INTRODUCTION
The Yayoi period (Figure 1) is a critical age for Japanese archaeology. At the beginning of this period wet paddy rice cultivation was introduced in northern Kyushu island of the Japanese archipelago and new farming implements and a new type of pottery called Yayoi Pottery appeared. The creation of irrigated rice paddies required extensive construction, leading to conflicts over water and land, which drove changes in societal organization and leadership. During the Yayoi period, people changed their way of life from hunting and gathering to farming.
Archaeological chronology of the mainland of the Japanese Archipelago. “Jomon” means cord-marked decoration found on the earthenwares in this period. “Yayoi” is the name of the town where different type of earthenwares from Jomon were excavated. “Kofun” means old burial mounds that were constructed in large numbers in this period.
Archaeologists have long believed that iron tools were introduced along with paddy rice cultivation around the beginning of the Yayoi period, which was considered to be around the 5th century BCE. However, accelerator mass spectrometry radiocarbon (AMS-14C) dating of materials related to early paddy field remains revealed that paddy rice cultivation had begun in northern Kyushu by the 10th century BCE (Fujio et al. Reference Fujio, Imamura and Nishimoto2005; Fujio Reference Fujio2021). This indicates that paddy rice cultivation was introduced earlier than the introduction of iron tools in Japan.
The chronology is still being debated, with some researchers questioning whether 14C dating of Japanese materials can be correctly calibrated with IntCal, which is based on data obtained from Western tree specimens. In response to this criticism, the National Museum of Japanese History (NMJH) has been working on the 14C dating of tree rings from 10th century BCE samples and comparing the data to those of IntCal (e.g., Sakamoto et al. Reference Sakamoto, Imamura, van der Plicht, Mitsutani and Sahara2003; Ozaki et al. Reference Ozaki, Imamura and Matsuzaki2007). Some results have been adopted into the latest calibration curve, IntCal20 (Reimer et al. Reference Reimer, Austin, Bard, Bayliss, Blackwell, Ramsey, Butzin, Cheng, Edwards and Friedrich2020). The shape of IntCal20 covering the period from the 1st to the 3rd century has been revised since the data from Japanese tree rings were included. This period comprises the transition period from the Yayoi to the Kofun period, when large chief tombs were constructed in Japan. The chronology of this period has also been a major research topic in Japanese archaeology. However, the trends observed in the 14C dating of Japanese tree rings may be attributable to regional effects. A better understanding and verification of these trends are required to establish more reliable 14C ages from the Yayoi period.
In the 2010s, a novel dendrochronological dating method using δ18O of cellulose in annual tree rings was developed in Japan and put to practical use. In humid regions such as the Japanese archipelago, cellulose oxygen isotope ratios are synchronized with annual precipitation. By applying this method, paleoclimate was successfully reconstructed precisely on an annual basis (Nakatsuka et al. Reference Nakatsuka, Sano, Li, Xu, Tsushima, Shigeoka, Sho, Ohnishi, Sakamoto and Ozaki2020). Since precipitation variations are shared across a relatively large area in East Asia, they can also be applied to dendrochronology (Sano et al. Reference Sano, Li, Murakami, Jinno, Ura, Kaneda and Nakatsuka2022), resulting in significant improvements in 14C dating of Japanese tree annual rings (e.g., Sakamoto et al. Reference Sakamoto, Hakozaki, Nakao and Nakatsuka2017). Some of the results were included in the new calibration curve, IntCal20 (Reimer et al. Reference Reimer, Austin, Bard, Bayliss, Blackwell, Ramsey, Butzin, Cheng, Edwards and Friedrich2020), as the first tree ring dated by δ18O. In order to contribute to the chronology of the Yayoi period, we conducted annual 14C dating of two Japanese tree rings.
SAMPLE PREPARATION AND MEASUREMENTS
Two wood samples were obtained from buried trees dating back to the beginning and end of the Yayoi period, respectively (Figure 2). The samples were designated as KGSR002 (from the site in Kagoshima) and MGSNMr-1 (from the site in Miyagi). Although conventional dendrochronological methods could not date these materials, we determined their ages by δ18O dendrochronology (Table 1).
Locations of the sites where the samples used in this study were excavated. Typical directions of summer monsoons around the Japanese Archipelago are also indicated.
List of samples for annual 14C measurement.
