Although it is easy to agree with Toby Pillatt that the role of weather in the archaeological record has been understated, it is more difficult to provide a clear method for incorporating the weather into archaeological interpretation. His concerns about the dominance of climate in interpretations are also acknowledged. Whereas climate and weather are indeed important to the interpretation of past ways of life, when we do try to recognize a direct correlation between, for example, a climate proxy record and any specific human behaviour, it is common to discover that any neighbouring proxy record will show a different set of relationships. Although there is a long history critiquing this approach, there is still a tendency for the uncritical to draw simplistic conclusions.

Of the many factors that can obscure the long-term relationship between humans and the environment, that of chronological imprecision continues to impede robust conclusions. Roberts (2011) has pointed out problems that arise from using environmental and archaeological chronologies, especially those that rely upon radiocarbon dating that has a precision of say ± 50 years. Although occasional varve chronologies provide annual laminations for proxy records from the lake basin sediments themselves, the archaeologist who strives to compare such records with terrestrial archaeological records (such as excavations or surveys) is then confronted with the problem raised by the lack of precision of the standard archaeological chronologies. On the other hand, when the indicators of human activity and climate derive from precisely the same sedimentary layers, there is a clear case for the proxy indicators being contemporaneous. In the case of the laminated varved sequence from Nar Lake in Cappadocia, Turkey, Roberts concludes,

The results from Nar Lake are not only informative concerning the lack of correlation between, in this case, an arid event (here aridity is derived from oxygen isotopes) and a period of large-scale abandonment; they are also starting to fall in line with a large corpus of evidence from archaeological landscape and settlement surveys that show the huge impact of Late Antique imperial policies on the landscape. One must emphasize, however, that by showing a lack of correspondence between an arid event and an abandonment phase, Roberts is not saying that climate is unimportant; rather, his point is that the relationships are not necessarily simple and straightforward. This is very much in agreement with Pillatt's article.

But in a world of proxy climate records (we must all have been rendered cross-eyed by the mind-boggling array of graphs in many recent reviews of climate change), how can we capture and incorporate the record of weather and relate it to human actions and landscape? I tend to agree with Pillatt that it is both worthwhile and relevant for the following reasons.

If we are to incorporate the weather, or perceive the landscape as ‘weather-worlds’, from where are we going to obtain our records? Pillatt is fortunate enough to have the assistance of Isaac Fletcher and Elihu Robinson, two 18th-century diarists who provide terse but valuable time-sequence records for Cumbria. Textual sources are, in general, the key to providing a record of past weather and these usually take the form of diaries or letters setting out specific events; rarely do they provide a longer-term record of climate. In the Cumbrian case, the records are relatively recent; however, similar, albeit more interrupted, records in the form of letters written in cuneiform script can be found in the Middle East going back to roughly 1200 B.C., or even to the 3rd millennium B.C. (Neumann and Parpola 1987; Cooper 1983; Widell 2007).

Closer to the period under discussion, the remarkable records of Michael the Syrian provide a graphic record of weather and other events, which until recently had not been set alongside archaeological evidence (Widell 2007). Such records provide an illuminating picture of the variety of challenges faced by ancient populations, but also their complexity. Michael the Syrian, elected Patriarch of Antioch from 1166 to 1199, provides an array of information concerning numerous climatic and agricultural catastrophes that took place in northern Syria and neighbouring areas from the sixth century A.D. until A.D. 1196. His records not only provide evidence on weather, they also show that humans in this semi-arid part of the Fertile Crescent experienced a wide range of disasters in addition to the atmospheric ones that have been the focus of many scientists and archaeologists in recent years.

Widell's synthesis of the diaries lists the following, in order of frequency of citation: severe winters, locusts, drought, snow, stormy winds, freezing conditions, hail, flood, plague, mildew, rain, rats and weevils. Of these, drought, which is normally the disaster of choice of climatologists, came third in frequency with 13 attestations over 276 years (Widell 2007, table 2), or one year in every 21. Of course, one might object that not all of those events that relate to the atmosphere are really weather: droughts might be rather longer-term climatic events or they may represent relatively short but severe episodes. Neither might they be seen as solely climatic; some may be due to political or economic circumstances. Nevertheless, they do provide a record of atmospheric events that were experienced by humans, and some, such as episodes of freezing conditions or ‘hail breaking trees and grapevines in the region of Militene’ (ibid., 52) come much closer to being weather events. Moreover, these records remind us of the importance of other weather, such as freezing conditions, that resulted in the catastrophic loss of livestock in the Aleppo region in 1911 (Lewis 1988).

