The Phase I model has been designed without a roof or cover over the stage. In part because there was no archaeological evidence of pillar bases for the early Rose, which would have been necessary for supporting a stage roof, or ‘heavens.’132 The cantilevered roof stipulated in the contract for Henslowe’s later Hope playhouse seems to have been particular to that building, conceived as it was as a dual purpose with a removable stage to make the space flexible enough for animal baiting as well as plays: ‘And shall also builde the Heavens all over the saide stage to be borne or carryed without any postes or supporters to be fixed or sett vppon the saide stage.’133 The cantilevered roof can be seen in Hollar’s panorama of London (which he mislabels ‘The Globe’, fig. 52).
Bowsher conjectures the possibility of a very short cantilevered canopy due to the presence of a single shingle tile found at the site, but which may instead relate to the roof over the stage which was erected during the alterations in 1591/2.134 A lack of superstructure over the stage explains why the foundations of the front wall appear to be shallower than those of the frame.135
As for a ‘heavens’ over the Phase I stage, Gurr questions whether Thomas Nashe’s observation, that ‘here you shal find a paper stage streud with pearle, an artificial heau’n to ouershadow the fair frame,’ in his Preface to the first quarto edition of Sir Philip Sidney’s Astrophel and Stella published in 1591, ‘may or may not mean that the Rose had the same kind of stage cover as the Theatre and Curtain, and therefore had the posts to support it,’ the only three purpose-built playhouses at the time of publication.136 No evidence of posts was found in the archaeology for Phase I to support such a heavens over the stage. It is just possible Nashe is referring to the newly modified Rose if such modifications took place in the summer of 1591 (see Introduction, note 1); in any case, Gurr’s argument for a roof over the stage in Phase I is predicated on the evidence of a drip line in the archaeology that Bowsher and Miller conclude actually belongs to Phase II (see section 5.8.5), which was known to have a stage roof held aloft by pillars.
In his analysis of staging in the plays at the Rose, which he categorises as pre- and post-1592, Gurr concludes that:
Much of the evidence [in the published play texts] … is too non-specific to be really useful. The best generalisation would be to say that a comparison of early Rose plays with the latter texts gives no indication that any major change in resources or machinery was introduced in 1592. I am inclined to favour the idea that the frons scenae at both had angled walls, and that the discovery space and the heavens and its windlass were used from early on, as was the ‘above’ over the stage platform. Certainly there is no clear indication that any of these features were added for the first time in 1592 … The playwrights do not seem to have targeted their plays to specific playhouses in the early years, and most likely not before 1594, when the Rose and the Theatre began to offer a stable pair of venues for the two stabilised companies performing at them. Nonetheless, it does seem that, from the late 1580s, most of the playwrights preparing scripts for the London companies expected their venues to offer an ‘above,’ a discovery space, two stage pillars, a stage trap, and possibly even suspension machinery in the heavens.137
However, Egan argues that ‘no surviving play first performed between 1576 and 1595 calls for the appearance of deities other than by “Enter” and “Exit,” except John Lyly’s The Woman in the Moon (circa 1590–95), which was “written for boys and court performance” and hence does not tell us about open-air amphitheatre conditions.’
Robert Greene’s The Comical History of Alphonsus, King of Aragon (1598 [1594-95]), which belonged to the Henslowe/Alleyn company at the Rose when printed in 1599, Gurr says is ‘almost certainly an early Rose play.’138 The play’s opening stage direction gives ‘et Venus be let downe from the top of the Stage’; at the end of the play, Venus exits from this scene with what she calls a ‘trudge to heavens againe,’ with a direction ‘Exit Venus. Or if you can conveniently, let a chaire come downe from the top of the stage, and draw her up.’139 Mention of a Mahomet play in Henslowe’s diary was performed by the Admiral’s Men at the redesigned Rose between August 1594–February 1595, although the relationship with Green’s pay is supposition.140 Greene clearly allows for the possibility that his ideal of flight might not be realized, and the direction may have been added when, at the time of printing in 1599, such mechanisms were available at the Rose and other playhouses.
