Journal of Archaeology in the Low Countries 4-1 (October 2012)

Celtic field banks and Early Medieval rye cultivation

Willy Groenman-van Waateringe

Keywords: Early Medieval times, cultivation, pollen analysis.


The discrepancy between 14C and OSL dates of Celtic field banks and the high values for Secale in these banks in Denmark, Germany and the Netherlands calls for an explanation. This article shows that downward movement of pollen must have caused this discrepancy. It also means that it will not be possible to get an impression of the cultivation of the Celtic fields in the Iron Age-Early Roman period by pollen analysis of these soils, because the pollen evidence from this period is overshadowed by the Early Medieval data. Only the old surface layers underneath the banks, sometimes with traces of early/pre-Celtic field cultivation, may provide some information concerning that period. However, in some cases this picture is also disturbed by the penetration of younger pollen.

The Early Medieval pollen data from the banks point to summer rye cultivation taking place on the banks, in accordance with high phosphate values, particularly in the top layers of the banks.

1 Introduction

Earlier descriptions of Celtic fields in the Netherlands range from Heydensche legerplaatsen en wallen (pagan settlements and walls; 17th c.), Roman camps (18th c.) to camps in general (19th c.). The interpretation as arable land was given by Van Giffen (1928) following English publications introducing the name Celtic fields. As early as 1911 these landscape structures had been described for Denmark by Sophus Müller and interpreted by him as arable fields. Hatt (1931) called the system Oldtidsagre (old arable fields) and gave an explanation for the formation of the banks by the accumulation of debris from the fields. Van Giffen (1944 a.o.) thought that the banks were gradually formed by the removal of the exhausted and thus infertile topsoil. Modderman (1955) strongly opposed this with the argument that it would mean removing the best part of the soil.[1] So did Jankuhn (1956-57): after taking away the humic upper layer and so exposing the infertile sand underneath, arable cultivation would have been made impossible.

According to Brongers (1976, Ch. V) the banks were formed by clearing the fields of stones and tree stumps. He also brought forward stratigraphical arguments for the transport of material from the fields on to the banks, and a mathematical model showing that in a later stage humic material must have been brought in from outside, to improve soil fertility.

Spek et al. (2003) concluded that the banks were deliberately raised at the field boundaries during the last stage of cultivation (Middle Iron Age, Early Roman period), by bringing in material both from inside the banks (the fields) as well as from outside. In the last stage (Late Iron Age to Roman period) cultivation would have taken place on the banks.

It seems that there was ample space for such practice: the width of the earthen banks measures between 3-16 m: Examples are: Øster Lem Hede, Denmark, 3-6 m (Helt Nielsen [2009]), Noordse Veld, Zeijen, 6-12 m, and Vaassen, c. 8 m in the Netherlands (Spek et al. 2003), Brongers 1976) and Flögeln Haselhörn, Germany, up to 16 m (Zimmermann 1976).

2 Celtic fields and rye cultivation

The oldest finds of rye in the Netherlands, NW Germany and Denmark date from the Late Iron Age/Roman period, and probably represent a weed grown in fields with other cereals. The start of rye cultivation in NW Europe has to be placed later, in the first centuries AD, with a rapid expansion in the Migration/Early Medieval period, and becoming most intensive in the 2nd millennium AD (Behre 1992, 2000; Behre & Kučan 1994; Mikkelsen 2003).

In the grain finds of the middle of the first century, rye is equal in numbers to or even exceeds other cereals (Behre & Kučan 1994, 39, tab. 6). It is assumed to have been grown as a summer crop. Summer rye should produce less pollen than winter rye (see also Behre & Kučan 1994, 36-41, 57-62, 69-70, 85-87).

