Th / U dating of interglacial and interstadial fen peat and lignite : Potential and limits

The state-of-the-art and potential of the 230Th/U disequilibrium method are discussed for the dating of fen peat and lignite. Recommendations are given for the collection of suitable samples. The numerous interfering factors in 230Th/U dating of fen peat show that a rigorous examination of the reliability of the measured data sets is required if reliable 230Th/U ages are to be obtained. The accuracy of such 230Th/U ages allows a reliable correlation of interglacial and interstadial deposits to the warm periods documented in the SPECMAP timescale but does not yet allow detailed temporal resolution. [230Th/U-Altersbestimmung interglazialer und interstadialer Niedermoortorfe und Ligniten: Potential und Grenzen] Kurzfassung: Der Stand und die Zukunft der Altersbestimmung von Niedermoortorf und Lignit mit der 230Th/U-Ungleichgewichtsmethode werden diskutiert. Empfehlungen zur Auswahl der Probeentnahmestellen und zur Entnahme von geeigneten Proben werden gegeben. Aufgrund der zahlreichen negativen Einflussmöglichkeiten ist eine gründliche Eignungsprüfung der gemessenen Aktivitätsverhältnisse von Uran und Thorium solcher Proben notwendig, um zuverlässige Alter bestimmen zu können. Die Datierungsgenauigkeit ist ausreichend, um interglaziale und interstadiale Ablagerungen sicher den Warmperioden der SPECMAPZeitskala zuzuordnen, noch nicht aber, um sie zeitlich auflösen zu können.


Basic principles of 230 Th/U dating
Peat deposits are important archives of Pleistocene climate change but correlation of individual records with long-term archives such as marine oxygen isotope records (SPECMAP; MARTINSON et al. 1987) remains tentative. Correlations based on palynological and stratigraphic studies are usually based on certain pre-assumptions and can hence not provide any independent age constrains. Sound numerical dating methods seem to be suitable tools to achieve a reliable correlation between the marine and terrestrial climatic records. One of them is the 230 Th/U dating method. It is suitable for all materials that accumulated uranium during their formation and have not been affected by post-sedimentary processes that would mobilise uranium and thorium. When this is the case, the sampled material can be expected to have behaved as a closed system with respect to uranium and/or thorium during ageing. Moreover, it should not be older than 350-500 ka (IVANOVICH & HARMON 1992;BOURDON et al. 2000). The 230 Th/U dating method is based on the radioactive decay series of 238 U. The first longlived daughter isotope 234 U (τ = 2.4525×10 2 ka, CHENG et al. 2000) decays into the isotope 230 Th (τ = 75. 69 ka, CHENG et al. 2000) with the emission of an alpha particle. The half-life of 238 U (τ = 4.468×10 9 a) is considerably longer than that of 234 U. As a consequence, radioactive equilibrium is established after about one million years in most geologically old rocks and sediments. At radioactive equilibrium, all members of the decay series have the same specific activity (which is the decay rate per mass unit), i.e., the activity ratios of any two of the series members are equal to one. Radioactive equilibrium can be disturbed by geochemical processes if the members of the radioactive series have certain geochemical properties. An example is the uranium and thorium isotopes of the uranium-238 decay series. U(VI) ions are soluble in oxygenated water while thorium is practically insoluble. Hence, during weathering, uranium tends to be dissolved in water and thorium bound by clay minerals. Uranium dissolved in groundwater can be incorporated into new systems. Examples are speleothems, foraminifera and fen peat. The most simple 230 Th/U dating model assumes that the new system initially contains only uranium. Radioactive disequilibrium exists between 238 U and 230 Th during ageing. The numerical 230 Th/U clock starts at zero and a 230 Th/ 238 U activity ratio (AR) = 0. The radioactive disequilibrium between 230 Th and 238 U continues and evolves during ageing. A new radioactive equilibrium is approached after 350-500 ka. This process can be used to date sample material and is described by Equation 1, which has to be solved iteratively. Activity ratios are given in square brackets and λ = ln 2/τ: The change in the 230 Th/ 238 U AR with age t is shown in Figure 1 assuming an initial 234 U/ 238 U AR of 1.15, which is characteristic for most marine carbonates. The 234 U/ 238 U AR of samples from terrestrial environments covers a wide range between 1 and 20. The most rapid change in the 230 Th/ 238 U AR takes place between 50,000 and about 200,000 years. The maximum age that can be measured by any dating method is the minimum age of the sample. For the radiometric and mass spectrometric 230 Th/U dating methods, this maximum age is about 350 ka and about 500 ka, respectively (Chapter 7).  230 Th/ 238 U and 234 U/ 238 U activity ratios with increasing age for an initial 234 U/ 238 U AR of 1.15 (corresponding to the mean of marine sediments).
