The purpose of this study is to see if it is possible to determine the course of forest succession over several hundred years by the dissection and close scrutiny of live and dead plant material. The study area was a one—tenth—acre (0.04—ha) square plot in a forest that had never been cut, located near Ashuelot in southwestern New Hampshire. Within it stem—location and growth—rate data were collected from three different classes of stems: 1) living trees, 2), dead unburied stems and stem fragments, and 3), woody remnants buried in the forest floor. The vegetational history of the plot was reconstructed from before 1665 to 1967. During this time, autogenic succession did not contribute significantly to compositional changes, whereas disturbance was an important mediator of such changes. Furthermore, changes in forest structure were often manifestations of species behavior: each tree species had a distinctive stem—distribution pattern in relation to microtopography, mode of germination, and growth. The approach seemed to provide a useful means to obtain evidence about forest change through time that may help prediction and illuminate ecological theory. Limits and modifications of the procedure are discussed.