The British actor and director Edward Gordon Craig (1872–1966) is known to have published more theoretical writings on the theatre and exhibited more set designs, than he actually produced plays: he took part in only 13 productions, either as a director or a set designer. Only 3 out of these 13 productions were stagings of plays by Shakespeare: Much Ado about Nothing in 1903, Hamlet in 1912, and Macbeth in 1928. Shakespeare was nevertheless one of his favourite playwrights, and he was obsessed with thinking about how to best put on his plays. The Tempest is just one example of a Shakespearean play to which he returned over and over again throughout his life, jotting down notes as to how he would stage it, should the opportunity present itself. He collected many of those notes in 1939 in a single manuscript, held by Bibliothèque nationale de France (hereafter abbreviated: BnF), département des Arts du spectacle (Performing Arts Unit, hereafter abbreviated: ASP), under shelf mark: EGC‑Ms‑B‑18. This manuscript cannot be regarded as some sort of a prompter book for a complete, consistent mise-en-scène, as Craig never bothered to homogenize his ideas about the play, which makes this manuscript all the more interesting to study. As Craig took pains, in most cases, to indicate the date of each individual annotation, it makes it possible to follow step by step the evolution of his conceptualization of a possible production. Most notably, it shows how Craig changed his mind about the treatment of the opening scene, in which Shakespeare depicts the tempest properly and the shipwreck that ensues, shifting over time from an almost realistic rendition of a ship, to radical stylization and abstraction through which Craig disposes of both sea and ship altogether.
But before focusing on act I, scene 1, it is important to have a clear overview of what The Tempest meant to Craig, and how he envisioned it as a play.
The place of The Tempest in Craig’s thought
When Craig was a young actor, in the 1890s, he had no liking for Shakespeare’s supernatural comedies such as A Midsummer Night’s Dream, The Tempest or The Winter’s Tale, nor for the character of Ferdinand, in which he could easily have been cast. In a later writing (Woodcuts, 12), reflecting on his youth, he claimed that at that time he did not “comprehend what [these] plays were about. They seemed too vague, mystic, bodiless”. This did not prevent him, however, from suggesting in September 1904 that he could produce The Tempest with Max Reinhardt (1873–1943), first in Berlin and then on a tour in England.1
By 1911, his interest in the play had grown to the point that he selected it for his demonstration that it is quite possible to produce a given script “in ten or even twenty different ways, and that each interpretation can be right” (Craig, Theatre Advancing, 192). In the same article, he highlighted the importance of understanding correctly such a play in order to give a good performance of it:
[…] the very best actors cannot hold up the weight of a great play like The Tempest if they are surrounded by what is called “noisy” scenery, by restless lighting or costumes, and if the stage manager has not understood and explained to his staff and performers the meaning of the play and the whole effect of the production. This meaning of the play is one of the things so often forgotten. (Craig, Theatre Advancing 192)
Is Craig sincere in this text, when he describes The Tempest as “a great play”? One can doubt it. The volume in which he collected all his staging ideas is filled with harsh criticism. On the whole, Craig does not deem the play worthy of Shakespeare, and he is convinced that Shakespeare simply strove to improve a poor play written by some inexperienced playwright (EGC‑Ms‑B‑18, 9r):
This is an old play rewritten by Shakespeare […] I bet that it is a play by a young man—very young—taken by S[hakespeare] who can invent no more plots but who can write as well as ever. Shak[espeare] comes across this and likes—rather likes—the boldness of the youth in taking fairy people, spirits, and magic for his stage.2
Elsewhere in the volume, Craig describes the play as “a queer piece, a thing of shreds and patches” (EGC-Ms-B-18, 8r)3 and “a very inferior affair” (EGC-Ms-B-18, 9r).4 He affirms that it was written by “two authors-muddlers: one poet attempting to link, to smooth, to save, [and] one Burbage bustling about, butting in, trying to ‘pull it together’; net result: a failure, a poor play, rich in some passages” (EGC-Ms-B-18, 114v).5
Other sources reveal that Craig suspected the play was not performable. On a copy of Horace Howard Furness’s variorum edition of The Tempest, he wrote that it is “a mysterious play which seems to deny all approach to it” (4-EGC-942(7), unnumbered half title page). In his daybook for 1957, Craig wrote (EGC-Ms-B-541(3), 55):
It’s well-nigh an impossible play to stage—it’s not of a piece—it has not the clearness of Hamlet or Othello or Midsummer Night’s Dream—it’s another dream and all dream […].
