We would like to return to compare the results presented in Sections 2-6
with the objectives outlined in our original research design (Section 1). It will be
immediately obvious that our original aim of studying change from the Iron Age to
the early medieval period has only been partially fulfilled, as the pottery collected
from the survey did not make it possible to address the question of the later Roman
to early medieval transition. Indeed, as noted by Steven Willis, there are major
contrasts between the Roman pottery assemblages found in this region and those
from the Tarragona area which also make it more difficult than originally
envisaged to offer direct comparisons between the regions. Nevertheless, the
distinctions in the Roman ceramic assemblages between the two areas are
themselves interesting (Section 5) and suggest a much lower level of economic
interaction between northern Portugal and the rest of the Roman world than might
have been expected.
The research design posed six specific questions to which we now turn:
a) What is the distribution of archaeological material across the valley within the
survey area? How does this vary through time?
There was a dense distribution of sites across the landscape (Table 7.2), although
the overall amounts of Iron Age and Roman pottery were lower than those found in
other surveys in the Mediterranean. The number of Iron Age sites was
unexpectedly large and there was a very strong pattern of continuity into the
Roman period (Table 7.1). The lack of closely dated Roman wares and the general
absence of dateable early medieval fabrics does not allow us to address the problem
of continuity in the middle ages.
b) How does this distribution of material relate to the known sequences of the
castros? In particular is there any evidence for the High Empire when many
castros appear to be deserted?
The most important new information relates to the numbers of newly discovered
Iron Age sites in the lowland areas around the castros. Although such sites were
known in the region prior to the survey (e.g. Martins 1990), the number of such
sites found and their density (an average of 4.94 sites per square km) is much
larger than previously estimated. Even allowing for this statistic being distorted by
the bias in our survey towards land in valleys, it completely changes any
understanding of the character of Iron Age settlement in the region. Our difficulties
in dating sites to narrow ranges makes it difficult to address the issue of shifts in
settlement use between the castros and surrounding areas. Two features emerge
clearly from the data. First, there is a very strong pattern of continuity of lowland
sites from the Iron Age into the Roman period. Second, the density of Roman rural
settlement is much higher than that of the Iron Age (averaging 8.42 sites per square
km in the Roman period compared with 4.94 in the Iron Age: see Table 7.2).
Whether this increase in settlement density results from population growth or
relocation from castros remains uncertain, but we suspect the former is the more
important factor.
c) Is material in the valleys the result of cultivation (manuring) and/or the
dispersed agricultural settlements?
Although there was a very light background scatter of material across the valley
floors, it was too little to suggest any systematic pattern of manuring. In general,
the volumes of pottery found were very small, suggesting a comparatively low
level of ceramic use with the result that any such manuring is unlikely to have been
archaeologically detectable. Equally, pottery of Iron Age date was not very robust
and is unlikely to have survived in the ploughsoil over long periods.
d) Does the distribution of material and the composition of assemblages vary
with distance from the coast up the valley?
The modification to our research design which resulted in the reduction of the
number of transects to two makes it difficult to answer this question with certainty.
The patterns of site distribution show a relatively even distribution of Iron Age
sites across the two transects, but a higher density of Roman sites in the coastal
transect (Table 7.2, Transect 1). This pattern of change is supported by the
larger proportion of new sites established in the Roman period which were noted in
Transect 1 (Table 7.1). Equally, the small quantities of traded Roman wares appear
to have been more common near the coast than inland. Although these data are
insufficient to sustain much confidence, they do suggest some shift in the balance
of activity towards the coastal zone during the Roman period.
e) Does the distribution of material and the composition of assemblages vary
with distance from Braga?
Again the quantity of material recovered is such that our results need to be
treated with care, but Steven Willis has noted a marked distinction between the
rural assemblages from the survey and those he was able to examine from Braga.
The town had a higher proportion of fine wares than are found in the rural sites, as
well as a broader ceramic repertoire. Although we cannot distinguish differences in
assemblage composition with distance from the town, a clear urban-rural
dichotomy is evident, suggesting differing supply patterns and varying cultural
norms, most likely reflecting the contrast between the continuities of settlement in
the countryside and the newly established lifestyles urban centre.
f) Are there geomorphological changes within the river valley which can be
correlated with changing land use in antiquity?