process can be used to date sample material and is described by Eq solved iteratively. Activity ratios are given in square brackets and � The change in the 230 Th/ 238 U AR with age t is shown in Figure 1 as 234 U/ 238 U AR of 1.15, which is characteristic for most marine carbo of samples from terrestrial environments covers a wide range betwe rapid change in the 230 Th/ 238 U AR takes place between 50,000 and maximum age that can be measured by any dating method is the mi For the radiometric and mass spectrometric 230 Th/U dating methods about 350 ka and about 500 ka, respectively (Chapter 7). KAUFMAN & BROECKER (1965) introduced an isotope-ratio evolutio 234 U/ 238 U AR is plotted versus the 230 Th/ 238 U AR. It shows the varia with increasing age. Each initial 234 U/ 238 U AR yields a separate evo point (1,1) at radioactive equilibrium, i.e., at infinite age. The ARs straight lines of their corresponding ages, which are termed "isochr 2 Fen peat and lignite for 230 Th/U dating process can be used to date sample material and is described by Equation 1, which has to be solved iteratively. Activity ratios are given in square brackets and � = ln 2/�: The change in the 230 Th/ 238 U AR with age t is shown in Figure 1 assuming an initial 234 U/ 238 U AR of 1.15, which is characteristic for most marine carbonates. The 234 U/ 238 U AR of samples from terrestrial environments covers a wide range between 1 and 20. The most rapid change in the 230 Th/ 238 U AR takes place between 50,000 and about 200,000 years. The maximum age that can be measured by any dating method is the minimum age of the sample For the radiometric and mass spectrometric 230 Th/U dating methods, this maximum age is about 350 ka and about 500 ka, respectively (Chapter 7). KAUFMAN & BROECKER (1965) introduced an isotope-ratio evolution plot in which the KAUFMAN & BROECKER (1965) introduced an isotope-ratio evolution plot in which the 234 U/ 238 U AR is plotted versus the 230 Th/ 238 U AR. It shows the variation of these two ARs with increasing age. Each initial 234 U/ 238 U AR yields a separate evolution line. All meet the point (1,1) at radioactive equilibrium, i.e., at infinite age. The ARs of coeval samples fit straight lines of their corresponding ages, which are termed "isochrons" (Fig. 2).

Fen peat and lignite for 230 Th/U dating
The history of 230 Th/U dating of fen peat began with a study by TITAYEVA (1966). Her efforts failed as the analyzed Holocene peat was too young for 230 Th/U dating. But TITAYEVA recognised that "excess" (detrital) thorium and uranium have to be taken into account for 230 Th/U dating of fen peat. The first 230 Th/U ages of this material were published by VOGEL & KRONFELD (1980). Because the ages were not corrected for detrital thorium and uranium, as suggested by TITAYEVA (1966), their results were not acceptable by Quaternary geologists. The first chronostratigraphically satisfactory 230 Th/U ages of fen peat were determined by VAN DER WIJK et al. (1986) and HEINJES & VAN DER PLICHT (1992). Fen peat and lignite have a potential for 230 Th/U dating because the fulvic and humic acids they contain have an extremely large capacity for forming complexes with uranium dissolved in groundwater. Moreover, in the reducing milieu of peat, soluble U(VI) is reduced to U(IV) and bound in very stable, immobile uranyl organic complexes. As a result fen peat may have uranium concentrations of up to 100 ppm. The following scenario of fen peat formation is the basis for 230 Th/U dating of fen peat and lignite. Figure 3 shows a schematic section of a growing fen moss in a shallow lake. During its growth, uranium dissolved in groundwater enters the moss and is tightly bound to fulvic and humic acids. Before the peat has been compacted, groundwater can enter even deeper parts of the moss. In the terminal stage, organic material fills the entire volume of the lake. In the succeeding glacial period, the organic deposits are covered by mineral soil and compacted. Figure 4 shows a section of interglacial fen peat sandwiched between detrital material above and below. The uranium dissolved in the groundwater seeping through the bottom and top mineral cover layers during the period of ageing is continuously absorbed in the bottom and top parts of the peat. It has been empirically found that the corresponding rim layers are seldom thicker than 10 cm. The organic material in the rim layers behaves as an open system with respect to uranium and is not datable. As a    230 consequence, groundwater entering the central part of fen peat more than 20-cm-thick should be free of uranium and the central part remains a closed system with respect to uranium. It is suitable for 230 Th/U dating. The uranium and thorium data determined in a chronological study of the Netiesos interglacial site in Lithuania by GAIGALAS et al. (2005) is used in the following as a case study to illustrate the 230 Th/U dating of fen peat and lignite. The authors determined an ESR age on freshwater mollusc shells of 101.5 ± 11.5 ka. Six samples were collected between 30 and 55 cm depth. All were analyzed using the L/L (leachate/ leachate) method (SCHWARCZ & LATHAM 1989;KAUFMAN 1993); the two outer samples (from 30 and 55 cm depth) were not analyzed with the total sample dissolution (TSD) method (BI-SCHOFF & FITZPATRICK 1991;LUO & KU 1991). The U and Th isotope ratios were measured radiometrically at St. Petersburg University and summarized with 2σ standard deviations in a research report ( Table 1). The standard deviations of these results deviate slightly from those published by GAIGALAS et al. (2005). The reason is unknown.

