On Performance Management,
or Notes from the SM

Derek Miller
Harvard University

This is a slideshow version of a presentation delivered November 17, 2018 at the American Society for Theatre Research.
Use the right arrow key to advance, left arrow key to go back. Best viewed in full screen.
Do not cite without express permission.

Here is a stage manager's report from a performance of the 1980 Broadway revival of Morning's at Seven.

There's a small note in the bottom left box of this report. It reads: "Show played well."

Good Performance Report

A lot of the reports for this show include that same phrase: "Show played well." The terse, middling endorsement feels underwhelming, inconsequential.

Show Played Well

On other days, the performance report looks like this. Cast illnesses, crew replacements, leaking roofs, the day Nancy Reagan was in the audience—you name it, it happened. These reports, when not affirming "show played well," catalogue forces that threaten to break the theatrical performance and to undermine the production's integrity.

Bad Performance Report

I think our tendency, as scholars of theater and performance, is to turn to the messy performance reports, the ones full of incident that aroused the stage manager's attention and left an unusual trace in the archive. But I've been arguing in my recent work against this form of archival exceptionalism and seeking a scholarship that better understands the average. In this paper, I want to outline a study of stage manager's reports focused on the normal, rather than the exceptional cases. And I'm going to demonstrate how that attention might shift the emphasis of some of our scholarship.

The show these particular reports document is rather unremarkable, except for its surprisingly long run. A small-town comedy of family and old age, Morning's at Seven by Paul Osborn, was revived at the Lyceum Theatre on April 10, 1980. It garnered wide-spread praise for its veteran cast, which included Nancy Marchand, Teresa Wright, Maureen O'Sullivan, and Gary Merrill, under the direction of Vivian Matalon.

Playbill Title Page

Marnel Sumner, Ellen Raphael, and Roger Franklin were the three stage managers who filled out these reports, for the information of and occasional action by the producers.

Sumner Signature Raphael Signature Franklin Signature

Morning's at Seven played 581 total performances: 16 previews, an Actor's Fund benefit, and 564 regular performances. I have 579 stage managers' performance reports from this revival, all collected at the New York Public Library for the Performing Arts.

All Reports

As a way into this archive, I want to start with the data in the upper left box: the performance timings. These reports record a curtain up and curtain down time for each of Morning's at Seven's three acts, which, when summed up, give a total playing time for the show.

Performance Report Timings

Here is a box plot, which shows the distribution of performance playing times. As you see, the show ran generally around two hours: the median was 1:59:30 and the mean (the black square) very close, a mere nine seconds longer. The fastest and shortest performances were 1:53:35 and 2:09:44, respectively, but they were very much outliers. Fully half of all performances were plus or minus 90 seconds, a variation around the mean of 1.26%. And 90% of all performances (the distance between the whiskers) lay in an eight-minute range, that is, the median plus or minus four minutes. That, to my mind, is amazingly precise.

Playing Times Boxplot

So the first lesson from these reports is that our scholarship should pay more attention to the remarkable precision with which professional artists execute their craft. By definition, precise performances are repetitive and minimize variation, thus lacking the unpredictability that marks performance's liveness. But we can and should admire precision, just as Sara Jane Bailes teaches us to admire modern theater's performance of failure and Stacy Wolf urges us to to admire the amateur's exuberant joy. These reports document professional work at the highest level, characterized chiefly by remarkable precision.

Now, because professional work is so precise, we can use variations in the performance to think about how companies maintain a production's integrity and, indeed, what a production, over time, actually is. The playing time's overall precision does not mean that the playing time did not vary. We can plot the performance timings in a line chart, from the first preview until closing night.

Playing Times Line Chart

The general pattern here might be described thus: First, there is random variation throughout; that is, the time series is noisy. We can add a trend line, a moving average of the past eight performances, to filter out some of that noise.

Playing Times Line Chart with Moving Average

Now we see that the performances clearly got faster from performance 1 to about performance 70. From there, the company mantained a general (fast) equilibrium until around performance 150, when the steady state shifted slightly slower. The timing is then generally consistent through hundreds of performances, until performance 534, when the show gets stunningly slower and stays at a slower level through the end of the run. What's going on here?

Playing Times Line Chart with Moving Average

Here is the same chart, but with vertical lines indicating some major events in the production's run: opening night; the departure of Nancy Marchand, replaced after four weeks' interval by Kate Reid; Shepherd Strudwick's replacement of Gary Merrill; King Donovan's replacement of Richard Hamilton; and, finally, the replacement of four other cast members at performance 534. To understand better the precision of performance durations, let's parse them out into production phases that account for these cast changes.

