one argument that is being made against the seos (seos12) is that it is too short (as measured from the throat to the mouth).
if it was too short, it would not load the compression driver through the passband, i.e. the frequency response would drop like a rock. this starts to occur around 800hz for the seos and is in full force by around 600hz, which is well out of the intended passband for the design.
the second argument that the seos is too short says that internal reflections will cause peaks and nulls in the passband. this is actually a very sound theoretical argument. the problem is the seos has a very large roundover, which makes for a very large transition region where these reflections will occur. where exactly is the front of the horn, 2" in front of the c.d., 2.5", 3.0", 3.5"?
for illustration, below is an approximate cross section of the 290c horn 9 (red) and the seos12 (blue). the 290 has a relatively small roundover relative to its depth, so the distance from the front to the back of the horn is fairly easy to measure and the transition region from horn to free air is fairly abrupt. the 290c is definitely longer (throat to mouth), but the benefit of that is that it loads to 800hz, where there is a fairly large peak in the frequency response. what benefit that provides is unclear. at least it is out of the pass band.
the seos is a little different. its large roundover makes calculating a distance from the throat to the mouth difficult and the transition region from horn to free air is large and smooth. almost a full 1/3 of its depth is a "transition region" from the pure catenary portion of the flare to the front of the horn. as a result, reflections have no real "wall" to bounce off of.
the result is that on the bottom end directivity collapses smoothly below the passband and there are no corresponding significant passband peaks and dips.