Factors that interfere with 230 Th/U dating of fen peat and lignite
There are two main factors which render 230 Th/U age determination difficult or even impossible: 1. Open-system conditions with respect to uranium and thorium, as well as 2. Detrital contamination with allochthonous 230 Th.
Open system conditions: Fen peat and lignite theoretically offers a chemically ideal milieu for closed-system conditions with respect to uranium. In contrast to carbonates (e.g., corals, foraminifera, speleothem), the reducing milieu of the peat means that any uranium present is reduced to insoluble U(IV), which form stable complexes with inorganic and organic ligands, such as humic and fulvic acids. Moreover, movement of humic acids over large distances and, therefore, post-depositional displacement of uranium can be largely excluded in the confined conditions of interglacial peat deposits. Likewise, α-recoil only moves the uranium daughter isotopes over distances that are smaller than the common sample volume of several cm 3 (HENDERSON & SLOWEY 2000). According to the above-described scenario for the formation of interglacial fen peat deposits, the central part may be considered as a closed system with respect to uranium and thorium. However, there are several possible situations in which this may not be true. Some fen peat deposits contain thin sand layers which may be conduits for oxygenated groundwater to the middle of the deposit. In the peat adjacent to these layers, open-system conditions with respect to uranium may have been present. Occasionally samples from the central part of a deposit without visible stratigraphic peculiarities yield outlier 230 Th/U ages. A possible explanation is that permafrost might have expanded the frozen fen peat and on warming voids were created into which groundwater containing uranium could have penetrated before the peat was recompacted. Detailed descriptions of the sampling site may help correlate outlier 230 Th/U ages to stratigraphic peculiarities. There are other scenarios which may occur in permafrost regions (SCHIRRMEISTER et al. 2002). For example, superficial peat layers may become dry down to a depth of 1-2 m during the summer months, allowing oxygenated melt water to enter the deep parts of the deposit. Many dated interglacial peat deposits in western Asia have been exposed along river banks. If the river rises above the level of the peat layers, samples collected from up to several decimetres depth may have been affected by river water. Exposed fen peat deposits well above the present river level might also have been affected by seeping rainwater. Only drill cores of sufficient length seem to guarantee suitable peat samples. Thorium typically exists in the +4 and +5 oxidation states. The ions are readily hydrolyzed and either precipitate or are adsorbed on detrital particulates (inorganic or organic), clay minerals and iron (oxy)hydroxides. Thorium ions can be transported with humic acids (RICHARDS & DORALE 2003). But movement of humic acids in confined peat beds can be excluded.   Detrital contamination with allochthonous 230 Th: Airborne dust or water-borne fine-grained material may become incorporated during the formation of fen peat. These contaminants usually contain thorium, which is often bound to detrital clay. The presence of 230 Th in the detrital material means that the 230 Th/U clock shows 230 Th/U ages that are too old. The allochthonous (detrital) 230 Th decays during ageing while the content of radiogenic 230 Th increases as a result of the decay of 234 U. Hence the 230 Th/U age error due to detrital contamination decreases with increasing age. The age error caused by an initial detrital 230 Th content is often large and may amount to several 1000 years up to 100,000 years. Therefore, the two or more sources of 230 Th (one radiogenic and at least one detrital) have to be distinguished and quantified for correction of the 230 Th/U ages.
Limits on the 230 Th/U method for the dating of peat and lignite: The above-described scenario of both accumulation and binding of uranium in fen peat limits the use of the 230 Th/U dating method for this material. Accurate 230 Th/U ages of sublayers cannot be expected. During the growth of the moss the peat remains very porous and groundwater or lake water containing dissolved uranium may enter even deeper layers. Hence the 230 Th/U ages will always be younger than the actual ones. Moreover, as fen peat formation is restricted to shallow lakes, which exist mainly at the end of the interglacial or interstadial periods, the corresponding organic deposits often represent the end of the peat-formation period or the interglacial period rather than the beginning. Hence, suitable material of palynologically distinct sub-layers of the first part of the interglacial period can seldom be provided.
Exclusion of raised-bog peat from 230 Th/U dating: Raised-bog peat receives uranium mainly from particulate matter in precipitation. Therefore, the concentration is usually low and there may be different sources of thorium and uranium with different isotopic compositions. Moreover, as raised bog peat is often only  weakly decomposed and the content of humic acids is low the ability to absorb actinides is limited. As a consequence, all attempts to date raised-bog peat using the 230 Th/U dating method have failed.

