Dating the Westminster passageway with isotopic dendrochronology By Dan Miles & Neil Loader
Mark Collins told us about the newly found passageway in Westminster Palace, and asked if we could study the ceiling lintels with a hope that tree rings might determine when the timbers were felled. This dating technique – known as dendrochronology – relies on the matching of sequences of growth rings of varying widths in a piece of wood of unknown age, against a reference chronology derived from the rings of many trees that has been independently dated. If successful, dating the Westminster lintels would imply when the blocked doorway was inserted, or at least enlarged.
We assessed timbers for dendrochronological potential in March 2019. There were five bearing-beams or lintels aligned north-south, all in oak wood. The western three appeared to be first-use timber, while the other two might have been reused from another structure. The latter were not suitable because of the possible secondary use, and anyway they had a very low ring count – probably less than 30. Two of the larger beams appeared to have around 50–60 relatively consistent rings, with some bark edge. This was not a particularly optimistic assessment (normally we need 80–100 rings), but Parliament asked us to go ahead. Certainly, these timbers were not of the quality of those used in Richard ii’s reconstruction of the Westminster Hall roof in the 1390s, which we successfully dated in July 2019 as part of a separate project.
In May 2019, with Mark Collins’ assistance, we sampled the second and third timbers from the west. From study of the pattern of shakes and knots, we determined that both were from one log, squared up with the side and sawn in half. Therefore, all of our five samples are considered to have come from the same parent tree (four from the second timber from the west, and one from the third).
Importantly for tree-ring dating, three samples retained bark edge. This was vital for the interpretation of any dendrochronological dating, as one wants to know the date the timbers were felled, not when they started to grow. The sapwood was noted to have considerably narrower growth rings than the heartwood, not an uncommon feature.
The individual ring sequences were compared with each other, but cross-matching was poor, especially considering they were from the same parent tree. It was only by both comparing statistically and confirming with graphic plots that all five sequences were correctly aligned and combined into one 92-ring mean sequence. As the pith was at one end of this sequence, and the bark edge at the other, the tree was probably less than 100 years old when felled.
We compared this mean ring sequence with the reference chronologies, but we found no
conclusive cross-matching. The longer individual sequences were also compared with the chronologies, as well as a reduced master of just the heartwood, but again no matches were forthcoming. The fact that the individual same-tree components matched so poorly would suggest that there were growth inconsistencies which led to ring widths not correlating very well with changes in climate. The reasons for this are not entirely clear, but it is not uncommon with ring-width dendrochronology.
This failure of conventional dendrochronology was not entirely a surprise, given the initial assessment. Nevertheless, Mark Collins was keen to try all avenues of scientific research to answer the primary research question. Fortunately, together with colleagues at Swansea and Oxford universities, we have been carrying out research into the development of a novel precision dating method: oxygen isotope dendrochronology.
Isotopic dendrochronology
Oxygen isotope dendrochronology relies upon many of the same fundamental principles, limitations and assumptions as conventional dendrochronology. However, instead of using widths to characterise annual growth rings, it uses the ratio of oxygen isotopes ( 18oand 16o) from the summer growth (known as δ18o). Variations in this signal largely reflect the isotopic signature of rainfall taken up by the tree, and are related to how wet the summers were.
Through an exacting laboratory process, each ring is prepared for isotopic analysis. They are separately dissected under a microscope to excise a sample of the late wood (summer growth); the early wood (spring growth) is largely formed from stored photosynthate, so tends to carry a chemical signal influenced by earlier years’ weather. The small individual samples are then purified to alphacellulose and analysed by isotope-ratio mass spectrometry. The resulting isotopic data are then compared against a reference chronology in a manner similar to ring-width dendrochronology. Where a strong and unambiguous statistical match is identified a date may be reported.
The samples from the two lintels were run individually. Both found a match with the reference chronology, with a probability that this occurred by chance of less than one in a million. The two δ18oisotope sequences also match each other, and were combined to form a 75-year mean series which dated with high confidence. As the isotopic matching of the two samples was absolute and definitive, it was possible to transfer these calendar dates to the ring-width sequences, and through the conventional dendrochronology, apply these dates to the other three bark-edge samples. As these three samples had complete sapwood, with an additional partial ring from spring growth in the outermost and youngest year, a precise felling date of Spring 1659 was determined for the parent tree, and thus for both beams.
Analysis of the wood grain and isotopic dates confirmed that the two timbers sampled are likely to have originated from a single tree. The two halves of the tree were opened out and placed crown down. Gaps observed between the mortar bedding and the edges of the timber, indicate that the lintels had shrunk a little after being put in place, and may suggest that the wood was used unseasoned, very shortly after felling in spring 1659.
Isotope dating was conducted by the Swansea University Isotope Dendrochronology Laboratory, in partnership with Darren Davies, Gareth James, Giles Young, Danny McCarroll (Swansea) and Christopher Bronk Ramsey (Oxford); development of the technique was supported by the Leverhulme Trust and Natural Environment Research Council. For a description of the method see “Tree Ring dating using oxygen isotopes: a master chronology for central England,” by NJD Loader & five others, Journal of Quaternary Science 34 (2019), https:// onlinelibrary.wiley.com/doi/full/10.10
02/jqs.3115. Dan Miles is a partner at the Oxford Dendrochronology Laboratory; Neil Loader is professor in the Department of Geography at Swansea University