What is particularly interesting about the record of Michael the Syrian is its richness: the communities had to cope with a much wider range of challenges than are normally included in many analyses of human–environment interactions. Moreover, as Widell concludes, ‘The constantly recurring catastrophes enumerated by Michael the Syrian demonstrate the resilient nature of the settlements in this region in antiquity’ (Widell 2007, 55). In other words, life in this populous region continued, despite the challenges – a point to which I return later.

Although it is relatively straightforward to incorporate climate or climate proxy records into the interpretation of the archaeological records, weather and short-term events are more difficult to use. As Pillatt observes, agent-based models provide one way of incorporating such episodic or short-term events and such methods offer a more nuanced and interesting response than simple cause and effect. For example, one modelled scenario employed for a Near Eastern model community attempted to estimate the responses of the model community to a run of dry years (Wilkinson et al. 2007, cited above by Pillatt). Because the community was not simply viewed as a physical system of production, but was endowed with a range of social behaviours such as allowing exchanges between families, offering loans, setting bride prices and so on, the responses to the ‘events’ were more complex and non-linear than the normal interpretations offered in much literature on human–environment interactions. One outcome was that the ‘cyber community’ reacted to the five-year dry spell with a flurry of exchanges within the community. In this case the affected households attempted to exchange what they had (animals) for what was in short supply (usually, but not necessarily, grain). In many cases this resulted in those households that were in a position of economic advantage gaining by trading grain for animals. Because many households were low on grain, animals were in surplus, therefore ‘elite’ households were in a position to accumulate large holdings of animals. These could then be exchanged later, and at an advantage, for more grain.

Such transactions do not assume the functioning of formal exchange economies, and in the case of the modelled non-market economies the models demonstrate that a climatic event was translated through social interactions to produce an economic output. However, the impact of such economic transactions would vary across the community and might result in very different outcomes for different households depending upon their initial circumstances or ongoing decisions. One can perceive a similar degree of complexity in the Cumbrian examples presented by Pillatt. Although agent-based methods could be employed in north-west England, it is crucial to emphasize that any formal models that incorporate large-scale processes such as urbanization in north-west England need to include a much greater degree of complexity than if models were simply restricted to a single farm or household. Nevertheless, the models do enable different scenarios to be run and allow the modeller to do ‘experiments’ to see which scenarios might have been more plausible. Agent-based models can also incorporate weather, not just climate.

An important feature of weather records is that, as such, ‘weather, as experienced by individuals, can, to a certain extent, be related to long-term climatic changes observed in proxy records’ (p. 39). Because they derive from the observations of humans, weather records can ‘humanize’ and add life to proxy records. For example, it is a fairly straightforward task to place the records of Michael the Syrian alongside the climate proxy record from, say, Nar Lake in Turkey, either to see how they correspond or to flesh out points of detail. Unfortunately, in this particular case the records are too far apart geographically to provide true complements to each other. Nevertheless, the principle that climate proxy records and human weather observations form complementary resources is important, because, by being interpreted through the lens of humans, weather records provide a different class of information. This is especially the case when the observer inserts an additional subjective interpretation that may shed light on how any particular weather may have influenced decision making at the farm level. Overall, weather records could serve to animate, validate and humanize long-term proxy records.

Although I find myself in agreement with most of Pillatt's points about the relationship between climate and weather, I am less convinced by the assertion that landscape archaeologists ‘need to completely rethink how to approach the idea of climate’ (p. 40), or that ‘in the realm of past climate studies, landscape archaeologists continue to outsource this work to their colleagues in the environmental sciences’ (p. 33). The field of landscape archaeology is so broad and there are so many different types of landscape archaeologist that although some might have taken to ‘outsourcing’, others explicitly include the use of climate proxy records within their investigations. This is particularly the case in the Middle East where many landscape study areas are situated within climatically marginal regions. Hence it is quite common in Middle Eastern regional, settlement or landscape studies to see climate proxy records used alongside landscape and settlement records and in some cases archaeologists have been involved in the collection of the record itself. Perhaps Pillatt's comments relate more to UK-based landscape archaeology, which, it is true, has a different history and methodological base than its Middle Eastern or Mediterranean counterparts.

