4. Scientific Dating Methods

4.1 Radiocarbon dating

Alex Bayliss and Peter Marshall

Radiocarbon (¹⁴C) is a naturally occurring radioactive isotope of carbon that is formed in the upper atmosphere when cosmic radiation interacts with nitrogen atoms. It is unstable, with a half-life of 5730±40 years. It is taken up by living organisms, but decays after death so that the proportion of ¹⁴C in the dead organism decreases over time. By measuring the proportion that remains, the elapsed time since death can be estimated.

In principle any organic material that was once alive can be dated, including bone, carbonised or waterlogged plant materials, and marine shell. Radiocarbon is, however, very difficult to measure, in large part because the ¹⁴C concentration in living material is extremely low (about 1 in every 1 million carbon atoms). This makes detecting a radiocarbon atom in a sample at the limit of detection (about 55 ka) equivalent to identifying a single specific human hair that might occur on the head of any of the human beings alive on earth today!

This means that it is much more difficult to date Pleistocene samples accurately than to date more recent samples (which contain more radiocarbon). This is illustrated in Table 2, which shows the impact on the reported radiocarbon age of modern contaminants on samples of different actual ages.

Since the introduction of Accelerator Mass Spectrometry (AMS), which dates samples less than 1g in weight, the absolute amount of contaminant needed to cause such offsets is tiny. Such contamination can cause samples that are much earlier than the limit of radiocarbon dating to produce finite ages (Busschers et al. 2014). The pressing need to avoid or remove contamination in older samples has practical implications for how Pleistocene samples are collected in the field and processed in the laboratory.

In the field extreme care should be taken to ensure that modern contaminants such as hair or hand-cream do not come into contact with samples. Bone, antler, ivory, charcoal and shell samples should be wrapped in tin foil and placed in clearly labelled plastic bags. Irreplaceable artefacts are often sampled by specialists from the dating laboratory to minimise intervention. Sediment samples must be securely wrapped in black plastic and refrigerated as soon as possible after retrieval.

Sub-sampling for radiocarbon dating, either by hand-picking macrofossils using tweezers or sieving in water, should be undertaken swiftly in a clean environment. Be particularly wary of fibres from paper towelling. Macrofossils should be stored with a small amount of water in a glass vial with a screw lid that has a foil liner and refrigerated. For all potential samples, organic consolidants, fungicides, etc. must be avoided.

Research continues to be undertaken on refining the chemical procedures used for preparing Pleistocene samples for radiocarbon dating. Ultrafiltration of gelatin extracted from bone, antler and ivory samples of this age is now routine (Brown et al. 1988; Jacobi et al. 2006), although other methods are also under development (Linscott et al. 2024; Deviese et al. 2018). Improved accuracy may also be obtained by implementing more complex pretreatment for charcoal samples (Bird et al. 1999; Ascough et al. 2009), and refined pre-screening and preparation methods for ornaments made from marine shell (Douka et al. 2010).

Materials selected for dating must not only contain sufficient carbon and be uncontaminated, but they must also have a secure association with the human activity or environmental event that is the target of the dating programme. Precision may be improved if sequences of related samples can be obtained (Bayesian Chronological Modelling).

Given the technical difficulties of accurate radiocarbon dating in this period, replicate measurements should be undertaken where sufficient material is available. Suitable datable material is often scarce on Pleistocene sites, but it is essential that the reliability of the chronologies of this period are not undermined by dating unsuitable material simply through the lack of better samples.

The following information must be published for each radiocarbon measurement:

• details of the facility/facilities that produced the results and how samples were pretreated, prepared for measurement, and dated;
• details of the radiocarbon results and associated measurements and how these have been calculated;
• details of the material dated and the context from which it came.

Bayesian Chronological Modelling provides examples of the information that should be provided for each radiocarbon date. Note that at the limit of the technique some radiocarbon ages may be quoted with asymmetrical error terms (for example, GrN-12876 from Lynford Quarry, which produced an age of 35,710 +930/–830 BP); others may produce minimum ages (for example OxA-11572 also from Lynford Quarry, which produced an age of >49,700 BP).

Radiocarbon calibration is now undertaken using a set of internationally-agreed calibration curves that extend back to 55,000 cal BP. Terrestrial samples from the northern hemisphere should use IntCal20 (Figure 11; Reimer et al. 2020) and marine samples should be calibrated using Marine20 (Heaton et al. 2020) with an appropriate local ΔR (‘Delta R’) correction (see Bayliss and Marshall 2022 section 1.6).

