Early Medieval Europe 2008 16 (2) 127­165
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd, 9600
Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA
Blackwell Publishing LtdOxford, UKEMEDEarly Medieval Europe0963-94621468-0254 2008 The Author. Journal Compilation  Blackwell Publishing LtdXXXOriginal ArticlesThe climatic anomaly of the early Middle AgesFredric L. Cheyette
The disappearance of the ancient
landscape and the climatic anomaly
of the early Middle Ages: a question
to be pursued
Fredric L. Cheyette
Archaeological surveys and rescue archaeology have now dated the disappearance
of occupied sites in late antiquity with considerable precision, especially
in the Rhône valley and northern Gaul. Landscape archaeology has shown
a conversion from arable to pasture and reforestation during the same
period. Recent studies of the climate of the first millennium show that this
was also an extended period of wet and cold climate. How these phenomena
were connected is an important research question. A preliminary suggestion
made here is that since reversion from arable to pasture affected regions
as far apart as Italy and Poland it cannot simply be ascribed to the political
and fiscal dislocation of the ancient world, but should be understood as
one effect of the climatic anomaly.
In the now substantial historical literature on the `transformation of
the Roman world' the subject of agriculture and the rural world in
what had been the western empire has been relatively neglected, despite
the plethora of archaeological investigations that directly bear on the
subject. In the series published under that commodious title, only one
volume out of fourteen has been devoted to the rural economy,
* Research for this article was supported by a faculty research grant from Amherst College and
by a Mellon Foundation Emeritus Fellowship. I have been much informed by conversations
with I. Catteddu, L. Bourgeois, D. Marguerie, M.-P. Ruas, J. Henning and P. Van Ossel,
and by the comments of an anonymous reviewer for this journal. I also owe much to the
hospitality of the American Academy in Rome, the C.E.S.C.M. at Poitiers, and the Columbia
Institute for Scholars in Paris.
128 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
and that one under the curious question The Making of Feudal
Agricultures?1
Almost all consideration of the late antique and early
medieval economies of what was once the western Roman empire
appears still to be in thrall to the questions and hypotheses first put
forward by Henri Pirenne over seventy years ago. Did cities remain
vibrant and economically active places? Did international commerce
survive? What were the rhythms of decline and revival of urban
marketplaces and the trade that fed them?2
There is no need to insist on the fact that urban centres and an urban
economy of whatever size and strength, not to mention armies, monasteries
and aristocratic courts, would not have existed without a rural world to
feed them. And although there is no way to calculate accurate figures
for late antiquity and the early Middle Ages, the examples of both early
modern European populations and modern third-world populations
suggest that an estimate of eight or nine agricultural workers for every
individual not working in the fields would be a reasonable guess.3
The
story of the rural world is therefore the story of the vast majority of the
population, in addition to being the necessary prerequisite to all those
studies of the urban world and the market economy which dominate
the literature.
In all likelihood, the reasons for this neglect are many. Of the most
important, surely the first is the belief that there is little or no new
evidence to make such a history possible, since all the written sources
­ the few relevant texts in the Roman law codes, the early `barbarian'
codes, sermon literature, a few Italian papyri, and so forth ­ have long
been worked over. Few conclusions drawn from them seem, however,
1
M. Barceló and F. Sigaut (eds), The Making of Feudal Agricultures? (Leiden and Boston,
2004). The article in this volume by A. Durand and P. Leveau, `Farming in Mediterranean
France and Rural Settlement in the Late Roman and Early Medieval Periods', is one of the
most important attempts in recent years to fill the gap. Because there is a vast bibliography
on nearly every subject touched on in this article, I will restrict my references to only a
few items, whose own notes and bibliographies will lead the interested reader on to a
larger range of literature. For a more extensive bibliography, regularly updated, see this
article on my web site: <http://www.amherst.edu/~flcheyette/Publications/Transformation
%20rural%20world.pdf>.
2
A recent exception is C. Wickham, Framing the Early Middle Ages: Europe and the Mediterranean
400­800 (Oxford, 2005), which considers many of the issues in the first part of this article.
Some of the data that I present here were published too recently to have been available
to him.
3
In 1821, in the midst of industrialization, some English counties still counted over 60% of
the population engaged in agriculture, while in Finland in 1805 it was still 82% of the male
labour force. E.A. Wrigley, Poverty, Progress, and Population (Cambridge, 2004), pp. 118, 297.
The first US census (1790) classified 95% of the population as `rural': Historical Statistics of
the United States (Washington, 1975), series A 67­72.
The climatic anomaly of the early Middle Ages 129
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
to be above controversy.4
Second, the continuing assumption among
many historians that the only changes that might have occurred would
have been in the manner in which elites exploited the peasantry, from
this to conclude that the only story worth telling, or the only story that
could be told, was how the Roman great estate turned into the monastic
and imperial great estates of the Carolingian world.5
Third, the belief
that if there were any changes in the way that agriculturalists ­ whether
called slaves, serfs, or simply peasants ­ worked the land, those changes
would have been mainly the result of technical or material losses. At the
head of the list of such changes would have been the now longdisproved
notion that iron tools disappeared from peasant households
(an idea popularized in particular by Georges Duby6
).
All of these beliefs have commonly fed the conviction that there was
very long-term continuity in the rural world between late antiquity and
the beginning of the great medieval expansion, usually dated around
the year 1000. From this, some have surmised that somewhere under
modern villages one would find the Roman villa that originally served
as the pole of attraction for medieval peasant households, while others
have speculated that the dispersed habitat of antiquity continued until,
sometime in the tenth or eleventh century or later, castle lords forcibly
4
Discussion of the literary evidence in T. Lewit, Agricultural Production in the Roman Economy,
A.D. 200­400, BAR International Series 568 (Oxford, 1991), Ch. 8; and in P. Van Ossel and
P. Ouzoulias, `Rural Settlement Economy in Northern Gaul in the Late Empire: An Overview',
Journal of Roman Archaeology 13 (2000), pp. 133­60. See also P. Ouzoulias, `La déprise
agricole du Bas-Empire: un mythe historiographique?', in P. Ouzoulias and P. Van Ossel
(eds), Les campagnes de l'Ile de France de Constantin  Clovis: Colloque de Paris 1996, Document
de travail 3 (Ile-de-France, 1997) pp. 10­20; P. Van Ossel, Etablissements ruraux de l'Antiquité
tardive dans le Nord de la Gaule, Gallia supplément 51 (Paris, 1992).
5
See the pertinent critical comments of Wickham, Framing the Early Middle Ages, pp. 259­65,
and for a recent example of this historiographical tradition, P. Sarris, `The Origins of the
Manorial Economy: New Insights from Late Antiquity', English Historical Review 119 (2004),
pp. 279­311.
6
Notably G. Duby, Rural Economy and Country Life (Columbia SC, 1968), pp. 20­2 and The
Early Growth of the European Economy (Ithaca, 1974), pp. 13­17, repeated almost verbatim
by R. Fossier, La terre et les hommes en Picardie (Paris, 1968), I, p. 236, both based essentially
on the inventory of Annapes. The theory was already severely criticized by R. Delatouche,
`Regards sur l'agriculture aux temps carolingiens', Journal des Savants (1977), pp. 73­100, at
pp. 78­9. Archaeologists have since found that the country smithy (evident in the slag
from his furnace) was far from unknown in the early Middle Ages: iron production is
present at 18 of the fifth- to eighth-century rural sites inventoried by E. Peytremann,
Archaéologie de l'habitat rural dans le nord de la France, Mémoires publiés par l'Association
française d'Archéologie mérovingienne 13, 2 vols (St-Germain-en-Laye, 2003). For the
presence in the early Middle Ages of iron-tipped ploughs capable of turning the soil, see
now J. Henning, `Germanisch-romanische Agrarkontinuität und -discontinuität im
nordalpinen Kontinentaleuropa', in D. Hägermann et al. (eds), Acculturation: Probleme einer
germanisch-romanischen Kultursynthese in Spätantike und frühem Mittelalter (Berlin, 2004),
pp. 396­435.
130 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
resettled peasants into nucleated villages where they could more easily
be controlled and exploited.7
Over the last twenty years or so, archaeological field surveys and rescue
archeology have begun to suggest a very different narrative. This narrative
has not been without its obscurities and points of contention, not
surprising in the case of a young and rapidly developing field of study.
The results of these studies, nevertheless, have clarified the nature,
rhythms, and chronology of changes in material culture and the agricultural
economy over the half millennium from the third through to the
seventh century. They have thus brought into sharper focus the vexed
issues of economic and demographic `decline' (whatever its ideological
baggage, still a useful term, I believe, to cover material impoverishment
and reduced population). They have also described the nature and
chronology of the stages through which the ancient rural landscapes of
both northern and Mediterranean Europe passed to become the very
different landscapes of the Middle Ages.8
In particular, when looked at
in detail, the research on the rural world points to an important distinction
between the developments of the third and fourth centuries and what
happened in the centuries that followed, a distinction that is obscured
by the far more complex cultural, political, and physical transformations
of the urban world. In this article I will briefly discuss the longer halfmillennial
narrative before turning to the period when the ancient
countryside vanished in the west and the first elements of what would
be the medieval countryside came into being: the period from the fifth
to the seventh century. A critical factor in this later period, I will argue,
7
The classic statement of the first idea is M. Bloch, French Rural History, An Essay on its Basic
Characteristics, trans. J. Sondheimer (Berkeley and Los Angeles, 1966), p. 1. The classic statement
of incastellamento is P. Toubert, Les structures du Latium médiéval, 2 vols (Rome, 1973), I,
pp. 330­54, generalized as `encellulement' by R. Fossier, Enfance de l'Europe, 2 vols (Paris,
1982), I, pp. 288­317, and since turned into a commonplace in French and Italian historiography.
Without denying the importance of tenth- to eleventh-century developments, both the
archaeological and documentary evidence point to proto-villages as geographical and social
entities and perhaps as nascent field systems at a significantly earlier date: see Wickham's
nuanced discussion, Framing the Early Middle Ages, pp. 516­17 and for Mediterranean Gaul,
A. Durand, Les paysages médiévaux du Languedoc (Xe­XIIe sicles) (Toulouse, 1998), pp. 77­95.
For Italy, see for example R. Francovich, `L'incastellamento e prima dell'íncastellamento
nell'Italia centrale', in E. Boldrini and R. Francovich (eds), Acculturazione e mutamenti:
prospettive nell'archeologia medievale del mediterraneo (Florence, 1995), pp. 397­406; É. Hubert,
L'incastellamento en Italie centrale (Rome, 2002); R. Frankovich and R. Hodges, Villa to
Village (London, 2003), Ch. 3; and for the area studied by Toubert, H. Patterson, H.
DiGiuseppi et al., `Three South Etrurian "Crises": First Results', Papers of the British School
at Rome 72 (2004), pp. 1­36.
8
I set out this last problem in F. Cheyette, `The Origins of European Villages and the First
European Expansion', Journal of Economic History 37 (1977), pp. 182­206, and for the
Mediterranean landscape in C. Amado and F. Cheyette, `Organisation d'un terroir et d'un
habitat concentré: un exemple méridional', in A. Bazzana, P. Guichard and J.M. Poisson (eds),
Habitats fortifiés et organisation de l'espace en Méditerranée médiévale (Lyon, 1983), pp. 35­44.
The climatic anomaly of the early Middle Ages 131
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
was a climate anomaly, largely ignored by historians but well known to
historical climatologists.
The evidence of field surveys
The general shape of this new archeology-based narrative was first
sketched by the South Etruria survey conducted in the years following
the Second World War by the British School at Rome. The results were
summarized by Timothy Potter in 1979, and brought to the attention
of historians a few years later by Richard Hodges and David Whitehouse.9
In his analysis of African Red Slip ware (ARS) and Forum ware from
the Ager faliscus, Potter showed that where 82% of all the known
antique sites were occupied in the second century CE, only 58% were
occupied in the third century and only 19% in the fifth and sixth.
When Helen Patterson and her team recently re-examined the material
from the survey, including so called `coarse ware' as well as ARS, they
found that Potter's broad narrative was substantially confirmed: they
identified 1,300 sites from the period 100­250, fewer than 400 from the
period 250­450, and a continued dramatic decline through the later
fifth, the sixth and seventh centuries.10
Since the publication of the South Etruria survey, other surveys have
confirmed equivalent declines in occupied sites between the second
century and the end of the third, and the continued disappearance of
sites from the fifth century onward in region after region, not only in
Italy but throughout the western empire. Many of these surveys have
recently been summarized elsewhere; there is no need to do so again
here.11
I will merely mention three recently published examples that
have particular value.
