None of my tests have shown any difference between using 'Arp Bypass' versus just using the 'Arp Master' switch to prevent arp output.

Are there specific use cases for using 'Arp Bypass' rather than just using 'Arp Master'? If so, what are they.

1. Neither switch setting is preserved on a power OFF/ON cycle

2. Neither setting permanently affects the stored/permanent value of a part Arp switch. The 'Arp Bypass' switch (but not 'Arp Master') DOES alter the current/effective value of the part Arp switches by using OFF for all parts even if the setting is ON.  

Are my assumptions regarding setting 'Arp Master' to OFF true? Testing seems to support them.

1. Any currently running arps, and the sound they produce stops immediately

2. No arps are triggered while the master switch is off

3. No polyphony resources are used by any arps

So my question is: are those assumptions also true for 'Arp Bypass'?

If they are then why is an 'Arp Bypass' switch even needed or useful given that you could just use the 'Arp Master' switch on the front panel?

Tests show that setting 'Arp Bypass' to ON:

1. does NOT permanently affect the stored Arp part setting for any part. If part 3 has Arp ON but you change 'Arp Bypass' to ON the screen will show part 3 as having Arp OFF. But if you now store the performance it will show part 3 with Arp ON (and 'Arp Bypass' as OFF) after you load it again because ON is the stored setting - it was just temporarily overridden in memory when 'Arp Bypass' was set to ON. 

2. immediately stops the sound of any currently sounding arps.

What my tests haven't been able to determine is whether the arps are actually stopped or if they keep running (but not sounding) in the background.

Nor can I determine if new arps are not started, and using polyphony, or if they are started but just can't be heard or have their midi exported to a DAW.

Why was 'Arp Bypass' added when it seems that the front panel master switch can already do the same thing?