The TH118 actually drops to 3 ohms at 67 Hz compared to the QW218 which is about 10 ohms at this frequency. Two TH118 speakers = 1.5 ohms @ 67Hz! Given this is right in the middle of its intended operating range I struggle to accept its nominal impedance being listed as 4 ohms but it’s probably OK...just.
The other thing to note is; some of the specifications listed in this thread are whole space measurements and some ½ space. There is quite a difference ... 6 dB - see attached link.
I have not looked at the curves of the other speakers-but will comment that it is VERY COMMON to use LOTS of smoothing to present a freq or impedance graph.
When you use smoothing-the peaks and dips get "rounded off" so they do not appear as high or as low as they actually are.
So it depends on where the numbers come from-a smoothed response or the actual response.
And when "smoothing" there are several different methods and sometimes both are used-so it is hard to know what is exactly going on.
First is the number of data points used. If a data point is not present at the particular dip-then that dip simply does not show up on the plot-even if it is actually there.
Then there is the smoothing of the overall plot.
A real loudspeaker has lots of ripples. Yes we have to use smoothing to make it make sense-but how much and where is the real question-and different people have different answers.
I have a problem with any measurement with subs in "whole space". yes it may be a "measurement", but where in real life is that ever going to take place?
An before you answer "flown subs", consider this. Where is the AUDIENCE?
??
In most cases they are on the ground or on a large surface. Most people do not listen to a sound system when they are standing on a 30' ladder.
So with the audience on the ground-THEY are in a half space situation-so the half space loading IS in effect-even if the subs are flow.
It does not matter if the subs are flown or not-you only get the half space loading ONCE.
So I don't see any condition in real life that a whole space measurement would be used.
Now I am talking about subs for this-because the wavelengths are so long that the space the average person is away from the boundary really doesn't matter.
With full range speakers it is a different situation in that the wavelengths are much shorter (at the higher freq) so a whole space condition is a good starting point.
If you put a full range speaker in a half space condition then the response will get '[skewed" by the lower freq having additional loading of the boundary, but the higher freq or those controlled by the horns will not get any additional level. As evidenced in the Peavey link you provided-the higher freq don't change with the loading.
Of course it also depends on the freq involved-the spacing from the boundary etc as to how much additional (or cancellation) is going on.
Once again-it is not always as simple as it looks or as people would like it to be