G 3. AN ELECTRONIC JOURNAL OF THE EARTH SCIENCES Published by AGU and the Geochemical Society

Transcription

1 Geosystems G 3 AN ELECTRONIC JOURNAL OF THE EARTH SCIENCES Published by AGU and the Geochemical Society Research Letter Volume 5, Number 3 20 March 2004 Q03005, doi: ISSN: Identification of Aniakchak (Alaska) tephra in Greenland ice core challenges the 1645 BC date for Minoan eruption of Santorini Nicholas J. G. Pearce Institute of Geography and Earth Sciences, University of Wales, Aberystwyth, SY23 3DB, U.K. (nick.pearce@aber.ac.uk) John A. Westgate and Shari J. Preece Department of Geology, University of Toronto, Toronto, Ontario, M5S 3B1, Canada Warren J. Eastwood School of Geography, Earth and Environmental Sciences, University of Birmingham, Edgbaston, Birmingham, B15 2TT, U.K. William T. Perkins Institute of Geography and Earth Sciences, University of Wales, Aberystwyth, SY23 3DB, U.K. [1] Minute shards of volcanic glass recovered from the 1645 ± 4 BC layer in the Greenland GRIP ice core have recently been claimed to originate from the Minoan eruption of Santorini [Hammer et al., 2003]. This is a significant claim because a precise age for the Minoan eruption provides an important time constraint on the evolution of civilizations in the Eastern Mediterranean. There are however significant differences between the concentrations of SiO 2,TiO 2, MgO, Ba, Sr, Nb and LREE between the ice core glass and the Minoan eruption, such that they cannot be correlatives. New chemical analyses of tephra from the Late Holocene eruption of the Aniakchak Volcano in Alaska, however, show a remarkable similarity to the ice core glass for all elements, and this eruption is proposed as the most likely source of the glass in the GRIP ice core. This provides a precise date of 1645 BC for the eruption of Aniakchak and is the first firm identification of Alaskan tephra in the Greenland ice cores. The age of the Minoan eruption of Santorini, however, remains unresolved. Components: 6520 words, 2 figures, 4 tables. Keywords: Aniakchak; tephra; Minoa; Santirini; ice core. Index Terms: 8404 Volcanology: Ash deposits; 1065 Geochemistry: Trace elements (3670); 3655 Mineralogy and Petrology: Major element composition. Received 4 December 2003; Revised 28 January 2004; Accepted 2 February 2004; Published 20 March Pearce, N. J. G., J. A. Westgate, S. J. Preece, W. J. Eastwood, and W. T. Perkins (2004), Identification of Aniakchak (Alaska) tephra in Greenland ice core challenges the 1645 BC date for Minoan eruption of Santorini, Geochem. Geophys. Geosyst., 5, Q03005, doi:. 1. Introduction [2] Currently, debate exists over the exact date for the Minoan eruption of the Santorini (Thera) volcano in the Second millennium BC. This date is of major importance in placing time constraints on the evolution of civilizations in the Eastern Mediterranean. While 14 C methods give calibrated ages for Copyright 2004 by the American Geophysical Union 1 of 10

2 the eruption around 1600 BC, these are not sufficiently precise to be of value in synchronizing the evolution of civilizations in this region. Thus to obtain a more precise eruption date, features of several proxy records have been related to the Minoan eruption, including the accurate dating of ice core acid peaks and tree-ring growth anomalies [Hammer et al., 1987; Kuniholm et al., 1996; Manning et al., 2002, 2001]. These proxy records, which can often be correlated over large distances [Manning and Sewell, 2002], reflect major and widespread environmental impacts; they do not, however, identify unequivocally the cause of a particular environmental disturbance. Combining tree-ring frost records, the acid peaks in ice cores and 14 C dating, a range of dates anywhere between BC has been proposed for the Minoan eruption, and the relative errors/merits of these ages have been widely argued [Hammer et al., 1987; Manning, 1998; Manning and Sewell, 2002; Wiener, 2003; Zelinski and Germani, 1998]. To be certain of linking a particular eruption with the ice core record would require the presence within the ice of juvenile volcanic materials which could be analyzed and correlated with a particular source [Zelinski and Germani, 1998]. Recently, analyses have been published of a series of minute (5 mm) glass shards recovered from the GRIP ice core from Greenland [Hammer et al., 2003]. These analyses are claimed to confirm a 1645 ± 4 BC age for the Minoan eruption of Santorini, however there are serious problems with the correlations suggested by Hammer et al. [2003]. The archaeological implications of this