Is there any methodological characteristic universal in Physics?
Even if some branches of Physics lose their reproducibility, their experimental testing, their deterministic predictivity isn't some core preserved? Isn't a stochastic, probabilistic, uncertain prediction still a prediction?
Being a phenomenological science that builds/develops its axiomatic systems inductively from observation.
Being a science whose conclusions follow deductively from the axiomatic systems inductively built.
Being a science whose conclusions explain and predict and don't simply record phenomena.
Being a science whose observer is made methodologically objective. As long as you do Physics the way it is prescribed and you observe exactly what is observed by any arbitrary person who is also following the rules you are absolutely guaranteed to reach the same conclusions with them. Even if the conclusion is not precise but a disperse probability distribution.