FROM ALLOY COMPOSITION TO ALLOYING PRACTICE: CHINESE BRONZES* A. M. POLLARD, R. LIU†, J. RAWSON and X. TANG School of Archaeology, University of Oxford, 36 Beaumont Street, Oxford OX1 2PG, UK We propose a new methodology based on standard statistical processes for displaying and rigorously comparing the alloy composition of archaeological bronze alloys. Although traditional approaches using visual comparisons of histograms of alloying elements in an assemblage of archaeological objects are adequate for observing differences between these distributions, we argue that differences in sample size cannot be adequately accounted for without using a statistical approach. We demonstrate this methodology by comparing the alloy composition of bronzes from the sequence of Bronze Age cultures in Central China—Erlitou, Erligang (Zhengzhou, or early Shang), Anyang (late Shang) and Western Zhou. We suggest that this approach allows the identification and rigorous comparison of ‘regional alloying practices’, which in turn enables us to link the alloy composition of the objects with the intentions and skills of foundry workers. KEYWORDS: ALLOYS, ALLOYING PRACTICE, CUMULATIVE FREQUENCY ANALYSIS, KOLMOGOROV–SMIRNOV TEST, BRONZE AGE CHINA INTRODUCTION Although the term ‘Bronze Age’ means different things in different parts of the world, and covers different time periods, one of the common features in Eurasia is the emergence of the practice of alloying copper with other metals some time during this period. A common sequence is the use of unalloyed copper, followed by arsenical copper and finally tin bronze (usually initially without lead, but later with; Tylecote 1992, 1–2, 18–44). Traditional explanations for this have tended to focus on either technological ‘improvement’ (i.e., the trend towards ‘better’ alloys) or geolog- ical determinism (the sequence of oxide and sulphide ores in an idealized ore deposit; e.g., Pernicka 2014). However, in many parts of Eurasia these technologies are not necessarily sequential or exclusive, although the widespread use of tin bronze does not usually dominate until late in the Bronze Age, apart from in China. The exact timing of the sequence in different parts of Eurasia is also often an open question (see, e.g., in China, Liu et al. 2015; Mei et al. 2015). The sequence of alloy usage in a specific region is usually determined simply by looking at the ma- jor element data in the objects, often displayed by plotting histograms of the tin or arsenic contents in these objects. Although this visual method is generally adequate for deciding the dominant alloying practice at a specific time and place (allowing for considerations of sample bias), it becomes less ideal for comparing the practices at different times or places. Visual examination of two or more histo- grams can give a reasonable impression of similarities or differences, but this is not adequate for rig- orous comparisons, which require statistical testing, primarily because the assemblages of objects to be compared typically contain widely differing numbers of samples. This paper proposes a new meth- odology for characterizing and statistically comparing alloying practices across time and space, using as a case study the development of leaded tin bronze alloys in China. As noted above, the Chinese use *Received 22 December 2017; accepted 17 May 2018 †Corresponding author: email ruiliang.liu@arch.ox.ac.uk Archaeometry ••, •• (2018) ••–•• doi: 10.1111/arcm.12415 © 2018 University of Oxford bs_bs_banner