Marco Valenti's fine-grained twenty-year study of 1,979 square
kilometres of the provinces of Siena and Grosseto (9% of Tuscany), which
involved both field surveys and excavations, revealed a decline in the
9
T.W. Potter, The Changing Landscape of South Etruria (New York, 1979), Ch. 6. R. Hodges
and D. Whitehouse, Mohammed, Charlemagne and the Origins of Europe: Archaeology and the
Pirenne Thesis (Ithaca, 1983), pp. 33­48.
10
Patterson et al., `Three South Etrurian Crises', p. 18. At the time this publication was prepared,
Patterson states, the survey data had not been completely analysed.
11
See especially N. Christie, From Constantine to Charlemagne: An Archaeology of Italy, AD 300­
800 (Aldershot and Burlington, 2006), Ch. 5; Wickham, Framing the Early Middle Ages,
pp. 465­518; B. Ward-Perkins, The Fall of Rome and the End of Civilization (Oxford, 2005),
especially pp. 138­46. They reach very different conclusions about the meaning of the data.
Major critiques of survey data at P. v. Dommelen, `Una riconsiderazione di ricognizioni
estensive: il caso dello Scarlino-survey', in M. Bernardi (ed.), Archeologia del paesaggio (Florence,
1992), pp. 859­76; P. Van Ossel and P. Ouzoulias, `Rural Settlement Economy'; E. Louis, `A
De-Romanised Landscape in Northern Gaul: The Scarpe Valley from the 4th to the 9th
Century AD', in W. Bowden et al. (eds), Recent Research on the Late Antique Countryside
(Leiden, 2004), pp. 479­504.
132 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
average density of occupation from one site every 1.27 square kilometres
in the early third century, to one every 4 square kilometres in the fifth
century, and one every 10 square kilometres in the sixth. By the fifth
century a little more than a fifth of the early third-century sites were
still occupied, and only about a third of these survived into the sixth.
Patterson's survey of the Tiber river basin showed a similar decline,
with a first reversal of expansion in the third century, a little more than
a third of the sites surviving into the fifth century, and only one eighteenth
of that number still occupied in the seventh.12
Though more exact dating
of pottery, especially of coarse or `common' ware, has refined the
chronology of older surveys, making the decline look less catastrophic
than it did to Hodges and Whitehouse in 1983, both the direction and
the rhythm of site abandonment have been confirmed time and again
in Italy, though with significant regional variation.13
Direction and rhythm have also been confirmed in southern Gaul,
where the international programme working under the name
`Archaeomedes' in their survey of eight separate regions of the lower
Rhône valley discovered 865 Roman period sites. For these sites the
group recorded not only the dates of occupation, size, and the nature
of material found, but also data on the environment. The dates of first
occupation from the Roman conquest to the fifth century CE are
presented in Table 1. Of these 865 sites, 780 gave no signs of occupation
prior to the arrival of the Romans.
Table 2 gives the number of individual sites occupied during the
same centuries.
12
Patterson et al., `Three South Etrurian Crises', and M. Valenti, `La formazione dell'insediamento
altomedievale in Toscana. Dallo spessore dei numeri alla construzione di modelli', in
G.P. Brogiolo et al. (eds), Dopo la fine delle ville: le Campagne dal VI al IX secolo, Documenti
di Archeologia 40 (Mantua 2005), pp. 193­219. Similar densities in the area of Claterna in
Emilia: M. Librenti, `Ricognizione di superficie e insediamento medievale nella pianura
emiliano romagnola: alcune considerazioni', in Secondo Congresso Nazionale di Archeologia
Medievale (Florence, 2000), pp. 170­4.
13
On the `catastrophic' interpretation see the cautious remarks by Patterson et al., `Three South
Etrurian Crises', pp. 22­3. In general Christie, Constantine to Charlemagne, pp. 412­27. For
differing rates of survival between areas as close as Parma and Cesena see now S. Gelichi et al.,
`La transizione dell' antichit al medioevo nel territorio dell'antica Regio VIII', in Brogiolo
et al. (eds), Dopo la fine delle ville, pp. 53­80, on which more below.
Table 1 Date of foundation of Roman era sites in the mid- and lower Rhône valley
BCE 1st century 2nd century 3rd century 4th century 5th century Total
252 462 36 21 58 26 865
The climatic anomaly of the early Middle Ages 133
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
That is, the number of sites occupied in the fifth century was only a
third of the number of sites occupied in the second century, and nearly
half of these could have been sites first occupied in the fourth or fifth
century. The published data unfortunately do not give numbers for
sixth-century occupation. For the period up to 500 CE, however, the
Rhône valley percentages are of the same order as those of the South
Etruria survey.14
A richer vision of developments in the lower Rhône valley, and one
extending beyond 500 CE, comes from an earlier survey that was part
of the same programme, a survey of the Lunellois plain between Nîmes
and Montpellier, just west of the mouth of the Rhône.15
Here, 150
ancient and medieval sites were intensively studied. The chronology of
their occupation is represented in Figure 1.
Not surprisingly, the rate of survival in this small sample reflects
those of the larger survey of which it was a part. Sixty-six of the sites
were created de novo in the first century. Two-thirds of these did not
survive into the third century, and three-quarters did not survive to the
end of the third. The fourth and fifth centuries mark a period of
renewal, though about half of these new sites do not survive into the
sixth. In terms of numbers of creations and disappearances, there are two
important `breaks' in the pattern of settlement: one in the passage
through the third century, the second in the sixth and seventh centuries,
by the end of which nearly all the ancient sites had been abandoned.
These three recent surveys are particularly important, first because of
the size of the areas covered and the number of sites investigated, several
thousand all told. They were conducted in areas where late antique and
early medieval fine and common wares have now been much more
securely dated. They have involved excavation of selected sites as well
as the gathering of surface pottery. And they have been exceptionally
attentive to all the factors that might bias their results. What might
their findings mean?
14
S. van der Leeuw et al. (eds), Archéologie et systmes socio-enviornnementaux: études multiscalaires
sur la vallée du Rhône dans le programme ARCHAEOMEDES, Collection de Recherches
Archéologiques. Monographies 27 (Paris, 2003), annexe 2. The original data are presented in
fifty-year periods. I have simplified by giving in Table 2 the higher of the two figures for each
century.
15
Archaeomedes, Des oppida aux métropoles (Paris, 1998), Ch. 5 and annex 1.
Table 2 Number of Rhône valley sites occupied by century
BCE 1st century 2nd century 3rd century 4th century 5th century
252 541 565 313 224 188
134 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
The earliest impulse when the results of the South Etrurian survey
were published was to conclude, to quote Tim Potter himself, that
`everything points to a level of depopulation that is as marked in the
towns as it is in the countryside'.16
Whitehouse and Hodges were
equally forceful.17
But not everyone was convinced. Some, in response,
16
Potter, Changing Landscape, p. 145.
17
`Depopulation . . . is not impossible; the ruined towns and wasted countryside suggest that
it happened in the Mediterranean at the end of the Roman period ­ and the burden of proof rests
with those who maintain that it didn't.' Hodges and Whitehouse, Mohammed, Charlemagne, p. 53.
Fig. 1 Numbers of sites in the Lunellois, by date of foundation and longevity.
Source: Archaeomedes, Des Oppida aux métropoles
The climatic anomaly of the early Middle Ages 135
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
tried moving what appeared to be the absent country dwellers to the
cities or to the villas and farmsteads that remained. This solution
proved implausible, for the supposed migrants would have had to be
fed by someone in the very countryside that instead appeared to be
emptying out; and though luxury expenditures are everywhere visible
in the great villas of the fourth-century countryside, there has been no
demonstration of their increased productivity, in fact ­ where attempts
have been made to answer the productivity question ­ just the opposite.18
Furthermore, with rare exceptions, such as Trier, which as imperial
capital enjoyed a vibrant fourth century (as did the countryside around it),
most western cities were being partially, or in some cases completely,
abandoned.19
Others have suggested that the people who had lived in
dispersed farmsteads (or their immediate descendants) moved to new
proto-villages (thus a hamlet of three farmsteads could have the same
population as three dispersed farmsteads, and in consequence, in the
tabulation of sites, three old ones would disappear and a single new
one be created without any necessary change in the total population
represented).20
Yet others have proposed that many of the smaller sites,
especially those of the first and second centuries, were not living spaces
at all, but modest structures to hold equipment or temporary quarters
for agricultural labourers whose residences were elsewhere.21
These objections to a simple demographic interpretation of the
evidence seem to be driven more by a desire to avoid a `catastrophist'
vision of late antiquity, than to open up the debate on the fate of the
late antique rural world. They do, however, have the heuristic advantage
of raising two fundamental questions about the survey data, which we
may put simply as, `What lies behind those black lines in Figure 1?'
Specifically, what constitutes a `site'? And what do we understand by
`continuous occupation'? It is only in recent years, as archeologists have
18
See the discussion of the villa of Pré-Bas, below. Older excavations of rural sites, unfortunately,
largely ignored their production areas, since these were unlikely to produce museum-quality
artefacts.
19
See in general J.H.W.G. Liebeschuetz, Decline and Fall of the Roman City (Oxford, 2001),
pp. 64­97, supplemented now by G.P. Brogiolo and S. Gelichi, La citt nell'alto medioevo
italiano (Rome, 1998), and by S.T. Loseby, `Decline and Change in the Cities of Late Antique
Gaul', in J.-U. Krause and C. Witschel (eds), Die Stadt in der Spätantike: Niedergang oder
Wandel? (Stuttgart, 2006), pp. 67­104. Italian exceptions are surveyed by F. Marazzi, `Cadavera
urbium, nuove capitali e Roma aeterna', in ibid., pp. 33­66. A striking example of the
synchronicity in the depopulation of a major urban centre and the depopulation and
reorganization of its rural environs is presented by V. Bel et al., `Réflexions sur une ville et
sa proche campagne dans l'Antiquité: le cas de Nîmes (Gard)', in A. Bouet and F. Verdin
(eds), Territoires et paysages: mélanges Philippe Leveau (Bordeaux, 2005), pp. 19­44.
20
A hypothesis recently proposed by Louis, `A De-Romanised Landscape'.
21
Proposed by F. Favory, J.-L. Fiches and C. Raynaud in `La dynamique spatio-temporelle de
l'habitat rural gallo-romain', in van der Leeuw et al. (eds), Archéologie et systmes socioenvironnementaux,
especially pp. 309­11.
136 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
begun at long last to pay attention to post-fifth-century levels of both
rural and urban sites and, in the case of villas, to the working areas as
well as the living areas where art objects of value were likely to be
found, that the nature and importance of these questions has become
clear.22
No single excavation can represent all the complexities that have
recently been found, but the work on one villa may at least suggest
what could be hiding behind one of those long lines, and why `site' and
`continuous occupation' are problematic terms.
Not far to the west of the Lunellois is the villa of Pré-Bas just outside
the village of Loupian, 4 kilometres from the port of Mze.23
There was
a farm here in the Republican period, of which naught remains but
some filled storage pits and ditches. The villa of the first century CE
was devoted almost exclusively to producing and exporting wine, with
substantial space given over to living quarters for the workforce and to
the warehouse for the wine. In the late second and third century, both
areas were allowed to shrink. An entire section of the inhabited space
fell to ruin. The remains imply a partial shift to cereal cultivation.
Among other changes, the storage area for wine was transformed into
an iron-working shop. One to two centuries later, in the second half of
the fourth century, the building's owner built new wine presses and
wine storage areas, but the volume of containers from that period
suggests a production that was only a half to a third of the production
of the villa's most profitable days in the first century. The remains of
the fifth century suggest a production that is in turn five to six times
smaller than that of the fourth. Nevertheless, it was in the fourth century
that the villa was rebuilt on an elaborate and luxurious scale, with
mosaic floors and painted walls. It is that fourth-century villa that one
visits now. Not more than a century later the villa suffered what its
excavator Christophe Pellecuer calls `a brutal and permanent decline'
(`une brutale et durable récession de l'occupation'). The occupants
continued to repair worn mosaics with mortar, but wall paintings were
allowed to fade and crumble, marble was removed from one wing, and
mosaics eventually suffered when heavy objects (roof beams, perhaps)
fell on them, after which they were covered over with a mixture of
cement and tile fragments. Finally in the sixth century the great courtyard
was colonized with buildings that made use of the remaining walls.
