In just the same way as it is impossible to isolate the Celts, so it is to determine a specific kind of agriculture which might be described as ‘Celtic’. That agriculture formed the basic economy of Europe and the Mediterranean zone by the first millennium BC is not in question. However, outside the classical world our knowledge of the nature of agriculture is severely restricted by the lack of any significant documentary sources. A few tantalizing references occur in the works of Greek and Roman commentators but they are barely enough to construct any kind of coherent picture. The practice of agriculture, probably more than any other industry, is constrained by the nature of the soil and vicissitudes of climate. It is, therefore, important to recognize that agriculture in Europe and particularly in Britain is quite different from the agriculture of the Mediterranean zones. In consequence the classical works on agriculture cannot be used to provide any kind of generalized insight into what happened in northern Europe. This applies equally to soil preparation and treatment and to the particular crops cultivated. Bearing in mind that agriculture in the sense of food production probably began in the latter part of the seventh millennium BC in the fertile crescent at the eastern end of the Mediterranean and gradually spread throughout Europe to include Britain and Ireland by the fourth millennium BC, specific Mediterranean practice would have been adapted and changed quite significantly as man responded to soil and climate change. Similarly within the Celtic world of the first millennium BC, arguably the land area stretching from the Pyrenees to the Rhine and from Ireland to Romania, contemporary farming practice would have varied quite considerably from one zone to another: differences would have been dictated by the varied climatic zones and soil types. Until quite recently, with the advent of agrochemicals, farmers have been able to grow only those crops which any particular landscape will allow them to cultivate. The ability to influence the natural prevailing conditions was extremely limited.
The construction of a picture of agriculture in the Celtic world has to be based upon the data extracted by archaeological excavation. Inevitably these data are fundamentally inadequate. Agricultural practice and its produce are by definition ephemeral and annual: ephemeral in the sense that agricultural operations are carried out day by day, ploughing, manuring, planting, hoeing and reaping; and annual In that the fruits of farming depend upon the seasons of the year to reach maturation.
To isolate an annual event in the archaeological data is virtually impossible. It is also true to say that climatically each year is a unique event having quite specific challenges and responses which normally defy clear identification. The data range from pollen grains, impressions of seeds fired into pottery, carbonized seed more often than not the result of an accident, desiccated and waterlogged plant remains, the former virtually non-existent in Europe, traces of ploughing left in underlying rock or subsoil of soil layers identified to the period, ancient fields surviving as monuments in the landscape, occasional tools and implements or fragments thereof and a limited range of iconography. In fact, it is this last type of data which gives some of our best source evidence for agricultural practice in north-west Europe, but the majority of it is to be found in Scandinavia rather than in the limited ‘Celtic’ zone. However, the close similarity of the other surviving data suggests that the agricultural responses were the same then and it would be foolish to deny such useful evidence simply because it falls marginally beyond the Celtic lands.
In any approach to understanding the remote past it is critically important that the argument or interpretation is directly driven by the archaeological evidence. Where there is an assumption of a practice which must have occurred to sustain the existing data, every effort must be made to Identify such a practice by exploring the processes which might have left physical traces previously unrecognized or not linked with such a practice. The provision of winter feed for livestock is such an example and is examined below. The following discussion will demonstrate how little Is known for this period and how much there is yet to discover.
Agriculture is traditionally divided into two general categories of arable and pastoral farming. There seems little doubt that the great majority of farms, with minor exceptions, practised a mixture of these two categories, any emphasis on one or the other being dictated by soil and climate. In broad terms Britain can be divided into two agricultural zones; the region south-east of a line from the Bristol Channel to north Yorkshire but including south-east Scotland Is primarily devoted to arable farming while north-west of that line pastoral farming is the norm. Given the minimal change in climate between the present day and two thousand years ago, the same constraints would have obtained for the Celtic farmers.
The single most significant element of arable farming is the plough itself. A full understanding of the technology of tillage is regarded as an Indicator of successful arable farming. The normal appreciation of Celtic or iron age farming falls somewhat short of this state, the plough being discussed simply as a stick ard which does little more than scratch the surface of the soil; hence the farmers merely scratched a living from the soil.
To compound the issue the assumption has normally been that the soil must also be light and therefore relatively poor. From the peat bogs of Denmark a number of these so-called stick ploughs have been recovered which, on close examination, rather belie their dismissive description. One typical example Is referred to as the Donneruplund ard, named after its find location. The reason for Its deposition in the bog is generally thought to be ritualistic but, since the tool was actually worn out and broken, it was most probably dumped there with a curse rather than a blessing. The simple difference between an ard and a plough is that the latter is fitted with a curved mouldboard which inverts the soil. Its probable introduction occurs in the tenth century AD. Nevertheless, it is a complex tool comprising a main share which is in fact a pointed stick, a heart shaped undershare fitted with spigots which hold the main share in position, a curved handle or stilt, all of which pass through a mortise joint cut in the foot of the main beam and locked into place with wedges (Figure li. i). The wear pattern on the undershare (one side was worn away to the spigot, thus causing the ard to be abandoned) strongly suggests that the ard was used in a particular and specific manner. One side was continually! n undisturbed soil, the other in disturbed soil. If this is the case, the ploughman must have ploughed the soil in ‘lands’ or blocks rather than going up and down the field laying one furrow immediately against another one. The interpretation of this sophisticated technique has been enhanced by the construction and testing of a full-scale replica. The angle of presentation of the main share (the pointed stick to the soil surface) is c.29° from the horizontal, which ensures it neither bounces along the surface nor digs Itself into the ground. Quite simply, it holds the implement at a steady level in the body of the soil. The heart-shaped undershare lifts the soil, which then flows past the foot of the main beam of the ard, having been thoroughly stirred. In practice it is extremely efficient and is able to cope with a wide range of soils including heavy loams as well as the light rendzinas. A large number of iron socks or sheaths designed to protect the end of the main share from excessive wear have been found on iron age sites throughout Europe. Without such a protection (none was found for the Donneruplund ard), the main share wears away at an average rate of c.6ij mm per hectare. Given the ease of adjustment for the main share this hardly represents a problem.
