A weedy window on medieval wheat farming


We have exceedingly little historical evidence for crop husbandry practices in Anglo-Saxon and medieval England, prior to the 13th century. But we do have an increasingly rich archaeobotanical record, including not only the charred grains and chaff of harvested crops, but also the seeds of many and various arable weeds. Since different weed species thrive in different conditions, their archaeological remains offer us a unique window onto arable environments – and, consequently, crop husbandry strategies – in the past. This piece illustrates the potential of archaeobotanical ‘weed ecology’ through an investigation of sowing seasons.

In an earlier article for the Year on the Field blog, I introduced the archaeobotanical evidence for wheat cultivation in Anglo-Saxon and medieval England, spanning the 5th to 13th centuries. Most of this period is very poorly documented in the historical record, but we’re gaining ever more insights from a particular kind of archaeological evidence: charred plant remains. Last time, I focused on the charred grains of free-threshing wheat, hulled barley, oats and rye – the chief crops of England in this period – and found that, while wheat did not rise to its modern pre-eminence in the medieval period, it seems to have been preferentially cultivated in the rich clay vales of central England.

But archaeobotany can tell us much more than simply which crops were cultivated, where, when, and in what proportions. For the thousands of charred crop deposits sampled at excavations across England include not only cereal grains and chaff, but also the seeds of arable weeds which were accidentally harvested along with the crop. While not every wild species in the archaeobotanical record was necessarily a weed – some are aquatic or woody, for instance – a great many are very plausible as arable weeds, and some are known as such in later documents, and indeed through to the present day: for example, stinking mayweed (Anthemis cotula) and corncockle (Agrostemma githago).

A nuisance to Anglo-Saxon and medieval farmers, they are invaluable to researchers as a sensitive proxy for growing conditions in the past, because we can observe their ecological preferences today, and so make inferences about the ecology of medieval fields. Such investigation entails reference to modern floras and botanical studies, but also original fieldwork, surveying the weed populations of traditional and organic farming regimes (Figure 1).

Botanical survey of field margins at Laxton, Nottinghamshire, England, by the ‘Feeding Anglo-Saxon England’ team from the University of Oxford (photograph by M. McKerracher)

This ‘weed ecology’ approach can be relatively simple, using the environmental preferences of individual species to make inferences about the past. For instance, where seeds of spike rush (Eleocharis palustris) contaminate a deposit of wheat, we could infer that the crop was grown in damp soils – since spike rush is a moisture-loving plant. But to gain a more sensitive and precise measure of past arable environments, we can go a step further and apply functional weed ecology, an approach first pioneered at the University of Sheffield.

The principle is to work out, not simply which weed species tend to occur under particular cultivation conditions, but which kinds of weed tend to occur there: that is to say, weeds with what ‘functional traits’, such as leaf area, or ability to regenerate from root fragments. The big advantage of this approach is that we don’t need past arable fields to have been infested with the exact same weed species that we can observe today. Instead, we can extract these modern functional trait data and compare them with those represented in archaeobotanical assemblages, using a suite of statistical methods.

So, what kinds of insights can functional weed ecology offer us, with regard to Anglo-Saxon and medieval crop husbandry? Recent work has explored factors such as soil disturbance (as a measure of heavy ploughing), for instance, and fertility (as a measure of manuring). But the example I’d like to share with you here concerns sowing times.

The received wisdom – based largely on later medieval documents – is that early medieval farmers sowed wheat and rye in the autumn, and barley and oats (and peas, beans, etc.) in the spring. From a biological perspective, the basic assumption is that barley and oats can make do with a shorter growing season. And by having two different sowing seasons, as part of a crop rotation regime, farmers could spread their labour more efficiently over a longer period. But it’s also widely acknowledged that all four of these cereals could, theoretically, have been sown in either autumn or spring. We can’t assume that wheat was inherently an autumn-sown crop in Anglo-Saxon and medieval England. So how can we find out?

This is where the weeds come in. A study by Amy Bogaard and colleagues (Bogaard et al. 2001) found that, against most expectations, germination time was not the main factor which determined whether a weed occurred amongst autumn- or spring-sown crops (although it was one of the factors). Rather, the biggest influence was flowering habit, i.e. when that species tends to start and finish flowering, as this dictates how long the plants have to set seed. Essentially, those species with an early and short flowering season have a competitive advantage in autumn-sown crops, but a corresponding disadvantage among spring-sown crops – because the springtime ploughing cuts them off before they have the chance to become established. By contrast, weeds with a late or long flowering season have an advantage in spring-sown fields, because they still have a chance to set seed after the springtime ploughing.

Using a special exploratory technique called correspondence analysis, we can then explore, on a statistical basis, whether – for a given archaeological site – any cereal’s grains and chaff tend to co-occur particularly with either autumn- or spring-associated weeds. And in one pleasingly clear case, the Feeding Anglo-Saxon England project has found just such a pattern, with wheat and rye appearing as autumn-sown crops, barley and (more tentatively) oat as spring-sown crops – in agreement with the later documented tradition (Figure 2). This case study is from early medieval Stafford, a fortified settlement and later town in the West Midlands of England, with both heavy clays and lighter gravels in its hinterland (Hamerow et al. 2020).

Correspondence analysis of crop and weed remains from Anglo-Saxon and medieval Stafford, western England; the closer together crops/weeds are on this plot, the more likely are their grains/chaff/seeds to occur together in charred deposits

But this certainly isn’t a universal pattern. In another case study – Mildenhall, Suffolk, in the east of England, where occupation starts rather earlier (around the 6th century) and the local soils are variable – there are no clear associations between particular cereals and autumn- or spring-associated weeds (Figure 3). In other words, there is no evidence to suggest that, at Anglo-Saxon and medieval Mildenhall, wheat (or any other cereal) was consistently sown in one season or the other. Perhaps pragmatism was the watchword of the day – maybe if a run of especially cold, wet winters was too harsh for the wheat, it was sown in the spring, but if conditions were right, it was sown in the autumn to benefit from a longer growing season?

Correspondence analysis of crop and weed remains from Anglo-Saxon and medieval Mildenhall, eastern England

Ultimately, the more sites we look at, the more variety we find. Wheat does not seem to have been always and everywhere an autumn-sown crop – as discussed above, we need not expect this to have been the case – and there is no clear indication of any ‘seasonality horizon’, i.e. a period in which particular associations arose between certain crops and certain sowing seasons, or strict regional patterns. There is more work to be done, but so far – thanks to the insights offered by weed ecology – we can suggest that Anglo-Saxon and medieval wheat may well have been a crop for all seasons.


A. Bogaard, G. Jones, M. Charles, J.G. Hodgson, On the Archaeobotanical Inference of Crop Sowing Time using the FIBS Method, Journal of Archaeological Science 28, 2001, 1171–1183.

H. Hamerow, A. Bogaard, M. Charles, E. Forster, M. Holmes, M. McKerracher, S. Neil, C. Bronk Ramsey, E. Stroud, R. Thomas, An Integrated Bioarchaeological Approach to the Medieval ‘Agricultural Revolution’: A Case Study from Stafford, England, c.AD 800-1200, European Journal of Archaeology 23(4), 2020, 585-609. Free to download at: https://doi.org/10.1017/eaa.2020.6

Dr Mark McKerracher Post-doctoral researcher, University of Oxford

Institute of Archaeology, 36 Beaumont Street, Oxford mark.mckerracher@arch.ox.ac.uk