The use of tree-ring dating has increased rapidly over the last two decades, and enough results have now been published to undertake an assessment of what has been learnt, to review the present position, and to consider future directions. The paper seeks to identify a number of issues currently concerning both dendrochronologists and those who use the data. The benefits of accurate dating are considerable. Precise dates are not only important for understanding individual buildings, but are leading to the reassessment of both regional chronologies and the development of construction and technique. It is suggested that we may now be in a position to start identifying general trends in the incidence of buildings of different types and status over the country as a whole. At present this kind of analysis can only be tentative, but in the long run it will enable us to place buildings in a wider historical context, and help us to cross the chasm between documentary historians and those who study buildings.
Tree ring dating is a technique with which most of us are now familiar, whether or not we have been in a position to use it and benefit from the results. Dates for British buildings have been published annually in Vernacular Architecture since 1980, and have also appeared elsewhere (Footnote 1). Enough has now been done for an attempt to be made to assess what has been learnt so far, to review current concerns and to make suggestions for the future.
In 1984 an unpublished report on dendrochronology to the CBA Historic Building Committee (Footnote 2) noted that over the previous fifteen years tree-ring curves had been established for several areas of the British Isles, and over 100 buildings or parts of buildings had been dated. However, much more needed to be done. At that time no curves had been successfully established for the south west, the south east, or East Anglia. Work by the laboratories had largely concentrated on three issues: on the statistical theory of sampling, the verification of the basic method, and the establishment of model curves. Very little sampling had then been undertaken for the purpose of architectural dating, and at that time only RCHME seemed likely to be in a position to commission such work. In addition, the collection of samples was not always combined with rigorous examination of the structures from which they were taken, which invalidated some of the results, while publication of those results was often inadequate.
While things can no doubt still be improved, we have certainly come a long way in the last thirteen years. Curves have been established for new areas and dates for 650 or more buildings have been listed in Vernacular Architecture alone - although it has to be said that others have not yet appeared. Rigorous correlation of samples and structures is increasingly taken as a matter of course, published results are far more informative, and the technique is starting to become available as a tool for building recorders of all kinds. It is used to advance knowledge of both individual structures and groups of buildings and, perhaps most importantly of all, is beginning to further our understanding of wider historical questions.
Many buildings have been selected on an individual basis, but since the mid 1980s a number of large-scale projects have got underway. Of these, only the conclusions for RCHME in Kent, and for RCAHMW in Wales, have yet been published (Footnote 3), although papers have been delivered based on the work of the Leverhulme cruck project in the Midlands, the Shropshire dendrochronological project, and the RCHME northern roofs project, and the dates of the individual buildings from these projects have been published in Vernacular Architecture. Without a full understanding of each area and its buildings, or of what was in the minds of those who commissioned the work, this wealth of material can only be given very summary treatment; it is too soon to draw firm conclusions. What follows is a preliminary and tentative assessment, indicating potentially interesting advances, which will require further research to substantiate. This paper concentrates on work on vernacular buildings, and does not discuss the results of important work on ecclesiastical roofs, particularly on the roofs of cathedrals ( Footnote 4).
The possibility of having precise dates for the timbers of standing structures is a heady inducement to building recorders to seek the help of dendrochronology in interpreting their material. But obtaining precise dates is not always simple. There are a number of reasons why timbers will not date, some of which are in the process of being overcome, but some of which may remain recalcitrant. Before discussing the results published so far, it is worth looking at some of the problems which can be encountered.
As the discussion below of geographical coverage suggests, climatic conditions mean that it has proved very difficult in certain parts of the British Isles to establish the local master chronologies that are necessary to date buildings. Dating in regions such as the south west (Footnote 5) and East Anglia has been notoriously difficult, although progress is now being made (Footnote 6). In these areas local climate patterns, or possibly the effects of woodland management, mean that the way forward may be through producing local chronologies for relatively small regions, and building outwards from them. This was the case in Essex where the breakthrough came from intensely detailed work on the Cressing Temple site; and in Devon, where work has concentrated on a small area north of Crediton (Footnote 7). Another very localised ‘ecological niche’ has been recognised around Durham ( Footnote 8).
Dating is done by cross-matching, or correlating, the pattern of ring-growth in a particular tree with a previously established master curve, possibly obtained from trees in another area. The quality of match is expressed in terms of ‘t-values’, with high t-values - say at least 5 - giving more cause for confidence in the date than low ones. Most people are aware that a certain number of rings are necessary in order to obtain a reliable date. However, there is some disagreement among dendrochronologists as to what the smallest number of rings may be for a reliable match. Some believe that 80 or more rings of surviving annual growth are required for a result; others that reliable results can be got from as few as 50 rings. In some places trees grew so fast that timbers were used which have far fewer than 50 rings surviving, and at present such timbers are unlikely to date. The question of whether such trees can be dated is one which is currently being debated and researched by the professionals. Those whose background lies in statistics take a more positive view of the potential for success than those who are primarily archaeologists or natural scientists. Such discussions are beyond the sphere of most clients but, since they can affect the results, it is useful to know that they are taking place.