They were perhaps timber-framed structures. Some may have been
22
On the difficulties of using published reports to say anything about post-fifth-century villas
as recently as a decade ago, see C. Balmelle, Les demeures aristocratiques d'Aquitaine (Bordeaux,
2001), pp. 118­22. Not much had changed since my lament in `Origins of European Villages'.
23
What follows comes from C. Pellecuer, `La villa des Prés-Bas (Loupian, Hérault) dans son
environnement: contribution  l'étude des villae et de l'économie domaniale en Narbonnaise',
thesis, University of Aix-Marseille I (2000). It is to be hoped that this important thesis will
soon be published.
The climatic anomaly of the early Middle Ages 137
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
`sunken huts'. The inhabitants threw their kitchen garbage just outside
these buildings, where it was eventually buried in the ruins. These scant
remains (only 3% of the potsherds collected on the site post-date the
year 450) are the last signs of occupation. At length a destruction layer
covered the entire site. After a glorious fourth, an impoverished but
hopeful romanitas persisted at Pré-Bas through part of the fifth century,
as fourth-century residential rooms were converted to utilitarian usage.
Those who lived here strove to maintain what they could of the old
structure. Then the will, or the means, no longer survived, and finally
the site was abandoned and buried.
Meanwhile, sometime in the fifth century, about 800 metres away a
Christian church went up, most likely at the behest of (and paid by)
the owner of Pré-Bas; a church that also served as a cemetery for the
villa, for the artisanal and fishing hamlet near the lagoon whose
products served the villa, and for the small community that quickly
gathered around the sacred space and remained there until at least the
Carolingian period. The original church probably fell to ruins by the
seventh or eighth century and was replaced by another, which itself was
replaced in the late Middle Ages by the gothic structure that is now on
the site.
Here, we see, is the problem with just counting sites. A villa such as
this, with a substantial work force in the first century, an uncertain
number in the second and third, and a renewed prosperity but still
perhaps only half the first-century work force in the fourth, is counted
as one site. It is still one site in its impoverished sixth-century state. At
the same time, small agglomerations (such as ancient Lunel), post-fifthcentury
rural churches, and small farms are also counted individually as
`sites'. Their presumed populations are very different. Furthermore, is
Pré-Bas, with its associated late antique church and artisanal-fishing
quarter one site, two, or three? Is the survival of the church after the
villa and artisanal quarter disappear an example of continuity or of
abandonment and new foundation? Indeed, given the changes through
all the centuries of its occupation, should we pay more attention to the
continued use of the site (or sites) or to the direction and speed of the
transformations? However these questions are answered, it is obvious
that a `site' is not an entity with any fixed physical meaning, still less
any demographic meaning, even if defined by the number of square
metres where the shard fragments were found. Pellecuer could estimate
the manpower required for production at Pré-Bas, based on the
containers used for stocking wine and grain, but it would be foolhardy
to extrapolate from that to all the large villas in the Rhône valley survey
or anywhere else. Thus, while it seems clear enough that behind the
drop in `occupied sites' lies a fall in population, we will probably never
138 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
be able to measure that fall with any precision or even set it within a
plausible range.24
The scale of population decline, however, may very well be a secondary
issue. What appears more important are the social and economic changes
over the long term ­ implied by the scale and direction of physical
changes at Pré-Bas and confirmed by similar changes at many other sites.
The fourth-century heyday of luxurious villas ­ a few newly founded at
that time, most of them rebuilt, sometimes after a period of abandonment
­ has long been a fixture of Roman archaeology.25
It is this feature of the
late Roman countryside that has always seemed to be a strong counterargument
to notions of late antique `decline'. More refined analysis of
survey data coupled with the identification of post-fifth-century pottery
sequences has gradually resolved this apparent contradiction. The Rhône
valley project, for example, by carefully identifying sites by their size,
demonstrated that the small and medium-sized farms were the most fragile;
they were the ones most likely to disappear after the mid-second century.
It was the larger villas that were most likely to survive or be reoccupied
in the fourth century.26
Surveys in Emilia, Tuscany and Friuli, have found
the same phenomenon.27
What we may be looking at in the late second
and third century is the collapse of a speculative commercialized agriculture,
a collapse that smaller farms were unable to withstand. The nature of
this economic crisis remains a problem for the specialists to solve.28
It is
clear from luxurious villas like Pré-Bas that the reforms of Diocletian
and his successors succeeded in returning wealth to the countryside, at
least in the hands of the well connected, the rich, and the super rich.29
What happened from the fifth century on is another story. As recently
as a decade ago, there was little but supposition and guesswork on which
24
For a much richer examination of the problems involved in deriving demographic data from
archeological surveys see F. Trément, `Études micro-regionales et paléodémographie', in F.
Gateau et al., L'Etang-de-Berre, Carte archéologique de la Gaule 13:1 (Paris, 1996), pp. 98­113.
My thanks to Paul Van Ossel for clarifying these issues in conversation.
25
Lewit, Agricultural production. J. Banaji, Agrarian Change in Late Antiquity: Gold, Labour, and
Aristocratic Dominance (Oxford, 2001), esp. p. 16. Balmelle, Les demeures aristocratiques. Recent
discoveries of second- to third-century abandonment and fourth-century reoccupation: L. Schneider,
`Dynamiques spatiales et transformations de l'habitat en languedoc méditerranéen durant le
haut Moyen Age (VIe­IXe s.)', in Brogiolo et al. (eds), Dopo la fine delle ville, pp. 287­312.
26
Favory et al., `Dynamique spatio-temporelle', pp. 297­9, 307­10. Archaeomedes, Des oppida,
pp. 110­14.
27
Gelichi, `La transizione dell' antichit al medioevo'. Librenti, `Ricognizione di superficie e
insediamento'. Valenti, `Formazione dell'insediamento altomedievale'. C. Magrini, `Il territorio
di Aquileia tra tardoantico e altomedioevo', Archeologia Medievale 24 (1997), pp. 155­71.
28
As suggested by Favory et al., `Dynamique spatio-temporelle', pp. 307­10. A start in considering
the general problem, though only regionally, is J.-L. Fiches (ed.), Le IIIe sicle en Gaule
Narbonnaise (Sophia Antipolis, 1995).
29
Banaji, Agrarian Change in Late Antiquity, pp. 83­120.
The climatic anomaly of the early Middle Ages 139
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
to base a narrative of rural change in these later centuries.30
That has all
changed as the careful identification of fifth- and post-fifth-century pottery
in many regions has established that the villas still occupied in the fourth
century were abandoned in the course of the fifth or early sixth centuries,
or were profoundly impoverished in ways that match the story of Pré-Bas.31
The crunch, or in less imagistic terms, the transition, is clearly visible
in the Lunellois data (Fig. 1). After the end of the first century `bubble'
had taken its toll of smaller villas and farmsteads, it was the turn of the
larger and more prosperous. The handful that did not vanish by 400
would probably prove, if carefully excavated, to have gone through an
impoverishment equivalent to the villa of Prés-Bas: parts falling to ruin,
other parts rebuilt with spolia, wood, wattle and daub, colonnades
divided into living spaces or artisan workshops, all summed up in Tamara
Lewit's colourful phrase, `bones in the bathhouse'.32
A second aspect of
the Lunellois data must be noted as well: new foundations of the fourth
and fifth centuries were relatively fragile, lasting only one or two centuries.
And, of course, a number of those `sites' may be churches or cemeteries,
with or without known attendant habitations.
Wherever archaeologists have recently looked, the situation is similar:
a further, and radical, thinning out of fourth-century habitation sites
during the fifth and sixth centuries and the impoverishment of the few that
remained. This has been confirmed in the region of Metz in northeast
Gaul, in the Eifel west of the Rhine, in the Paris basin, along the
Danube frontier, in northern and southern Italy.33
The ruin of baths,
30
Lewit, Agricultural Production, Ch. 5.
31
For example, M. Py (ed.), Dicocer: dictionnaire des céramiques antiques en Méditerranée nordoccidentale:
VIIme s. av. n.  ­ VIIme s. de n. ., Provence, Languedoc, Ampurdan (Lattes, 1993).
32
T. Lewit, `Bones in the Bathhouse: Re-Evaluating the Notion of "Squatter Occupation" in
Fifth to Seventh Century Villas', in Brogiolo et al. (eds), Dopo la fine delle ville, pp. 251­62.
33
T. Lewit, `"Vanishing Villas": What Happened to Elite Rural Habitation in the West in the
5th­6th C.?', Journal of Roman Archaeology 16 (2003), pp. 260­74, whose argument that this
was only a `cultural change' and not a sign of impoverishment and depopulation, I find
implausible. Metz: G. Halsall, `The Merovingian Period in North-East Gaul: Transition or
Change?', in J. Bintliff and H. Hamerow (eds), Europe between Late Antiquity and the Middle
Ages (Oxford, 1995), pp. 38­52. Eifel: K.H. Lenz, `Late Roman Rural Settlement in the
Southern Part of the Province of Germania Secunda in Comparison with other Regions of
the Roman Rhineland', in P. Ouzoulias, P. Van Ossel et al. (eds), Les Campagnes de la Gaule
 la fin de l'Antiquité (Antibes, 2001), pp. 113­46. Paris basin: P. Van Ossel, `Structure,
évolution et statut des habitats ruraux au Bas-Empire en Ile-de-France', in Ouzoulias and Van
Ossel (eds), Campagnes de l'Ile de France, pp. 94­118, and P. Van Ossel and P. Ouzoulias,
`La mutation des campagnes de la Gaule du Nord entre le milieu du 3e sicle et le milieu du
5e sicle. O en est-on?', in M. Lodewijckx (ed.), Belgian Archaeology in a European Setting,
2 vols (Louvain, 2001), II, pp. 231­45. Danube: C. Siffre, `Kontinuität und Bruch entlang der
Donau (4­8 Jahrhundert)', in S. Biegert et al. (eds), Kontinuitätsfragen Mittlere KaiserzeitSpätantike,
Spätantike-Frühmittelalter (Oxford, 2006), pp. 71­5. Northern Italy: S. Gelichi et al.,
`La transizione dell' antichit al medioevo nel territorio dell'antica Regio VIII', in Brogiolo
et al. (eds), Dopo la fine delle ville, pp. 53­80. Southern Italy: A.R. Staffa, `Riasseto urbano,
trasformazioni territoriali, forme di acculturazione nell'Abruzzo Bizantino (secc. VI­VII)', in
Boldrini and Francovich (eds), Acculturazione e mutamenti, pp. 315­60.
140 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
the subdividing of mosaic floors with light partitions made of wood and
scavenged materials, the planting of `sunken huts' in courtyards, and the
disappearance of all signs of luxury in an accumulation of garbage
and crumbling walls, may be a sign of `squatter' occupation or not:
archaeology cannot tell us anything about the legal rights of the people
who lived in these places, nor their connection to previous owners.
They are all surely signs of impoverishment. The major exceptions to
this story so far discovered in the west (and their location is significant)
were along the coasts of the Adriatic and the Gulf of Taranto.34
I will
return to these in a moment.
If we compare the data from northern Gaul to that of the Mediterranean
we see the same pattern emerging. In 1992, P. Van Ossel summarized
the results of a multiplicity of surveys in north-eastern Gaul.35
They
showed that around Trier, on the rich soil of the Rhineland west of
Cologne, in the Pays de France between the Seine and the Oise, and
doubtless other regions of the Picard plain and the north of France,
about three-quarters of the sites occupied in the third century were still
occupied in the fourth. In contrast, around Nijmwegen and on the
sandy soils of the northern Rhine, perhaps 80% of the inhabited sites
were abandoned between the third and the fourth century. Then, even
where antique occupation had held its place, as in the region of Hambach
(east of Jülich), desertion followed in the fifth century.36
During the
course of the fifth century almost all of the last Roman-style villas and
farmhouses disappeared in north-eastern Gaul. Van Ossel concluded
that in a vast region stretching from the Ardennes and central Belgium
through northern France, it is difficult to find establishments even in
the fourth century that presume a high economic productivity. This is
true both of buildings in stone and buildings in wood. By the fifth
century such impoverished sites had become the substantial majority.
And this economic downturn was not compensated by increased
activity in the larger villas. In the archaeological material, he says, one
witnesses a levelling to the bottom.
