”Repertoire” was a keyword at the Comédie-Française Registers Project Conference in Cambridge, MA. Spurred by a number of papers (including those by Jan Clarke, Bill Weber, and Pannill Camp), I developed four attempts to visualize the stability and variety of the troupe’s repertoire from 1680 to the Revolution. As a newcomer to the Comédie-Française and to eighteenth-century French theater generally, I offer these charts with some trepidation. Consider them experiments in seeing how repertoire evolves, rather than answers specific questions about the Comédie-Française and its repertoire. The latter requires longer study and more expertise than one weekend and my limited knowledge can offer.
First, two glimpses of how the company develops repertoire. Figure 1 shows each decade’s repertoire, segmented by the decade in which plays premiered at the Comédie-Française. The columns move from left to right by decade, from the 1680s to the 1790s. The y-axis records the percent of the season’s repertoire from a given decade. In the first column, the 1680s, 100% of the repertoire “premiered” in the 1680s. (This is true by definition, because the Comédie-Française itself was founded in 1680, and thus any work performed that decade premiered that same decade. With slightly more data, one could remake this chart to account for the many works that actually premiered prior to the Comédie-Française’s founding. But for this analysis, the definition is axiomatic.) In the second column, the 1690s, approximately 50% of the company’s repertoire was from the preceding decade (the grey segment) and 50% premiered in the 1690s (the blue segment). During the 1700s, the percent of the repertoire from the 1680s remained stable, but the 1690s premieres now occupied a further 15% of the repertoire, leaving only 35% of works to premiere in the 1700s themselves (the orange segment). And so on across the chart.
A few points suggest themselves immediately. First, of course, we see the long dominance of the 1680s repertoire (again, including a significant amount of pre-1680 work), which seems to drop only barely below 20%, and then only in one decade (the 1780s). Second, the top segment of each column is always a decade’s premieres. A quick glance across the top of the chart reveal the 1730s as an unusually productive decade in the eighteenth century. Third, the 1690s, 1700s, and 1710s repertoire seem, as a group, to occupy a relatively stable percentage of the repertoire from the 1720s through the 1750s (with temporary backsliding in the 1740s). (These are the blue, orange, and green segments.) To maintain that repertoire, the troupe sacrificed some of its durable 1680s repertoire in the 1730s to keep in rotation the still-popular works from 1690–1719. This is clearest in the steep drop in 1680s repertoire from the 1720s, where the 1680s account for approximately 40%, to the 1730s, where the 1680s occupy closer to 30% of the repertoire.
The calculations in the above chart count the repertoire as a unique set of works. Whether a work is performed 1 or 30 times, it counts as an equal portion of the company’s repertoire. One might reconstruct this chart counting performances instead, as in Figure 2. This version measures not only the variety of the company’s repertoire but also the actual use of the repertoire. The picture appears, at first glance, not drastically different from the previous image. The most obvious difference is in the use of the 1680s repertoire. From 1700 on, the decline in the use of the 1680s repertoire’s appears relatively steady, from its peak of over 60% of performances to the nadir of 20% in the 1790s. Compared to the fortunes of the 1680s, premieres from other decades occupy a much more stable slice of any given decade’s performances. One possible feature of interest: for some decades, the premieres seem to skip a season before reacquiring favor with the troupe. For instance, the 1730s repertoire (the brown segments) appear to comprise only 10% of performances in the 1740s. But in the 1750s and for a number of decades following, the 1730s plays seem to have increased their presence by around half.
The next chart will try to look at the impact and durability of each season’s premieres in more detail. This is a complicated chart, so let’s construct it step by step. Consider the 1750 season. (While seasons actually cross the calendar year, for convenience I name seasons by the first year. The 1750 season thus should be understood as the 1750–51 season.) We can make a simple chart displaying every season in which a play that premiered in 1750 was subsequently performed. It might look something like Figure 3. Moving from left to right, we go season by season, starting in 1750 and ending in 1792. For each season, if a work that premiered in 1750 was performed that season, we mark it with a black circle. So, in 1753, 1763–1772, 1776, 1777, etc., no works from the 1750 season were reperformed.