One problem that is being encountered is that in some regions, timbers of certain periods simply will not date. As the results published in Vernacular Architecture indicate, a relatively large number of late thirteenth and early fourteenth-century buildings have been dated from many parts of England. But not from Kent. In that county buildings which are almost certainly of this period have so far proved to be undateable, even when there are a considerable number of rings (Footnote 9). Similar situations occur elsewhere. In part the answers may be related to climate, but there are also theories that it may be due to woodland management. The problem does not concern the period of felling, but rather the period when the trees germinated. A twelfth-century gap in the germination date of trees in the Midlands has led the Nottingham laboratory to set up a project specifically to examine Sherwood Forest oak (Footnote 10). In other laboratories the same issues are being looked at during the course of other work. It has also been suggested that the way trees were grown in the fourteenth century affected the quality and availability of fast-grown timber in the first half of the fifteenth century ( Footnote 11), although the last section of this paper suggests that social and economic explanations may also be likely causes for a relative dearth of certain types of standing buildings in this particular period. However, the whole question of the way trees were grown is one which is directly relevant to all dendrochronology. Some research has already been undertaken on cross-matching timbers from living trees, where ecological and genetic variations can be studied in a controlled manner (Footnote 12). When more research has been done the results are likely to be of considerable significance for those who study standing structures.
Further complications arise in regions where much of the building timber was elm or other wood. Elm was frequently used in Somerset in both medieval and later buildings (Footnote 13); it was used in medieval buildings in Warwickshire (Footnote 14) and in parts of Kent, notably in the north east of the county where timber was scarcer (Footnote 15), and after the Middle Ages is found in abundance in the Cotswolds, the Chilterns and parts of the Midlands (Footnote 16). As yet, chronologies have only been established for oak, and there are differing views as to whether these can be used for dating elm, which anyway tends to be fast grown and to have ‘complacent’ ring patterns which are less easy to date. Several laboratories are looking into the problems associated with dating both elm and beech, and English Heritage is currently funding work on conifers, which will be relevant primarily to post-medieval buildings. So far, in some areas samples from other types of tree have cross-matched with oak, while elsewhere they have not. Thus the dating of timber other than oak is a subject for future research.
A good match with a high t-value against a local master will date all the rings present in the timber, and in particular will give the date of its last ring. However, this last ring may not be the last ring of the tree from which it came, since some of the soft, outer, sapwood rings may have been removed by the carpenter or lost over the years. The felling date of timbers which have bark remaining can be precisely ascertained, whereas the felling dates of those which have only sapwood, or the heartwood/sapwood boundary, can only be given an approximate date, the degree of certainty depending upon the number of sapwood rings remaining and the accuracy with which the likely number of sapwood rings has been calculated. Current estimates of these vary, with some researchers proposing that it is 95% certain that sapwood rings range between 10 and 55 per tree (Footnote 17), and others that there are more likely to be between 15 and 50 sapwood rings (Footnote 18). The question of estimation, and the presentation of such results for public consumption, is discussed in more detail in Dan Miles’s paper below. It is important because estimated felling dates are usually clear guides to construction dates, and are therefore critical to the analysis of the results of the last few years.
The situation is further complicated by the fact that trees do not all have the same average number of sapwood rings. In certain areas, in Brittany for example, and in Kent, it has been shown that there were fewer than in some other regions. Calculation of the number, and therefore estimation of the felling dates of timbers, may need to be refined as work in an area proceeds. This is precisely what occurred in Kent. At the start of that project the number of sapwood rings in trees was estimated as thirty, within a range of fifteen to fifty for 95% of all mature trees (30, -15 + 20). By the end of the project, the evidence from firmly dated timbers suggested that the number should be revised to twenty five, within a range of fifteen to thirty five (25, -/+10). The effect was to pull all the estimated dates back by 5 years. If one is going in for accurate dating, this matters. In the meantime the earlier dates had been published and may well be inaccurately repeated (not surprisingly, they crept into the first draft of the database used in the preparation of this article). In Brittany, two sets of sapwood estimates have been calculated, the first on the basis of the historical material, the second using good evidence from modern trees. Dates calculated by each method are different, and the dendrochronologists are reluctant to throw their weight wholly behind one set or the other (Footnote 19). It is likely that as more work is done, local variations will increasingly be recognised and taken into account when dates are calculated.
One view is that where felling dates are not precise, that is, where bark or evidence of bark is not present on a sample, only date ranges should be given. But even then people may, for the purposes of calculation, turn that range into a single, notional date in order to create tables of the sort illustrated later in this article. Since the most likely felling date does not necessarily lie in the centre of the range, calculations made by the ignorant will be less accurate than calculations made by the professionals themselves. These are tricky issues which are currently engaging the attention of dendrochronologists. Those for whom dates matter need to be aware that there are very real problems to be resolved, and our eagerness to accept precise dates which may not be correct is a worrying concern.
Large-scale projects and individual sampling have between them covered considerable parts of the country but, as Figure 1 indicates, there are still areas of Britain where few results have so far been obtained.
Location of dated buildings, published in Vernacular
Architecture , 1980-1996. [GB Coastline Copyright Bartholemew
Database]
Hardly any buildings have yet been dated on the
west coast, from Scotland to Lands End. Although not shown on this map, this
lacunae extends also to the Channel Islands, where only one building has so far
proved dateable (Footnote 20). The
situation in the west is improving, but still has a long way to go. Much of the
dating in eastern Wales and Shropshire has taken place in the last three or four
years, successful sampling in Cornwall and Somerset has only really begun in the
last two, and the Sheffield laboratory is currently devoting considerable effort
to building a chronology for Devon. The situation on the eastern coast of
England is much the same. Very little work has been done successfully in the
north east, although this has recently begun to be rectified through projects
devoted to truncated principal roof trusses (Footnote 21) and
the buildings of religious estates in Durham (Footnote 22).
