Construction of efficient identification schemes using batches of discrete-valued tests

Well established methods are available for identifying a specimen from a known set of taxa by applying either a single set of tests, or a hierarchical sequence of single tests. These are relevant respectively when the cost of the tests or the time required for identification is most important. Often, however, a mixed strategy is required which aims to reduce the cost of identification while avoiding taking too much time. Methods are derived for constructing efficient identification schemes which involve the sequential use of batches of tests. Also, the special case is considered where an initial batch of tests is used, followed by a second batch whose composition depends on the results observed for the initial batch.