Dating Techniques in Archaeological Science PDF

Dating Techniques in Archaeological Science 3159 been turned to a benefit, where interaction areas developed radiocarbon, replaced seriation as a have been defined by the ability to obtain a preferred dating method, at...

Dating Techniques in Archaeological Science 3159 been turned to a benefit, where interaction areas developed radiocarbon, replaced seriation as a have been defined by the ability to obtain a preferred dating method, at a cost of clarity. As seriation (Lipo et al. 1997). groups, phases change their definition with each Why does seriation work? It was recognized new assemblage added to them, resulting in some early on (Phillips et al. 1951) that the traits best situations where the similarity within phases is suited for defining historical classes were those less than the similarity between them. Neither with no functional values, sometimes called style. cross-dating, which is analogous to the use of If the environment exercises no influence over index fossils in geology, nor phase construction, style, which is therefore adaptively neutral, then moreover, solved the problems of spatial variation D the distribution of stylistic traits is controlled only and the time-transgressive nature of types, and by chance and transmission processes, which both therefore have produced chronological con- entail historical relations. Simulations of trans- fusion. The rigor of seriation, where one has to mission processes of traits involving only drift produce a seriation that works, is much needed. (chance) and innovation produce the same unimodal curves seen in seriation (Neiman 1995). Seriation has not been used as much as the Absolute Dating benefit of its analytical rigor might recommend. In the USA in the early 1900s, there was a per- Methods Based on Cyclical Variations ception that the American past had little time depth, particularly when compared with the scale Varves of change documented in Europe (The discovery The first absolute dating method was the use of of Folsom did not occur until 1927.). Small-scale varves, which are horizontal sedimentary layers of change, however, was being documented, partic- varying structure based on annual depositions. ularly in the Southwest, using seriation and stra- They were first recognized in Sweden (varve is tigraphy. But no one understood why seriation Swedish for cyclical layer) in the late nineteenth worked, and it was believed that stratigraphy century and were first used for chronology by the was necessary to validate seriation. Kroeber Swedish geologist Gerard de Geer, who used gla- famously wrote “the proof is in the spade.” Stra- cial lake varves in 1912 to determine the age of tigraphy got credit for demonstrating change in glacial retreat in Scandanavia. Clastic varves are the American past, also reinforcing a “perception created by the annual deposition of different grain that artifacts within vertically bounded recovery sizes. For glacial lakes, during spring runoff units comprised real, discrete entities rather than coarse particles are deposited, while in late sum- accidental (results of deposition) chunks of a cul- mer and fall, slower water carries only fine parti- tural continuum” (Lyman and O’Brien 1999). cles and allow clays to settle out. This results in Dendrochronology was also beginning to provide annual bands of lighter coarse grains and darker absolute dates in the Southwest, and archaeolo- fine grains. The thickness of the layers is depen- gists found that it could be used to infer dates of dent on temperature so that sequences of thick and pottery, even though it dates a different event. The thin layers can be cross-referenced from one lake painted designs on Southwest pottery proved to to another. In Scandanavia a cross-referenced have limited temporal spans, so cross-dating varve sequence extends with one-year resolution replaced seriation, despite having much less to 13,000 years. Biogenic varves are created by rigor. Seriation was widely used in the Eastern seasonal pulses of aquatic biological layers. The woodlands in the 1920s and 1930s, but difficulties best known are varves in the Cariaco Basin in in spatial variation led to comparisons among Venezuela where lighter layers, representing areas using stratigraphic-bound parts of the deposition of diatoms during winter and spring seriation, which were called “phases.” Phase con- upwelling, alternate with darker layers, struction, a grouping method based on similarity representing sediment run-off during rainy sum- of artifacts and anchored by dates from newly mer and fall. Varves have limited applicability in 3160 Dating Techniques in Archaeological Science archaeology aside from environmental recon- thickness, density, and cell structure as a function structions and radiocarbon calibration. They can of several environmental variables, mainly tem- date pollen sequences which are then compared perate in cold climates and precipitation in dry with pollen from archaeological sites. climates. Other variables affecting rings include frost, fire, insects, and floods. Patterns of thick- Dendrochronology ness or density variation in rings can be correlated More applicable to archaeology is dendrochronol- from tree to tree. If cross-references can link mod- ogy, also developed in the early twentieth century, ern trees with ancient ones, a master curve can be by Andrew E. Douglass, an astronomer working built, allowing absolute dates at one-year resolu- in the Southwest. He was interested in the cycle of tion (Fig. 1). If one cannot match ancient curves sunspots as might be recorded in tree rings but with modern trees, then only a floating chronol- realized the variable ring widths could be used for ogy is obtained, but which can still be useful for dating. He produced his first archaeological chro- relative dating. The history of dendrochronology nology in 1921. Europeans did not get started is replete with search for missing links to connect until the 1930s, and it was not until last 25 years floating curves with the master. Gaps in the curves that long chronologies have been developed. often reflect periods of tree depletion, so are them- The principle behind dendrochronology is sim- selves interesting. Best-known sequences are ilar to that for varves (Baillie 1995). Cross- those built on conifers from the American South- sections of trees display successive layers of west and on oaks in Northern Europe (both extend annual xylem growth, which in many trees form back about 10–12 ka), but the method has been rings. The inner band of light colored cells is extended to nearly every region in the world. called early wood, and the outer band of dark- Two requirements limit the applicability of colored cells is called late wood. Trees vary in dendrochronology in archaeology: prehistoric their usefulness for dating. Some have rings that use of wood with suitable rings for dating and do not vary much but others vary dramatically in preservation of that wood. These are large outermost 1765 ring 1875 ring width 1 mm 1 mm 1 mm 1822 annual 1851 1 mm 1 mm overlapping overlapping overlapping the year in which the tree was felled is known 1790 1800 1810 1820 1830 1840 1850 1860 1870 1950 1960 1970 1980 Dating Techniques in Archaeological Science, Fig. 1 Schematic showing how cross referencing works in dendro- chronology. This is a reproduction of a figure from Schweingruber (1989: 51) Dating Techniques in Archaeological Science 3161 requirements and so the exceptional precision of such as settlement changes in the Southwest – dendrochronology can seldom be exploited. Even have been correlated with climatic variables where dating is possible, there are a number of gleaned from tree-ring data (Dean et al. 1985). potential problems. Missing rings or double rings, The one-year resolution has also made it possible both caused by stress, affect counting and cross- to correlate cultural change with singular events referencing. More serious are mismatches such as a large volcanic eruption creating a dust between the dating event and the target event. veil. An eruption in AD 536 has been correlated Wood may be cut long before it is used. It can be with Sierra Nevada cold years, European dry fogs, incorporated into earlier features for repair rea- Scandanavian cold summers, the Irish failure of D sons (although interesting in itself), and it can be break, famines in China, stars not