INTRODUCTION
As part of an effort to enhance the time resolution of the present international calibration curve (IntCal13, Reimer et al. Reference Reimer, Bard, Bayliss, Beck, Blackwell, Bronk Ramsey, Buck, Cheng, Edwards and Friedrich2013), and to increase the potential for investigating detailed changes in past solar activity, a substantial amount of single-year tree ring radiocarbon (14C) measurements have recently been produced (e.g. Jacobsson et al. Reference Jacobsson, Hamilton, Cook, Crone, Dunbar, Kinch, Naysmith, Tripney and Xu2017; Miyake et al. Reference Miyake, Masuda, Nakamura, Kimura, Hakozaki, Jull, Lange, Cruz, Panyushkina, Baisan and Salzer2017b; Sakamoto et al. Reference Sakamoto, Hakozaki, Nakao and Nakatsuka2017; Marshall et al. Reference Marshall, Bayliss, Farid, Tyers, Bronk Ramsey, Cook, Doğan, Freeman, İlkmen and Knowles2018). New annual 14C data are important since these give new insights into both the short- and long-term solar behavior, which are important to unraveling the processes governing the solar dynamo and to better understanding the link between solar variability and Earth’s climate (Gray et al. Reference Gray, Beer, Geller, Haigh, Lockwood, Matthes, Cubasch, Fleitmann, Harrison and Hood2010). Equally important, annual 14C data will refine the present calibration curve by providing more detailed structure to the curve itself, which, in turn, will provide new possibilities for high-precision dating of various materials, such as the 14C age calibration of archaeological artifacts (e.g. Pearson et al. Reference Pearson, Brewer, Brown, Heaton, Hodgins, Jull, Lange and Salzer2018). Hence, it is of great importance that new tree-ring data submitted for future calibration curves are of sufficient quality in terms of the robustness of the dendrochronological age constraints, internal and external validation of 14C measurements within each laboratory, and that the new 14C ages are compared with existing 14C data in order to ensure the calibrated ages are reliable and accurate. Furthermore, annual 14C data have contributed to the findings of so-called solar cosmic-ray events, which are characterized by an abrupt increase in the amount of 14C in Earth’s atmosphere for one or several years (Miyake et al. Reference Miyake, Nagaya, Masuda and Nakamura2012, Reference Miyake, Masuda and Nakamura2013, Reference Miyake, Masuda, Hakozaki, Nakamura, Tokanai, Kato, Kimura and Mitsutani2014, Reference Miyake, Jull, Panyushkina, Wacker, Salzer, Baisen, Lange, Cruz, Masuda and Nakamura2017a; Usoskin et al. Reference Usoskin, Kromer, Ludlow, Beer, Friedrich, Kovaltso, Solanki and Wacker2013; Jull et al. Reference Jull, Panyushkina, Lange, Kukarskih, Myglan, Clark, Salzer, Burr and Leavitt2014, Reference Jull, Panyuskina, Miyake, Masuda, Nakamura, Mitsutani, Lange, Cruz, Baisan, Janovic, Varga and Molnár2018; Wacker et al. Reference Wacker, Güttler, Goll, Hurni, Synal and Walti2014; Güttler et al. Reference Güttler, Adolphi, Beer, Bleicher, Boswijk, Christl, Hogg, Palmer, Vockenhuber, Wacker and Wunder2015; Rakowski et al. Reference Rakowski, Krąpiec, Huels, Pawlyta, Dreves and Meadows2015; Fogtmann-Schulz et al. Reference Fogtmann-Schulz, Østbø, Nielsen, Olsen, Karoff and Knudsen2017; Wang et al. Reference Wang, Yu, Zou, Dai and Cheng2017; Park et al. Reference Park, Southon, Fahrni, Creasman and Mewaldt2017). Annually resolved 14C records have also been used to search for similar excursions in the atmospheric 14C content that may be caused by near-Earth supernova explosions, but so far no clear indications of such an event have been found (Damon et al. Reference Damon, Kaimei, Kocharov, Mikheeva and Peristykh1995; Menjo et al. Reference Menjo, Miyahara, Kuwana, Masuda and Nakamura2005; Dee et al. Reference Dee, Pope, Miles, Manning and Miyake2017).