The regional story is picked up by Édith Peytremann's catalogue of
308 excavated sites in an area that covers the northern regions of modern
France (thus not including the Belgian and Rhineland areas in Van
Ossel's summary, but encompassing regions west and south of the Paris
34
See the articles by S. Gelichi, G. Volpe and P. Arthur in L'Archeologia dell'Adriatico dalla
preistoria al medioevo (Florence, 2003). G. Volpe, `Villaggi e insediamento sparso in Italia
meridionale fra tardoantico e altomedioevo: alcune note', in Brogiolo et al. (eds), Dopo la fine
delle ville, pp. 221­49; Gelichi et al., `La transizione dell' antichit al medioevo'.
35
Data in Van Ossel, Etablissements ruraux de l'Antiquité tardive. Comment in Van Ossel and
Ouzoulias, `Mutation des campagnes'.
36
Lenz, `Late Roman Rural Settlement'.
The climatic anomaly of the early Middle Ages 141
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
basin).37
Her tabulation of occupation dates can be represented graphically
in a manner that allows a direct comparison with Figure 1.
Because a number of these sites were excavated before the pottery
chronology of the fifth, sixth, and seventh centuries was clarified, the
dating of a number of them can be fixed only within a range of several
centuries. Those sites are indicated in gray. There is a further caution:
most of these sites were discovered in rescue operations. It is therefore
possible that site portions where occupation could have persisted lay
beyond the areas available for excavation.38
The graph nevertheless
shows the relative fragility of new foundations through this transitional
three-century period as does the data from the Lunellois. Of the
nineteen sites in Peytremann's survey that were already established in the
fourth century (or newly created then), four lasted two centuries or less.
One more was re-occupied after a possible hiatus of a century or more.
Of the remaining fourteen, three had vanished before the end of the
sixth century and three more by the end of the seventh. Four survived
into the tenth century and the remaining four into the eleventh,
twelfth, and thirteenth. Twenty-four new sites were first established in
the fifth century. Of these, eight were abandoned by the end of the
sixth century. To these we can add another four sites less securely dated
but occupied sometime during the fifth to sixth centuries. Fifty-nine
new sites were first established in the sixth century, to which we can
add another twenty-two of possible sixth-century foundation. Nearly
half of these were likewise short-lived: only forty-five showed signs of
occupation beyond the end of the seventh century, and most of these
went on into the tenth century or beyond. For many of these new early
medieval creations of the fourth to sixth centuries, something created
a bottleneck. Only those that passed through this transition period
survived to near or beyond the year 1000.
In Peytremann's data, as in that from the Lunellois, it is clear that
we are not witnessing a desertion of the countryside. The rural world
remained inhabited from the fifth century to the seventh (otherwise
who would have been buried in all those cemeteries that for a long time
were our only archeological evidence for the Merovingian period in
northern Gaul?). It was, however (and with only localized exceptions),
far more thinly populated than it had been at the end of the second
century or even in the fourth. Most important of all, some distinguishing
features of that antique countryside ­ and, as we will see, perhaps all were
vanishing. Whatever new structures were taking their place, they
37
Peytremann, Archéologie de l'habitat rural.
38
My thanks to Isabelle Catteddu for clarifying all these issues and for identifying the
problematic dates in Peytremann's catalogue. In the following discussion I have included only
those sites whose dating is relatively certain (i.e., indicated in black on the graph).
142 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
Fig. 2 Numbers of sites in northern Gaul by date of foundation and longevity
The climatic anomaly of the early Middle Ages 143
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
were very fragile, both in terms of their physical properties (light and
perishable construction) and in terms of their continued use over time.
The most striking change is that all the settlements in Peytremann's
catalogue, even those dating from the fourth century, were built in a
manner that resembled the Iron Age settlements beyond the Roman
limes. Even when those of the fourth and fifth century (and even some
of the sixth) occupied the sites of earlier Roman farmhouses, they were
not built in the Roman manner. They were timber-framed one-room
or two-aisled structures, and sometimes nothing more than a `sunken
hut' (a form that is found over almost all the surveyed area by the sixth
century), occasionally associated with other structures that archaeologists
have interpreted as granaries or workshops. From the point of view of
imperial Rome, these settlements were something radically new in the
landscape. From the point of view of Iron Age Europe beyond the
Roman frontier, they are a sign that northern Gaul had rejoined their
world. And not just northern Gaul, for the same type of structures were
being built in rural Italy, and by the sixth century in urban Italy as
well.39
Peytremann points out, furthermore, that the passage from the seventh
to the eighth century was also the period when a major change occurred
in the plan of these settlements. At its beginning a majority of the
newly created rural habitations were still dispersed farmsteads following
the antique pattern; at its end almost all new sites were groups of
farmsteads that shared production and storage areas, truly proto-villages.
In the sixth century, rural settlements still followed the alignments and
the field boundaries that survived from the ancient world; by the
eighth, peasants were creating new and different field plans. And it is
in the eighth century that one begins to find churches and chapels in
the midst of farmyards and the living sharing their space with the
dead.40
Again, this transformation is sooner or later common to all the
European lands that had once been part of the Roman empire.41
After
a long decline in inhabited sites since the second or third century, with
39
M. Valenti, `Forme abitative e strutture materiali dell'insediamento in ambito rurale toscano
tra Tardoantico ed Altomedioevo', in G.P. Brogiolo (ed.), Edilizia residenziale tra 5 e 8 secolo
(Mantua, 1994), pp. 179­90. G.P. Brogiolo and S. Gelichi, La citt nell' alto medioevo italiano:
Archeologia e storia (Rome, 1998), Ch. 4.
40
Peytremann, Archéologie de l'habitat rural, I, p. 335.
41
Here my interpretation differs from Wickham, Framing the Early Middle Ages, pp. 476­81.
His focus, however, is not on numbers and physical properties but rather on the villa as a
cultural and economic phenomenon. Compare Peytremann's data to the schematic flow chart
in Christie, Constantine to Charlemagne, p. 493. Francovich and Hodges, Villa to Village.
Gelichi, `Transizione dell'antichit al medioevo'. S. Gutiérrez-Lloret, `La experiencia
arqueológica en el debat sobre las transformaciones del publamiento altomedieval en el SE de
al-Andalus: el caso de Alicante, Murcia y Albacete', in E. Boldrini and R. Francovich (eds),
Acculturazione e mutamenti (Florence, 1995), pp. 165­89. Schneider, `Dynamiques spatiales'.
144 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
a revival in some regions in the fourth, the passage through the fifth
and early sixth century is the beginning of a major transition. The last
surviving major physical structures of the antique rural world are in
ruins or disappear by the end of the sixth century. After a difficult
passage through the seventh century, a very different organization of the
rural world begins to appear in the eighth.
The evidence of field structures
Up to now we have been looking at buildings in the landscape, dwellings
and farmhouses. The rural landscape, however, is far more than farm
buildings. There is also the constructed terrain, consisting of roads, field
boundaries (fenced and/or ditched), terraces or other structures to hold
soil in place, ditches to drain and ditches to irrigate. The very existence
of arable, in fact, implies constant upkeep, implies structures, in the
sense of human actions to arrest natural processes. Signs that such
upkeep has been abandoned implies, at the very least, a change in land
use, for without means to drain off excess rainfall or to irrigate when
and where necessary, crops will fail; where fields have not been cleared
and ploughed there will be no domesticated plants and natural succession
takes over, as any backyard gardener knows ­ grass and weeds, then
shrubs and trees. This succession also leaves its marks over time and can
be read ­ as superimposed field patterns, as crop marks, damp marks,
or frost marks in aerial photographs, in soil stratigraphy, in fossil pollen
samples.
Where these changes can be dated, they give us direct evidence for
human impact on the rural environment and tell us what happened in
the increasingly impoverished and thinly populated west of the fifth,
sixth, and seventh centuries. And what they reveal is a decrease in
arable, a reversion of ploughed lands to pasture for cows, sheep and
goats, or, often enough, to forest for pigs, as mixed agriculture turned
increasingly to stock raising. Here again, the evidence points to fewer
mouths to be fed, fewer stomachs to be filled, and both as cause and as
consequence, fewer hands working the soil. As we explore this narrative
of fields, we will once again move from a broadly structured storyline,
made evident now by radical changes in field plans, to the type of data
that could in time spell out more precisely the rhythms and chronology
of change, region by region.
Some of the most striking examples of the disappearance of Roman
field systems and the physical structures that implanted them in the
landscape are in the same general area where the Archeomedes survey
was made and in neighbouring areas west along the Mediterranean
coast. Roman centuriation practices had long been known from the
The climatic anomaly of the early Middle Ages 145
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
texts of Roman surveyors, and recognized in aerial photographs taken
during the Second World War. But it was in Orange in the lower Rhône
valley that André Paganiol discovered the first physical documentary
record.42
Since then, historical geographers have discerned in the
landscape the phantom traces of a number of Roman centuriations
belonging to the ancient cities of Orange, Nîmes, Béziers and Narbonne,
and the search for others, not only in Mediterranean regions but in the
north as well, has become a minor industry. It is important to emphasize
the adjective `phantom' in the previous sentence, for in contrast to
some regions of Italy (in the Po valley, for example), to spot what may
remain of the alignments of those ancient landscapes has sometimes
demanded sophisticated computer analysis of aerial photographs.43
The Mediterranean landscape in Figure 3, a mosaic of photographs
taken in 1968 by the Institut Géographique National, shows the
beginning and the end of this narrative of transformation and, by
implication, sets many of the questions that need to be answered.44
The
area covered is the countryside south-east of Béziers in southern France,
between the city, just off the top of the photograph and the Mediterranean,
just off the bottom. The field patterns (primarily vineyards
when the photographs were made) are dominated by the roads radiating
out from three villages: Vendres on its lagoon (just off the bottom left),
Sauvian, and Serignan, both on the Orb river. (The radiating fields of
Sauvian are partially, and those of Serignan are principally on the other
side of the river from the villages because they were laid out before
the river changed its course in the mid-thirteenth century.) There are,
however, a few field-roads that appear to belong to another non-radiating
structure. They are orthogonal and run north­south. The longest, to
the left of the picture, turns off the Vendres-Sauvian road and runs
straight to the river flood plain just south of Béziers. At the time the
photo was made it was still a dirt field-road; when the autoroute was
built two decades later this field-road was paved to link the highway to
the beaches. This road and some much shorter parallel roads and field
boundaries are all that remain here of the ancient centuriated landscape,
intensively studied by Monique Clavel.45
Yellow arrows point to traces of
this centuriation. Orange and green arrows point to medieval radiating
42
A. Piganiol, Les documents cadastraux de la colonie romaine d'Orange, Gallia suppl. 16 (Paris,
1962).
43
Centuriated landscapes: J. Bradford, Ancient Landscapes: Studies in Field Archaeology (London,
1957), and above all, Misurare la terra: centuriazione e coloni nel mondo romano. Catalogo del
Mostra (Modena, 1983). Polemical critique of the ancient cadaster scholarly `industry' by one
of its early promoters: G. Chouquer, L'etude des paysages: essais sur leurs formes et leur histoire
(Paris, 2000).
44
For a detailed analysis of this landscape see Amado and Cheyette, `Organisation d'un terroir'.
45
M. Clavel, Béziers et son territoire dans l'Antiquite (Paris, 1970).
146 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
field-roads from Sauvian and Serignan, cut by the Orb river when it
changed its course.
At the very top of the picture, where in 1968 the first major urban
expansion was taking place, one can see significantly more of the ancient
centuriated alignment in an area called St-Jean-d'Aureilhan. Here a
field-road runs due east from the city, and then for no apparent reason
turns resolutely ninety degrees to the south, in the same orientation as
Fig. 3 Béziers (Hérault, FR) (Photo:  IGN)
The climatic anomaly of the early Middle Ages 147
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
the orthogonal north­south field-roads south of the city and the river.
Immediately to the south of the main road are several shorter roads at
precisely the distance (twenty actus) of the standard Roman centuriation.
Some of these roads are now city streets, others have disappeared as the
area has been completely urbanized. But rescue archeology in different
sectors, including the vicinity of the medieval chapel that gave the area
its name, found that one field-road was superimposed on a Roman era
ditch. Excavations nearby found other ancient ditches aligned with this
centuriation (which Monique Clavel labelled `Béziers A.'), while yet
others (probably later) were oriented in a different direction, suggesting
two different phases of antique occupation. These excavations confirmed
that the visible alignments indicated in Figure 3 do indeed follow that
of the first Roman centuriation. The fill in the earlier ditches could not
be dated; that in the later ditches contained pottery from the fourth to
the early sixth century. At the same time, the excavators showed that
on this site there was a break in occupation during the seventh to ninth
centuries. Three hundred metres to the north, however, there was an
ample spread of pottery from the fifth/sixth centuries to the ninth/tenth,
associated with storage pits, postholes, and ditches.46
Documentation of this area in the cartulary of Béziers cathedral
begins in the tenth century.47
By then the present-day villages ­ as well
as one that vanished before the mid-twelfth century ­ already existed.
