In prehistoric times, oaks growing on the surface of Irish raised bogs were recording rare extreme events.
These extreme events are characterized by the simultaneous occurrence of the narrowest rings in the lifetime of trees throughout a wide geographical spread of sites.
Significantly the dates of these extreme "narrowest ring" events coincide with the estimated dates for major volcanic eruptions as recorded in the Greenland ice-cap. One of these events occurs in the decade of the and coincides very precisely with a previously suggested volcanic event at (now ) put forward by LaMarche.
One is led to the inevitable conclusion that some major hemispheric event took place in the decade of the , and a strong circumstantial case can be constructed that the event was volcanic in origin. Since the dating is based on precisely dated tree rings no further refinement of the date of the event — probably — is required. There remain several points to be addressed.
Is the event, identified by the dendrochronologists, really Santorini, or did Santorini erupt at some other time, as suggested by some archaeologists? Do other tree-ring records assist in our understanding of the problem? Are there other ways of refining the evidence — i.e., are there isotopic alterations within the tree rings which might allow additional fingerprinting of the event? This paper will review progress in at least some of these areas.
Three lines of evidence — ice cores, bristlecone frost rings and an Irish oak minimum-growth event — combine to make it virtually certain that there was a large volcanic dust-veil event in . None of these lines of evidence is capable of specifying the volcano involved. They cannot therefore prove that Santorini erupted in . However, none of these continuous records, which between them appear capable of detecting most large eruptions, indicate any other consistent event between and . How therefore are we to date Santorini?
There seems to be a consensus opinion that the eruption of Santorini was violent. Watkins et al. (1978) went so far as to say that it was one of the largest explosive eruptions in post-glacial times and indeed that "eruptions of this magnitude are yet to be chronicled." To the outside observer it appears that the main event must have happened in a particular calendar year. It seems unlikely that either archaeology or radiocarbon dating can ever specify that year precisely. Yet, in the overall scheme of things, the date "exists" and, sooner or later, it ought to be possible to pin it down. This article aims to define the problem from a chronological point of view and detail the evidence for the event, seen in the Irish tree rings and elsewhere, in the .
The starting place has to be recognition that the widespread occurrence of narrowest growth rings in Irish bog oaks — minimum-growth events — coincides with the suggested dates of major dust-veil events derived from the Greenland ice cores (Baillie and Munro 1988; Hammer et al. 1980, 1987). Table 1 lists the two sets of dates which have already been tabulated by Hughes (1988). In addition, two of the dates, and , coincide remarkably with two bristlecone frost-ring events in and respectively. (For the former see LaMarche and Hirschboeck 1984; Hughes 1988; for the latter LaMarche pers. comm.)
Acidity peaks over 4 microequivalents/Kg ice
Low growth events in Irish oaks
So the worst growth conditions recorded by the Irish bog oaks tend to occur at points in time suggested by both ice-core and bristlecone-pine workers as being periods when major dust veils occurred.
Now there are several questions to be addressed. What are we seeing in the tree rings? What mechanisms might be involved? What is the most likely date for the event being recorded in the tree rings, whether or not that event is specifically Santorini?
Let us accept for the purposes of argument that the coincidence between the ice-core evidence and that from the bristlecones and Irish oaks is the result of a major eruption. The ice-core acidity peak at ± 20 tells us there was a volcano; the brisdecone pine and Irish oak effects tell us that something "hemispheric" happened within the limits of the ice-core date range — specifically at the beginning of the decade of the , most probably in . Let us look at what is seen in the tree-ring record. LaMarche and Hirschboeck reported a notable frost event in the growth ring for (now ) in high altitude bristlecone pines in the White Mountains of California. At the time the article was written this was the only notable frost event between and . Since then LaMarche has uncovered another frost event at . There is no question of the accuracy of the bristlecone tree-ring record, so we can take it that there was a rare and extreme freezing event, in at least 50% of the trees studied, sometime in the growth season of the year .
When we look at the Irish record we see another rare event occurring in the . Six trees on four different bogs produced their narrowest growth rings in . Such extreme single-year events occur only nine times in 6000 years (Baillie and Munro 1988). However, it is clear from examination of the Irish record that the "event" at is not restricted to a single year. In reality is the low point of a reduction in growth which in general appears to start around and lasts for a number of years.
The first observation which drew attention to the minimum growth events in the Irish oaks centred on a small group of trees from Garry Bog, Co. Antrim. Several of these trees (see Fig. 1) showed a band of very narrow rings in the . Another tree, Q1863, had originally been assessed as "impossible" from onwards (from a chronology-building point of view there is little point in attempting to measure such rings) and indeed re-examination showed that the ring structure broke down completely at . Other Garry Bog trees, Q1933 and Q2032, showed a distinctly narrow band of rings beginning in . In trees from other northern bogs a similar picture emerged. For example, Q5392 from Sentry Hill showed a colour change associated with a narrow band of rings starting in . These findings were the first indication that LaMarche's event was not something localized to the White Mountains of California, but was potentially part of a hemispheric phenomenon. Overall there is a body of northern Irish oaks which show a dramatic effect starting a few years before the rare narrowest-ring event of . Having isolated the event, it seems reasonable to look at tree-ring patterns from that immediate period in order to obtain the clearest picture of what happened to the trees in question.