All radiocarbon results within this range should be calibrated, and details published of the calibration protocols used, including any reservoir corrections employed. Calibration in this period is, however, likely to be subject to significant refinement over the coming decades, so it is essential that laboratory codes and uncalibrated radiocarbon ages are also published to enable them to be recalibrated with new calibration curves in due course.

Where Bayesian Chronological Modelling is employed, calibration is simply part of the modelling process and it may be more appropriate to quote posterior density estimates rather than simple calibrated date ranges.

4.2 Uranium-Thorium dating

Alistair Pike

Uranium-Thorium (U-Th) dating exploits the build-up of the isotope ²³⁰Th (itself radioactive) from the decay of ²³⁸U to ²³⁴U to ²³⁰Th, within the Uranium-series radioactive decay chain. Over time, the activity ratio ²³⁰Th/²³⁸U builds up until radioactive equilibrium is reached, which gives a practical older limit to the method of around 500 ka. The younger limit is constrained by our ability to measure low abundances of ²³⁰Th. This depends on the sample size and its Uranium concentration, but dates typically can be produced on samples a few centuries old.

The technique is suitable for calcium carbonate (calcite) precipitates such as stalagmites, stalactites and flowstones (collectively known as speleothems), and for travertines and tufa (for example Richards and Dorale 2003). Speleothems can occur associated with archaeology in cave deposits, and travertine and tufa occasionally in open air sites.

The error on a U-Th date depends on its age. Under ideal circumstances, measurements of the isotopic ratios using modern mass spectrometric methods can be made to less than ±0.5% (at 2σ), which can lead to uncertainties of less than 100 years in 10 ka. But as the sample age approaches the limit of the method, the errors can get far larger. For example, a 0.5% measurement error (on each isotopic ratio) translates to errors of ±1.2 ka at 100 ka and of +40/−31 ka at 400 ka. Note that the errors are noticeably asymmetrical towards the limit of the technique.

Problems are commonly encountered from detrital contamination of the calcite (for example by cave sediments and particulates). Not correcting for such contamination would lead to older apparent dates. The level of detritus is monitored by measurement of the common isotope of Thorium, ²³²Th, usually expressed as the activity ratio ²³⁰Th/²³²Th. High values (for example >100) indicate low levels of contamination, whereas values <5 indicate severe contamination.

For low and moderate levels of contamination a correction can be applied using an assumed ²³⁰Th/²³²Th ratio for the detritus with a large uncertainty propagated to the calculated date. For highly contaminated samples, the errors on corrected ages may become so large that the dates are not useful. An alternative strategy is to take multiple same-age samples (for example from a single growth layer of speleothem) to construct an ‘isochron’ to correct for detritus. Again, the errors will increase, sometimes drastically.

An additional, though apparently rare, problem is the leaching of Uranium or Thorium from the calcite (open-system behaviour), which can give older or younger apparent dates. Where this is suspected, speleothems can be sampled sequentially along their growth axis. U-Th dates not conforming to their stratigraphic order may indicate open-system behaviour.

When selecting samples, it is worth noting that dates on calcite are only indirect dates for the associated archaeology, but can provide maximum, minimum or bracketing ages for archaeological deposits (Table 3). Securely demonstrating the stratigraphical relationship between the samples dated and the archaeology is of utmost importance.

When taking samples, it is worth considering the worst-case scenario: that the samples will be detritally contaminated and possibly be open system. Samples should be taken that are suitable for multiple sub-sampling in the laboratory. This will enable the construction of an isochron and/or checks for any open system behaviour, even if these steps are not eventually required.

An ideal sample would be the complete sequence of growth layers of a flowstone floor that formed directly over or between two archaeological layers. These can be detached as a block, cut with a grinder or cored with a coring drill (Figure 12).

Photographic and other documentation of the position of the sample, and especially its relation to archaeological layers, is essential, as well as is indicating the uppermost (youngest) layer on the sample.

Where speleothem formation is very active, long sequences of samples bracketing different layers can produce detailed chronologies for sites (for example Hoffmann et al. 2013). Sample storage is straightforward and can be in individual plastic bags, or for small samples, clean plastic tubes.

Occasionally, it is not possible to remove complete samples without undue damage to the archaeology or to the cave (for example in the case of calcite deposits on top of cave paintings; Pike et al. 2012). In these cases, the calcite should be sampled in situ. This provides fewer opportunities to control for open-system behaviour and increases the potential for contamination from the sampling equipment and other complexities, so it is best to consult with a specialist and arrange for them to take the samples.