Around AD 1000 there is a strong increase of rye in the pollen diagrams in NW Germany, together with an explosion of Rumex acetosella (sheep’s sorrel), one of the most important weeds of rye fields. Typical weeds of winter rye fields, such as Agrostemma githago (corn cockle) and Centaurea cyanus (cornflower) follow in the course of time. Mikkelsen (2003, 152-154) summarizes the introduction of rye in Denmark, already starting in the pre-Roman Iron Age (2nd and 1st c. BC), again probably as weeds in cornfields. From the middle of the first millennium onwards there are numerous finds of carbonized rye grains, here supposed to be of winter crops, because of a much earlier rise in Rumex acetosella in Danish pollen diagrams. However, this species is furthered by, but not dependent on, rye and may also occur in abundance in sandy grassy areas. Moreover, the similarity between weeds in summer and winter corn fields in the same habitat is great, the difference lying mainly in the cover percentages (Haveman et al. 1998, 206). Pollen data from Western Jutland, show low values for Cerealia around the beginning of our era, the start of rye cultivation in Jutland around AD 300, gradually increasing in the second half of the first millennium (Odgaard 1994, 155-157; 2010).

Rye pollen data from archaeological sites in the central part of the Netherlands show a slight increase from the Roman period to the 12th century and from then on a steep increase (fig. 1) (Groenman-van Waateringe 2010, fig. 3.9). The latter is in accordance with Behre’s ideas on the link between winter rye cultivation and the development of plaggen soils.


Figure 1 Increase in cereal pollen over time in archaeological sites on the Veluwe, central Netherlands.

3 Palynological data from Celtic fields

Stratigraphical data, ceramic finds, 14C and OSL dates place the origin of the earthen banks somewhere in the Iron Age (Zimmermann 1976; Behre 2000; Spek et al. 2003; Helt Nielsen [2009]).

There is ample evidence underneath the excavated Celtic fields in NW Europe for earlier pre-Celtic field arable activity Brongers 1976; Zimmermann 1976; Spek et al. 2003; Eriksen 2006). This can be in the form of ard traces underneath the walls or pollen data. However, in the pre-Celtic field period the amount of cereal pollen is extremely low, since the cereals grown were self-pollinating, producing low amounts of pollen which would not have been released in the air, and were only to be found in quantity where threshing took place.

This changes drastically in pollen samples taken from within the banks. The total amount of cereal pollen – part of which is clearly identified as Secale (rye) – has increased considerably, sometimes up to more than 20%, in relation to the numbers in the pre-Celtic field layers. In Vaassen (Casparie 1976) and Flögeln-Haselhörn the increase in cereal pollen can also be observed in layers interpreted as pre-Celtic field layers. How can this be explained, since the formation of the banks took place hundreds of years before rye was cultivated in NW Europe?

To solve this problem four sample series from Celtic field banks will be discussed in detail.

1 Øster Lem Hede (Eriksen 2003; Helt Nielsen [2009]; Helt Nielsen & Groenman-van Waateringe, 2012). Analysed by the author (fig. 2).


Figure 2 Øster Lem Hede. Cereal pollen from four banks and a lynchet.

In 2001 samples for pollen analysis were taken from top to bottom in consecutive layers with distances of 2-5 cm from two banks and two fields in one section. In 2007 the same was done from two banks and one lynchet (2007/2) and from three fields in three different sections. Cereal pollen from underneath the banks was sparse. The banks themselves have high percentages (up to 20%) of cereal pollen, as far as identifiable, nearly all Secale. Values for Poaceae (grasses), Asteraceae liguliflorae (Composite family) and Spergula/Spergularia (spurry species) and for herbs in general are high.

In the samples from the fields and from the lynchet the amount of cereal grains is lower than in the samples from the banks in the strict sense.

2 Noordse Veld Zeijen (Spek et al. 2003). Analysed by the author (fig. 3).[2]


Figure 3 Noordse Veld, Zeijen. Cereal pollen from a bank.

Samples were taken as in Øster Lem Hede from a bank and field. Again the bank is characterized by rather high cereal percentages, partly identified as Secale (up to 10%, but total cereal percentages up to c. 20%). Values for Poaceae, Asteraceae. Spergula arvensis type, Chenopodiaceae (Goose-foot family), Polygonum persicaria type (lesser persicaria), Rumex-a type (sorrel) and Succisa pratensis (devil’s bit scabious) are high. The topmost samples of the field show the same high herb percentages. As stated “The combination of pollen types together with high values for Ericales strongly recalls the pollen assemblages of medieval plaggen soils” (Spek et al. 2003, 165). Also here are higher cereal values in the bank than in the field samples.

