Sentences in rhetoric texts were used in this study to determine a way in which thetorical syntactic parallelism can be analyzed. A tagmemic analysis determined tagmas which were parallel or identical or similar to one another. These were distinguished from tagmas which were identical because of the syntactic constraints of the language (invariable tagmas). Parallel constructions ranged from compound parts of speech to a virtual network of parallelism in which a given strand or chain (descending patterns of tagmas) could be considered to be parallel to several other strands, each in varying relationships to the original and to one another. This study made it possible to compare parallel constructions according to: (1) the amount of parallelism as determined by the number of identical or similar tagmas in each sentence, (2) the "depth" of each construction as determined by the number of tagmas in the longest strand, (3) the "breadth" of each construction as determined by the number of strands in each construction, (4) the "effectiveness" of each construction as determined by the number of variable tagmas, (5) the "strength" of each construction as determined by the amount of lexical identity, and (6) the "rigor" of each construction as determined by the number of identical rather than similar tagmas in each. (Author/HOD)