Historians and sociologists of psychiatry often claim that psychiatric diagnoses are discontinuous. That is, a particular diagnoses will be described in one way in one era and described quite differently in a different era. Historians and sociologists often draw epistemic consequences from such discontinuities, claiming that truth is pluralistic, provisional and historicised. These arguments do not readily fit in with how analytical philosophers of science approach scientific realism. I show how the pessimistic meta induction does not capture the point which historians and sociologists are making but scientific perspectivism seems to capture their point much better. I then highlight conceptual innovations which scientific perspectivists add. They demarcate between truth and objective reality, they specify which type of truth they endorse and they put down constraints on possible truths. This blocks an anything goes relativism which historians and sociologists can be in danger of falling into. I highlight my argument by discussing a discontinuous episode in the history of autism. I discuss three aspects of this discontinuity and show how scientific perspectivism can portray each aspect as non-trivially true. My argument shows that we can be scientific realists about autism even if we can formulate notions of autism in quite different ways.