3 Flögeln-Haselhörn (Zimmermann 1976; Behre & Kučan 1994; Behre 2000) (fig. 4).


Figure 4 Flögeln-Haselhörn. Cereal pollen from four sections through banks.

Behre & Kučan (1994) published four sections through Celtic field walls. In all four diagrams the percentages for Cerealia are high (with percentages of c. 28 and 34 in two of the samples, stratigraphically assigned to the pre-Celtic field phase.[3] Not much can be said about the herbs, because the published diagrams are a selection of curves. The highest Cerealia percentages go together with rather high values for Succisa.

4 Vaassen (Casparie in Brongers 1976) (fig. 5).


Figure 5 Vaassen, Cereal pollen in four samples from banks.

Pollen analysis in Vaassen comprises eight samples taken from various layers, spread over different parts of the excavation. Six samples belong to the pre-Celtic field phase. Cerealia were sparse. Only two samples (4 and 5) above each other in one of the banks – sample 4 from the old arable layer beneath the bank and 5 from within the bank – show slightly higher values for Cerealia, c. 5%. Casparie writes about these two samples: “From the very high percentages of Cerealia in spectra 4 and 5 it may be inferred that cereals, probably wheat, were grown here”. Spectra 7 (pre-Celtic field) and 8 (bank) have, according to Casparie, relatively many agriculture indicators, such as Plantago lanceolata (ribwort plantain), Rumex and Artemisisa (mugwort). Percentages for Poaceae, Polygonum persicaria and Asteraceae are specifically in sample 8 higher than in 7. He also mentions the presence of the hepatics Anthoceros laevis and A. punctatus (hornworts) (samples 4-5, 7-8).

4 Cultivation on the banks of the Celtic fields

Brongers (1976, 71) was one of the first to point to the use of the banks for cultivation, considering “the rather high percentage (40%) of land otherwise not used”. After a period of lying fallow the soil in the banks should have recovered and could be used for agriculture.

The data of Gebhardt (1976) showing the highest phosphate values occurring in the banks and not in the fields, made Zimmermann (1976) argue for the possibility that the banks, in stead of the spaces in between, had been used for agriculture. The same holds true for the Celtic field of Zeijen (Spek et al. 2003). The analyses of Øster Lem Hede excavations in 2001 and 2007 have not yet been published. However, there is a list of analyses carried out by K. Dalsgaard (oral communication), concerning the 2001 excavation, pointing in the same direction.

Micromorphological research in the Celtic field of Zeijen has shown that regular soil tillage took place at the ridges, with an intensification in the top layers, which also show the highest phosphate values.

But when did this cultivation of the banks begin? Zimmermann (1976, 88) writes that the use of the banks for arable, wide and humic through long-lasting traditional cultivation could only have occurred in the last phase of the Celtic field. They were then probably used for cultivation of specific plants and prepared by necessary manuring, while the fields in between were used for extensive agriculture.

However, Behre (2000, 140) and followed by Spek et al. (2003) suggest another scenario. According to Spek et al. the start is with fields without banks, but probably some other kind of boundary in the Late Bronze Age and Early to Middle Iron Age. Zimmermann has proposed to call these early Celtic fields instead of pre-Celtic fields (Behre 2000). The development of the banks would have occurred in the last phase (Late Iron Age, Early Roman period), and it was mainly in these last phases that the banks were intensively cultivated. Behre (2000) is less specific concerning the dating of this cycle, he writes only of a later phase for the formation of the banks and for cultivation on them in a younger phase.

5 Discussion

In the following I will show that it is not possible through pollen analysis to get an idea of the cultivation during the existence of the Celtic fields in general or to determine the start of the cultivation on the banks. The pollen so far found in the banks has to be dated to the Migration/Early Medieval period and it overshadows all traces of cultivation from earlier periods.