Further south, little has taken place in Lincolnshire, other than in Lincoln
itself, while in East Anglia less than half a dozen buildings have been dated in
Norfolk and Suffolk. Although a number of results have recently been obtained in
western Essex (Footnote 23), the
prevalence of timber-framed buildings in that part of the country means there is
obviously a great deal more to be learnt. As Figure 1 shows, the bulk of the
results obtained so far are concentrated over a wide area of the Midlands, in
part of the Thames valley, in Kent, and latterly in Hampshire.
In the early stages, in order to establish local chronologies, it may be important to get buildings dated, no matter what kind of buildings they are (Footnote 24). So the first buildings to be dated in a region are often of limited value in sorting out architectural chronology. As more buildings are dated, dating becomes easier, and the selection can be more refined. But as long as the number of buildings dated in the whole of the British Isles remains limited, there are bound to be biases. Work has been funded by researchers with specific questions in mind, and their attention has frequently been directed to buildings of a particular type or date. This has produced a preponderance of dated crucks in the Midlands, of a particular roof type in the northern counties, of early barns in Essex, and of rural medieval houses in Kent. To the results of these and other projects can be added a growing number of results from ‘one-off’ sites, which are making a welcome extension of coverage to buildings of all kinds and dates. Nonetheless, anyone using the published dates needs to be aware of the biases that are presently built into them.
A further point to be noted, as the maps in Figure 2 indicate, is that most of the buildings date from the Middle Ages.
Maps showing the chronological distribiution of dated buildings, published in Vernacular Architecture, 1980-1996. [GB Coastline Copyright Bartholemew Database]
There are a number of reasons for this. Firstly, the desire to sort out the early chronology of timber-framed construction, eg. aisled and base cruck buildings, has led to a concentration of effort on the earliest surviving structures. The fact that buildings of the twelfth, thirteenth and fourteenth centuries are, relatively speaking, extremely well-represented on the maps, is due to this research aim. Secondly, the large-scale research projects which have used dendrochronology have almost all been concerned with medieval themes. This is hardly surprising, for it is clear that the value of tree-ring dating in the Middle Ages, when documents, date stones and stylistic detail are scarce, can hardly be over estimated. For the first time ever, researchers have the ability to date their subjects with some confidence. It is noticeable that the majority of seventeenth and eighteenth-century dates are clustered in parts of the north east Midlands where more work has been initiated and funded by local authorities, whose agendas are perhaps not confined to pure historical research. Finally, medieval buildings were largely built with virgin timber, whereas later buildings often reused earlier material, which means that successful tree-ring dating of later buildings may be more patchy.
In a number of cases dendrochronological dates have been found to coincide very closely with documented dates for building. Such confirmation of the accuracy of the technique is immensely reassuring for all who use tree-ring dates. A good match occurred, for example, at the Abbots House, Shrewsbury, where precise felling dates of 1457 and 1458 were obtained, and in 1459 a building ceremony was recorded, attended by the Abbot of Lilleshall, his carpenter and borough officials (Footnote 25). In Hampshire, the barn of Overton Court, belonging to the Bishop of Winchester, was erected with timber felled in 1496, while the house was built with timber felled in 1496, 1504 and 1505. These dates gave the researcher, Edward Roberts, an indication of where to search in the accounts, leading to the discovery that timber was acquired both in 1496/7 and in 1505/6, while the building of the barn and house were respectively documented in 1497/8 and 1506/7 (Footnote 26). This is one of several instances in Hampshire where tree-ring dating facilitated documentary research, leading to significant advances in understanding the buildings.
Even when timbers have been more tentatively dated than in these cases, they can tie in well with documentary evidence. In Canterbury, it is known that the Sun Inn in Burgate was under construction in 1437-8 ( Footnote 27). Tree-ring dating provided no precise date, but a range of 1425-60, later revised to 1425-45 (Footnote 28) . Within the range the most likely felling date was first estimated as c1440, later as c1435. The date ranges encompassed the documented date, and both the estimated dates within the ranges were close, but it is somewhat heartening to find that the revision brought it to what must be about the most probable felling date on grounds other than dendrochronology. This kind of evidence inevitably gives the user confidence in the methods of estimation.
Dating of individual structures, often undertaken as part of the restoration programme of large and important buildings, has contributed significantly to the understanding of those buildings, and often enabled them to be placed more accurately in a wider context. Many of these projects have been funded by the owners of the sites, or by English Heritage, Cadw or local authorities. The contribution official bodies have made to the understanding of the buildings in their care, or upon which they are giving advice, is considerable. In Nottinghamshire, for example, relevant buildings are now recorded and tree-ring dated as a regular part of County Council conservation casework. The results often lead to a building being upgraded, and therefore eligible for grant aid. They have contributed to an increased understanding of the county’s heritage, which is of considerable value to conservation officers as well as to historians, and they have helped to interest the general public in historic buildings, often raising funds for restoration in the process ( Footnote 29).