Figure ii. i Replica of the Donneruplund ard. (Copyright; Peter Reynolds.)
In addition to the actual ard itself there are a number of rock carvings, primarily in the region of Bohuslan in Sweden, which show such an ard being drawn by a pair of horn-yoked cattle. The great majority of these rock carvings date from the Late Bronze Age to the Early Iron Age. Some even show the vertical bar between the main beam and the share, suggesting either a method of adjusting the angle of penetration or the presence of a coulter, in effect a vertical knife. One major problem encountered in the use of the Donneruplund replica was the bulk of vegetation and roots which collected between the angle of the share and the main beam. A coulter would have been a useful addition. An edged iron bar, however, even should it survive, is unlikely to be identified as a coulter out of context. A replica of this so-called stick plough has become an extremely successful and useful implement, so successful in the lighter soils, in fact, that it produced furrows in the ploughsoil up to 300 mm deep, which meant that the field area had to be smoothed or levelled out before it became a seed bed.
A second ard of totally different design was recovered from another peat bog in Denmark at Hvorslev. Quite simply, the main beam is an appropriately curved tree branch and the trunk from which the branch grew was fashioned into a horizontal share. At the rear a mortise joint was cut into which the handle was fixed. This ard, too, was worn out and most probably thrown away. Trials with a full-scale replica proved quite disconcerting in that it failed totally as a tillage implement. However, it too is represented on a rock carving scene from Littlesby in Sweden. This depicts as a ritual what can only be a spring sowing scene. Both the ploughman and the bulls are shown with rampant phalluses, a bag the ploughman carries is interpreted as a bag of seed and two horizontal lines below are thought to be the furrows waiting to be sown. Changing the kind of trials with the Hvorslev ard from ploughing the soil to drawing seed-drills in a previously ploughed soil demonstrated quite clearly that this was its primary function. The furrow it produced averaged just 200 mm deep, the ideal depth for seeding in north-west Europe. If such a practice was the norm, first ploughing and then seed drills, another major reassessment is necessary. If the seed Is sown directly into a prepared drill, the total germinability of the seed, normally in excess of 95 per cent, is enjoyed by the farmer. In other words the input is total, unlike broadcasting the seed, which has a loss rate of up to 75 per cent, as both biblical parable and practical trials confirm, requiring considerable over-input to achieve the same end product. Without increasing the input, lower production is the inevitable result.
There remains yet a further problem posed by the archaeological data. This is the plough or ard marks found on prehistoric and later sites in all types of soil. They comprise interrupted score marks in the underlying rock, whether that is chalk, clay, sand or loam. They often indicate multiple ploughing and occasionally cross-ploughing. Repeated trials with the above ards completely failed to produce any kind of comparative evidence. Indeed, only when things went terribly wrong, when the ard tip buried itself in the soil, with commensurate risk and danger to the ploughman, did any kind of mark In the subsoil occur.
Unfortunately no physical plough or ard like those above has yet been discovered but further prehistoric rock-art scenes perhaps hold the key. There are three specific examples, one from Sweden, one from southern France and one from northern Italy, which depict an ard scene with a share set at an extremely steep angle to the ground.
None of these could be used as a regular plough to create a tilth because the angle of presentation is such that the implement would bury itself almost immediately. In north-west Spain in the province of Galicia a similar type of plough or ard was in use in this century. In effect it was an oak hook, the point tipped with an iron sheath, the upper curve attached to a straight plough beam fitted at the rear with two grab handles. It was used specifically to break up ground previously uncultivated or which had lain fallow for many years. The manner of use Is especially interesting. The implement was attached to a pair of bulls (each district kept such a pair for this purpose as well as the more prosaic reasons of husbandry), a ploughman firmly grasped the handles at the rear of the beam, the point of the hook just locked into the ground surface, two further men armed with goads simultaneously jabbed the bulls’ rumps. They in turn lunged forward to escape the goads, the hook was driven fiercely through the ground, effectively burying itself in uprooted vegetation and soil, and the whole ensemble came to a juddering halt after 2 to 4 metres’ progress. At this point the hooked share, for such it was, was wrestled out of the soil and the operation repeated again and again. The resulting upheaved clods were broken down with mattock hoes Into the semblance of a tilth which was then ploughed with the regular ard. Excavation of this process revealed a typical plough or ard mark in the underlying surface.
Re-examlnatlon of the actual prehistoric ard marks indicates the average length to be between 2 and 4 metres, often with an area of greater disturbance at one end. It would seem that these marks rather correlate with the kind of action of what can best be described as a rip ard or sod buster. It is most unlikely they are the result of regular ploughing, which can occur several times a year since, within a relatively short time, they would be cancelled out. The implication of the ard marks, however, does not cease with the argument for a rip ard but also supports the idea of fallowing land to allow It to recover.
Finally with regard to ards and ploughing. If the evidence of the rock art is admissible, then a remarkable ploughing scene from Krokholmen In Bohuslan discovered in 1971 but previously unpublished gives even more Insight into agricultural practice (Figure 11.2). The scene clearly shows a double team of cattle, undoubtedly cows, pulling an ard, with a ploughman at the rear holding the stilt of the ard and another figure midway between the teams seemingly in close attendance to help steer the cattle. The major importance of this scene is the use of a double team of cattle, in effect increasing the traction power presumably to cope with a heavier soil. Apart from this one example to date, the increasing of the cattle to two or more spans is thought not to have occurred until the Middle Ages. The evidence is quite clear that a panoply of ploughs or ards existed in the Celtic period, raising the level of tillage technology far above that implied by the description of a stick ard. Given this level of equipment and skill, there is clearly no particular landscape or soil type which could not be tackled successfully.
