Repeatable applies to experimentation. It does not apply to a good deal of what we might call observational evidence and evidence obtained by interpretation, reasoning and fair extrapolation. If I find a dinosaur femur in the cliff over the beach, it may be subjected to dating methods, interpretation of the context of its site, comparative anatomy and its context within the geographical area (etcetera - I'm no Stu). I may never find another specimen, so it's not "repeatable", but it is evidence and, given correct application of the scientific process to it, it's perfectly good scientific evidence.