Here, we present two new single-year 14C records with annual to biannual time resolution for the periods AD 692–790 and AD 966–1057. We compare our measurements with existing data and the IntCal13 curve (Reimer et al. Reference Reimer, Bard, Bayliss, Beck, Blackwell, Bronk Ramsey, Buck, Cheng, Edwards and Friedrich2013) in order to validate the quality of our data. We further present relevant statistics to report on the dendrochronology of the wood pieces used in this study and the accelerator performance of the Aarhus AMS Centre (AARAMS) at Aarhus University, Denmark.
METHODS
Oak tree samples (Quercus sp.) from four locations in Denmark were collected and dendrochronologically dated by the tree-ring laboratory at the Danish National Museum. All tree-ring records were dendrochronologically matched with the oak master curves for West Denmark, Zealand, and Schleswig-Holstein using the program DENDRO for Windows (Tyers Reference Tyers1999) that relies on the methods outlined by Baillie and Pilcher (Reference Baillie and Pilcher1973). The investigated pieces of wood originate from four different wood samples (Table 1). Ring widths for the wood records used in this study are listed in the supplementary table (Table S1).
Table 1 Information on wood records used in this study.
a West Denmark master chronology (National Museum of Denmark, Environmental Archaeology and Materials Science).
b Schleswig-Holstein master chronology (University of Hamburg, Department of Wood Science).
c Sjælland master chronology (National Museum of Denmark, Environmental Archaeology and Materials Science).
All tree rings were separated into early (EW) and late wood (LW) fractions whenever possible. For the years where early and late wood could not be separated, the measurements are based on whole wood (WW) instead of late wood, which otherwise is used for all samples. The time resolution of the two datasets is predominantly biannual. Each data point associated with a year is based on at least two measurements in order to enhance the accuracy and precision. Whenever there was insufficient material to perform at least two measurements for the year of interest, one sample for the particular year and the subsequent year were measured.
α-cellulose was extracted from all wood samples using a combination of the methods proposed by Loader et al. (Reference Loader, Robertson, Barker, Switsur and Waterhouse1997) and Southon and Magana (Reference Southon and Magana2010). The wood samples were bleached at 70°C with 1M sodium chlorite (NaClO2), acidified with 1M hydrochloric acid (HCl), then treated at room temperature with 17% sodium hydroxide (NaOH) to dissolve hemicellulose. Lastly, they were treated with 1M hydrochloric acid at room temperature to remove any contamination from atmospheric CO2. Subsequently, the α-cellulose was sealed in glass tubes containing copper oxide (CuO) and combusted to CO2 at 900°C, cryogenically purified, and then reduced to graphite using hydrogen and iron as catalyst. All samples were analysed using the 1MV High Voltage Engineering accelerator mass spectrometer system at the Aarhus AMS Centre, Aarhus University, Denmark (Olsen et al. Reference Olsen, Tikhomirov, Grosen, Heinemeier and Klein2017). 14C ages are reported as conventional 14C ages in 14C yr BP based on the measured 14C/12C ratio corrected for both natural and process isotopic fractionation by normalizing the result to the standard δ13C value of –25‰ VPDB using the 13C/12C ratios measured during AMS analysis (Stuiver and Polach, Reference Stuiver and Polach1977). The 14C ages are further converted to decay-corrected Δ14C values using
$${\Delta ^{14}}C = \left\lceil{ exp( - t_{C14}/8033) \over exp(- t_{cal}/8266)}- 1 \right\rceil \times 1000\permil$$where tC14 is the 14C age and tcal the calendar age. All 14C data are presented in Table 2.
Table 2 14C tree-ring analysis: 14C ages of late-wood (LW) or whole-wood (WW) rings presented in this study. The 14C Agemeancolumn provides the weighted mean 14C age with a reduced χ2 test (95.4% confidence level) given as X≤ Y. The χ2 is passed if and only if X≤Y (Bevington and Robinson Reference Bevington and Robinson2003). The deviation in terms of standard deviations (dev σ) is calculated as (14C Age – 14C Agemean)/ σ[14C Age].
RESULTS AND DISCUSSION
The pieces of wood cover the two periods AD 692–790 and AD 966–1057 (hereafter referred to as the AD 750 and AD 1000 dataset, respectively). Tree-ring widths associated with each piece of wood used in this study are shown along with the master curve in Figure 1. Only a limited part of each wood piece was used for 14C measurements where tree rings were sufficiently preserved and broad enough for separation of late and early wood (Tables 1 and S1). Student-t test values of matches between the wood pieces and the dendrochronological master curve of West Denmark are 11.0, 5.7, 4.4 and 6.4 for Ravning Enge (RE1), Østergård (ØG), Haderslev Fjord (HF) and Mojbøl (MOJ) respectively. The best Student-t match for Gråbrødre Kloster (GBK) was the Zealand master curve with a Student-t test value of 6.0 (Table 1).
