Radiocarbon dating of bulk sediment has long been used as a method of last resort when reliable wood, charcoal, or plant macrofossils are not available for analysis. Accurate dating of sediment is complicated by the presence of multiple organic carbon fractions, each with a potentially different 14C activity. Additionally, the presence of carbon bound by clay minerals can significantly reduce the accuracy of a sediment age determination, with the oldest 14C ages seen in samples with the highest clay content (Scharpenseel and Becker-Heidmann 1992).

We have constructed a system to extract carbon from meteorites using a vacuum-tight RF melting method in order to study radiocarbon activities in meteorites. The extraction system was examined using iron standards of known carbon content. The carbon extraction efficiencies and 14C ages of the iron standards by this method were compared with the results obtained previously by our older melting system and a wet oxidation method. Higher collection efficiencies of about 90% for the iron samples of relatively high carbon content were achieved by the new system. The efficiency of extracting a small amount of carbon is also near 90% after improving the extraction procedure. The 14C ages of the iron standards were compared to the ages by the wet method. The results indicate that contamination by modern carbon is negligible in the system. Furthermore, terrestrial 14C ages of two Antarctic meteorites, Y-75102 and ALH-77294, from the Yamato and Allan Hills ice fields, respectively, were determined. The age of Y-75102 is estimated 4.0 ± 1.0 ka, and the age of ALH-77294 is 19.5 ± 1.2 ka. The 14C ages on the meteorites roughly agree with the literature value. However, further study is needed in improvement on reducing a background value and of complete fusion of a meteorite in the extraction system.

An error source in radiocarbon dating of ancient mortar is dead carbon of limestone mixed in the matrix. To eliminate the influence of limestone the difference in feasibility to react with acid between mortar and limestone is used. Since the rate of reaction depends on grain size use of a well-defined grain size can give a better separation between mortar and limestone. We present results for the grain size dependence of reaction rates for several mortar and limestone samples and discuss the application for dating.

We present routine methods of target preparation for radiocarbon analysis at the ANTARES Accelerator Mass Spectrometry (AMS) Centre, as well as recent developments which have decreased our procedural blank level and improved our ability to process small samples containing less than 200 μg of carbon. Routine methods of 14C sample preparation include sample pretreatment, CO2 extraction (combustion, hydrolysis and water stripping) and conversion to graphite (graphitization). A new method of cleaning glassware and reagents used in sample processing, by baking them under a stream of oxygen, is described. The results show significant improvements in our procedural blanks. In addition, a new graphitization system dedicated to small samples, using H2/Fe reduction of CO2, has been commissioned. The technical details of this system, the graphite yield and the level of fractionation of the targets are discussed.

A line for preparation of graphite targets for accelerator mass spectrometry (AMS) radiocarbon dating has been built in the Gliwice 14C Laboratory. The AMS 14C measurements of our targets are performed in the Leibniz-Labor für Altersbestimmung, Kiel, Germany. The quality of our line has been tested in two series of AMS 14C measurements of background and Ox-II standard samples and by measurements of the amount of CO2 released during combustion of sample-free quartz tubes. Most background contamination in the first series was introduced during combustion, which has been greatly reduced by baking quartz tubes vacuum-sealed with CuO and Ag. The residual contamination (ca. 1.5 μg C) seems to come mostly from the quartz tubes themselves. At present, most of the contamination of the background is introduced during graphitization. The reproducibility of background preparations is satisfactory, especially for samples larger than 1.5 mg, when it is better than ± 0.09 pMC. Despite still significant contamination with low-14C carbon during the graphitization process (corresponding to 1.2 ± 0.2% of 14C-free carbon), the good reproducibility of the results allows us to use our line in routine 14C dating.

In Groningen, all organic samples for accelerator mass spectrometry (AMS) are combusted in an automatic Elemental Analyzer, coupled to an Isotope Ratio Mass Spectrometer and Cryogenic Trapping System. The Gas Chromatographic (GC) column, part of the Elemental Analyzer system, appeared to be the main cause for memory effects. Therefore we modified the Elemental Analyzer, such that the trapped CO2 no longer passed the GC column. Our system modification reduced the memory effect significantly, as shown by lower radiocarbon concentration values for anthracite backgrounds, and a much smaller spread in these values. Our modified system can perform up to 40 combustions unattended in about 6 hr.