But, wait: what if more than one work from 1750 were reperformed? How do we know how many works we’re talking about here? Let’s remake the chart, this time setting the circle’s color (on a grayscale from 0% equals white to 100% equals black) to the percentage of 1750 premieres that are performed in a season, as in Figure 4. The first circle in 1750 is, by definition, black; one hundred percent of the premieres from 1750 were performed in 1750. But the next season, the circle is much lighter (1/8 of the way to black), because of the 8 works that premiered in 1750, only 1 was reperformed in 1751. In 1761 and 1762, two works from 1750 were reperformed, and the circle is 25% of the way to black.
One more wrinkle: how can we tell whether or not the reperformed works are the same subset? That is, imagine that five plays, A, B, C, D, and E premiered in a season. In the next season, A and B are reperformed, giving us a 40% black circle. But in the subsequent season, C and D are reperformed. Our circle is again 40% black, but we’ve lost any sense that the repertoire over this two-year period includes not 40% of our plays, but actually 80%. So we will make for our chart one more piece, a horizontal line for each work, extending from the premiere year and continuing to the last year in which the play was performed. Again, the darkness of the line is a percentage of the plays reperformed, from zero to one-hundred, white to black. The result is Figure 5. Of the four works from 1750 to be reperformed in subsequent seasons, one was dropped from the repertoire in 1763, one in 1774, and the last two in 1784 and 1785. Thus the line is 50% black until 1763, then 37.5% black until 1774, then 25% to 1784, and finally 12.5% to 1785, when the line disappears and the 1750 premieres began gathering dust on the company’s shelves. Finally, we can simply combine our circle and line charts, to Figure 6.
Ok, that’s just the 1750 season. Let’s do the same thing for every season from 1680 through 1792 and put them on a single chart. In Figure 7, moving up each row along the y axis are the seasons from 1680 to 1792. (The blue seasons mark every fifth year, to improve chart legibility. You can zoom in on the page to enlarge the image.) As you move across a given row along the x axis, you see the durability of that row’s (i.e., season’s) premieres in later seasons. So, at the x, y point 1720, 1720, the dot represents all the plays that premiered in the 1720 season. The point 1721, 1720 gives all the plays that premiered in 1720 that were reperformed in 1721. And so on. Reading the 1720 season, for example, we see immediately that the season’s premieres did extremely poorly in 1721. The light dot at 1721, 1720 reveals that a very small percentage of premieres from 1720 were part of the 1721 season. But the line extending out along 1720 is much darker than the dot in 1721, recording that, although the 1720 premieres were extremely unpopular in the first decade or so, a number of them resurfaced in later years. By contrast, the 1731 premieres were more immediately popular in subsequent seasons, with more frequent and darker dots along the x axis. However, a smaller percentage of premieres from 1731 remained in repertoire, as that season’s horizontal line is lighter than that for 1720. This complicated graph tries to capture, then, both how and when a given season’s new works were reperformed, as well as the percent of plays in a given season that remained viable in later years. Among the other details visible here are the abysmal records of the 1761 and 1762 seasons (after an unusually successful 1760). Contrast the strange 1728 season, which bore little to no fruit for decades until a sudden and persistent resurgence of interest in a number of works from that year in the late 1780s. No overall patterns are clear, but the graph offers a glimpse of the success of each season’s novelties in detail.
