Draught cattle, used for ploughing and carting, contributed to drive social transformations in prehistoric societies by replacing or complementing human power. However, identifying draught cattle from archaeological sites has proven challenging due to the dearth of direct evidence. This paper presents a biometric approach to identifying draught cattle in archaeological assemblages based on metapodials, and first and second phalanges. The analysis includes over 150 modern non-draught cattle encompassing various species and breeds, along with a smaller group of draught cattle. Statistical tests and multivariate analyses were first adopted, revealing distinct morphological differences between the two groups, which do not appear to be highly correlated with age. Although cattle limb bone morphology could vary between breeds, the principal component analysis suggests traction is the predominant factor distinguishing between modern draught and non-draught cattle. Biometric data from metapodials and phalanges were then applied to construct five predictive logistic regression models, with the first phalanges showing the highest balanced accuracy for separating the draught and non-draught groups, a clear advance from Lin et al. (2016)'s pioneering work. The predictive models were then applied to data from two British sites to demonstrate the applicability of the proposed approach to archaeological assemblages. The results show that this biometric approach has the potential to significantly enhance our understanding of draught cattle exploitation in the past.

中文翻译：

---

寻找役牛：一种鉴定方法


用于犁地和马车的役牛通过取代或补充人类力量，为推动史前社会的社会变革做出了贡献。然而，由于缺乏直接证据，从考古遗址中识别役牛已被证明具有挑战性。本文提出了一种生物识别方法，用于识别基于后足、第一和第二指骨的考古组合中的役牛。该分析包括 150 多头现代非吃奶牛，包括各种物种和品种，以及一小部分吃奶牛。首先采用统计检验和多变量分析，揭示了两组之间明显的形态学差异，这些差异似乎与年龄不高度相关。尽管牛肢体骨骼形态可能因品种而异，但主成分分析表明牵引是区分现代吃水牛和非吃水牛的主要因素。然后应用来自掌足和指骨的生物特征数据来构建五个预测性逻辑回归模型，其中第一个指骨在区分吃水组和非吃水组方面显示出最高的平衡准确性，这是 Lin 等人（2016 年）开创性工作的明显进步。然后将预测模型应用于来自两个英国遗址的数据，以证明所提出的方法对考古组合的适用性。结果表明，这种生物识别方法有可能显着增强我们过去对役牛剥削的理解。