The minimum required data for reporting a U-Th date are:

• sample code;
• laboratory code;
• U concentration;
• ²³⁴U/²³⁸U ± error;
• ³⁰Th/²³⁸U ± error;
• ²³⁰Th/²³²Th ± error;
• uncorrected U-Th age ± error;
• corrected U-Th age ± error.

There is no convention on reporting ages relative to a datum, though BP (before AD 1950) has been used, as has b2k (before AD 2000), but most commonly no datum is stated and the date is assumed as years before the publication date. Dates are in calendar years and do not require further calibration. In addition, provide the half-lives used (or published source) for the date calculations, along with details of the method of correcting for detrital contamination and the ratios used.

If isochron dating is used, include a graphical plot of the isochron and associated statistics (as produced by software such as Isoplot), either in the publication or as supplementary information.

4.3 Luminescence dating

Geoff Duller

Luminescence dating methods use naturally occurring minerals to calculate the time since a sample was last exposed to daylight or was last heated above about 250°C (Duller 2008). It has become a key geochronological method for studies of the Middle Palaeolithic, especially in Africa, Australasia and Europe (for example Jacobs et al. 2008; Roberts et al. 2015).

When minerals such as quartz and feldspar are exposed to radioactivity from the natural environment, a small proportion of the energy is stored in the crystal structure. At some later date, the energy can be released and produces light; this is the luminescence signal used for dating (Figure 13).

There are several luminescence dating techniques, based on different minerals and different signals. Quartz dating using optically stimulated luminescence (OSL) has been well established since 2000. Infrared-stimulated luminescence (IRSL) from feldspars has become established since 2008. This has led to the development of post-infrared IRSL methods (pIR-IRSL).

Other luminescence signals are also available, each with different strengths and weaknesses. For instance, infrared radiofluorescence (IR-RF) and infrared photoluminescence (IRPL) from feldspars have been developed, as has the use of the TL signal from biogenic calcite. These methods, however, are still in the early stages of development and application (e.g. Key et al. 2022; Duller and Roberts 2018). Advances in methodology are constantly being made and close collaboration with a laboratory is strongly recommended.

Luminescence methods can date the last time that the mineral grains in a sediment were exposed to daylight (optically bleached). This is normally when the sediments were deposited by a river, by the wind or by some other geomorphological process. When the mineral grains are exposed to daylight any energy stored in them is released, and this sets the ‘clock’ to zero. Once mineral grains are buried by further deposition energy starts to accumulate within them, and this continues until they are collected for measurement.

Sediments suitable for dating should contain either fine silt (4–11µm) or sand grains (90–300µm). Aeolian sediments are ideal (Text Box 5), but fluvial and some colluvial materials are also suitable, especially with the use of single-grain luminescence methods. The key consideration is whether there is a high probability that the mineral grains were exposed to daylight at or before deposition. Also consider the mixing of deposits through processes such as bioturbation or coversand reactivation.

The speed with which signals are bleached varies between different signals. Generally, the OSL signal from quartz bleaches the quickest and is therefore best suited to a situation where bleaching at deposition may have been limited. IRSL from feldspars bleaches more slowly; and pIR-IRSL and IS-RF signals bleach slower still.

Luminescence dating can also be used to date the last heating of stones and flints. Heating to more than about 250°C will release the energy stored in the mineral grains. Hearth stones, or flints that have been inadvertently burnt in hearths, have been targeted from Palaeolithic sites (for example Preece et al. 2007; Richter 2007).

Samples for luminescence dating can be collected by non-specialists, but it is preferable for a luminescence practitioner to do this. The luminescence signals used for dating are sensitive to light, and thus samples must be collected in such a way that daylight is excluded. Red light, such as that from the LEDs used for rear bicycle lights, does not affect the signal, and can be used where limited illumination is needed during sampling.

For sediments a common method of sampling is to hammer a metal or plastic tube (typically 30–70mm in diameter and 150–200mm in length) into the sedimentary unit. The ends of the tube should be packed with plastic and sealed using tape to avoid movement of the sample during transportation to the laboratory, and to avoid moisture loss. If this way of collection is not possible an alternative method is to use a large sheet of black plastic to exclude daylight from the section and to collect the sample in an opaque bag using a trowel.

Intact borehole and vibrocore sequences can also be sampled for luminescence dating in the laboratory, as long as they have been retrieved and stored appropriately and sediment shielded from light is available.

Measurement of a luminescence signal from sediment can now be made with portable OSL instruments, which may enable differentiation of sediments of radically different ages. The equipment currently available is, however, unable to replicate the procedures undertaken in the laboratory. The signals obtained are complex to interpret and commonly need to be used in tandem with laboratory measurements.

