Fossils appear when erosion or digging uncovers them. Erosion goes on all the time, in total disregard of the presence or absence of fossils. Hence many fossils are uncovered and may be casually destroyed. The ideal fossil is one whose position in its surroundings is still identifiable; and it represents as complete a record as possible of the original organism. Isolated bones and teeth are important--they have told us much about human evolution--but complete skeletons are, of course, much more informative, though much less likely to be found.
Sedimentary formations are naturally the richest source of fossils for the paleontologists who seek them. Here fossils were formed and preserved through deposition of muds and silt. When such a bed of fossils is found great care is taken in the removal of specimens. Often, chunks of the sedimentary material are removed and worked on with special tools in a laboratory so as to assure intact specimens. Depending on the size and nature of the fossil, the paleontologist's tools are appropriately varied. In looking at microscopic fossils, the techniques of a gem cutter and polisher are utilized. This assures thin, transparent preparations for the microscope. Larger fossils may require accurate splitting of sedimentary layers followed by the addition of preservatives, e.g., plastics, to protect the fossil remains. And large dinosaur bones are often covered with plaster at the site of their excavation to protect them from rain and other wear and tear. Most fossils are destroyed before they catch the eyes of fossil-hunters. Sedimentary rocks are changed through movements of the earth's crust or through heating, and this damages or destroys fossils. When we put this fact together with the realization that relatively few organisms are successfully fossilized, we can conclude that the fossils we have recovered are a very limited sample of the fife of the past. But that conclusion protects against the extremes of wild speculation orpessimistic conservatism in interpreting the fossil record. Neither the extreme of assuming we have a complete record of the past nor the extreme of assuming we can say nothing of the past is justified.