The discrepancy between the archaeological dating of the banks somewhere in the middle of the Iron Age and the high amount of Secale pollen can be explained either by tillage activities, mixing older and younger material, or by intrusion of younger pollen in older layers as result of percolation and bioturbation. Tillage activities cannot be excluded, but are dependent on the type of activity and its depth. As soon as the banks have reached a certain height and the spade or plough (ard?) no longer reaches into the lowermost layers the situation in these layers will be fossilized. This process must have taken place in the banks, otherwise the 14C dates and the OSL dates would have shown a rather mixed picture. This is not the case (Helt Nielsen 2009). The rate of downward movement of pollen is dependent on the type of soil and the soil fauna, reaching less deep in more acid soils (Stockmarr 1975, Groenman-van Waateringe 2012). After the formation of a humus podzol and an iron pan the downward movement stopped and the situation became frozen.

A dating of the rather high Secale values in the Early Medieval period, higher compared to the Roman period, but not as high as in the later plaggen soils, corresponds well with Behre’s idea of summer rye cultivation in the 1st millennium AD and is clearly shown in the pollen diagrams IV and VI from Flögeln (Behre and Kučan 1994). Considering the distance between the Celtic field of Flögeln-Haselhörn and these two pollen diagrams, c. 800-1000 m as the crow flies, the increase of cereal pollen from the Roman to the Migration period matches well with the Early Medieval data from the walls of the Celtic fields (cf. Behre & Kučan 1986, Table 1).

Zimmerman’s (1976) and Behre’s (2000) questions as to where to find the fields of the Late Roman/ Migration period (the settlement of Flögeln-Eekhöltjen (1st-5th c. AD) may now have received an answer. Mikkelsen (2003, 126) asks for more research concerning the supposed abandonment of the Celtic field system in the early Roman period, because in his view this is still an unanswered question.


Willy Groenman-van Waateringe


Review data:

Submission: 25/2/2011

Revised submission: 12/3/2011


1. See for more details and references Brongers 1976, 18-29.

2. In figs 2,3,4 Secale pollen and Cerealia in general have been taken together. The rather poor preservation of the pollen hampers the certain identification specifically of cereal pollen, because most of them are heavily crumpled. High values in arable fields point to the wind-pollinated Secale.

3. The higher values for Pinus (pine) and the presence of Centaurea cyanus and Fagopyrum (buckwheat) in some of the top samples point to pollen from Late Medieval and even younger periods.


The Foundation for Anthropology and Prehistory in the Netherlands made the journey to and stay in Denmark possible with a grant for the Dutch partners in the excavation at Øster Lem Hede 2001. Thanks are due to KNNV Publishing Zeist, the Netherlands for permission to reproduce fig. 1. I also want to thank Bent Vad Odgaard for sending me his thesis and some articles concerning vegetation development in Jutland. Christine Jefferis is thanked for improving my English, and my colleagues Dr. Jan Albert Bakker and Professor Leendert Louwe Kooijmans for general improvements of the text.


Behre, K.-E. 1992, The history of rye cultivation in Europe, Vegetation History and Archaeobotany 1, 141-156.

Behre, K.-E. 2000, Frühe Ackersysteme, Düngermethoden und die Entstehung der nordwestdeutschen Heiden, Archäologisches Korrespondenzblatt 30, 135-151.

Behre, K.-E. & D. Kučan 1986, Die Reflektion bekannter Siedlungen in Pollendiagrammen verschiedener Entfernung – Beispiele aus der Siedlungskammer Flögeln, Nordwestdeutschland. In: K.-E. Behre (ed.), Anthropogenic indicators in pollen diagrams, Rotterdam (A.A. Balkema), 95-114.

Behre, K.-E. & D. Kučan 1994, Die Geschichte der Kulturlandschaft und des Ackerbaus in der Siedlungskammer Flögeln, Niedersachsen, seit der Jungsteinzeit, Oldenburg (Probleme der Küstenforschung im südlichen Nordseegebiet 21).

Brongers, J.A. 1976, Air photography and Celtic Field research in the Netherlands, Amersfoort (Nederlandse Oudheden 6).

Casparie, W.A. 1976, Palynological investigation of the Celtic Field near Vaassen, the Netherlands. In: J.A. Brongers, Air photography and Celtic Field research in the Netherlands. Amersfoort (Nederlandse Oudheden 6), 105-113.