An early but notable result from sampling individual buildings with the help of local organisations came through work on the timbers of the great hall in Leicester Castle, whose structural history was complex and ill-understood (Footnote 30). Although no precise date for the early work was obtained, due to the absence of sapwood, it became clear that the original aisled hall was built in the twelfth century but was wholly reconstructed in the sixteenth century, retaining only the arcade posts. The revised comprehension of the structure made it possible to place the hall in the milieu of other early halls, such as those at the Bishop’s Palace, Hereford, and Burmington Manor and Temple Balsall in Warwickshire. All these buildings have now been tree-ring dated, with greater or less degrees of precision, and gradually our understanding of the development of these early buildings is advancing.
Tree-ring dating has likewise helped to clarify complex situations in later buildings whose dates were the subject of debate, and sometimes also to pinpoint the likely builders. Examples are Gainsborough Old Hall in Lincolnshire, of the 1460s and 1480s, Oakwell House, Birstall, in West Yorkshire, of the 1580s, and parts of Daneway House, Bisley, in Gloucestershire, of 1674, where a conflict of views over the dates of plasterwork and panelling was resolved by dating the timberwork (Footnote 31). In some cases tree-ring dating has simplified what appeared to be a complex sequence of building; so that what was thought, on structural evidence, to be of two dates, has been shown to be of one, even if constructed in two phases. Charlton Court Barn, Steyning, in West Sussex, of 1404-6, and Tickenham Court, Clevedon, in Avon, of 1471-6 are both examples of this. In other cases, as in the former infirmary of Wherwell Abbey in Hampshire, what had been assumed to be one phase turned out to be two (Footnote 32). While dating major buildings is valuable in terms of the immediate work in hand, the results also contribute to our general understanding of architectural development. This cumulative effect will become increasingly apparent as time goes by.
A recent development in the application of tree-ring dating has been the sampling of small features, such as partition staves. Small timbers often have as many, or even more, rings than larger ones, and at Three Chimneys in Mapledurham the question of whether a partition was original or not was resolved in this manner. Another potentially important development is measurement by specialised photography. This has been used for some time for prepared end grain; but the reading of rings from a radial section on the face of a plank or beam is now being explored, for example at Stokesay Castle. At the moment such work is very experimental, but if it can be developed further, then it may become possible to date important parts of buildings which cannot be cored in the normal manner (Footnote 33).
Nearly everyone who has experience of tree-ring dating will have found themselves in for some surprises. Buildings which one felt reasonably confident in dating, turn out to be later or earlier than expected. With a single building it may be difficult to turn this new knowledge to constructive use, but with a group of buildings it can affect one’s picture of chronological development and enable one to look at one’s material in a new way. It is perhaps invidious to pick out examples, for the phenomenon is widespread. Certainly, in my work in Kent I initially made mistakes in dating, and learnt a great deal from being able to initiate a substantial tree-ring dating programme which allowed the construction of a framework for dating buildings throughout the Middle Ages. As a former member of the RCHME Threatened Buildings team I, like my more illustrious seniors, was a late dater. This no doubt arose from the experience of working across England, of knowing a little about many areas, but few areas in depth. An example from Kent is Clakkers Hall, Plaxtol, published in English Vernacular Houses as c1500, but now known to be built of timbers felled between 1442 and 1462, with a likely felling date of c1452 (Footnote 34).
One interesting point to emerge from the lists in Vernacular Architecture is that a number of urban buildings, once thought to date from the fifteenth century, were built of timber felled at a much earlier period. The Cross Keys Inn, Leicester, dates from after 1309, probably c1334 ( Footnote 35), while ‘Severns’, formerly in Middle Pavement, Nottingham, is now firmly dated to 1335 (Footnote 36). Assuming the buildings were erected shortly after the felling dates, these are substantial and significant revisions, and the whole question of urban dates will be touched upon again in a later section of this paper. A look at the lists also indicates that larger buildings in Shropshire have produced some unexpected dates. On the one hand, the Prior’s Hall in Much Wenlock, generally considered to date from c1500, has been shown to have been erected with timbers felled in 1425. On the other, some earlier halls such as Upton Cressett Hall and Moat House, Longnor, previously thought to date from the late fourteenth century, are now known to have been built with timbers felled only in the second and third quarters of the fifteenth century ( Footnote 37). Thus the timbers of buildings which had been considered to be a hundred years apart were in fact felled at more or less the same time, and the sequence of their dates has even been reversed. Whether the felling date for the timbers of the Prior’s Hall is the date of building may be another question, and the conclusions of the researchers must be awaited with interest. If so, this kind of redating may lead to a new view of the development of medieval building in the county. Shropshire is one of the few counties where medieval houses of all kinds and of varying status have been sampled systematically, so the conclusions will be particularly valuable. While results of this sort may mean a revision of views on chronological development in some areas, it will be less obvious in regions where attention has focused on a narrower range of building types, and likely to be most marked where research and publication have been going on for a long time. In areas where intensive research is more recent, changes in dates will probably be less dramatic, since the lessons of tree-ring dating, even if applied from results in other regions, are likely to form part of the initial analysis.