The documentation is sufficiently dense to allow the identification of
field-roads mentioned in tenth-century charters with those visible in the
twentieth-century photographs. By the tenth century, and probably at
least several generations earlier, the basic structure of the medievalmodern
landscape had been created, with its central villages and their
radiating field-roads. With the exception of the few remaining traces
just described, the structures of the ancient landscape had vanished.48
How had this happened? And when?
The most likely sequence is the following. The physical structures roads
and ditches ­ that fixed Roman field boundaries in the landscape
and provided the drainage so essential to agriculture in a Mediterranean
climate were no longer maintained. They were allowed to fill up with
alluvial soil, especially those ditches that did not run with the slope of
46
O. Ginouvez and H. Pomaredes, `Premieres observation sur les sites antique et médiéval de
Saint-Jean d'Aureilhan commune de Béziers', Archéologie en Languedoc 17 (1993), pp. 157­71.
Supplemented by Ginouvez et al., `Saint Jean d'Aureilhan, fouilles programmées 1994,
Parcelles HX 241, 258', manuscript report, Direction Régional des Affaires Culturelles, Service
Régional d'Archéologie [hereafter ms. DRAC SRA], Montpellier (1994). Ginouvez et al.,
`St Jean d'Aureilhan, Sauvetage Archeologique', ms. DRAC SRA, Montpellier (1993). And
P. Chevillot et al., `Beziers ZAC de Montimaran', ms. DRAC SRA, Montpellier (1998).
47
J. Rouquette (ed.), Cartulaire de Béziers: livre noir (Paris, 1918).
48
See Amado and Cheyette, `Organisation d'un terroir'.
148 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
the land. This would explain why some north­south ditches and
associated roads would survive or occasionally be cleared, for that is the
direction of natural drainage, while east­west ditches and roads, without
practical purposes, were abandoned to natural processes. This could
only mean that fields were no longer cultivated but were allowed to
return to grass, and probably here and there to scrub and wood. Roman
boundaries, no longer meaningful, vanished from the land (though
perhaps not always from memory). Place names, however, survived here
and there. An early eleventh-century charter mentions a substantial
tract of land called Licinianum (Licinius's place?) east of the flood plain
of the Orb, between the villages of Cers and Villeneuve. Its eleventhcentury
boundaries, however, had nothing to do with ancient centuriation.
They were determined by field-roads that radiated from the nearby
village, by the river, and by a wandering `carraria que discurrit per loca
multa' which can still be identified in the photos of the 1968 aerial
survey.49
When medieval peasants ploughed this land once again, it
would have been easy for them to spot where ancient drainage ditches
once had been. They would stay waterlogged for a longer time after
heavy rains, and in time perhaps be overgrown with shrubs and trees.
Yet it would have been easier to re-dig the old ditches than to dig new
ones where none had been before. In this way ancient alignments could
reappear here and there after centuries of abandonment, but now in the
context of very different field structures with very different field shapes
and property boundaries.
This narrative is not entirely imagined. Exactly such a sequence of
events has been uncovered on the plain of Orange in the middle Rhône
valley and dated with relative precision. In the commune of Lapalud, a
few kilometres north of the city, archeologists working in advance of
the construction of the TGV were able to test the identification of the
hypothesized `centuriation B' of Orange and not only confirm its
existence and orientation but also identify two other ancient cadasters.50
A complex sequence overlay the earliest antique cadastration. The two
`centuries' uncovered (whose exact numbering could be extrapolated
from the recovered fragment of the Roman cadastral map discovered by
Paganiol) were originally co-planted in grain and vines, but by the last
quarter of the first century CE had been converted to pasture. Already
49
Cartulaire de Béziers, no. 57.
50
J.-F. Berger and C. Jung, `Developing a Methodological Approach to the Evolution of Field
Systems in the Middle-Rhône Valley', in P. Leveau et al. (eds), Environmental Reconstruction
in Mediterranean Landscape Archaeology, The Archaeology of Mediterranean Landscapes 2
(Oxford, 1999), pp. 155­67, and reports on geomorphology by Berger and Jung in P. Boissinot
et al., `Lapalud: Les Girardes. Archéologie et TGV: Lot 21. Rapport de Fouille', ms. DRAC
SRA, Aix-en-Provence (1998) and M. Goy and I. Remy, `Montboucher-sur-Jabron: Constantin.
Archéologie et TGV: Lot 12. Rapport de Fouille', ms. DRAC SRA, Lyon (1996).
The climatic anomaly of the early Middle Ages 149
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
in the early second century this was a less humanized landscape and was
showing signs of heavy erosion. The ditches that defined the centuriation,
write the excavators, appear to have silted up `almost instantaneously'.
In the third century, with a rising water table, the drainage system was
completely abandoned, and 15 to 20 centimetres of soil slowly accumulated.
Then, in the fourth century, cereal pollen, missing since the late first
century, reappears in the samples. Drainage ditches were re-cut. Arboreal
pollen, however, is now over 30% of the total, and towards the end of
the fourth or in the early fifth century, alluvial soil once again filled
these abandoned ditches. The ditches were invaded with hydrophilic
plants and pioneering species, and over the ensuing centuries the soil
level rose another 20 to 40 centimetres. Eventually, perhaps as late as
the twelfth or thirteenth century, new ditches were cut where those of
antiquity had been.51
A similar sequence was discovered not far away at
Pierrelatte, `les Malalognes', where the entire sequence of early ditches
was covered in the seventeenth and eighteenth century by alluvial soil
carried by a local stream, yet a modern hedge was planted barely 2 metres
from the antique ditch.52
In this manner the antique alignment was
repeatedly recreated through abandonments and reoccupations, over
two millennia, while agriculturalists were succeeded by pastoralists, the
land abandoned, and reoccupied over several cycles. The story is enough
to make one wonder whether the apparent maintenance of centuriated
fields in the Po valley, for example, is an illusion that masks another
case of recreation or even repeated recreations, rather than continuity
of arable use.53
Ideally, one would like to have such careful stratigraphy of field
structures from the Béziers plain, indeed, from all over Europe. In its
absence one can only hypothesize, as I have done, that the superposition
of medieval field patterns over radically different antique patterns hides
the same story. It is a story of decay and abandonment of the physical
infrastructure of production that follows the same complex curve as the
site data summarized in the first section of this article. Field ditches
were abandoned and grain fields and vineyards were converted to pasture
as the first-century `bubble' was reaching its end. Erosion followed,
and a slow reforestation. Then renewed investment of labour in arable
agriculture in the fourth century, only to be followed by abandonment,
51
Boissinot et al., `Lapalud: Les Girardes'.
52
J.-F Berger et al., `Données paléogéographiques et données archéologiques dans le cadre de
l'opération de sauvetage archéologique du TGV-Méditerranée', in J. Burnouf et al. (eds), La
dynamique des paysages protohistoriques, antiques, médiévaux et modernes (Sophia Antipolis,
1997), pp. 155­84.
53
See for example, M. Cremaschi, `Il territorio mantovano fra il Boreale e l'et romana:
l'evoluzione ambientale', in Misurare la terra, III, pp. 13­17, describing medieval field patterns
overlying centuriation-oriented ditches filled with alluvial soil.
150 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
alluviation, and a rise in the water table. Either money, or manpower,
or the will, was lacking to keep these fields in production.
In the plains of northern Gaul and Britain the evidence of a sharp
break between antiquity and the Middle Ages is no less clear. Aerial
photographs showing Iron Age and Roman period enclosures, farmsteads
and villages, as well as `Celtic fields' eventually overlain by the long
strips of medieval open fields have often been published, especially from
sites in the English midlands. The same phenomena are visible all across
the northern plains of Europe and have been confirmed by excavation.54
Figure 4 is an example from Brittany. It shows proto-historic enclosures
and traces of antique field systems revealed by aerial photographs, here
overlain on an early nineteenth-century cadastral map whose boundaries
are directly inherited from the strips of medieval open fields.55
What we are seeing both north and south is thus a sharp reversal of
a development that began with the spread of arable agriculture in the
late Bronze Age. In the passage from one set of structures to others that
are very different, the rural economy of mixed arable and stock-raising
abandoned much of the arable (that, among other purposes, fed the
ancient cities and the imperial armies) and moved sharply in the direction
of a silvo-pastoral economy.
For the regrowth of forests there is plentiful evidence from all over
western Europe. From the English midlands to Picardy to the Po valley,
evidence has long since emerged for extensive Roman farming in land
where forests were being cleared in the twelfth century.56
Two striking
examples come from the Rhineland where two supposedly `primeval'
forests turned out to be regrowth of the early Middle Ages. In the Eifel,
east of Jülich, the clearing of the forest of Hambach for coal mining
revealed that it had been cleared and densely occupied in the Roman
period and abandoned by the sixth century, for the few Merovingian
sites in the region were all outside the limits of the forest except for two
glass-making establishments. The Kottenforst not far to the south was
likewise densely occupied in the Roman period and abandoned by the
54
In northern Germany and Scandinavia the change has been dated to the seventh century,
though there is some evidence from southern Scandinavia that in some locations it may have
occurred as early as the third; in Anglo-Saxon England it has been dated to the eighth
century: H. Hamerow, Early Medieval Settlements: The Archaeology of Rural Communities in
Northwest Europe 400­900 (Oxford, 2002), pp. 140­4, 152.
55
D. Marguerie et al., `Bocages armoricains et sociétés, gense, évolution et interactions', in T. Muxart
et al. (eds), Des milieux et des hommes: fragments d'histoires croisées (Paris, 2003), pp. 115­31.
56
A representative sample: C. Lewis, P. Mitchell-Fox and C. Dyer, Village, Hamlet and Field:
Changing Medieval Settlements in Central England (London, 2001), pp. 76­81. The photographs
of Estrées-s-Noye in R. Agache, Atlas d'archéologie aérienne de Picardie: le bassin de la Somme
et ses abords  l'époque protohistorique et romaine (Amiens, 1975), pp. 317, 385 where the relic
of the medieval forest and even a charcoal-burning pit impinge on a great Roman villa.
G. Rippe, Padoue et son contado: Xe­XIIIe sicle (Rome, 2003), pp. 45­57, especially the case
of the forest of Viminario, pp. 54­5.
The climatic anomaly of the early Middle Ages 151
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
sixth century. Unfortunately only 19 of the well over 200 Roman sites
in and around the forest of Hambach were excavated, and there is no
published data on the pottery collected during the rescue operations in
either of the two forests.57
A similar phenomenon with similar dating
57
W. Janssen, `Römische und frühmittelalterliche Landerschliessung im Vergleich', in W. Janssen
and D. Lohrmann (eds), Villa­curtis­grangia: Landwirtschaft zwischen Loire und Rhein von
der Römerzeit zum Hochmittelalter (Munich, 1983), pp. 81­122.
Fig. 4 St-Brieuc-de-Mauron (Morbihan, Brittany). Source: Marguerie et al.,
`Bocages armoricains'
152 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
has been documented, mainly by studies of pollen, from the Rhineland
all the way east to modern Poland.58
But it was not limited to areas that
ancient and modern myth-making has long associated with primeval
impenetrable forests. It was equally true of central Gaul, where a study
of eighty-eight lake and bog sites indicates reforestation beginning in
the fourth century and lasting until the middle of the ninth.59
So, too,
for areas along the Mediterranean coast and the Cantabrian mountains
of north-west Spain.60
As might be expected, the dates for these developments
vary from place to place, since so much depended on local factors.