Determination of the reliability of data sets for the 230 Th/U dating of fen peat and lignite
Based on the above-described scenario for fen peat formation and the numerous factors that interfere with 230 Th/U dating of fen peat, it is necessary to identify and to discard data sets from samples that are not suitable for 230 Th/U dating. Such sample material does not fulfil the two basic premises of this dating method: (1) It has behaved as an open system with respect to uranium (and thorium) and/or (2) it contained more than one detrital contaminant. Three simple tests are used for the initial reliability determination: depth profiles of (i) uranium concentration or specific 238 U activity, (ii) 234 U/ 238 U AR, and (iii) ash content. Uranium-depth profiles may show up samples which behaved as an open system with respect to uranium on the basis of elevated uranium concentration relative to the base level in the central part of the peat profile. The 234 U/ 238 U AR in the top and bottom rim layers can be expected to differ from the central part if 234 U was preferentially leached or accumulated. The third test is determination of the ash content of the sample: an elevated ash content may indicate the presence of sand layers. Figure 5 shows the results of the three tests for the Netiesos site (GAIGALAS et al. 2005). The 238 U activity of samples L-4 and L-5 (45 and 50 cm) are 30-40 % above the other samples and therefore these samples potentially violate the two basic premises for 230 Th/U dating. The authors do not describe any disturbance of the stratigraphy, and therefore other information is necessary to determine which samples need to be excluded from the age determination. For example, the decrease in ash content with depth points to the possibility of open-system conditions in the upper part of the peat.
Examination of 222 data sets obtained from 33 fen peat sections in Eurasia (e.g. ARSLANOV 2005) has yielded evidence that 230 Th/U ages can be satisfactorily correlated with the SPEC-MAC chronology even from samples with considerably elevated or depleted uranium concentrations, or from samples with anomalous 234 U/ 238 U ARs or that contain a high ash content. Hence, the three tests do not reliably identify samples that do not fulfil the basic premises for 230 Th/U dating. More sophisticated tests are required. The theoretical background was developed in the 1960s. KAUFMAN & BROECKER (1965) introduced the concept of a binary mixture of radiogenic 230 Th and allochthonous detrital 232 Th. If the latter has only one source, the proportion of detrital 230 Th can be estimated via the thorium concentration or the 232 Th activity. As the half-life of 232 Th is very large (τ = 14.01×10 9 a), it behaves similar to a stable isotope. Hence, its activity has not changed during the aging of the sample and hence is a measure of the initial detrital 230 Th content in the sample. One of the main tasks of 230 Th/U dating of fen peat and lignite is to determine the initial 230 Th/ 232 Th AR (thorium index) in order to correct the corresponding ages for the detrital contamination (Chapter 5). The principal feature of binary mixing is that any two properties of the two components show a linear correlation. Hence, the plot of any two isotope ratios of uranium and/or thorium in peat samples -according to the KAUFMAN & BROECKER concept -must also yield a straight line. If this requirement is not fulfilled, the basic premises for 230 Th/U dating are not fulfilled: The material did not behave as a closed system with respect to uranium and thorium and/or there were more than two sources of detrital 230 Th. In 1970, OSMOND et al. (1970     Th/ 232 Th und 234 U/ 232 Th sowie die Y-Achsenabschnitte die 230 Th/ 238 U-und 234 U/ 238 U-Aktivitätsverhältnisse. 230 Th/U dating of interglacial and interstadial fen peat and lignite:Potential and limits 85 ( 234 U/ 232 Th AR versus 238 U/ 232 Th AR; Figs. 6 and 7 top). The Rosholt-I plot is often incorrectly referred to as an "isochron" plot. The slope of the least-squares fitted straight line in a plot of the 230 Th/ 232 Th versus 238 U/ 232 Th activity ratios of coeval samples equals the 230 Th * / 234 U AR of the radiogenic 230 Th * and its intercept on the y-axis equals the decay-corrected initial 230 Th/ 232 Th AR (detrital correction factor = f). The Rosholt-II plot yields the present 234 U/ 238 U AR. Both ARs are required to calculate the 230 Th/U ages (Eq. 1). An "isochron" slope of 0 corresponds to a 230 Th/U age of zero; an "infinite" age has a slope of 1 (radioactive equilibrium). Figures 6 and 7 show two sets of the four Rosholt-Osmond plots of the ARs obtained from the uranium and thorium analyses of the fen peat at the Netiesos site using the L/L and TSD methods. The displacement of any pair of AR values from any of the four mixing lines indicates that the corresponding sample may not fulfill the basic premises for 230 Th/U dating. Especially, the Rosholt I and II plots give "a misleading impression of an well-defined isochron for data of little statistical power" (LUDWIG 2003). LUDWIG mentions three factors as the reason. (1) The 238 U/ 232 Th and 234 U/ 232 Th ARs tend to reflect the U/Th element ratio. The abundances of both 234 U and 230 Th of non-zero-age materials are generally highly correlated with the concentration of uranium. Therefore, the scatter of the 238 U/ 232 Th AR greatly amplifies even a dubious isochron trend. As a result the correlation coefficient of the Rosholt isochrons are usually higher than 0.90 and also considerably greater than those of the Osmond plots. (2) The ARs, especially if measured radiometrically, are highly correlated if the denominator isotope 232 Th activity is low. Due to the large analytical error the data scatter along lines subparallel to most isochrons and paradoxically strengthen the visual impression of an isochron. 3) The data points of radiometrically measured ARs plotted with error ranges appear to fall more precisely on the regression line than the error ranges imply, even if there is much more scatter than the errors would otherwise indicate ("zero"-correlation problem;CHAYES 1949CHAYES , 1971. To obtain a well defined mixing line, (LUDWIG & TITTERINGTON 1994), data sets from at least two coeval samples with widely differing detrital 230 Th content must be available. In the case of fen peat, a much larger number of samples should be analysed. Coeval samples with similar detrital contamination yield a data cluster, which excludes the possibility of a mixing line. Moreover, the y-axis intercept of the mixing line must be positive, as a decay-corrected negative initial 230 Th/ 232 Th AR is theoretically excluded. The mean of negative decay-corrected AR values can be accepted only if its confidence interval includes zero. The four mixing plot tests -especially those by Osmond -are a sensitive check for identifying unsuitable data sets. However, it is recommended to include the three simple tests described above in a rigorous examination of the reliability of U/Th age data. Moreover, supplementary tests should be found which also allow an objective and reproducible identification of suitable and unsuitable data sets.

