Extract from the publication

DATED WOOD FROM GOTLAND AND THE DIOCESE OF SKARA

by Alf Bråthen

This volume contains dating of wood from medieval buildings on Gotland and churches in the diocese of Skara.

Dates of medieval stone buildings in Visby may be useful in the study of trade and cultural connections in the Baltic Area and the conditions which led to the rise of the Hansa League.

Dating of parts of structures from gotlandic churches mainly concerns the 13th and 14th centuries and may possibly reflect economic conditions, but most rewarding may be the comparisons the reader makes between the architecture and the decoration of the churches on the continent and on Gotland.

The dating of churches from the diocese of Skara was published in 1983 in a small edition which is out of print. The reason for this reprint is that it will be a convenience for those readers who make comparisons between the romanesque parts of churches on Gotland and in Västergötland. These two provinces belong to the areas in Scandinavia which in the Middle Ages got their main influence from the continent where centra such as Lübeck, Bremen, Hamburg, Clairvaux and Cluny are of importance. As many medieval buildings in Scandinavia remain nearly unchanged in their original structure and reflect stylistic features from the continent they may also be of interest for research of the continental centra of influence. The knowledge of the pace of foundation of churches in Scandinavian areas also makes a contribution to the knowledge of the activity of the archbishopric of Bremen.

The original publication, and this internet condensed version, is directed to readers with a wide knowledge of medieval north european architecture and therefore does not point conclusions or analysis which are obvious to this category of readers.

Wood from Gotland is difficult to date

When travelling across Gotland it is immediately evident that in most areas the forests are exposed to severe disturbances in growth. A considerably high percentage of the trees have bumpy trunks and many dry branches in their crowns. By the middle of the 1980:s about 10 samples from newly felled pine trees were collected from three forest areas. One area with a shallow substrata on calcareous rock in the vicinity of Hangvar on the northern Gotland, one marshy ground area north of Etelhem in the middle of the island and one gravel ground area south of Östergarn. The number of annual rings in the samples varied and did not exceed 70. The felling year was known but samples from the first two sites could dendrochronologically not be coupled to each other. One sample from each of the sites could be coupled to a reference chronology from Gotland. From the site at Östergarn 4 of 12 samples could be coupled to each other and to a reference chronology.

The outcome of the investigation indicates that when taking samples from an establishment of wood on Gotland one ought to select carefully among the timbers and take many samples in order to get a sufficient result. One should also have in mind that a dating trial may fail in spite of the fact that accessible timber scrutinized in situ may seem to be promising.

In areas where there are deep deposits of soil on the rock, for instance in areas at Hemse and Sproge, anomalies in the growth of annual rings seem to be considerably less frequent than in areas with thinner deposits of soil. Scientific investigations have not been accessible but root diseases, mostly caused by fungi may be a dominant cause to the frequent presence of trees which abruptly are inhibited in their growth. For most trees the annual growth is without major disturbance during the first decades of their lives, and suddenly the growth decreases to an extremely low level, where the width of the annual rings is less than 0.3 mm. That low level of annual growth often proceeds for many decades and is only to a low degree affected by the variations of climate. In cases of extremely slow growth one or another of the annual rings may be only partly developed. In the upper part of the soil deposit water with dissolved salts has ordinary values of pH, but when the root system with increasing age expands and penetrates into cracks of the calcareous rock it encounters alkaline values of pH, which alters the normal conditions of life of the trees. The production of protective substances like terpene and resins is hampered and the tree is more exposed to attacks. The abundance of dry branches may partly have other explanations, for instance injuries by salty sea moisture which in wintertime is carried by strong winds to the forests. From what is presented above it is apparent that attempts to date wood from Gotland get involved in difficulties. If only one sample from a wood structure is taken the probability is great that when the sample is analyzed it will reveal severe patterns of anomalies in the growth and therefore cannot be dated. Apart from that the probability for dating increases when a number of samples are taken from an establishment, because some are better than the others, it is often possible to couple data from some of the samples so that a longer sequence of data is obtained which reflect the annual variations of climate.