In a draft letter intended for Peter Brook (born 1925), and written on 20 April 1956, Craig called for a staging that would “lead to the massacre of the awful rubbishy lines and ideas”, and expressed his empathy for Shakespeare who had had to devote time and energy to improving such a bad play (EGC-Ms-B-18, 123v): “What Sh[akespeare] must have suffered over this horrible work is a crushing thought.”
If Craig held The Tempest in such low esteem, why was he so much interested in rereading and annotating it, in the first place? Precisely because of the challenge it represented for stage directors. As a play that deals primarily with dream—“We are such stuff as dreams are made on”—it compels directors to surpass themselves in imaginative qualities, and to surpass the Bard himself in poetic qualities. In 1924, Craig published an article in which he elaborated on the potential that The Tempest represents for an audacious stage director. In that article, Craig fancies that the action of The Tempest takes place undersea, and that all the lines delivered by the characters in act II, scene 1 “are issuing like bubbles from the mouths of six drowned live men sunken to the bed of the sea and wearily talking in their deadly sleep” (“On The Tempest”, 161). In other terms, the world expounded in The Tempest is the world of the afterlife, and the stage director is in the tricky position of having to materialize on stage all the wonderful beauties of that uncharted world (“On The Tempest”, 163–164):
In such an isle full fathoms five indeed our fathers lie. […]
Something very beautiful to see and to hear must have been what [happened there].
[…] What happened under the sea in an island […] is what I should like to make visible in The Tempest upon a stage, were I content to work to no purpose, to fashion what I fear would for ever fail to please you […].
Set designs drawn by Craig and now held in Paris, Vienna, and Osaka, show that as early as in 1905 he had intended to locate the action of The Tempest at the bottom of the sea. Such a reading of the play implies that the shipwreck depicted in act I, scene 1 is an actual event: unless a ship has actually wrecked, there is no obvious reason why all her passengers should have drowned. But this is not the only reading of the play that Craig had in mind. As already mentioned above, Craig opined that The Tempest can be produced “in ten or even twenty different ways”, and he had at least two other understandings of the play: one in which the ship actually exists, but the shipwreck only took place in the imagination of the protagonists; and one in which the ship herself is no physical object, but only part of Prospero’s dream. In both cases anyway, there is no need to show the shipwreck, as it never took place. At an unknown date after 1936, Craig wrote, referring to Miranda (4-EGC-942(7), 22): “She (as they in the ship) is possessed by this dream of a wreck—where none was in reality.”6 On 18 December 1956, Craig made it explicit, in a draft letter intended for his cousin John Gielgud (1904–2000), that “since that old magician [i.e. Prospero] kept all the wreck neat and trim it was and could only be in idea that the dam[n] wreck ever existed” (EGC-Ms-B‑18, 17r).
Craig had therefore, it seems, at least three possibilities in mind: 1. A real ship and an actual shipwreck, both being only suggested; 2. A real ship, but no shipwreck; 3. No ship and no shipwreck.
He never really made his choice between these three radically distinct ways of envisioning the play. Of course, had he had an opportunity to stage it, he would have had to make a decision; but as the mise-en-scène of The Tempest remained throughout his life a mere exercise for himself, he felt free to experiment with all kinds of ideas, without the necessity to pick one and develop it. In all cases anyway, Craig felt challenged by Samuel Taylor Coleridge’s (1772–1834) assertion that The Tempest
[…] addresses itself entirely to the imaginative faculty; and although the illusion may be assisted by the effect on the senses of the complicated scenery and decorations of modern times, yet this sort of assistance is dangerous. (66)
Such an assertion seems to ruin all stage directors’ efforts to put on the play. Of course Craig could not be contented with such a notion, and felt all the more compelled to strive to find several ways of directing it. He commented on Coleridge’s words as follows (4-EGC-942(7), 9): “Yes, but you should tell us how to deal with act I, scene 1, for example—for after all we have only our eyes and ears to help us when in a theatre.”
