Mechanical seed identification [User-computer dialogue at the U.S. National Seed Herbarium]

A user-computer dialogue has been established at the U.S. National Seed Herbarium to identify isolated seeds and fruits. Four major data processing programs are used to make identifications or to solve taxonomic problems. Properly prepared documentation sheets facilitate the recording and numeric coding of diagnostic seed-fruit characters. These coded characters are data banked as dichotomous, multistate, or quantitative observations. Because the manual completion of these documentation sheets may be quite time consuming, systems are being considered to automatically generate computercompatible data. Elements of one system are the scanning electron microscope (SEM) and an optical scanner with a direct computer feed. SEM photographs could be optically scanned to determine seed shape, size, and surface topography as well as the relationships of the hilum, lens, and aril to seed shape and size. The computer is programmed to suggest useful characters for an identification, to recognize character and taxon weights, and to permit correction of user errors while an identification is in progress.