There Is an abundance of evidence for prehistoric fields and field systems to be found in Britain. As a general rule they have survived as field monuments on hlllslopes delineated by the low banks or lynchets which formed through soil creep at the lower side of the field, during their use. Abandonment has allowed them to become stabilized by vegetation, and subsequent grazing means that the majority are under grass.
Figure ii. i Rock carving from Krokholmen in the region of Bohuslan, Sweden, showing a double team of cattle pulling an ard. (Copyright; Peter Reynolds,}
Unfortunately, the agrochemical revolution of the past forty years has seen great swathes of these fields destroyed by the plough. Many can still be seen outlined by soil marks where once the lynchets stood but even these are disappearing at a depressing rate. Ironically, the prehistoric fields had survived on the poor light soils particularly in Wessex and Yorkshire, and without chemical boosting of these soils they would survive still. The fact that iron age farmers were cultivating the poor soils themselves raises a question about the extent of land under cultivation at that time. The distribution of iron age sites across all soil types in all regions clearly denies that the fields which survive as monuments were the only areas cultivated. Undoubtedly the whole landscape was under intensive and necessarily diverse use. Because these fields have survived on the poor soils rather implies that during the Iron Age pressure on cultivable ground was greater than at any other time until the present century.
The fields themselves tend to be square rather than rectangular in shape, and given the ard marks described above, were probably cross-ploughed as a rule. Certainly a better tilth is gained by cross-ploughing, the clods of earth being attacked from two directions. Also because the organic content in the soil is high, the roots and plant material tend to be streamed in the direction of the plough. Turning at right angles to the stream does break the material down more successfully and in addition brings persistent root masses to the surface, allowing them to be pulled out more easily. One of the greatest enemies to the farmer, a great colonizer of cereal fields is couch grass {Agropyron repens), which needs continuous rooting out if the cereals are to thrive. The field sizes range in extent from o. i6 to 0.25 hectares and broadly represent an agricultural day’s work. Such an area can be ploughed, sown, hoed and reaped within a working day. There is seemingly no other reason to offer for their size since the farmers were technologically fully equipped to make much larger fields. The larger fields of the Roman period, for example, were cultivated with exactly similar equipment.
Very few fields have been examined archaeologically and evidence for field boundaries is extremely slight. Some fields have been edged with a continuous wattle fence. Perhaps hedges were set between the fields. Recent evidence supports the possibility of hedges on top of enclosure banks around settlement sites. Perhaps the cultivated fields were simply left without specific physical boundaries like those to be seen in Galicia in north-west Spain. Alternatively, blocks of fields may have been fenced in. By the same token, if the fields were fenced, no clear evidence of gateways has yet come to light.
Our knowledge of the crops cultivated in the latter part of the first millennium BC comes almost exclusively from carbonized seed, seed accidentally burned and turned into charcoal within the settlement zone, except for representations on the reverse of some Celtic coins. In this latter case there is a stylized ear of cereal which is most probably emmer wheat rather than barley, which is the more usual interpretation. Because both are bearded cereals, the confusion is easily understood. If the representation is to indicate wealth or even to advertise a product like the representation of vines on Roman coins, the likelihood of its being emmer wheat is reinforced by Strabo’s comment that this was a major export from Britain to the Continent. The seed evidence, however, is comparatively slight and gives at best only a presence and absence listing. The critical point is that carbonized seed is invariably recovered from the settlement zone, and therefore has had to have been moved from the production zone, the fields, into the settlement area, probably during harvesting, before it could have suffered the accident which led to its carbonization. That representatives of all the plants within the cultivated areas were brought back into the settlement is extremely unlikely.
The list of cereals available to the Celtic farmer differs little from that of today. There were four types of wheat, four types of barley, oats, rye and probably millet.
The finds argue for wheat and barley as the predominant crops. Their presence and diversity, however, give no insight into how they were actually cultivated, nor is there any documentary evidence. Britain has a climate distinctly different even from that of the near Continent and undoubtedly this would have been exploited to the full. Caesar describes our winters as less severe {remissioribus frigoribus) (v. l2) and the humid temperate climate, driven as it always has been by low pressure from the Atlantic, provided ideal conditions for cereal production. The principle of autumn sowing is traditional in the Mediterranean zones and presumably with the arrival of the first farmers to Britain in the Neolithic this practice was continued. However, because the winter here tends to be over by early March and summer is considerably less severe and barely arrives until late June or even July, spring sowing of cereals is a positive option. The advantages of two sowing seasons are not inconsiderable since the harvesting time is staggered, the winter-sown crops being ready before the spring ones; the work load similarly is spread and - of economic interest - the yield from autumn-sown crops (because of winter frosts checking growth and subsequently increased tillering by the plants) is greater. In addition some cereals, like millet, are frost-sensitive and can be sown only in the spring. Specifically, it allows for greater areas of land to be cultivated and therefore for greater returns.