Playing Times Line Chart with Moving Average

Here is a set of boxplots, both for the overall performance timings, and for each of the cast phases indicated in the preceding figure. One doesn't wish to over-interpret these data; however, we can also see that cast changes had at least some effects. Most notably, the four new cast members performed a much slower show than the original performers they replaced, clocking in at a mean of two hours and four minutes. That's almost three standard deviations from the mean of all previous performances, a very big difference.

And now I'll make a second interpretive assertion: the changing performance timings indicate a shift in the ontology of the production itself. In other words, we should think of theatrical productions not only as collections of texts, designs, people, and interpretations, but also as the average of a collection's iterative realizations in performance. A production is an average, normal performance. When a performance deviates too far from the average, the production itself becomes foggy and harder to discern. And when the production's average changes, the production itself also changes.

Let me work through my thinking more slowly. Performance timings give us one view of a production's general shape, of what the production, qua production, is. (Obviously, a production of Morning's at Seven that regularly ran three hours, even if we knew nothing else about it, would be very different from this production.) Now, on the one hand, that shape has to be fuzzy: our definition of a production cannot require absolute repetition at every performance—we must allow for some variation in execution. On the other hand, some variations will exceed acceptable boundaries. If those variations are temporary, they might indicate an aberrant performance. But if the changes seem more lasting, they indicate a real change in the production itself.

Playing Times Line Chart with Moving Average

This latter kind of change is clear, for example, in the wide variability of the preview period, during which the very text of the show was in continual flux and the production had not yet resolved itself into a re-presentable shape. King Donovan's arrival with the company clearly made the show faster. Notice, for instance, that Donovan's maximum (two hours and fourteen seconds) is faster than the seventy-fifth percentile (the top of the boxes) for all other periods. Such a shift suggests that Donovan altered the shape of the original production significantly. By contrast, Kate Reid's replacement of Nancy Marchand does not seem to have changed the playing time. So by that metric alone, the production was the same with either Marchand or Reid in the role of Ida. And, as already noted, the arrival of four new cast members made the production significantly different from what preceeded it, at the very least in its duration.

Thus, not every personnel change disturbs the production's integrity; some changes might redefine a production, others might not. But personnel changes do place a significant strain on production integrity and pose a particular challenge to the company.

Personnel changes are a surprisingly persistent challenge in these reports. Consider just the actors. Morning's at Seven has nine roles, understudied by five performers at a time. Ultimately, twenty-one actors performed in this revival on Broadway.

Morning's at Seven Cast Members

ThorMaurice Copeland (1-581)
CoraTeresa Wright (1-533)
Carmen Mathews (534-581)
ArryElizabeth Wilson (1-533)
Nancy Culp (534-581)
IdaNancy Marchand (1-269)
Harriet Rogers (270-298)
Kate Reid (299-581)
CarlRichard Hamilton (1-501)
King Donovan (502-581)
HomerDavid Rounds (1-405, 438-533)
Robert Moberly (406-437, 534-581)
MyrtleLois de Banzie (1-533)
Charlotte Moore (534-581)
EstyMaureen O'Sullivan (1-581)
DavidGary Merrill (1-453)
Shepperd Strudwick (454-581)
UnderstudiesDaniel Ziskie (Homer), Frances Helm (Esty), Robert Moberly (Homer), Jonathan Farwell (Thor, Carl, David), Dan Desmond (Homer), Martha Miller (Myrtle, Arry), Harriet Rogers (Cora, Ida, Esty, Arry)

Those twenty-one actors combined to form thirty-eight unique cast configurations on stage during the run. The following table shows each of those casts and the number of times they performed in that arrangement. I've ranked each configuration from 1 to 38.