But the tendency can be found everywhere. To be sure, reforestation
did not mean an absence of human activity. Quite the contrary, there
are many signs of forest use, for charcoal, for pottery-, glass-, and
iron-making, for construction, and, perhaps most importantly, for
feeding pigs. Yet the decline in human impact is evident.61
Even the pastures, apparently, were not maintained, or at least not
sufficiently to support the more robust cattle and sheep that Iron Age
and Roman pastoralists had bred. The study of animal bones in fifthcentury
and later archeological sites reveals a steady decline in the stature
of sheep, cows, oxen, and pigs (though not of horses).62
The most likely
explanation is that larger and more robust animals were no longer
58
Lenz, `Late-Roman Rural Settlement'. B. Zolitschka and J.F.W. Negendank, `A High
Resolution Record of Holocene Palaeohydrological Changes from Lake Holzmaar, Germany',
in S.P. Harrison et al. (eds), Palaeohydrology as Reflected in Lake-Level Changes as Climatic
Evidence for Holocene Times, Paleoclimate Research 25 (Mainz, 1998), pp. 37­52. `Environmental
Change and Human Impact in the Middle Lahn Valley (Hessen) during the
Middle and Late Holocene', <http://web.uni-frankfurt.de/fb11/ipg/spp/Postergallery/Poster_pdf/
Poster_Urz_etal.pdf>. C. Siffre, `Kontinuität und Bruch entlang der Donau (4­8 Jahrhundert)',
in S. Biegert et al. (eds), Kontinuitätsfragen Mittlere Kaiserzeit-Spätantike, Spätantike-Frühmittelalter
(Oxford, 2006), pp. 71­5. S.T. Anderson, `Local and Regional Vegetational Development in
Eastern Denmark in the Holocene', Denmarks Geologiske Undersogelse. Arbog (1976), pp. 5­27.
K.E. Barber, F.M. Chambers and D. Maddy, `Late Holocene Climatic History of Northern
Germany and Denmark: Peat Macrofossil Investigations at Dosenmoor, Schleswig-Holstein,
and Svanemose, Jutland', Boreas 33 (2004), pp. 132­44, at p. 142. See also the pollen diagram
from Lake Woryty, Poland in B.E. Berglund (ed.), Handbook of Holocene Palaeoecology and
Palaeohydrology (New York, 1986), Fig. 22.10.
59
J.-L. de Beaulieu, A. Pons and M. Reille, `Histoire de la flore et de la végétation du Massif Central
depuis la fin de la dernire glaciation', Cahiers de micropaleontologie, ns 3 (1988), pp. 5­35.
60
C. Raynaud, Le Village gallo-romain et médiéval de Lunel-Viel (Hérault) (Paris, 1990), pp. 322­3,
327. So also in the Cantabrian mountains of north-west Spain: C. Muoz Sobrino et al., `Palynological
Data on Major Holocene Climatic Events in NW Iberia', Boreas 34 (2005), pp. 381­98.
61
L. Bourgeois, `Espaces boisées, pôles d'habitat, et occupations marginales de l'Antiquité au Haut
Moyen Age', in Ouzoulias and Van Ossel (eds), Les Campagnes de l'Ile de France, pp. 32­4.
62
S. Lepetz et al., `Culture et élevage en France septentrionale de l'Age du Fer  l'an Mil', in
A. Belmont (ed.), Autour d'Olivier de Serres: pratiques agricoles et pensée agronomique, Bibliothque
d'Histoire Rurale 6 (Paris, 2002), p. 96. S. Deschler-Erb et al., `La crise de l'empire romain au
IIIe sicle aprs J.-C: les données archéozoologiques en Suisse du nord', in H. Richard and A.
Vignot (eds), Equilibres et ruptures dans les écosystmes depuis 20 000 ans en Europe de l'ouest, Annales
littéraires de Besançon 730 (Paris, 2002), pp. 281­90. F. Salvadori, `Resti osteologici animali:
elementi di continuit e discontinuit tra tardoantico e altomedioevo', in R. Francovich and
M. Valenti (eds), IV Congresso Nazionale di Archeologia Medievale (Florence, 2006), pp. 520­4.
The climatic anomaly of the early Middle Ages 153
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
selected to be fattened for market; meanwhile, in pasture no longer
regularly hayed but invaded by woody plants, the lower metabolic needs
of smaller animals would have given them a competitive advantage.63
In all this we are seeing exactly what we might expect from the siteoccupation
data. The rural population was thin and widely dispersed,
and it was far less capable of supporting a non-agricultural population
than its fourth-century predecessor had been. If Roman material culture
is taken as the standard of wealth, it was a far more impoverished
population as well.
This correspondence with site-occupation data is suggestive, as is the
chronological correspondence of the two carefully excavated sites in the
lower Rhône valley. Are there any other independent data that might
eventually give us more precise dates for this transformation and enrich
our understanding of the changes occurring in the environment? The
answer is yes, in studies of fossilized pollen, such as those just mentioned,
which allow paleobotanists to investigate the environment surrounding
the sites from which the samples have been taken.
Figure 5 comes from the site of Glatinié, near the city of Laval, west
of Paris.64
Pollen diagrams are made by extracting a soil core from a
bog or other anaerobic environment, cutting it into thin slices, then
identifying and counting the fossilized pollen in each slice. The diagram
63
See M.W. Demment, and P.J. Vansoest, `A Nutritional Explanation for Body-Size Patterns
of Ruminant and Nonruminant Herbivores', American Naturalist 125 (1985), pp. 641­72.
64
D. Barbier, J. Burnouf and L. Visset, `Les diagrammes Société/Végétation: un outil de dialogue
interdisciplinaire pour la compréhension des interactions Homme/Milieux', Quaternaire
(2001), 12, (1­2), pp. 103­8. Interpretation of the data in this diagram in D. Barbier,
L. Visset, J. Burnouf, `Une source pollinique et son exploitation ­ A propos de la tourbire
de Glatinié (Mayenne)', Histoire et Sociétés Rurales 18 (2002), pp. 137­58.
Fig. 5 Pollen diagram from Glatinié. Source: Barbier et al., `Diagrammes Société/
Végétation'
154 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
is made by representing the percentage of each species at each level from
the present (top) to the most ancient (bottom). Absolute dates are derived
by carbon dating when suitable material is present in the core. In this
diagram the different species have been gathered into general categories
to indicate not only the extent but also the nature of human impact on the
area vegetation. Because not all plants produce airborne pollen, and those
that do so produce it in different quantities, the percentages represented
do not exactly represent the percentage of each species in the local
landscape. They do, however, signal the appearance, growth, decline, and
disappearance of the pollen-spreading species. Because bogs have their
own specialized species, it is necessary to filter those out of the totals
before calculating percentages in order to represent the wider environment
around the bog. The diagram reproduced in Figure 5 presents the
filtered data to the left of the central axis and the unfiltered to the right.
I have reproduced here only the section of the diagram pertinent to
the present argument, between two calibrated C-14 dates: CE 73  91
at the bottom and CE 989  35 at the top.65
An intermediate date of
857  87 is also marked. The diagram is arranged to read groups of
plants outward from the central axis: closest to the axis are cereals and
other cultivated plants, then weeds of grain fields, followed by weeds of
disturbed land (roadsides, around houses and farmyards, on abandoned
sites), grasses, and forest cover. As we should now expect, the agricultural
impact on this environment was at its height in the first century CE,
with mixed cereal, orchard and garden crops and extensive pasture land.
It was a very open landscape (indicated by the percentage of ruderal
species). From then on, the species indicating human presence decline
rapidly. Since there is no proof that soil was accumulating in the bog
at a steady rate it is impossible to date even approximately when forest
took over much of the landscape, only that it did at some moment
between about 400 and about 700. This was followed by a brief revival
of agriculture (peaking around 700­750?) with a notable extension of
pasturage, followed by a less significant retreat (note the continuation
of cereals and accompanying weeds) before the rapid and more securely
dated deforestation and extension of pasture and ploughed land in the
late ninth to tenth centuries. Here again, the uncertain dating of the
earlier changes may not be all that important, since we might expect
that local ecological conditions as well as human decisions would have
a large part in both the chronology and the extent of pasture and forest
expansion and would therefore result in significant regional differences.
65
I have calibrated the raw BP dates given in Barbier et al. using U. Danzeglocke, O. Jöris,
B. Weninger, CalPal online. http://www.calpal-online.de/, accessed 2007/30/1.
The climatic anomaly of the early Middle Ages 155
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
The role of climate
What caused the transformation in productive practices, the decline in
cereal cultivation, and the end of the ancient rural landscape?
Without a doubt, constructing a fully satisfactory answer to this
question would require a complex flowchart with numerous feedback
loops, for given the centuries over which this transformation took place
and the economic and ecological complexities of any agricultural
economy, especially one as highly commercialized as was that of Rome
during the early centuries of the empire, we are doubtless looking at
many causes that were also consequences, tendencies that were selfreinforcing.
Any satisfying answer, furthermore, would have to take into
account the simultaneous changes north and south. It would also
have to take into account changes both within what had once been the
limits of the western empire as well as beyond the camps and fortifications
that marked the limes, beyond even the reaches of Mediterranean
commerce. For, as we have just seen, pollen diagrams from as far east
as Poland show the same rise and fall of cultivated species occurring at
around the same time as the same movements occurred in the west.
The most frequent answer to be found in the literature, whether
archeological or historical, is local political and military events. It is
the invasion of the Saxons (Britain), the invasion of the Franks (Gaul),
the Gothic-Byzantine wars or the invasion of the Lombards (Italy).
Yet, curiously, where careful excavation of late Roman sites has been
undertaken, signs of violent destruction range from exceedingly rare
to non-existent. And often enough, neither the signs nor the direction
of transformation change across such `important' events.66
If warfare
disrupted commerce, as it most likely did, why did the economy not
quickly recover, and the population with it, after marauding armies had
passed or had settled down? Furthermore, the supposed influx of new
immigrants is hard to locate, now that grave-goods are no longer taken
to be `ethnic' identifiers, and the population loss suffered by Europe
beyond the limes is not reflected in population gain where they settled.
Local political conditions ­ the retreat of Roman armies from Britain,
for example, or the `barbarians' crossing the Rhine, the Danube, or the Alps,
whatever the antiquity of this explanation ­ cannot possibly be sufficient.
Other explanations, less localized this time, are the Justinianic plagues
and their sequels, and the so-called `536 event'. Both undoubtedly had
66
For a new version of the old political explanation see Ward-Perkins, Fall of Rome, pp. 128­34.
Yet the argument of P. Delogu, `La fine del mondo antico e l'inizio del medioevo: nuovi dati
per un vecchio problema', in R. Francovich and G. Noyé (eds), La Storia dell'alto Medioevo
italiano (VI­X secolo) alla luce dell'archeologia (Florence, 1994), pp. 7­29 has only been
confirmed by more recent research, and seems as valid for the north as for Italy.
156 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
severe and long-lasting consequences, the plague especially. But they hit
a west that was already well on the road to impoverishment and depopulation.
And they did not long interrupt the sixth-century economic
prosperity of the east, which, far more densely populated than the west,
surely suffered more.67
The cloud of volcanic dust that covered the earth
in 536, a `year without a summer', was recorded in texts from Ireland
to China. Major famines followed in Europe, perhaps continuing as late
as 541.68
While the short-term consequences of this `dry fog' were surely
important, it seems unlikely that the long-term changes in the European
landscape we have seen can be attributed to this one event.
Indeed, there were even stronger volcanic events noted worldwide that
occurred at the beginning of what would be two periods of agricultural
expansion, the eighth century and the eleventh.69
Once again, a particular
event or sequence of events, no matter how dramatic and devastating in
the short term, seems inadequate to the explanatory demands. Something
in the west prevented the region from recovering for centuries.
Was it the break in the commercial links that bound rural producers
to cities and long-distance trade?70
This explanation cannot be dismissed
out of hand, even though, as we have just seen, the decline in cereal
cultivation (and apparently in population as well) affected lands far to
the east of the Roman limes, far beyond the reach of Mediterranean
commerce. Nevertheless, there were a few areas of the old western
empire that, however impoverished, continued to survive long after
others went into decline, and those were the cities of the Adriatic coast,
Emilia-Romagna, and the Gulf of Taranto, as well as Rome and ­ to a
67
Brief narrative in L.K. Little, `Life and Afterlife of the First Plague Pandemic', in L.K. Little
(ed.), Plague and the End of Antiquity: The Pandemic of 541­750 (Cambridge, 2007), Ch. 1.
Literary evidence assembled and translated in D.C. Stathakopoulos, Famine and Pestilence in
the Late Roman and Early Byzantine Empire (Aldershot, 2004). On the prosperity of the east,
see C. Foss, `The Near Eastern Countryside in Late Antiquity: A Review Article', in The
Roman and Byzantine Near East: Some Recent Archaeological Research, Journal of Roman
Archaeology, Supplementary Series 14:1 (Ann Arbor, MI, 1995), pp. 213­34; and Wickham,
Framing the Early Middle Ages, pp. 443­65; disputed by H. Kennedy, `Justinianic Plague in
Syria and the Archaeological Evidence', and P. Sarris, `Bubonic Plague in Byzantium: The
Evidence of Non-Literary Sources', both in Little (ed.), Plague and the End of Antiquity, Chs
4 and 6 respectively.