The evidence of the seed-drill ard discussed above suggests that seed potential was maximized but it gives no real indication of seed input. The fundamental assumption must be that an adequate seeding rate had evolved in the sense of minimum expedient input to perceived maximum output. Research into prehistoric crop yields at Butser Ancient Farm (Figure 11.3) spanning more than twenty years has been based upon this premise. The minimum input assumed in the research programme is a mere 50 kilos per hectare, approximately a quarter of the modern sowing rate. The other issue of paramount interest with regard to crop yields is whether fields were manured or not. The general assumption is that manuring was practised from the Bronze Age onwards, based upon abraded sherds of pottery being recovered from field areas. Tantalizingly, however, very little evidence has been found for the presence of middens or manure heaps within enclosures, though present research into the trace evidence of lipids may alter this in the future. The difficulty lies in the vCry organic nature of the material and its rapid dissolution and disappearance. In consequence the Ancient Farm research programme has examined a range of treatments including manuring and non-manuring practice. The results averaged across two decades suggest surprisingly good yields of both emmer and spelt for manured fields of 3.5 tonnes per hectare and non-manured fields of 1.7 tonnes per hectare for autumn-sown fields and slightly less for spring-sown fields. These figures correlate favourably with modern yields prior to the introduction of chemicals. All of which suggests that surplus production was well within the grasp of the Celtic farmer, especially as the results quoted are gained from a worst option, since the trials were conducted on the poorest of soils, a friable rendzina over middle chalk on a north-facing hill slope. Given a good soil in a protected river valley, the results would have been commensurately improved. In addition to the cereals, the evidence from carbonized seed indicates the presence of several legumes in the Late Iron Age. Primarily the Celtic or tic bean (Vida faba minor) is represented along with vetch (Vi saliva) and with the very occasional pea {Pisum sativum).
Figure ii. j General view of Butser Ancient Farm, nr Chalton, Hampshire.
(Copyright: Peter Reynolds.)
Conditions for the accidental carbonization of vegetables are seemingly more rare. The presence of these leguminous crops, well attested, of course, in subsequent periods, rather complicates the treatment options open to the prehistoric farmer. Crop rotation must be regarded as a likely treatment, with the legume crop fixing nitrogen in the soil to the advantage of any following cereal crop. Results from this treatment at the Ancient Farm suggest a regular cereal return year on year of 2,6 tonnes per hectare. A third option is also not unlikely: the growing of beans in particular inter-rowed with the cereal. The major benefit from this is not only the simultaneous deposit and utilization of nitrogen after the first year but also the stouter stalks of the bean plants literally holding up the cereals in bad weather conditions and preventing lodging. This symbiosis of crops can be extended to include both vetch (VI saliva) and tufted vetch (VI cracca), though if the growth of the vetch is excessive it can actually cause lodging. Traditionally rye {Secale cereale) and vetch (V. saliva) have been grown together but in recent times primarily as a fodder crop for livestock.
Besides the major food crops, evidence abounds for the growing of flax {Linam Hsitatissimnm). Whether this was specifically for the stem fibres to manufacture into linen or for the oil which was obtained by crushing the seeds is difficult to assess, since there is virtually no surviving evidence in Britain for the post-harvest processing. In all probability flax was grown for both purposes. Another oil-producing plant, gold-of-pleasure (Camelina saliva), is also found, though it may have been a weed of the flax crop itself.
One particular plant, fat hen {Chenopodium album), occurs very regularly in the seed evidence from iron age sites. Today it is universally regarded as a weed but in times past the young plant has been used as a vegetable like spinach for human consumption; the mature plant can be treated like hay for winter animal fodder and the seeds can be ground up into a flour for bread-making. Its frequency suggests it could well have been a serious crop plant in prehistory, especially with regard to its germination time and short life-cycle. It normally germinates in early June and can be harvested in early September. Given its diversity of uses, it could have been employed as a catch crop, being planted when a cereal crop had failed. Alternatively it could have held its place as a cropping plant in its own right.
The wealth of cereals, legumes and other plants clearly indicates that the Celtic farmer had a wide variety of choice. In addition, given the knowledge of the microclimate and soil types available to him, there can be no doubt that land was used optimally. It requires but little experience not to plant specific crops where they won’t thrive.
The methods of harvesting crops, especially cereals, offer a number of choices in that the resources a crop offers are quite considerable. A reference by Strabo which describes the Celtic practice of specifically harvesting the ears of the cereals focuses attention upon the problem. No doubt Strabo mentions the practice simply because it was so different from the Roman harvesting methods. If he was correct in his observation, and the subsequent Celtic invention of the harvesting machine {vallus) in the second century AD which strips the heads off the cereals supports him, then the direct result (substantiated by experiment) is a virtually pure harvest of the cereal in question. When both emmer and spelt wheats are ripe and ready to harvest, the joint between the cereal stem and ear, the rachis internode, becomes extremely brittle and breaks off very easily (Figure 11.4), so easily, in fact, that the use of a sickle is made redundant since the ears literally come off in the hand. Impurities in the crop are represented primarily by black bindweed {Polygonum bilderdykia) amd common cleavers {Galium aparine) which entwine themselves around the cereal and its ears and are extremely difficult to separate during harvesting. These too are found ¦dth carbonized cereal grains. Common cleavers is particularly interesting since it might well be an indicator of the autumn sowing of cereals. It rarely appears as a weed of a spring-sown crop.
The obvious second crop of a cereal field is the straw itself. In the case of barley straw, the crop is a significant source of winter fodder, while the wheat straw, less palatable to livestock, is important for thatching, animal bedding, perhaps for matting and even basket-making. But there is potentially a third crop to be considered. Inevitably the fields were infested with arable weeds even if the spaces between the rows were carefully hoed during the growing season. A common ratio of arable weed to cereals even in a managed field, as revealed by experiment, is 2:3. Of these arable weeds all of which germinate after sowing and come to fruition before harvest, quite a percentage are food plants. The vetches, cleavers, oraches, bindweeds and fat hen, amongst others, are all worth collecting as storable food supplies. It is not unreasonable, therefore, since all these seeds are found in the carbonized seed record, to suppose a triple harvesting, first for the ‘sport’ food plants, second for the cereal itself and finally for the straw.
Figure 11.4 Emmer wheat {Tritieum dieoecum) at harvest time. {Copyright: Peter Reynolds.)