seen in China, reused. For these reasons date clustering is often high mortality rates in Ireland and England, and used to estimate building construction, but even the beginnings of the Justinian plague (Baillie then stockpiling, and multiple reuse, and repair 1995). Cultural events have also been correlated can produce a broad array of dates difficult to with climatic variables dated by marine oxygen interpret. Another problem is non-cutting dates, isotope records and ice-core layering produced by which are age estimates from wood missing its annual dust cycles. outer rings. Sapwood is the outer living part of the Various marine fauna also exhibit annual trunk, while heartwood is the hard, dense, physi- growth patterns that are dependent on environ- ologically dead inner part. Sapwood is more prone ment conditions. These have been widely used to decay and is often removed initially during for reconstructing past environments and season- construction for that reason. It is also more likely ality of collection. Very recently attempts have to be burned. Non-cutting dates are attempts to been made to use them for dating by developing salvage dates by estimating the number of outer a cross-referenced master curve that can be rings lost, possible by regression using living trees assigned dates by 14C wiggle matching. Work if all the sapwood is not removed. If some of the has been done on bi-valve shells, fish otoliths, heartwood has been removed, only a maximum and corals, but the dating range has been limited date is possible. Non-cutting dates are not uncom- to just 100 s of years. mon. Of 194 available dendro dates from the Chama River region of New Mexico, only 55 are Methods Based on Radioactive Decay cutting dates. A final problem with dendrochro- Radioactivity was discovered in the late 1890s nology is the attempt to use it to date other things. by Henri Becquerel, Marie Curie, and others. In For example, one might try to date the manufac- 1902 Ernest Rutherford and Frederick Soddy ture of ceramics found in a room by the age of established that radioactivity resulted from trans- beams in the walls of the room. This obviously formation of unstable elements into new forms. In requires a bridging argument, and not always an 1906 Rutherford proposed that the He/U ratio easy one to make because the firing of the ceramic could be used to determine the age of the earth. and the construction of the room can be separated The next year Bertram Boltwood proposed using by significant time. Where these problems can be Pb/U ratio toward the same end. The Pb/U method controlled, dendrochronology provides spectacu- was improved by Arthur Holmes, who established lar resolution. Dating of timbers from lake settle- a geologic time scale in 1911, which is in general ments in Central Europe, for example, showed agreement with modern values. Isotopes were dis- that settlements were not continuous but episodic, covered in the 1920s, and improvement of instru- something that would be completely blurred by mentation to measure them resulted in the mass less precise techniques such as radiocarbon. spectrometer in the 1930s. In the 1940s, U-series Another use of dendrochronology, only briefly dating, K-Ar dating, and 14C dating were all mentioned here, is for environmental reconstruc- developed. While there are many other such dat- tion by correlating ring widths and densities with ing methods, most are relevant only in geology, known climatic indices. Many prehistoric events – whereas these three have seen common use in 3162 Dating Techniques in Archaeological Science archaeology. The methods are all similar in con- with time as a function of the 234U half-life. cept. The ratio of a radioactive element to a stable There are other ratios within the series, and also one is related to age by the half-life of the the 235U series that can be used for dating, but this decaying element. is the most common in archaeological applica- tions. The event being dated is a depositional Uranium Series Dating event. It can be related to archaeological events Two of the most common natural radioactive only by the relationship of the archaeological elements are 238U and 232Th. Both have half- material to the deposit. For example, archaeolog- lives (how long it takes for the concentration to ical materials sealed by a travertine layer must be be reduced by half) of the same magnitude older than the age of the travertine. A unique (109 years) as the age of the universe. They application in this regard is the dating of rock art decay so slowly that their concentrations are by calcite layers that have grown over it. nearly constant over the time span of human evo- There are four requirements for U-series dating lution. Neither decay directly to a stable element to work. First, the initial fractionation must occur but rather decay through a series of daughter over a short period of time relative to the half-life radioactive elements before reaching a stable iso- of the daughter. Second, the initial abundance of tope of lead. The half-lives of the daughters are the daughter must be negligible or known. If the very short compared to the half-lives of the par- sediment is not initially free of 230Th, this can be ents, so that after some period of time, if there are estimated by measuring 232Th. Third, all U must no perturbations to the system, equilibrium is be derived from the ambient water in the deposi- reached where the production rate of the daughter tional habitat. Adjustment for 234U/238U fraction- equals its decay rate. The parent and each daugh- ation is needed. 234U is more susceptible to ter will then have the same activity, a condition leaching because the energy from alpha recoil known as secular equilibrium. Because of the injects 234Th (the short-lived parent of 234U) into different half-lives, the parents will have a much pore space or damages the 234U site. The ratio of 234 higher concentration than any of the daughters. U to 238U in sea water, for example, is about There is also a third decay chain headed by 235U, 1.147. Both the initial 234U/238U and the 230 but this one is much less common. Th/232Th ratios can be estimated by isochron If some geological process disrupts the decay dating. Fourth, after fractionation the system must chain, disequilibrium results. This is the basis remain closed. for uranium-series dating. 238U decays into U-series dating does not work as well with 234 Th and then into 234Pa, both short-lived bones, teeth, and shell, because of U uptake after daughters that remain in equilibrium with the the initial fractionation, violating the closed sys- parent, then into 234U with a relatively long tem requirement, unless the U uptake can be half-life (2  105 years), then 230Th (half- modelled. One recent attempt at this is the absorp- live ¼ 7  104 years), and then through a series tion/diffusion model. If one knows the U diffusion of short-lived daughters until the stable 206Pb is co-efficient in the bone, the bone porosity, and the reached. Uranium has two oxidation states, a tet- environmental concentration of uranyl ions, one ravalent and a hexavalent one. In oxidizing con- can predict distribution profiles of U and U-series ditions, it forms a hexavalent uranyl ion complex, isotopes transversely across the bone. The profiles which is soluble. Thorium occurs only in the are u-shaped and gradually flatten out with time. tetravalent state and is insoluble. Water will con- Traditionally, U series measurements were tain mainly U because Th will settle out. When made by alpha spectrometry (count decays), but water-lain materials such as stalactites, stalag- better precision and a wider age range is possible mites, travertines, and corals are deposited, they with mass spectrometry (count atoms), particu- will contain only U without the daughters. After larly thermal ionization mass spectrometry deposition, the daughters begin to rebuild. The (TIMS) and multicollector inductively coupled ratio of 230Th to its parent 234U will increase mass spectrometry (MC-ICPMS). The age range Dating Techniques in Archaeological Science 3163 is about 500,000 years (controlled by the half-life K-Ar dating is restricted to volcanic materials of 230Th) to a few hundred years (depending so can only date archaeological materials that are on the ability of mass spectrometer to resolve bracketed by volcanic depositions. It has seen disequilibrium). most use in early human evolution. Its upper age limit is the age of the earth. Young ages are diffi- Potassium-Argon Dating cult because of low signal to noise ratios, but the Potassium-argon dating is based on the decay of eruption of Mt. Vesuvius in AD 79 has been dated. 