To calculate an age, luminescence measurements are made to calculate the total dose received by the sample during burial (known as the equivalent dose (D_e) or palaeodose). Separate measurements are needed of the natural radioactivity at the site. This enables determination of the amount of energy delivered to the sample per year (known as the dose rate). The age is calculated by dividing the equivalent dose by the dose rate.

Some dose rate measurements can be made in the laboratory, but in situ measurements using a gamma spectrometer are preferable, especially where sediments of differing radioactivity occur (Figure 14).

Where in situ gamma spectrometry is not possible it is important to consider whether the nature of the sediments varies within 300mm of the sample. Where large variations are seen, sub-samples of the different sediments should be collected for dose rate measurements in the laboratory, and their location relative to the luminescence sample noted. These dosimetry samples can be exposed to daylight since they will not be used for luminescence measurements. In addition, the thickness of the overburden should be noted, and an estimate of the water content during burial will be required.

For burnt objects, shield the artefact from as much light as possible; complete exclusion of light is unnecessary, as the inside of the artefact is normally used for measurement. Collect a representative sample of the sediment surrounding the artefact along with the artefact. The same issues about measurement of the gamma dose rate apply for burnt samples as they do for sediments

Luminescence can be used to date events from decades to more than 100 ka. The upper limit is determined by saturation of the luminescence signal — the point at which no additional energy can be stored in the mineral grains (Duller 2008). This varies from one sample to another, from one mineral to another, and is dependent on the dose rate.

It is common to be able to date to 100 ka, not unusual to be able to reach 200 ka, and ages of 400 or 600 ka are feasible. Feldspars are often able to date older samples than quartz, but this comes at the cost of greater methodological uncertainty as feldspar methods are more complex. Precision better than 5% (at 1σ) is normally unrealistic because of uncertainties in the dose rate. At ages of 100 ka and above, uncertainties of 10% are common.

Ages are normally given in kilo annum (ka) before the date of measurement. No agreed datum exists for luminescence ages, but it is good practice to report the date when a luminescence age was measured (Brauer et al. 2014). When reporting luminescence ages avoid using the term ‘BP’, which is used for radiocarbon ages.

Supporting information required for luminescence ages includes:

• the sample code;
• the laboratory code;
• the mineral and analytical method used for luminescence measurement;
• details of any statistical analysis of the luminescence data;
• the equivalent dose (D_e) for the sample.

Show on a radial or an Abanico plot if combining multiple D_e values to obtain the final age. It is also good practice to publish an example of the dose response curve (the growth of the luminescence signal with laboratory radiation) to illustrate whether the signal is close to saturation.

Include details of which methods were used to measure the dose rate, the water content used in calculation, the individual dose components (alpha, beta and gamma) and the cosmic dose rate.

4.4 Amino Acid Racemisation

Kirsty Penkman

Amino Acid Racemisation (AAR) dating relies on the time-dependent breakdown of proteins (and their constituent amino acids) in fossils such as shells.

It covers the date range from 10 years ago up to as long ago as 3 Ma, and thus is applicable to the whole of the Quaternary. However, it is most useful in the British context for dating Palaeolithic sites and Pleistocene deposits older than c. 40 ka.

A simplified overview of the technique is given below; further details can be found in Lowe and Walker (2015, 332–9).

Amino acids are the building blocks of proteins. They are found in all living tissues and can be preserved in fossil biominerals such as shells or coral. Most amino acids can exist in two forms, which are non-superimposable mirror images of each other (Figure 15), and are designated left-handed (laevo, L-form) and right-handed (dextro, D-form). In living organisms, proteins are almost exclusively made from the L-form.

After death, however, a spontaneous reaction starts called racemisation. This leads to a progressively increasing proportion of the D-form in direct relation to the time elapsed, until the D and L forms are present in equal quantities. Depending on the amino acid, this process can take thousands or millions of years and therefore is applicable over Quaternary timescales (Figure 16a).

Different species break down at different rates, so analyses are undertaken on monospecific samples (usually individual mollusc shells, 1–5mg in weight). The extent of amino acid racemisation (AAR) in a sample is recorded as a D/L value, and its age can thus be determined based on (a) which amino acid it is, (b) the species (of mollusc or other biomineral) being analysed, and (c) a baseline reference framework of comparative data from independently dated sites (an aminostratigraphy).

Protein degradation consists of a series of chemical reactions that are dependent not only on time, but also on environmental factors (such as pH, availability of water), which can confound the time signal. These difficulties in AAR’s early applications have led to a focus on analysing ‘closed-system’ protein from fossil samples (Towe 1980) — those where the fraction of protein analysed is physically or chemically shielded from the environment.

