Dendrochronology, or Tree-ring dating, has emerged in recent decades of the 20th century as one of the most important dating tools for a number of disciplines such as archaeology, climatology, botany, and the history of art and material culture. While there are numerous publications and informative websites concerned with the technical aspects of dendrochronology and its application, this introduction is designed as a guide for medievalists with a general interest in the subject.Briefly stated, dendrochronology creates a statistically based calendar of tree felling dates derived from the meticulous measurement of the variation of annual growth rings. Each 'ring' represents a single year's growth. Since tree rings reflect seasonal changes, only trees from temperate and arid regions have identifiable annual rings. Species like oak, chestnut, elm, beech, juniper, and conifers produce annual rings whose width and character depends mainly on climatic conditions, such as moisture and temperature. Thus, one would expect to find similar ring patterns for the same species in wetland habitats of similar altitude and growing conditions. When measurements of the variations in the size of the annual rings have been accumulated for large numbers of trees from the same species and region, a chronology of ring widths derived from these dimensions can be established. This methodology, for 'dendro dating' by means of a reference chronology was developed in the early 20th century by A.E. Douglass who used yellow pine from ancient pueblos in Flagstaff, Arizona. Oak, however, was the premier building material for large-scale carpentry in pre-modern Europe, and early buildings have furnished a considerable amount of specimens. Oak chronologies for Europe were developed mainly in the last quarter of the 20th century, such as: Hollstein's north German oak chronology (1980), Becker and Delorme's south German oak chronology (1978 -1981); there are exceptionally long chronologies for Ireland, extended via fossil material and anchored by living trees (Bailie and Pilcher, 1977 and 1980); numerous chronologies were developed in Britain, but the sequences for oak are generally shorter, e.g. Fletcher (1977-78) and Laxton and Litton, East Midlands (1988). These tree ring chronologies, including others from Scandinavia, Holland and Belgium span most of the medieval period and have been used as the foundation oak chronologies for those developed later for Europe, including parts of France, such as the early Northern French Chronology AD 1274-1979 (Pilcher, 1987) and in eastern France (Lambert et al., 1989). Of particular interest for the exploitation of Baltic oak is the north Polish chronology (Wazy, 1986). It should be pointed out, however, that while there are hundreds of dated tree-ring sequences throughout the world, many are designed for post-medieval climatic and ecological studies and involve species other than oak.
With the aid of computerized chronologies and sponsorship of national organizations and universities, a dramatic increase has occurred in the number of dates obtained for ancient and medieval sites. It must be understood, however, that dendrochronology provides us with the felling date, or sometimes a felling date range for the earliest possible construction date (called the terminus post quem). This date or range of dates is not the same thing as a date for a building, and one must be cautious in accepting felling dates without understanding how to use the information. On the other hand, medievalists can be confident that in the period before 1300, timber was normally used within a year or two of felling. When the exact felling season can be established (often possible through dendrochronology) it confirms the traditional pattern of cutting in winter or early spring before the onset of the building season. Given the nature of medieval building practice, dendrochronology has become highly significant for architectural and technological history. It gives us reliable and objective (scientific) dates for a wide range of medieval artefacts and buildings against which we can compare our knowledge gained from documentary sources, stylistic change, and technological evolution from ship hulls to cathedral spires. Dendrochronology is particularly useful both in establishing phases of construction in high status, stylistically dated buildings and also for dating lower status, vernacular structures for which it may be the only tool available.
Architectural historians, historians of science, and medievalists not fully informed about this dating tool are often confused by statistical ranges for dates and ask how such a seemingly straightforward technique like counting and comparing tree rings can often produce such varied results? Perhaps when the methodology and variables involved are understood this will be less perplexing. At the outset, we must understand the fundamental difference between building a reliable reference or master chronology and the complex process of obtaining a sequence of felling dates for oak extracted from sites. Dating will also likely involve such standard research methods as stylistic, typological, structural and documentary analysis or, looking toward the future, the use of advanced techniques of botanical research such as genetic analysis for refining ring characteristics for individual tree species (See Pearson, VA, 28; Groves, VA, 31, 6).