The three preceding charts all offer a look at how successful works from a given decade or season were in subsequent years. They try to capture the persistence of works from a given period over the full time period of the data. But we might also think about repertoire on a smaller scale, focusing on the year-to-year turnover. Figure 8 charts the percent of a season’s repertoire that was also part of the preceding season. Think of this figure as the conservation of repertoire on a seasonal basis. The maximum comes somewhere around 1710, when a full 90% of the season’s plays had been performed in the season immediately prior. The minimum comes in the 1730s, at around 65%. Now it is entirely unclear from this chart whether the relative novelty of a given season compared to its predecessor derives from greater innovation (i.e., more new works) or simply from the less frequent repetition of old works (i.e., transforming a perennial favorite to a biennial or triennial specialty). The trend itself, however, is apparent. From 1680 to 1710, seasonal repetitions were around 80%, with variations of between 5 and 10% on either side. Seasonal repetitions trended down to the 1730s, then hovered around 70%, before rising again almost to the earlier level in the 1750s. Another peak in the 1770s was followed by a second decline through the end of the data in the early 1790s. These patterns are far less pronounced (though still somewhat discernible) if we take a three-season window instead, as in Figure 9. Here, the real difference appears to be between the seasons from 1685 to 1765, which hovered around 85% conservation of repertoire from the past three years, and from 1765 to the Revolution, when 90% or more was the norm.
Finally, let’s take a close look at the single most important “premiere” season, namely the company’s first season, 1680. Again, by definition I’m exploring only company premieres, the first season in which a work was performed by the Comédie-Française, constituted as such. Thus all the works by Molière, for instance, that the company performed in 1680 “premiered” in 1680, although they had actually premiered years or even decades earlier. In other words, we’re examining, by and large, the established, neo-classic repertoire of the French theater as utilized by the Comédie-Française up to the Revolution. (Some pre-1680 works first appeared in the company’s repertoire in later years; I do not account for them here.) Figure 10 gives a play-by-play look at the 1680s repertoire and its persistence in later seasons. Each play has its own row, each column is a season, beginning as usual in 1680, ending in 1792. A filled box in the grid indicates that the play was performed at least once in that season. An empty box means the play was not performed. The plays are sorted by the number of seasons in which they were performed. And I have excluded plays that appeared in fewer than 10 seasons after 1680.
The resulting chart divides roughly into thirds. At the bottom of the chart we have the true company mainstays. L’École des femmes, Le Médecin malgré lui, Tartuffe, each performed at least once in every season to the Revolution. Another dozen plays (Le Cid, Britannicus) with only intermittent performance gaps of a season or two, and then starting only in the 1720s. And then a handful more with somewhat longer periods of non-performance, but otherwise strong consistency from year to year. The middle third, say from around Le Menteur to Sgnarelle, has long stretches of popularity, but then disappears for 5 or 10 years at a time, often more than once. These plays are popular enough to sustain interest, but not so essential that they cannot be set aside. And then we have the top of the chart, successful or popular plays that, for the most part, had a more temporally limited appeal. Generally, these plays were successful to the 1720s, when the company either dropped them entirely or revived them only occasionally, but without any sustained interest.
Now, within this general tripartite division, individual plays follow their own unique patterns, patterns scholars better versed in the period’s dramatic literature would recognize more readily than I. For example, Jan Clarke noted to me during the conference that many of the company’s true mainstays are petites pièces, performed as a second or third play in an evening. (This is something one might verify, of course, with the data.) Les Femmes savantes is an example of a play for which a specific look at company practices might be fruitful. The five-year hiatus in the late 1740s marks a clear break in the company’s use of the play. Why? And what explains the sudden dismissal of Héraclius, empereur d’Orient in the 1700s, and the play’s relatively consistent return a decade later? Seeing the 1680 repertoire as a whole, the treatment of individual plays stands out more starkly and invites us to tell new stories about the Comédie-Française and its production practices.
I hope these charts contribute to that endeavor.
To download png files of the main figures, select the links below.
Derek Miller is an Assistant Professor of English at Harvard University. His research focuses on the intersections between art and industry in the nineteenth and twentieth centuries. More information at http://scholar.harvard.edu/dmiller.