This essay describes a novel dataset that facilitates the quantitative analysis of eighteenth and nineteenth-century French painting. Based on titles listed in the Paris Salon livrets, the dataset assigns detailed keywords indicating the content for each of the more than 148,000 paintings shown at the Salon—the principal French art exhibition of the era—from the seventeenth to nineteenth century. To demonstrate the interest and utility of this dataset, we present a case study about a genre that has traditionally been neglected by both art historians and cultural economists: portraiture. Our analysis shows portraiture was ubiquitous, usually representing 27% of all paintings exhibited in a year—more than any other genre. We also trace the changing demographics of sitters. There were, for example, dramatic increases over time in how many images of women were displayed. We also chart the rise of quasi-anonymous portraiture, where names of sitters do not appear in paintings’ titles but audiences from certain social classes could identify subjects. We ultimately demonstrate how quantitative methods can be fruitfully applied to this art historical dataset, which is now available freely online, and is just one of many similar datasets that can be digitized and studied.

本文描述了一个有助于对 18 世纪和 19 世纪法国绘画进行定量分析的新数据集。根据巴黎沙龙livrets 中列出的标题，该数据集指定了详细的关键字，指示从 17 世纪到 19 世纪在沙龙（该时代的主要法国艺术展览）上展示的 148,000 多幅画作中每一幅的内容。为了展示这个数据集的兴趣和实用性，我们提出了一个关于传统上被艺术史学家和文化经济学家忽视的流派的案例研究：肖像画。我们的分析表明，肖像画无处不在，通常占一年展出的所有绘画的 27%——超过任何其他类型。我们还追踪保姆不断变化的人口统计数据。例如，随着时间的推移，展示的女性图像数量急剧增加。我们还绘制了准匿名肖像画的兴起，绘画的标题中没有出现坐者的名字，但来自某些社会阶层的观众可以识别主题。我们最终展示了如何将定量方法有效地应用于这个艺术历史数据集，该数据集现在可以在线免费获得，并且只是许多可以数字化和研究的类似数据集之一。