We constructed a semi-automated system to transform carbonate samples to CO2, as a means to increase sample-processing capacity. The physico-chemical process includes hydrolysis of carbonate, quantification of the mass of carbon and CO2 collection in a glass ampoule. The system is computer-controlled and monitored, and all the data are stored. A single run of five consecutive samples requires about 3.5 hours. Measurements of 14C concentrations were made on samples of IAEA C-1 Carrara marble to test the reliability of this semi-automated system. These measurements have allowed the determination of the total system background and the memory effect of our system.

Processing and measurement of 200 biomedical samples has provided an opportunity to better understand the characteristics of accelerator mass spectrometry (AMS) analysis of such samples. We have utilized established procedures (Vogel 1992) and developed new methods for handling various biological samples. We have included secondary standards of known isotope ratio for all assays. A method of determining maximum precision for each unknown sample value is also reported. The presented data are an update of the ongoing radiocarbon AMS biomedical program at Purdue University.

We summarize the basic operation of accelerator mass spectrometry (AMS) systems used to measure radiocarbon and discuss the calculations used to convert AMS output to 14C data.

For more than 15 years, the radiocarbon community has participated in a series of laboratory intercomparisons in response to the issue of comparability of measurements as perceived within the wider user communities (Scott et al. 1990; Rozanski et al. 1992; Gulliksen and Scott 1995; Scott et al. 1997). In this report, we provide an update on the current 14C laboratory intercomparison and reflect on future issues linked to the laboratory intercomparison program, not least those resulting from a significant growth in the number of accelerator mass spectrometry (AMS) facilities providing routine dating of small samples (milligram size).

We present a data model designed for laboratories in which many different methods are used. The main feature of the model is the recursive relationships of data records in the subsample table. This makes it possible to model all steps from the preparation of the sample to the final value and enables the raw data to be stored together with the final values, even if many preparation steps or many subsamples are involved. We show ways to adopt this basic model to laboratories with many years of data. The data model is extended to include laboratory records and attribute data, e.g. the geographical coordinates of the sampling site or links to the various projects for which the samples are used.

Accelerator mass spectrometry radiocarbon (AMS 14C) dating of the late glacial section of laminated sediments from Lake Gościąż and Lake Perespilno, Poland, performed with time resolution of 20–30 years suggests quasi-periodic oscillations of δ14C. The regularity of oscillations has been checked by the Maximum Entropy and Fast Fourier Transform methods (MEM and FFT), which revealed peaks at 200 and 230 yr. These periods are similar to those found in the high-precision Holocene δ14C record, and attributed to the changing sun. The analytical assessment of the significance of the FFT and MEM peaks is problematic because of non-uniform spacing and various uncertainties in the input data. The significance of the peaks has therefore been studied by the Monte-Carlo method. Because the original data were approximated with a spline function, the amplitude of the MEM and FFT peaks depends on the stiffness of the spline, which is strongly related to the “real” uncertainty of 14C ages. The Monte-Carlo experiments demonstrate that the significance levels of the MEM and FFT peaks also depend on the spline stiffness. Therefore, the existence of solar 14C variations in the Late Glacial remains an open question.

A heuristic approach, nicknamed “transition dating,” was used to date sequences of early Iron Age contexts using a series of 14C determinations. The basic principles of transition dating are simple and intuitive: 1) attempt to date transitions between periods, phases, etc. rather than the phases themselves, and 2) the most plausible date for that transition is one that is later than the dates from contexts preceding it, and is still earlier than the dates succeeding it. Hypotheses regarding the actual date of each transition may be evaluated using an appropriate loss function. These loss functions can also be adjusted or weighted by the user to account differentially for the various factors causing the distortion or “fuzz” in the dates.

This paper highlights some of the main developments to the radiocarbon calibration program, OxCal. In addition to many cosmetic changes, the latest version of OxCal uses some different algorithms for the treatment of multiple phases. The theoretical framework behind these is discussed and some model calculations demonstrated. Significant changes have also been made to the sampling algorithms used which improve the convergence of the Bayesian analysis. The convergence itself is also reported in a more comprehensive way so that problems can be traced to specific parts of the model. The use of convergence data, and other techniques for testing the implications of particular models, are described.

Nowadays, there is a large number of time series of natural data to study geophysical and astrophysical phenomena and their characteristics. However, short length and data gaps pose a substantial problem for obtaining results on properties of the underlying physical phenomena with existing algorithms. Using only an equidistant subset of the data with coarse steps leads to loss of information. We present a method to recover missing data in time series. The approach is based on modeling the time series with manifolds of small dimension, and it is implemented with the help of neural networks. We applied this approach to real data on cosmogenic isotopes, demonstrating that it could successfully repair gaps where data was purposely left out. Multi-fractal analysis was applied to a true radiocarbon time series after recovering missing data.

