intersubjective colour assessment is made. The ceramic fabric is inspected for inclusions
and the relative number, size, and colour of the inclusions is noted. The sherd
itself is measured for thickness and its shape is described using standardized
terms such as rolled rim or strap handle. And finally the surface of the sherd
is inspected for treatment such as glaze, slip, incisions or impressed design. All
of these inventory features are readily apparent on the potsherds. The inventory reports are the data in this case of archaeological
testing, but they are not
the evidence because they make no contact with the objects and events
described by theory. They are, as yet, not relevant to the phenomena
predicted
by the hypothesis.
This requisite contact comes with the next step in the treatment of the finds,
the preparation
of a catalogue. This requires substantial
background
knowledge on the part of the individual cataloguer and the archaeological
community as a whole. A sherd which has been inventoried
is now evaluated by the
experts in an effort to determine the type of vessel it is from, the use of the
vessel, and, most importantly,
its date. The type of vessel is judged by
comparing
dimensions,
fabric, and cross-sectional
shape of the sherd to
corresponding
features of whole or nearly-whole
vessels found elsewhere. The
use of the vessel can then be inferred from its shape and from the context in
which it and its comperanda
are found.
Dating the sherd is also dependent on the identification
of the type of vessel
(or in some cases on the physical features of the material or surface treatment).
The sherd is distinguished
by the vessel it is from, or by distinctive decoration
SHIPS 24:2-B

178

Studies in History and Philosophy of Science

or glaze, and can be then associated with similar pieces that have been found
elsewhere in datable circumstances. Some excavations, notably in Athens at
the Agora and the Kerameikos, and at Corinth, have uncovered specimens of
pottery in context with dated inscriptions, with coins, or associated with
buildings that are datable by historic reference. From these excavations come
source-books of pottery types and their dates. This punctuated information
can be filled out through studies of stylistic evolution and through seriation
studies, attending to relative amounts of ceramic styles as a function of time.
The accuracy of dating these reference materials is of course dependent on a
thorough understanding of their context. An object kept for generations as an
heirloom, for example, might eventually be discarded with a coin of much later
period. A casual analysis of the tailings of an archaeological dig, for another
example, might find a beer bottle next to an overlooked neolithic potsherd, but
it would be a mistake to date one artifact by its association with the other.
Thus, the datable circumstances of the Athenian Agora require an understanding of the texts and buildings, including the intent and accuracy of the
writing of the texts found with the pottery or of texts about the buildings, and
the circumstances of deposition.
By attending to the comparable features of the pottery found on the
Paximadhi peninsula and by matching these features to pottery found in
datable contexts, the catalogued sherds bear information of their own dates information, that is, of activities in the past. Soft, powdery, orange fineware
with black glaze, for example, is indicative of activities in the Classical period.
This kind of pottery means occupation of the site during that time period. But
it is so meaningful only with the support of the antecedent claims which make
the general connections between the ceramic data and the evidential dates. In
this case these are claims about the coincidence of pots and texts or pots and
historic buildings in the Athenian Agora, the dating of coincident inscriptions
or coins, and the accuracy of this dating as secured by knowing the circumstances of deposition, the intent of the writing (that it is not itself a fiction) and
the like. Its not that these claims are suspect or complicated. The reason for
pointing them out is simply to show that they are being relied on and that they
are, at least in part, claims about the past. They are, in the archaeological
context, theoretical claims. And since they are being used to make the
connection between the manifest archaeological record and the information it
has to offer about the past, these are middle-range theories. They are used to
tell the SEEP archaeologists what the sherds mean for the cleruchy hypothesis.
These middle-range theories are all pretty simple claims, about the veracity
of the dating material and the legitimacy of dating a particular pot by its
association with a particular text, but simple accounting claims may be all that
are needed in this case because it is pretty simple information that is being
sought in the pottery from the Paximadhi. All we are getting by way of