Cellulose extraction from tree rings was carried out by NMJH using the procedure described by Sakamoto et al. (Reference Sakamoto, Hakozaki, Nakao and Nakatsuka2017). Buried wood samples were cut into blocks of proper size and polished manually to determine the tree-ring number and widths. The blocks were further polished to a thickness of a few millimeters. After ultrasonic cleaning using acetone and a 2:1 v/v chloroform-methanol mixture, each tree-ring plate was slipped between perforated PTFE sheets and stitched up with PTFE string. The sample was put into a test tube and bleached four times (1 hr each time) with a solution of sodium chlorite and concentrated HCl at 70°C. Hemicellulose was removed using a 17.5 wt% NaOH solution (1 hr, three times at 80°C). After neutralization, the samples were vacuum freeze-dried. Bleached annual rings were separated under a microscope and stored in pre-baked glass tubes.
Annual ring cellulose was divided into aliquots of about 0.1 mg and wrapped in 7 mm × 7 mm silver foil square for δ18O measurement. The measurement was performed using a continuous flow mass spectrometer coupled to a pyrolysis-type elemental analyzer at Nagoya University. Age determination by δ18O was compared with the master chronology of central Japan. The δ18O variation of KGSR002 was found to be perfectly synchronized with that of a second KGSR sample, and their average value was compared to the master chronology.
Graphitization of the remaining annual ring cellulose was carried out at the Laboratory of Radiocarbon Dating, the University of Tokyo, which also performed AMS 14C measurements. The tabulated tree-ring 14C ages of this study are presented in the supplemental material. To examine the repeatability of the measurements, some of the annual rings of KGSR002 were measured two or three times. Except for a couple of cases, most results were within the measurement error of below ±25 14C years.
RESULTS AND DISCUSSIONS
KGSR002: 1053–921 BCE
This sample is a buried Japanese bead tree excavated from the Shiromizu B site in Kagoshima Prefecture in the southwestern region of the Japanese archipelago. Preliminary 14C dating returned an age of 2820 ± 30 14C BP (IAAA-143218), suggesting that the sample dates to around the 10th century BCE. 14C measurement was carried out after determining tree-ring age using δ18O. The results are shown in Figure 3, along with the ±1σ of the IntCal20 calibration curve and original datasets cited from https://www.intcal.org.
Comparison between 14C ages of KGSR002 (black data points with uncertainties), IntCal20 (solid lines), and the original datasets of IntCal20 (gray data points with uncertainties).
The data from this period used in IntCal are coarse and the calibration curve is relatively gentle. However, the behavior of KGSR002 shows a clear trend of 14C concentrations during this period. That is, the period around 1000 BCE shows 14C ages that are older than expected, while the period after that exhibits 14C ages on the lower side of the calibration curve. The age deviation ΔZ for all samples was calculated according to
$$\Delta {\rm{Z}} = {{T{{\left( t \right)}_{sample}} - T{{\left( t \right)}_{IntCal}}} \over {\sqrt {\sigma \left( t \right)_{sample}^2 + \sigma \left( t \right)_{IntCal}^2} }}$$
where ΔZ is the normalized difference, Tsample and TIntCal are respectively the 14C ages of the tree-ring sample and the corresponding IntCal of t, and σsample and σIntCal are their respective uncertainties. A histogram of the age deviation values is shown in Figure 4. The ΔZ data were fitted to a normal distribution using the least-squares method; the resulting distribution has a central value of 0, but consists of two overlapping peaks with different centers. Clarifying the detailed behavior of the calibration curve is expected to lead to more precise dating methods, such as the 14C-wiggle matching (Pearson Reference Pearson1986).
Normalized distribution of differences (ΔZ) between 14C ages of KGSR002 and IntCal20. The solid line shows the best fit to a normal distribution.
MGSNMr-1: 41 BCE–130 CE
This sample is a buried Japanese zelkova excavated from the Nakazaike Minami site in Miyagi Prefecture northeastern Honshu island of the Japanese archipelago. Preliminary 14C dating returned an age of 1826 ± 26 14C BP (Beta-361508). Annual 14C measurement was carried out after determining tree-ring age by δ18O. The results are shown in Figure 5, along with the ±1σ of the IntCal20 and IntCal13 (Reimer et al. Reference Reimer, Bard, Bayliss, Beck, Blackwell, Bronk Ramsey, Buck, Cheng, Edwards and Friedrich2013) calibration curves.
Comparison between 14C ages of MGSNMr-1 (data points with uncertainties), IntCal20 (solid lines), and IntCal13 (dashed lines).