Act I, scene 1 of The Tempest becomes thus the issue at stake: how to present it on a stage without jeopardizing the whole play’s spiritual value? How to avoid the lavish sensationalism of 19th‑century productions, while appealing to a 20th‑century audience’s senses? Quite obviously, the huge difference between Craig’s three readings of the play resulted in three distinct strains of practical solutions when it comes to the staging of this particular scene. These three strains can be reduced to two: one in which the ship is to be seen, one in which there is neither ship nor sea on stage.
1905–1939: the actuality of a ship
In Craig’s earliest preserved sketch of a set design for The Tempest, dated 1905, the ship is not only present, she is even treated in a relatively realistic way (EGC-Ms-B-18, 130r). Although the pencil strokes are rather faint and difficult to interpret, there is one prominent, easily distinguishable detail: several lines clearly depict parts of the rigging of a ship. Those rigging elements form a diagonal across the sheet, dividing it into two distinct, equal sections. These two sections are unfortunately virtually indecipherable. The overall impression conveyed by this sketch is however somewhat reminiscent of Craig’s 1906 design for Dido and Æneas, published in Towards a New Theatre, in which a dozen sailing ships are visible (Towards …, 56).
When Craig copied his stage directions in a single volume in 1939, he introduced the various ideas he had had so far about act I, scene 1 as follows (EGC-Ms-B‑18, 13v):
This [scene] can be done several ways—excluding the building up of a facsimile of a ship (Italian) of the period and putting it through its paces on the stage. We are left with: 1. Suggestion by lights and this and that of a storm and wreck; 2. The hypnotic powers of Ariel seen at work upon the 8 or 10 [or] 20 more passengers. I have not heard of, nor seen, either of these two possibilities attempted.
The stage direction that elaborates on the first possibility has no less than 4 distinct dates attached to it (1905, 1921, 1930, and 1939). The set represents either the interior of a cabin or some portion of the deck—Craig does not choose between these two locations—and a lantern swings in all directions. It is the only source of light against a background characterized by Craig as “pitch black”. As a consequence, dancing shadows are cast erratically throughout the scene. In addition to the constant movement of light and shadows, the floor itself is designed so as to move beneath the characters’ feet: it consists of “a double-way which clanks this and that way [i.e. from front to back, and from left to right] and […] all four slides slope a little towards centre—result effect of some sort of bridge all dusk and indigo […]” (EGC-Ms-B‑18, 13v).
This production concept relies therefore entirely on light and motion: the details of the set are not visible, and the storm is suggested through the characters’ wild gestures inevitably induced by the moving floor, and the wild dancing of their shadows cast by the constantly swinging lantern. Although Ariel is not listed among the characters in this scene, Craig insists that his lines in scene 2: “I boarded the king’s ship; now on the beak, / Now in the waist, the deck, in every cabin, / I flamed amazement” (Shakespeare 109, I.1.196–198), clearly indicate that he has to be present in scene 1 as well, and the lantern symbolizes Ariel’s presence on the ship. Most certainly, Craig draws here on Francis Douce’s (1757–1834) interpretation of Ariel’s lines in scene 2. Douce commented on those lines as follows: they are, he says, “a very elegant description of a meteor well known to sailors. It has been called by the several names of the fire of Saint Helen, Saint Elm, Saint Herm, Saint Clare, Saint Peter, and Saint Nicholas” (vol. I, 3).
Craig was particularly interested in Douce’s remark concerning the fire of Saint Elm that “is also supposed to lead people to suicide by drowning” (vol. I, 4).7 Perhaps he felt that this remark could provide some logical justification for a production concept in which it was possible to show the passengers of the ship as drowned people on the seabed, while no shipwreck had occurred.