The probabilistic radiocarbon calibration approach, which largely has replaced the intercept method in 14C dating, is based on the so-called Bayes' theorem (Bayes 1763). Besides single-sample calibration, Bayesian mathematics also supplies tools for combining 14C results of many samples with independent archaeological information such as typology or stratigraphy (Buck et al. 1996). However, specific assumptions in the “prior probabilities”, used to transform the archaeological information into mathematical probability distributions, may bias the results (Steier and Rom 2000). A general technique for guarding against such a bias is “sensitivity analysis”, in which a range of possible prior probabilities is tested. Only results that prove robust in this analysis should be used. We demonstrate the impact of this method for an assumed, yet realistic case of stratigraphically ordered samples from the Hallstatt period, i.e. the Early Iron Age in Central Europe.

This paper covers three different methods of matching radiocarbon dates to the ‘wiggles’ of the calibration curve in those situations where the age difference between the 14C dates is known. These methods are most often applied to tree-ring sequences. The simplest approach is to use a classical Chi-squared fit of the 14C data to the 14C curve. This gives the calendar date where the data fit best and allows tests of how good the fit is. The only drawback of this method is that it is difficult to ascertain the uncertainty in the date found in this way. An extension of this technique uses a Monte-Carlo simulation to sample possible 14C concentrations consistent with the measurement made and for each of these possibilities performs a Chi-squared fit. This method yields a distribution of values in the calendrical time-scale, from which the overall dating uncertainty can be derived. A third, rather different approach, based on Bayesian statistics, calculates the relative likelihood of each possible calendar year fit. This can then be used to calculate a range of most likely dates in a similar way to the probability method of 14C calibration. The theories underlying all three methods are discussed in this paper and a comparison made for the fitting of specific model sequences. All three methods are found to give consistent results and the application of any one of them depends on the nature of the scientific question being addressed.

Comparison of two sets of marsh-accumulation records from each of three Connecticut (USA) salt marshes, one based on individually calibrated dates and the other on wiggle-match dating of the same series of dates, shows that wiggle-match dating results in more precise and objective reconstructions of longer-term (102–103 yr) changes in accumulation rate. On (sub-)century time scales, wiggle-match dating can reveal steps in the calibrated marsh-accumulation envelope as artefacts of the calibration curve, but may also leave real short-term changes in accumulation rate undetected. Wiggle-matches are non-unique, being dependent on the number, quality and distribution of radiocarbon dates in a sequence, how a series of dates is subdivided into groups (representing intervals of uniform accumulation rate), and what is considered a “best match”. Samples from the studied salt-marsh deposits required no correction for reservoir effects prior to calibration.

The earliest dating of samples taken from logboats found in the area of Poland was done at the Gliwice Radiocarbon Laboratory in the late 1970s and early 1980s. After a 10-year break, the study of their chronology was renewed. The 14C dates (56) include all previously published and new, unpublished results obtained during last several years. Here, we discuss and provide probabilistic interpretation of the calendar age of the dated boats. The calibration of 14C dates was done with the OxCal program for dates less than 300 BP, and with the GdCALIB program for all remaining dates. In distribution of calibrated dates we find a lack of samples between the ages of around 800 BC and 300 AD. This result is surprising and differs from results observed for Central Europe. The remaining age ranges, with high frequency of dates, are in good coincidence with similar periods obtained for Central Europe.

Tree-ring dating of oak logboats was carried out on 60 growth sequences, dated against standard chronologies defined for the area of Poland. The results of 14C dating and tree-ring analyses give consistent chronologies.

We propose a new method of cross-dating the wood samples based on the classical methods of spectral estimation. This method uses the average cross-spectral density as a function of the relative position of the series. Because it is not sensitive to phase shifts in data it is appropriate for cross-dating samples originating from geographically distinct areas.

The accuracy of cross dating depends on the integrity of the samples used, and in the case of well-preserved wood samples, the precision of relative age comparison may reach a single year. The method was tested on two dendrochronological series from Scythian barrows of known age in Southern Siberia: the Pazyryk barrows (the Altai Mountains) and the Dogee-Baary-2 burials (Western Sayan Mountains) separated by 450 km. The analysis has shown that the Pazyryk barrow is younger by 80 ± 4 yr than the Dogee-Baary −2 burials. This result is in agreement with the new chronology of Scythian-related sites suggested for Southern Siberia and Central Asia.