Estimating the population of urban agglomerations over historical time periods is a major challenge due to the complexity of the urban growth process. Villages can become towns, towns can grow into cities, and cities can be transformed into urban agglomerations in a number of ways: They may increase due to natural population growth - that is as a result of a larger number of births than deaths; they may grow due to rural-urban or urban-urban migration; or they may emerge as a result of administrative changes. These administrative changes can also involve several different processes: They may include the incorporation of sub-urban areas or neighboring towns into a larger city or the foundation of a completely new city - as was the case with several newly established national capitals. Administrative changes also include the re-naming of urban agglomerations - particularly those that had foreign names been assigned during the colonial period.

The figure above tries to highlight some of these processes. The challenge of estimating the population of human settlements arises from the fact that both their name and their physical extend may have changed over past decades - sometimes several times. Different towns (with different names) may have been combined into one urban agglomeration - with the name of one of the original towns or with a completely new name. Population statistics referring to a "city proper" may be discontinued and replaced by data for the much larger area of an "urban agglomeration" with the same name - which may, or may not, include other neighboring cities.

In the World Urbanization Prospects we make an effort to prepare consistent population estimates over historical time periods for human settlement that follow the same definition. This often requires that historical population estimates have to be re-adjusted to a current definition that may not necessarily coincide with other popular definitions for that city. The definition used in the World Urbanization Prospects is documented for each city in a database of Data Sources.