In his depiction of the early Rose playhouse (Speculum Britanniae, printed 1593 but drawn sometime earlier), Norden draws an open roof with no ‘hut’ or roof above the stage (fig. 53a). It is distinctly different to his depiction of the Rose after the alterations to it in 1591/2, which observes a noticeable change (fig. 53b).
Texts of plays thought by Gurr to have been written for or performed at the early Rose appear to call for posts or stakes to be used, but these aren’t necessarily in reference to stage pillars. The 1592 edition of Thomas Kyd’s The Spanish Tragedy or, Hieronimo is Mad Again, which Gurr suggests is ‘an early Rose play’, includes a scene that calls for a stake for tying a prisoner to: ‘They binde him to the stake.’141 Egan suggests, ‘since the order given is to burn him at the stake it would surely make better sense to use a property-stake,’142 although the only record of performance in Henslowe’s accounts date between 1591-1593, when pillars were available.143 Gurr suspects that Robert Wilson’s (?) The Three Lords and Three Ladies of London (1590) was written in around September 1588, although it ‘does not turn up, as [Robert] Wilson himself does later, in the lists of the Henslowe enterprises, but it was possibly a Rose play nonetheless.’144 The final scene contains dialogue and a stage direction that refer to a ‘post’ and to ‘the contrary post’.145 This, Egan says, ‘certainly suggests use of the two stage-posts’; however, as Wickham (cited by Egan) explains, this play probably belonged to the Queen’s Men and ‘two posts had to be forthcoming wherever they presented it—at court or on provincial tours as well as in a London playhouse: so the ambiguity cannot be removed entirely.’146 Or, as Gurr suggests, ‘[e]ven if not [available at the first Rose playhouse], this is evidence that the Queen’s Men expected to be able to perform at venues which had stage posts in the late 1580s.’ In Robert Greene’s The History of Orlando Furioso (1594), in which the lover pins his roundleys on to trees,147 Gurr conjectures these ‘would more readily be pinned on the stage posts than the tiring house wall,’ is not concrete evidence for pillars at the early Rose.148
Perhaps in the absence of stage posts and a roof, the first Rose playhouse had a canvas cover over the stage to protect the performers, and more importantly their costumes, and even the painted frons scenae from the elements?
Ronayne suggests that a velarium cloth, traditional in Roman theatres and later used at the Teatro Olympico in Vicenza, is unlikely at outdoor playhouses as it would ‘billow distractingly,’ although precedents exist.149
One of the earliest modern theatres in Rome, commissioned in 1513, was a temporary structure of wood painted to look like marble, built in the classical style with no roof but nevertheless featuring a ‘ceiling’ of blue and white cloth. In 1520, Henry VIII’s temporary theatre at Calais, built for the entertainments of Charles V, had an elaborately painted and studded blue cloth with ‘starres, sonne and mone,’ stretched, like a pavilion, over the circular theatre, supported by a large wooden column, underneath a more substantial waterproof roof. The theatre was remarkably similar to Elizabethan round playhouses: a sixteen-sided polygon with the spectators arranged in three tiers of galleries surrounding the stage on three sides. In 1527, Henry VIII commissioned Holbein to build a temporary theatre with a double layered roof—one of thick canvas over buckram and painted with the twelve signs of the zodiac.150 In 1572, on the occasion of the visit of the French embassy, a Banqueting House was erected and ‘the Covering thereof with Canvasse.’151 In 1581, in connection with the visit of the embassy from France to negotiate a marriage between Elizabeth and the Duc d’ Alencon, a banqueting house was erected at Whitehall ‘the walls of this howse was closed wth canvas … most cuninglie painted, the cloudes wth the starrs, the sunne and sunne beames.’152
Canvas awnings, then, seem to be the preserve of royal entertainments, with no documented evidence of their use in public playhouses. It is possible that Henslowe’s ‘clothe of the Sone & Mone’ was intended as some sort of stage cover forming the heavens in the Phase I Rose, although it only appears on an inventory of playhouse properties dated 1598.
