It actually seems fairly obvious to me also, but unfortunately, I don't think my appraisal is going to agree entirely with TJHUB's conclusion, so you guys can do whatever you like with my comments here.
After reading through everything in the thread and looking all of the plots over I am seeing essentially what I would expect to see based on what I have read (with one exception, which I will get to in a minute).
First, regarding the mic calibration file, the Behringer ECM8000 is a fairly flat mic to begin with. Even to use it without a calibration file will not result in these apparent discrepancies, in fact I doubt you would notice the difference. Rather, I think the explanation is much simpler -
From what I gather, these are in-room response measurements measured at the listening location ( I did not see the actual distance specified, unless I missed it somewhere). If this distance is 8-10 feet then I would expect to see:
....room gain in the lowest octaves
....boundary cancellation and peaks due to the room response
....and a falling top two octaves due to directivity and absorption
And this appears to be exactly what we see. There is a difference between the two speaker, but he commented that he did not move them to the same location, so that can easily explain these differences too (at least partly).
Now, about the falling top-end response - I do not know how well the mic is set-up on the tweeter axis. If it is off-axis much it will contribute to an even greater drop, so this is something to consider as well.
I always check my speakers in my room using an RTA with pink noise to see how the balance looks in-room. When the balance is very flat at one meter it will still show a significant loss in top octave energy at 8-10 feet, but it doesn't sound that much different, because at that distance we are hearing a greater proportion of the speaker's power response, which the mic does not pic up. Again, what I see in these graphs correlates well with what I see in listening position measurements in my own room, with one exception.....
The dip at 2kHz in the one speaker should probably not be present in one and not the other. If his measurments were vertically off-axis (from the tweeter axis) then I could see picking up the beginning of a null in the lobing along this axis, but the top-end should be more rolled off if this were the case. Based on this, I tend to agree with Dennis, I think the tweeter in that speaker may be connected with the wrong polarity compared to the other speaker. Otherwise, I don't have much of any issue with what I am seeing here.