The annually formed layers of sediment from Lake Suigetsu have the potential to reveal information about environmental change in Japan over the past ∼50,000 years with chronological precision as good as, or better than, the Greenland ice cores (1). Such precision is invaluable for synchronizing records and evaluating leads and lags in the global climate system. Perhaps even more important, the fragile leaves and seeds hidden within these layers (see the figure) provide a record of past atmospheric concentration of 14C, the radioactive isotope of carbon. As reported by Bronk Ramsey et al. on page 370 of this issue (2), this record stretches back over the full length of the radiocarbon age scale. The results are invaluable for improving the accuracy with which radiocarbon dates can be converted to the calendar time scale.

Radiocarbon (14C) is one of the main dating methods in archaeology and Earth science for the late Quaternary period back to ∼50,000 years ago. The method is based on the predictable radioactive decay of 14C, which forms when neutrons generated from cosmic ray bombardment collide with nitrogen in the upper atmosphere. However, the 14C concentration in the atmosphere is not constant, because varying amounts of cosmic rays reach the atmosphere and carbon storage and release from the ocean and biosphere change with time. To estimate the true age of an unknown sample, its radiocarbon age must be corrected for these changes.

To ensure that radiocarbon dates are corrected consistently, scientists have been building internationally recognized calibration curves for many decades. The latest such curve, IntCal09, covers the whole radiocarbon age scale to ∼50,000 years ago (3).

IntCal09 relies heavily on tree rings, which are ideal for building a correction (or calibration) curve, because the year of growth of the ring can be determined precisely, provided that the tree section to be measured can be matched to a master chronology. The oldest securely dated master chronology goes back to 12,593 calendar years before the present (cal B.P., where “present” is set as 1950 A.D.). For calibration beyond this date, IntCal09 had to rely on 14C measurements of corals and the tiny shells of planktonic organisms found in marine sediments (3).

Radiocarbon correction beyond tree rings. Bronk Ramsey et al.'s atmospheric 14C record, derived from measurements on terrestrial macrofossils (right) from the laminated sediments (left) of Lake Suigetsu, will help to improve the IntCal calibration curve used to estimate the true age of radiocarbon dated samples. CREDITS: (LEFT) GORDON SCHLOLAUT/HELMHOLTZ CENTRE POTSDAM-GFZ GERMAN RESEARCH CENTRE FOR GEOSCIENCES, GERMANY; (RIGHT) RICHARD STAFF/OXFORD RADIOCARBON ACCELERATOR UNIT/UNIVERSITY OF OXFORD, UK

However, using the marine data to calibrate the atmospheric radiocarbon record is not straightforward. There is an apparent radiocarbon age difference of ∼400 years on average between an organism living in the surface ocean and one living on land at the same time. This discrepancy arises because the ocean is a large carbon reservoir that responds more slowly than the atmosphere to changes in 14C production. The deep ocean also contains carbon that has become depleted in 14C through radioactive decay since it was entrained into the global circulation. Eventually, deep water upwells and mixes with the surface ocean, leaving it depleted in 14C relative to the atmosphere. Because changes in ocean circulation can result in variable upwelling of this deep water, the 14C record in marine organisms is attenuated and offset from the atmosphere by an amount that changes with time and location. To use the marine records in the terrestrial calibration curve for periods before the tree rings, the offset must be estimated, increasing the uncertainty of the calibration.

A direct atmospheric 14C record like that from Lake Suigetsu can be used to decrease the uncertainty in the calibration curve. It will also be invaluable for capturing higher-frequency oscillations in atmospheric 14C, which are not resolved in the IntCal09 curve for the time period before the tree rings. For stratigraphical series of radiocarbon ages—such as those used to date the earliest pottery from the Xianrendong Cave, China, to 20,000 years ago (4)—matching the radiocarbon dates from sediment layers to these oscillations will improve age estimates.

However, laminations in lake sediments are seldom, if ever, perfect. In the dark clay sediments of Lake Suigetsu, white layers are formed by the deposition of silica cases of algae (diatoms) in spring and summer; each layer is counted as 1 year (5). However, not all layers are preserved, and, in some cases, two diatom layers may be produced in a single year. Counting the layers thus has an associated uncertainty, which accumulates with depth. Bronk Ramsey et al. circumvent this problem to an extent by matching the 14C of uranium-thorium–dated stalagmites from the Bahamas (6) and China (7) to the Suigetsu 14C record, thereby reducing the uncertainty on the Suigetsu time scale.

The IntCal09 calibration curve assumes a constant atmospheric-surface ocean age offset (R) to estimate the atmospheric 14C levels from the marine calibration records, which were selected from regions where ocean currents are believed to have been stable over the past 12,000 years. Through comparison with the Lake Suigetsu record, Bronk Ramsey et al. show that for these oceanographically stable regions, R was larger ∼20,000 years ago than during the past 12,000 years, but not to the extent estimated by some authors (8, 9). This estimated variation in R will provide more realistic values for the marine data in the calibration curves.

Earth's geomagnetic field shields against cosmic rays, which produce 14C and other isotopes such as 10Be in the atmosphere. About 40,000 years ago, a substantial decrease of the geomagnetic field strength, the Laschamp event, nearly doubled the production rate of 10Be (10) and would have had a similar effect on 14C. The 14C production spike is attenuated in the IntCal09 marine records due to the large carbon reservoir. This has led to debates about whether radiocarbon calibration for atmospheric samples would be invalid for this time period (11, 12). In the Suigetsu record, the Laschamp effect is within 2 standard deviations of the IntCal09 calibration curve. Given radiocarbon measurement uncertainties of hundreds to even a few thousand years in this time period, the Laschamp effect will have minimal effect on calibration.

The Lake Suigetsu 14C record reported by Bronk Ramsey et al. does not have sufficient data density to provide a stand-alone calibration curve, but it will substantially augment and improve the atmospheric radiocarbon calibration curve. Work is in progress to update the IntCal calibration curve, which will include the Lake Suigetsu data. The resulting improvement in the accuracy of calibrated radiocarbon dates will greatly affect studies of past climate and environmental change and human response to these changes.