A spark-gap transmitter is essentially "white noise" being radiated into space, the earliest versions being used for CW morse-code telegraph-keyed modulation.
For RF, this was usually tuned, or resonated to a transmitting antenna, for greater efficiency.
Back in those days, interference with other stations or radio-frequencies was non-applicable, as the telegraph office / telecoms / post office was the sole source.
Nowadays, however, RF bandwidths are narrower, closely packed, hence the strict control of 'willy-nilly' random transmitters.
While on the spark-gap topic, an interesting experiment I tried out of curiosity:
My 180W arc welder has an output no-load voltage of 48V AC, at 50Hz (mains frequency).
As welding, especially gouging, at about 5mm distance, creates a plasma arc, I wondered if this would cause 'spark-gap' interference on my LW/SW/MW/FM world scanner radio (portable).
Using the counter/timer on my mobile phone, I recorded the earphone output from radio to my PC's soundcard input as MP3.
Each 'weld' or 'arc' test was done at fixed times, to differentiate from random RF noise.
To my surprise, not the entire AM spectrum is affected, rather 'grouping' around areas such as 6MHz, 12MHz, etc.
Longwave was swamped, and MW quite noisy, so maybe the lower 100-1600KHz frequencies are strongest.
FM was not affected, except when tranny is very close to arc.
At about 10-15m, the 'Arc' hum tapers off.
ADD: The frequency spectrum is affected by the length of the arc, so can be considered a 'white noise' wideband radiated source.