68
See in general, J.D. Gunn (ed.), The Years Without Summer: Tracing A.D. 536 and its Aftermath,
BAR International Series 872 (Oxford, 2000). Tree-ring evidence analysed in M.G.L. Baillie,
`Dendrochronology Raises Questions about the Nature of the AD 536 Dust Veil Event',
The Holocene 4 (1994), pp. 212­17. A sceptical assessment of volcanic effects (though not
addressing the 536 event) is J. Grattan, `Aspects of Armageddon: An Exploration of the Role
of Volcanic Eruptions in Human History and Civilization', Quaternary International 151:1
(2006), pp. 10­18.
69
See the `volcanic explosivity index' in A. Robock and M. Free, `The Volcanic Record in Ice
Cores for the Past 2000 Years', in P.D. Jones et al. (eds), Climatic Variations and Forcing
Mechanisms of the last 2000 Years (Heidelberg, 1996), pp. 533­46.
70
The explanation favoured by Ward-Perkins, Fall of Rome, esp. pp. 132­7, which must be
considered apart from his blaming it all on the `barbarians'.
The climatic anomaly of the early Middle Ages 157
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
lesser extent ­ Marseille, reconnected to Byzantine trade routes by
Justinian's reconquest.71
Any explanation of the general pattern will also
have to include these exceptions. The possible mechanisms connecting
long-distance trade to rural production have yet to be explored. I will
offer a few hypotheses later. For the moment, it is sufficient to note that,
whatever those connections, the decline of commerce is not sufficient
to explain what is happening in the countryside. Again, the geographical
range of the phenomena remains to be accounted for.
There remains the climate, which, of course, is European-wide
despite the obvious differences between the Atlantic and continental
regimes of the north and the Mediterranean regime of the south. What
follows is a brief review of some of the climatological evidence for the first
millennium CE and some hypotheses to serve as a future research agenda.
Climate change, or more precisely a two- or three-century climate
anomaly has occasionally been evoked by scholars studying the last centuries
of Roman Britain and early Anglo-Saxon agriculture.72
It has also been evoked
by some scholars working on Mediterranean agriculture (though just as
forcefully denied by others).73
Recent research by climatologists has allowed
more precise dating of this anomaly as well as clarifying its nature.74
Since the Bronze Age, cereals have been a large and fundamental
part of the western diet. Two aspects of climate have a determining
71
E. Zanini, `Racontando la terra sigillata africana', Archeologia Medievale 23 (1996), pp. 677­
88. G. Volpe, `San Giusto e l'Apulia nel contesto dell'Adriatico tardoantico', in L'archeologia
dell'Adriatico, pp. 507­36. A. Toniolo, `Importazioni tra IV e VIII secolo d.c. nella laguna di
Venezia', ibid., pp. 616­21. P. Arthur, `I Balcani e il Salento nel medioevo', ibid., pp. 654­65.
Older survey by A.R. Staffa, `Riasseto urbano, trasformazioni territoriali, forme di acculturazione
nell'Abruzzo Bizantino (secc. VI­VII)', in Boldrini and Francovich (eds), Accultrazione e
mutamenti, pp. 315­60. Comparison of site survival around Modena and Cesena with that
elsewhere in Gelichi, `La transizione dell' antichit al medioevo'. L. Sagui, `Indagini
archeologiche a Roma: nuovi dati sul VII secolo', in P. Delogu (ed.), Roma medievale:
aggiornamenti (Florence, 1998), pp. 63­78. Patterson, `Three South Etrurian Crises'. F. Trément,
Archéologie d'un paysage: les étangs de Saint-Blaise (Paris, 1999).
72
Notably by Jones, End of Roman Britain. See also E. Jones, `Climate, Archaeology, History
and the Arthurian Tradition: A Multiple-Source Study of two Dark-Age Puzzles', in Gunn
(ed.), The Years without Summer, pp. 25­34.
73
A. Durand, `Les milieux naturels autour de l'An Mil: approches paléoenvironnementales
méditerranéennes', in P. Bonnassie and P. Toubert (eds), Hommes et sociétés dans l'Europe de
l'An Mil (Toulouse, 2004), pp. 74­9. Durand and Leveau, `Farming in Mediterranean
France', in Barceló and Sigaut (eds), The Making of Feudal Agricultures, pp. 177­241. J.-F. Berger,
`Les étapes de la morphogense holocne dans le sud de la France', in van der Leeuw et al.
(eds), Archéologie et systmes, pp. 144­7, with a curious contradictory assertion by Favory et al.
in the same volume, `Dynamique spatio-temporelle', p. 311. P. Leveau et al., `La crise environnementale
de la fin de l'Antiquité et du Haut Moyen Age: définition d'un modle et retour aux
milieux réels', in Richard and Vignot (eds), Equilibres et ruptures, pp. 291­303, is a sceptical view.
74
The period was labelled the `Vandal Minimum' by R.A. Bryson, `Orbital History, Volcanism,
and Major Climate Changes: On Integrating Climate Change and Culture Change', Human
Ecology 22 (1994), pp. 115­58, but this name has not gained popularity. It is usually now called
simply `the early medieval cold period': see for example M. Benvenuti et al., `Late-Holocene
Catastrophic Floods in the Terminal Arno River (Pisa, Central Italy) from the Story of a
Roman Riverine Harbour', The Holocene 16 (2006), pp. 863­76.
158 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
effect on their production: precipitation and temperature. All plants
are genetically programmed to produce seed within a relatively narrow
range of both water availability and of optimal degree days during the
growing season. Too much or too little rainfall or bad timing of that
rainfall during the growing season will lead to seed rotting or failing to
sprout or, later on, to plants rotting or drying. If frost comes too late
in the spring or too early in the autumn or the growing season is too
cold, the plant will not have time to ripen its seeds. Grains are particularly
sensitive to wet, cold winters.75
There are, of course, no direct instrument
records of either precipitation or temperatures before the nineteenth
century to tell us when such events happened, so climatologists have
been forced to turn to proxy records of a physical nature whose traces
are left in such places as the soil, and arctic and alpine glaciers. Such
data are significantly less precise than instrument records, but with the
development of increasingly sophisticated dating techniques they are
starting to become useful for historical analysis.
Indirect evidence for an extended period of heavy rainfall has long
appeared in the archeological record of Mediterranean Europe, from
the `newer fill' of Claudio Vita-Finzi to recent analyses of accumulated
sediments in Mediterranean river beds.76
The research of J.-F. Berger
and his colleagues along the TGV right-of-way in the Rhône valley has
given dates from the late fourth to the seventh century for an extended
period of torrential rains, erosion, and a rise in the water table, with
fields becoming waterlogged.77
Dating of such soil strata, however, is
75
For scientific details see J.F. Bierhuizen, `The Effect of Temperature on Plant Growth,
Development and Yield', in R.O. Slatyer (ed.), Plant Response to Climatic Factors (Paris, 1973),
pp. 89­98. For historical data from the twentieth century: C. Pfister, Bevölkerung, Klima und
Agrarmodernisierung, 1525­1860, 2 vols (Bern, 1985), II, pp. 35­7.
76
C. Vita-Finzi, The Mediterranean Valleys: Geological Changes in Historical Times (London,
1969). C. Giraudi, `Late-Holocene Alluvial Events in the Central Apennines, Italy', The Holocene
15 (2005), pp. 768­73. C. Morhange, C. Vella et al., `Human Impact and Natural Characteristics
of the Ancient Ports of Marseille and Fos in Provence, Southern France', in Leveau et al.,
(eds), Environmental Reconstruction, pp. 145­53. For the Tiber, see D. Camuffo and S. Enzi,
`The Analysis of Two Bi-Millennial Series: Tiber and Po River Floods', in Jones et al. (eds),
Climatic Variations, pp. 433­50. For the Arno: M. Benvenuti et al., `Late-Holocene Catastrophic
Floods'. And in southern Gaul, the small Vidourle river: J.F. Berger et al., `Villetelle/
Ambrussum', in Service Régional d'Archéologie Languedoc-Roussillon, Bilan Scientifique
(2001). Compare: Mark G. Macklin et al., `Pervasive and Long-Term Forcing of Holocene
River Instability and Flooding in Great Britain by Centennial-Scale Climate Change',
The Holocene 15 (2005), pp. 937­43. B. Vannire et al., `Land Use Change, Soil Erosion and
Alluvial Dynamic in the Lower Doubs Valley over the 1st millennium AD', Journal of
Archaeological Science 30 (2003), pp. 1283­99. F. Trément et al., `Mutations environnementales
et systmes socio-économiques en Grande Limagne (Massif Central Français) de l'Age du Fer
au Moyen Age', in Hervé and Vignot (eds), Equilibres et Ruptures, pp. 269­79.
77
Among others: J.-F. Berger et al., `Géoarchéologie du bassin valdainais', in J.P. Bravard and
M. Prestreau (eds), Dynamique du Paysage: Entretiens de Géoarchéologie, DARA 15 (Lyon,
1997), pp. 103­28. Berger and Jung, `Developing a Methodological Approach', pp. 155­67.
J.-F. Berger and C. Jung, `Données paléogéographiques et données archéologiques', in
Burnouf et al. (eds), Dynamique des paysages, pp. 175­6.
The climatic anomaly of the early Middle Ages 159
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
always problematic. First, because the accumulation of water-borne
soil depends on the plants and man-made structures (such as drainage
ditches, dams, terracing) upstream. Soil without plant cover is eroded
rapidly by excess rainfall. As grasses, shrubs and trees take hold, erosion
will lessen, even in times of torrential rains. On the other hand, the
abandonment of ditches and terrace walls may lead to increased erosion.
Therefore, the accumulation of soil downstream is usually not only an
indirect but also an incomplete proxy record of when and where the
rain was falling. Furthermore, with the exception of rare finds, such as
that of the buried ships in Pisa's ancient harbour, the accumulation
cannot be dated directly by the man-made material it may contain,
since that material has been washed there from somewhere else; it can
only be dated by what is in the strata sealed beneath the alluvion and
the more recent material in the stratum that seals the alluvion in turn.
The error bar therefore may be very wide.78
Recently, however, a group of climatologists have found proxy data
with remarkable time resolution for rainfall along the upper Rhône, in
the north-western Alps. Before 1870, when attempts began to bring the
river under control, with the construction of dikes, and eventually of
dams, the Rhône at flood stage poured its waters into Lake le Bourget,
a little to the south and west of Annecy. The flood waters carried
silicate sediment which settled to the bottom of the lake. By measuring
the relative amount of these sediments along a 9 metre core extracted
from the lake bottom, F. Arnaud and his colleagues were able to reconstruct
periods of flooding over the last 7,200 years, with a resolution of
8 years.79
The portion of their graph relevant to this paper ­ the two
millennia of the Common Era ­ is reproduced in Figure 6. The dating
is in calibrated calendar years.
Increased silicate sedimentation in the lake could come either from
land clearing or from increased precipitation. Part of the sediment
documented here may thus be due to the clearing of the countryside
between Lake Geneva and le Bourget and in the Arve and Fier river
catchment areas during the Roman centuries. This may account at least
in part for the two small spikes on the graph dated to the first and third
78
See Leveau et al., `Crise environnementale', pp. 291­303. For Pisa, see Benvenuti et al., `LateHolocene
Catastrophic Floods'.
79
F. Arnaud, M. Revel et al., `7,200 Years of Rhône River Flooding Activity in Lake Le Bourget,
France: A High-Resolution Sediment Record of NW Alps Hydrology', The Holocene 15
(2005), pp. 420­8. For the hydrology of the lake, see E. Chapron et al., `Climatic Variability
in the Northwestern Alps, France, as Evidenced by 600 Years of Terrigenous Sedimentation
in Lake Le Bourget', The Holocene 12 (2002), pp. 177­85. For a description of the catchment
area see M. Revel-Rolland et al., `Sr and Nd isotopes as tracers of clastic sources in Lake Le
Bourget sediment (NW Alps, France) during the Little Ice Age: Palaeohydrology implications,
Chemical Geology 224 (2005), pp. 183­200.