The harvest, whether it was double or triple, spanned most of August and September and involved its transfer from the fields into the settlement area. Bearing in mind that the focus of archaeological attention is invariably upon the settlement, only that plant material which is transferred from the fields has any chance of being represented in excavated data. The incompleteness of that data is emphasized when one examines a harvested field after the removal of the crop. A large range of low-growing arable weeds is present but unrepresented in the harvest Itself. Typical examples include the corn pansy {Viola arvensis) and scarlet pimpernel {Anagallis arvensii). In fact, the overall view of a harvested field immediately suggests its value as animal fodder, especially as the grass grows only poorly at this time of year. The principle of turning livestock out into the stubble, both to clean the fields and manure them at the same time, is self-evident. The other alternative of burning the stubble is a real possibility but to prove it further work needs to be done in examining surviving and undisturbed prehistoric fields.
Similarly it is, as yet, impossible to identify the methods or zones of treatment of the harvest itself. The critical process is the preparation of the cereals, in particular for storage. All the cereals in question are bearded, and for practicable storage it is necessary to remove the beards or awns to reduce the bulk. The beards may have been singed off or alternatively beaten or flailed off. The presence of the flail is argued as early as the Neolithic in Switzerland. The former system might well lead to carbonized seed as the result of too enthusiastic processing and certainly leaves the ears entire, which means a second breaking-down process into seeds or spikelets. The second and more likely system, certainly if the traditional treatment of cereals has its beginning in prehistory, achieves both ends in one process. The cereal is heaped up and beaten with flails or even sticks and subsequently winnowed. Thereafter it can be stored. The archaeological evidence for storage is of two major types. For the Middle Iron Age in particular there is an abundance of pits which have been determined to be grain-storage pits. These are generally cylindrical or beehive-shaped with a diameter of c.1.50 m and a depth between i. o m and 2.0 m. Exceptionally, pits deeper than 3.0 m have been found. Long series of experiments have demonstrated that storing grain in such pits is extremely successful. The practice is referred to by both Tacitus and Pliny. Quite simply, the pit is filled with grain and the mouth is sealed with clay or even dung and covered with soil. The clay or dung, provided it is kept damp, makes an hermetic seal for the pit. The grain immediately adjacent to the seal and the walls of the pit begins to germinate, using up the oxygen and giving off carbon dioxide. Within the space of three weeks the atmosphere within the pit has become loaded with carbon dioxide which inhibits any further germination in the bulk of the grain. The loading by volume can reach as much as 20 per cent (in air the normal carbon dioxide content is 0.006 per cent by volume). The germinated grain dies and forms a thin skin against the pit surface, representing a loss rate of less than 2 per cent of the quantity stored. Provided the seal remains intact, grain can be stored in this way for long periods. However, in all probability it was stored only for the winter period. Again experiment has shown that grain stored in this way retains its germinability quite remarkably at levels over 90 per cent. In consequence these storage pits require careful consideration. The average pit volume holds approximately 1.5 tonnes of grain. That grain can be either food grain, enough to feed at the least thirty people eating a mixed diet or seed grain, enough at the assumed sowing rate above to seed 25 hectares. What is certain is that the whole contents of the pit have to be removed once the pit is opened, since resealing is impossible. These pits, therefore, may represent the safe warehousing of grain, probably seed grain, for the export to which Caesar refers. Major sites like Danebury hill-fort, where great numbers of pits were found, could represent collection centres, although most minor sites of this period have one or more such pits. More mundanely the major sites could be controlling grain supplies in the sense of collection and redistribution. Whatever the management might have been, the pits clearly represent the storage of grain surplus for the immediate requirements of the ensuing winter and underline the success of arable farming.
It has been argued that the other system of storage comprised small granaries set on large posts above ground very much like the small buildings set on staddle-stones still to be seen in the modern landscape. The primary purpose of these buildings is to allow air circulation all around the structures and secondly to inhibit access to rodents. These buildings are likely to have been storage sheds not only for grain but also for other materials. Their average size is some 2 m x 2.5 m, giving a potential capacity of over 7 cubic metres, which is virtually impossible to exploit fully because of problems of access and management. It is also likely that any grain stored in such buildings was kept in sacks or bins. With regard to the actual bulk of the grain needed for human consumption, half a tonne is small enough to keep within the domestic house, the grain being ground into flour as required.
Livestock was without doubt important to the Celts but it is virtually impossible to quantify that importance. The documentary evidence is slight and devolves primarily upon Caesar’s comments that grain and leather were two principal exports. Britain lends itself to both cereal production in the south-east and pastoralism or stock-raising in the north-west. Perhaps it is not beyond the realms of possibility that in Celtic Britain prior to the Roman conquest cattle drives were made from the northern regions to the south-east ports. If stock is raised for leather only, it is much easier to move on the hoof and process at the latest possible stage. Perhaps the return trade was in cereals, needed but difficult to produce in the north-west. Leather, of course, need not imply only cattle. Sheep and goatskins are equally of value and wool would logically have been another trading item.
Archaeological evidence is restricted to the usual principal sources; bones, coprolites, representations like rock carvings and figurines. Occasional discoveries of hoofprints have been made but these are more curiosities rather than specific evidence. The bone evidence itself, discovered during site excavations, is not unexpectedly relatively sparse. In fact, it is surprising that any does survive, given the ways in which all parts of an animal carcass can be put to good use. Though quantification of bone evidence is carried out with painstaking care and skill, it is difficult to relate the actual evidence itself with the organized running of an agricultural unit. Since it is virtually impossible to date the bone assemblages to a particular century, let alone assign any contemporaneity within the assemblage itself, it is important to remember that a decade in farming, like a week in politics, is a long time, certainly long enough to see shifts of emphasis in a farm’s livestock holding - whether by choice or by external constraint, like disease or extreme climatic conditions or a combination of both. In recent cool humid summers in the 1990s farmers have lost 50 per cent and more of a flock to fly strike. Other fatal diseases, not yet eradicated, could well have been present in the Iron Age. Lung worm and liver fluke were certainly present; if they are unchecked, these diseases will debilitate sheep to the point of death. Less dramatic in terms of maintaining livestock numbers is the sheer necessity of providing winter fodder. If the summer harvest of grass and leaf hay is inadequate, then stock numbers most certainly would have been reduced in the autumn. There is little point in eating an animal which has starved to death rather than culling it at its prime in the early autumn.
