Roman to early medieval cereal farming in the Rhineland: weeds, tillage, and the spread of the mouldboard plough

ABSTRACT

A new model for gauging levels of soil disturbance (i.e. tillage) by analysing arable weed assemblages from archaeological contexts is applied to an extensive Roman-to-early medieval archaeobotanical sequence from the region west of Cologne. It tests the hypothesis that increasing use of the mouldboard plough, especially in a three-field system, would result in increased levels of soil disturbance which would be reflected in the kinds of weeds that grew in arable fields. The results point to clear differences in tillage regimes during the Roman period, providing support for the view that military sites were not provisioned by the same networks that supplied the civilian market. They also reveal generally low disturbance levels for the fifth and sixth centuries, indicating a continuing predominance of ard cultivation in the post-Roman period. The majority of seventh- to eighth-century samples had, however, been grown in ‘high disturbance’ conditions, a pattern that continued through the eighth and ninth centuries. Although use of the mouldboard plough within a fully developed three-field system may not have become widespread until the tenth or eleventh century, our evidence suggests that a plough capable of turning over the soil was in use in the Rhineland at a much earlier date.

KEY WORDS:

• Roman
• early medieval
• farming
• mouldboard plough
• Rhineland
• weed ecology

INTRODUCTION

The present paper applies a recently developed model for gauging levels of soil disturbance (i.e. tillage) by analysing arable weed assemblages from archaeological contexts to a Roman-to-early medieval archaeobotanical sequence from the region west of Cologne. Not only has this region produced an exceptional archaeobotanical dataset that enables long-term agroecological change to be traced, but it also saw the early adoption of the mouldboard plough and three-course rotation, traditionally dated on the basis of written sources — above all, Carolingian polyptychs — to the later eighth and ninth centuries, although fully developed open field farming involving communal management of arable is not thought to have emerged before the tenth century at the earliest (Devroey 2003; Hildebrandt 1988).² Our working hypothesis was that increasing use of the mouldboard plough in the early medieval period, especially in a three- field system that involved fallow ploughing, would result in increased levels of soil disturbance which would be reflected in the kinds of weeds that grew in the arable fields. Tracing the spread of the mouldboard plough is important because, unlike the ard, which scratches a furrow into the surface of the field, the mouldboard plough turns the soil over, improving tillage and enabling farmers to cultivate heavier, more productive soils. The kinds of arable weeds that thrive in highly disturbed, well-tilled conditions, such as those produced by a mouldboard plough, differ from those that prefer less disturbed conditions, such as those produced by ard cultivation. Perennial weeds that regenerate easily from tiny root fragments (vegetative propagation) thrive in heavily disturbed conditions, for example. A second trait, flowering duration, predicts the ability of (annual or perennial) species to recover from high levels of disturbance. Other traits such as canopy dimensions and ratios of leaf area to leaf thickness and to leaf dry weight are indicators of species’ potential in relation to soil fertility, and so reflect other agronomic practices such as manuring (see Bogaard et al. in McKerracher & Hamerow 2022, for a more detailed explanation of this approach). Weed seeds that were harvested and preserved (by accidental charring) together with cereal grains therefore provide an invaluable tool for reconstructing the soil conditions in which crops grew; interpretation of such evidence using weed species’ functional traits is an approach known as functional weed ecology.

THE HIGHGROVE-LAXTON MODEL

In order to reconstruct levels of medieval soil disturbance accurately, we needed to compare the weed flora from medieval samples to a baseline provided by modern, organically farmed analogues. These came from Home Farm, Highgrove, in Gloucestershire, where ‘heritage’ cereals are grown organically and crop rotation is practiced, and from the village of Laxton in Nottinghamshire, where crop rotation is practiced within a still-functioning open field system, but cereals are no longer grown organically. Quadrat surveys of weed flora were undertaken at Highgrove, along with botanical surveys of both the Laxton ‘sykes’ — unsprayed strips of meadow at the interstices of the open fields — and the arable fields before they were sprayed with herbicide. Comparing the functional traits of weeds that grew in the undisturbed conditions of the sykes with those that grew in arable fields allowed us to assess the impact of different levels of mechanical disturbance on the weed flora. Using discriminant analysis, it was possible to distinguish between the two sets of samples — those from the sykes and those from the fields — with a high degree of accuracy on the basis of two functional traits reflecting tolerance of disturbance: flowering duration and vegetative propagation (Bogaard et al. 2022).

ANALYSIS AND INTERPRETATION

Using this model, it was possible to interpret the weed seed assemblages from over twenty medieval sites in England. Our analysis showed increasingly ‘disturbed’ weed profiles through time (between c. 500 and 1300) in the region that came to be dominated by open fields, the so-called Central Zone (Hamerow 2022). This increase suggests more frequent tillage, increasing use of the mouldboard plough, or both. Remarkably, the levels of tillage indicated by the medieval samples were comparable to those achieved by modern, mechanised systems such as those used today at Highgrove and Laxton.