Middle-Range

Theory in Historical Archaeology

179

evidence is a date. More complicated information, such as who made a

particular pot or exactly what it contained, would demand more complicated
middle-range theorizing. Dating a pot (and in turn the site where it is found) is
not unlike sizing a microbe by positioning the little creature next to something
else of otherwise known size. It is in the process of otherwise knowing the size
of the comparanda where middle-range theories (accounting theories) are used
in the case of the microbe. The relatively simple question of size may call for
only simple middle-range theories, while more involved information, such as
composition of the microbe, would involve more middle-range theorizing.
Ascertaining the dates of occupation of the sites on the Paximadhi peninsula
then makes use of some amount of middle-range theory. So too does another
aspect of the evidence used to test the cleruchy hypothesis: the analysis of
ceramic fabric to see if it is of Attic origin or from local Karystian kilns. This
analysis is limited to coarseware pottery, the stuff with many large inclusions,
because it is the inclusions themselves which are characteristic of the place of
manufacture. The ceramic samples for this petrographic analysis are still at the
lab, but there is no question that the cleruchy-relevant evidence as to whether
the Palio Pithari coarseware is or is not from clay deposits in Attica will be
heavily dependent on middle-range theories. The actual data in this case will be
some sequence of chemical events in the Fitch lab of the British School in
Athens and these must be made to speak of ancient pottery manufacture. A
theoretical understanding will be needed to ensure that the petrochemical
similarities or dissimilarities in the Euboean and Attic ceramics must be due to
similar or dissimilar origin and not simply inevitable environmental processes
that show their effects over time. Chemical analysis speaks of associations
between ancient communities only with a lexicon linking chemical phenomena
to mineralogical composition to deposits of clay and the manufacture of pots.
The results of the fabric analysis are just coming in, and they are bound to
be somewhat ambiguous when they are completed. If the results are negative in
the sense of showing no greater proportion of Attic fabric at the purported
cleruchy sites than at other sites around the Karystian bay, this will not
necessarily tell against the cleruchy hypothesis. No one really knows that
cleruchs tended to use nonlocal coarseware. In fact, fleshing out the understanding of the cleruchs behaviour is part of the project. And if the evidence is
of more Attic pottery, there remains the ambiguity as to whether this is pottery
brought from home or pottery bought in trade by some group of residents on
the Paximadhi. The problem here is that there is no middle-range theory to
link the ceramic feature of being an Attic fabric and the feature of being
brought over with Athenian citizens. Such a relation could only be established
by studying what was already known to be a settlement of Athenians such as a
cleruchy. But that would be begging the question in the Euboea case as it
would breech the independence from theory for which the evidence is a test.

180

Such is the difficulty in trying

determine its properties.

Studies in History and Philosophy of Science

to detect

an object

and

at the same

time

There is a similar kind of ambiguity in the evidence of the groundplan

of the
Palio Pithari site. There is no disputing the observation
that the plan resembles
those of typical Classical farms in Attica. What is missing is the proof that this
indicates any sort of influence of Attic farm design on the residents of the
Paximadhi
peninsula,
or that the design identifies the Palio Pithari site as a
farm. The similarity of design could be a result of Attic influence or it could be
an independent
development
under the influence of similar environmental
conditions in the two places. Persistent north winds lead to a predominance
of
residential sites with small rooms tucked against the north wall and courtyards
on the sheltered south, and this shared design has nothing
to do with
communications
between the residents. What is missing in the cleruchy hypothesis case is a middle-range
theory linking groundplan
to use and/or origin of
design of a building.
The point is that the data are informative
evidence when there are independently justified middle-range
theories to make the connection
to the hypothesis, and not otherwise. Furthermore,
where these theoretical links between
data and phenomena
are at work, they can be uncovered from their normally
covert activities and made explicit. Thus in the dating of pottery there must be,
and there are, claims linking descriptions
of sherds to dates of pots, the step
from inventory
to catalogue.
These simple middle-range
theories are often
presented as pictures with terse captions rather than as discursive theories, but
they are nonetheless
claims which connect the information
in the present
archaeological
record with information
about the unobservable
past. And
these claims which account for the evidence used to test the cleruchy hypothesis are accountable
to their own evidential and theoretical
support. The
pottery dating catalogues
are supported
by evidence that is believable and
meaningful
in light of theoretical claims about the texts, and buildings that
provide a datable context.
The middle-range
theories, in other words, are just like any other theories.
They are nothing special when it comes to justification.
They help and they
need help. The pottery studied in its context in southern Euboea and dated
under the influence of previous ceramic theorizing will itself be catalogued and
published
to serve in interpreting
ceramic data found elsewhere. The theoretical claims which are the accomplishment
of the southern Euboea work will
become the tools of evidential accounting
in some other situation. They too
can be used as middle-range
theories.
Note also that the accounting
theories (the middle-range
theories) used in
the dating of pottery in this case are independent
of the cleruchy hypothesis
itself. There is no cleruchy information,
no presumption
of anything about the
particular hypothesis in the dating of pottery from its context in the Athenian