If the bones cannot give a realistic idea of proportions of stock, at least they tell us what kind of stock was kept. That it was ultimately kept for food is shown by the occasional discovery of butchery marks.
The cattle were by modern standards relatively small. The medium-legged Dexter cattle are the modern equivalent of the Celtic shorthorn. The Dexter, bred in the nineteenth century from the Kerry cattle of Ireland and the Welsh Black cattle, themselves probably descendants of the Celtic cattle, has a number of characteristics likely to have been present in its remote ancestor. It is a tough, powerful animal capable of thriving on relatively poor pasture in challenging conditions. Experience in training Dexters to the yoke ard to ploughing with replicated iron age ards has shown them quite capable of ploughing a fifth of a hectare a day. Cattle management can only be guessed at in the context of the prehistoric period. From the many rock carvings it can be seen that both bull and cow were horned, which, while useful for yoking, leads to difficulties in winter housing. On the Continent the long-houses indicate the use of individual stalls. In Britain evidence of indoor overwintering is inconclusive. The reason for separating cattle when they are kept in close proximity is the dominance factor. In every herd of cattle or any other group of farm livestock there is a strict order of dominance, with usually a lead cow. Even with a yoked pair of cattle, one of the pair will dominate the other, a fact exploited by the ploughman by putting the dominator on the land side of the work. The working pair of cattle undoubtedly received different treatment from that of the general breeding herd. They were probably housed within the farmstead, specially fed and watered and, most importantly, they were tame. They represented the power unit of the farm. The remainder were kept for milk, beef and hides. Cows mature at about two and a half years old, at which time they can be put in calf and subsequently provide milk. The gestation period is nine months and most cows will calve annually if managed in that way. To obtain all the dairy products, there must have been some kind of organized management. Critically, those animals deemed to be worth keeping, as opposed to culling as calves, had to be kept as unproductive animals for over two years. It is likely that the working pair, probably cows rather than bulls or steers, were selected from the herd at 5 or 6 years old to maximize their value. It is interesting that in the Celtic legends of a thousand years later cattle were regarded as being at their prime at 7 years of age.
Gourmet connoisseurs of today bemoan the modern tendency to describe 3-year-old cattle as beef and indeed 3-year-old sheep as mutton. It would seem neither beef nor mutton grace the modern table as they surely did the Celtic feast.
The difficulty of distinguishing sheep and goat bones has led to a strange hybrid referred to in specialist reports as a caprovid. However, sufficient evidence has been recovered to identify both bronze age and iron age sheep. The typical sheep of the Bronze Age was the Soay, a breed which has survived in the Hebrides. Finds of both wool and bone identify it accurately. It is a small but athletic animal, both female and male usually horned, and the wool is plucked or rooed in the early summer. Wool colour ranges from dark brown to oatmeal with occasional white. In the Iron Age the sheep were slightly heavier boned and larger. The probable breeds were the Hebridean and the Manx Loughton, survivors respectively in the Hebrides and the Isle of Man. Both breeds occasionally have four horns in male and female. The wool colour of the Hebridean is normally dark brown and for the Manx a fawn; their fleeces, a longer staple than that of the Soay, are shorn. Their arrival coincides with finds of sheep shears. At the end of the first millennium BC the Shetland sheep is identified: it has a much longer stapled wool ranging in colour from white to moorit. While it is neat to docket each breed into a specific time slot, the reality was probably entirely otherwise. A flock of sheep at the end of the Iron Age would have been a mixture of all three breeds, some characteristic of just one type, others crosses between the breeds.
The primary value of sheep is for meat and wool, though they might have been milked as well. In terms of bone survival it is quite remarkable that any escaped the omnivorous attentions of self-respecting dogs. Breeding maturity for sheep is normally reached in the second year, along with the first fleece. Like cattle, sheep need to be foddered over winter and the same considerations apply for them.
The probable descendant of the prehistoric goat is the breed known as the Old English goat. Relatively small and tough, the goat undoubtedly had its place in the Celtic farmstead. Far less fussy than cattle or sheep, the goat will eat almost anything. In addition, having kidded, it will continue to produce milk well beyond the kid’s weaning time.
The management of sheep and goats is difficult to assess with any accuracy. There is a need to excavate areas beyond settlements in order to attempt to discover the presence or absence of grazing paddocks. It would seem from the abundance of settlement sites and their close juxtaposition, along with the focus upon cereal production, especially in the south, that open grazing areas where flocks of sheep and herds of goats might browse were at a premium. The pastoral idyll has the shepherd or goatherd wandering about the landscape with his charges, perhaps playing a note or two on the pipe, returning to the fold each evening: the sort of thing to be seen to this day in the Mediterranean where the maquis abounds. In temperate Europe, however, there is no maquis. By the same token sheep and goats must be kept off cereal fields and, indeed, freshly coppiced woodland too, where they will, if given the opportunity, destroy tree shoots with relish. The question focuses upon the nature of the landscape. Was it ordered and totally managed or was it farmed in tiny pockets surrounded by rough uncultivated land? The evidence to date indicates the former. In consequence it is likely that cattle, goats and sheep were kept in some form of paddock system, which in turn led to grazing management regimes.
