The Cartesian product file is an efficient method of handling the partial match queries to retrieve a record from a file, which has the specified values for the specified attributes. The optimal processing efficiency for the query is realized in an environment where multiple disks with parallel access are employed by allocating the Cartesian product file among the disks in a distributed way so that the buckets which may contain the record satisfying the specified condition are read uniformly from the disks. Recently, Faloutsos and Metaxas proposed a method whereby the Cartesian product file is allocated to the multiple disks using the error-correcting code. In this paper, using their method of allocating the Cartesian product file in the multiple disks, the following is presented: (1) the sufficient condition for the queries so that the processing efficiency is strictly optimal; 92) the necessary and sufficient condition for the code that realizes the allocation with the strictly optimal efficiency for any queries; and (3) the expression for the average response time for the query by which we can easily compute the average response time when the weight distribution of the code (or that of the dual code) is known. Through several examples, it is shown that the average processing time determined from the expression is very close to the theoretical lower bound.