claim are discussed by Wiener [2003] and the interpretations of Hammer et al. [2003] have already been refuted on statistical grounds [Keenan, 2003]. Here, the analytical data of Hammer et al. [2003] are considered alongside new analyses of tephra from the Aniakchak volcano on the Alaska Peninsula, which is proposed as the source for the glass shards from this layer of the GRIP ice core. 2. Material From the GRIP Ice Core [3] Hammer et al. [2003] recovered hundreds of microparticles from the 1645 BC acid spike layer of the GRIP ice core (sample A1340-7) of which 174 analyzed grains were silica-rich volcanic glass. Major element analyses were performed using an analytical scanning electron microscope (ASEM) for which details are not given, although Hammer et al. [2003] state that the small sizes and irregularity of the particles prevents ideal calibration. Hammer et al. [2003] also analyzed 8 larger (about 10 mm) shards by secondary ion mass spectrometry (SIMS) using a Cameca ion microprobe for a range of trace elements. At the same time they analyzed 38 grains of glass for major elements from a comminuted sample of the Upper Pumice layer of the Minoan eruption (Bo-1) collected from Santorini, but for this material, only 3 grains were analyzed for their trace elements. These data are presented in Table 1, with the major element analyses having been normalized to an anhydrous basis. 3. Comparisons of the Major Element Chemistry [4] Comparing the ASEM major element chemistry from Hammer et al. [2003] of the A ice core glass with the Bo-1 Minoan ash shows large differences in composition (see Table 1) although in many cases these are within the large errors of their analyses. Significant differences in composition exist for SiO 2 which is 3.6 wt% higher in Bo-1 than A1340-7, Al 2 O 3 is 0.8wt% higher in A1340-7, FeOt (total Fe represented as FeO) and TiO 2 are both 50% higher in A than Bo-1, MgO is almost twice as high in A1340-7, and Na 2 O is marginally higher in A1340-7, although due to the small sample size and irregular shape, Na is unlikely to have been determined accurately [Hunt and Hill, 2001]. As many rhyolitic eruptions are similar in major element chemistry [Westgate et al., 1994], broad or general similarities between glass from separate deposits are insufficient for accurate correlation. In such tephrochronological studies, the major element chemistry must be essentially identical and, as such, acceptable variations are 1 2% relative for SiO 2 and Al 2 O 3, and in the 5 10% range for the remaining, less abundant elements [Pearce et al., 2004]. The differences in the data from Hammer et al. [2003] are too great to allow a 2of10

3 Table 1 (Representative Sample). Major and Trace Element Data for the Bo-1 Glass From Santorini, the Minoan Ash From Gölhisar Gölü, SW Turkey, A From the GRIP Ice Core, Hayes Volcano (UT710), Veniaminof (Samples 5Rj... ), Mount St. Helens Y Tephra, and Aniakchak (UT2011) a (The full Table 1 is available in the HTML version of this article at Hammer et al. [2003] Eastwood et al. [1999] This Study GRIP Ice Core A (n = 174) Minoan Bo-1, Santorini (n = 38) Relative Difference A1340-7/ Bo-1 Minoan, Gölhisar Gölü (n = 67) Aniakchak UT2011 (n = 14) Hayes Volcano, UT710 A Calibrated and Normalized Aniakchak Normalized Relative Difference A1340-7/ Aniakchak SiO (1.82) (1.62) 5% (0.59) (0.19) % TiO (0.60) 0.60 (0.52) 49% 0.29 (0.06) 0.47 (0.08) % Al2O (0.98) (0.84) 6% (0.18) (0.11) % FeO 3.34 (1.11) 2.22 (0.88) 50% 2.04 (0.12) 2.45 (0.08) % MnO 0.36 (0.40) 0.18 (0.29) 100% 0.07 (0.04) 0.13 (0.05) % MgO 0.63 (0.48) 0.34 (0.33) 83% 0.28 (0.04) 0.52 (0.02) % CaO 2.12 (0.60) 1.82 (0.55) 17% 1.40 (0.10) 1.74 (0.09) % Na 2 O 3.74 (0.88) 3.23 (1.21) 16% 4.76 (0.64) 5.13 (0.13) % K2O 3.64 (0.54) 3.73 (0.96) 3% 3.24 (0.12) 2.93 (0.10) % Cl 0.31 (0.04) 0.20 (0.03) 0.4 F 0.07 (0.09) O=F,Cl 0.09 (0.04) Cr2O (0.46) 0.17 (0.24) Total H 2 O diff (1.53) a Major and trace element data for the Bo-1 glass from Santorini [Hammer et al., 2003], the Minoan ash from Gölhisar Gölü, SW Turkey [Eastwood et al., 1998; Eastwood et al., 1999; Pearce et al., 2002], A from the GRIP ice core [Hammer et al., 2003], Hayes volcano (UT710, this study and [Riehle, 1994], Veniaminof (samples 5Rj... [Riehle et al., 1999]), Mount St Helens Y tephra [Westgate, 1977] and Aniakchak (UT2011, this study). All analyses have been recalculated to 100% excluding H2O. The A analyses have been recalibrated using the reported Bo-1 analyses from Hammer et al. [2003] and the analyses for the Minoan tephra from Eastwood et al. [1999] for those elements common to both data sets, and the UT2011 analyses have been recalculated to 100% for the same elements for comparison. Numbers in parentheses are analytical precision at ±1s where available. 3of10