The other production concept, based on the “hypnotic powers” of Ariel, is dated either 1922 or 1939 (this is unclear from the manuscript). In this concept, Ariel is physically present, and is accompanied by a group of musicians and singers. The scene takes place by day light, under a “pale blue sky” in which “white clouds” are to be seen; the rest of the set consists of “yellow sands” on the foreground, and “hills” in the background. In this setting, here is what the audience is shown (EGC-MS-B‑18, 13v):
[C]haracters all lined up on deck (about 20) and Ariel with his musicians creating in the hypnotized 20 a sense of storm, calamity and wreck. They sway like waves with chorus of voices. […] They listen as the boatswain prone calls to his men to do this, do that. All whisper or yell or chatter as in their sleep.
The overall effect is totally different here: this concept does not rely so much on the visual elements as on the sounds; the only movement that can be seen on stage comes from the swaying of the hypnotized passengers. The immobile boatswain seems to be a mere instrument through which Ariel communicates with the other characters and creates in them the sensations provoked by a storm. Craig was aware of the difficulty that actors might have experienced in performing his stage directions: he made the remark, in 1939, that “perhaps only the Habima group could carry out this idea seriously and well”, referring to the Hebrew-speaking company that had operated under the auspices of the Moscow Art Theatre from 1918 to 1926, and that he admired very much. In 1956, he added the following words to that remark: “helped by Peter B[rook]”, the young and innovative stage director whom he had just met.
In two drawings dated 1935 (EGC-MS-B-18, 16r), Craig seems to return to a more “realistic” treatment of the ship: two distinct levels are clearly materialized, the lower level corresponds to the cabin, the upper level corresponds to the deck, on which a mast is to be seen. A winding staircase leads from one to the other, and the noble passengers use it throughout the scene, “running up and down all the time”, according to Craig’s specification. This almost realistic production concept contrasts sharply with everything else Craig has ever envisioned for the opening scene of The Tempest.
1942–1956: the empty stage of Prospero’s mind
During World War II, Craig had the revelation of a completely different treatment. This was on 3 May 1942, in Paris; Craig subsequently copied his new ideas in his Tempest manuscript by the end of November 1956 (EGC-MS-B-18, 16v, 18r).8 This treatment is based on the notion that the opening scene is one “where the words must all be heard above any howl of the winds and the roar of the waves—for the words are the Essence of the scene”. The focus is therefore once again here on the textual and sonic aspects, rather than on the visual elements. Prospero himself becomes the main protagonist of act I, scene 1; Ariel is “nearby”; in a later drawing, dated 1956, Ariel is absent, but in his stead Miranda is represented sitting on the floor, asleep, her back leaning against the armchair in which Prospero “can (if he wish) sprawl”. Craig’s vision for the scene has changed radically: “Now I see no more a ship (mast, sailors, etc.), I hear no more howls and roars nearby.” The action does not take place aboard the ship, but on the island. What Craig wishes to highlight is how Prospero is responding to the events that are happening in the distance:
I see Prospero, Ariel nearby; Prospero alone on his island, and afar off the howls, roars, cries, diminuendo.
Rather nearer, the voices of the mariners, crew, boatswain, etc., and the passengers, there to tell clearly the tale of the Disaster. The face and movements of Prospero tell us of his reaction to the unseen action going on off the stage.
Prospero as he listens in…9
At this point however, Craig changes his mind and thinks suddenly of another possibility, far more radical, far more audacious: he imagines that, perhaps, Prospero himself could deliver the text of act I, scene 1, as:
[…] a receiving instrument speaks in a room. As he listens in, as he looks on, hearing and seeing and reporting as one who is mesmerized reports in regular, quiet, unemotional tones—a monotone—till the climax comes: “We split—we split—we split.”
A wail (recorded on gramophone).
The term “gramophone” makes him change his mind once again, and he thinks of a third possibility, less radical, making use of state-of-the-art technical devices, and questioning the very notion of liveness in performance: “In fact, we will try the whole 65 lines of text as a record—and let it slowly out (close to Prospero) who notes each sentence: Prospero the listener…”
The use of the adverb “slowly” in this context is rather puzzling: what does Craig mean here? Should the recorded text be at lower speed, or is it just a loose, incorrect way of meaning that the sound should not be too loud? Craig may also indicate here that the recorded text should not be heard all at once, but that the lines should be interspersed with silence; in a following sentence, he makes it explicit that there should be “short or long pauses between the several bursts of speech”. As a matter of fact, Craig admits that the actual sonic qualities of the recording are relatively unimportant, as long as the audience’s attention is focused on Prospero: “Slow or rapid, loud or soft, jerked or smooth; maybe something in the lights, colours, shades coming and going. But Prospero remains still, and the commanding presence.”
