The Year Written In The Wood

AA tree in a temperate climate lays down one layer of wood a year, and the width of that layer records the season which produced it: broad after a mild wet spring, narrow after a cold or a dry one. Because every tree in a district lives through the same weather, the sequence of wide and narrow rings is shared property, and a pattern taken from a timber of unknown age can be slid along a sequence of known age until the two lock together. This is crossdating, and where it works it delivers what no other archaeological technique offers: not a range but a year. The oak chronology assembled from trunks buried in the gravels of the Rhine now runs without a break to 8,480 BC, and a beam from a farmhouse roof can in principle be assigned to the winter its tree came down. The promise is exact; the conditions on which it depends are narrower than it sounds.

BThe first difficulty is that the annual ring is not reliably annual. A tree driven to the edge of survival by drought may add no measurable wood at all, leaving a locally absent ring; a tree checked by a June frost and then recovering may lay down two bands in a single season, a false ring which an inattentive reader counts twice. Marianne Ostrow, who directs the Kelvedon Laboratory of Tree-Ring Research, examined 1,240 cores from pines on the arid fringe of the Ebro valley and found rings absent in 6.4 per cent of the years sampled, rising to 19 per cent across the three driest decades in the record. Oaks on deep well-watered lowland soils showed the same fault in fewer than one year in five hundred. What matters is not the size of the error but its behaviour: a ring that was never formed always shortens the count and never lengthens it, so the mistakes of a long sequence accumulate in one direction instead of cancelling.

COstrow's larger worry is not counting but matching. A ring pattern fixes a date only if the timber grew close enough to the reference sequence to have shared its weather, and medieval builders floated wood down rivers for hundreds of kilometres. In 2019 her laboratory assembled 380 beams whose felling years were fixed independently by building accounts, and tested them blind against master chronologies from three regions. Where the timber had grown within 200 kilometres of the chronology used, the true year was recovered in 87 per cent of cases; beyond 600 kilometres the figure fell to 41 per cent, and in 22 per cent of those distant trials the laboratory returned a date that was both wrong and statistically confident. The failures were not spread evenly. They fell overwhelmingly on beams carrying fewer than 70 rings; samples with more than 120 rings were dated correctly whatever the distance, since a longer pattern has more shape and fewer places to fit.

DAurel Vargha, an environmental archaeologist at the Sandhaven Institute, accepts the audit and disputes what is drawn from it. A dating method, he argues, is never fairly judged against perfection; it is judged against the method that would otherwise be used, and for wooden structures that method is radiocarbon. Working with 96 timbers of known age, he found that radiocarbon returned ranges averaging 84 years wide at 95 per cent confidence, while ring-matching, counting its failures as failures, placed 71 of the 96 in the correct year. The two are also increasingly used together, a radiocarbon range narrowing the field of candidate matches before the rings are read. Vargha volunteers two limitations. Every timber in his sample was oak carrying more than 150 rings, which is not what a waterlogged excavation usually yields; and he compared how precise the two methods were, not how often each was believed by excavators when the two disagreed.

EThe argument has left the laboratories. A rule of the Nordic Heritage Council in force since 2022 obliges any laboratory supplying a date for a publicly funded excavation to deposit its raw ring-width measurements in an open archive within two years, and two other national bodies have drafted something similar. Ostrow regards the duty to deposit as the most valuable part of such rules, because a contested date can then be re-tested by somebody with nothing to lose by overturning it. Vargha, whose figures are the more often quoted by defenders of the method, is here the more sceptical of the two. An archive, he observes, records the match a laboratory accepted without recording the matches it rejected or why, and a laboratory in the habit of accepting weak fits will satisfy the rule comfortably. Both dislike the practice of publishing a felling year without stating how many rings the sample carried.

FWhat is emerging is a division of the question rather than an answer to it. Where a tree grew under a single dominant constraint, at the cold limit of its range or on a dry margin where every millimetre of growth waits on rain, the ring series carries a strong shared signal and a match is either convincing or plainly absent. Where growth is governed by several factors at once, as on sheltered lowland soils with water and warmth to spare, the series records the accidents of one tree's life as loudly as the climate of its district, and adding more samples will not clean it. Whether the timbers that survive in excavation come mostly from the first kind of place or the second is disputed, and neither researcher claims to know. What both reject is the framing that dominates popular argument: that the choice lies between a method which gives an exact year and one which gives only a range, when the exactness itself is conditional on where the tree stood.