Morning's at Seven Cast Configurations

Rank Thor Cora Arry Ida Carl Homer Myrtle Esty David Performances
1 Copeland Wright Wilson Marchand Hamilton Rounds de Banzie O'Sullivan Merrill 195
2 Copeland Wright Wilson Reid Hamilton Rounds de Banzie O'Sullivan Merrill 60
3 Copeland Wright Wilson Rogers Hamilton Rounds de Banzie O'Sullivan Merrill 52
4 Copeland Wright Wilson Reid Hamilton Moberly de Banzie O'Sullivan Merrill 31
5 Copeland Mathews Culp Reid Donovan Moberly Moore Helm Strudwick 28
6 Copeland Wright Wilson Reid Hamilton Rounds de Banzie Helm Merrill 21
7 Copeland Wright Wilson Reid Hamilton Rounds de Banzie O'Sullivan Strudwick 20
8 Copeland Rogers Wilson Marchand Hamilton Rounds de Banzie O'Sullivan Merrill 18
9 Copeland Mathews Culp Reid Donovan Moberly Moore O'Sullivan Strudwick 18
10 Copeland Wright Wilson Marchand Hamilton Rounds de Banzie Helm Merrill 15
11 Copeland Wright Wilson Reid Donovan Rounds de Banzie O'Sullivan Strudwick 12
12 Copeland Rogers Wilson Reid Hamilton Rounds de Banzie O'Sullivan Merrill 11
13 Copeland Wright Wilson Rogers Donovan Rounds de Banzie O'Sullivan Strudwick 10
14 Copeland Wright Miller Reid Hamilton Rounds de Banzie O'Sullivan Merrill 9
15 Copeland Wright Wilson Reid Hamilton Moberly de Banzie Helm Merrill 9
16 Copeland Wright Wilson Reid Farwell Rounds de Banzie O'Sullivan Strudwick 8
17 Copeland Wright Wilson Reid Hamilton Rounds Miller O'Sullivan Strudwick 7
18 Copeland Wright Miller Marchand Farwell Rounds de Banzie O'Sullivan Merrill 6
19 Copeland Wright Miller Rogers Hamilton Rounds de Banzie O'Sullivan Merrill 6
20 Copeland Wright Wilson Rogers Hamilton Rounds de Banzie Helm Merrill 5
21 Copeland Wright Wilson Rogers Hamilton Rounds de Banzie O'Sullivan Strudwick 5
22 Copeland Wright Miller Marchand Hamilton Rounds de Banzie O'Sullivan Merrill 4
23 Copeland Wright Wilson Reid Hamilton Rounds de Banzie Helm Strudwick 4
24 Farwell Wright Wilson Reid Donovan Rounds de Banzie O'Sullivan Strudwick 4
25 Copeland Wright Wilson Marchand Farwell Rounds de Banzie O'Sullivan Merrill 3
26 Farwell Wright Miller Reid Donovan Rounds de Banzie O'Sullivan Strudwick 3
27 Copeland Wright Wilson Reid Hamilton Rounds Miller O'Sullivan Merrill 2
28 Copeland Wright Wilson Reid Farwell Rounds de Banzie O'Sullivan Merrill 2
29 Farwell Wright Wilson Reid Hamilton Rounds de Banzie O'Sullivan Strudwick 2
30 Farwell Wright Wilson Reid Donovan Rounds Miller O'Sullivan Strudwick 2
31 Copeland Mathews Culp Reid Donovan Moberly Miller O'Sullivan Strudwick 2
32 Copeland Wright Miller Marchand Hamilton Ziskie de Banzie O'Sullivan Merrill 1
33 Copeland Wright Wilson Marchand Hamilton Rounds de Banzie O'Sullivan Farwell 1
34 Copeland Wright Wilson Rogers Hamilton Moberly de Banzie O'Sullivan Merrill 1
35 Copeland Wright Wilson Reid Hamilton Moberly Miller O'Sullivan Merrill 1
36 Copeland Wright Wilson Rogers Hamilton Rounds de Banzie Helm Strudwick 1
37 Copeland Wright Wilson Rogers Hamilton Rounds Miller Helm Strudwick 1
38 Farwell Wright Wilson Rogers Donovan Rounds de Banzie O'Sullivan Strudwick 1
Cast Configuration Bar Chart

This bar chart shows the percent of total performances featuring each cast configuration. Note that the original cast (rank 1) appeared together in only a third of the 579 documented performances—meaning two thirds of the performances deviated from the work built in rehearsal at least in their casting. More startlingly, fully a quarter of all performances featured cast configurations that played together fewer than 16 times (that's rank ten or higher). In other words, one out of every four performances took place with a set of actors that played those specific parts for under two weeks of performances.

Many of those cast changes, of course, featured understudies. Almost two in every five performances—or, in an eight-performance week, three performances per week—included an understudy. And no matter how able, understudies disrupt any company's work. As stage manager Marnel Sumner put it in a note on performance 135, "the first performance for any understudy […] tests the mettle of not just the understudy in question […] but also the stage managers, the other members of the cast, the wardrobe and hair departments—everyone involved in getting the show on." Crew replacements could pose similar challenges. "A new man on house curtain and it was slow!" Sumner complained in a technical note. When the flyman changed again later in the run, the same problem arose, leading Sumner to arrange a "curtain rehearsal" at half-hour. Understudies thus challenge the production's ability to function, requiring extra labor to maintain production integrity.

Understudies represent only one form of temporary variation that might disrupt a performance. The stage managers' reports catalogue a stream of mistakes and disturbances, though of varying significance, for a total of 511 recorded incidents.