40 K, which has a half-life of 125  109 years. It has two decay paths. Almost 90% of the time it Radiocarbon Dating D decays to 40C with the emission of a beta particle. This is the most common chronometric method About 10% of the time it decays to 40Ar via used in archaeology. Carbon has three naturally electron capture with the emission of a neutrino occurring isotopes: 12C (99%), 13C (1%), and 14C and a gamma ray. The ratio of 40K/40Ar is related (one part in million million). Only 14C is unstable. to time since a degassing event when all the argon It is continually formed in the upper atmosphere from a sample is removed. This occurs during a by collisions between cosmic-ray produced neu- volcanic event when heat brings all Ar in the trons and nitrogen. It rapidly combines with oxy- sample into equilibrium with the atmosphere. gen to form carbon monoxide and then carbon The method assumes there is no Ar escape after dioxide. Carbon dioxide mixes in the atmosphere, cooling, but this is rare because the Ar ion is large dissolves in the oceans (the hydrosphere), and enough that it becomes trapped in the solid crystal enters the biosphere through photosynthesis. lattice. Initially the 40K/40Ar ratio was measured There is a somewhat delayed and indirect entry on two aliquots, one for K, measured on a flame into a fourth reservoir, the sedimentary sphere photometer assuming natural atomic abundance (Fig. 2). 14C decays at a statistically constant rate (40K is 0.0117% of total K), and one for Ar, (half-life of 5730
30 years) into 14 N, emitting a measured on a mass spectrometer. Nowadays, beta particle. The production and decay are in the sample is placed in an atomic reactor and equilibrium so that there is a more or less constant bombarded with neutrons to transform 39K (the amount in each of the carbon reservoirs. The event most common isotope of K) to 39Ar. From atomic dated is when carbon material is isolated from the abundance, 39Ar serves as a proxy for 40K. All carbon cycle ending any replenishment with new 14 measurements can then be made on a mass spec- C. This is ultimately an end of photosynthesis, trometer using only one aliquot. Because only the corresponding to the death of an organism. ratio needs to be determined, this avoids trying to Assumptions of the method (Bowman 1990) compute absolute abundances, increasing accu- include: (1) the concentration of 14C in each car- racy, and precision. Because a correction must bon reservoir has remained constant through time; be made for atmospheric Ar, five different iso- (2) there is complete and rapid mixing of 14C topes of Ar are typically measured, and isochron throughout the reservoirs; (3) there is the same 14 dating used. The method now is commonly called C concentration in all parts of the biosphere; argon-argon dating. (4) 14C concentration has not changed in the sam- Measurements are commonly done using step ple since isolation by any other mechanism than heating, which allows diffusion of Ar out of the radioactive decay; (5) natural levels of 14C can be sample successively. Because diffusion occurs at measured with sufficient accuracy and precision; a lower temperature on the surface than the core, and (6) the event dated is close in time to the event step heating can show any evidence of Ar loss or of interest. The first three of these assumptions are gain since the time of interest, using a plateau test not strictly true and the last three are sample for the best estimate of age. Laser ablation is used specific. to date single grains, allowing assessment of het- Constant concentration of 14C in the atmo- erogeneous materials such as lavas which can be sphere was proven incorrect by comparison of contaminated with older grains. tree rings dates with 14C assays on the same 3164 Dating Techniques in Archaeological Science Cosmic radiation Proton Production 14 N nucleus 14 C Oxidation Spallation products 14 Distribution CO2 Photosynthesis Dissolved CO2 14 Decay: C → 14N + ß-1 + ve Dating Techniques in Archaeological Science, by dissolution into the ocean. The decay equation is Fig. 2 Illustration of carbon cycle. Production occurs in entered at the bottom, where ß1 is a beta particle and v¯e the upper atmosphere. Distribution occurs through photo- is an antineutrino. The drawing is based on similar ones synthesis, and then by transfer through trophic levels, and found on radiocarbon lab web sites rings. There is a long-term systematic underesti- and interglacial periods can affect the level of mation of 14C dates, which correlates with shifts 14 C. For all these reasons, radiocarbon dating in the earth’s magnetic field strength. Cosmic rays cannot produce calendar dates. There must be a are charged particles deflected by the magnetic calibration. For terrestrial samples, this has been field. The stronger the field, the more deflection. done by carbon dating tree rings of known age. Short-term trends, which limit precision, are oscil- Rings from bristlecone pine in North America and lations superposed over the long-term trend. oaks and pines in Europe have allowed calibration Called “wiggles,” or the de Vries effect, they are to 13.9 ka. For marine samples, carbon dating thought to be caused by sunspot activity which of foraminifera from varved sediments in the influences the magnetic field between the earth Cariaco Basin and carbon dating of U-series and sun. Over the past couple of centuries, there dated corals have been used. The most common has also been a decrease in concentration of 14C in method for calibration is a probability method the atmosphere, called the Seuss effect, due to using Bayesian statistics. Bayesian statistics, burning of fossil fuels releasing old carbon. In using stratigraphy as aprior, are also used to con- the 1950s and early 1960s, there was an increase strain dates from multiple intercepts when the in 14C due to atomic bomb testing. This has been calibration curve is relatively flat. used to study mixing of 14C in the biosphere, for Reservoir effects refer to problems in example, to show that adjacent tree rings do not extending calibration to other reservoirs besides exchange carbon. It has also been used to produce the atmosphere. Mixing rates of dissolved carbon very accurate dates in the last half of the twentieth in the oceans is slow so that carbon is “older” by century. Finally, the solubility of carbon dioxide the time it reaches the bottom. Upwelling can into water is affected by temperature, so glacial bring this carbon to the surface, so that surface Dating Techniques in Archaeological Science 3165 waters can have a modern age, on a global aver- introduced. Currently, the sensitivity is not as age, of about 400 years (Rg). Additional offsets, high as AMS, but may eventually allow sample called DR, are controlled by local factors and can sizes as small as 10 mg. vary by up to 1000 years and can vary temporally, Charcoal has been a mainstay of radiocarbon making dating of marine organisms difficult. dating. Because long-lived samples have the old Another reservoir effect is a 30-year offset of the wood problem, preference is for short-lived southern hemisphere from the Northern Hemi- species, although with increasing risk of contam- sphere because of direction of trade winds and ination from rootlets and humic acids. Other the greater surface area occupied by 14C–depleted short-lived flora such as grass, seeds, leaves, and D ocean in the Southern Hemisphere. Rivers and even pollen are good dating candidates as long as lakes also exchange carbon dioxide with the atmo- fractionation is controlled. For shells, the inor- sphere requiring some correction. Dissolution of ganic fraction is usually dated but can exchange calcium carbonate and soil humic material into carbonate with the environment. Recrystallization freshwater also introduce old carbon. from aragonite to calcite is often a sign of possible Concentration of 14C in living organisms dif- contamination. Marine and fresh-water shells fers from the atmosphere due to differential require reservoir correction. Terrestrial shells and uptake of carbon dioxide during photosynthesis. insects are less desirable because they interact Lighter