How crossdating fixes a ring to a year

Crossdating is matching the pattern of wide and narrow rings in a piece of wood against dated samples from the same region, until every ring has exactly one calendar year. It is what turns ring counting into dating, and it catches rings that a plain count would miss or invent.

The Laboratory of Tree-Ring Research at the University of Arizona calls it the field’s most fundamental technique. Its own wording is that dendrochronologists demand a single calendar year for each ring. A. E. Douglass began the work there in 1904, with specimens collected in the lumber yards of Flagstaff, and the laboratory itself was approved as a university department in December 1937.

A count gives the number of rings, not their years. A tree can skip a ring in a bad year or lay down a false one after a dry spell, so even a clean sample can be wrong by one or two. Matching against dated wood exposes both mistakes. That is why dating is pattern-matching, not counting.

Trees do not always form a complete ring. In a severe drought, or in a year of insect attack, part of the cambium, the layer that makes new wood, can stay dormant all season. The ring is then present in some places on the trunk and absent in others. St. George and Anchukaitis call this one of the main reasons tree-ring dating is done by pattern-matching and not ring counting.

The oldest drawing method for doing that matching is the skeleton plot. Only the narrow rings are marked, as lines on a dated axis, and the lines from an undated sample are slid along until they sit under the lines of a dated one.

A master chronology is an average ring-width record for one species and region, built by crossdating living trees and then older timber that overlaps them. The longest regional ones reach back more than 12,000 years. A new sample is dated by sliding it along the master until the patterns agree.

The reference records are shared. NOAA’s International Tree-Ring Data Bank, the world’s largest public archive of tree-ring data, lists more than 5,000 sites on six continents, with ring-width, wood-density and isotope series. Anyone can search it by species, place or investigator.

Software compares ring widths and reports a t-value. English Heritage reports often treat a t-value of about 3.5 or more as a match, if the curves also look alike, and prefer 5 or 6 and higher because lower values can be spurious. A date is stronger when it matches more than one reference chronology.

The visual check matters as much as the number. Two curves can reach a respectable t-value by chance, in a short sample especially, so the curves are plotted and read by eye, and the date has to make sense for the timber. When it does not, the honest result is “no secure date”.

Further reading: Stokes and Smiley, An Introduction to Tree-Ring Dating (University of Chicago Press, 1968).

• Laboratory of Tree-Ring Research, University of Arizona: What is dendrochronology?
• Laboratory of Tree-Ring Research: lab history
• NOAA NCEI: International Tree-Ring Data Bank
• St. George and Anchukaitis, “On the AD 1815 Tambora eruption and the matter of misplaced tree rings”, PAGES Magazine 23(2), December 2015
• Archaeology Data Service: English Heritage dendrochronology reports