It would be wrong to give the impression that all dating is in the melting pot and that all chronologies must be revised. In Shropshire, as elsewhere, tree-ring dating has in many cases confirmed dating done by traditional methods of stylistic and structural analysis. As Christopher Currie remarked recently, after thirty years’ research and a tree-ring dating programme in Oxfordshire, his conclusions remain much the same (Footnote 38) . This does not imply that much of tree-ring dating is a waste of time. On the contrary, the fact that previous dates were obtained by less scientific methods means that it was not possible to treat them with the confidence that dendrochronology inspires. If buildings are to be used to further historical research, then accurate dating is critical, and confirmation of previous dating is just as important as correction.
The generally accepted chronological development of early timber-framed structures has been largely confirmed by tree-ring dating, although the tendency has been to push some of the smallest early survivors back in date. However, the results of the only large-scale programmes of tree-ring dating specifically aimed at studying construction: the Leverhulme cruck project in the Midlands and the RCHME truncated principals project in northern England, have not yet been published. Until that time it is not really possible to assess the full implications of what has been learnt from extended programmes of this kind, and a discussion of what dendrochronology has to teach us about structure may be premature.
One important aspect of construction which is likely to be considerably advanced through tree-ring dating is the typology of timber jointing techniques. Pioneered by Cecil Hewett, and based upon his extensive knowledge of early Essex buildings, later augmented by analysis of roofs elsewhere, the evolution of jointing techniques has had a significant effect on the understanding and dating of timber buildings throughout England. While most researchers now accept the broad outlines of Hewett’s typology, doubts have been expressed about the detailed chronology and close dating of certain features, and therefore of some buildings. Just as tree-ring dating is clarifying and refining the dating of buildings generally, so it is likely to refine the date ranges of jointing techniques. Since many of the key buildings lie in Essex where dating has proved d One important aspect of construction which is likely to be considerably advanced through tree-ring dating is the typology of timber jointing techniques. Pioneered by Cecil Hewett, and based upon his extensive knowledge of early Essex buildings, later augmented by analysis of roofs elsewhere, the evolution of jointing techniques has had a significant effect on the understanding and dating of timber buildings throughout England. While most researchers now accept the broad outlines of Hewett’s typology, doubts have been expressed about the detailed chronology and close dating of certain features, and therefore of some buildings. Just as tree-ring dating is clarifying and refining the dating of buildings generally, so it is likely to refine the date ranges of jointing techniques. Since many of the key buildings lie in Essex where dating has proved difficult, it has not until recently been easy to apply this test. Where results have been obtained, the dates of some in the typological sequence have been confirmed. The dates of the barns at Cressing Temple, for example, are only slightly later than those suggested by Hewett: the first phase of the Barley Barn, previously thought to have been erected in the late twelfth century, is now known to have been constructed between 1205 and 1235, and the earliest phase of the Wheat Barn, previously dated to c1255, now dates to c1257-80 ( Footnote 39). In both cases the dates have shifted later, although not by any appreciable amount. Other buildings, however, have proved to be considerably later than was originally thought: it is now known that the stave church at Greensted, Essex, and the ‘rhenish helm’ spire at Sompting church, Sussex, both thought to be pre-Conquest, were built after 1063 and in the early fourteenth century respectively (Footnote 40), while the earliest phase of the barn at Belchamp St Paul, is not eleventh or twelfth century, but 1240-75 (Footnote 41). The importance of redating these and other buildings lies in the fact that they formed part of a closely argued typological sequence with chronological implications, and the redating has reversed the order of some examples. The main outlines of the sequence, which is founded upon detailed knowledge of the structures, is likely to stand, but the longevity of particular jointing forms may make a difference to the chronological profile of the typology. Its importance and influence is such that, when more dates have been obtained, a major re assessment considering the implications of the changes should be undertaken. This will almost certainly improve our perception of the development of medieval construction.
The chronological implications of the typology of aisled halls, base crucks and crucks were never as closely argued; so, while we may learn a great deal about their development and distribution, this is less likely to require a major reassessment of received views on dating. That aisled halls survive from an earlier date than base crucks has long been suggested, and both this and the fact the earliest surviving aisled buildings, dating to the late twelfth or thirteenth centuries, are as likely to lie in the Midlands as in the South East (where they might once have been expected), are confirmed by tree-ring dating (Footnote 42). Examples of both base crucks and crucks date from the middle years of the thirteenth century, with the base cruck at Siddington barn in Gloucestershire dating to 1245-7 (Footnote 43), and two cruck blades from Upton Magna in Shropshire producing a date of 1269 ( Footnote 44). These dates are possibly a little earlier than was previously thought, but documentary dates for crucks of some sort, as we have recently been reminded (Footnote 45), are known from yet earlier in the thirteenth century. In the past it has been suggested that one type of cruck structure developed from the other - different researchers holding different views of the sequence. So far, the earliest dates for the two types are so close that if one developed from the other this must have occurred well before the first surviving examples. Thus, until further analysis is undertaken by those more qualified than the present writer, the question of origins remains as obscure as it was in 1981 ( Footnote 46). Tree-ring dating has, however, confirmed that the bulk of surviving base crucks date to the fourteenth century, while cruck numbers peaked in the mid fifteenth, and continued in some regions throughout the sixteenth. But before any really constructive comments can be made it is essential that the geographical location, function and social status of these structures are considered (Footnote 47). The extent to which other aspects of the development of cruck construction, such as the typology of apex forms, have been advanced through the recent dating programme must await publication of the results.