A master, or reference chronology (a reliably dated sequence) begins, in principle, with living trees of a particular species, such as oak or pine. In order to construct this critical resource for dendro dating, numerous trees are felled or bored through to obtain a core sample (Fig. 4, below). The rings inside the bark are analysed to determine the precise age of the tree. Each ring represents a year's growth. Thus, every ring width, measured in hundredths of a millimeter can be assigned to a particular calendar year. By cross matching a great number of samples from trees of various ages, the ring widths are averaged to establish a mean sequence of size and years. This long and arduous process, if successful, results in a highly reliable Master chronology that covers a particular region and time span and is the foundation for dating unknown specimens. The range and usefulness of a particular chronology depends on establishing a link from living trees backward in time. For example, a chronology that has been critical for dating medieval buildings in Continental Europe is the German Master Oak Chronology, A.D. 822-1964 (Hollstein, 1965-1980). This chronology has been the cornerstone for much of the dendro dating in parts of France and Belgium and has led to the development of chronologies for the Ardennes, A.D. 1146-1991, and the Mosan (Meuse 5) chronology, A.D. 672-1991(the region of the Meuse river between France and Germany). These sequences have, in turn, proved useful in dating buildings in Flanders and Picardy (Hoffsummer, 1995, 30- 40).
By working gradually backwards to join undated sequences (floating chronologies) to a dated master chronology, the time span of the original sequence continually expands and thus, numerous versions of master chronologies are produced as research has progressed. Moreover, bridging opens the possibilities for the addition of new geographical regions and hence an extension of the sequences for oak and other woods. Sites once updateable by this method are now part of a global research database. Medievalists, in particular, appreciate the relatively recent addition of the Baltic States of north Eastern Europe and the extension of the Irish oak (Dublin and Belfast) chronologies. (See below)
Individual samples of unknown date are compared to reference chronologies by a combined process of physical (dimensional) matching and statistical analysis using computer programs such as Baillie's CROS program. The rough data of ring-width variations in a given sequence is indexed, or smoothed, by using a moving average of five adjacent widths to estimate the general trend of the raw ring measurements. The correlation between the cross-matched samples and several reference chronologies is qualitatively evaluated by the so-called Student's t value (a statistical method devised by William Sealy Gosset, 1876-1937, who used Student as a pseudonym). A widely used adaptation of the Student's t is the Belfast CROS program designed specifically for the 'high frequency' data from tree rings (Baillie, 1982, 84-85). This program is most recently discussed by Oxford dendrochronologist, Dan Miles (VA, 28, 40-41). Essentially, the calculations describe the degree of similarity or difference between pairs of ring widths (indices) at positional offsets along the sequence of data; thus, the total length of the sequence as well as the number of rings are factors in obtaining a good match (Fig. 3). When the correlation coefficient (which lies between -1 and +1) is nearer to +1, the match is closest. Taking the logarithm of this value and dividing it by N-1 (where N is the number of rings compared), the t-value is obtained.
T-values of 5.0 or better indicate good correlation, and 10 or better for English oak would tend to indicate the same parent tree (Miles, personal communication). What is important to bear in mind is that the calculations are enormously sensitive to both the number of rings averaged and the size of the rings, so that even a change of just a few hundredths of a millimeter in one ring size within a sequence can have significant consequences for t-values. According to Dan Miles, the quality of the sample is paramount. Even measuring the same sample along different or adjacent radii can produce highly variable t-values ranging from 20 to 40 and for poor samples, as low as 4 or less! Thus the dendrochronologist's dream would be oak specimens with upwards of 100 well-formed and relatively narrow rings (no doubt this is a happy tree as well). Recently, outstanding t-values of circa. 16 to 22 were obtained from imported oak from Salisbury and Peterborough Cathedrals (see below). When t-values are this high, the timber likely derives from slow growing oaks in an undisturbed site (Ireland or northeastern Europe).