The pig, both domestic and wild, was equally important in the Celtic world. A large number of figurines of wild boar have been found, including on shields as emblems. There is no doubt that it was revered for its ferocious fighting characteristics as well as it wondrous feasting qualities. The later legends of boar hunts suggest that the chase was an important element of the boar’s status. Perhaps the wild boar was particularly important because the hunt for it represents a major leisure activity, a time within the welter of farming activities when a man could choose a particularly dangerous way to prove his manhood. The domesticated version of the wild boar was undoubtedly kept but exactly how remains a problem for archaeologists to solve. Pig bones are regularly well represented m assemblages but evidence for housing or control is at present lacking.
Bone evidence for poultry is meagre. Caesar remarks that geese were kept for pleasure {animi causa) but makes no reference to chickens. Since chickens were widespread throughout the Mediterranean countries, their presence in Britain probably warranted no special mention. Geese, however, held a special place in what for Caesar was contemporary Roman history. Exactly what is meant by animi causa is difficult to interpret since the real meaning is about spiritual pleasure. Our knowledge of the importance of birds in the Celtic spiritual world barely ranks the goose as especially significant. Nonetheless the image of a Celtic farmyard must be populated by free-range chickens and geese. As for specific types it is attractive to think of the chickens as being Old English game fowl. These birds have a reputation for hardiness and aggression. The cocks have been much sought after as fighting birds. It is interesting to speculate whether some of the circular buildings were not houses but cock pits. This would, indeed, have been animi causa and fits into a long tradition of the sport. The geese could well have been the grey lag, an elegant, medium-sized bird also given to a degree of territorial aggression but not against its own kind as in the case of fighting cocks.
Poultry management is an area of pure speculation. The basic requisite is protection from predators, particularly the fox. Perhaps the fouls were rounded up each evening and housed safely. Interestingly, in contrast to modern poultry which lay virtually all the year round, these early types lay eggs only in the Spring. Egg collection lengthens the laying period slightly but not enough to include eggs in the Celtic diet as other than a seasonal luxury. The approach might well have been not to collect eggs but allow the hens to sit and produce more birds.
Finally, with regard to livestock mention must be made of the horse. There is no doubt that the horse played an important role in the Celtic world especially with regard to the warrior aristocracy. It is most unlikely to have been an agricultural animal in the sense of working on a farm. Caesar refers in his battles with Cassivellaunus to being faced by 4,000 chariots. Numbers are always to be treated with a degree of suspicion, especially when they are referring to battles won and lost. However, given the size of the Caesarian legion, this figure is not unreasonable. The implication is for 8,000 trained war horses. To keep such a number in the field, at least another 8,000 must be in reserve in the sense of breeding stock, foals and animals in training. And this specifically in south-east England. The raising of horses, therefore, must have been a not insignificant agricultural operation. The infrastructure needed to produce such numbers argues for specialist ranches with all the problems of grazing, winter foddering, housing and necessarily breaking in and training. That they were status animals and were held in high esteem is evidenced at the very least by the chariot burials both in Britain and elsewhere in Europe. The animal itself was probably very similar to the Exmoor pony, a tough, uncompromising beast capable of carrying a man all day across rough country.
Mention was made above of coprolites or faeces as an important source of archaeological evidence for livestock. The analysis of the faeces allows insight into feeding regimes. It has proved possible, for example, to prove that both hay and leaf foddering, including twigs of hazel and alder, were used as early as the Neolithic in Switzerland. This kind of evidence has implications for the way in which the total landscape was employed.
Farming is, by definition, a system devised to produce a reliable and organized food supply throughout the year. With regard to plants it involves the growing of essentially storable foodstuffs, fruits which can be dried and kept in reasonable condition for at least a year. For human consumption these broadly comprise cereals, pulses and legumes. The maintenance of livestock for food as well as other products requires similar attention for the provision of fodder with virtually the same rules. The material must be capable of being dried and stored successfully, this time for a minimum of six winter months, i. e. hay, some cereals, straw (especially barley and oat straw and leaf fodder). Given all the archaeological evidence for the prehistoric Celtic period, it is certain that the Celtic farmer not only grew all these products and maintained a healthy herd of livestock, but that he did it remarkably successfully. One suspects the real economic reason for the Roman conquest of Britain in the first century AD was the agricultural wealth of the country.
Celtic farmsteads and farmhouses present us with yet more difficulties inasmuch as what could be described the average, the typical for any region, has yet to be established. A considerable number of enclosures have been excavated in Britain, ranging in size from great hill-forts or hill towns of many hectares to small banjo-shaped enclosures of less than a hectare. These latter, the small enclosures, are the target sites in that the few that have been examined carefully are usually associated with traces of field systems and often, though not invariably, contain elements of what one might expect of a farmstead. The problem lies in the size of the sample, which is too small to allow generalizations. Ironically, in 1993 at Lavant in West Sussex, in the shadow of the Trundle hill-fort a totally unenclosed group of several iron age round-houses and four - and six-post structures were discovered prior to the extension of a reservoir. The site extends beyond the limit of the excavated area, so further research is planned. The nature of the evidence, in fact, comprised the bases of postholes, arguing for an overburden of some 450 mm of topsoil and therefore the greatest percentage of the evidence will be earthfast. The disturbing aspect of this particular site lies first in the lack of an enclosure ditch, a feature that is likely to be picked up in aerial photographs, and second, that the evidence lay in the soil overburden. Identification of such sites by present prospection methods is virtually Impossible. A major area survey of the region around the Danebury hill-fort in Hampshire is currently in train following the intensive excavations of the hill-fort itself. The objective is to determine the nature of the feeder landscape for the hill-fort where considerable provision for grain storage in the form of pits and four-post granaries were identified (Figure 11.5). If the typical feeder farms were unenclosed sites like that of Lavant, the difficulties of executing such a survey so that it has real significance have been immediately compounded if not made insurmountable. Logic would suggest that within the purview of a major powerful site like the Danebury hill-fort farmers might well have dispensed with any enclosure ditches and even perhaps have initiated an early form of monoculture in response to supply and demand, some perhaps specializing in cereal production where enclosure ditches were not needed, while others concentrated upon livestock where ditches and banks provide valuable stock control elements. The normal enclosure ditch is usually 1.50 m wide and 1.50 m deep with a V section. The bank is made from the upcast material and most likely surmounted by a wattle or living fence. Such a ditch can hardly be regarded as a significant military defence of any kind and is best regarded as a system of livestock control which has even survived as a recommended system into the nineteenth century.
