REVIEW // Painting by Numbers: Data-Driven Histories of Nineteenth-Century Art

Eliza Goodpasture

Painting by Numbers: Data-Driven Histories of Nineteenth-Century Art
Diana Seave Greenwald
Princeton and Oxford: Princeton University Press, 2021. Pp.256. £28 (hardcover)

Diana Seave Greenwald’s Painting by Numbers: Data-Driven Histories of Nineteenth-Century Art takes a striking approach to the study of art history. She uses extensive nineteenth-century exhibition records from the Paris Salon, a group of American museums and institutions, and the British Royal Academy Summer Exhibitions, to statistically analyse a large pre-existing dataset on paintings exhibited in the nineteenth century. ‘Through the lens of social scientific sampling,’ she writes, ‘the canon can be analysed as a potentially biased sample…. Data can chart how systematic behaviours of actors in the art world have influenced the inherited art historical canon.’[1] Greenwald’s assessment of the prevalence of sample bias within the art historical canon provides a new terminology for critiques of the canon that go back to feminist art historians in the 1970s, such as Linda Nochlin and Griselda Pollock. Greenwald’s goal of using quantitative data to analyse all the works that were created in a given context, rather than those that became famous, is a productive one. Though she is not the first scholar to put quantitative methods to use in art historical investigation, she is the first to focus on the nineteenth century and to utilise the existing data sets of academy exhibitions, focusing her analysis on works of art, rather than a larger social context. Her work is an innovative and welcome approach to art historical research, though it does not always do justice to its methodology.

The book is divided into two sections. The first two chapters explain Greenwald’s methods of analysis in straightforward terms for the reader. She follows this with three case studies – one focusing on France, one on America, and one on Britain. Greenwald makes the point repeatedly throughout her book that her ‘data-driven’ approach to art does not seek to replace or circumvent traditional approaches to art history, but rather to add to them. Much of her analysis offers fascinating new insights and ways to reframe common assumptions about the art historical canon. For example, in her first case study on the French Salon, Greenwald analyses the relationship between paintings of rural France and the cost and ease of transport from Paris to the countryside. She concludes that ‘a decrease in the cost of travel from Paris is correlated with greater amounts of rural genre painting,’ and that ‘agricultural employment share is positively correlated with an increase in the amount of rural genre painting displayed.’[2] These conclusions stand in contrast to the usual assumptions in art historical literature, which has traditionally seen the increase in rural genre paintings in nineteenth-century France as a reaction to urbanisation and the decline of agricultural economies. Greenwald states that this ‘nuts-and-bolts finding’ that a ‘relatively easy commute’ between city and country drove increases in rural genre painting confirms the belief held by many economic historians that ‘often, modernization had its greatest effect on people simply by making life more convenient.’[3] This case study is Greenwald’s strongest, and it not only offers the reader a statistical analysis of exhibition records of the Paris Salon but also incorporates other economic data and aspects of socio-historical research, including an analysis of several hundred letters written by painter Jean-François Millet. Drawing from such an array of information, Greenwald’s conclusions are both intriguing and lay the groundwork for much further investigation.

However, her readiness to draw firm conclusions based entirely on quantitative analysis occasionally feels premature. In her third case study, using exhibition catalogues from the Royal Academy’s Summer Exhibitions between 1796 and 1914 as her dataset, Greenwald inventories the number of depictions of different countries within the British Empire with the intention of drawing conclusions about the ways in which the Empire was visualised on the walls of the Royal Academy. She bases her argument on an analysis of paintings that specifically include geographical locations in their title, which she then categorises by country and analyses quantitatively. I found Greenwald’s basic principle of only including artworks whose titles include placenames remarkably narrow. As Greenwald herself recognises, within such a rubric many paintings do not register as images of Empire, yet their subjects clearly bring up issues relating to colonisation. For example, as a work of portraiture, Johann Zoffany’s The Family of Sir William Young does not overtly reference a specific British territory but, to no less an end, pictures a family made wealthy through colonial connections stood alongside their Caribbean servant. It is precisely this sort of nuance – that acknowledges the different forms of visually representing Empire – that Greenwald’s methods lack, and which ultimately shrinks her data set to the approximately twelve percent of paintings shown in the Summer Exhibitions that include placenames in their titles. This, unfortunately, leads the author to somewhat suspect conclusions. For example, Greenwald claims that only about five percent of paintings in the Summer Exhibitions between 1796 and 1914 were representations of Britain. However, surely many more paintings pictured subjects set in Britain, despite the country simply not being named in the title?[4] Victorian genre paintings, for example, tended to have narrative titles, yet obviously depicted modern British life. Furthermore, many British landscape paintings of the period had generic pastoral titles that did not indicate the exact location of their subject, even though the setting would have been recognised by contemporary viewers.

Though I found myself unaccustomed to taking a data-driven approach to art, I agree with the author that her methodology, combined with traditional qualitative methods, is an innovative way to broaden our understanding of the canon and the art historical record. Greenwald points out that despite ‘critiques of the concept of genius and of the canon, little systematic effort has been made to define what is not genius as embodied by the artworks that have not survived or have faded into such obscurity that no images of them are readily available.’[5] This is an excellent observation, and raises exciting questions about art that has been lost or forgotten, and how we might use the tools Greenwald offers to recover its place in art history. Lamentably, however, Greenwald rarely achieves this goal of recovery. The case-studies within Painting by Numbers centre on established artists – Edgar Degas, Jean-François Millet, Johann Zoffany, Thomas Daniell, Lilly Martin Spencer, and John Constable, for example – and include many extant works. It is, needless to say, challenging to study works that no longer exist, and though Greenwald’s data analysis is able to partially recuperate artworks that we no longer have access to, she has not found a way to bring these into deeper discussion. I look forward to the ways in which Greenwald, and other scholars building on her work, continue to take pioneering data-driven approaches to the history of art in the future that push the questions Greenwald has raised even further.


  1. Diana Seave Greenwald, Painting by Numbers: Data-Driven Histories of Nineteenth-Century Art (Princeton: Princeton University Press, 2021), 7.
  2. Greenwald, 68.
  3. Greenwald, 82
  4. Greenwald, 146.
  5. Greenwald, 154.