The understanding of the glass trade in the first millennium CE relies on the characterisation of well-dated compositional groups and the identification of their primary production sites. 275 Byzantine glass weights from the British Museum and the Bibliothèque nationale de France dating to the sixth and seventh century were analysed by LA-ICP-MS. Multivariate statistical and graphical data analysis discriminated between six main primary glass types. Primary glass sources were differentiated based on multi-dimensional comparison of silica-derived elements (MgO, Al2O3, CaO, TiO2, Fe2O3, ZrO2) and components associated with the alkali source (Li2O, B2O3). Along with Egyptian and Levantine origins of the glassmaking sands, variations in the natron source possibly point to the exploitation of two different natron deposits. Differences in strontium to calcium ratios revealed variations in the carbonate fractions in the sand. At least two cobalt sources were employed as colouring agents, one of which shows strong correlations with nickel, indicating a specific post-Roman cobalt source. Typological evidence identified chronological developments in the use of the different glass groups. Throughout the sixth century, Byzantine glass weights were predominately produced from two glasses that are probably of an Egyptian origin (Foy-2 and Foy-2 high Fe). Towards the second half of the sixth century a new but related plant-ash glass type emerged (Magby). Levantine I was likewise found among the late sixth- to early seventh-century samples. The use of different dies for the same batch testifies to large-scale, centralised production of the weights, while the same die used for different primary production groups demonstrates the co-existence of alternative sources of supply. Given the comprehensive design of our study, these results can be extrapolated to the wider early Byzantine glass industry and its changes at large.