Figure 1 Measured ring widths of the pieces of wood used in this study. Indices are abbreviations of the location of each wood piece. RE1: Ravning Enge (green), GBK: Gråbrødre Kloster (blue), ØG: Østergård (red), HF: Haderslev Fjord (yellow) and MOJ: Mojbøl (purple). Black line is the master curve of West Denmark used for the dendrochronological dating. (Please see electronic version for color figures.)
The new single-year measurements generally agree with the main structures in the IntCal13 curve, but these also add new detailed structures to the curve (e.g. around AD 700, 762, 1014 and 1040) (Figure 2). The new datasets capture two rapid solar cosmic-ray events in AD 775 and AD 993. In the AD 750 dataset, an offset towards lower Δ14C values exists between the new data and the IntCal13 curve between AD 725 and 760 and just before the onset of the AD 775 event. The mean Gaussian filters of the measurements in Figures 2 and 3 are based on calculated weighted means (weighted by error) and show the general fluctuations of the new data. The z-scores of replicate measurements (n = 196) for the combined AD 750 and AD 1000 dataset follow a normal distribution with a mean value of –0.02 ± 0.86. The replicate z-scores pass a reduced χ2 test (0.7 ≤ 1.2), indicating that we are able to reproduce measurements from the same year. Although the total number of replicates passes a χ2 test, some disagreements between several replicate years still persist (AD 724, 766, 781, 1021, 1045 and 1055, Table 2), highlighting the advantage of measuring more than one sample for each year that is analyzed. Four international comparison standards have been used to confirm the accuracy during analysis: IAEA-C3 (F14C = 1.2959 ± 0.003, n = 111, F14CTRUE = 1.2941 ± 0.0006), IAEA-C7 (F14C = 0.4954 ± 0.0002, n = 157, F14CTRUE = 0.4953 ± 0.0012), IAEA-C8 (F14C = 0.1503 ± 0.0001, n = 158, F14CTRUE = 0.1503 ± 0.0017) and FIRI-D (F14C = 0.5695 ± 0.0002, n = 194, F14CTRUE = 0.5705 ± 0.0002). Whereas IAEA-C7 and C8 are agreeing well with the true F14C value, the IAEA-C3 is too young and the FIRI-D is slightly too old. The reason for the IAEA-C3 discrepancy is unknown, whereas as we suspect that the pretreatment method may explain the FIRI-D discrepancy. All FIRI-D samples are pretreated using our standard acid-base-acid pretreatment, and two recent analysis of FIRI-D on extracted cellulose provided an F14C value of 0.5704 ± 0.0011, in perfect agreement with the expected FIRI D F14C value. However, more data is required in order to be able to conclude whether or not the pretreatment may explain the FIRI-D discrepancy. Furthermore, samples from two broad rings have provided enough material for several 14C analyses, which have been measured in different measurement batches.
Figure 2 Combined annual/biannual Δ14C values plotted together with the IntCal13 calibration curve (Reimer et al Reference Reimer, Bard, Bayliss, Beck, Blackwell, Bronk Ramsey, Buck, Cheng, Edwards and Friedrich2013). The grey curve is constructed using a running Gaussian mean filter of length 3 and is shown as guide for comparison with IntCal13.
Figure 3 Weighted means and Gaussian mean filter of data from this study (black dots and grey curve) along with the IntCal13 curve (blue curve) and published data.
