The above image shows the initial impedance response of a tapped horn I built. Note that while the first two impedance peaks look like fairly close matches, some things stick out: (1) the actual impedance peaks are slightly higher in frequency than the predicted ones, suggesting that the path length may be slightly shorter (I predicted the possibility of it being off by around 3%~4%). This could also possibly be due to other things, like difference between HornResp's value for c and actual c in my location. However, more important though are the greatly reduced third peak and some ripples above 100 Hz, and the difference between predicted and actual impedance between 20 and 40 Hz is a bit higher than it appears to be elsewhere in the response graph. This suggests that the enclosure is not sufficiently braced and the panels are not properly damped.
The image above shows the same tapped horn, this time with a bit of bracing added to the first segment. The upper impedance peak is easier to distinguish, but it's not very high, and there are still a few ripples in the response that are not predicted by HornResp.
The image above shows the same tapped horn, this time with a lot more bracing added to the first segment (basically an additional layer of plywood was glued and screwed to the external panel at that point). Some bracing was also added to the third segment of the horn (found at the bottom of the box). The upper peak is now higher and very clearly defined and the ripple @200 Hz has disappeared.
The above image shows the same tapped horn, this time with more bracing added to the third segment of the horn (two 5" x 20" strips of ply added to the bottom of the box). A small brace was also added internally to the second segment of the horn. Not much has changed, but it looks like there's now closer correlation between the predicted and actual impedance between 20 and 40 Hz. Further bracing is likely not going to make any significant difference, unless perhaps I manage to find what panel resonance is causing the little blip @ 100 Hz.
Finally, the image above shows the same tapped horn, with 185 Lbs of weight added to the enclosure (basically me sitting on top, LOL). All panel resonances are now sufficiently damped to the point of having little or no visible impact on the impedance curve.