181

Middle-Range Theory in Historical Archaeology

Agora.

So while there is a clear theoretical

test the theory about

The

theory

does

not

a cleruchy,
sponsor

influence

on the evidence

it is not a case of circular,

its own

evidence.

self-serving

It is rather

used to
testing.

independent

evidence and in this sense the testing is objective.

The same is true in the potential evidence of analysing the ceramic fabric.
The middle-range
theories are amenable to articulation
and they are normal
theories which are the accomplishments
of chemistry and geology, but they are
in this case used to link singular properties of the present to general information of the cleruchy hypothesis. These theories are clearly independent
of the
cleruchy hypothesis, and thats important.
Having presented the hypothesis, the evidence, and the middle-range
theories which give meaning to the evidence, the final question is of the status of the
hypothesis in light of the testing. The evidence from the Palio Pithari excavation, in particular
the dating of the abandonment
of the site, tend to disconfirm the hypothesis. Put more symmetrically,
there is tension in the network of
claims which includes the hypothesis, evidence, data, and middle-range
theories. This incompatibility
alone does not indicate which of the claims must be
revised or rejected. Following
Quine and Duhem, no claim is immune to
revision or rejection, and the hypothesis
is potentially
retainable
if there is
reason to doubt some other parts of the network. For example, the auxiliary
theories, claims that have not been discussed here but which are used to draw
implications
from the hypothesis, are part of the tension. Perhaps the proposed
dates of the Karystian cleruchy are off. The start date is informed by relatively
clear and secure textual evidence and is therefore well entrenched.
But the
termination
date is more flexible. If, for example, the cleruchs did not farm
their own land but only profited by renting it to locals, then it is not
unreasonable
to expect the farming activities
to continue
even after the
cleruchs left for home. In this case, only the start date of the farmsite is
definitive enough to be telling evidence for or against a cleruchy, and the start
date matches expectation.
The data claims, the pottery inventories
for example, could also be questioned. If the pottery is wrongly described it is likely to be wrongly dated. But
this is a worry and a potential saving of the hypothesis that will last only as
long as it takes to have another look and recheck the sherds.
But the evidential claims, supported as they are by middle-range
theorizing,
are more vulnerable
to lingering doubt. The date assessments of the standard
comparanda
pottery could be questioned,
though this is unlikely because of
the widescale upset this would cause in the larger web of beliefs about the
ancient world. In other words, this would result in more tension rather than
less. But perhaps the propriety of using Athenian or Corinthian
stylistic dating
in the context of Euboea could be questioned.
Perhaps pottery styles change
across social class as well as through time, and the class of people on the

