Explanation:
It's possible that sometimes the LGWR process must wait because DBWn has not compressed check pointing a file. To avoid this situation you need to increase the number of database writer (DBWn) processes.
Incorrect Answers:

1: You may also find that contention exists between your ARCH and LGWR processes during checkpoints at log switches. To resolve this performance problem, you must do the following. Place your redo log files on separate disk resources so that ARCH can copy the recently filled online log to the archive destination. Also, place your archive destination on yet another separate disk resource, away from all the online redo logs. But question says nothing about log switches, so you can consider this answer as incorrect, because other answer is correct.
2: The number of members for all log groups has no influence on the frequency of checkpoints.
3: By increasing the number of log writer (LGWR) processes you will not fix problem, because it is related with database writer processes: they don't have enough time to write all dirty buffers into data files.
Oracle 8, DBA Certification Exam Guide, Jason S. Couchman, p. 902. Chapter 19: Tuning Disk utilization

---