Samples AAR-27091 and AAR–27770 have been analyzed in 6 (one outlier) and 4 batches, respectively, providing evidence for good reproducibility for the duration of the 14C analysis of tree-ring batches (Table 2). The dendrochronological dating of the wood piece RE1 is verified by the AD 775 event, as the timing of the event recorded in this study is in full agreement with existing 14C records from all over the world (Figure 3A). The period spanned by AD 980–1006 in the AD 1000 dataset is plotted along with published 14C data by Fogtmann-Schulz et al. (Reference Fogtmann-Schulz, Østbø, Nielsen, Olsen, Karoff and Knudsen2017) in Figure 4. The wood piece used in the study by Fogtmann-Schulz et al. (Reference Fogtmann-Schulz, Østbø, Nielsen, Olsen, Karoff and Knudsen2017) was dendrochronologically dated by the Danish National Museum and prepared and analyzed at AARAMS like the wood pieces used in this study. Some disagreements between the AD 1000 dataset of this study and the data by Fogtmann-Schulz et al. (Reference Fogtmann-Schulz, Østbø, Nielsen, Olsen, Karoff and Knudsen2017) are apparent for the years AD 1001–1006. In this period, the measurements by Fogtmann-Schulz et al. (Reference Fogtmann-Schulz, Østbø, Nielsen, Olsen, Karoff and Knudsen2017) are based on single measurements of whole and late wood material, while the new data presented in this study are based on two measurements of late wood per reported year. The new data presented here appear to agree better with existing data for this period, and we therefore consider them more accurate than the data presented by Fogtmann-Schulz et al. (Reference Fogtmann-Schulz, Østbø, Nielsen, Olsen, Karoff and Knudsen2017) for the years AD 1001–1006 (Figure 3B). A rather striking difference between the AD 1000 dataset and published data by Fogtmann-Schulz et al. (Reference Fogtmann-Schulz, Østbø, Nielsen, Olsen, Karoff and Knudsen2017) concerns the occurrence of the AD 994 cosmic-ray event. The data presented by Fogtmann-Schulz et al. (Reference Fogtmann-Schulz, Østbø, Nielsen, Olsen, Karoff and Knudsen2017) suggest the event occurred after the formation of early wood in AD 994 (i.e. in agreement with published data by Miyake et al. [Reference Miyake, Masuda and Nakamura2013, Reference Miyake, Masuda, Hakozaki, Nakamura, Tokanai, Kato, Kimura and Mitsutani2014]), whereas the AD 1000 dataset suggests the event took place between the late wood fractions in AD 992/993, i.e. one year earlier than proposed by Miyake et al. (Reference Miyake, Masuda and Nakamura2013, Reference Miyake, Masuda, Hakozaki, Nakamura, Tokanai, Kato, Kimura and Mitsutani2014) and Fogtmann-Schulz et al. (Reference Fogtmann-Schulz, Østbø, Nielsen, Olsen, Karoff and Knudsen2017). In the period between AD 991 and 995, the AD 1000 dataset is based on two measurements of the late wood fraction, i.e. the most robust wood fraction for 14C measurements (Kudsk et al. Reference Kudsk, Olsen, Nielsen, Fogtmann-Schulz, Knudsen and Karoff2018). This discrepancy concerning the timing of the AD 994 event prompted us to scrutinize our laboratory methods and procedures as well as the dendrochronological age constraints. Careful remeasuring of the ring widths of the MOJ and ØG wood pieces showed that the dendrochronological ages of the MOJ record were displaced by –1 year. Thus, although the overall conclusions of Fogtmann-Schulz et al. (Reference Fogtmann-Schulz, Østbø, Nielsen, Olsen, Karoff and Knudsen2017) remain valid, the timing of the event was one year too young. The updated information on the MOJ wood records has therefore been added to the present paper (Tables 1 and S1). Based on the corrected dendrochronological ages for the MOJ wood records, and the combined evidence from the two wood records (MOJ and ØG), we conclude that the cosmic-ray event took place during the summer of AD 993 (Figures 3 and 4).
Figure 4 Weighted means of data from this study along with EW, LW and WW data published by Fogtmann-Schulz et al. (Reference Fogtmann-Schulz, Østbø, Nielsen, Olsen, Karoff and Knudsen2017) across the AD 992/993 event.
The two new datasets presented in this study are shown in Figure 3 along with published tree-ring data in the literature. Whenever the published data are based on wood samples that cover more than one year, the age of the midpoint of the time span is used (e.g. for the published IntCal13 data (10-yr blocks), Sakamoto et al. Reference Sakamoto, Imamura, Van der Plicht, Mitsutani and Sahara2003 (10-yr blocks), Güttler et al. (Reference Güttler, Wacker, Kromer, Friedrich and Synal2013) (2-yr blocks)). The differences between the two new 14C records presented here and published tree-ring data are generally small. The detailed structures found in the new AD 750 dataset around AD 690–720 agree well the 10-yr data blocks of Sakamoto et al. (2003). In the AD 750 dataset, we observe offsets toward lower values compared to IntCal13 in the period between AD 725 and 760. This offset is in agreement with the data published by Sakamoto et al. (2003) and Park et al. (Reference Park, Southon, Fahrni, Creasman and Mewaldt2017) which both show a similar offset compared to the IntCal13. However, the new data presented here have slightly lower values compared to those of Park et al. (Reference Park, Southon, Fahrni, Creasman and Mewaldt2017) but fit well with published data for the AD 775 event.