Detrital contamination and the correction factors
According to the KAUFMAN & BROECKER (1965) concept, the radiogenic 230  amplifies even a dubious isochron trend. As a result the correla isochrons are usually higher than 0.90 and also considerably gre plots.
(2) The ARs, especially if measured radiometrically, are h denominator isotope 232 Th activity is low. Due to the large analy along lines subparallel to most isochrons and paradoxically stre of an isochron.
3) The data points of radiometrically measured appear to fall more precisely on the regression line than the erro is much more scatter than the errors would otherwise indicate ( CHAYES 1949CHAYES , 1971).
To The four mixing plot tests especially those by Osmond are a identifying unsuitable data sets. However, it is recommended to described above in a rigorous examination of the reliability of U supplementary tests should be found which also allow an objec identification of suitable and unsuitable data sets.

Detrital contamination and the correction factors
According to the KAUFMAN & BROECKER (1965) with the concentration of uranium. Therefore, the scatter of the 238 U/ 232 Th AR greatly amplifies even a dubious isochron trend. As a result the correlation coefficient of the Rosholt isochrons are usually higher than 0.90 and also considerably greater than those of the Osmond plots.
(2) The ARs, especially if measured radiometrically, are highly correlated if the denominator isotope 232 Th activity is low. Due to the large analytical error the data scatter along lines subparallel to most isochrons and paradoxically strengthen the visual impression of an isochron. 3) The data points of radiometrically measured ARs plotted with error ranges appear to fall more precisely on the regression line than the error ranges imply, even if there is much more scatter than the errors would otherwise indicate (zero-correlation problem;CHAYES 1949CHAYES , 1971. To obtain a well defined mixing line, (LUDWIG & TITTERINGTON 1994), data sets from at least two coeval samples with widely differing detrital 230 Th content must be available. In the case of fen peat, a much larger number of samples should be analysed.
Coeval samples with similar detrital contamination yield a data cluster, which excludes the possibility of a mixing line. Moreover, the y-axis intercept of the mixing line must be positive, as a decay-corrected negative initial 230 Th/ 232 Th AR is theoretically excluded. The mean of negative decay-corrected AR values can be accepted only if its confidence interval includes zero.
The four mixing plot tests especially those by Osmond are a sensitive check for identifying unsuitable data sets. However, it is recommended to include the three simple tests described above in a rigorous examination of the reliability of U/Th age data. Moreover, supplementary tests should be found which also allow an objective and reproducible identification of suitable and unsuitable data sets.

Detrital contamination and the correction factors
According to the KAUFMAN & BROECKER (1965) where g o is initial 234 U/ 232 Th AR and g is the decay-corrected g o . Both of these equations to correct for detrital thorium and detrital uranium assume there was only source of contamination. No detrital correction is needed if most of the samples contain little or no detrital contamination and inferring effects can be excluded, like in the case for speleothems and or corals. In such cases the slope of the mixing lines of the Rosholt-I and Rosholt-II plots equal the 230 Th/ 238 U AR and 234 U/ 238 U AR, respectively, which are required to calculate the 230 Th/U age (Equation 1). The computer program ISOPLOT (LUDWIG 2003) can be used to calculate the corrected 230 Th/U age and its standard deviation. The theoretical background for calculating the "error-correlated" standard deviation is given by TITTERINGTON (1994) andLUDWIG (2001;2003).