4 correlation to be proposed between the ice core glass and the Minoan Bo-1 deposit. Keenan [2003] has used a statistical method involving t-tests on the standard errors (not standard deviations) of the analyses of Hammer et al. [2003] to show that the Minoan Bo-1 sample and the A glass cannot be the same, and the interpretation here, based purely on the compositional data, reinforces this conclusion. 4. Wider Correlation of the GRIP Ice Core Glass [5] The major element analyses reported by Hammer et al. [2003] for the Bo-1 Minoan tephra are similar in SiO 2,Al 2 O 3, FeOt, MgO and CaO to published Minoan tephra analyses [Eastwood et al., 1998, 1999; Vinci, 1984; Vitaliano et al., 1990; Warren and Puchelt, 1990], but are generally higher in the minor elements (see Table 1). These differences probably arise from calibration problems in the ASEM analyses (irregular shard geometry may generate problems with corrections for mass absorption, fluorescence etc). The same relative errors will also be present in the analyses of the glass from the GRIP ice core. To enable a more accurate comparison with other likely sources for the ice core glass, a correction factor can be calculated for the ASEM major element analyses, using the 38 ASEM analyses of the Bo-1 Minoan glass and one of the published electron microprobe analyses (EPMA) of the Minoan tephra [Eastwood et al., 1999; Vinci, 1984; Vitaliano et al., 1990]. This has been done using the average of 59 wavelength dispersive spectrometry (WDS) EPMA major element determinations of the Minoan glass recovered from Gölhisar Gölü in southwest Turkey [Eastwood et al., 1999] to give a corrected concentration for the glass in A This has then been normalized to 100% for those elements common to both the Gölhisar Gölü and Bo-1 data sets (i.e., excluding F, Cl and Cr 2 O 3 ). The choice of published analyses for normalization makes only minute differences to the corrected A data. This recalibrated analysis can now be compared with data from other known volcanic sources active at around 1645 BC. [6] Several large volcanic eruptions occurred in North America in the early to mid-second Millennium BC, their ages making them possible correlatives of the glass from the 1645 BC layer of the GRIP ice core. Veniaminof, on the Alaska Peninsula, underwent a large (>50 km 3 ) caldera forming eruption at about 3700 ± C year B.P. [Miller and Smith, 1987] which gives a 2s calibrated age [Stuiver and Reimer, 1993; Stuiver et al., 1998] of between BC. Aniakchak, again on the Alaska Peninsula, also underwent a caldera-forming eruption at this time, for which a range of dates has been produced. These include 3430 ± C years BP [Miller and Smith, 1987], 3380 ± C years BP [Vogel et al., 1990] and the precise age of 3345 ± C years BP [Begét et al., 1992], which give 2s calibrated age ranges of between BC, BC and BC respectively. Aniakchak erupted more than 50 km 3 of magma [Miller and Smith, 1987], and tephra from this eruption traveled northeast across Alaska [Begét et al., 1992]. The Hayes volcano, Alaska, erupted a series of 7, or possibly 8, closely related tephra beds at 3650 ± C year BP [Riehle et al., 1990], or at 3660 ± C years BP [Begét et al., 1991], which give 2s calibrated age ranges of BC and BC respectively. The Hayes tephras are estimated to have a total volume of approximately 10 km 3, these also being deposited predominantly to the northeast of the source [Riehle, 1994; Riehle et al., 1990]. The Y tephra set (Yb, Yn and Ye tephras) from Mount St. Helens was erupted in a series of events approximately 1560 BC [Pringle, 1993] with a bracketed 14 C date at c BP (average 3410 ± 80 BP) [Luckman et al., 1986] giving a 2s calibrated age of BC, although these eruptions involved only approximately 4 km 3 of magma. The Yn tephra of this set was blown north-east and was deposited over a wide area of north-western Canada [Westgate, 1977]. [7] Major element analyses of tephras from Hayes volcano [Riehle, 1994] and UT710, this study), Mount St. Helens [Westgate, 1977], Aniakchak (UT2011, this study) and Veniaminof (recalculated to an anhydrous basis and Fe 2 O 3 converted to FeOt from [Riehle et al., 1999]) are listed for comparison 4of10