160 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
centuries, though these also correspond to flooding of rivers in the
Appenines (perhaps, to be sure, caused by the same process of land
clearing). The large spike that begins between 450 and 500 CE, however,
can only be due to a major long-term increase in precipitation,
since it corresponds to a period of large-scale abandonment of agriculture
in the region and an increase in forest cover.80
It should not be imagined that this increase in precipitation concerned
only the north-west Alps. There is evidence from elsewhere, notably the
Italian peninsula, that increased rainfall was widespread. The spikes in
the first and third centuries CE correspond, as I have said, to an increase
in the recorded flood events of the Tiber river. So does the spike from
the fifth to eighth. There were also intense alluvial phases of rivers in
the northern Appenines in the sixth and seventh centuries.81
There is
equivalent evidence of colder and wetter weather in the same period in
Denmark and Germany, associated with a decline in agriculture.82
The
reason for this change in precipitation would have been a long-term
80
See W. Tinner et al., `Climatic Change and Contemporaneous Land-Use Phases North
and South of the Alps 2300 BC to 800 AD', Quaternary Science Reviews 22 (2003), pp. 1447­60.
81
C. Giraudi, `Late-Holocene Alluvial Events in the Central Apennines, Italy', The Holocene 15
(2005), pp. 768­73; and Benvenuti et al., `Late-Holocene Catastrophic Floods'.
82
K.E. Barber et al., `Late Holocene Climatic History of Northern Germany and Denmark:
Peat Macrofossil Investigations at Dosenmoor, Schleswig-Holstein, and Svanemose, Jutland',
Boreas 33 (2004), pp. 132­44.
Fig. 6 Flooding of the Rhône river. Source: Arnaud et al., `7,200 Years of Rhône
River Flooding'
The climatic anomaly of the early Middle Ages 161
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
shift in the North Atlantic Oscillation (NAO), the relative strengths of
the wintertime high pressure system near the Azores and the low pressure
system near Iceland. When both pressure systems are weak (a `negative'
NAO in the jargon of climatologists), the jet stream shifts south, bringing
wet winters to southern Europe and cold winds out of the Arctic across
northern Europe. There is a strong correlation between this record of
high flood water on the Rhône and some reconstructions of the NAO
since 1500, so we may hypothesize that the same was likewise true for
the early medieval change in the frequency of flood events.83
Was there also a cooling of the climate? The evidence is reasonably
good, to begin with because the increase in rainfall itself implies lower
temperatures. The most direct sign of a cooling climate, however, is the
advance of Alpine glaciers, as represented in Figure 7. In the sixth and
seventh century the Aletsch advanced almost as far as it did during the
coldest cycles of the Little Ice Age (the higher on the chart, the further down
the valley the glacier extended); the Gorner advanced only a little less.
The lower Grendelwald glacier shows a similar chronology.84
Studies
of the advance and retreat of the Rhône glacier and others in the Alps
in modern times have shown that the daily mean temperature for July
and August accounts for 58% of their variance, while precipitation
between October and June accounts for 21%.85
That is, advancing glaciers
are, as one might suspect, largely the consequence of wetter winters and
colder summers, but especially the latter. How much colder did summers
get? One German study has estimated a decline of 1 degree Celsius in
average summer temperature.86
That may not appear to be very much,
but it has been estimated that if the average fell that much during the
early stages of the Little Ice Age, the frequency of crop failure in northern
England would have increased from one year in twenty to one in
three.87
More relevant, perhaps, is the clear coincidence of the weather
83
Compare Figure 6 to the reconstructed phases of the NAO in J. Luterbacher et al., `Extending
North Atlantic Oscillation Reconstructions back to 1500', Atmospheric Science Letters (2002)
at <http://www.atmos.colostate.edu/ao/other_papers/ASL_NAO.pdf>. Also M.F. Glueck and
G.W. Stockton, `Reconstruction of the North Atlantic Oscillation, 1429­1983', International
Journal of Climatology 21 (2001), pp. 1453­65. For general information on the NAO consult
<http://www.ldeo.columbia.edu/NAO/>.
84
H. Holzhauser et al., `Glacier and Lake-Level Variations in West-Central Europe over the
Last 3500 Years', The Holocene 15 (2005), pp. 789­801.
85
J. Grove, `The Century Time-Scale', in T.S. Driver and G.P. Chapman (eds), Time-Scales
and Environmental Change (London, 1996), p. 41.
86
G. Patzelt, `Die klimatischen Verhaltnisse im südlichen Mitteleuropa zur Römerzeit', in H.
Bender and H. Wolff (eds), Ländliche Besiedlung und Landwirtschaft in den Rhein-DonauProvinzen
des Römischen Reiches (Espelkamp, 1994), pp. 7­20.
87
M.L. Parry and T.R. Carter, `The Effect of Climatic Variations on Agricultural Risk',
Climatic Change 7 (1987), pp. 95­110. M.L. Parry presents the theoretical relationship between
the principal climate factors and the growth and maturation of food crops in Climate Change
and World Agriculture (London, 1990), pp. 41­52. Although his focus is on global warming,
the analysis is suggestive for cooling as well. See also, Pfister, Bevölkerung, Klima, II, pp. 35­7.
162 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
anomaly of 1816­17, caused by the eruption of Tomboro in 1815, with
the great European subsistence crisis in those same years.88
One of the consequences of a wetter and colder climate is immediately
visible, however, in at least one aspect of the archeological record ­ the
abandonment of settlement and agriculture in river valleys that had
88
J.D. Post, The Last Great Subsistence Crisis in the Western World (Baltimore, 1977).
Fig. 7 Advances of two Alpine glaciers. Source: Holzhauer et al., `Glaciers and
Lake-level Variations', reproduced with permission from SAGE Publications
The climatic anomaly of the early Middle Ages 163
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
become waterlogged and a move to hill slopes or hilltops. The
phenomenon has often been noted both in Italy and in Mediterranean
Gaul.89
Water-logging was also probably the reason for the abandonment
of heavy clay soils, as it was for the spread of rye, a grain somewhat
more adaptable to colder and wetter soil.90
The comparison to the first wave of the Little Ice Age (c.1300­c.1460)
implied by the graphs of glacier extension suggests some reflections
more directly relevant to the subject of this article. As serious as plagues
and famines were in the fourteenth and early fifteenth centuries, and as
serious as was the resulting demographic decline, the basic structure of
the medieval rural world survived. Some villages were abandoned, but
most were not, and at least some of those abandoned were reoccupied
once the population began to recover. Field systems remained, even if
some fields reverted to fallow and to `waste'. In Britain it was the
enclosure movement of the early modern period that transformed the
open fields (and fossilized many of them under grass), not the troubles
of the fourteenth and fifteenth centuries. In some areas on the Continent,
as we have seen, tenth-, eleventh-, and twelfth-century fields could still
be photographed from the air in the mid-twentieth century. In contrast,
the climatic downturn that began around 500 CE wiped the slate clean
and what began to emerge two hundred years later was completely new.
Why was this so? I would suggest that the answer is connected
indirectly to the longer survival of a few exceptional antique landscapes
and economies along the Adriatic and Tyrrhenian coasts, the exceptions
to the general picture of impoverishment and depopulation elsewhere.
We saw earlier that one of the consequences of the bursting of the early
imperial `bubble' in commercial agriculture was apparently a disinvestment
in rural infrastructure, revealed archaeologically in the filling of drainage
ditches with alluvial soil. In some areas field structures may have been
recreated, at least in part, during the fourth-century revival, but they
too were allowed to fill and be covered over. Was the same true for
other necessary infrastructures of the agricultural economy? If so, a
worsening of the climate, especially increased precipitation, would have
been fatal, washing away the soil on slopes until grass and woodland
89
This has been studied most thoroughly in the middle Rhône valley: Berger and Jung, `Developing
a Methodological Approach'. It has been noted from the Ile de France to Italy, though
without specific reference to hydrological causes: L. Bourgeois, Territoires, réseaux et habitats:
l'occupation du sol dans l'ouest parisien du Ve au Xe sicle, 3 vols, thesis, Paris I ­ Sorbonne
(1995), I, pp. 75, 199. Frankovich and Hodges, Villa to Village, Ch. 3.
90
Abandonment of heavy clay soils: private communication from P. Van Ossel apropos of the
differences in late antique settlement survival between the Pays de France (between the Seine
and Oise valleys) and Marne-la-Vallée. See also F. Gentili and N. Mahé, `Serris: Les Ruelles.
Village Du Haut Moyen-Âge', ms. DRAC SRA, St Denis (1997). Spread of rye cultivation:
K.E. Behre, `The History of Rye Cultivation in Europe', Vegetation History and Archaeobotany
1 (1992), pp. 141­56. Hamerow, Early Medieval Settlements, pp. 135­7, 140.
164 Fredric L. Cheyette
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
began to fix it in place, water-logging low lying plains and bottom lands
(naturally among the richest soils). Near to cities where the rural economy
remained commercialized, whether through commercial exchange or
through the transfer of wealth from Constantinople to clients in the
western outposts,91
there would have been incentive to continue or to
begin again to invest in rural production, not least because the presence
of Byzantine troops and administrators in the cities provided a ready
market. Here a tenuous romanitas maintained itself. It is thus not
surprising that among the best survivals of centuriated landscapes
are those around Modena and Cesena, where, as we saw, survival or
reoccupation of older rural sites in the fifth and sixth century was
exceptionally high.
In most of the west that was not the case. Here political and military
shocks, then the coming of plague, only exaggerated the effects of failed
crops and famine in a world where even in the best of times human
fertility barely held its own against mortality.92
The archaeology of the rural
world reveals an infrastructure in decay or in ruins, drainage and irrigation
ditches clogged, terrace walls collapsing, roads no longer maintained.
This development in the countryside occurs at the same time that
portions of many urban centres begin to decay and are eventually
abandoned. We are surely looking at a circle, or in systems language,
a `feed-back loop'. Urban centres were fed by commercial grain, meat,
and wine production in their countryside, or, in the case of the greatest
urban centres, from around the Mediterranean. Fewer urban mouths
meant a smaller market, less investment, and, for independent peasants,
reason enough to shift towards subsistence agriculture, away from
labour-intensive grain production towards a pastoral economy requiring
fewer hands. And even the producers of meat eventually stopped selective
breeding and the production of good forage.
Without stimulus from outside, the commercial agricultural economy
would have spiralled downward. This is exactly what we see in the
gradual disappearance of the physical remains of that economy ­ the
commercial pottery ­ from the archeological record. With a sparser
population, infectious disease may not have spread as rapidly, but
commercial circuits would no longer have functioned to supply those
areas where crops had failed. It was in these circumstances that colder
and wetter weather would have had major consequences, `marginalizing'
soils that had once been prime producers of grain, inducing agriculturalists
91
On such transfers see Banaji, Agrarian Change in Late Antiquity.
92
Stathakopoulos, Famine and Pestilence. For an overview of ancient demography and a critique,
see W. Scheidel, Measuring Sex, Age and Death in the Roman Empire: Explorations in Ancient
Demography, Journal of Roman Archaeology, Supplementary Series 21 (Ann Arbor, 1996), and
W. Scheidel (ed.), Debating Roman Demography (Leiden and Boston, 2001).
The climatic anomaly of the early Middle Ages 165
Early Medieval Europe 2008 16 (2)
 2008 The Author. Journal Compilation  2008 Blackwell Publishing Ltd
to move to locations more suited to the new productive mix of animals,
products of woodland and marsh, and enough grain for their own
consumption. That this happened at different rates in different regions
should not be surprising, for the commercial circuits did not subside
everywhere at the same time or at the same rate. But eventually it
happened even in Byzantine Italy. The ancient landscapes vanished
and, in time, the new landscape, the medieval landscape emerged.
It is traditional to end an article of this sort by saying that only
further research will confirm or disprove this hypothetical narrative. In
this case, however, the data may already exist or soon come out of the
ground in the programmes of rescue archeology associated with the
construction of rapid train lines and superhighways. It is to be hoped
that the problems raised in this article will help shape the research
agenda of those programmes. What is needed above all is the rapid
dissemination of the data in a form that historians can use. Needed as
well is the rapid dissemination of data from the study of fossil pollen,
again in a form accessible to non-specialists and, where available, with
a temporal resolution useful for historians whose interest is in changes
over decades, or from century to century, and not periods of five hundred
or a thousand years. Much of this data probably already exists in desk
drawers, filing cabinets, and digital data sets. May we hope to see it
emerge to the light of day? So, too, with the data series produced by
climatologists, especially those who claim increasingly refined temporal
resolutions. It is to be hoped that they also will make their data accessible
with the claimed temporal resolution made visible. For further progress
in refining and correcting the story of this important and still obscure
period of European history will demand the communication and
cooperation of specialists in all these fields. Historians with their
traditional tools can no longer go it alone.
Amherst College