DFSORT initially intended to use only x work data sets for this sort application. However, the work space requirements for the sort were unexpectedly larger than anticipated, so DFSORT used y additional work data sets.