The chemically isolated ‘intra-crystalline’ fraction found in some biominerals forms such a closed system, meaning that the AAR within this fraction is solely time and temperature dependent, and therefore predictable (Penkman et al. 2008; Dickinson et al. 2019; 2024). This technique has been particularly successful in dating carbonate and phosphate fossils (shells, tooth enamel, eggshells, foraminifera, ostracods, earthworm granules) and long-lived biominerals (corals). It can be used to provide age information within an individual sample (Hendy et al. 2012).

AAR laboratories have developed dating frameworks for a large number of commonly-occurring ‘closed-system’ species, but tests can be undertaken on additional species to examine whether they would be suitable for AAR dating.

In a British Palaeolithic context, the most suitable materials for AAR dating are tooth enamel, Bithynia opercula (Figure 17) and Bithynia, Valvata, Littorina, Nucella, Patella and Pupilla shells. The crystal phase of calcite biominerals (such as opercula or eggshell) are more stable over longer timescales and are therefore preferred for sites of Early and Middle Pleistocene age.

The rate of breakdown towards D/L equilibrium in the intra-crystalline fraction is still affected by temperature, so comparative frameworks need to be applied from regions with a broadly similar temperature history. For instance, it is not appropriate to compare D/L results from tropical material to a framework based on sites from southern England, but any material from England can be interpreted within the same comparative framework.

In Britain analysis of amino acids in Bithynia opercula can be used to correlate deposits with the Marine Oxygen Isotope stages (Figure 16b), to a sub-MIS level for at least the Late Pleistocene (Penkman et al. 2011).

A non-specialist can collect material and/or sediment samples in the field. Sometimes molluscs, teeth or other suitable remains will be directly visible, but as it is not always possible to tell whether a sediment body contains suitable material for AAR dating, it may be necessary to collect a preliminary sample and then subsequently assess its potential for AAR dating.

Material for AAR dating is typically collected from wet-sieved residues of sediment samples. The only special sampling and pretreatment considerations are that the temperature-dependence of the racemisation reactions means it is important that any material submitted for dating has not been treated in any way that compromises its temperature history. For example, do not sieve with hot water or dry in an oven. Suitable material for AAR dating in the residues can be identified to species level (e.g. vertebrate or mollusc) by a faunal specialist or by the AAR laboratory.

Analyses are routinely undertaken on the total hydrolysable amino acid fraction (THAA, which includes both free and peptide-bound amino acids), and often also on the free amino acid fraction (FAA, produced by natural hydrolysis).

AAR laboratories tend to issue results in a report, with laboratory codes identifying samples, the relevant D/Ls and concentrations where appropriate. These data should be included in any publications, and it is also important to publish full sample information (including species), provenance information on the dated material and the provenance of material contributing to the reference framework.

4.5 Palaeomagnetism

Chuang Xuan

The Earth’s magnetic field intensity and direction are constantly changing at various temporal and spatial scales. Beyond historical observations of the last few hundred years, our knowledge of past field behaviour is mainly derived from natural remanent magnetisations (NRM) preserved in geological and archaeological archives.

These archives record palaeomagnetic field information mainly through two mechanisms:

• igneous rocks (e.g. lava, volcanic glass)
• and fired archaeological features acquire NRM through thermal remanent magnetisation (TRM).

This is when magnetic minerals cool from high temperatures to below the Curie point, locking in a magnetic signature. In contrast, sedimentary rocks formed in a marine or lake environment record palaeomagnetic field information through (post) depositional remanent magnetisation (DRM). This is when magnetic particles in the sediments align themselves to the ambient magnetic field during or shortly after sediment deposition.

Palaeomagnetism has been widely used for dating Pleistocene sedimentary sequences. The process typically involves the measurement of palaeomagnetic directions and/or intensity preserved within stratified samples, which are then compared to well-dated palaeomagnetic reference records (Hounslow et al. 2022).

Reversals in the Earth’s magnetic polarity (swapping of north and south poles), referred to as chron and sub-chron boundaries, provide a key method for correlating and dating sedimentary sequences worldwide (e.g. Opdyke et al. 1966). Changes in polarity within a sedimentary sequence are identified by measuring the NRM directions: declination and inclination (Figure 18).

The measured chron pattern of a sedimentary sequence is then compared with a geomagnetic polarity time scale (GPTS) which provides ages for when corresponding reversals occurred.

Interpretation of a magnetic polarity record often requires verification from other independent dating methods (e.g. biostratigraphy, Marine Oxygen Isotope curves; Happisburgh 3), especially when the top of a sequence does not have a modern age, or when a sequence contains a hiatus.

