[old_tv_nut]

As I probably explained in another thread, the transmitter side Q filters were really not properly specified by the NTSC/FCC. To guarantee good results, Q baseband should have a notch at 900 kHz, so that in the receiver, when the upper sideband is killed by the receiver 4.5 MHz sound trap, the lower Q sideband also has no energy at 900 kHz below chroma and there would be no quadrature distortion, which was the aim of I/Q modulation in the first place. So, when I/Q demodulation was tried on some later sets with larger screens, the quadrature distortion could be quite visible, and results depended on the Q response of the particular color encoder in use at the studio. So, I/Q was a great idea, but suffered from under-specified filter standards. I believe none of the experiments that led to the choice of I/Q actually used an end-to-end system including the RF/IF parts. Later, when the whole system was being tested, there were a million problems to fix, and if the quadrature distortion was visible (say on a large-screen trinescope), it would have been easy to ignore as possibly a small receiver issue that could be fixed later.

Edit: Also, probably posted by me elsewhere, I did an extensive series of tests on I/Q demodulation when I worked for Zenith in the 80s, and even using baseband I/Q with no quadrature distortion, there were some effects that people weren't used to, like the vertical strokes of yellow movie titles turning orangish because the I signal was full amplitude for narrow objects while the Q signal was smeared and reduced in amplitude.