Archaeologists make a particular effort to place material evidence from the past within a chronological framework. Every object—whether a wall, mosaic, ceramic vessel, coin, ditch, temple, house or inscription—must undergo internal criticism to determine what it is, by describing its immediately observable physical and morphological characteristics, in the hope that these bear a definite relationship to time. In addition, all available information about its setting must be gathered: location, depositional context and association with other objects. Once these conditions have been met, a chronological construction may temporarily claim a degree of validity, pending further evidence.
«Relative» and «absolute» chronologies
Although somewhat inappropriate, the expressions relative chronology and absolute chronology nevertheless describe two equally valid ways of referring to time. The former is, so to speak, natural to archaeologists, especially those with few or no written sources—for example, specialists in prehistory and protohistory. In this sense, related objects, such as ceramics or archaeological layers, are ordered so as to reveal the relations of anteriority, posteriority or simultaneity between them. It is in this sense that the chronology is said to be «relative». Three well-established methods make it possible to establish this type of relationship: seriation, stratigraphy and spatial analysis.
«Absolute» chronology is an approach whose aim is to place material evidence within the quantified time of the calendar. For schoolchildren, it is the time represented by the timeline running around the classroom walls: linear, calculable time. A few rare archaeological objects, such as coins or lapidary inscriptions, can be read in a way that provides relatively easy access to a date—or rather, to a range of possible dates. However, methods based on laboratory physical measurements, obtained through radiocarbon dating, archaeomagnetism or thermoluminescence, are those most readily associated with absolute chronology in everyday life.
Bayesian statistics (see the preceding pages) can help reconcile these two approaches. Unlike classical frequentist statistics, it makes it easy to take account of information associated with the context of events, in addition to dating measurements. This information, such as temporal ordering relations inferred from the position of one event relative to another, is incorporated via the prior distribution of the Bayesian model. This technique is particularly valuable when the material culture of the civilisations under study is entirely devoid of dating artefacts. Faced with this lack, it may be tempting to seek analogies with neighbouring civilisations whose chronology is better established because they have more complete sources, or simply to rely on a dating tradition. In either case, the proposed chronologies would rest on fragile foundations far removed from the actual context of the discovery.