The pattern in the new AD 1000 dataset spanning the AD 994 event fits well with the existing data published by Miyake et al. (Reference Miyake, Masuda and Nakamura2013, Reference Miyake, Masuda, Hakozaki, Nakamura, Tokanai, Kato, Kimura and Mitsutani2014), which shows a similar decrease in 14C after the cosmic-ray event. The relatively good agreement between the AD 1000 dataset and published data continue between AD 1000 and 1025, although the published data show that some discrepancies exist among the various datasets between AD 1040 and 1050, where they almost show opposite patterns. The new data are also in good agreement with data by Damon et al. (Reference Damon, Kaimei, Kocharov, Mikheeva and Peristykh1995), particularly around the peak at AD 1015. Between AD 1025 and 1040, the new data are consistent with the data published by Güttler et al. (Reference Güttler, Wacker, Kromer, Friedrich and Synal2013), but some discrepancies exist between the two datasets between AD 1040 and 1050, where they almost show opposite patterns.
Several studies have searched for fingerprints of proximal supernova by producing annual 14C records across years associated with known supernova explosions (Damon et al. Reference Damon, Kaimei, Kocharov, Mikheeva and Peristykh1995; Menjo et al. Reference Menjo, Miyahara, Kuwana, Masuda and Nakamura2005; Dee et al. Reference Dee, Pope, Miles, Manning and Miyake2017). So far, no clear excursions have yet been found in any 14C record that could be linked to a past supernova explosion. Two supernova explosions are known to have occurred in AD 1006 and 1054, and these events are thus covered by the new AD 1000 dataset. Although an increase in the atmospheric 14C content can be observed at AD 1055 (Figure 2), there are no clear indications of any of the two supernova explosions.
CONCLUSION
In this study, we present two new 14C records based on single-year measurements of Danish oak. The two new records, which span the periods AD 692–790 and 966–1057, have an annual to biannual time resolution. Each data point presented in this study is based on two separate samples and two separate measurements. Statistics on replicate measurements indicate that our data are reproducible, which increases the robustness and reliability of the new data. The two new records span the two rapid cosmic-ray events in AD 775 and 994. The timing of the AD 775 event in the data presented here is in full agreement with the timing reported in the literature, but the AD 994 event occurred one year earlier compared to existing records. Careful reanalysis of the dendrochronology that underpins the annually resolved 14C records based on Danish oak published here and in Fogtmann-Schulz et al. (Reference Fogtmann-Schulz, Østbø, Nielsen, Olsen, Karoff and Knudsen2017) suggests unequivocally that the AD 994 event actually took place in the summer of AD 993. The new data presented here generally agree with the trends in the current international calibration curve, although with a minor offset between AD 725 and 760, an offset that is also observed in other published annually resolved 14C records. Finally, to meet the criteria for data to be included in the next international calibration curve (e.g. ring widths, t-values and achieve numbers), we also present all relevant information concerning the wood material on which the new 14C records are based.
ACKNOWLEDGMENTS
Sabrina GK Kudsk, Alexandra Fogtmann-Schulz, Mads F Knudsen, and Christoffer Karoff sincerely acknowledge the financial support from the Villum Foundation (VKR023114 and VKR010116). Bente Philippsen is supported by the Danish National Research Foundation under the grant DNRF119 – Centre of Excellence for Urban Network Evolutions (UrbNet). UrbNet further funded analysis of the RE1 wood record. Funding for the Stellar Astrophysics Centre is provided by the Danish National Research Foundation (Grant agreement No.: DNRF106).
SUPPLEMENTARY INFORMATION
Tree-ring widths of the wood piece used in the study by Fogtmann-Schulz et al. (Reference Fogtmann-Schulz, Østbø, Nielsen, Olsen, Karoff and Knudsen2017) (MOJ) have been applied in the supplementary table along with tree-ring widths of the wood pieces used in this study in order to make the information public. Dendrochronological information of the MOJ wood piece is given in Eriksen et al. (Reference Eriksen2004) (wood piece 50851229).
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit https://doi.org/10.1017/RDC.2019.107