The scale of geomorphological change observed within the landscape were not
such as to suggest any major phases of landscape change. Rather, we would see a
pattern of continuous geomorphological change through time. There was little
evidence for radical alterations to the river valleys. However, there may have been
substantial erosion of the schist areas. Although it is by no means clear whether the
general absence of Iron Age and Roman settlement in areas on the schist is real or
a result of more recent erosion, it is likely that these areas have been more altered
by erosion than others in the region.
More generally we set out to apply the techniques of fieldwalking and detailed
surface survey to an area where the techniques had not previously been applied.
We found that the landscape was more intractable than others in which we have
worked. The small fields, intensive agriculture and relatively small numbers of
dated finds meant that the results were less spectacular than had been hoped.
Equally, the problems of soil type, terracing and field size meant that the results of
the geophysical surveys were disappointing, although by no means negative. In
contrast to these rather negative impressions, we have found that the analysis of
the results from such a topographically varied area has greatly benefited from the
use of GIS. This has enabled us to obtain new perspectives and to address
questions which it would otherwise have been difficult to approach. We trust that
the insights drawn from these analyses will encourage others to apply the
techniques in other studies of Iron Age and Roman rural settlement.
Finally, some broader contrasts with the Tarragona region as originally envisaged
are instructive. Three contrasts can be drawn.
First, the density of sites found in the
Ave Valley field survey (Table 7.2) was greater than that noted in the Tarragona
area where there were 3.22 Iberian (i.e. Iron Age) sites per km² and 3.125 sites per
km² of the Republic and Early Empire (Carreté et al. 1995, 248). This
is certainly partially a result of the more constrained landscape in this survey area,
meaning that settlement sites were preferentially constructed where the survey was
concentrated in the valleys. Nevertheless, this high density of sites suggests a
much more prosperous agrarian society in the Iron Age and Roman periods than
has often been suggested.
Second, the range of sites represented in the Ave Valley
is much more limited than found in the Tarragona region. Although it has not been
possible to measure site size in this survey, the overall impression is
predominantly of small farmsteads (aside from the castros). In contrast, the
Tarragona survey found a wider range of site sizes and a pattern of increasing
differentiation of sizes and forms through time (Carreté et al. 1995,
2323, 249).
Finally, we should note that in comparison with the Tarragona area,
there is a stronger pattern of site continuity in this area (Table 7.1). These last two
features together suggest a much more stable society, with the establishment of
new sites in the Roman period but little evidence for the emergence of dominant
rural sites (i.e. villas) or the accumulation of large land holdings. Although our
evidence is insufficient to allow much further speculation, we can surely point
towards a society in this part of Portugal that was far less influenced by the Roman
world than that around Tarragona. Nevertheless, the all too common stereotype of
a backward and poor region has been comprehensively demolished by the work in
this survey. In the Roman period, as today, the lower lying parts of the region were
heavily occupied, arguably by groups who retained social structures focused on
local centres at the castros. Whether or not the latter remained in occupation, they
certainly seem to have remained in the view of the occupants.
Some broader issues concerning the transition from Iron Age to Roman in this
part of Portugal are developed further elsewhere (Millett forthcoming).
7.1 Future work
The extensive data from the survey offer much scope for future work, although
such studies will probably rely on the future analysis of excavated material from
the region providing a more refined chronology. Equally, more spatial analysis of
material in the GIS would be worthwhile, investigating, for instance, the influence
of drainage across the study area. Further viewshed analysis and cost analysis
would also be worthwhile, using improved models which take into account the
influence of changing climate, vegetation and watercourses.
In a broader context, additional fieldwalking surveys would certainly improve our
knowledge of the region. Such work is certainly needed in the areas where
development pressures are most intense, as our work has shown that there are
many more sites to be found than previously believed. Such work might be best
targeted on discrete regions rather than sample transects. It would also benefit from
being undertaken over longer seasons throughout the year. Finally, any further
work in the region would undoubtedly benefit from a detailed geomorphological
assessment of the areas studied, as we still do not fully understand the processes of
very localised erosion and deposition which characterise this area.