Any significant geological rotation or tilting caused by tectonic events should also be considered, but these are usually negligible for Pleistocene-aged sequences. The resolution and accuracy of palaeomagnetic dating based on geomagnetic polarities is determined by the number of reversals available and on uncertainties associated with the age of these reversals in the GPTS.

For the Pleistocene, the major reversal is between the Matuyama and Brunhes chrons (see Figure 19). Subchrons represent shorter-term reversals lasting tens or hundreds of thousands of years. GPTS ages for all Pleistocene reversals have been calibrated by astrochronology (see Ogg 2020) and should have uncertainties of less than 10–20 ka.

The more frequent geomagnetic excursions are defined as brief (<10 ka) events during which geomagnetic poles significantly deviate (up to 45°) from the background pole positions (Figure 19), though high-resolution studies indicate that at least some excursions are associated with 180° directional changes that lasted a few hundred to a few thousand years (Laj and Channell 2015).

Polarity reversals and geomagnetic excursions provide detailed insights into the Earth’s magnetic field behaviours and offer valuable opportunities for correlation and the establishment of isochrons for Pleistocene sedimentary sequences around the globe. NRM preserved in sediments records not only reversals and excursions, but also detailed changes in field strength and directions. These can be reconstructed through estimates of relative palaeointensity (RPI) and through palaeo-secular variations (PSV).

RPI records are usually constructed by normalising NRM of a sample by laboratory-introduced magnetisation to compensate for the ability of the sample to acquire magnetisation. Various criteria have been proposed to ensure the quality of the RPI records (e.g. Tauxe 1993).

RPI records constructed from different worldwide sedimentary sequences appear to record a dominantly dipolar geomagnetic signal and are generally coherent at a scale of a few tens of thousands of years. These RPI records can also be correlated to palaeointensity changes estimated using other methods, such as cosmogenic nuclides (e.g. Simon et al. 2016; Text Box 3) and marine magnetic anomaly profiles (Gee et al. 2000). Geomagnetic polarity reversals and excursions are usually associated with dominant lows in RPI records.

The use of RPI to constrain the chronology of a sedimentary sequence is usually referred to as palaeointensity-assisted chronology (PAC). Detailed RPI stack records that can be used as global or regional templates now span the entire Pleistocene (e.g. Valet et al. 2005; Yamazaki and Oda 2005; Channell et al. 2009) (see Figure 19, bottom).

In addition, PSV (i.e. declination and inclination) records have also been widely used to provide centennial-millennial scale age constraints, especially for late Pleistocene and Holocene sequences. These usually compare the PSV records to a regional reference curve or geomagnetic field model prediction for a location (e.g. Avery et al. 2017).

Samples used for palaeomagnetic dating are typically oriented according to their dip and strike directions (i.e. deviation and angle from horizontal plane). Samples are taken as discrete cubes or cylinders (Figure 20a), or as continuous U-channel sections (Figure 20b) marked with a reference orientation (true north, direction of top of sample).

For core samples where orientation is difficult to track during coring, a straight reference line should be marked on the core liner to guide subsequent cutting and splitting of the core and to facilitate declination corrections later on. Sediment samples are typically enclosed in plastic containers and stored in a fridge (set to ~4°C) away from strong magnetic sources to suppress any physical or chemical alternations.

Measurement of NRM and laboratory-introduced magnetisations of the samples are often conducted on a superconducting rock magnetometer capable of resolving weak magnetisations (i.e. 10-5 A/m level) (Figure 20c).

Samples are usually measured before and after stepwise heating or alternating field (AF) demagnetisation treatment to remove secondary magnetisations presumably carried by magnetic minerals with lower blocking/unblocking temperatures or lower coercivity.

Although geomagnetic polarity chrons, excursions, and RPI and PSV have become widely used for dating Pleistocene sequences, the detailed mechanism through which sediments acquire magnetisation is still poorly understood.

The sediment magnetisation ‘lock-in’ process may introduce a smoothing effect and centennial- to millennial-scale time offsets to sedimentary palaeomagnetic records (see Roberts et al. 2013). Such offsets might define the ultimate resolution of palaeomagnetism dating for Pleistocene sequences.

4.6 Tephrochronology

Rupert Housley and Ian Matthews

When volcanoes erupt, they disperse ash over thousands of kilometres in a matter of days or months. When identified in Pleistocene deposits they provide time-parallel marker horizons called isochrons. Tephrochronology is the use of these volcanic ash layers (tephras) to infer the age of associated sediments.