5 in Table 1. It is clear that the Aniakchak tephra (UT2011) and the recalibrated ice core glass (A1340-7) are extremely similar, showing only minor differences (<8% relative) for all elements except FeOt, which shows a 25% relative difference. FeOt is the most imprecisely determined of the ASEM major elements (i.e., >1 wt%), and has a relative standard deviation of around ±40% in Bo-1. This may in part be responsible for the larger difference in FeOt between A and UT2011. In all cases the 1s error on the ASEM analyses of A encompasses the analysis of UT2011. The major element chemistry strongly suggests that the glass in the ice core is derived from Aniakchak. All the Hayes volcano and Mount St. Helens tephra beds of the appropriate age are too siliceous to correlate with A1340-7; differences in FeOt, alkalis and Al 2 O 3 likewise exclude the possibility of correlation. The andesitic character of the Veniaminof samples rules out a correlation with the ice core glass. 5. Trace Element Comparisons [8] Hammer et al. [2003] produced ion microprobe (SIMS) trace element analyses of 8 glass shards from the A ice core sample and compared these with 3 analyses of glass from Bo-1 and these are listed in Table 2. In such small samples this is a remarkable achievement. Unfortunately, the analysis of only 3 grains of Minoan material is statistically insufficient to gain an impression of the accuracy of the SIMS analyses when the heterogeneity of the Minoan pumice, as shown by the large inter-shard variation described recently [Pearce et al., 2002], is considered [cf. Keenan, 2003]. Despite this observation, the trace element analyses of Hammer et al. [2003] for the Bo-1 glass are broadly comparable with other published analyses of Minoan material for many elements [Eastwood et al., 1999; Pearce et al., 2002; Vinci, 1984; Vitaliano et al., 1990; Warren and Puchelt, 1990]. As with the major elements, there are again considerable differences between the trace element analyses of Bo-1 and A given by Hammer et al. [2003], with factors of up to 2.6 for differences in Sr, around 1.5 for Ba, Sm, Nb and Rb and around 1.2 for the LREE present in the data set. Differences in trace element concentrations of this magnitude rule out a possible correlation [Pearce et al., 2004]. At the reported concentrations (a few tens to a few hundred ppm), elements such as Ba, Sr, Rb and the LREE should be determined with good accuracy and precision (perhaps around ±10%) even in such challenging materials, but the differences between the Bo-1 and A analyses are well beyond any realistic analytical errors, and far too great for these deposits to be considered the same. Thus a simple consideration of the relative differences between the ice core glass and the Minoan tephra can be used to rule out a relationship between these samples. This difference has also been demonstrated using t-test statistics on the standard errors of the analyses presented by Hammer et al. [Keenan, 2003]. Nonetheless, Hammer et al. [2003], believe these data sets show a remarkable resemblance. [9] The differences between A and the Minoan ash are further highlighted when compared with 56 single glass shard analyses of the Minoan ash performed by laser ablation ICP-MS [Pearce et al., 2002] (Table 2). The most apparent differences are in the concentrations of Ba, Sr, Rb and the LREE. It is clear that the REE concentrations determined by Hammer et al. (Bo-1) and Pearce et al. (Gölhisar Gölü) for the Minoan deposit are similar (Table 1), but there is a marked difference in slope of the LREE between the Minoan sample and the ice core glass (Figure 1 and Table 2), the latter having a much lower La/Sm ratio. For robust correlations to be made, it is vital that incompatible trace element ratios are identical between samples and that the shape of the REE profiles is the same [Pearce et al., 2004]. [10] Table 2 also presents new trace element analyses of a sample from the Hayes volcano (UT710) and Aniakchak (UT2011). The Hayes volcano glass has much higher Sr than the A glass and lower REE, with a very steep REE profile (defining this as Type II Alaskan volcano [Preece et al., 1999]), and thus the ice core glass is again shown not to be from Hayes volcano. In contrast, however, is the data for the Aniakchak tephra (UT2011), which has almost identical Sr, Rb, Zr, REE to A1340-7, as well 5of10

