Analogue TV encoders took a signal with a dynamic range of "black to white" (e.g. 0V to 5V) and, before modulating it onto a carrier frequency, first compressed the signal upwards, making it into a smaller range of e.g. 1V to 5V, making it into a signal that's actually more like "dark grey to white." Analogue TV decoders did the opposite, treating 1V as the "black level", stretching the 4V "dynamic range" back into the full black-to-white intensity range.
Doing this allowed "true 0V" (i.e. gaps cut out of the modulated signal by electronic post-processing, below the "minimum" intensity of 1V, where you're just getting carrier) to act as a form of in-band signalling. 0V was used "outside" the image to both synchronize the TV's raster electronics, and inform said electronics of the period of time they should be off altogether, rather than firing even a weak electron beam. (This most noticably prevents your CRT from drawing a subtle diagonal swoosh when it starts each new frame.)
Doing this allowed "true 0V" (i.e. gaps cut out of the modulated signal by electronic post-processing, below the "minimum" intensity of 1V, where you're just getting carrier) to act as a form of in-band signalling. 0V was used "outside" the image to both synchronize the TV's raster electronics, and inform said electronics of the period of time they should be off altogether, rather than firing even a weak electron beam. (This most noticably prevents your CRT from drawing a subtle diagonal swoosh when it starts each new frame.)
See also: the terms "front porch" and "back porch" (in e.g. https://electronics.stackexchange.com/questions/201011/what-...), referring to the parts of each scanline that are 0V.