Apart from the basic processes just described, a critical part of dendro dating concerns the heartwood-sapwood boundary (H/S) of oak (Fig. 4a). There is an easily visible physical and chemical change that distinguishes the live sapwood, or outermost growing layers where the sap flows, in contrast to the darker heartwood (dead inner wood). Sapwood is highly variable and rarely evenly distributed even in a single tree; thus, the high degree of variability of this boundary predicates that dendrochronologists have a sufficient range of samples. Moreover, the presence of sapwood within the heartwood, called included sapwood (or "moon rings"), has been estimated to occur in about 2% of European oak and is caused by hard winters and early frosts. These included rings (or, a single ring) appear as light areas like the sapwood seen in the oak slice in Fig. 4 (a). Historians of technology interested in structural matters involving strength of materials should note that timber with included sapwood produces regions of weakness, since this wood is less dense, less elastic, and more susceptible to decay than the surrounding heartwood. According to recent observations by Marek Krapiec, occurrences of included sapwood have been traced in England, France, the Baltic basin, and Russia extending over a considerable period of time; he suggests that these signature years of hard winters may become a useful index for recording trends both for climatology and potential dating (Krapiec, 1999).If the bark or the last annual ring is preserved, the precise year and season in which the tree was felled can be determined; however, a dating problem often arises when the tree is converted into usable timber. To become part of a structure, the tree is felled, trimmed and cut in a particular way, termed conversion. For example, the preferred (and most expensive) conversion method for primary structural members of a medieval roof frame often involved the use of an entire tree (pre-selected for size) converted to a squared, heartwood section of oak whose bark and sapwood had been mostly removed. This is called box-heart conversion. Conversion removes a percentage of the critical dating evidence, especially in more complex carpentry whose jointing requirements would demand squared timbers, as for example, in the assembly of the arcade plate, collar beam, and hammer post in the early hammer-beam roof at Pilgrim's Hall in the Winchester Cathedral Close, recently re-dated to circa 1305 (VA, 2001).
The question then arises: how can an approximate felling date be determined from partial evidence? The solution to this problem relates to a botanical understanding of oak. Since the transition between the porous sapwood and the dense heartwood of oak is climate-related, the amount or missing number of sapwood rings can be estimated from statistics based on complete sapwood for the region where the tree was grown.
Dendrochronologists use two methods of determining the earliest plausible felling date for timber with incomplete sapwood. In England, where dendrochronology has been particularly advanced for several decades, the tendency is to use regional microclimates to determine the number of rings of sapwood normally present. For example, the chronologies for Warwickshire in the West Midlands, indicate that sapwood rings for oak vary from 15 to 40 in number and a general average of approximately 30 rings (30 years) has been used when more refined data is absent.
Elsewhere, the calculation of the amount of sapwood has been estimated by using actual measurement of ring width, as in John Fletcher's early work, where an average of about an inch of sapwood (assuming a consistent average ring width of 1mm) equals about 25 years growth. The sapwood-width method has been recently applied to dating oak in France, Switzerland, and Belgium. For example, for eastern Belgium, Patrick Hoffsummer calculates that the average amount of sapwood from the last hardwood ring to be 2 cm wide, which in turn averages to 16 rings, or 16 years with an uncertainty of +/- 5 years for the earliest possible felling date. While this method uses a shorter span of ring years, it makes assumptions based on averaged ring widths over a fairly wide region.
If the ring marking the boundary between heartwood and sapwood (H/S) is present, then a felling date range can be given; for example, from a H/S boundary ring dated to AD 1300, the average range for a Warwickshire building would be 1315 to 1340. If only heartwood is present, with a last ring dated to 1300, then the earliest possible felling date would be 1315, usually printed as 1315+, but in reality the missing ring sequence with no fixed H/S boundary could extend much further, and felling could have taken place much later than 1340. In the worst case scenarios with no sapwood at all, these methods are still reliable in distinguishing building phases by generations, and as research progresses the date ranges may well become far more refined, especially in England, Belgium, and Germany.
As a caveat, historians of medieval architecture need to realize that dendrochronology at best tells us exactly when a particular tree was felled, but it does not indicate when it may have been actually used or re-used in construction. Thus, one also needs to look carefully at the integrity and consistency of the carpentry per se. There are some cases where timber has been stockpiled in advance of construction, although the normal procedure with oak was to use it green (soon after felling), since it was easier to work before the timber dried, shrank, warped, or hardened. Nonetheless, in large and elite architecture constructed over a considerable period of time, an hiatus in building campaigns might easily occur after a certain amount of timber had already been felled and partially converted.
