When split-file processing is in effect, the first
variables in the matrix data file will be split variables, followed
by ROWTYPE_, the grouping variable, VARNAME_, and then the variables used to form
the correlation matrix.
A full set of matrix materials is written for each
subgroup defined by the split variables.
A split variable cannot have the same variable name
as any other variable written to the matrix data file.
If split-file processing is in effect when a matrix
is written, the same split file must be in effect when that matrix
is read by another procedure.