I will repeat this: the entire band of distortion products from 200 Hz to 4.7 kHz in Fig. Z DO NOT LIVE AT THOSE FREQUENCIES, they exist because of IM between two other frequencies far removed from that locale, and any local deviation in FR is not tied directly into their levels! Remember, we are looking at the current distortion IN THE CABLES, not anything else at this point.
I will now go over exactly how the distortion manifests.
How can the distortion in the speaker cables manifest to our ears? The woofer only has significant acoustic output up to about 9 or 10 kHz, while the tweeter only has significant acoustic output down to about 1 kHz. Any distortion that would be audible could only come from output below the 9 or 10 kHz that the woofer can have output at, and above 1 kHz for the tweeter.
The crossover IS rolling off the drivers output, there is no doubt about this. What may get overlooked, is that any distortion reduction can only come between the acoustical roll off and the crossover roll offs for each driver. There is indeed a limited range where the distortion will manifest for each driver.
For the tweeter of this particular speaker system, under these conditions, that range is from 1 kHz to about 6 kHz, and for the woofer, that range is 6 kHz down to about 3 kHz, with the crossover at about 3 kHz for this system. At 1 kHz, the crossover voltage tweeter output will be down approximately 20 dB, and approximately 10 dB at 3 kHz. For the woofer, it is down 10 dB at 3 kHz, and 20 dB at 6 kHz.

The crosover roll off itself may be considered as part of the reduction, but this roll off is applied to both the signal coming from the tweeter bi-wire speaker cable AND to the single speaker cable, so that any advantage this represents will be present for both cables equally. So the amount of reduction in distortion is maintained, despite the roll off inherent in the crossover.
If all the various factors are taken into consideration, as I have explained them, the amount of distortion reduction could be said to be about 15 dB in a region from about 2 kHz to about 5 kHz for the tweeter, and for the woofer, a difference of about 3 to 5 dB in a region from 3 to 6 kHz or so. There may be lesser reductions outside these bands, and this is not taking into account other factors which I feel might make these levels, regions and amounts more significant than they seem at first glance.
So the actual amount of distortion reduction is not actually 20 dB, even though there were regions on Fig. Z that had 25 dB differences at frequencies that might have been relevant with a casual analysis. It is more like a range from about 15 dB to 3 dB over a frequency band from 2 kHz to 6 kHz in terms of what would possibly be audible.
Note that these levels and range is just for this particular test signal, some of the other Phi Spectral test signals gave different results, being based on different frequency mixes. The Phi Tri-Band for instance gave a wider frequency range, while the Phi 12 Revised gave a few isolated level differences that were at the 15 dB point or more, all things taken into consideration as above.

Tech Details; Drive level for Fig. Z was 2 VRMS into the 4 Ohm nominal load the speaker system presents, so that the RMS level of the signal was at 1W, due to the crest factor, the signal peaks would be hitting 80W or slightly more. 0 dB on the graph equals 1 ampere. The noise floor of this particular measurement is approx. -90 dB.
Repeatability of the measurements made using the current probe, the 16 bit FFT and the entire measurement chain were fairly consistent, with up to a maximum of 1 to 2 dB of variation in the higher frequencies, and up to 2-3 dB of level variations in the lower frequencies. Frequencies below the 100 Hz test band tended to vary more due to the FFT bin size and LF hum and noise.

End

All applicable copyright laws apply, all rights reserved, except transmission by USENET and like facilities granted.  Any use or inclusion in print or other media are specifically prohibited.  The informational content is not warrantied in any way or form, and any use of said content are at the reader's own risk, the author shall not be held responsible in any way for any damages or injuries arising from the content of this web site.  Common safety practices are encouraged at all times.  Do not fold, spindle or mutilate.