The "scientific method" -- which I commented on -- needs to formulate hypotheses, test them and so on. That can't reliably be done in any mostly historical science, by definition.
(And of course you can randomly test some things in historical sciences too. E.g. finding historical trading patterns by looking at materials used for artefacts, diet by looking at enamel of skeletons, etc etc. Other things won't find any possible test solution, ever.)