There is plenty of evidence in both ecclesiastical and domestic buildings for the re-use of timber, which remained a valuable commodity throughout the Middle Ages and was often salvaged. While in many cases, re-use is obvious from context (Fig. 6) dendrochronology has made re-use more apparent when tell-tale signs like inappropriate joints and empty mortises are not conspicuous. In sum, if the timber in question is suitable for this dating method, dendrochronology can be extremely reliable and an invaluable aid to a variety of medieval studies and can date many objects and structures to within a generation up to about 5,000 years ago.
SUMMARY: Non-specialists need to be aware of the critical importance and complexity of the H/S boundary and t-values and especially to remember:
|
In dating historic structures, eight to twelve core samples are normally taken to date the timber in each area of construction of a building, so that different building campaigns and resources are included in the dating process. It is also important to insure that each sample has enough rings to establish a reasonably accurate match, i.e. core samples with less than 50 rings are considered unsuitable. Taking the core samples at the correct angle and position requires skill and knowledge of the structure and construction process.
As an important by-product of dendro dating, we are beginning to learn much more about medieval woodlands, wood species exploited, and the management of these vital resources in the wider medieval economy of supply and demand. Taking a broader, interdisciplinary perspective, this information applies to material resources not just to high status buildings but for a variety of structures such as mills, manufacturing technology, fuel and transport. Also, both up and down the social scale, we can ascertain the choices carpenters and patrons made concerning the quality and quantity of timber employed in a given building (Cf. Courtenay, 1997).
In this ecological context and with the development of regional chronologies, dendrochronologists and botanists are exploring the potentialities of locating timber sources, or timber provenancing. This yields a good deal of information about the types of woodlands in which trees were grown and by implication, patterns of woodland exploitation and the marketing of resources, as for example, oaks from Ireland, oak, fir, and pine from Scandinavia and woodland oaks from the eastern Baltic areas of Poland and Russia, whose trees exhibit quite different ring patterns from the traditionally coppiced oak woodlands of France, the Rhineland, Britain, and Switzerland. Gavin Simpson (Nottingham), Ian Tyers (Sheffield), and Cathy Groves (Sheffield) have documented early importation of Baltic oak panels into England in distinction to softer woods such as Norway spruce, fir, and Scots pine also documented to the 13th century but rarely surviving and imported mainly in the later Middle Ages. In this area of research, one of the most important contributions to medieval studies is the development of the Danzig oak chronology.
Looking at sources in the far west of Europe, Dan Miles has recently published the results from a long-term study of the roof carpentry of Salisbury Cathedral. For example, his analysis of timber from the surviving eastern chapel roofs that remained intact after Price's reconstruction in 1736 indicates that much of the oak was imported from Ireland. Dendro dating and provenancing prove that a large proportion of the timber came from the Dublin area, and a spring felling date of 1222 with high t-values confirms documentary evidence for the construction dates for the original roofs of the eastern chapels built between 1222 and AD 1225, the date of consecration of the chapels. It is also notable that the carpenters' marks are in Arabic rather than the traditional Roman numerals, and these are to date, the earliest known in England. (Miles, Ancient Monuments Lab Reports (AML), forthcoming; Miles and Worthington, VA 31, 2000, list 107).
Other well-known monuments also contain imported oak; for example, the backs of the choir stalls at Ely Cathedral dating to 1345 were made of Baltic oak boards of the highest (carving) quality (Groves, VA 31, 59). Most recently, Groves et al. have established from the initial phases of the examination of the roof and great painted ceiling at Peterborough Cathedral the spectacular use of Baltic oak panels. The eastern bays of the nave date to circa 1230 (with superb t-values ranging from of 9.7 to 15.2! (Groves, AML 10 and 37 (1999-2000) and VA 31, 119). This research is still in progress and will appear in subsequent issues of Vernacular Architecture. Stylistically then, this magnificent canted ceiling can be placed in the context of extant, mature Romanesque and early Gothic compositions dated to the end of the 12th and first half of the 13th centuries in the tradition of the nave of St. Michael's at Hildesheim and continued later in parish churches.
