Is there an argument, apart from experiments, that we know this is true?
And if we only know it by experiment, how do we know the experiments are precise enough to conclude this?
Stated differently, what argument is there that prevents us from replacing c in the Lorenz transformations by $c' = c + \epsilon$