IRISH OAK RING PATTERNS
It is immediately clear that not all trees show a dramatic effect. This is almost certainly due to the fact that some trees were growing on mineral soils while others were growing rooted on peat. Trees growing on mineral soils should be much less susceptible to down-turns in growing conditions than the marginal trees rooted on deep peat. Looking at a selection of trees from the north of Ireland it is fairly clear that the real downturn is around . Fig. 1 shows some of the original Garry Bog trees together with a mean chronology for nine of the trees which show the most dramatic effects. There is a very definite drop in mean ring width after .
Since the initial observation of the event in the Irish trees in 1985-6 it has been possible to look at other chronologies across the same period from England, from Germany and from further south in Ireland. The results are presented below.
From 1985-88 efforts were being made to outline a prehistoric oak chronology in England. Material was drawn from two main areas, East Anglia and Lancashire. In both areas the oaks were preserved in peat and in the case of the Lancashire material it is fairly clear that these are bog oaks in the Irish sense, i.e. they were rooted on peat. The situation with the oaks from East Anglia is not so clear and at least some of them may have grown on mineral soils with subsequent burial in peat.
The approach to building the English chronology was to construct floating chronologies and to date these against the pre-existing Irish and German chronologies. This proved highly successful and a continuous chronology is now available back to (Baillie and Brown 1988). One interesting finding from this work was the observation that all of the East Anglian oaks, which could be cross-dated, fell into one continuous chronology spanning to . This exactly paralleled an independent chronology from the main Lancashire site, Croston Moss, which spanned to . In this case an extension of the Croston Moss chronology ran on from to .
Here then are two independent chronologies which are neatly bracketed by two of the narrowest-ring events in the Irish oak chronology (the dust-veil events in the and ). In the case of Croston Moss the only break in an otherwise continuous chronology, over 22 centuries long, falls across the . Clearly something happened in the which adversely affected bog oaks in England.
Fortunately one group of bog oaks covering the relevant period was discovered by J. Hillam at Sheffield (Hillam pers. comm.). The Hasholme chronology spans to and shows an extremely narrow band of rings in the (curve (a) in Fig. 2).
At the time of writing there are five oak chronologies from Germany relevant to this period. Of these, four show definite evidence for reduced growth at exactly the same time as the Irish and Hasholme trees (Fig. 2, (b)-(e)). These chronologies were constructed at Köln (Schmidt pers. comm.) and Göttingen (Delorme and Leuschner pers. comm.).
The fifth chronology, which derives from the south of Germany (Becker pers. comm.) shows no dramatic effect in the 1620s, though it does, on average, show narrower rings in the 1620s than in the preceding decade. As the trees included in this southern chronology are known to have grown on mineral soils, this may explain their lack of sensitivity. Referring to Fig. 2, it is clear that the event begins at essentially the same time as the event recorded in the Irish trees. However comparison of Fig. 1 and 2 show that while the Irish ring widths tend to drop after , the English and German trees show 1628 as already a narrow ring. By summing the year-to-year trends in each area and converting the excess trend to a percentage, it is possible to mean two equally weighted data sets. The "Trend Mean" curve in Fig. 2 is therefore an "average" picture of the trend in oak growth over the relevant period. Again it is clear that the drop in oak is dramatic from on.
Most of the bog oak material used in the construction of the Belfast 7200-year chronology was derived from sources in the north of Ireland. However, in the course of dating oak timbers for interested archaeologists and others, several timbers have come to light which span the . Fig. 3 shows three of these timbers.
Timber (a) is a single, definite bog oak from Derrycashel, Co. Longford. While its mean ring width is around 1.0 mm until , thereafter its ring width decreases steadily. By its rings are less than 0.5 mm in width and this continues until . Here we are seeing a random sample from a new area going into decline after .
Timber (b) is another random sample, this time from a bog trackway at Corlona, Co. Longford. Since in this case the sample is archaeological we might expect that it would originally have grown on mineral soil. Again its mean ring width drops to less than 0.5 mm by and does not recover for some twelve years.
Timber (c), from another trackway at Timahoe, Co. Kildare, repeats the phenomenon. This timber never recovers from whatever happened in the 1620s and its ring width averaged less than 0.5 mm for the following century.
SUMMARY AND CONCLUSION
As has been demonstrated, the saw a traumatic episode for oaks not just in Ireland but in England and Germany. Given that the oaks studied by dendrochronologists represent only a tiny random sample of the oaks which were growing at the time, we can conclude that there must have been a very general reduction in oak productivity. We can also see that the onset of this reduction is around . If LaMarche's date is superimposed on Fig. 1-3 it is clear that he specified a year not just of trauma for bristlecone pines but for European oaks as well.
LaMarche's frost rings are strongly correlated with dust veil events. Irish narrowest-ring events are also strongly correlated with dust-veil events by way of the Greenland ice-core acidity dates. Hammer et al. see the most prominent Dye 3 acidity peak for several centuries at ± 20. It is hard to imagine that there can be any other explanation for the accumulated tree-ring effects.