In the case of fen-peat dating a statistical check of the individual 230 Th/U ages is recommended in order to identify outliers. This cannot be done with the ISOPLOT program. Therefore, Equations 2 and 3 should be applied to each data set of peat and lignite samples (GEYH 1994;. The required f and g values and their standard deviations are obtained from the intercept of the mixing lines with the y-axis of the Rosholt-I and Rosholt-II plots or the iterative approximation of the decay-corrected initial 230 Th/ 232 Th AR described below. The standard deviation of the detritus-corrected 230 Th activity is calculated using the usual error propagation equations. In most cases the correction for detrital uranium is not necessary because the 234 U/ 238 U AR and 232 Th/ 238 U AR do not correlate for most peat samples. This is also the case for the Netiesos site (Figs. 6 and 7 bottom). An independent approach to the determination of the detrital correction factor f is to approximate the decay-corrected initial 230   Th/U dating of interglacial and interstadial fen peat and lignite:Potential and limits 87 of coeval samples containing contamination from different sources scatter widely. Iteratively increasing the detrital correction factor f causes the points to move to the left in the isotope-ratio evolution plot of KAUFMAN & BROECKER (1965). The higher the detrital contamination the faster the points move to the left. The optimum detrital correction factor f is obtained when the points of the cluster are the closest together. The points in the cluster move further apart when the f value is increased beyond the value for the minimum width of the cluster. An elliptical shape of the cluster is caused by differences in the initial 234 U/ 238 U AR (Fig. 8). An effect resulting from -recoil as observed in carbonates (HENDERSON & SLOWEY 2000) cannot be completely excluded for fen peat. A plot of the standard deviation of the 230 Th/U age versus the f value provides a measure of the scatter of the individual ages. For the case study from the Netiesos site , the scatter of the L/L and TSD 230 Th/U ages approaches a minimum at f(L/L) = 0.412  0.080 and f(TSD) = 0.534  0.011 (Fig. 9). The plots again show that sample L-6 does not fulfill the premises for 230 Th/U dating. An iterative method developed to determine the decay-corrected initial 230 Th/ 232 Th AR provides an additional test for a rigorous examination of the reliability of the data sets (GEYH 2001). The iteratively detritus-corrected 230 Th/U ages plotted against the iteratively determined detrital correction factor yields lines with different slopes. If all samples behaved as closed systems with respect to uranium and only one detritus component was present in the samples, the lines should intersect at only one point -the actual f value. If one or more lines do not pass through this point the corresponding sample is identified as unsuitable.
For the case study of the Netiesos site, the bold line for sample L-6 in the L/L plot belongs to outlier data which do not fulfill the premises of the 230 Th/U dating method. This line does not pass through the common intersection point.

230 Th/U age and random uncertainty
The data sets suitable for calculation of a 230 Th/U age and its standard deviation are selected on the basis of the rigorous examination of their reliability described above. The detrital correction is applied to each measured 230 Th activity (Equation 2; Chapter 5) and the 230 Th/U ages are calculated using Equation 1. The 2 test is then applied to the corrected 230 Th/U ages, outliers are discarded, and the mean 230 Th/U age is calculated. The standard deviation of the mean 230 Th/U age is obtained using the Gaussian method for error determination of the 230 Th/U ages of the samples.

Minimum 230 Th/U age of samples
The limited precision of uranium and thorium isotope activity measurements means that there is an upper limit of the 230 Th/U age that can be determined. This limit is the minimum 230  The decay-corrected initial 234 U/ 238 U (Y) is given by The uranium and thorium ARs of an infinitely old sample are given by a straight line in the isotope-ratio evolution plot. The equation for this isochron is derived from Equation 4 assuming an age t = ∞: The ellipse defined by the X and Y random uncertainties of any data point should touch but never cut the infinite-age isochron. When the infinite-age isochron is tangent to the ellipse of the X and Y random uncertainties, the minimum 230 Th/U age t min of the sample can be determined from the X min and Y min (Fig. 11). The equation of the ellipse with maximum axis lengths of ± 2 X and ± 2 Y around X min and Y min is given by Equations 4-7 are solved iteratively by decreasing or increasing the age until the error ellipse of a date touches the infinite-age isochron. A detritus-corrected 230 Th/U date is considered reasonable if it is smaller than the corresponding minimum 230 Th/U sample age t min (GEYH &   MÜLLER 2005). Otherwise the result has to be expressed as the minimum 230 Th/U age.

