I bet that you'd say that you were driving both speaker simultaneously. Jeff - there are two aspects of your measurement strategy which are, as far as I'm aware unique to you, and not widely used in the speaker industry - and certainly not by Harbeth.
1. You drive both speakers simultaneously with a test signal even though they are in different physical places in the room radiating sound along different pathways to the microphone and via all reflective surfaces.
2. You are measuring a long, long way away from the speakers (13 feet, about 4m) at your listening position.
I would never attempt such a test setup because (in my opinion) although it will draw some sort of response curve on your analyser screen that curve cannot be interpreted adequately. Or at all. It is as good as meaningless. Why? First, if you drive both speakers, and they are in different places so (obviously) at some frequencies there will be constructive and at others destructive interference. The mic hasn't the intelligence of the human ear to resolve that - and you've already said that you are doing averaging in your analyser - so most likely, at the low frequencies where the wavelengths are long, the dissimilar path lengths from the speaker to mic will effectively add the L + R sounds together, whereas the different path lengths will cancel in the middle and especially upper frequencies. That explains to me the left-to-right downward slope of your curve.
Second, measuring so far from the speakers just isn't going to tell you what you want to know about the speakers (which you are criticising) at all. I fully appreciate
why you want to measure at your hot spot; because that's where you listen but you will not get a useful result with your method. Or any method across a useful frequency band without an anechoic chamber for a room. Or at least, I couldn't if I replicated your setup. I think you mentioned Atkinson/Stereophile, so I had a look at how he measures 'in room'.
Typical example here: note also bass lift. As he says, he measures the speaker at a number of points in an arc, about three feet (not thirteen feet) to about 4 feet from ONE driven speaker then averages the measurements together over the arc. That
will yield a useful response of how the speaker drives the room, but of course, in the speaker's nearfield. I use a similar method. That's the standard way of measuring a speakers 'in-room' response. Yes, it's most probably not the situation at the hot spot, but that point is extensively polluted by the room's contribution.
Finally, we can again refer to the BBC's work in this area. Bob Walker - brilliant engineer, last audio boffin and acoustician recently retired from Research Dept. wrote in his AES paper about the realities of speakers in rooms and the resulting distribution of their sound in the room:
-----------------------------------------
"In any partially or fully enclosed space that uniform spreading (of sound from a speaker) proceeds for only a short time until part of the sound wave strikes some acoustically significant object. What then happens is always complicated. Sound propagates as a wave function and demonstrate all the properties usually associated with the interaction of waves and objects -- reflection, refraction and absorption. What happens when a sound wave meets a discontinuity in the medium depends on the acoustic properties of the boundary materials and the size of the discontinuity in relation to the wavelength of the sound wave. Over the normal audio frequency span, wavelengths range from about 15mm to 7m. That nicely encompasses most sizes of objects within rooms, and even the room itself. Plus, the interactions between sound waves and the room and its contents cover the whole gamut of reflections and refraction effects, as well as absorption. It is that complexity which renders a real sound field impractical to treat analytically.
In a typical room there is usually at least the floor surface within about 2m of the source. Therefore, from a maximum of about 6ms onwards, the sound field (even outdoors) contains components which have interacted with some surfaces or objects. After 30 ms in a small room with a sound wave front will have travelled
in every direction to the boundaries of the room and will have interacted at least once with every object contained therein."
----------------------------------------
So, putting a microphone far away from any speaker in a non-anechoic room will swamp the microphone with initially the direct sound and then in just a few millseconds, reflections from all and every object in the room, and the room walls. What the mic sees will be the (small) contribution direct from the speaker plus a vast amount of reflected sounds, all added together. You will not be able to tease the speaker from the room because your averaging analyser is not intelligent; what you have is a composite measurement of speaker + room. I would be extremely cautious indeed about trying to equalise the response at your hot spot by slavish attention to the graph - I just wouldn't do it myself. By all means apply a little EQ to taste and by ear, but I strongly recommend that you do not frighten yourself with the graph and certainly do not use it as a guide to eq: use your ears for that
only. I suspect that if you EQ for a flat graphed response at 13' the result will be bass-light and too bright i.e. lean and thin.
I have a very busy week ahead with X-Factor TV this week but I think this is plenty coverage of this topic for now from my side - over to others.
Source: R. W . Walker, BBC. 'A simple acoustic room model for Virtual Production'. 106th AES Convention, Munich, May 1999. Reprint #4937