6 Table 2. Trace Element Analyses of Bo-1 Glass From Santorini, the Minoan Ash From Gölhisar Gölü, SW Turkey, A From the GRIP Ice Core, Hayes Volcano (UT710), and Aniakchak (UT2011) a Hammer et al. GRIP Ice Core A (n = 8) SIMS Hammer et al. Santorini Bo-1 (n = 3) SIMS Pearce et al. Minoan Gölhisar Gölü (n = 56) LA-ICP-MS Pearce et al. Individual Shard, Range (min/max) Aniakchak UT2011 (n = 4) Solution ICP-MS Hayes UT710 INAA Eastwood et al. Minoan Gölhisar Gölü Solution ICP-MS Sc 13.9 (6.8) 2.57/ (0.65) V 12 (2) 13 (2) 12.2 (0.20) 39.4 Cr 9.6 (1.7) 1.7 (0.5) 2.02 (0.16) 6.20 Rb 53 (9) 77 (12) 113 (21.9) 56.4/ (2.29) Sr 190 (29) 72 (11) 55.5 (9.17) 37.1/ (3.62) Y 35 (5) 32.5 (5) 35.4 (6.33) 24.3/ (2.54) 35.4 Zr 253 (39) 292 (44) 281 (48.8) 201/ (3.92) 260 Nb 21 (4) 15 (3) 9.85 (1.91) 6.58/ (0.15) 11.2 Cs 3.8 (0.9) 4 (0.9) 3.46 (0.81) 1.87/ (0.11) Ba 1020 (150) 690 (100) 450 (79.6) 297/ (12.5) La 24.5 (3.8) 29.5 (4.5) 31.3 (6.24) 20.2/ (0.19) Ce 49.4 (7.5) 60.5 (9.2) 49.3 (8.40) 32.4/ (1.96) Pr 7.4 (1.2) 7.1 (1.2) 5.88 (1.24) 3.28/ (0.19) 6.62 Nd 28 (4) 25.2 (3.9) 24.0 (4.91) 15.5/ (0.09) Sm 8.2 (1.6) 5.3 (1.2) 5.26 (1.37) 2.49/ (0.24) Eu 0.78 (0.28) 0.00/ (0.10) Gd 5.5 (1.6) 5.8 (1.4) 5.22 (1.73) 2.43/ (0.20) 6.48 Tb 0.91 (0.31) 1.1 (0.3) 0.98 (0.32) 0.00/ (0.06) Dy 8.1 (1.4) 6.7 (1.2) 6.46 (1.61) 3.31/ (0.19) 6.79 Ho 1.7 (0.4) 1.4 (0.3) 1.52 (0.32) 0.99/ (0.08) 1.55 Er 5.5 (1) 5.1 (0.9) 4.41 (1.21) 2.53/ (0.08) 4.53 Tm 0.59 (0.2) 0.94 (0.24) 0.72 (0.28) 0.30/ (0.03) 0.72 Yb 5.3 (1.1) 4.3 (0.9) 5.17 (1.75) 1.59/ (0.19) Lu 0.89 (0.34) 0.30/ (0.02) Hf 6.88 (2.34) 3.39/ (0.20) Ta 0.99 (0.87) 0.24/ (0.01) Th 16.2 (3.79) 8.73/ (0.07) U 5.39 (1.05) 3.30/ (0.12) Ba/Rb Rb/Sr La/Nb La/Sm a Trace element analyses of Bo-1 glass from Santorini [Hammer et al., 2003], the Minoan ash from Gölhisar Gölü, SW Turkey [Eastwood et al., 1998, 1999; Pearce et al., 2002], A from the GRIP ice core [Hammer et al., 2003], Hayes volcano (UT710, this study), and Aniakchak (UT2011, this study). Figures in parentheses for the data of Hammer et al. [2003] are their 1 standard deviation error, and for the LA-ICP-MS data of Pearce et al. [2002] are ±1 standard deviation of all single shard analyses, which represents the range of the data and is not a measure of analytical precision, which would be approximately ±5% for Ba and Zr; ±20% for Tb, Ho, Tm, Lu and Ta; and ±10% for the remaining elements. All concentrations in ppm. as very similar Ba, Nb, Y and Cs, particularly when some allowance is made for the differences in the SIMS analyses of Bo-1 when compared to other analyses of Minoan tephra [Eastwood et al., 1999; Pearce et al., 2002; Vinci, 1984; Vitaliano et al., 1990; Warren and Puchelt, 1990]. The Aniakchak REE analyses are compared with the A analyses in Figure 1, where it is evident that the 2 data sets have an indistinguishable REE profile, markedly different from the Minoan tephra. Figure 2 shows an incompatible element spidergram that compares the Minoan and Aniakchak tephras with the ice core glass. Again, the Aniakchak tephra and A glass are identical, and they both differ considerably in Rb, Sr, Ba, Ti, and Sm from the Minoan tephra. 6. Statistical Distance [11] An appropriate statistical method to explore the possibility of a correlation between tephra beds uses a measure of the statistical distance between pairs of samples. This methods has been developed by Perkins et al. [1998, 1995] in their correlations 6of10

7 Figure 1. Chondrite normalized REE spidergrams for glass from the GRIP ice core (A1340-7) [Hammer et al., 2003], the Minoan tephra deposited at Gölhisar Gölü, Turkey [Pearce et al., 2002] and tephra from the caldera forming eruption of Aniakchak (UT2011, this study). Normalisation factors from Sun and McDonough [1989]. of silicic tephras. Perkins et al. [1998] give the distance formula as D 2 ¼ Xn k¼1 ðx k1 x k2 Þ 2 = s 2 k1 þ s2 k2 where n is the number of elements used in the comparison, x k1 and x k2 are the concentrations of ð1þ the kth element in the first and second samples, and s k1 and s k2 are the precisions of the determinations of the kth element in each sample (i.e., the standard deviations of analyses). Perkins et al. [1995] have shown that the calculated value of D 2 has a Chisquared distribution amongst compositionally identical samples. Here, this method is applied to Figure 2. Chondrite normalized incompatible element spidergram for glass from the GRIP ice core (A1340-7) [Hammer et al., 2003], the Minoan tephra deposited at Gölhisar Gölü, Turkey [Pearce et al., 2002] and Aniakchak (UT2011, this study). Normalization factors from Thompson [1982]. 7of10