The detection of tephra layers is achieved by extracting the volcanic material from the host sediments — usually the glass fraction — and then classifying it chemically. This chemical dataset is then matched to a particular eruption (a correlative) by comparing its chemical signal with those from previously recorded eruptions in an international database (Figure 21 and Figure 22; for a full review see Lowe 2011).

Tephra research is primarily stratigraphic, but calendar dating for Pleistocene sequences can be acquired by two methods: direct dating of the tephra itself or dating of associated material.

Direct dating of volcanic deposits using Argon-Argon, Uranium-series or fission track methods, requires large amounts of material (Walker 2005). However, such quantities are usually not available in areas distant from the source volcano. England receives ash from Iceland and from Continental Europe, but the ash concentrations are too low and they lack the relevant mineral material to be directly dated.

Tephras are more commonly dated by determining the age of the layer in which they are found. The layer can be directly dated using datable material within it, or by obtaining a series of dates from a stratigraphic sequence of deposits — for example using age-depth Bayesian modelling of a series of radiocarbon dates (see Gransmoor). In some cases, the date of the layer can be estimated directly by counting annual laminations in lakes and in ice cores. Tephra isochrons allow this calendar dating to be transferred to deposits wherever the ash is detected.

Tephrochronology is a viable dating technique for the entire Pleistocene Epoch and is only limited by the reference datasets available for comparison. So far, there have been only a limited number of studies applying tephrochronology to English archaeological sites. However, distribution maps of ash fall suggest there is good potential to apply tephra studies to sites across the entire country (Figure 21).

Tephra studies in Europe have focused on the Late Pleistocene. There is a robust tephrochronology for northern Europe consisting of c. 20 tephras between 15 and 11.5 ka. A developing tephrochronology of c. 58 tephras has been established for the remaining Late Pleistocene (c. 120–15 ka) (Blockley et al. 2014; Davies et al. 2014). There is no reason why tephras should not be detected in Early or Middle Pleistocene deposits but these earlier periods have not received the same level of research. One example has been identified in the Middle Pleistocene West Runton Freshwater Bed in Norfolk (Brough et al. 2010).

The precision and accuracy of tephra ages are limited by the dating techniques and age models used for the type sites. During the Late Pleistocene, precision can be as good as 1–2% of the determined age (Bronk Ramsey et al. 2015a).

You should provide the following information when reporting tephra data:

• the tephra counts;
• the chemical data and the chemical standards;
• the analytical conditions used;
• the proposed correlative;
• how the age estimate is derived.

Ash layers — called cryptotephra — are usually of shards less than 125µm and therefore invisible to the naked eye. In England, it is likely that any tephras encountered will not be visible in the field owing to their small shard sizes.

Sampling for cryptotephra on archaeological sites requires the collection of a continuous sediment record covering the entire studied sequence (Figure 23). Tephra are often unevenly represented on a site so it is advisable to sample two or more sections.

In fine-grained sediments, take the samples using overlapping monolith tins. Where coarse clastic material predominates, collect samples in clean small bags from an exposed face, in contiguous 10–20mm intervals working from the section base upwards. In exceptionally clastic-rich sediments taking a full sample may not be possible; or a lower resolution (such as 50–100mm) must be accepted.

The relationship of the samples to geological layers and the archaeology must also be recorded.

Cryptotephra processing (Lane et al. 2014) in the laboratory involves (Figure 23):

• screening of samples c. 300mm³ in size from 50–100mm contiguous sediment blocks;
• if tephra is present, a series of contiguous 10mm samples are processed to pinpoint the highest concentration, often interpreted as the isochron;
• separating sufficient vitreous tephra shards for major (using EPMA (Electron Probe Microanalyser)) and trace element (using Laser Ablation Inductively Coupled Plasma Mass Spectrometry (LA-ICPMS) or SIMS (Secondary Ion Mass Spectrometry)) chemical analysis;
• compare these chemical results with databases of chemical signatures of known tephra horizons (Tephrabase: Newton et al. 2007; RESET: Bronk Ramsey et al. 2015b).

The current application of tephrochronology to archaeology can be classified into three categories: (1) wetland archaeology; (2) open-air ‘dry’ sites; and (3) rock shelters and caves. No formal assessment has been made of the likelihood of finding tephras in any of these kinds of deposits in Britain, but Housley et al. (2015) demonstrated that 22% of open-air ‘dry’ sites and 34% of rock shelters and caves produced tephras in Continental Europe.