The concept of ‘weather-worlds’ opens up many possibilities, especially in the context of Ingold's observation that ‘the land responds in countless ways to the weather's myriad expressions as the medium in which we live’, as Pillatt summarizes it (p. 34). This requires that we consider soil climate, a neglected field that provides insights into subterranean landscapes as well as practices of human use of the land.

Although there is an assumption that crops respond to atmospheric climate in a fairly lock-step manner, roots are encapsulated within the soil and so crops are more likely to respond to soil climate. ‘Soil climate’ is therefore a useful concept because it relates to conditions that actively influence crop growth. Soil climate includes soil moisture, which in the Middle East is crucial to crop development and yields. In the case of cereal cultivation, autumnal and winter rainfall nourishes the crop and if there is enough soil moisture, crops grow and yields may be sufficient for seed, consumption and some storage. However, traditional and ancient practice inserts a fallow year into the seasonal cycle (i.e. biennial fallow), with the result that a small percentage of soil moisture is carried over during the fallow year into the cropping year. This results in a bonus in soil moisture, which might be anywhere from 5 to 20 per cent and increases crop growth as well as acting to stabilize crop yields (Janssen 1970; Wilkinson 1994). The practice of violating fallow has been noted to contribute to greater instability of cropping in the former Soviet Union (Parry 1990) and could have amplified any effects of climate change in the ancient Middle East. Therefore a simple change in the land-use regime from biennial to annual cropping can result in a decrease in soil moisture and loss of yield. Overall, human practices that vary the fallowing interval can have a significant effect on the soil climate and sustainability of the cropping system, perhaps even equivalent to many climatic ‘events’.

In Britain, soil climate will vary spatially because of the variegated nature of the cultural landscape itself. Because the inhabitants of the countryside, including the Cumbrian farmer, are usually very knowledgeable about their local soils and micro-climates, any records might include both weather and comments on the response of the soil to such weather. Perhaps, as Tom Williamson (2003) has argued, landscape archaeologists need to be more sensitive to soils and their patterning.

In addition to the points made by Pillatt, archaeologists could shift away from the use of atmospheric climate change alone and engage with at least three components of climate: climate, weather and soil climate. To these one could add, at the most general level, a fourth component, namely a cumulative measure of climate such as the particularly informative ‘cumulative deviations from the mean approach’ employed by McGovern et al. (2007) (cited by Pillatt). By including a sequence of cumulative climatic records it is possible not only to incorporate a ‘memory effect’ but also to provide evidence of major turning points in the atmospheric record.

Finally, I agree that there has been a tendency for archaeologists and environmental scientists to be too keen to find causal relations from correlations between climatic events and human societies. This has often led to the ‘cherry picking’ of data to draw particular attention to apparent correspondences between climatic downturns (droughts in the Middle East) and social devolution or even collapse, at the expense of other relationships. What is often of much greater interest (to this respondent at least) is how humans manage to maintain their societies through major periods of stress, be they from weather or climatic events. If a community manages to survive a climatic event of some severity by adopting a suite of responses, that is every bit as interesting as if they had succumbed. Such results not only tell us about community resilience, they also demonstrate strategies that may be useful for survival: ‘human resilience in the face of adverse circumstances’ is surely every bit as useful as the recognition of societal collapse or devolution. More efforts should therefore be focused on recording what happens archaeologically across periods of significant climate change. Questions to ask should include: what changes in land use and crop type occurred? Did any other forms of agrarian practice change (e.g. fallowing, manuring)? In the Near East, was there an attempt to introduce irrigation? And so on.

To conclude, climate change has for too long dominated the agenda of global-change archaeology, and it is certainly time for weather (and weather-worlds, soilscapes) to be examined alongside it. However, rather than jettison climate or climate proxy records, weather and climate (and, indeed, soil climate) should be seen as complementary facets of the atmosphere, and bringing them into a broader perspective of long-term socio-economic change is to be welcomed.