Vol. 2 No. 2 - 1973
Historic buildings in Ireland often lack any real evidence of building date. This is in sharp contrast to the general situation in England and on the Continent where documentary evidence is very much more complete. In the case of many Irish castles, monasteries and tower houses where only fragments of broken walls remain, nothing will ever be known apart from what can be recovered through archaeological excavation.
Until recently the same comment would have been valid for most of the extant early buildings, at least in the North of Ireland. The farmhouse or cottage of a peasant farmer constructed from local materials and built by local craftsmen will, in general, have given rise to no documentation or record. However, the development of dendrochronology or tree-ring dating has for the first time in this country opened up the possibility of establishing accurate dates for the structural timbers of a whole range of buildings as well as sites of a purely archaeological character.
The basic theory of the dendrochronological method is of extreme simplicity. Most trees in a temperate climate develop one growth increment or ring per year, thus over their lifetime they leave a record of the number of years they have been growing. This record is preserved as a series of concentric rings exhibited in a cross-section of their main stem. The dendrochronological method depends not on the number of rings but on the relative widths of these rings. The annual ring is put on immediately under the bark and is due to the growth and division of a thin layer of cells called the cambium. In a timber such as oak, which is of primary interest to European work, the cambial activity is responsible for a burst of large vessels as the leaves expand early in the growth season. These vessels, formed immediately under the bark, carry liquids from the roots to the leaves. After this initial vessel formation the cambium produces smaller vessels during the remainder of the growth season and the ring ends abruptly with a line of very small and dense cells at leaf fall. The tree is then dormant throughout the winter until growth is triggered in the spring of the following year.
The early wood containing the large vessels is of lower density than the increasingly fibrous summer wood. It is this variation in density and hence in wood colour which makes the growth rings clearly visible. As stated above, the most significant factor about these growth rings is the variation in their width from year to year.
No two growing seasons are ever exactly the same and this variation in the complex interbalance of climatic and other factors which limit the tree's growth is reflected in the relative width of the annual ring. That is, in a good growing season a tree will put on a relatively wide growth ring while in a bad growing season it will put on a relatively narrow ring. To give some idea of the complexity of the factors governing the width of a tree ring, the following relationship was developed by Danish workers for oak (Jones, 1960).
Ring width = –52.26 + 0.0833 (mm rain June + July)
+1.494 (mean temp. ºC June - September)
+1.280 (mean temp. ºC Aug. - Oct. of preceding year)
It can be seen from this that in its lifetime a tree incorporates into its structure a pattern of wide and narrow rings which are a reflection to some extent of the variations in the external factors affecting its growth. This leads directly to a basic statement of the dendrochronological method "that two trees growing under the same conditions over the same period should exhibit, to some extent, similar tree-ring patterns".
The original testing of this hypothesis was carried out shortly after the turn of the century, in America, by A. E. Douglass, an astronomer interested in sunspot cycles. He discovered that trees growing under very similar climatic conditions in the semi-arid regions of southern Central America showed marked similarities in their patterns of wide and narrow rings. Since in his area of study the amount of sunshine was almost constant from year to year) he was able to show that the ring widths of trees were closely correlated with the amount of rainfall in any particular year. That is, in a year of extremely low rainfall a very narrow ring would result and conversely in a year of relatively high rainfall a wide ring would result. Significant years, where all the trees studied exhibited either wide or narrow rings) were called signatures. By virtue of the annual nature of the rings he was able to count back in time from the felling date of the trees and establish the calendar dates of the wide or narrow signature years. Thus all of his yellow pine samples showed 1898 as a noticeably wide ring and 1851 as noticeably narrow (Douglass 1919).
The limit of his primary information on tree-ring patterns was determined by the age of the trees on which he was working. However, he found that the sequences of rings could be extended back in time by using the patterns of tree-rings contained in historic timbers from settlers' cabins and Pueblo Indian dwellings. This was possible because the patterns of signature years from the outside of the historic timbers could be matched with the patterns from the inner end of the living trees. By repeating this process with successively older pieces of timber from the same region, he was able to build up a chronology of many hundreds of years in which each tree-ring could be related to a calendar year (Bannister, 1969).
Having constructed this master chronology he was then able to date pieces of timber of unknown age by comparing the tree-ring pattern in the undated sample with the master chronology and finding the position of best fit. It was the success of this method which led to the establishment of absolute dates for American prehistoric sites for the first time. With the successful proving of the method by Douglass, the American workers turned their attention to the establishment of very long sequences of tree-rings. Their success with the Giant Redwoods (Sequoia Gigantea) which yielded tree-rings dating back to 1500 B.C. has in recent years been enhanced still further with the exploitation of the Bristlecone Pines (Finns Aristata). These trees, a single one of which had lived for over 4,000 years, have yielded a continuous tree ring chronology of 8,000 years (Ferguson 1972).
While this work was continuing in the United States the successes prompted workers in the Old World to begin investigations in the same field. However, there were two highly significant differences between the available timbers in Europe and America, Firstly, there were no trees in Europe with the lifespans of the order of those of the Redwoods and Bristlecone pines. This meant that the building of sequences of tree-rings necessitated the cross-matching of large numbers of samples, each of relatively short length, mostly between 100 and 300 years. Obviously this made the construction of European sequences much more laborious than those of America. Secondly, the simple stress conditions of the American trees, which made their growth directly dependent on the quantity of rainfall, did not exist in the more temperate and lower altitude areas of North-West Europe. Instead, as mentioned above, the growth of trees was the result of a complex interbalance of climatic factors. In addition, while virtually all the American work had been carried out on softwoods, in Europe apart from Western Russia and Scandinavia the only existing timber used consistently from early periods was oak (Quercus L.). However, regardless of the difficulties successful chronologies have been established in Russia and Scandinavia using pines and in Germany, Belgium and England using oaks. Several of these chronologies extend back as far as the mid-first millennium A.D. (Thompson 1967, Eckstein 1970).