We were able to refine this picture further with weed data provided by colleagues at the Lauresham Laboratory for Experimental Archaeology in southern Germany, adjacent to the World Heritage Site of Lorsch Abbey. At Lauresham, experimental fields have been cultivated for several years according to a three-field rotation system using a set of reconstructed mouldboard ploughs pulled by cattle (Pl. I) (Kropp 2022). A detailed survey of the kinds of weeds growing in these fields provided us with the floristic data needed to compare these experimental fields with the Highgrove-Laxton model. The Lauresham fields all resemble the arable spectrum in the Highgrove-Laxton model, as predicted, and provide an indication of the minimum disturbance levels we should expect if a mouldboard plough was being used within a three-field system.

Plate I. Experimental ploughing of the fallow field at the Lauresham Open-Air Laboratory, using a reconstructed mouldboard plough (Photograph: Staatliche Schlösser und Gärten Hessen).

Due to a lack of readily available samples from England, it was not possible to compare the medieval disturbance levels to those seen in Roman Britain, where, it is assumed, the ard was the primary tillage implement (Lodwick & Brindle 2017, p. 42). In collaboration with the University of Cologne, we therefore applied the Highgrove-Laxton model to a Roman-to-early medieval archaeobotanical sequence from the region west of Cologne, where an exceptionally extensive archaeobotanical dataset enables agroecological change to be traced over time (Fig. 1; Table 1), (Hamerow et al. 2022). This part of the Rhineland lay, furthermore, in the hinterland of the Roman limes and contained not only large numbers of troops needing to be fed, but also two sizeable urban populations at the coloniae of Cologne and, further north, Xanten (Brüggler et al. 2017; Reddé 2018).

Fig. 1. Topographic map of the Rhineland with location of excavations that yielded archaeobotanical samples. Roman sites in red, medieval sites in blue. Roman sites are differentiated by type: diamond = colonia, rectangle = castrum/vicus, dot = villa/rural. Colour gradation of the topographic map from green to brown according to altitude: 0, 85, 170 and 255 m above sea level (Map: R. Lubberich, LVR-Amt für Bodendenkmalpflege im Rheinland).

TABLE 1: ROMAN AND MEDIEVAL SAMPLES INCLUDED IN DISTURBANCE WEED ANALYSIS

The Roman samples analysed (n= 78) came from sixteen sites (Fig. 1): eight villae or farms, grain, probably representing the harvests of several productive units.

Figure 2 shows the results for Roman archaeo-botanical assemblages, classified by site type. The castra/vici and villae sites all reflect a wide range of disturbance levels, including low disturbance levels consistent with ard ploughing. Low disturbance levels are especially dominant (around 75 per cent) in the assemblages from castra/vici. The coloniae, by contrast, reflect higher disturbance levels, mostly exceeding the Lauresham baseline and hence conditions consistent with mouldboard ploughing. These results have two key implications. First, the extensive, low-input cereal husbandry regimes geared towards surplus production that provisioned the castra/vici documented by a previous study of weed flora (primarily based on weed traits reflecting soil fertility) were achieved largely by ard cultivation (Hamerow et al. 2022). Second, the cereal husbandry regimes that provisioned the coloniae (Xanten and Cologne) — which were shown by the previous study to be highly variable, ranging from very ‘high’ to very ‘low’ input — benefited from distinctly higher disturbance levels, and potentially used some form of heavy, perhaps wheeled, plough technology capable of turning over two coloniae (Xanten and Cologne) and six castra or related vici. Most of the archaeobotanical samples represent the fills of pits or postholes and include some grain-rich deposits likely to represent redeposited storage material. One set of samples, from Krefeld Gellep, derived from a horreum. The medieval samples (n=75) came from seven settlements: one urban site: Duisburg, and four rural sites: Bornheim-Walberberg, Inden-Pier, Erkelenz-Tenholter Strasse. and Bonn-Bechlinghoven (Table 1). Many of the charred plant remains derive from Grubenhäuser, although some came from pits. None of the medieval samples analysed appears to represent stored cereals charred in situ, although sample composition suggests that some if not most represent redeposited stored the soil (ibid.). The heavy wheeled plough was used in parts of the Roman West, and a recent survey has concluded that ploughs with the capacity for turning over the soil existed in north-east Gaul (Henning 2004; Marbach 2001). Such ploughs seem to have been used outside the Empire as well as within it: in Denmark, well-preserved traces of ‘turned furrows’, evidently produced by ploughs with mouldboards, have been radiocarbon dated to between the first century b.c. and the late Roman Iron Age (Larsen 2013).