The truncated principal roofs project is concerned with dating a range of roof types which occur in the north, and particularly the north-east, of England between the mid fourteenth and the seventeenth centuries. The earliest examples are combined with crown posts, the later are purely of side-purlin construction. Such roofs are often difficult to date from stylistic evidence, and since they are of varied form, it was thought worthwhile to try and trace their evolution. This has now been achieved, and the results should soon be published (Footnote 48). Hopefully, the information will enable roofs of similar construction to be dated more accurately. The roofs are used over buildings of differing type and status, and it seems likely that further advances in knowledge may be made by not only considering the type of roof, but also the status, location and function of the buildings in which the roofs occur. Thus, as in the case of crucks, the initial analysis of results will almost certainly require more contextual research for the full benefits of tree-ring dating to become apparent.
In a number of areas enough work has now been done for comments to be made about what has been discovered. On the whole the results have not been startling, in the sense of overturning previously accepted theories. For this we must be thankful: it is clear that most regional recorders already had a good grasp of typological development, and indeed also of the general outlines of regional chronology. Tree-ring dating has enabled this to be substantiated and refined.
One of the few dramatic findings relates to the work undertaken on agricultural or service buildings belonging to religious estates in and around Durham (Footnote 49). Seven buildings, which have none of the architectural detail required by traditional methods of dating, have now been tree-ring dated. The result has been to all but double the number of known late-medieval barns in the North East, and provide a framework which may be used to analyse other structures in an area where medieval buildings are rare and extremely difficult to recognise.
In Wales, RCAHMW initiated a programme to date seventeen phases on eleven sites in the north and east of the principality (Footnote 50). The aim was to date a range of timber houses, of varying types and social status, between their first appearance and the end of the Middle Ages. The dates fall between the 1430s and the mid sixteenth century (with later dates for inserted ceilings). Not only has a framework been established which will enable local researchers to date other Welsh structures with more confidence, but Welsh buildings can now be compared more readily with dated English ones. For example, base crucks of some distinction were still being constructed there in the mid fifteenth century when they had ceased to be erected in most parts of England; and it is argued that no truly vernacular buildings survive before the mid sixteenth century. The chronological staggering of buildings according to size and social class in Wales is quite striking in contrast to what we know from elsewhere.
In Nottinghamshire, geographical and chronological patterns of dated houses have been published, without regard to size or type of building ( Footnote 51). Sampling has been undertaken on a very wide range of house types and dates within the county, with much of the work having been initiated by an enlightened County Council. It is clear that the earliest houses lie in the south and south east of the county, in the Trent valley and the towns of Nottingham and Newark. The numbers of dated buildings in the larger upland region to the north of the county are both fewer and later. It is likely that this pattern is a true reflection of the surviving buildings on the ground. Perhaps these conclusions were to be expected, but they are none the less welcome, and one suspects that those working in conservation derive major benefits from the increased understanding of the development of timber framing within the county.
Other counties where major programmes are producing results are Derbyshire, Hampshire and Shropshire, while work in Somerset and the Channel Islands is just beginning. The most beneficial results will almost certainly arise in those counties where the sample is taken from the full range of building types, ie. urban and rural, agricultural, domestic and ecclesiastical.
On a smaller scale, Dan Miles is attempting to date the majority of historical structures within the Oxfordshire parish of Mapledurham. Buildings range from the fourteenth to the eighteenth century, and the results, when combined with information from other sources, will add a new dimension to the understanding of the history of the parish.
Tree-ring dates for some 650 buildings have been published so far in Vernacular Architecture. Many of the buildings have had more than one phase dated, so that dates for over 800 phases are now known. The lists include cathedrals and churches, monastic ranges, public buildings, houses, both great and small, agricultural buildings, and a few industrial structures. The dates range from the twelfth to the nineteenth centuries, although by far the greatest number fall between the fourteenth and sixteenth centuries. Within this late medieval period there are two noticeable peaks, the first occurs in the early fourteenth century, the second and larger begins in the second half of the fifteenth, reaches a high point by the end of the century, and dies away gradually over the course of the sixteenth century. It may be argued that this chronological pattern of dates indicates little else than the research interests of the clients. Up to a point this must be true since, as discussed above, the technique is most valuable as a research tool for the Middle Ages. But in Shropshire, 62 buildings, comprising 104 phases, have been dated between 1247 and 1666, and a sharp decline in numbers in the first half of the sixteenth century is followed by a marked rise from about 1550 (Footnote 52). This suggests that the national decline in numbers noted in the sixteenth century may reflect more than just bias on the part of researchers. Elsewhere, not enough later buildings have been sampled for conclusions to be drawn. But for the medieval period as a whole the time may already have come when the body of evidence is large enough to allow a preliminary assessment of the pattern of building.
However, to make sense of trends it is essential to break the buildings down by type or by social status, and this in itself is fraught with ambiguity. Separating houses from barns or churches is simple. Separating greater houses, built by the aristocracy, ecclesiastical institutions or the gentry, from those which may be termed ‘vernacular’ is far more difficult, particularly since the relevant information may be lacking from the published results. It has, however, been attempted in the following section, which relates only to houses and public buildings. Although some mistakes will have been made, it is hoped that the resulting patterns of building are robust enough to absorb them.
Figures 3-5 illustrate three types of building, plotted in blocks of 33 years, between c1200 and 1600.