8 Table 3. Distance Function Calculations for a Range of Possible Correlative Pairs of Samples in This Study for the 12 Trace Elements Rb, Sr, Zr, Nb, Y, Cs, Ba, La, Ce, Pr, Nd and Sm a Sample Pairs D 2 Calculated Accept/Reject Ho (D 2 critical 99% = 26.22) Bo-1 (Hammer et al.) and Gölhisar Gölü (17.46) Accept, samples are possible correlatives LA-ICP-MS (Pearce et al.) Bo-1 (Hammer et al.) and A (Hammer et al.) Reject, samples are statistically different Bo-1 (Hammer et al.) and UT2011 Aniakchak (this study) Reject, samples are statistically different A (Hammer et al.) and UT2011 Aniakchak (this study) Accept, samples are possible correlatives A (Hammer et al.) and Gölhisar Gölü (67.02) Reject, samples are statistically different LA-ICP-MS (Pearce et al.) UT2011 Aniakchak (this study) and Gölhisar Gölü LA-ICP-MS (Pearce et al.) (810.9) Reject, samples are statistically different a The critical value of D 2 at 99% confidence level is 26.22, and thus where the D 2 calculated >D 2 critical, the null hypothesis, that the two samples are identical, can be rejected. For those comparisons including the LA-ICP-MS analyses from Gölhisar Gölü, the first value of D 2 is calculated using the standard deviation of all the single shard analyses (which reflects the natural variation between individual shards, not the analytical precision) and the D 2 figure in parentheses is calculated using an estimate of the analytical error for the LA-ICP-MS analyses (i.e., 10% relative for all elements, except Zr and Ba at 5% [see Pearce et al., 2004]). comparisons between the trace element analyses given by Hammer et al. [2003], Pearce et al. [2002] and the new analyses of the Aniakchak tephra. The heavy REE have been excluded from these calculations due to the relatively large uncertainties associated with their analyses, and in the LA-ICP-MS data from Pearce et al. [2002] the standard deviation of all single shard analyses (which reflects the large, natural, inter-shard trace element variation) as well as an estimate of the true analytical error [see Pearce et al., 2004] have both been used (see Table 3). The results are presented in Table 3, and show, at the 99% confidence level, that neither the Bo-1 [Hammer et al., 2003] nor the Gölhisar Gölü [Pearce et al., 2004] samples can be correlated with the ice core glass. The data for the Aniakchak tephra (UT2011) and the ice core glass can however, by this analysis, be regarded as possible correlatives. Calculations of statistical distance could not be applied to the major element data because of the large standard deviations present in the Hammer et al. analyses, which gave no discriminations between any sample pairs. 7. Conclusions [12] Careful consideration of the major and trace element analyses of glass shards from the GRIP ice core shows that they are compositionally distinct from the Minoan tephra from Santorini. Comparison with new analyses clearly demonstrates that the ice core glass from A was derived from Aniakchak. The chronology of the ice core defined by Hammer et al. [2003] thus places a firm date of 1645 ± 4 BC on the caldera-forming eruption of Aniakchak. This agrees very well with the published age ranges for the Aniakchak eruption from 14 C dating, and particularly with the calibrated age of BC from Begét et al. [1992]. The eruption of Aniakchak at 1645 BC is also consistent with the suggestion that the acid volcanic signal recorded in the GRIP, North Grip and Dye 3 ice cores is the result of a major volcanic eruption south of Greenland, but north of 30 N [Hammer et al., 2003]. The presence of glass shards trapped within the ice core in this case directly links the Aniakchak eruption with the acid peak at 1645 BC. Furthermore, this is thought to be the first firm identification of an Alaskan tephra in the Greenland Ice Sheet, and clearly, the techniques and potential exist for identification of more such deposits. Whilst this identification provides an accurate and precise age for the Aniakchak eruption, and a source for one of the acid spikes in the Greenland ice cores, it leaves the controversy on the age of the Minoan eruption unresolved. 8. Methods [13] Major element analyses of the UT710 (Hayes tephra) were performed using an ETEC EDS electron microprobe at the University of Toronto. Glass separated from the Aniakchak tephra (UT2011) was mounted in a resin block, polished 8of10

