Paleodemography is the study of past populations that have left no or few written documents containing evidence of their demographic behaviour. Its objectives are similar to those of historical demography, the difference being more the nature of sources used than the periods of study. Whereas historical demography is based on written sources, paleodemography relies on material sources. Epistemology of the Discipline The paleodemographic approach defined here is based on a single source: human bone remains, which are a precious biological archive. As it must be done for all historical data, the limits of sources and of their representativeness are discussed, as well as ideas about age. The Osteological Data To adopt an effective demographic approach, the anthropologist attempts to determine the age at death and the sex of skeletons, while the datation dating of the settlement is provided by the archaeologists from stratigraphic and typological observation. The chapter examines various methods in estimating sex and age at death, which all assume that the indicators remain constant over time, under the principle of the biological uniformity of Homo sapiens sapiens, and which all present statistical biases due to the poor/weak correlation between biological age and civil age. Establishing a Reference Population This chapter explains the construction of the reference populations we will use, after examining their intrinsic qualities: one for children, other for adults, both based on a specific and single biological criterion. In order to be as close as possible to the mortality patterns of pre-industrial populations, albeit not ancient or medieval were largely non-urbanised and had little or no access to modern medicine, these reference populations could be modified following the age at death distribution observed in pre-transitional population of Lisbon (Portugal) at the end of 19th century. Age at Death: Current Approaches and Methods To propose paleodemographic reconstitution from skeleton data, it is necessary to use criteria accessible from historical and archaeological data and, as a demographic study directly based upon archaeological data is not possible, it is necessary to define tools which take account of osteological constraints. This chapter examines available tools and propose some adaptations. Current Demographic Models The purpose of paleodemographic analysis is to understand the population as it was, within a given socio-environmental context, where the individuals it comprised formed a dynamic group marked by births, deaths and sometimes migrations, and when all we have, at best, is the bones of those who died. This is no easy task and turning to (requérir à) population models and life tables or parametric models is necessary. The hypothesis required for such models are remembered and the main flife tables actually used by demographers and paleodemographers are reviewed. Finding the Right Models for Pre-industrial Populations As contemporary model tables, or mathematical models based on current conditions, are not the best tools for describing pre-transitional human behaviour, characterized by very high rates of fertility and child mortality, it appears necessary to develop for preindustrial populations. These models have to present some characteristics as to be rested on (s’appuyer sur) a large corpus of tables statistically representative of mortality among pre-industrial populations, to propose inputs easily accessible from historical or osteological sources and to take account of the growth rate, either positive or negative, of the population. Model Tables for Pre-industrial Populations Following the previous recommendations and using the linear regression between the logarithms of the probabilities of dying and some paleodemographic indicators, we propose a network of life tables specially developed for preindustrial populations, under the properties of stable populations, and discuss benefits and limits. Definition and Exploration of a Pre-industrial Standard The previous models, based on are of poor quality for the early ages because of the wide distribution of observed values. In this chapter, life tables are constructed by the logit method, based on the same sample of observed tables and on the mean of the probabilities of dying by age. This model appears to be one of the best ways of studying infant and child mortality. Final Overview An archaeological excavation has delivered a set of skeletons. What conditions must be met if we wish to use this set of skeletons in a demographic study? And if these conditions are met, what options should be chosen and what are the basic stages in the process? Methodological proposals and tools developed in the previous chapters are recapitulated in order to define a protocol for paleodemographic studies. Examples of Archaeological Applications Demographic approach to buried populations can be now reexamined with the new “tools” proposed in this handbook. Some of the recurrent questions of paleodemography, such as the problem of estimating the proportion of under-20s, and new research topics can be explored from four case studies: for two of them only biological data are available (burial grounds from Late Antiquity comprehensively excavated and covered by thorough historical, archaeological and anthropological studies). The last two are sites with extensive written documentation that provide information on the demographic behaviour of the exhumed populations. The analysis of these sites is not necessarily complete nor is the discussion closed. Critique of Current Methods To extend the reflection introduced by Isabelle Séguy and Luc Buchet, we propose a critical overview of the methods used by paleodemographers the main approaches used by paleodemographers to estimate the age structure of a population for which they only have biological indicators measured from skeletons. None allows to solve solving the composed statistical problem in a satisfactory way. A more satisfactory method should use a full Bayesian model, considering the random nature of both the reference and osteo-archaeological data. A New Method for Estimating Age-at-Death Structure This chapter presents methodological principles of a new strictly fully Bayesian method, based on a precise statistical model taking into consideration the uncertainty of all the data upon which the estimation is based, in order to solve this major and recurrent problem for palaeodemographers. The main advantages of this approach, compared with the previous ones, are discussed and paleodemographers are encouraged to propose any improvements further insights that may be necessary to complement this first approach.