Geoffrey Bowker’s Memory Practices in the Sciences argues that science depends not only on discovery, experiment or theory, but on organised practices of remembering. Scientific knowledge presents itself as timeless, objective and universally valid, yet Bowker shows that this “eternal present” is produced through archives, classifications, standards, records and databases. The central issue is memory practice: the social and technical work through which traces of the past are stored, ordered, forgotten, recovered and made usable. Bowker rejects the idea that memory is simply conscious recollection. Instead, memory operates through institutions, procedures, environments and technologies, from written records and geological strata to computer databases. His examples show that memory is distributed: prisons remember through rules, landscapes remember through paths and traces, and science remembers through time standards, archives and classification systems. This matters because scientific facts often appear detached from history, as if they were true everywhere and always, but they require extensive infrastructures to become stable. Bowker’s discussion of Lyell’s geology is especially revealing: the earth itself is treated as an archive, yet scientists interpret that archive through the record-keeping habits of their own institutions. Memory is therefore never neutral. It selects, formats and excludes; it creates continuity while concealing discontinuity. In contemporary science, the digital database becomes a new archival form, promising total recall while also casting certain spaces, organisms, practices and histories into oblivion. Bowker’s argument ultimately reframes knowledge as an archival achievement. Science does not merely discover the past; it constructs the conditions under which the past can be remembered, synchronised and made authoritative. The archive is thus not a passive storehouse, but an active machinery of knowledge, power and forgetting.