9 Table 4. Comparison of Certified Concentrations for USGS Reference Materials QLO-1 and RGM-1 With Values Determined From Analyses of These Materials Performed Alongside UT2011, on 2 Separate Occasions, 3 Months Apart, as Part of This Study a QLO-1 Accepted Conc. QLO-1 This Study (n = 2) RGM-1 Accepted Conc. RGM-1 This Study (n = 2) Sc (0.23) (0.06) V (4.03) (0.14) Cr (0.02) (0.03) Rb (1.56) (2.83) Sr (17.0) (1.41) Y (1.06) (1.13) Zr (1.41) (3.54) Nb (0.07) (0.05) Cs (0.05) (0.30) Ba (16.3) (6.36) La (0.28) (0.35) Ce (0.14) (1.91) Pr (0.30) (0.18) Nd (0.21) (0.07) Sm (0.19) (0.11) Eu (0.07) (0.01) Gd (0.27) (0.04) Tb (0.01) (0.03) Dy (0.11) (0.09) Ho (0.03) (0.01) Er (0.04) (0.09) Tm (0.03) (0.00) Yb (0.00) (0.17) Lu (0.01) (0.02) Hf (0.00) (0.16) Ta (0.03) (0.03) Th (0.01) (0.21) U (0.04) (0.07) a Numbers in parentheses are analytical precision at ±1s. QLO-1 and RGM-1 concentrations in bold font are certified values, in normal font are suggested values [Govindaraju, 1994]. All concentrations in ppm. and analyzed at the University of Toronto, using a Cameca SX-50 WDS electron microprobe with a 15 mm defocused beam accelerated to 15kV with a 6nA current. Calibration was against a series of mineral and glass standards. Full details of the EPMA methods are given in Pearce et al. [1997]. [14] Trace elements were determined by INAA at the University of Toronto for UT710, and by solution ICP-MS at the University of Wales, Aberystwyth for UT2011. At Aberystwyth, 0.2g of a cleaned glass separate of UT2011 was digested using hot HF/HClO 4, and the solution produced was analyzed by fully quantitative ICP-MS, calibrated using multi-element standards prepared from single element solutions. The Aniakchak tephra sample was analyzed twice on 2 separate occasions, 3 months apart to give the average concentration presented here. On each occasion, alongside the Aniakchak sample, the USGS rhyolitic Certified Reference Materials RGM-1 and QLO-1 were analyzed as a check on data quality. These were prepared at the same time and by the same method as the Aniakchak sample. These analyses are presented in Table 4 and agree extremely well with the certified concentrations given for these materials [Govindaraju, 1994]. [15] Full descriptions of all analytical methods used are given in more detail elsewhere, including EPMA, solution and laser ablation ICP-MS methods [Eastwood et al., 1999; Pearce et al., 2002, 1997, 2004] and INAA analyses [Westgate and Gorton, 1981]. Acknowledgments [16] We thank Tom Ager (USGS, Denver, USA) and Owen Mason (Anchorage, USA) for providing samples of Aniakchak tephra, and Jim Riehle (USGS, Reston, USA) for providing sample UT710 from Hayes volcano. Thanks are due to Jim Begét for alerting us to some 14 C dates we had overlooked and to both he and David Pyle for their positive comments on the first version of this paper. This work was supported by research grants from Natural Sciences and Engineering Research Council, Canada to JAW. References Begét, J. E., R. D. Reger, D. Pinney, T. Gillispie, and K. Campbell (1991), Correlation of the Holocene Jarvis Creek, Tangle Lakes, Cantwell, and Hayes tephras in southcentral Alaska, Quat. Res., 35, Begét, J. E., O. Mason, and P. Anderson (1992), Age, extent and climatic significance of the c BP Aniakchak tephra, western Alaska, USA, Holocene, 2, Eastwood, W. J., N. J. G. Pearce, J. A. Westgate, and W. T. Perkins (1998), Recognition of Santorini (Minoan) tephra in lake sediments from Gölhisar Gölü, southwest Turkey by laser ablation ICP-MS, J. Archaeol. Sci., 25, Eastwood, W. J., et al. (1999), Geochemistry of Santorini tephra in lake sediments from southwest Turkey, Global Planet. Change, 21, Govindaraju, K. (1994), 1994 compilation of working values and sample descriptions for 383 geostandards, Geostandards Newsl., 18, Hammer, C. U., H. B. Clausen, W. L. Freidrich, and H. Tauber (1987), The Minoan eruption of Santorini in Greece dated to 1645 BC?, Nature, 328, of10

10 Hammer, C. U., G. Kurat, P. Hoppe, W. Grum, and H. B. Clausen (2003), Thera eruption date 1645 BC confirmed by new ice core data?, paper presented at the SCIEM EuroConference, SCIEM, Vienna. Hunt, J. B., and P. G. Hill (2001), Tephrological implications of beam size - sample size effects in electron microprobe analyses of glass shards, J. Quat. Sci., 16, Keenan, D. J. (2003), Volcanic ash retrieved from the GRIP ice core is not from Thera, Geochem. Geophys. Geosyst., 4(11), 1097, doi: /2003gc Kuniholm, P. I., et al. (1996), Anatolian tree-rings and the absolute chronology of the eastern Mediterranean BC, Nature, 381, Luckman, B. H., M. S. Kearney, R. H. King, and A. B. Beaudon (1986), Revised 14 C age for St. Helens Y tephra at Tonquin Pass, J. Earth Sci., 23, Manning, S. W. (1998), Correction. New GISP2 ice core evidence supports 17th century BC date for Santorini (Minoan) eruption: Response to Zelinski and Germani (1998), J. Archaeol. Sci., 25, Manning, S. W., and D. A. Sewell (2002), Volcanoes and history: A significant relationship? The case of Santorini, in Natural Disasters and Cultural Change, edited by