If I recall correctly...
In J1 alone, the two 68k Resistors are in parallel so their effective Resistance is 34k.
In J2 alone, one of the two 68k Resistors is tied to ground so the input signal is no longer in parallel (rather, it is in series with the one 68k Resistor) and the effective Resistance is 68k. Twice that of J1 alone.
Therefore, J1 alone is louder because there is less resistance, 34k ohms (34k is less than 68k), to the input signal in the parallel circuit feeding the V1 input grid.
The 1 Meg Resistor that is tied to ground is placed in the input circuit to help establish the high input impedance similar to other electrical devices like many of the Volt-Ohm meters where 10 Meg Resistors are placed at the positive terminal test jack input.
But when both J1 and J2 are plugged in at the same time, each has its own 68k Resistor in series (no longer in parallel). In essence, this makes them both the same level as J2, not J1, because each is offering 68k ohms of Resistance in each input.
In other words, if both J1 (Hi) and J2 (Lo) are used at the same time they are at the same J2 (Lo) level.
If my memory is incorrect or my explanation is incorrect please correct me with a detailed explanation.
This is fascinating, because I've never even noticed input #2. Plugging into #2, you'd turn the volume on the amp up a little and get a little more preamp gain, and/or add gain to the inputted signal to equal the volume you'd get using input #1. But the power amp would see the same level of signal coming out of the preamp. So you conceivably get a little more preamp distortion but, at the same output volume, you would not get any more power amp distortion. Do I have this correctly?