Question about the LX521's crossover

The LX521's crossover involves a 1st order crossover between the lower and upper midrange.

How is dipole EQ applied to the dipole rolloff of the upper midrange when the passive crossover is 'in the way'?
(Dipole rolloff for the upper midrange starts sooner than the lower midrange due to the narrowing of the baffle between upper and lower midrange.)

Does SL just arbitrarily EQ the resultant dip flat instead of precisely applying a 6Db/octave EQ to the upper midrange?

I read SL's commentary on his web page but dont seem to find the explanation.

Trying to do a similar design using a 'filler driver'.
 
Making a clone requires lots of test measurements and basic knowledge about dipole behaviour.

Im not making a clone at all. But i want to apply the same 'filler driver' concept, its a great concept.

I think that was only the first iteration of the LX521 that used a passive 1st order filter between the 2 mids, he later went to 4 way active using the minidsp.

Of course that will always be sonically better - but I rather get rid of an expensive set of amps/dacs