Paul, N1BUG, recently conducted some tests alternating transmissions of JT9-1 and JT9-2 on 2200m. This was possible as recent releases of the slow JT9 software have been more stable, thanks to the continued efforts of the author. I participated in the tests.
The results of the experiment are here: JT9-2 vs JT9-1 Report
The conclusion from the report is as follows:
Closing Remarks --------------- On the basis of this test it is evident JT9-2 mode in SlowJT9 decodes at significantly lower signal level than JT9-1 can. Even considering signal fading with the longer transmission periods, JT9-2 gets the message through more often.