Context. In the context of structure formation and galaxy evolution, the contribution of magnetic fields is not well understood. Feedback processes originating from active galactic nucleus (AGN) activity and star formation can be actively influenced by magnetic fields, depending on their strength and morphology. One of the best tracers of magnetic fields is polarised radio emission. Tracing this emission over a broad redshift range therefore allows an investigation of these fields and their evolution. Aims. We aim to study the nature of the faint, polarised radio source population whose source composition and redshift dependence contain information about the strength, morphology, and evolution of magnetic fields over cosmic timescales. Methods. We use a 15-pointing radio continuum L -band mosaic of the Lockman Hole, observed in full polarisation, generated from archival data of the Westerbork Synthesis Radio Telescope. The data were analysed using the rotation measure synthesis technique. We achieved a noise of 7 μJy beam −1 in polarised intensity, with a resolution of 15″. Using infrared and optical images and source catalogues, we were able to cross-identify and determine redshifts for one-third of our detected polarised sources. Results. We detected 150 polarised sources, most of which are weakly polarised with a mean fractional polarisation of 5.4%. No source was found with a fractional polarisation higher than 21%. With a total area of 6.5 deg 2 and a detection threshold of 6.25 σ , we find 23 polarised sources per deg 2 . Based on our multi-wavelength analysis, we find that our sample consists of AGN only. We find a discrepancy between archival number counts and those present in our data, which we attribute to sample variance (i.e. large-scale structures). Considering the absolute radio luminosity, we find a general trend of increased probability of detecting weak sources at low redshift and strong sources at high redshift. We attribute this trend to a selection bias. Further, we find an anti-correlation between fractional polarisation and redshift for our strong-source sample at z  ≥ 0.6. Conclusions. A decrease in the fractional polarisation of strong sources with increasing redshift cannot be explained by a constant magnetic field and electron density over cosmic scales; however, the changing properties of cluster environments over cosmic time may play an important role. Disentangling these two effects requires deeper and wider polarisation observations as well as better models of the morphology and strength of cosmic magnetic fields.