Precision and accuracy of 230 Th/U dating of fen peat
The precision and accuracy of 230 Th/U dating of fen peat may be illustrated by three case studies. GAIGALAS et al. (2005) published L/L and TSD 230 Th/U ages of 108.8 ± 8.7 ka and 80.3 ± 5.9 ka, respectively, for fen peat deposits at the Netiesos site in Lithuania. The higher value from the L/L data was obtained after rejection of the data from the uppermost (L-1: 30 cm) and lowermost (L-6: 55 cm) peat samples. A rigorous examination of the sample data sets was not carried out. The decision to reject these two values was based neither on an elevated 238 U activity nor on ash content (Fig. 5). A rigorous check of the reliability of the L/L data sets identified only sample L-6 as unsuitable for 230 Th/U dating. The L-6 data do not fit the mixing lines of the Rosholt-I and Osmond-I plots (Fig. 6), no minimum cluster size was found when the L-6 data was included in the evolution plot (Fig. 8), and in the iterative plot the L-6 line does not pass through the common intersection point of the L-1 to L-5 lines. Sample L-2 has a slightly elevated ash content (Fig. 5) and its AR values do not fit well the mixing lines of the Rosholt-I and Osmond-I plots (Fig. 7). Other indications that L-2 is unsuitable were not found. A rigorous check of the reliability of the TSD data sets does not identify any outliers. The 230 Th/U age was calculated with the Equations 1 and 2 and yielded 78.4 ± 4.3 ka and a detrital correction factor of 0.519 ± 0.161. The χ 2 test applied to the four data sets yields χ 2 = 0.4. Iterative evaluation yielded an age of 76.4 ± 1.3 ka with a detrital correction factor of 0.534 ± 0.010. A 230 Th/U age of 72.6 ± 10.0 ka and a detrital correction factor of 0.268 ± 0.365 and a χ 2 = 0.8 were calculated using Equations 1 and 2, discarding the data from samples L-2 and L-6. Iterative evaluation yielded a detrital correction factor f = 0.456 ± 0.060. An age of 77.0 ± 11.1 ka and a detrital correction factor f = 0.199 ± 0.381 (χ 2 = 0.4) were obtained when only the data from sample L-6 was discarded and the iterative evaluation yielded f = 0.419 ± 0.080. Disregarding sample L-6 shifts the calculated lines in the Rosholt-I and Osmond-I of the L/L data plots to the right and downwards, respectively. This yields an acceptable 230 Th/U age of 77.0 ± 11.1 ka. It is in agreement with the TSD 230 Th/U age of 78.4 ± 4.3 ka. This age falls within the time span of MIS 5a (Brørup interstadial) between 76 and 78 ka of the SPECMAP time scale (MARTINSON et al. 1987).  The second case study presented here was the first attempt to check whether it is possible to date fen peat and lignite by the 230 Th/U dating method (GEYH 2001). Eemian peat deposits in Europe seemed to be the most suitable for this as they contain the only Quaternary interglacial material that can be palynologically firmly identified and related to the SPECMAP timescale (GEYH 2001). Unfortunately, no cores were available and all samples were collected from exposures with the associated risk of interfering processes. Seven sites in Germany yielded a mean 230 Th/U age of 111.9 ± 1.2 ka for the detritus-corrected 230 Th/U ages and 114.4 ± 1.9 ka using Equations 1 and 2. The agreement is good but the ages are below what is expected from SPECMAP chronology. There are several possible reasons for this. The data sets were not subjected to a rigorous reliability test. The samples might not have completely fulfilled the two basic premises for 230 Th/U dating due to the lack of optimum sampling conditions. In any case, the 230 Th/U ages and their standard deviations for these seven sites may be considered as representative of the attainable accuracy of 230 Th/U dating of fen peat. The precision, however, is good enough to firmly assign the dated Eemian interglacial deposits to MIS 5e but is not sufficient to assign a precise chronological age to sub-layers. In the third case study, the 230 Th/U ages for Holsteinian/Hoxnian Interglacial peat were obtained from drill cores and agree better with the SPECMAP chronology (GEYH & MÜLLER 2005). The Hoxnian Interglacial in England has been correlated with the Holsteinian Interglacial (TURNER 1970). A mean 230 Th/U age of 317 ± 14 ka was obtained from the fen peat at Tottenhill in the Nar Valley of NW Norfolk (ROWE et al. 1997). The 230 Th/U ages from the Marks Tey site in the UK, which has been palynologically classified as Holsteinian, (ROWE et al. 1999), have a very low precision and could be interpreted only by using a probabilistic approach. The authors concluded that these deposits correlate with MIS 11 or some older stage with 87 % confidence. Using Equations 4 to 7, 230 Th/U minimum ages of >165 to >245 ka were obtained for their data (GEYH & MÜLLER 2005). This implies that the measured U and Th data reliably allow only the statement that the geological age of the Holsteinian samples is greater than MIS 7. B The Bossel site about 30 km west of Hamburg contains two organic layers: one above and one below the sediments of the Holsteinian Sea. This location became the reference site for the Holsteinian Interglacial in northern Germany as declared by the European Commission on the Stratigraphy of the Quaternary. The "isochron" plot of the AR of ten samples from both peat layers yielded a mean detritus-corrected 230 Th/U age of 323 ± 5 ka with individual detritus-corrected 230 Th/U ages between 298 and 347 ka. The lower peat layer yielded a mean detritus-corrected 230 Th/U age of 312 ± 3 ka with individual detritus-corrected 230 Th/U ages between 293 and 369 ka. The four 230 Th/U ages of the upper peat layer yielded a detritus-corrected mean age of 327+50/-37 ka. These case studies provide evidence that 230 Th/ U dating of fen peat and lignite does not yield high-precision ages. The accuracy may be around 10,000 a. But this is adequate to reliably relate any interglacial peat deposit to only one of the documented warm periods of the SPECMAP chronology (MARTINSON et al. 1987).