Eriksen, P. 2006, RIM 7806 - Øster Lem Hede. Undersøgelse af ager med digevoldinger. Sønder Lem sogn, Hind herred, Ringkøbing amt. Stednummer 180412-163, Ringkøbing (Internal report Ringkøbing Museum).

Gebhardt, H. 1976, Bodenkundliche Untersuchung der eisenzeitlichen Ackerfluren von Flögeln-Haselhörn, Kr. Wesermünde, Probleme der Küstenforschung im südlichen Nordseegebiet 11, 91-100.

Groenman-van Waateringe, W., 1986, Grazing possibilities in the Netherlands based on palynological data. In: K.-E. Behre (ed.), Anthropogenic indicators in pollen diagrams. Rotterdam (A.A. Balkema), 187-202.

Groenman-van Waateringe, W. 2010, Man and vegetation on the Veluwe in six time slices. In: J. Fanta & H. Siepel (eds.), Inland drift sand landscap, Zeist (KNNV publishing), 49-64.

Groenman-van Waateringe, W. 2012, Thousands years of vegetation history revealed by pollen in a sandy soil, central Netherlands, Environmental Archaeology 17, 2012.

Hatt, G. 1931, Prehistoric fields in Jylland, Acta Archaeologica 2, 117-158.

Haveman, R., J.H.J. Schaminée & E.J. Weeda 1998, 30 Stellarieta mediae (klasse der akkergemeenschappen). In: J.H.J. Schaminée, E.J. Weeda & V. Westhoff (eds.), De vegetatie van Nederland 4, Leiden (Opulus Press), 199-246.

Helt Nielsen, N. [2009], Udgravningsberetning. RIM 9140-Øster Lem Hede. Sønder Lem sogn, Hind herred, Ringkøbing amt. Sted-nr: 180412, Sb-nr. 163. Ringkøbing (Internal report Ringkøbing Museum).

Jankuhn, H. 1956-57, Ackerfluren der Eisenzeit und ihre Bedeutung für die frühe Wirtschaftsgeschichte. Berichte der Römisch-Germanischen Kommission 37-38, 148-214.

Mikkelsen, P.H. 2003, Agerbruget. In: P.H. Mikkelsen & L. C. Nørbach, Drengsted. Bebyggelse, jernalderproduktion og agerbruget i yngre romersk og ældre germansk jernalder, Moesgård (Jysk Arkæologisk Selskab), 115-225.

Modderman, P.J.R. 1955, Oudheidkundige verschijnselen. In: W.J. van Liere & G.G.L. Steur, Een bodemkartering van de gemeente Epe en een bodemkundige verkenning van een deel van de gemeente Heerde, Wageningen (De bodemkartering van Nederland 16), 58-63.

Odgaard, B.V. 1994, The Holocene vegetation history of northern West Jutland, Denmark, Copenhagen (Opera Botanica 123).

Odgaard, B.V. 2010, Quantitative landscape dynamics in Denmark through the last three millennia based on the Landscape Reconstruction Algorithm approach, Vegetation History and Archaeobotany 19, 375-387.

Spek, Th., W. Groenman-van Waateringe, M. Kooistra & L. Bakker 2003, Formation and land-use history of Celtic fields in north-west Europe – an interdisciplinary case study at Zeijen, the Netherlands, Journal of European Archaeology 6, 141-173.

Stockmarr, J. 1975, Retrogressive forest development, as reflected in a mor pollen diagram from Mantingerbos, Drenthe, the Netherlands. Palaeohistoria 17, 37-48.

Van Giffen, A.E. 1928, Prehistoric fields in Holland, Antiquity 2, 85-87.

Van Giffen, A.E. 1944, De tumuligroep bij de zgn. heidensche legerplaats op het Emmerveld, gem. Emmen, Nieuwe Drentsche Volksalmanak 62, 134-141.

Zimmermann, W.H. 1976, Die eisenzeitlichen Ackerfluren – Typ ‘Celtic field’ – von Flögeln-Haselhörn, Kr. Wesermünde, Probleme der Küstenforschung im südlichen Nordseegebiet 11, 79-90.