isotopes (12C and 13C) are taken up pref- with the sedimentary reservoir which often is not erentially, so plants will have a 14C deficiency. in equilibrium with the atmosphere. Because of This is routinely corrected by measuring the environmental exchange in the inorganic fraction, ratio of 13C/12C and assuming the effect of 14C most effort in dating bone uses organic collagen. is double that of 13C. Differences among plants Collagen does not readily exchange carbon but in the carbon uptake, used to advantage in degrades easily and for older bones there may reconstructing diet, create complications in dating not be sufficient amounts. It has a characteristic animals that draw food from plants with different amino acid profile, so if this is intact, it is not fractionations. Long-lived trees can be depleted in likely the collagen has been altered. Work has 14 C because the inner heartwood has ceased to even been done on dating individual amino exchange carbon with the atmosphere, creating acids. There are a variety of ways to extract col- the old wood effect. lagen or individual amino acids, all with the aim to In preparing samples, care must be taken to reduce contamination. The most common is ultra- remove recently incorporated, but loosely bound, filtration which takes advantage of the large mol- contaminates. The acid-alkaline-acid pretreatment ecules of collagen, allowing separation from is routinely used. A more rigorous pretreatment, smaller molecules with a filtering membrane. called ABox, involving additional chemical oxi- The inorganic fraction of bone may be dated if it dation stages, is used for charcoal near the age is heated high enough to remove carbonates and limit of the method, where contamination has the upon cooling recrystallizes, becoming resistant to largest effect. Measurement of 14C is challenging. environmental exchange. This is called calcined Low concentration increases the difficulty as the bone. One problem is that some of the carbon in age limit of about 40–50 ka is neared. Beta parti- calcined bone may come from the fuel, which is cles are also weak. The conventional method was subject to the old wood effect. Organic matter is to use gas counters or liquid scintillation. In both incorporated into soils from a variety of sources, cases, decays are counted. The advent of acceler- most during pedogenesis. This can be dated but ated mass spectroscopy allowed the counting of the date reflects the mean residence time, a vari- atoms, which permits higher sensitivity (by 1000 able dependent on rates of deposition, pedogene- to 10,000 times) and smaller sample sizes. AMS sis, and depth of burial. Longer the mean may not be the last word. Recently, laser-assisted residence time, less meaningful is the date. Dates isotope ratio analysis (LARA), based on isotope provide a minimum date for events associated specificity of laser resonance in CO2, has been with the initiation of soil development or 3166 Dating Techniques in Archaeological Science maximum age for burial by overlying sediments. Methods Based on Radiation Exposure Different fractions of the soil, e.g., the insoluble residue or the base-soluble humate, can some- Luminescence Dating times give different dates. Organics in ceramics The second most used absolute dating method and can be dated. These have disparate sources one that can complement radiocarbon well is (original organics in clay, temper, soot from luminescence dating. Understanding lumines- fires, and food residues) creating severe associa- cence as a phenomenon was achieved in the first tion problems. Best targets are food residues, par- half of the twentieth century, and suggestions of ticularly lipids. its use for dating were made as early as 1953. Radiocarbon will probably remain the most A dating method was worked out on pottery valuable absolute dating method in archaeology, by an Oxford group in the 1960s and 1970s. but it is not a panacea. As Christopher Bronk Extension to dating sediments came in the Ramsey (2008) of the Oxford radiocarbon lab 1980s. Major strides in method development has noted, “If we are to interpret radiocarbon occurred in the last 20 years. measurements on samples in terms of chronology, Luminescence dating has several advantages. there is much more that needs to be considered First, the target event and the dating event are and thought about. There is an expectation that as often the same. The dating event for pottery and the method has developed and become more pre- burned lithics or rocks is the last heating event, cise, all of this has become routine and no longer usually when made. The dating event for sedi- needs to be considered. This is far from the truth.” ments is the last exposure to sunlight, often This is an exhortation to the archaeological com- when the sediment was deposited. Second, it munity not to treat radiocarbon as a black box. All works on materials that are ubiquitous: quartz dates must be evaluated, particularly in how it and feldspar, the two most common minerals in relates to the target event. Ranked criteria for the world. Third, it has a wide dating range, from judging radiocarbon dates, assuming all technical a few years before the present to more than matters have been accounted for, have been pro- 500,000 years. The drawback is that it is a very posed. From best to worse, these are (1) congru- complicated method, perhaps the hardest for ence of target and dating events, (2) direct archaeologists to understand, and dates need functional relationship between target and dating proper evaluation. Because it is complicated, event (e.g., hearths in floor of houses), (3) quantity error terms are propagated over a number of mea- of organics such that relationship seems reason- surements, reducing precision. able (e.g., charcoal concentrations), (4) just spatial Luminescence is defined as the emission of association, and (5) just assumed association. light from a material following the initial absorp- Obviously there is a large difference from dating tion of energy from an external source (Chen and use of corn by dating a cob to dating an occupa- McKeever 1997). To distinguish it from similar tion by a random piece of charcoal found in the phenomenon such as fluorescence, there is a time vicinity. Bridging arguments need to be stated delay between absorption and emission. This time clearly and their validity needs to be evaluated, delay, from a few seconds to millions of years, on a case by case basis. Some have applied strict implies the storage of energy in some metastable hygienic rules to radiocarbon dates and have state before return to ground state. The source thrown out 80% of the available ones. Others of this energy, for dating, is naturally occurring argue this may be going too far and valuable radioactivity. Release of the energy from the meta- information is being thrown out with the bad. stable state requires an additional stimulus of Perhaps the best solution is not to rely solely on energy, usually heat or light. If the stimulus is radiocarbon dates but to consider other dating heat, it is called thermoluminescence (TL); if the methods as well. This of course is the basis for stimulus is light, it is called optically stimulated the Bayesian analyses that are increasing in luminescence (OSL) or infrared-stimulated lumi- popularity. nescence (IRSL), depending on the wavelength. Dating Techniques in Archaeological Science 3167 Luminescence is usually observed in ordered therefore proportional to amount of trapped crystals and the metastable states are associated charge and to the amount of radiation the sample with defects in the crystal lattice that produce a has received. The “palaeodose” is the amount of charge deficiency. Radiation ionizes atoms creat- dose the sample has absorbed through time. The ing an electron (negative charge) and hole “equivalent dose,” often abbreviated De, is the (positive charge) pair. These are attracted to laboratory measurement of the palaeodose, deter- defects of opposite charge, where they are held mined by calibrating the measured luminescence until sufficient stimulus releases them. Those signal against laboratory irradiation. The unit of defects where the stimulus energy required to absorbed dose is the gray (Gy), or one joule per D liberate the charge carrier is relatively small are kilogram. The dose rate is the amount of radiation called traps. Those where the stimulus energy is absorbed