In SAS 9.2^®, when you try to merge two data sets and you have set the SORTSEQ= option globally to specify an alternative character-collating sequence, you might receive the following error message stating that the input data sets cannot be combined because they have different collating sequences:

ERROR: Input data sets cannot be combined because they have different 
collating sequences (SORTSEQ).

This error can happen if one of the data sets involved has not been explicitly sorted but it is using a default collating sequence (for example ASCII on an ASCII-based machine). No explicit collating sequence is set in the data-set header in this case. This occurs because SAS^® is running on a platform that uses ASCII-based session encodings, and the ASCII collating sequence (SORTSEQ=ASCII) is considered redundant. As a result, SAS sets the collating sequence to all blanks to indicate that the default collating sequence for that platform was used. The following example provokes this error:

The error occurs to prevent data that has not been sorted with the same collating sequence from being merged with other data.

To fix this you need to ensure that both data sets are sorted. Alternatively, if you know that the data set is sorted, you can apply a weak assertion, which you set via a data set option. In the following example, the weak assertion is set with the SORTEDBY= option:

You can also apply a strong assertion, which is set by the SORT procedure. In the following example, the strong assertion is indicated with the PRESORTED option. PROC SORT skips the sorting process when it encounters a strong assertion that the data has been sorted.

Operating System and Release Information

options sortseq=national; 
data test1; 
input v1 $ v2 $; 
datalines; 
a 1 
b 1 
c 1 
; 
data test2; 
input v1 $ v3 $; 
datalines; 
c 1 
b 1 
a 1 
; 
run; 
proc sort data = test2; 
by v1; 
run; 
data test3; 
merge test1 test2; 
by v1; 
run;