Errors and Disruptions Categories Bar Chart

Here's a bar chart of errors reported during the run in a set of controlled categories. As you can see, by far the most frequent problems arose with lines—lines garbled (as when cast members replaced one character's name for another) or dropped entirely, as well as vocal trouble that came from the strain of eight shows a week or lingered after the frequent illnesses that plagued the cast. Late entrances were also a common error.

Errors and Disruptions Causes Bar Code

If we can look at a breakdown of errors at each performance, subdivided by their cause, we see that Gary Merrill and Kate Reid both had concentrated periods of erratic behavior that risked throwing performances off track. In Merrill's case, he seems to have had manic episodes, particularly after the surprisingly successful opening, that lead to strange offstage incidents and an unreliable monologue on stage. Kate Reid's early performances as a replacement cast member were notably uneven as she struggled with lines and seemed, from Sumner's perspective, not to understand how her part fit into the rest of the show. Fundamentally, however, the company behaved with high professionalism, focused on identifying and mitigating problems, not creating them.

Many of the problems the company managed came from exogenous disruptive forces. Audiences, for instance, caused 68 distinct incidents noted in these reports and were thus the biggest challenge to the show. Disruptions ranged from noisy drunks to an ill patron who vomited in the mezzanine during the first intermission.

Even when the audience behaved, the theater itself often did not. The backstage temperature became a near constant source of complaint in the winter, often solved by turning on a heating system that caused loud clanging in the pipes. Sumner thought temperature very important to the company's performance and to the show's reception. As he noted after performance 363: "It's the first time this week that the auditorium was comfortable. We also gave our best show and got our best reaction. I wonder if these facts are related."

Keeping up the production, of course, was not simply a matter for daily management, but required occasional course-corrections from someone with a more remote perspective on the proceedings. Thus director Vivian Matalon attended performances to take and give notes. The notes, however, often threw the company off its focus at the subsequent performance. "The first performance following a director's note session is a poor one," Sumner observed. "One in which the actors play the notes instead of the character and/or the play. One could almost feel a hesitation before each line that contained a note and then the actor proceeded to louse it up." This reveals an important distinction between the company's success in any given performance and the production's general integrity. Clearly, the work required to hone the production over time was worth an occasional stiff and uncertain performance.

Such maintenance, of course, was also literal. The set included grass, which, due to daily trampling by the actors, required frequent regluing. Wooden stairs broke, lighting equipment fell out of alignment, wigs grew disheveled, understudies went through their paces in anticipation of the next illness or accident. The day-to-day work of keeping the production up, keeping things functioning, consumed enormous time and effort. Few of these problems would have distorted the production so as to make it incomprehensible: the "noise" of any one flubbed line or missed light cue would not, by itself, distort the "signal" of any given performance. But taken cumulatively, these problems could eat away at the production until they did distort that signal and the information the company had labored to communicate through its rehearsal and preview period would have been lost amidst sloppy, error-ridden work. (This is, of course, the comic structure of Michael Frayn's great backstage farce, Noises Off.)

I want to recognize that many of my readers likely have an intimate knowledge of the professional theater's operation and thus the details I've highlighted from Morning's at Seven's long run are familiar and unsurprising. But as much as we might know these things about how theater works, I suggest that we, as a discipline, tend not to foreground their importance to theater's operation. And here I want offer a final interpretive frame for considering this archive and return to the point with which I began.

Viewed one way, these stage manager's reports document an endless series of threats to the production's integrity—the exceptions, the failures that hover over every performance and give performance its frisson of liveness. But difference here is not the company's organizing principle—rather, they seek sameness. Too much emphasis on performance's variability occludes the true goal of so much labor: performance sustainability. A history of Morning's at Seven's exceptionally unruly or erratic performances would miss its real achievement, namely, the precision with which the company maintained an integral production. That production is a moving target, responsive to changes in the cast, in particular. But executing that production with precision and care is, these reports make abundantly clear, the primary focus of the company's work together.

So I invite you to join me in thinking more about the repetition inherent in so many theater practices, exemplified here by the long-running commercial show. When you repeat a production, at some point the work of performance has less to do with art and interpretation than with that word suggested by the stage manager's title: management. Yes, a theatrical performance is a system of signs. Yes, a theatrical performance is a phenomenal and affective experience. But a theatrical performance also realizes a theatrical production, and the efforts to do that, as documented in stage managers' reports, reveal the performance as a managed event. Well-managed performances seek a precision sufficient to maintain the production's integrity.

From this perspective, the best performances are not the ones that arouse the stage manager's attention and earn a page of furiously written notes. No, the best performances are the ones that leave the least trace in this vast archive, the performances for which the stage manager can just write:

Show Played Well
Show Played Well Reports

On Performance Management, or Notes from the SM