The number of observable properties of sedimentary rocks is large and numerous classifications have been proposed for describing them. Some rock classifications, however, may be disadvantageous in situations such as logging rock core during coal exploration programs, where speed and simplicity are the essence. After experimenting with a number of formats for logging rock core in the Appalachian coal fields, a method of using color photographs accompanied by a rock name and numeric code was selected. In order to generate a representative collection of rocks to be photographed, sample methods were devised to produce a representative collection, and empirically based techniques were devised to identify repeatedly recognizable rock types. A number of cores representing the stratigraphic and geographic range of the region were sampled so that every megascopically recognizable variety was included in the collection; the frequency of samples of any variety reflects the frequency with which it would be encountered during logging. In order to generate repeatedly recognizable rock classes, the samples were sorted to display variation in grain size, mineral composition, color, and sedimentary structures. Class boundaries for each property were selected on the basis of existing, widely accepted limits and the precision with which these limits could be recognized. The process of sorting the core samples demonstrated relationships between rock properties and indicated that similar methods, applied to other groups of rocks, could yield more widely applicable field classifications.