This brings us to the question of mechanism. What is the link between a large volcanic dust veil and reduced oak growth in northern Europe? Sear et al. (1987) have shown that there is noticeable cooling in the northern hemisphere for many months following major eruptions in recent times. Kelly and Sear (1985) have also shown significant low-pressure anomalies over northern Europe after major eruptions. Given that bog trees and some trees from mineral soils are badly affected by the conditions, it seems increasingly likely that the oaks were responding to raised water tables. This would account directly for the growth reduction in bog oaks and in those oaks on mineral soils susceptible to flooding.
So, we have to recognize the general problem of attempting to date volcanic events by dendrochronology. Since the likely mechanism is cooling, giving rise to climatic upset, giving rise to local difficulties for trees, there may be a time lag between the eruption and what is seen in the ring patterns. This possibility was addressed by LaMarche when he suggested that the eruption which caused the frost effect in bristlecone pines may have actually taken place one or two years before . European oaks might again be expected to suffer a time lag, especially if the mechanism did involve the raising of water tables. Additional attractions of this mechanism are that it would allow different areas to be affected at slightly different times — which we appear to see — with the overall effect lasting for a number of years. It might also explain the "damage" effects suggested by Fig. 3 if water table alterations actually killed the deeper root systems of some trees.
So while the visible effects in the European trees suggest something happening to the trees around it has to be recognized that the actual date of the eruption could be either in that year or one or two years before.
The way to answer this question is almost certainly the detailed analysis of stable isotope ratios (deuterium and carbon-13) in the annual rings of trees which grew across the later . Since stable isotopes might give some direct measure of cooling associated with particular growth rings, study of trees from a wide geographical area might give a clear picture of the actual conditions prevailing from the 1630s to and answer the question of the exact date of the cooling associated with the eruption. Preliminary results of analyses by Ramesh (pers. comm.) have shown that there is relevant information in Irish tree rings of the period.
If we review the "narrowest ring" story as it developed after 1986 it is possible to see the strength of the circumstantial case for a highly significant event at . Until 1986 only LaMarche and Hirschboeck's 1627 frost ring and Hammer et al.'s 210 ± 30, 260 ± 30, 1100 ± 50 and 1390 ± 50 dates were available to assess volcanic impact in the period to .
In 1986, looking for support or otherwise for LaMarche and Hirschboeck's date, three narrowest-ring events were identified in Irish oaks with start dates at , and . The coincidence of these and other events with Hammer et al.'s date ranges for acidity layers in Greenland suggested strongly that the Irish trees had been recording volcanic events.
In 1987 Hammer et al. published their new ± 20 acidity-layer date and withdrew their former 1390 date.
In 1988 the Irish event dates were published. The relevance of two additional lines of evidence has since become apparent. (1) LaMarche added another frost-ring event at . (2) Pang and Chou (1985 and pers. comm.) provided Chinese Dynastic-change dates. Their evidence included famines, , obscured stars, and Dynastic change, .
This extremely tight clustering of observations in the decade suggests a link between dust veils and the concept of the "Mandate of Heaven" associated with Chinese Dynastic change. When it is realized that suggested dates for the start and end of the Shang Dynasty can be found at variously and (Pang and Chou 1985) and (Kuniholm, this volume) and (Bernal pers. comm.) respectively, it is clear that the accumulated bristlecone, Irish oak, ice-core and Chinese evidence clusters dramatically.
Since in the all the evidence clusters within a mere 10 years, it can be inferred that any lack of precise correlation in the is simply due to the poorer dating resolution associated with greater age in all but the tree-ring evidence.
Such is the case for a hemispheric — volcano-related — environmental event at . How does this relate to the Thera debate? It is clear from the proceedings of the Third Thera and the Aegean World Congress that Thera is the only eruption known to occur in the bracket . None of the other suggested candidate eruptions have been proven to lie within this bracket. Thus, notwithstanding the volcanological assessment that Thera's sulphur output is incompatible with the 1644 ± 20 ice acidity, it must remain a possibility that Thera (or Thera in combination with some other eruption) caused the ice-core signal and contributed to the event. It is this strong circumstantial case, only available within the last few years, which prompts the suggestion that " should become the working hypothesis for the date of the Theran eruption until proven otherwise."
Figures mentioned in this paper
Ring width plots of four trees from Garry Bog. Co. Antrim. Shaded areas indicate reduced growth in the . The mean curve is derived from nine trees which all show similar growth reduction. Arrows indicate when six trees on four bogs exhibit narrowest growth rings.
Rings width plots for: (a) Hasholme, England (Hillam); (b) Bronzes 9, Germany (Schmidt); (c) Baum/ge 13, Germany (Schmidt); (d) Göttingen, Germany (Lauschner and Delorme); (e) Rammsee 5, Germany (Schmidt); showing reduced growth in the . The lower curve is a mean trend European master for the period where Irish and English/German components carry equal weight.
Ring width plots for three trees from southern Ireland showing massive ring width reduction following . (a) Derrycashel, Co. Longford; (b) Corlona, Co. Longford; (c) Timahoe Co. Kildare.