One mistake that had the hugest impact in physics come from a very simple phenomenon in appearance : the reflection of light on a mirror. That said the error may be due precisely to this impression of simplicity…
In early optical experiments, at the end of the nineteenth century, we postulated (without explicitly saying it, as it was so obvious (!!)) that the light was reflected by mirrors at the same speed it arrived… As if it was a simple « ball bouncing on a wall » in classical mechanics. All experiments at that time have be made on this postulate, and therefore all conclusion and current theory have been build upon it… It’s a little scarring if we become aware that this postulate is probably false, and that for several reason :
First the light has nothing to do with a ball made of matter, it has no mass, it’s just energy and therefore it obeys to none of the effects of inertia, main responsible of the conservation of speeds in rebound of objects with mass at our scale.
Then the apparently simple reflexion of the light on a mirror is in fact a relatively complex process, a sort of « reaction » with the material. The light, composed of electromagnetic wave, will not always be reflected, depending on the nature of the material it touches it can also be absorbed, thus « disappear » (in fact turn into another form of energy) or refracted. In short light « interferes » with the atoms of the material it encounters in a totally different way than a ball against a wall.
The postulate of the conservation of speed in the reflexion of light is then probably false, and should be replaced by this one : whatever is the speed of light when it attain a mirror, it will interact with its atoms in the same way, and will be re-emitted at a speed of light « relatively » to the mirror itself.
Note that the frequency, quantifying the energy of an electromagnetic wave, is preserved since the « retransmission » is made at the same « rhythm » of the incident light, therefore the energy is preserved also. And since the light has no mass, this speed reduction (or increase if the light was « relatively » less quickly than the object) does not cause either variation of energy.
Here we are, we always considered that light « keeps » its incident speed « bouncing » on a mirror but it is almost certainly not the case and this error had important consequences because after some important experiences » we postulated nothing less than « the speed of light C was always the same, whether emitted by an object in motion or at rest »… Yet we understand why as the assumption is « half true » : the light is effectively always ré-emitted (or produced) at the speed of light C in the reference frame of the emitter (which is in motion or at rest), but the speed of light is certainly not the same in ALL reference frame.