per unit time (Gy/unit time). The age is relatively large are called recombination centers. determined by dividing the De by the dose rate. When released, the trapped charge recombines The age is in terms of years before present, where with opposite charge in the centers. The process the present is the date of the laboratory radiation. of recombination results in release of energy, Unlike radiocarbon, no further calibration is some of it luminescence (Fig. 3). The intensity needed. of the luminescence signal is proportional to the Terrestrial natural radiation consists of alpha, rate at which the system moves from the metasta- beta, and gamma particles. They are distinguished ble state to the ground, or equilibrium state. It is by the penetrating power of the emissions, short for alphas (~20 mm in sediment), medium for betas (~2–3 mm), and long for gammas Conduction Band (~30 cm). There is also a minor contribution α ß γ from cosmic radiation, which is deeply penetrat- Traps ing. The main producers of the terrestrial radiation Center are the 238U and 232Th decay chains and 40K, all discussed earlier. There are minor contributions Velence Band from the 235U decay chain and 87Rb. Because of the long half-lives, the concentrations of these are Trapping by ionizing radiation effectively constant for the time range of lumines- cence dating. This is convenient because if allows Conduction Band estimation of the dose rate from current radioac- Trap tivity, unless there has been some geological hν disruption. The latter includes processes like Heat or light Center leaching and changing burial depths, but fortu- nately significant changes in the dose rate, Velence Band although always a consideration, are not common. Most materials found in samples being dated have Release as luminescence similar coefficients of absorption, but not so in Dating Techniques in Archaeological Science, water. It will absorb radiation at a different rate Fig. 3 Illustration of the luminescence process using than the sample, so estimation of average water energy band theory where the valence band represents the content through time is a requirement. This some- ground state and the conduction band represent the lowest excited state. Traps and centers have energy levels in the times hinders dating, but the usual practice is to gap between the two bands. Ionization lifts the electron to make an estimate based on measured contents and the conduction band and it and the hole are attracted to any climatic considerations, and then apply wide traps and centers. Heat or light liberates the electron from error bars to cover most possibilities. The range the trap and it recombines at the center to yield lumines- cence. The process is simplified and does not include all of radiation influences the size fraction of the possibilities for recombination routes. The drawing is sample that is dated. Quartz and feldspar have adapted from a similar drawing by Dorothy Godfrey-Smith negligible internal sources of alpha radiation. 3168 Dating Techniques in Archaeological Science Using sand-size particles (etched in the case vastly different), (b) the luminescence response of quartz) limits the alpha contribution, so that is the same during measurement of natural and the dose rate is mainly betas, gammas, and laboratory induced signals, and (c) the traps are cosmic. This simplifies procedures, and so is the stable over the relevant time period (Murray and most common form of dating. Where the sample Wintle 2000). The latter is usually assured by lacks coarse grains, fine-grain procedures are preheat to remove unstable signal. Two basic followed where particles are limited to 1–8 methods for measuring De are additive dose and microns, small enough not to attenuate the alphas. regeneration. In additive dose, increments of dose Then the dose rate must include the alphas as well. are added to aliquots that still retain their natural This is sometimes advantageous where the signal. A growth curve pairing dose versus lumi- gamma rate is poorly known, because including nescence intensity is extrapolated to the dose axis alphas reduces the proportion of gammas to the to estimate De. The method assumes that the fitted dose rate. Because of their short range, alphas do function for the known points is appropriate for not use all their energy before the sample becomes the extrapolated region, but this is often not the locally saturated (all traps are filled). Thus they case. This and large errors in extrapolation, espe- are less efficient at producing luminescence. An cially for old samples, limit the usefulness of efficiency factor, sometimes called a-value or additive dose. In regeneration, increments of b-value, must be determined to correct the alpha dose are added to aliquots that have had their dose rate in fine-grain dating. Because of the long natural luminescence removed (by heat or light). range of gamma radiation, much of the gamma The natural signal is then interpolated into the contribution to the sample arises outside the growth curve, avoiding the problems of additive sample. In the case of discrete pieces such as dose. The problem with regeneration is sensitivity ceramics or lithics, a sample of associated soil is change caused by emptying aliquots of natural often collected to measure the gamma dose rate. dose. One solution to the problems of additive Alternatively, gamma radiation can be measure in dose and regeneration is to combine them so that situ by placing dosimeters or using a portable one corrects the problems of the other. Multi- gamma spectrometer. Dose rate is measured in a aliquot techniques such as supralinearity correc- number of ways: alpha counting, beta counting, tion and the slide method are two examples. gamma spectrometry, and flame photometry (for Improved solutions have followed efforts to deter- K). Methods such as ICP and INAA are also used, mine De on single aliquots, in particular the single but these usually only measure the parent of the aliquot regeneration (SAR) method developed by decay chains and can produce misleading results Murray and Wintle (2000). This involves measur- if there is disequilibrium. ing the luminescence after successive regenera- Determination of De involves measuring the tion doses (including the natural dose). The natural intensity and the sample’s sensitivity in problem of sensitivity change is solved by terms of luminescence induced per unit radiation inserting a test dose of equal magnitude after dose. Sensitivity depends not only on the number each step. Dividing the signal from the regenera- of trapped charge carriers but the proportion of tion dose by the signal from its corresponding test them that are emptied during read-out and the dose provides a normalization of each regenera- proportion of recombination centers that emit tion signal. The regeneration/test dose signal light. Some centers release the energy as phonons. ratios are then plotted in a growth curve, and the Changes in sensitivity occur as a function of dose, natural/test dose signal is interpolated into it, pro- heat, light exposure, and time. Calibrating the viding a De. The main assumption of SAR is that natural OSL signal against that produced by the sensitivity of the test dose signal is propor- known laboratory dose will provide an accurate tional to the sensitivity of the regeneration dose De only if (a) any competition for charge during signal for any given dose. The method, and vari- trap filling is the same for laboratory and natural ations of it, has proven remarkably robust and is irradiation (especially considering the rates are the most common method used. Dating Techniques in Archaeological Science 3169 Deep traps in some minerals seem to lose elec- played a major role in deciphering the chronology trons at ambient temperatures even though kinetic of the Middle to Upper Palaeolithic transition, considerations suggest very low probability of often beyond the reach of radiocarbon. Dating the energy barriers being breached. This is called fire-cracked rock is an under-utilized application. anomalous fading. It appears commonly in feld- Insufficient heat of the rock at the time of interest spars but is rarely observed in quartz. A widely is a problem (rocks have a steep thermal gradient, accepted explanation is quantum tunneling. In any and the inside of the rock is usually sampled for sample that exhibits such fading, steps need to be dating). A