Although it is virtually impossible to identify archaeologically the typical farmstead, there is an abundance of evidence for Celtic houses. In contrast to the prehistoric long-house found on the Continent, the Celtic houses of Britain and Ireland are traditionally round. This particular feature, a round house with a conical thatched roof, has unfortunately led to the rather dismissive description of such dwellings as ‘huts’ and, given the normal walling material of wattle and daub the description worsens to ‘mud huts’, a definition which is belied by the sheer scale and intricacy of some of the houses. Construction materials, in fact, vary considerably
Figure 11.( Two four-post overhead granaries under construction.
(Copyright: Peter Reynolds.)
According to the region and range from dry-stone walled houses in Cornwall, Wales, the Cotswolds and Scotland to plank walls and wattle and daub in other areas and even chalk walls in southern England. However, the reality of such houses needs to be fully appreciated if the description of mere ‘hut’ is to be dispelled.
In genera] terms there are three basic forms of house construction revealed by excavation and, indeed, tested by empirical constructs. The evidence, with rare exceptions, is normally in negative form in that all that is found are the stake holes, postholes and foundation layers of stones or chalk blocks. The simplest form of round-house is evidenced by a single ring of stakeholes, the doorway only being distinguished by a pair of postholes. The regular occurrence of daub fragments, occasionally burned and thus preserving the impressions of wattles, argues that such houses were made of a wicker wall in the form of a circular basket, the break for the doorway comprising two major posts surmounted by a lintel mortised and tenoned into place. The doorway, in fact, has to be substantia! to counteract the outward thrust exerted by the interwoven wall. The height of the walls of such houses is to a large extent conjectural although an experimental construct of the second type of house discussed below indicated a height of i. jo m. The waterlogged remains of an Early Christian round-house in Northern Ireland supports this estimate. In practical terms such a height obviates unnecessary stooping within the building. Of particular note is the sheer strength of this type of wall, especially when newly built.
Although the component elements are themselves relatively weak, the stakes average 80-100 mm while the hazel rods or willow withies at the thickest point are no more than 25 mm in diameter; once woven into place, the opposing tensions create an extremely powerful structure. Over time the wattles dry, become brittle and lose their strength but the power of the wall now lies in the brittle strength of the daub which is plastered into the wattles both inside and out. Daub itself is a specific amalgam of 30 per cent clay, 60 per cent earth and 10 per cent straw, grass, hair or any other fibrous material. Initially it is mixed with water to apply to the walls. Gradually it dries out and provided the mixture is correct, there is little cracking and ultimately the fibres both hold it together and reinforce it. It is not unlikely that a lime wash was finally applied to give a waterproof and, incidentally, an attractive finish. The roofs of such houses described by Caesar as thatched perforce have to have been conical. The other alternative of a domed roof, inspired by the native houses of Swaziland in South Africa, is most unlikely given the average rainfall in Britain. The Swazi houses leak abominably when it rains. There is unfortunately no archaeological evidence for roof construction but a cone presents only a limited number of variables. The greatest problem is offered by the peak or point of the cone in that only a certain number of rafters can actually form it. If too many meet at the apex the point of the roof is lost in a jumble of timber and becomes impossible to thatch. In addition, because a thatched roof has to have a minimum pitch of 45° and a maximum pitch of 55°, then there is a tendency for the rafters to sag along their length under the weight of the thatch. A device which is critical to counteract any potential sag is a ring-beam made of hazel rods set one-third down the slant height of the roof. This also serves to support the supplementary rafters which make up the rest of the cone. All the rafters are secured in place by concentric rings of hazel rods tied to each rafter. These are correctly determined as purlins, since they are contructional and physically hold the cone together. An equally strong alternative is to interweave the rafters with hazel rods, creating a conical basket. The final effect is to convert any lateral thrust exerted by the rafters on the wall stakes, to which they are simply notched, into vertical thrust. All the weight of the roof including the thatch is directly downwards onto the wall. This type of house has, therefore, the same life expectancy as the walls of the house. Once the wall deteriorates the building will collapse. How long that should take is difficult to determine. There is no real reason why such a building should not last many decades provided the thatch is replaced at regular intervals. The type of thatch rather dictates its own lifespan: wheat straw, for example, lasts usually for fifteen years or so before it needs either to be replaced or another layer applied, river reed (commonly known as Norfolk reed) can last as long as eighty years, and ling or heather forty years or more. In none of the excavated examples to date is there a central post to hold up the roof. Where one would logically be found is the normal location of the hearth. The size of this type of house ranges from 4 metres to 9 metres in diameter. However, to put this into a more comprehensible context, the floor areas range from 12.6 square metres to 63.6 square metres. An average modern house has a ground-floor area of c.54 square metres. There are many perfectly adequate houses with smaller floor areas.
The second major type of Celtic round-house is widened by a double ring, an outer ring of close-set stakeholes and an inner ring of more widely set substantial postholes. Usually on the south-eastern quadrant is an arrangement of postholes suggesting the presence of a porch, the width of the doorway being twice the depth of the porch. This suggests a pair of doors which swing back into the porch flat against the walls. These houses range in size from to metres to over 15 metres in diameter, with respective floor areas excluding the porch of 78 square metres to over 180 square metres. The latter would accommodate one and a half average modern houses!
World History