Rural Aristocratic and Gentry Houses (165 examples)
Rural Vernacular Houses (166 examples)
Urban buildings, status not differentiated (79 examples)
The choice of 33 year intervals is a compromise between the wish to have periods small enough to be significant, but not too small to introduce a spurious sense of accuracy in the case of date ranges or some estimated dates. Where a date range spans two periods, the building has been placed in the period of its estimated date, or, where no estimated date was given, of the centre of the date range. The charts have not been taken beyond 1600 because at present too few later buildings have been dated for any meaningful trends to emerge.
The first chart, Figure 3, illustrates the chronological spread of aristocratic, ecclesiastical and gentry houses. It excludes monastic ranges and royal castles, and also what were obviously town houses, but includes all private residences built by the Crown and by ecclesiastics or ecclesiastical institutions. Clearly, there is a vast difference in status between houses built for the aristocracy and those built for minor gentlemen. And the latter buildings fall into a grey area which may overlap with some of the rural vernacular houses in Figure 4. Despite the potential overlap, the patterns of the two are so clearly defined that the risk of some distortion is worth taking. Figure 5 illustrates the spread of urban buildings of whatever status. It was not deemed feasible to break these down further, since the numbers are small, and it includes public buildings, such as guildhalls, as well as houses.
The ‘supra-vernacular’ houses in Figure 3 produce the same chronological profile as that indicated when buildings of all kinds are looked at together. While a handful of houses were erected throughout the thirteenth century, the numbers begin to increase in the last third, and show a marked rise in the early fourteenth. The lack of buildings during the rest of the troubled fourteenth century comes as no great surprise. But the dated buildings suggest that aristocratic and gentry building generally did not begin to pick up again until the middle of the fifteenth century. Moreover, it should be noted that eleven out of twenty - or 55% - of the examples in the period 1434 - 1466 are from Shropshire and Wales, with all but two of the rest coming from northern and midland counties such as Cumbria, Derbyshire, Lincolnshire and Nottinghamshire. As Richard Suggett has pointed out ( Footnote 53), the earliest gentry houses dated in Wales were only built in the 1430s, a conclusion which he claims can be applied to houses other than those which were tree-ring dated. At first sight this is surprising, given that close at hand in Shropshire gentry dwellings survive from the time of the early fourteenth-century peak which is apparent in other parts of England. He accounts for it by suggesting that earlier property was destroyed in Owain Glyndwr’s revolt. This might explain the lack of fourteenth-century building in Wales, but the evidence of Figure 3 suggests that the start of the fifteenth-century boom in large houses may have been a wholly northern and western affair in England as well. In the south of England the tree-ring dates imply that the late medieval upsurge in building high status houses was deferred until the 1470s, and this is born out by other research. In Kent, very little high class building is known between the mid fourteenth and the late fifteenth century (Footnote 54), and in Hampshire there was a similar dearth and late fifteenth-century resurgence, which has been attributed to the introduction of demesne leasing on the great estates (Footnote 55). One relevant point which might be made regarding the hiatus in gentry building relates to base crucks. If it is generally accepted that base crucks appear in houses built by those of superior status, then the fact that they die out in most areas after the 1330s is hardly surprising. By the time the gentry were building again, building styles in most regions had moved far beyond that stage.
The second chart, Figure 4, shows the chronological distribution of rural houses likely to be of below gentry status. A few buildings which may not be gentry houses survive from the thirteenth and early fourteenth centuries. Apart from a single example in Shropshire, they currently lie in southern England, primarily in Oxfordshire and Berkshire. There is no evidence for an early fourteenth-century peak, but clear signs of an increase in the last third of the century, continuing into the fifteenth. This is in marked contrast to the pattern of the known gentry houses, and seems to indicate a change in the circumstances of a number of peasants, perhaps at the expense of the landlords. In discussing this inference from the tree-ring dating of buildings, Christopher Dyer has related it to changes in prices and wages, and a renegotiation of the relationship between peasants and lords during and after the period of the Peasant's Revolt (Footnote 56). So far the dates show that the late fourteenth-century upsurge in vernacular building was largely confined to the south east, and that building in the rest of the south and the midlands followed only in the early to mid fifteenth-century. Surviving vernacular buildings in the north have only been dated in the second half of the century, and in Wales they apparently do not occur until well into the sixteenth. Wales apart, there is a massive increase in fifteenth-century rural houses, beginning in the middle years of the century, the tree-ring dating evidence suggesting that prosperity for a substantial number of peasants may have taken off rather earlier than historians have often supposed ( Footnote 57). The fact that Figure 4 shows a slight decline in dated buildings during the sixteenth century may simply be a reflection of work done, or it may be a general indication of the trend noted in Shropshire. It will be interesting to see what the national picture is like when a more truly representative sample has been obtained.
The chronological pattern of urban buildings is plotted in Figure 5. Since the numbers are fewer no attempt has been made to separate polite from vernacular, or even to distinguish houses from public buildings. When more urban examples have been sampled this should certainly be done, and at present aristocratic town houses, guildhalls and large inns form a high proportion of the dated buildings. No urban buildings have yet been dated before the last third of the thirteenth century, but when they appear they increase quickly, and are by no means all of high status. In Hampshire the earliest small houses so far dated are urban rather than rural, including one of 1292/3 and others of 1300,1335 and 1340 (Footnote 58). Despite the small size of the sample, the numbers seem to increase significantly in the early fourteenth century, and in contrast to gentry building they continue to be built well into the middle of the century. Unlike rural buildings, there is far less disparity between north and south, and the urban buildings plotted from the first two thirds of the fourteenth century include examples from midland and northern towns such as Chester, Leicester, Ludlow, Nottingham, Shrewsbury and York.
