Now if you look at the comment about measurement systems, Wayne makes a good point, although perhaps not as big of deal as it might seem. I believe Jeff and bill are using omni mic, Matt uses REW, and I use HolmImpulse. All the other designs I've seen by a few of the more casual guys have been with REW. Ill speak to HolmImpulse. It uses a swept sine and I've adjusted the program settings to get something like 20,000 data points. I use the windowing about 1ms before the first reflection in the impulse response, usually 8ms as I usually measure outdoors. I don't use any smoothing. Occasionally ill show a full range smoothed nearfield response only to show the low end response.
This technique has the advantage of removing reflections, no smoothing over important detail, at the cost of a smeared low end response. My measurements are useless below 200hz, unless I take specific measurements to get that data.
If ever you see evenly spaced dips and peaks you gotta ask, is this a reflection. Often it is. Next, where is the reflection? If its a nearby boundary in the setup, look for better data. If its a reflection in the waveguide, it better be very small, otherwise its a problem. Foam in the throat anyone