My favorite linguistic/sound story is this:
In 1878, a student called Ferdinand de Saussure proposed that PIE had only one vowel sound "e", and that all other vowel sounds in PIE languages came that single sound. How? He proposed that all other vowels came through a consonant "colouring" the original vowel "e". For example, the usage of e + w in words gave birth to the sound "u".
But this doesn't explain root words like "pas" that don't contain the core vowel "e" or any evidence of a rogue sound like "w" colouring it.
Here I will quote from the book:
"Saussure conjured up a daring hypothesis. Once upon a time, he argued, verbs like pas did indeed have a rogue sound after the core vowel e, and this sound was responsible for 'colouring' the vowel from e to a. But at a later stage, and after it had wreaked all this havoc, the rogue sound disappeared from the scene. If one were to use the symbol X to represent
this elusive rogue sound, then Saussure's claim was that pas originated as a regular root peXs, but that some time in prehistory X coloured the vowel e into an a, so peXs became paXs. Much later on (but still well before the earliest records), the rogue X was itself worn away because of [some other changes]. Schematically, then, the whole development can be represented as peXs -> paXs -> pas."
This remained a cool theory, outside mainstream linguistics. Then after Saussure died, archaeologists dug up some tablets and the oldest PIE language Hittite was deciphered. And there appearing in the very place where he had predicted it to be, the rogue consonant was found written down in clay.
Source: around page 109 from "The Unfolding of Language" by Guy Deutscher.