Fig. 2. Discriminant function scores for samples from Roman sites. (a) The relationship of Laxton sykes (white squares) versus Laxton and Highgrove arable fields (other symbols) to the discriminant function; (b–d) the relationship of archaeobotanical samples from Roman sites to the discriminant function: (b) castra/vici; (c) villae; (d) coloniae (larger symbols indicate centroids for the modern groups). The dotted line represents the ‘Lorsch baseline’, the minimum discriminant score to be expected under effective mouldboard ploughing. Samples to the right of this line are consistent with the use of a mouldboard plough.

Figure 3 shows how the early medieval Rhineland sequence develops from the fifth–sixth to the seventh–eighth centuries. Whereas samples of the fifth–sixth centuries reflect lower disturbance levels consistent with ard ploughing and with Roman villae, castra and vici, those of the seventh–eighth centuries, though variable, predominantly fall above the Lauresham baseline, indicating conditions consistent with mouldboard ploughing. This shift in disturbance levels between the fifth–sixth and the seventh–eighth centuries may therefore reflect a change in tillage practice, from predominant use of the ard to increasing use of the mouldboard plough and, potentially, more frequent tillage. Moreover, this shift is apparent for both of the two main early medieval archaeobotanical assemblages available, from the settlements of Bonn-Bechlinghoven and Bornheim. It is worth noting that at Bonn-Bechlinghoven, this shift in disturbance levels coincides with an isotopic change in barley grain δ¹⁵N values consistent with the introduction of systematic crop rotation in the seventh–eighth centuries (Hamerow et al. 2022).

Fig. 3. Discriminant function scores for samples dating to the fifth–eighth centuries. (a) fifth–sixth centuries; (b) sixth–seventh centuries; (c) seventh–ninth centuries.

The assemblages from the sites of Bonn-Bech-linghoven and Bornheim include few eligible samples from later centuries (Fig. 4). Samples dating to the eighth to ninth centuries derive from a range of sites but predominantly fall at or above the Lauresham baseline (Fig. 4). Those of the tenth century and later are similarly mixed and fall more or less equally on either side of the baseline, although whether this is due primarily to a persistence of arding or to an expansion of mouldboard-ploughed fields onto former pasture and the consequent persistence of a ‘pasture signal’ in the weed flora is impossible to say. It is notable that the few samples from the fifteenth-century town of Duisburg reflect particularly elevated disturbance levels (Fig. 4).

Fig. 4. Discriminant function scores for samples dating to the eighth– tenth centuries and later. (a) eighth– ninth centuries; (b) tenth century and later.

CONCLUSIONS

When considered together with the results of our earlier investigation of soil fertility in this part of the Rhineland (Hamerow et al. 2022), the results of the present study lead us to make two key observations: First, a clear contrast is apparent between the provisioning of Roman castra and vici on the one hand, and of coloniae on the other. The former were provisioned largely with cereals grown in extensive, low-fertility conditions geared towards surplus production using predominantly ard cultivation, broadly comparable to the conditions seen at villae, as might be expected. The majority of cereals consumed in the coloniae, on the other hand, grew in distinctively ‘high disturbance’ conditions comparable to those achieved using a mouldboard plough. This evidence casts serious doubt on the assumption that they were provisioned with cereals grown on villae where, as shown above, conditions were quite different. The samples from the coloniae also reflect greater variability in terms of soil fertility, including both extremely high-input (i.e. relatively small-scale, intensively manured) and low-input conditions. As with the evidence for contrasting levels of ‘intensity’ (Hamerow et al. 2022), these results support the view that military sites were for the most part not supplied by the same networks that provisioned the civilian market, the former presumably being in receipt mostly of grain extracted as tax in kind.

Second, the generally low disturbance levels seen for the fifth and sixth centuries are comparable to those seen at Roman villae, castra and vici, indicating — as expected — a continuing predominance of ard cultivation. The majority (around 70 per cent) of seventh- to eighth-century samples, however, had been grown in ‘high disturbance’ conditions, a pattern that continued through the eighth and ninth centuries. A marked change in cultivation practices around the seventh to eighth centuries is further indicated by evidence produced by our previous study for the introduction of systematic crop rotation around this time (Hamerow et al. 2022). Although use of the mouldboard plough within a fully developed three-field system may not have become widespread until the tenth or eleventh century (Verhulst 1990, pp. 22–3; Devroey 2003, p. 110), our evidence suggests that a plough capable of turning over the soil was in use in the Rhineland at a much earlier date (cf. Henning 2004). More frequent ploughing, and potentially fallow ploughing, could have further contributed to this shift to ‘high disturbance’ conditions.

Notes

1. This work was supported by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme under grant agreement No 741751.

2. The model was developed by one of the authors [AB] as part of a larger project to generate direct evidence for cultivation conditions in England during the so-called medieval ‘Agricultural Revolution’. The project ‘Feeding Anglo-Saxon England: the bioarchaeology of an “Agricultural Revolution”’ (‘FeedSax’; PI: Hamerow) received funding from the European Research Council.