Looking to the future, dendro provenancing will no doubt be of increasing interest to social and economic historians. Thus far, the tree-ring evidence reinforces documentary references to timber imported into Britain as early as the first half of the 13th century. Gavin Simpson has made it clear that both oak and conifers were shipped at approximately the same time and to the same ports, mainly along the eastern coast of England, where timber resources were in decline. Knowledge of where the timber came from for medieval buildings thus widens our perspective considerably with regard to oak supplies and marketing. As in modern applications of this field, medieval trees may provide insight into the relationship between climate and population density, resource management and long-distance trade as well as unravelling the more immediate dating problems of architectural history. All this said, dendrochronology must surely now takes its place in the 'toolbox' of disciplines for medievalists.
|
*Author's Note I am indebted especially to Dan Miles for supplying me with information and comments, to Nat Alcock for his helpful suggestions and information, and to Gavin Simpson for his illustrations of core samples. The drawings and photographs are my own, apart from the engraving of Peterborough and the Calendar detail from the Duke de Berry, Trés Riches Heures. The web sites listed are based on personal choice with a medieval focus, especially buildings. Anyone who investigates the more general sites and on-line bibliographies will discover an enormous amount of material related to earth sciences and archaeology from prehistory to the present. This listing is thus only a springboard to go in a variety of directions. I have added a short bibliography that will lead the interested reader into this fast-growing field, since what is presented here is only the tip of the iceberg. |
Baillie, M.G.L. (1982). Tree ring dating and archaeology. Chicago: Chicago University Press.
Baillie, M.G.L. (1995). A slice through time : dendrochronology and precision dating. London: Batsford.
Becker, B. (1993). 'An 11,000-year German oak and pine dendrochronology for radiocarbon calibration'. Radiocarbon 35: 201-213.
Bourquin-Mignot, C., Lambert, G., Lavier, C., Perrault, C. (1996). 'Comparison between oak and beech ring series from medieval sites and living trees in France', in Tree rings, environment, and humanity : proceedings of the international conference, Tucson, Arizona, 17-21 May 1994, ed. Jeffrey S. Dean, David M. Meko, and Thomas W. Swetnam (Tucson, AZ: Radiocarbon, Dept. of Geosciences, University of Arizona), pp. 485-490.
Courtenay, Lynn (1997). 'Scale and Scantling: Technological Issues in large-scale Timberwork of the High Middle Ages', in Technology and Resource Use in Medieval Europe, Cathedrals, the Mills, and Mines, ed. Elizabeth B. Smith and M. Wolfe (Aldershot: Ashgate), pp. 42-75.
Delorme, A. (1973). 'Aufbau einer Eichenjahrringchronologie für das südlichen Weser- und Leinebergland', Forstarchiv. 44: 205-209.
Delorme, A. and Leuschner, H.H. (1983). 'Dendrochronologische Befunde zur jungeren Flussgeschichte von Main, Fulda, Lahn, und Oker', Eiszeitalter und Gegenwart, 33: 45-57. ['Dendrochronological discoveries of the earlier histories of the [Noachian] Flood by Main, Fulda, Lahn, and Oker']
Fletcher, John, ed. (1978). Dendrochronology in Europe, Principles, Interpretations and applications to Archaeology and History, British Archaeological Reports (BAR), International ser., 51 (Greenwich). [This is a classic discussion for Britain, but now rather dated.]
Grissino-Mayer and Butler, David R. (1988). Tree-Ring Analysis: A Bibliography with special emphasis on Applications in Physical Geography. Athens, GA: University of Georgia, Department of Geography.
Groves, C. (2000). 'Belarus to Bexley and beyond: dendrochronology and dendro-provenancing of conifer timber', Vernacular Architecture 31: 59-66.
Grynaeus, A. (1995). 'Dendrochronological Research in Hungary (Present Status as of May 1995 and Future Development)', Dendrochronologia 13: 135-138.
