Except for its bulk, a coax line with a (mainly) air dielectric using a separate outer tube, with the inner supported inside it, could be designed &/or "tweaked" to be a very precise 50
, & would be unlikely to change to any significant degree over time.
The larger diameter 3
1/8inch, or the like in air dielectric coax lines would probably be the easiest to configure, but, at the same time the bulkiest.
As an aside, back in the day Marconi preferred 51.5
coax & the 1959 vintage TV transmitter setup at the ABW2 site, duly used that coax standard for the 3
1/8inch rigid cable sections involved in all the "plumbing" involved in getting the TV signal "up the spout".
These early installations used discrete Vision & Sound transmitters, the outputs of which were combined in "filterplexers", which as well as combining V & S, carried out a respectable amount of the sideband shaping of the VSB signal.
This particular installation used 2 Vision & Sound Transmitter pairs, with the output of the appropriate filterplexers being combined in a further device called a combining unit.
For the colour modifications in 1974, all of the equipment between the transmitters & the antenna feeder had to be replaced by tighter spec stuff, which was duly supplied by Amalgamated Wireless Australasia Ltd.
All the new stuff was 50
, so AWA also supplied, at great expense, a number of 1/4 wavelength impedance matching sections sections of 50.7445
.
Of course, being curious, we opened one up, & found that the weird impedance was achieved simply by turning the centre conductor down in a lathe!