摘要:Context. Reconstructions of past irradiance variations require suitable data on solar activity. The longest direct proxy is the sunspot number, and it has been most widely employed for this purpose. These data, however, only provide information on the surface magnetic field emerging in sunspots, while a suitable proxy of the evolution of the bright magnetic features, specifically faculae/plage and network, is missing. This information can potentially be extracted from the historical full-disc observations in the Ca II K line.Aims. We use several historical archives of full-disc Ca II K observations to derive plage areas over more than a century. Employment of different datasets allows the identification of systematic effects in the images, such as changes in instruments and procedures, as well as an assessment of the uncertainties in the results.Methods. We have analysed over 100 000 historical images from eight digitised photographic archives of the Arcetri, Kodaikanal, McMath-Hulbert, Meudon, Mitaka, Mt Wilson, Schauinsland, and Wendelstein observatories, and one archive of modern observations from the Rome/PSPT. The analysed data cover the period 1893–2018. We first performed careful photometric calibration and compensation for the centre-to-limb variation, and then segmented the images to identify plage regions. This has been consistently applied to both historical and modern observations.Results. The plage series derived from different archives are generally in good agreement with each other. However, there are also clear deviations that most likely hint at intrinsic differences in the data and their digitisation. We showed that accurate image processing significantly reduces errors in the plage area estimates. Accurate photometric calibration also allows precise plage identification on images from different archives without the need to arbitrarily adjust the segmentation parameters. Finally, by comparing the plage area series from the various records, we found the conversion laws between them. This allowed us to produce a preliminary composite of the plage areas obtained from all the datasets studied here. This is a first step towards an accurate assessment of the long-term variation of plage regions.