> I don't quite see the issue here. Can you please elaborate? Note the timestamps. In the first example, the last entry was from May 29 23:05:13 when the last 14 entries were requested, but from May 30 07:05:03 when the last 12 were. The expected behaviour is that the last entry is always the same (assuming that no new matching entries were added between journalctl invocations), however many of the most recent entries is requested.
In the second example, there seems to be a pattern: it skips the same amount of repeated messages from the end as the number of entries that is requested. The expected behaviour is that it doesn't skip anything from the end, and prints the most recent entries, as described in journalctl(1). I've checked the second example again now, it still behaves that way here -- just the timestamps get shifted, as was shown previously.