In a manner similar to high status rural houses, there seems to be a falling off of urban building in the later fourteenth century, but the following two periods, from 1400 to the 1460s, have produced a substantial crop of dated examples. Thereafter, the numbers drop, and never recover. Regional variations in economy are likely to produce different patterns in different towns, and there are still too few examples spread over the whole country for any but the most cautious suggestions. However, these first results make one wonder whether tree-ring dating will not allow buildings to play a much more significant part than hitherto in arguments concerning urban prosperity and decline in the later Middle Ages (Footnote 59) . So far, and in marked contrast to the chronological spread of rural houses, the evidence suggests there was a great deal of urban building at all levels in the early fourteenth century, and again in the early fifteenth century, followed by a significant decline in the later fifteenth and early sixteenth centuries. As Christopher Dyer has remarked, buildings may have much to contribute to the study of urban-rural relationships which are currently concerning historians, and this first analysis of tree-ring dates bears out his suggestion (Footnote 60). So far, few building recorders have engaged with this issue.
It is hoped that this paper has provided a glimpse of the enormous amount that has been achieved in tree-ring dating in a little over a decade. In terms of work likely to be undertaken from now on, the future looks bright. It is not just building recorders who see the value of tree-ring dating. More and more people are interested in using the technique for a variety of purposes. Not only national bodies, but regional and local authorities and organisations, are realising the potential of the technique in relation to the management and conservation of historic buildings. In the long run they will no doubt benefit by obtaining a more balanced view of the development of timber buildings in their counties sooner than that achieved in areas where research has sometimes had narrower aims.
In purely practical terms there are perhaps two main concerns. The first is that as the amount of work increases, so ought the number of professional dendrochronologists. Since several of the laboratories are located in universities it is to be hoped that plans for training new practitioners are in hand. The second problem is funding. At the time of writing the ‘ball park’ figure for dating a building or phase of building is £500 - £750. The main grant-giving bodies, such as the Leverhulme Foundation and the British Academy, have been generous in their support, and will no doubt continue to look favourably on suitable projects. Some of the large-scale projects without a single sponsor, for example those in Hampshire and Shropshire, have complicated and ingenious funding arrangements, mixing contributions from grant-giving bodies, local authorities, local societies, businesses and private individuals. This is probably the way forward, although it causes a great deal of work for the co-ordinators.
A number of potential areas of research have been high-lighted in this paper. Work on some of them depends largely on the professional dendrochronologists, who still have to solve outstanding problems relating to short ring sequences and the accuracy of estimation. Together with ecologists and building recorders, they need to continue to build chronologies in regions where these have proved difficult to establish, explore the dating of timber other than oak, research woodland management, and find answers to the task of dating buildings in awkward periods and parts of the country.
Other topics of research should be initiated by building recorders more directly. In the first place, projects already begun need to have their results written up and published for the benefit of everyone concerned. Secondly, those areas of the country where little has so far been attempted need to be explored, and those areas where only buildings of certain kinds or periods have been dated need to have the coverage extended, so that the resulting national sample is more truly representative. Thirdly, work should be done to test the reliability of older typologies and chronologies of building types and methods of construction, and the results analysed and published. Finally, it is conceivable that documentary historians have questions that buildings, once accurately dated, may be able to solve. Among these are questions relating to the pace of regional development, the changing fortunes of different social classes in the Middle Ages, the rise and decline of urban prosperity, and the differences in the development of town and country. Another area of research where tree-ring dating could play a significant part is in learning more of the organisation of the building trade itself. These are a few topics which are being, or could be, explored, and to them could be added many more by researchers whose perspectives are quite different from that of the present writer.
In an article in this volume, Matthew Johnson asks why it is that, despite the huge amount of research over the last twenty years, there is still no national synthesis to rival Eric Mercer’s English Vernacular Houses of 1975 (Footnote 61). The answer to his question relates to larger issues than tree-ring dating, but it could be argued that it is no bad thing that such an important enterprise has not been attempted before now. Speculations of the kind ventured above, vulnerable to amendment though they are, have only become possible since the advent of tree-ring dating. The chronology rests on a different, potentially more accurate, basis from that used by Mercer. We no longer have to use terms such as ‘late medieval’, which cover a wide range of historical phases and experience, but can be far more specific. If we seek to use buildings as historical evidence, then getting the dating right is a pre-condition of fundamental importance.
The compilation of this paper would not have been possible without the help of a number of people. Nat Alcock compiled the database which made analysis of the results possible. The maps in Figures 1 and 2 were kindly prepared by Frances Halstead. Nat Alcock, Graham Beaumont, Martin Bridge, Nigel Fradgley, Bob Hook, Bob Laxton, Cliff Litton, John McCormack, Gwyn Meirion-Jones, Dan Miles, Madge Moran, John and Jane Penoyre, Edward Roberts and Ian Tyers have commented on various aspects of tree-ring dating. I am grateful to them all and hope I have in no way misrepresented their views.