The notion of optimum coding is traditionally associated in linguistics with a binary record of the phonetic level units and/or other equivalent symbols, e.g. letters (Jassem 1974: 239-297; Herdan 1966: 259-303; Hammerl, Sambor 1990: 418-420). Its purpose is to demonstrate that variable frequencies of occurrence of the examined symbols may reduce the length of the coded message, i.e. indirectly reduce the time of its processing (understanding, reproduction and/or sending). A similar technique may, however, also be applied to the subsystems of the language. In particular, with the use of this method, one may code lexical fields, treated here as subsets of the lexical subsystem.