Sampling, analysis and isotope ratio determination
To fulfil the requirements for isochron 230 Th/U dating the samples must be coeval from the same interglacial or interstadial. For this it is sufficient to take samples from anywhere within the centre of the fen peat deposit. Rim layers of at least 10 cm thickness have to be excluded from the analysis. It is recommended to take the samples from a monolith or a drill core. At least four samples with a wide range of concentration of the detrital contamination, but even better more should be dated. The sample preparation is more or less the same for both measurement methods for determining the ARs and consists of several steps. Ultrapure 230 acids which are free of even traces of uranium and thorium have to be used. The chemical preparation should be done in a clean-air laboratory in order to lower the risk of any detrital contamination. This is especially important if small samples are treated for measurement in a TIMS (thermal-ion mass spectrometer) or MC-ICP-MS (multi-collector inductively coupled plasma mass spectrometer). In radiometric laboratories repeated and comprehensive contamination tests have to be carried out. Preferentially, peat should be treated by "total sample dissolution" (TSD) (BISCHOFF & FITZ-PATRICK 1991;LUO & KU 1991). This method includes both lattice-bound and adsorbed thorium. Empirically it has been found that 230 Th/U ages obtained by the "leachate/leachate" method (SCHWARCZ & LATHAM 1989;KAUFMAN 1993) often deviate from the palynologically expected age and do not fit the SPECMAP timescale. Selective leaching of peat samples does not adequately separate the radiogenic and detrital 230 Th components. In addition, the radiogenic 230 Th may become readsorbed on detrital material during incomplete dissolution.
The following steps are taken in the "total sample dissolution" method: − The samples of 3-5 g of fresh or 0.3-0.5 g of dry peat are taken from visibly undisturbed micro ranges within monoliths or cores. − The surface of each sample is removed. The peat is then broken into pieces, dried and ignited in a quartz tube at a maximum temperature of about 800 °C in a stream of oxygen to burn all organic material. Insoluble glass beads do not form below this temperature and, therefore, loss of uranium and thorium is avoided. − The weighed peat ash is treated with NaOH in order to remove traces of humic acids. The residue is completely dissolved in a con. HF/HNO 3 /HCL mixture to completely dissolve uranium and thorium. If the L/L technique is applied, the sample is leached with a HNO 3 /HCl mixture for at least 6 h. Next, 0.5 mL of a 229 Th spike and 0.5 mL of a 233 U/ 236 U double spike are added as a check of thermal fractionation in the ion source.
The solution is heated with an infrared lamp for up to 20 h in order to ensure perfect mixing of sample and spike. − The leachate is separated from the residue by centrifuging. − Uranium and thorium are co-precipitated with Fe(OH) 3 . − The precipitate is dissolved in con. HNO 3 . − Uranium and thorium are separated from the mixture by standard ion-exchange chemistry using an actinide-specific resin (DOWEX 1×8 100-200 mesh). Thorium is eluted from the column by dilute HCl, uranium is then removed with HBr. 230 Th/U age determinations are preferentially done now by TIMS and MC-ICP-MS. Details are given by SCHOLZ & HOFFMANN (2008). Radiometric measurements of the ARs require at least ten times more material and are less precise by one order of magnitude. The extracted uranium and thorium are electroplated on stainless steel discs. The latter are put into an alpha spectrometer and the emitted alpha particles are counted. Due to the long 230 Th half-life of 75,690 ka, only one out of 6×10 6 atoms of 230 Th decays during the common measurement time of one week. Hence, especially the counting statistics limits the precision of radiometric 230 Th/U ages.

Conclusions
Interglacial fen peat and lignite with ages of up to 350-500 ka can be dated with the 230 Th/U dating method. The main prerequisite is that the samples are collected from the parts of a deposit that behaved as closed system with respect to uranium during ageing and have no more than one detrital component. In the case of fen peat, the bottom and top rim layers of about 10 cm have to be discarded. Drill cores yield better samples than monoliths collected from exposed outcrops. Small samples of a few cm 3 (about 1 g of dry peat) are sufficient for TIMS and MC-ICP-MS 230 Th/U dating. Radiometric 230 Th/U dating requires at least ten times more material. The coeval samples should have a wide range in the amount of detrital contamination. A MEBUS A. GEYH single uranium and thorium isotope analysis of peat does not yield a reliable 230 Th/U age. The TSD analytical method seems to be superior to the L/L method. Reliable 230 Th/U ages, however, always require a comprehensive and rigorous reliability check of the measured data in order to identify samples that were unsuitable for 230 Th/U dating. The accuracy of 230 Th/U ages of fen peat and lignite is around 10 ka, which is sufficient for reliable correlation with the warm periods of the global marine δ 18 O chronology.