Sequence number format and modification level

Sequence numbers can be generated in the standard sequence field, the COBOL sequence field, or both:

The standard sequence field is the last 8 characters for fixed-length records, or the first 8 characters for variable-length records, regardless of the programming language. Use NUMBER ON STD to generate sequence numbers in the standard sequence field.
For members of partitioned data sets, the format of standard sequence numbers depends on whether statistics are being generated. If statistics are being generated, standard sequence numbers are 6 digits followed by a 2-digit modification level number. The level number flag reflects the modification level of the member when the line was created or last changed. If, for example, a sequence number field contains 00040002, the line was added or last changed at modification level 02. The sequence number is 000400.

If STATS mode is off, or if you are editing a sequential data set, standard sequence numbers are 8 digits, right-justified within the field.
The COBOL sequence field is always the first 6 characters of the data and is valid only for fixed-length records. Use the NUMBER ON COBOL or NUMBER ON STD COBOL to generate COBOL sequence numbers.
Attention:
If number mode is off, make sure the first 6 columns of your data set are blank before using either the NUMBER ON COBOL or NUMBER ON STD COBOL command. Otherwise, the data in these columns is replaced by the COBOL sequence numbers. If that happens and if edit recovery or SETUNDO is on, you can use the UNDO command to recover the data. Or, you can use CANCEL at any time to end the edit session without saving the data. COBOL sequence numbers are always 6 digits and are unaffected by the setting of STATS mode.

Sequence numbers usually start at 100 and are incremented by 100. When lines are inserted, the tens or units positions are used. If necessary, one or more succeeding lines are automatically renumbered to keep the sequence numbers in order.