good approach is single-grain dating on taken to correct for it or to circumvent it. A cor- coarse grains to separate out those grains that were D rection was proposed by Huntley and Lamothe heated sufficiently from those that were not. (2001), which uses fading rates observed at labo- Recent work has also aimed to date the surfaces ratory time scales and extrapolates them to a geo- of rock to when they were last exposed to sun- logic time scale. This seems to work where the light, useful for rocks used in structures. growth curve is in the linear region, up to about The largest application of luminescence dating 50 ka for feldspars. More recently (Buylaert et al. is in dating sediments to their last exposure to 2009), high-temperature stimulation using IRSL light. Luminescence dating of sediments has two following a low-temperature stimulation has been difficulties not often encountered in ceramics. shown to tap a signal that does not fade or fades One is the possibility that the sediment was not only weakly (called post-IR IRSL, or pIRIR). The exposed or insufficiently exposed to sunlight at low stimulation preferentially empties traps that the time of deposition. Insufficient exposure is are close to recombination centers and thereby called partial bleaching. The other is the possibil- more susceptible to quantum tunneling, and the ity that grains of different ages have been mixed succeeding high-temperature stimulation empties post-depositionally. One way to deal with the traps that are far from recombination centers and possibility of partial bleaching is to try to isolate much less likely to fade. These traps are also more a signal that is rapidly bleached. Quartz bleaches difficult to empty by sunlight, so that a residual more readily than feldspar, and within quartz the signal remains, making the method somewhat fast-bleaching component is preferred. Single- problematic for younger samples. But overall, grain dating, or dating of aliquots that have a the method seems to work over a wide age range. very small number of grains on them, is utilized In dating ceramics or lithics, the dating event is to separate bleached grains from partially the last exposure to at least 450  C heat. Materials bleached or unbleached grains. Single-grain dat- that can be dated include not only pottery but ing, using the smallest fundamental unit of analy- structural materials (bricks, daub), fire-cracked sis, is the only method for parsing mixed-aged rocks, and heated cherts or other micro-crystalline samples. Single-grain dating also has the advan- silicates. Glass, porcelain, and slag are more tage of allowing smaller sample size and of problematic. Traditionally, ceramics were mea- removing from analysis inappropriate signals sured by TL on multi-aliquots, but increasingly (e.g., from feldspars contaminating a quartz sam- OSL is now being used on single aliquots using ple) that are averaged into multi-grain aliquots, SAR. The ability to date individual ceramics thereby avoiding potentially misleading results. directly (to their time of manufacture) has large But single-grain dating has some disadvan- potential for sorting out mixed assemblages, mea- tages. First, because the signals may be small, suring duration of occupation, and dating small poorer and more variable precision will result. sites difficult to date otherwise. The main draw- Second, even though De is measured on single back is the expense, but with prudent sampling grains, dose rate can only be measured currently strategies, it can serve to validate other less expen- on bulk samples (although research into determin- sive dating techniques. TL is still the preferred ing dose rates at a single-grain level is beginning). method for cherts or flints, as there is some ques- If the dose rate is heterogeneous, grains of the tion about the reliability of OSL. TL on flints has same age might have different De values. The 3170 Dating Techniques in Archaeological Science biggest problem, however, is that single-grain preferred to turbulent water. Colluvial sediments, dating produces a distribution of De values that including in rock shelters, can be dated, although can be hard to interpret, and that are the subject of short duration episodes such as landslides and considerable debate. Variation in De has both tsunamis can be problematic. Palaeosols are dat- intrinsic and extrinsic causes (Thomsen et al. able by isolating the youngest grains that were 2005). Intrinsic causes include statistical varia- brought to the surface by turbation and bleached tion, instrumental error, and variation in lumines- while the soil was on the surface. Ocean and lake cence properties of individual grains. Extrinsic sediments have also been dated, as has tephra. causes include partial bleaching, mixing, and het- Luminescence has also been applied to dating erogeneous dose rates. A number of statistical structural features. Dating structures has been a methods have been developed to evaluate these persistent problem in archaeology. Traditional distributions (Galbraith and Roberts 2012). dating approaches suffer from unconvincing Measure of central tendency is commonly bridging arguments. Materials in the structure assessed by the central age model. The model can be much older than the structure, and stuff also specifies an over-dispersion statistic that can around the structure can also be of different age. be interpreted as the amount of dispersion beyond Dating palaesols that underlie the structure is one that accounted for by differential precision. The solution, in effect treating the structure as a burial debate is over how much over-dispersion is allow- event. Dating construction materials, such as able for a single-aged sample. One procedure is to bricks, adobe, mortar, or rock surfaces, is another do a dose recovery test. This involves removing solution. Bricks were heated in antiquity but the the natural signal from an aliquot, administering a other materials need to be dated to last exposure to known dose, and then applying the SAR protocol light. Dating agricultural features, such as canals, to see if the known dose can be obtained. While a dams, or terraces, has also been tried, but finding test of procedures, the dose recovery on single- sediments that directly address the construction grains because every grain is known to have event is a problem. received the same dose, can also be used to assess A final issue in luminescence is whether to date intrinsic variation. Any over-dispersion in the nat- quartz or feldspar. (Calcium carbonate has also ural distribution beyond that obtained in dose been used in selected applications.). Both have recovery can be attributed, at least in part, to advantages and disadvantages. The luminescence extrinsic causes. If partial bleaching is suspected, properties in quartz are fairly well known and it a minimum age model can be applied to isolate does not suffer from anomalous fading, but it has statistically the youngest grains. If post- two drawbacks. One is that it saturates relatively depositional mixing is suspected, a finite mixture early (less than 100,000 years) and the growth model can be applied to divide the sample into curve is linear over a shorter range. Dating quartz single-age components. Determination of partial near saturation has several complications that can bleaching or post-depositional mixing requires result in under-estimations of age. The other is not only assessment of the distribution but also that quartz can be quite insensitive, with signals consideration of context and other factors. There 10–15 times dimmer than feldspars. In many parts is little consensus in the luminescence community of the world, particularly mountainous areas or about this, although convincing arguments have areas in northern latitudes, quartz is too insensi- been made in many individual cases. tive for practical use. Feldspar saturates much Many kinds of sediments have been dated. later and the growth curve is linear over a longer Aeolian sediments are considered the best choice period. But feldspar composition is more diverse because of the high probability of air-borne parti- (usually only potassium and sodium feldspars are cles being fully bleached, although post- datable) and luminescence mechanisms, particu- depositional mixing can be a problem. Fluvial larly charge transitions, are more complicated. sediments are prone to partial bleaching; sedi- Plus, anomalous