Hackelberg, D. and Tegel, W. (1996). 'Neuentdeckte Fragmente eines hochmittelalterlichen Schiffes aus Uberlingen, Bodenseekreis' Archeologische Ausgrabungen in Baden-Wurttemberg ??: 260-264 ['Newly discovered fragments of a high medieval ship from Uberlingen, Bodensee']
Hillam, J., Morgan, R.A., and Tyers, Ian (1987). 'Sapwood estimates and the dating of short ring sequences', in Ward, R.G.W., ed. (1987)., pp. 165-85.
Hillam, J. (1998) Dendrochronology: guidelines on producing and interpreting dendrochronological dates. London: Ancient Monuments Laboratory, Conservation and Technology, English Heritage. [A basic concise guide widely used in UK]
Hoffsummer, Patrick (1995a). Les charpentes de toitures en Wallonie, Typologie et dendrochronologie (XIe - XIXe siècle). Namur: Études et Documents, Monuments et Sites.
Hoffsummer, Patrick (1995b). 'L'evolution des charpentes de toitures en Belgique et dans le Nord de la France, rapports recents de la dendrochronologie', in Le bois dans l'architecture: Entretiens du patrimoine (1993 : Rouen, France), Direction du patrimoine : Caisse nationale des monuments historiques et des sites: Association pour la connaissance et la mise en valeur du patrimoine (Paris :: Diffusion, Picard)
Hollstein, E. (1980). Mitteleuropäische Eichenchronologie: Trierer dendrochronologische Forschungen zur Arch‰ologie und Kunstgeschichte. Mainz am Rhein: P. von Zabern.
Kaennel, Michele and Schweingruber, Fritz H. (1995). Multilingual Glossary of Dendrochronology. Berne, Switzerland: Paul Haupt Publishers. [German, French, Spanish, Italian, Portuguese, and Russian; references, figures, and a list of species in Latin]
Krapiec, Marek (1999). 'Occurrence of Moon Rings in Oak from Poland during the Holocene', in Tree-Ring Analysis, ed. Wimmer, R. and Vetter, R.E. Wallingford (Oxford: CAB International), pp. 193-203.
Laxton, R.R. and Litton, C.D. (1989). 'Construction of a Kent master dendrochronological for oak, AD 1158-1540', Medieval Archaeology 33: 90-98.
Pearson, Sarah (1997). 'Tree-ring Dating: A review', Vernacular Architecture 28: 25-40. [web: <http://www.vag.org.uk/pearson.htm>; Updated in VA 32]
Simpson, W.G. and Litton, C.D. (1996). 'Dendrochronology in Cathedrals', in The Archaeology of Cathedrals; Oxford University Committee for Archaeology, Monograph No. 42 (Oxford: Oxford University Press), pp. 183-207.
Trenard, Y. and Duchateau, J.L. (1985). 'Dendrochronologie du chene dans la region de Paris', Dendrochronologia 3: 9-23.
Tyers, I., Hillam, J., and Groves, C. (1994). 'Tree and woodland in the Saxon period: the dendrochronological evidence', in Environment and Economy in Anglo-Saxon England, ed. J. Rackham, CBA Research Reports, 89 (York : Council for British Archaeology), pp. 12-22.
Walker, John (1999). 'Late twelfth and early thirteenth-century aisled buildings: a comparison', Vernacular Architecture 30: 21-53.
Ward, R.G.W., ed. (1987). Applications of Tree-Ring Studies, Current Research in Dendrochronology British Archaeological Reports (BAR) International Series, 3 (Oxford).
Wazy, T. (1986). 'Dendrochronologie in Nordpolen', Acta Interdiscipliaria Archaeologica 4: 123-8.
Wazy, T. (1992). 'Historical timber trade and its implications on dendrochronological dating', in Tree Rings and Environment, Proceedings of the International Symposium Ystaad, South Sweden, 1990 (Lund), pp. ??.
Webb, G.E. (1983). Tree Rings and Telescopes: The Scientific Career of A.E. Douglass. Tuscon, AZ: The University of Arizona Press.
[TOP]
