The process of selecting part of a site for excavation or an area for fieldwork, preferably according to a strategy which allows statistical estimates and generalizations of the relation of the sample to the unexplored parts of the whole site or area. In a way, all archaeological fieldwork and excavation is sampling, since it is impossible to collect all the data from the complex mass of an archaeological site. Selection may be arbitrary or nonarbitrary - perhaps by the need for particular evidence for a specific question (a 'judgment sample'); the question itself will be determined by the existing framework of archaeological thought. In a more specific sense, sampling or probabilistic or random sampling, uses the theory of probability to make estimates of how closely the observations obtained from the part examined ('sample') represent the characteristics of the whole group being studied ('population'), by using fixed rules of random selection so that each unit is given a known chance of selection. The area under study may be divided into sub-zones (strata) and each stratum can be sampled separately to give a more precise estimate of the whole population. The choice of sample design, the size of the sample units, and the proportion of the population sampled (the sampling fraction) will all affect the result, but even with quite small fractions accurate estimates of the entire population of sites within an area can be obtained. The method is particularly good at estimating the number of different types of site within the area. Methods are also being developed for the sampling of large groups of artifacts; excavations frequently produce very large quantities of bone or flint, and it has been shown that often it is necessary to study only a small sample of the whole population to obtain a reliable estimate of its character.