Why it matters when you are writing
What counts as primary depends on the question you are asking, which is why the category confuses people. A 1954 newspaper editorial is a secondary source about the event it reports and a primary source about mid-century editorial opinion. A review article is secondary evidence about the trials it summarizes and primary evidence about how the field understood those trials in the year it was written. Ask what your sentence is claiming, then ask whether the source witnessed that thing.
In empirical fields the practical version is simpler: the paper that first reports a study is the primary source for that study’s findings. Everything downstream — a textbook, a review, a press release, a systematic review’s summary table — is secondary, no matter how authoritative. Citing the review for a specific effect size passes along whatever transcription or interpretation error the review introduced, and reviewers do check.
Historians and qualitative researchers work with primary material constantly, and the standard there is provenance. Where did the document come from, who produced it, for whom, and what has happened to it since. A transcription in an edited collection is one remove from the manuscript, and saying so in a footnote is not pedantry — it tells a reader how much weight your reading can bear.
