Sampling a site is a necessary step in almost all archaeological excavations. All archaeologists do this at some point in their careers, and as such it is no surprise that several methods for sampling have developed. These methods are listed below along with a brief summary.
1) POINT/SPOT : In this type of sample, very small areas (or points, or spots) are taken at regular intervals. If enough funds are available to sample 5% of a site 1000 meters square, this method would perhaps explore 100 points a half meter square. (If my math is correct, of course). This is an advantage because a larger number of different locations can be sampled. However, this method could miss by inches a larger find, or could turn up numerous locations with nothing.
2) QUANDRANTS : This is a similar method, only instead of using small points, slightly larger squares are used. Taking the same 5% and 1000 meter square area, 10 sites of approximately 5 meters square might be sampled. This would produce the same 50 square meter investigative area as above, but would be a much more hit-or-miss approach. It is useful if a general idea of where to look is available, or if fairly random distribution of artifacts is ensured through some means.
3) TRANSECTS : This method is like the quandrant method in that a smaller number of larger samples are made, but the shape of the samples differ. Rectangles are generally used instead of squares. This is useful for determining distribution of artifacts from a town center, a river, and so on. Instead of having 10 five square meter samples, an archaeologist might take 10 one meter-by-five meter samples. This is particularly useful if an archaeologist has a spatial-related question in mind.
Also, like surveying, sampling can be probabilistic or non-probabilistic. However, unlike surveying, sampling can occasionally be COMPLETE. Often funds are not available for this, but it is an archaeologist's dream, and, for some, a reality, or as much of one as a site being completely uncovered.