To summarise: successful tephrochronological dating is dependent on the presence of an eruption coincident at the time the sediment accumulated. Longer records are more likely to include tephra and/or cryptotephra, so undisturbed deposits are preferable. Even if tephra shards are isolated on a site, they can only be assigned to an isochron if suitable dated geochemical reference data is available. Currently, these criteria are more likely to be met in the Late Pleistocene than in earlier periods.

4.7 The ‘Vole Clock’

Danielle Schreve

The recovery of small vertebrate remains is now a routine procedure when investigating sites of Quaternary age. Any calcareous fine-grained deposits may be suitable for sampling (sands, silts and clays, or seams of these within coarser gravel bodies).

Excavation with a trowel will often damage fragile specimens, so it is best to collect bulk samples of sediment, either as a column (to investigate any change up through a sequence) or as block samples around particular features of interest.

The samples should be wet-sieved individually through a half-millimetre mesh size before the residue is dried and then scanned under a low-power binocular microscope, extracting bones and teeth using foil tweezers. Clay-rich sediment samples should be air-dried or soaked with a dispersant such as 1% sodium hexametaphosphate before wet-sieving. This procedure will help to weaken the hydrophilic bond of the clay particles and enable easier processing.

Small vertebrate remains can offer highly detailed insights into many aspects of past environments and climates, food webs and evolutionary trends. In particular, the vertebrate fauna in Britain have been profoundly affected by Late Pleistocene long-term glacial to interglacial climate change and the succession of abrupt (decadal to centennial) climatic changes (Schreve 2001).

These cycles have influenced vertebrate species’ biogeographical ranges and have driven evolutionary trends and extinction events (Lister 1992). Taken together, these changes can be used to establish the relative ages of fossil assemblages — their biostratigraphy.

One notable example of a biostratigraphically-significant evolutionary trend is that seen in the water vole lineage, sometimes referred to as The ’Vole Clock’. Remains of fossil water voles are common in Quaternary deposits, thereby providing a large sample of teeth through which quantifiable changes can be observed. This is important because morphological change is often small over Quaternary timescales and tooth morphology between individuals can be highly variable. Large samples are required to capture variation within a population accurately.

The genus Mimomys appeared in Europe about 4 million years ago and evolved through several species. The genus survived until about 600 ka, when the final representative, Mimomys savini, was replaced by the modern genus Arvicola.

The key dental features of interest reside in the first lower molar (m1) — a ‘cloche-hat-shaped’ anterior loop (the anteroconid complex, ACC) and a series of three closed interlocking triangles and a posterior loop (Figure 24).

In the transition from Mimomys to Arvicola during the early Middle Pleistocene (late Cromerian Complex), an important change occurred in the switch from rooted teeth to permanently-growing molars (Figure 25). This apparently rapid change provides a significant biostratigraphic marker throughout western Eurasia.

In Britain, this dental transition shows a clear separation of older and younger sites. An ‘old’ group of early Middle Pleistocene sites are characterised by Mimomys (e.g. West Runton, Norfolk and Pakefield, Suffolk) and a ‘young’ group of early Middle Pleistocene sites have Arvicola (e.g. Westbury-sub-Mendip, Somerset and Boxgrove, West Sussex) (Preece and Parfitt 2012). This advantageous mutation provided Arvicola with extra tooth life, allowing them to extend their life span and breeding opportunities, thus perpetuating the mutation.

Within the genus Arvicola, further trends have been noted over the last half million years. There were two subspecies in Britain: Arvicola terrestris cantiana (also known as Arvicola mosbachensis) and the modern Arvicola terrestris terrestris (also known as Arvicola amphibius). The m1 lengthened and there was an increase in the ratio of the ACC to overall tooth length. The Mimomys fold, an archaic feature in the ACC, also became progressively uncommon in younger samples until finally disappearing (Figure 24).

The key trend, however, are differences in enamel thickness on the leading and trailing edges of the molars. Mimomys and early forms of Arvicola have thicker enamel on the trailing edges of the lower molars than on the leading edges. Over time, this trend reverses, so that in modern populations of Arvicola terrestris terrestris from Western Europe, the enamel is thicker on the leading edges of the lower molars (Figure 24) (Hinton 1926).

A method known as the Schmelzband-Differenzierungs-Quotient (SDQ or enamel differentiation ratio) was proposed by Heinrich (1982) to quantify this progressive trend. The method uses measurements of the combined trailing edge thickness from established points on the molar, divided by the combined leading-edge thickness, multiplied by 100.

The SDQ can then be compared with those of different Arvicola populations to establish the relative age of the sample. This technique has been widely applied in Britain to provide an independent chronology for many Quaternary sites (for example Schreve 2001; Roe et al. 2009).

Back to contents | Bayesian Chronological modelling | Case studies