Although Craig does not choose between those three scripts, he is confident that “this way we can reveal the idea in Shakespeare’s mind”. But he does not make explicit what Shakespeare’s “idea” consists of, according to him. Does the shipwreck actually take place, and Prospero hears the noise it makes and responds to it? Or is the shipwreck entirely imagined by Prospero? Should Craig’s production concept be staged, both interpretations would be possible for the audience.
This ambiguity was solved fourteen years later, when Craig met Peter Brook in 1956. Peter Brook and Natasha Parry (1930–2015) had come to visit Craig in Southern France in April 1956, and Craig was completely under the young couple’s spell, to the point that he shared some of his secret ideas about Macbeth with Brook, and allowed him to use them in his own production, if he was ever to direct that play (EGC-Ms-B-540(2), 5). Encouraged by that mark of confidence (quite unusual from Craig), Brook, who was to direct The Tempest at Stratford-on-Avon in 1957, and who by then had not yet become a convinced Artaudian, sent him a letter in August 1956, asking for a piece of advice, and Craig pasted that letter on his own note‑book (EGC-Ms-B-18, 127–128):
Have you any wise words on the play you’d care to drop this way? It’s fearfully difficult. Somehow all the masques have to be unified into the whole structure / conception of it. A lot of it must be very moonstruck and sinister, I feel. It ends in harmony, but surely should not be too harmonious from the start. It seems to me a mistake for the island to be peaceful and idyllic as soon as the first scene is over. And unless Prospero is a bit of a black magician tempted by his power he’s just Father Christmas. And how to suggest an island without depicting, without illustration? A ship can easily be evoked by its movement—a city by the essential lines of architecture, and so on. But an island—what is its essence? It forbids all constructions, scaffolds, bridges, steps—all unislandy. Perhaps I’ll have to knock down the back of the theatre and let in the Avon!
Send me a clue!
Craig’s draft response to Brook, dated 20 August 1956, is part of his Tempest manuscript. It is impossible to determine whether the letter he actually sent Brook was identical with this draft, and we do not know how Brook reacted to it, as there is no trace in the archive of a letter Brook would have sent Craig in response. In his draft letter, Craig suggested the play should begin with “my stage absolutely empty;10 not dark, not light: sleepy light” (EGC-Ms-B-18, 121r). Then Craig elaborates on his 1942 idea for scene 1, making it clear that the shipwreck only occurs in Prospero’s dream (EGC-Ms-B-18, 121–122):
All is still—but unbearably still; and then a figure, Prospero—not a ship, not a storm; scene 1: all the words printed as act I, scene 1 are now spoken by the mouth of Prospero, and since he makes the wretched wreck he will be at home. He seems to be … asleep: he (as ’twere) talks in his sleep … He seems to me to be seated, sprawling in a rocky armchair, his elbows on the arms and his hands held in air; I see them swinging gently from side to side […].
While Prospero is delivering the text of scene 1, visual elements appear in the dim light, transforming the empty stage into Craig’s recollection of the theatre on the boards of which he had learnt the skills of acting, many years before (EGC-Ms-B-18, 121–122):
[…] the whole of this stage is an island; you see boards, and ropes, and litter: it’s only your fancy, it’s the empty Lyceum Theatre […].
A big stage—pale, grey, brown shot with all the undersea pale greens and blues and crimsons. Yellows here and then—shot—with these—not spread. Fish seem to be swimming in and out of the ropes … I saw them anyhow. All vague apparitions. Dream place.
But I saw the figure in the rock seat and only later the bits of wreckage did form slowly, imperceptibly drift into a sort of undersea scene […].