fading is ubiquitous in feldspars ments laid down by relatively calm water are and must be addressed. Dating Techniques in Archaeological Science 3171 Electron Spin Resonance Dating Based on Weathering Electron spin resonance (ESR) is based on the A number of methods are based on chemical or same principles as luminescence but instead of biological changes in materials that accumulate releasing charge as a way of measuring it, ESR from exposure to weathering or during faunal measures the electron concentration in the traps. diagenesis. The rates of change for these methods More specifically, ESR studies paramagnetic are highly dependent on local environmental fac- centers with permanent magnetic dipoles. tors, making them difficult to apply. These are aligned randomly in the overall struc- ture so there is no magnetism, but they can be Obsidian Hydration D aligned in the presence of a magnetic field. Mag- Perhaps the most controversial is obsidian hydra- netic dipoles are produced by spin of unpaired tion. Developed in the mid-twentieth century, it is quasi-free electrons. Electrons in traps, or point based on the observation that water diffuses into defects, are not paired. Spin has two energy the surface of freshly broken obsidian and forms states, spin up and spin down. The difference a hydration rind that can be measured. Thickness between these two levels equals gmBH, where g of the rind can be related to age after taking is a characteristic factor of a particular paramag- into account composition, temperature, burial netic center, mB is quantum of magnetic moment depth, and other variables to estimate a hydration (Bohr magneton), and H is the magnetic force. If rate. There are two approaches. The empirical in addition to the magnetic field, electromagnetic approach determines the rate for any given obsid- radiation in the microwave frequency is also ian type either (a) by reference to independently applied, a strong absorption of energy results dated material in the same area or (b) by measur- where the microwave energy is equal to the ing the rate directly in laboratory experiments amount of energy needed to flip spin from up to using high water pressure and elevated tempera- down – a resonance effect. Microwave energy is ture. A problem with the latter is that short-term expressed as hv (Planck’s constant times fre- experiments may not provide a suitable model for quency), so the basic resonance equation is long-term processes. The intrinsic approach tries hv ¼ gmBH. Experimentally hv is kept constant, to estimate various parameters experimentally while H is varied. The resonance intensity of g is to calculate a hydration rate. The empirical proportional to the amount of trapped electrons, approach has limed applicability, while the intrin- and thus to the time since the electrons were sic approach faces a host of difficulties. emptied out. ESR is less sensitive than lumines- There is little agreement on any aspect of cence, and this limits dating to samples more obsidian hydration dating: how to measure the than about 20 ka. Most dating is done on calcium rind, what is the proper rate equation, and how carbonates where the dating event is precipita- to measure the various parameters of the equation. tion. It is less precise than U-series dating but has Despite the problems, a big advantage, making a longer age range (2 Ma). The most common continued research worthwhile, is that the target medium is tooth enamel, because it is resistant to event equals the dating event. Obsidian hydration diagenic change. ESR of carbonates suffers from measures when a piece of obsidian was last U uptake through time, which must be modelled. chipped. Often U-series dating is performed on the same Traditionally the rind, thought to have a sharp sample to help restrain when uptake might have boundary, was measured using an optical micro- happened. Because of its long time range, ESR scope. More recently, secondary ion mass spec- has played an important role in dating early troscopy (SIMS) has shown that the boundary is human evolution. ESR has also been employed gradational and that the diffusion of water extends to date heated and sedimentary quartz. Sufficient beyond the sharp optical border, which may have bleaching is often a problem in ESR dating of other causes. Rate equations are based on diffu- quartz, but it does have a longer dating range sion formulas called Fick’s laws. The original than luminescence. formula was x ¼ Dt1/2, where x is depth, D is 3172 Dating Techniques in Archaeological Science the diffusion coefficient, and t is time. Since D is SIMS) may be related to time through modified temperature dependent, it follows the Arrhenius Fick equations, but others are less optimistic, equation, D ¼ AeE/RT, where A is a constant opting for empirical approaches., which have peculiar to a given type of obsidian (related to been successful in a few studies. Perhaps the structural water content), E is the activation greatest utility of obsidian hydration currently is energy for diffusion to begin, T is temperature, as an ordinal measure, assessing the contempora- and R is the universal gas constant. In an influen- neity of artifacts in surface assemblages, tracing tial paper, Anovitz et al. (1999) argued that this dimensions of trade through time, evaluating arti- formulation is an inappropriate application of fact reuse and stratigraphic mixing, and determin- Fick’s laws because it does not take into account ing how well certain point types can be used in water concentration. The hydration rate changes cross-dating. as the absorbed water content changes. In plotting water against distance, the original formula has an Amino Acid Racemization exponential shape. Accounting for changes in Another method based on diagenesis is amino water concentration results in a sigmoid curve acid racemization. Original work on bone was (Fig. 4). Neither curve predicts a sharp diffusion not successful, but more recent work on shell boundary. The mechanism of hydration is not well has shown promise. Amino acids are the building understood, but it has been suggested that diffu- blocks of proteins. Most contain an asymmetric sion is slow at first because of high water concen- central carbon atom (called chiral carbon) to tration at the surface but then becomes faster when which four different side atoms or molecules are the glass structure undergoes hydrolysis. The dif- linked. This structure can be arranged in two fusion layer advances not according to t1/2 but tn, different ways, called isomers, which are mirror where n is empirically derived. Some argue that images of each other. All living organisms have parts of the sigmoid depth profile (measured by L-amino acid forms. When the organism dies, these forms spontaneously switch to D-amino acid forms, and then switch back. The proportion of L-amino acid forms steadily decline until some equilibrium is reached. The conversion process is called racemization, a term which refers to a change in optical properties from active (when one or the other is dominant) to inactive (when the ratio is 1:1). Some amino acids have more than one chiral carbon. When only one undergoes con- version it is called epimerization. The rate of conversion is dependent on many environmental variables, predominantly temperature. The time dependence of the reaction rate is the basis for dating. Depending on the type of amino acid and the environment, the maximum age range until equilibrium is achieved extends from 50 ka to 1 Ma. An intrinsic method of dating is too com- plicated, requiring detailed knowledge of reaction Dating Techniques in Archaeological Science, kinetics and temperature history. Rather, an Fig. 4 A depth versus concentration plot for obsidian empirical method is commonly used. Determine hydration. The solid line represents a profile from the original hydration formula. The dotted line is a sigmoid- the racemization rate at a site with samples of shaped curve representing a profile where the diffusion rate known age, and then apply it to nearby sites. changes with water concentration. The drawing is adapted Measurements is by high-pressure liquid from Fig. 2 in Anovitz et al. (1999) chromatography. Dating Techniques in Archaeological Science 3173 Bone is a poor candidate for amino acid race- Archaeomagnetism mization dating because it is an open system, The earth’s geomagnetic field changes signifi- subject