These European sequences have been used successfully to establish the dates of numerous Medieval buildings and confirm the known date of others, by comparing the ring plots of constructional Timbers with the established tree-ring master chronologies.
In Ireland there had been no serious attempt to use the dendrochronological technique before 1968. This was largely due to the entrenched belief that the wet climate of Ireland would lead to complacent tree-ring records. That is, the very slight variations in the weather year to year would not be sufficient to significantly alter the ring widths. In fact, reference to the existing weather records for Ireland, which have been kept in some cases back to A.D. 1700, show that the county as a whole has suffered considerable climatic variation especially in winter severity, summer cold and the occurrence of droughts (Wilde 1851). Further, observations on sections of both felled oaks and historic timbers show very considerable variations in the yearly ring widths, regardless of period.
With the European work as a precedent, a dendrochronological project was undertaken in the Palaeoecology Laboratory at Queens University, Belfast. Since oak is the only native timber commonly available from all periods, it was decided to use it as the basis of study. Samples of recently felled oak trees were obtained, from widely separated localities in the North of Ireland, for comparison of their ring patterns. Significant correlations were obtained between all of the samples studied and a master chronology was established on the basis of the information contained in their tree-ring patterns. The results obtained from this study showed conclusively that the method was valid in the small climatic area contained within a 50-mile radius of Lough Neagh.
This work gave a total tree ring sequence of 321 years running back in time from 1970 to 1650. This sequence was extended back into the fourteenth century by cross-correlation of the tree-ring patterns in the master chronology with the ring patterns of oak timbers from seventeenth century buildings. Up to the present this chronology has allowed accurate dates to be established for a dozen houses of the seventeenth and early eighteenth centuries which had utilised native oak timbers in their construction. Previously only two of these houses had documented dates and in both cases considerable doubt existed about the actual year of building.
The techniques employed in the cross- correlation of timber samples from temperate climates such as Germany or Ireland are different to those outlined for the semi-arid regions of America. As stated above, the American cross-correlations were obtained originally on the basis of the matching of signature rings, i.e. years where all trees in an area showed either a significantly wide or narrow ring. However, in European conditions trees do not reflect a single component of the climate and hence cross-correlations must take into account the whole ring pattern. This means that, instead of attempting to find signatures to cross-correlate two tree-ring patterns, plots of the widths of all of the rings are used. The ring widths are measured consecutively along a radius from the pith to the bark .The resulting jagged curves are then cross-matched on a statistical basis which takes account of all of the maxima and minima on the curves. Since the pattern of ring-width in any period of, say, 100 years is related to the climatic variations of those 100 years and is hence unique, it is to be expected that a statistically significant correlation will be obtained between two ring patterns which were formed over the same period.
The use of statistically backed computer programmes allows the subjective human element to be removed from the visual matching of tree-ring patterns. In comparing tree-ring patterns, the computer is programmed to move one curve past the other. Each time the second curve moves forward one year, a statistical test is performed to measure the correlation between the overlapping portion. A significant result is expected only when the portions being cross- matched had originally grown over the same set of years.
An extension of this process is the case where an established tree-ring chronology has been produced, for example the 590 year Belfast chronology, where each ring of the chronology can be assigned to a calendar year. In this case, a sample of oak timber can be dated by sliding its tree-ring pattern past the whole master. At the point where a significant cross-correlation is obtained between the un- dated sample and the master chronology, the date of the outer ring of the sample can be read off against the corresponding ring of the master. In oak where the sapwood is of a different colour and texture to the heartwood it is possible to obtain a close estimate of the date of felling of the tree from which the sample came, provided some or all of the sapwood is preserved. Of the material investigated to date, 70% of all undoubtedly native oak timbers have been found to agree significantly with the established standard chronology for the Northern of Ireland. The remainder have for the most part been unsuitable due to short tree-ring records or distortion of the ring pattern.
With the available 590 year chronology for the North of Ireland, this technique is capable, at the present time, of elucidating a number of the problems faced by the vernacular architect or building historian in this area. These problems centre on the almost total lack of accurate dating information relating to early buildings. Provided that oak timbers are available, dendrochronology can establish, with a high degree of certainty, a date for the oak timbers used in buildings of otherwise unknown age. This technique has already established relative dates for a number of constructional details, previously only vaguely related. Ultimately, with the further extension of the standard tree-ring chronology back in time, it will become possible to establish dates for crannogs and other timber-rich sites of pre-plantation date.
In the Craigavon area with its Plantation background and its scatter of oak framed buildings (U.A.H.S. 1969) whose timbers were all derived from the local oak forests, a detailed dendrochronological study could produce an invaluable series of dates which would trace the development of the area in the seventeenth and eighteenth centuries. In practice, thin sections form the ends of oak beams or slices from replaced oak timbers are all that are needed for study and any information on available samples of ancient timbers forwarded through the Editor would be most useful.