The shape of the IntCal20 calibration curve for this period was modified using the cedar (246 BCE–200 CE) and the cypress (52–542 CE) data from central Honshu island (Reimer et al. Reference Reimer, Austin, Bard, Bayliss, Blackwell, Ramsey, Butzin, Cheng, Edwards and Friedrich2020). In both cases, however, five or ten annual rings were used as a single measurement sample. Figure 6 shows a histogram of δZ values compared to the IntCal20 and IntCal13. Although the present measurements did not reach the plateau period of the calibration curve seen in the second half of the 2nd century, the distribution of the δZ values appears closer to that of IntCal20 than IntCal13. However, the first half of the 2nd century is equivalent to IntCal13.
Normalized distribution of differences between 14C ages of MGSNMr-1 and IntCal13 (above) and IntCal20 (below). The solid and dashed lines show the best fit to a normal distribution for each calibration curve.
The behaviour of 14C concentrations in the atmosphere around the Japanese archipelago has often been discussed. Kigoshi and Hasegawa (Reference Kigoshi and Hasegawa1966) measured the annual rings of the cedar, Yakusugi, on Yakushima Island located south of Kyushu Island and found that the 14C concentration was 0.5% to 1% lower than that of tree rings from Arizona and California during the same period, suggesting the possible influence of the ocean. Nakamura et al. (Reference Nakamura, Masuda, Miyake, Nagaya and Yoshimitsu2013) also measured Yakusugi cedar and other Japan-sourced trees. They showed that the 14C concentration in the atmosphere around the Japanese archipelago is positioned between the Northern Hemisphere (IntCal) and the Southern Hemisphere (SHCal), suggesting the involvement of the Pacific High caused by solar activity.
Monsoons around the Japanese archipelago may bring low atmospheric 14C concentrations from the Pacific ocean. They include the East Asian monsoon, which brings a rainy season to East Asia in the summer; typhoons over the western Pacific, which carry strong winds and heavy rainfall; and the cool monsoon called Yamase, which blows from the North Pacific to northeastern Japan during the summer. Hakozaki (Reference Hakozaki2013) has conducted 14C dating of biennial tree rings from the northernmost part of Honshu Island and found that the estimated ages around 1400 CE about 35 years older than those given by IntCal on average, suggesting that the Yamase may have had an effect. On the other hand, Hong et al. (Reference Hong, Sakamoto, Kim, Park, Sung, Park, Park, Hakozaki and Sonsubmitted) measured the 14C ages of annual tree rings excavated from the southern coast of the Korean peninsula in the 1st and 2nd centuries. They found that ages given by 14C dating were relatively close to those given by IntCal13 or were between the IntCal13 and IntCal20 values. The latitude of both regions is almost the same and both areas are similarly affected by the East Asian monsoons and typhoons. Atmospheric 14C concentration in these East Asian sites may be influenced by the distance from the North Pacific, which would reflect the influence of the Yamase. To elucidate this, areal 14C dating of tree rings from East Asia is required.
CONCLUSIONS
14C dating of Japanese tree rings in specimens from around 1000 BCE and around 100 CE was conducted. These periods mark the beginning and the end of the Yayoi Period, a pivotal era that is important to Japanese archaeology. Knowing the detailed behavior of 14C dating during these periods helps construct a more accurate chronological perspective. 14C ages obtained from tree rings of specimens from the southern Japanese archipelago dating from the late 11th century BCE were slightly above the IntCal20 calibration curve. In contrast, those from samples dating to the first half of the 10th century BCE tended to be below it.
The 14C ages of tree rings from northeastern Japan around the beginning of the common era show a behavior similar to that of IntCal20, which includes data of Japan-sourced trees. However, 14C ages of tree rings from the southern coast of the Korean Peninsula, located at similar latitude, are relatively close to IntCal13. These variations may reflect the influence of monsoons, which can dilute the atmospheric 14C concentration around the Japanese archipelago from the ocean at different times of the year.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit https://doi.org/10.1017/RDC.2023.50
ACKNOWLEDGMENTS
We are grateful to the Kagoshima Prefectural Foundation for Cultural Promotion, Buried Cultural Property Research Center (KGSR002), and the Sendai City Board of Education and Cultural Property Center of the Pasco Inc. (MGSNMr-1), for providing us with samples. Ms. R. Yamamoto helped us with sample preparation. Ms. Y. Saito reviewed the English text. JSPS KAKENHI Grant Numbers JP18H03594 and JP22H00026 supported this work.
Open access