to chemical exchange with the environ- cantly on scales of decades to centuries. Certain ment, introducing a host of additional variables minerals retain a record of these temporal to consider. Most studies focus on carbonate changes, called remanent magnetism. By compar- biominerals (aragonite and calcite): shell ing this magnetism to a dated record of change, (mollusk, brachiopods, ostracods), eggshell, the event when it was recorded can be dated. corals, and foraminifera. These systems are more There are three kinds of changes: direction and closed so mainly temperature affects the rates. intensity changes which are both regional in D Eggshell from ratite birds (ostriches, emus, and scope and polarity reversals which are global. rheas) has a very stable structure and works well. Directional and intensity changes are called It is common at many archaeological sites where secular variation. Thermoremanent magnetization these birds live. Exposure to fire, as in hearths, (TRM) is acquired during heating events of however, can dramatically affect rates. Over lim- ceramics, burned rocks, burned earth, hearths, ited geographic range and elevation, where shells kilns, and furnaces. The heating is the event can be expected to experience similar tempera- dated. Detrital remanent magnetization (DRM) is tures, shells with similar D/L ratios are all the acquired during aqueous depositional events of same age. This is called aminostratigraphy and is clastic materials. For both TRM and DRM, the used to correlate raised marine terraces and other target event and the dating event are often the kinds of formations. Mollusks are generally used. same. Archaeomagnetism is a derivative dating Changes in rates can be used to infer changes in method in that a master curve of magnetic changes temperature, helping to identify glacial and inter- must first be established and then dating is done glacial cycles. by matching the sample to this reference. The geomagnetic field is defined by three Ceramic Rehyroxylation parameters: declination, which is directional devi- This is a relatively new method based on the ation on a horizontal plane from geographic north; rehydroxylation of clays after they have been inclination, which is the angular dip from the fired. When ceramics are fired between 450  C horizontal plane; and intensity. The magnetic and 900  C, clays lose structurally bound field is produced by movement of electrical hydroxyl (OH) ions. After cooling, ceramics, sub- charge. Diamagnetic substances have outer elec- ject to even modest moisture levels, will slowly trons shells filled. When placed in a magnetic regain the hydroxyl ions. Data indicate the field, the electron orbits create magnetic field in rehydroxylation follows a power law after an the opposite direction of the applied field. Para- initial rapid stage. The equation is (m[t] – magnetic substances have incomplete outer ß/m0) ¼ a(T)t1/4, where m ¼ mass, m0 is mass electron shells and a magnetic moment due to after heating, ß is a constant related to the first uncompensated electron spin. When placed in a stage, a is a kinetic constant, T is temperature, and magnetic field, the internal magnetic field aligns t is time. The rate is determined experimentally by in the same direction as the applied field. In ferro- first weighing the sample before (ma) and after magnetic materials, magnetization from unpaired (m0) heating. Then the sample is continually electron spins can couple between adjacent atoms weighed while exposing it to moisture and keep- within regions called magnetic domains. The ing it at the temperature assumed to be average for extent to which they are susceptible to become the period in question. When rate of mass change magnetized is much higher than diamagnetic or becomes constant, then one can calculate the time paramagnetic materials. Their magnetization is require to reach ma. Original studies showed that retained even when the external magnetic field is the method agreed with independent evidence removed. The magnetism of ferromagnetic sub- over 2000 years of prehistory in Britain. The stances is dependent on inter-atomic distances. method is still undergoing development. These increase with temperature, so above a 3174 Dating Techniques in Archaeological Science certain temperature, called the Curie temperature reversal, the Brunhes-Matuyama, occurred at (about 570  C for iron), coupling breaks down and 780 ka and is important for archaeology. Short material becomes paramagnetic and any remanent episodes are called subchrons, and there are also magnetization is removed. As the material cools excursions which are not reversals but large devi- below the Curie temperature, the magnetic ations from normal secular variation that may not domains align in the direction of the earth’s mag- be global in extent. Polarity changes are dated by netic field over a relaxation time. Below some K-Ar. “blocking temperature” (somewhat lower than the Curie temperature), the magnetic alignments will become effectively frozen (TRM). When fer- Conclusion romagnetic grains settle in quiet water, they will align partially with the magnetic field. Subsequent With a wide assortment of dating methods avail- compaction can stabilize this orientation (DRM), able, how do archaeologists choose which method although the alignment can be disturbed by post- to use? The first step is to define clearly what the depositional disturbances. target event is, and then to choose which method Archaeological dating mostly uses TRM. An best addresses that event at the necessary preci- oriented specimen (hearths, kilns, bricks from sion. This will involve constructing bridging argu- walls) uses both direction and intensity for dating. ments to link the dating event to the target event. Direction is expressed as both declination and Often more than one method will be desirable to inclination, using a mathematical transformation build a convincing case and to increase temporal to define the latitude and longitude of a virtual information. Combining dates requires mapping geomagnetic pole (VGP). Unoriented specimens, out the relationships between all target events and such as pottery, use only intensity. To construct a all dating events: using the lowest common master curve of secular variation requires many denominator for resolution, defining limiting samples that are independently dated at fairly high points (terminus post quem or terminus ante precision. One method is graphical where VGP quem), and applying Bayesian statistics to con- positions are plotted and a curve is drawn through strain dates and identify outliers. A criticism points to minimize the variance. A second method might arise that dating methods lose their inde- is a smoothing technique called the moving- pendence if their evaluation depends on other window, where VGP values for overlapping dating methods, but are any methods truly intervals of time are averaged and weighted by independent in their application? Convincing precision and amount of overlap. There can a chronological arguments require convergence of problem of damping where the curves are sharp. different sources of information. A separate curve must be drawn for each region. About 1000 km is the estimated maximum for a valid region. Aside from producing absolute References dates, archaeomagnetism is useful for sorting out Anovitz, L.M., J.M. Elam, L.R. Riciputi, and D.R. Cole. intrasite chronology and assessing contemporane- 1999. The failure of obsidian hydration dating: ity between sites. There are some problems in Sources, implications, and new directions. Journal of assigning dates. Because the curves loop around, Archaeological Science 26: 735–752. there can be a problem of multiple intercepts, Bailey, G.N. 2007. Time perspectives, palimpsests and the archaeology of time. Journal of Anthropological which must be distinguished by archaeological Archaeology 26: 198–223. context or by intensity measurements. Baillie, M.G.L. 1995. A slice through time. London: B. T. Polarity changes occur at irregular intervals, at Batsford. least 23 times in the last 5 million years as Bowman, S. 1990. Radiocarbon dating. London: British Museum Publications. documented by cooled magmas erupted from Bronk Ramsey, Christopher. 2008. Radiocarbon dating: mid-ocean ridges. Long intervals or relatively Revolutions in understanding. Archaeometry 50: uniform polarity are called chrons. The last 249–275.