Reference chronologies

Soon after the work on dating wood from Gotland started it appeared necessary to produce special reference chronologies for Gotland. They contain characteristic patterns and signature years caused by the extraordinary conditions for growth on the island and, concerning pine, increases the datability of samples containing not too disturbing deviations from a standard growth. A reference chronology on oak, which only covers 652 years from A.D. 1330 - 679 (counted backwards in time) has no better qualities than a corresponding part of a chronology from the swedish mainland. A pine sample which can be dated against a reference from Gotland also has passable agreement with the pine chronology from Södertörn (south of Stockholm) but the sufficient level of continuous correlation which is required for dating normally is lacking. Hardly any of the more than 2000 samples from Gotland have been completely without parts of disturbances of growth, while from the Swedish mainland samples which reflect the growth in high and straight pines are easy to find.

It is of no use to estimate the fraction of datable samples from Gotland, because already when taking samples a selection is made where the estimation of the quality of dating is dependent on personal judgement. Two examples illustrates the variety of datability: Of 8 samples taken from a construction none was datable; in a favourable case 7 out of 10 samples were dated and in addition to that, among the dated samples the felling year of the trees could be stated because of presence of waney edge.

Some conception of the degree of datability is obtained when looking at the result from 92 samples of pine from Bulverket, a medieval defence construction in the middle of lake Tingstäde Träsk. Of these samples 40 were dated, which makes 45 % datability. Of the dated samples only two had waney edge, which is the surface of the outer ring close to bark. The felling year of the younger of these timbers was the cold season (autumn - winter) 1124 - 1125.

The reference chronologies covering pine from Gotland may be treated as partial sequences put together into two major continuous chronologies. One covers the period A.D. 1976 - 1269 and needs some comment. As the investigation of wood from Gotland started as late as 1979 it may seem a weakness that the felling year, the cold season 1976 - 1977 of the youngest wood is before the investigation started, but the year 1976, which was denoted by the forester in charge, has been confirmed by samples, taken near Östergarn. These have given a sufficient proof, and comparison with references from the mainland has given the same convincing result.

The second continuous collection of partial chronologies of pine covers the period 1366 - 814. The older part of that chronology which covers the 9th century is of limited use because of anomalies in the growth and few samples.

At the beginning of the dating activity on Gotland it was not clear that because of the divergent qualities of the wood a special chronology covering Gotland was needed. Some samples, which were dated relatively to each other, had an agreeing enhanced correlation with a chronology from the mainland. Hopefully waiting for samples which could be dated it was convenient to state a basic year on a relative scale. That year was chosen to be correct to a high probability. On that scale the year 1200 was marked, and as an example a piece of wood from the chancel of the church of S:t Göran got the year 1239 on the same scale. Since then during the last 10 years collection of samples has proceeded and data along that relative scale of time have been well established. From experience it seems to be difficult to find on Gotland pieces of pine wood which with full evidence can be dated against a reference chronology from the mainland. Even though a repeated analysis would give a number of good arguments in favour of that the year 1200 is correctly marked on the relative scale, it would give a deceiving impression of proof. The promising method of transforming the relative scale to an absolute reference chronology was to prolong the younger chronology until it sufficiently overlapped and dated the relative scale. When this is written the degree of overlapping is 97 years. The younger chronology ends at 1266 and the medieval (above denoted as relative) chronology starts at 1366.

The degree of overlapping

When scrutinizing a report on investigated and partly dated pinewood, edited by a european university, it was found that the oldest part among more than 1000 presented samples only had about three decades of overlap to the younger part. Undoubtedly many readers of that paper also have drawn a diagram on the degree of overlapping and discovered the same remarque without putting any questions in public. This means that the correlation between the patterns of growth between the samples compared to each other must have been almost perfect.

The samples from Gotland, presented in this paper, vary in quality, and it may be of some importance to know that the number of years of overlap is more than sufficient.