R.TorranceandJ.P.Grattan,pp ,Routledge, London. Manning, S. W., B. Kromer, P. I. Kuniholm, and M. W. Newton (2001), Anatolian tree-rings and a new chronology for the east Mediterranean Bronze-Iron Ages, Science, 294, Manning, S. W., et al. (2002), New evidence for an early date for the Aegean Late Bronze Age and Thera eruption, Antiquity, 76, Miller, T. P., and R. L. Smith (1987), Late Quaternary calderaforming eruptions in the eastern Aleutian arc, Alaska, Geology, 15, Pearce, N. J. G., et al. (1997), A compilation of new and published major and trace element data for NIST SRM 610 and NIST SRM 612 glass reference materials, Geostandards Newsl., 21, Pearce, N. J. G., W. J. Eastwood, J. A. Westgate, and W. T. Perkins (2002), The composition of juvenile volcanic glass from the c. 3,600 B.P. Minoan eruption of Santorini (Thera), J. Geol. Soc., 159, Pearce, N. J. G., J. A. Westgate, W. T. Perkins, and S. J. Preece (2004), The application of ICP-MS methods to tephrochronological problems, Appl. Geochem., 19, Perkins, M. E., W. P. Nash, F. H. Brown, and R. J. Fleck (1995), Fallout tuffs of Trapper Creek, Idaho: A record of Miocene explosive volcanism in the Snake River Plain volcanic province, Bull. Geol. Soc. Am., 107, Perkins, M. E., F. H. Brown, W. P. Nash, W. McIntosh, and S. K. Williams (1998), Sequence, age, and source of silicic fallout tuffs in middle to late Miocene basins of the northern Basin and Range province, Bull. Geol. Soc. Am., 110, Preece, S. J., J. A. Westgate, B. A. Stemper, and T. L. Péwé (1999), Tephrochronology of late Cenozoic loess at Fairbanks, central Alaska, Geol. Soc. Am. Bull., 111, Pringle, P. T. (1993), Roadside geology of Mount St. Helens National Volcanic Monument and vicinity, Wash. Dept. of Natl. Resour., Olympia, Washington. Riehle, J. R. (1994), Heterogeneity, correlatives and proposed stratigraphic nomenclature of Hayes tephra set H, Alaska, Quat. Res., 41, Riehle, J. R., P. M. Bowers, and T. A. Ager (1990), The Hayes tephra deposits, an Upper Holocene marker horizon in southcentral Alaska, Quat. Res., 33, Riehle, J. R., C. E. Meyer, and R. T. Miyoak (1999), Data on Holocene Tephra (Volcanic Ash) Deposits in the Alaska Peninsula and Lower Cook Inlet Region of the Aleutian Volcanic Arc, Alaska, U.S. Geol. Soc. Open File, Stuiver, M., and P. J. Reimer (1993), Extended C-14 data base and revised CALIB 3. 0 C-14 age calibration program, Radiocarbon, 35, Stuiver, M., et al. (1998), INTCAL98 radiocarbon age calibration, 24,000 0 cal BP, Radiocarbon, 40, Sun, S.-S., and W. F. McDonough (1989), Chemical and isotopic systematics of oceanic basalts: Implications for mantle compositions and processes, in Magmatism in the Ocean Basins, edited by A. D. Saunders and M. J. Norry, Geol. Soc. Spec. Publ., 42, Thompson, R. N. (1982), British Tertiary Volcanic Province Scottish, J. Geol., 18, Vinci, A. (1984), Chemical differences between the island and submarine pumice layers of Thera, Mar. Geol., 55, Vitaliano, C. J., S. R. Taylor, M. D. Norman, M. T. McCulloch, and I. A. Nicholls (1990), Ash layers of the Thera volcanic series: Stratigraphy, petrology and geochemistry, in Thera and the Aegean World III, edited by D. A. Hardy and A. C. Renfrew, pp , The Thera Foundation, London. Vogel, J. S., W. Cornell, D. E. Nelson, and J. R. Southon (1990), Vesuvius/Avellino, one possible source of seventeenth century BC climatic disturbance, Nature, 344, Warren, P. M., and H. Puchelt (1990), Stratified pumice from Bronze Age Knossos, in Thera and the Aegean World III, edited by D. A. Hardy and A. C. Renfrew, pp , The Thera Foundation, London. Westgate, J. A. (1977), Identification and significance of late Holocene tephra from Otter Creek, southern British Columbia, and localities in west-central Alberta, Can. J. Earth Sci., 14, Westgate, J. A., and M. P. Gorton (1981), Correlation techniques in tephra studies, in Tephra Studies, edited by S. Self and R. S. J. Sparks, pp , D. Reidel, Norwell, Mass. Westgate, J. A., W. T. Perkins, R. Fuge, N. J. G. Pearce, and A. G. Wintle (1994), Trace element analysis of volcanic glass shards by laser ablation inductively coupled plasma mass spectrometry: Application to Quaternary tephrochronological studies, Appl. Geochem., 9, Wiener, M. H. (2003), Time out: the current impasse in Bronze age archaeological dating, in METRON. Measuring the Aegean Bronze Age, edited by K. P. Foster and R. Laffineur, pp , Univ. of Tex., Austin. Zelinski, G. A., and M. S. Germani (1998), New ice core evidence challenges the 1620s BC age for the Santorini (Minoan) eruption, J. Archaeol. Sci., 25, of 10