There is no trace in the archive of any further conversation about the play between Craig and Brook, but Craig pasted into his Tempest manuscript a second draft letter, dated 18 December 1956, which he wrote to his cousin John Gielgud, who was to act as Prospero in Brook’s production the following year. Here, Craig confirms even more forcefully that neither the ship nor the wreck ever existed outside Prospero’s mind, and he insists that all the value of the opening scene relies entirely on the actor’s skills (EGC-Ms-B‑18, 17r):
Prospero (stands) or sprawls sleeping, alone on the stage … He moves a hand, maybe; he is such thing as Dream is made of, and he dreams the wreck.
The words are shot out by several voices; all the scene is in sound only: mumblings and cries, the words, maybe noises and music: hautboys, flutes, and singing, the voices do everything. Prospero listens in his sleep; his face (some acting for J[ohn] G[ielgud]—what!), rather a wicked face; he is motionless; the sounds increase; he laughs; he does what you will; but he does not move.
The dam[n] silly imitation of a wreck on the boards is swept away—the labour, the expense, the puzzlement all avoided.
Once again, Craig has changed his mind: here, the text is no longer delivered by Prospero himself, but by external voices, in the wings. The visual elements are focused on Prospero’s face. There is something Beckettian about this production concept: this Prospero looks indeed like some forerunner of Winnie in Happy Days, the Woman in Rockaby, Joe in Eh Joe, or the Listener in That Time. Craig was aware of Samuel Beckett’s (1906–1989) beginnings: he possessed a copy of Waiting for Godot, and in his daybook for 1957 he entered as an important fact that, on 8 April, he read a “report about Beckett’s new play” in The Sunday Times (i.e. Endgame; EGC-Ms-B-541(1), 71).
How did John Gielgud react to his old cousin’s suggestions? No reply from him is preserved in the Craig Collection in Paris, although we can infer from Craig’s daybook for 1956 that Craig did send him his letter.11
There is no evidence that Craig ever worked again on The Tempest after 1956. On 29 April 1957, he entered in his daybook (EGC-Ms-B-541(1), 88):
Then this evening looking at a line of The Tempest, I read on and on and on, the hour glided by. What Brook thinks he can make of this poem on a stage quite beats me. I wrote him it’s all a dream, nothing actual, till I suppose a dull quiet awakening at the end.
After Brook’s production opened at Stratford, Craig was eager to read reviews in the press, and was disappointed by the apparent lack of enthusiasm on behalf of critics, which he tried to explain as follows, in an entry dated 24 August 1957 (EGC-Ms-B-541(3), 55):
If the press notices on Peter and John’s attempts on Shak[espeare]’s Tempest do not read that hearty as they might, it’s because Tempest is a real problem for the stage, and I have doubts about P[eter]’s and J[ohn]’s ability to solve this problem. […] it’s another dream and all dream and Peter has failed to see this.
However, he also had the satisfaction to learn from the reviews that Peter Brook had perhaps used one of his ideas for the most difficult scene in the play, the opening scene, about which he had been thinking for so many years and for which he had envisaged so many distinct solutions (EGC-Ms-B-541(3), 55): “He seems to have used my idea of the swinging lantern in scene one.”
By way of conclusion
The earliest known drawing by Craig for act I, scene 1 of The Tempest is dated 1905, and his draft letter to John Gielgud about “how to open the play” is dated 1956. Craig spent thus over fifty years of his life thinking now and then about the difficulty of putting on this particular scene, the traps of a realistic treatment, and the necessity of avoiding the literal imitation of a shipwreck at sea. In an effort toward maximal abstraction and symbolism, he finally came up with a production concept akin to what he had wished to achieve with his 1912 Hamlet in Moscow, co‑produced with Konstantin Stanislavsky (1863–1938). His Hamlet was to be understood as a monodrama; he insisted that the audience should be made aware that what they were witnessing were not actual events in Hamlet’s life, but how those events resonated within Hamlet’s psyche, in his inner world.12 With The Tempest act I, scene 1, Craig reached another level of abstraction: Prospero is treated as some Beckettian figure ante litteram, whose dreams and thoughts the audience is invited to look directly in. While Craig is often perceived and introduced as a deadly foe of both playwrights and actors, here he relies entirely on Shakespeare’s poetic words and on the actor’s skills to convey the essence of the dream of a shipwreck, with no ship, and no sea: just with the sound of the words, and the mimics of a human face on stage.
