A finding isn’t settled until someone else gets the same answer. Authors, journals and funders can post a replication call for one specific claim. Labs register a plan before they start, and every attempt, consistent or not, is recorded on the original paper.
The registration is what makes the result usable. A lab says in advance which claim it is testing, what counts as a match, and what it will report either way, so a null result cannot be read as a failed experiment or quietly dropped. The original paper then carries both outcomes, and anyone citing it can see how many groups tried and what they found.
Credit is the part that makes labs willing to do it. Repeating someone else's experiment has traditionally earned nothing, which is why so little of it happens. A registered attempt is recorded as the lab's own contribution, named and dated, whether it confirmed the result or not, so the work counts toward a public record rather than disappearing into a footnote.
Example records. The entries below illustrate the format. No replication requests are open yet.