 In an early analysis of the excavation findings, Gurr and Orrell suggest that the box drain discovered ‘which apparently leads from the rear of the early [Phase I] stage … was intended to take water from a cover over the stage,’ although they concede that this doesn’t agree with ‘the absence of any column base to support such a roof’ (Gurr and Orrell, ‘What the Rose can tell us,’ 425–26); later analysis by Bowsher and Miller concluded that the drain belonged to the Phase II stage, which is known to have had a roof over it for which column bases were present (Bowsher and Miller, The Rose and the Globe). Furthermore, Bowsher argues that ‘[t]he presence of a stage roof [in Phase II] not only explains the wider shape of the new plan, but conﬁrms the absence of a stage roof in Phase I. The widening of the gallery walls either side of the new stage was made so that sightlines from the upper galleries to the stage below the new roof could be maintained. Conversely, if the Phase I building had had a stage roof, the proximity of the upper galleries either side of it would have precluded any view of the stage’ (Bowsher, ‘The Rose and its Stages,’ 47).
 Greg, Henslowe Papers, 19–22.
 Bowsher, The Rose Theatre, 41; Bowsher and Miller, The Rose and the Globe, 47.
 Bowsher and Miller, The Rose and the Globe, 47.
 Andrew Gurr, ‘The Bare Island,’ Shakespeare Survey 47 (1994), 29–43.
 Andrew Gurr, ‘The Rose repertory: what the plays might tell us about the stage,’ in New Issues in the Reconstruction of Shakespeare’s Theatre, ed. F.J. Hildy (New York: Peter Lang, 1990), 132–33.
 Gurr, ‘The Rose repertory,’ 128.
 Robert Greene, The comicall historie of Alphonsus, King of Aragon (London: Thomas Creede, 1599). STC 2nd 12233.
 Foakes, Henslowe’s Diary, 23-27. In Green’s play, ‘Mohamet’ speaks to his acolytes through a ‘Brazen head’, which may have led Greg to suggest that the play recorded by Henslowe may identify with Green’s Alphonsus (Greg, Henslowe Papers, 116-117, n. 65).
 Gurr, ‘The Rose repertory,’ 126–27. Thomas Kyd, The Spanish Tragedie, containing the lamentable end of Don Horatio, and Bel-imperia: with the pittifull death of olde Hieronimo (London: Edward Allde, ), STC 2nd 15086.
 Egan, ‘The Theatre in Shoreditch, 1576-1599’, 168–85.
 Henslowe records Lord Strange’s Men staged a play called ‘Don Horatio’ (‘spanes comodye [of] donne oracioe’)’ on 23 February and 13 March 1592, with subsequent performances of a play called ‘Hieronimo’ (‘Jeronymo’ or ‘the comodey of Jeronymo’) between March 1591 – January 1593; the play was reprised by the Admiral’s Men for a number of performances between January-June 1597 (Foakes, Henslowe’s Diary, 16-19, 55-60).
 Gurr, ‘The Rose repertory,’ 126–27. Published by R. W. (Robert Wilson?), Gurr suggests that the play may have been written for the Rose c. 1587–90.
 Robert Wilson, The Three Lords and Three Ladies of London (London: R. Jones, 1590), Sig. F3v (STC 2nd 25783).
 Egan, ‘The Theatre in Shoreditch, 1576–1599,’ 168–85, citing Wickham, ‘“Heavens”, Machinery, and pillars,’ 8.
 Robert Greene, The Historie of Orlando Furioso (London: 1594), Sig. D1r (STC 2nd 12265). Henslowe records a performance of ‘Orlando’ on 21 February 1592 (Foakes, Henslowe’s Diary, 16).
 Gurr, ‘The Rose repertory,’ 126–27.
 Ronayne, ‘Totus Mundus Agit Histrionem,’ 139.
 Robert B. Graves, Lighting the Shakespearean Stage, 1567–1642 (Carbondale: Southern Illinois University Press, 1999), 45–46.
 Nichols, Progresses of Queen Elizabeth, I. p. 305n, cited in ‘The Banqueting House,’ in Survey of London: Volume 13, St Margaret, Westminster, Part II: Whitehall I, ed. Montagu Cox and Philip Norman (London, 1930), 116–39.
 B.M. Harl. MS. 293, f. 217; cited in Montagu and Norman, Survey of London: Volume 13, 116–39.