182

Studies in History and Philosophy of Science

Paximadhi, these cleruchs, used each particular style at a slightly shifted time
period than did the Athenians. Unfortunately, the most coherent of these
pictures would have the lower class, displaced cleruchs getting the pottery
styles later than the resident Athenians, thereby giving the Palio Pithari pottery
even later dates and worsening the fit between evidence and hypothesis.
The fate of the cleruchy hypothesis is still up in the air. The outcome though
will have a relevance beyond writing the history of southern Euboea. The
degree of certainty in our understanding of whether or not there is a cleruchy
near Karystos will influence our understanding of the texts which originally
motivated the hypothesis. It will add meaning to the figures of the Athenian
Tribute Lists and assist in interpreting the ancient authors. In this way, a
dialectic of assistance between textual and material records is established to
take advantage of the independence of the two sources of information. As
Leone and Potter suggest each source can be profitably used in the role of
middle-range theory for the other. It is a back-and-forth sort of negotiation
between the historical and archaeological claims. Reading the texts will suggest
an interpretation of the material data (as the reading of Diodorus and Plutarch
motivated the hypothesis that the Paximadhi sites were part of a cleruchy), and
in turn the archaeological evidence may influence an interpretation of texts.
(The dating of the Palio Pithari site makes it less likely that the Euboea
described in the sending of cleruchs to Andros, Naxos, and Euboea was
southern Euboea. Or perhaps Diodorus, Plutarch, and Pausanias got the whole
thing wrong and there were no cleruchs at all.) Thus the theory of Euboean
cleruchs can participate with middle-range theories in other circumstances. So
too with the evidential claims discussed in this case, for example the pottery
catalogue. What is an accomplishment on this project can be a tool on
another.
The point in presenting these details of evidence in the test of the cleruchy
hypothesis is simply to show that in this test-case the middle-range theories,
the accounting theories, are at work and can be identified and talked about.
They are the required informational links between the archaeological record,
the data, and the events and objects of the past which are relevant to the
hypothesis and are the evidence. They are what you have to know to make the
connection between what is on the ground and the objects and events of
interest, that is, those things which make contact with the theory being tested.
Furthermore the example shows that the middle-range theories are simply
theoretical accomplishments from other concerns. They endure their own trial
by evidence and must stand by their own justification.
5. Summary and Generalizations
This paper began with some predictions about the nature of archaeological
and scientific evidence, predictions based on a conceptual argument. Claims

Middle-Range

Theory in Historical Archaeology

183

about the unobserved require evidence of the unobserved, and this in turn calls
for connections between the data of observation and these evidential claims
about phenomena. Claims about these connections are theoretical claims in the
sense of being descriptive of unobservable events and objects. To account for
the formation of the observed data, to show what the data mean with respect
to the hypothesis, requires some mention of what is going on behind the
scenes. In this sense there must be middle-range theorizing. The prediction is
that all evidence will be made meaningful and credible with the support of
middle-range theories.
This prediction has been put to the test by evidence from historical archaeology. The analysis showed that middle-range theories are at work and can be
articulated. Furthermore, they are not of a special kind of theory. Any theory
could be used in the role of middle-range and in the examples we have caught
the theoretical accomplishments of one context in the act of accounting for the
evidence in another. And finally, these accounting theories are not, as the idea
of middle-range is sometimes interpreted, of middling generality. Some
accounting claims are very general, as about physical interactions, while others
are specific as about a particular detector or a particular ancient author. This
distinguishes Binfords use, and the use here, of the label middle-range theory
from alternative applications by, for example, Schaffe34 who means exactly
theories of mid-generality, regardless of their use.
The philosophical point in all of this is not to show that archaeology makes
hypotheses and tests them against the evidence as is the practice in natural
sciences. The point is rather about the evidence used in this process and, for
that matter, in any process of acquiring knowledge beyond the manifest. The
evidence will be influenced by theory and the influencing theories can be
articulated. And insofar as we can insist on independence between these
influencing theories and the object theory, the hypothesis, the evidence will be
objective and the testing will be meaningful. Thus it is not that we rely on only
the best theories to account for the evidence, if by best is implied some
dispensation from doubt. We rely on regular old theories, themselves part of
the web of scientific beliefs. The examples show that beliefs which give
meaning to the observations are themselves part of the network, and their
epistemic contribution does not come from some epistemic priority but from
an epistemic independence.

Acknowledgements - I am
Southern Euboea Exploration

pleased to be able to thank the archaeologists

of the
Project, Don Keller, Cynthia Kosso, Roz Schneider and

*See K. Schaffer, Theory Structure in the Biochemical Sciences, The Journal of Medicine and
Philosophy 5 (1980), 57-95.

184

Studies in History and Philosophy of Science

Mac Wallace, for their help toward my understanding and enjoying archaeology, and
for letting me carry the zambilis.
The conclusions expressed in this paper, such as the status of the cleruchy hypothesis
in light of the current evidence or the ambiguity of certain evidence, are entirely my own
and do not necessarily match the views of the archaeologists of the Southern Euboea
Exploration Project.
My work has been supported by a grant from the National Science Foundation,
number DIR-8917989.