1 For late twentieth-century debate, see Courtenay, Lynn T. and Mark, Robert, ‘The Westminster Hall Roof: A Historiographic and Structural Study’, Journal of the Society of Architectural Historians, 46.4 (Dec. 1987), pp. 374–93CrossRefGoogle Scholar; letters from Heyman, Jacques, Mainstone, Rowland, and Courtenay, Lynn T. and Mark, Robert, Journal of the Society of Architectural Historians, 47.3 (Sept. 1988), pp. 321–23Google Scholar; Huang, Yun Sheng et al., ‘Westminster Hall's Hammer-Beam Roof: A Technological Reconstruction’, APT Bulletin, 20.1 (1988), pp. 8–16 CrossRefGoogle Scholar. Courtenay, Lynn T., ‘The Westminster Hall Roof: A New Archaeological Source’, Journal of the British Archaeological Association, 143 (1990), pp. 95–111 CrossRefGoogle Scholar; Morris, E. Toby, Black, R. Gary and Tobriner, Stephen O., ‘Report on the Application of Finite Element Analysis to Historic Structures: Westminster Hall, London’, Journal of the Society of Architectural Historians, 54.3 (Sept. 1995), pp. 336–47CrossRefGoogle Scholar; Mainstone, Rowland, ‘Structural Analysis, Structural Insights, and Historical Interpretation’, Journal of the Society of Architectural Historians, 56.3 (Sept. 1997), pp. 316–40CrossRefGoogle Scholar; Waddell, Gene, ‘The Design of the Westminster Hall Roof’, Architectural History, 42 (1999), pp. 47–67 CrossRefGoogle Scholar. The most recent discussion of the roof is Munby, Julian's, ‘Late 14th Century Reconstruction of Westminster Hall’, Westminster II. The Art, Architecture and Archaeology of the Royal Palace, ed. Rodwell, Warwick and Tatton-Brown, Tim (Leeds, 2015), pp. 120–32Google Scholar.

2 My approach is an adaption and extension of that of Harvey, William, ‘Westminster Hall and the Woodman’, The Builder, 121 (1921), pp. 220–21Google Scholar. It is also close, despite some significant points of disagreement, to that of Waddell ‘The Design of the Westminster Hall Roof’.

3 Harvey, John, The Perpendicular Style 1330–1485 (London, 1978), p. 118Google Scholar.

4 Courtenay, Lynn T., ‘Westminster Hall and its Fourteenth-Century Sources’, Journal of the Society of Architectural Historians, 43.4 (Dec. 1984), pp. 295–309 (pp. 307–08)CrossRefGoogle Scholar.

5 Harvey, John, English Mediaeval Architects, 2nd edn (Gloucester, 1984), p. 139Google Scholar; Courtenay, ‘Fourteenth-Century Sources’, p. 304. The original roof is lost and the present hammer-beam roof is a reconstruction of c. 1931 by William Weir, not intended as a replica of the original; see Emery, Anthony, Dartington Hall (Oxford, 1970), pp. 161–64Google Scholar.

6 Harvey, English Mediaeval Architects, pp. 138–39.

7 The History of the King's Works, ed. Colvin, H. M., 2 vols. (London, 1963), I, pp. 529–30Google Scholar; Calendar of the Close Rolls, Richard II, 5, 1392–1396 (London, 1925), p. 352Google Scholar. Of course, Herland's choice of Farnham as the site to frame the Westminster roof may have been due to its proximity to the woodland sources of the timbers.

8 Pre-Westminster Hall hammer-beam roofs: the Bishop's Kitchen, Chichester, c. 1300; Pilgrims’ Hall, Winchester, 1310–11; Tiptofts Manor, Wimbish, Essex, c. 1325; Balle's Place, Salisbury, c. 1370–85.

9 The form and structure of the roof carpentry of the great hall, as built 1097–99 by William II (Rufus), is unknown. Roland B. Harris and Daniel Miles have recently proposed a tie-beam roof: ‘Romanesque Westminster Hall and its Roof’, in Westminster II. The Art, Architecture and Archaeology of the Royal Palace, eds Rodwell and Tatton-Brown, pp. 22–71 (pp. 44–68). Aisled carpentry, however, remains a possibility; see Robert Beech, ‘The Hammer-Beam Roof: Tradition, Innovation and the Carpenter's Art in Late Medieval England’ (doctoral thesis, University of Birmingham, 2015), Appendix 2.

10 Harvey, English Mediaeval Architects, p. xlvi; for Winchester Castle, see Portal, Melville, The Great Hall of Winchester Castle (Winchester, 1899)Google Scholar; Biddle, Martin and Clayre, Beatrice, Winchester Castle and the Great Hall (Winchester, 1983)Google Scholar. The date of the roof is problematic. Biddle and Clayre suggested possible dates of 1348–49 or 1390–1404 (p. 28) whereas Portal leaned more to the 1390s (p. 64). Extensive works were certainly executed in the early 1390s: during 1391–92 Richard II paid 13 carpenters for 74 days work; in the thirteenth year of Richard II's reign, the hall was described as ‘quasi de novo constructa’ (Foreign Accounts, 17, Rich. II, rot. Gd; see Colvin, King's Works, II, pp. 863–64). In 2000, two relatively minor roof timbers were tree-ring dated (Vernacular Architecture, 31, p. 86) and yielded an imprecise felling date of 1462–1562. This, however, is not necessarily the period in which the roof was constructed. As the report states, ‘these timbers may have come from the same phase of construction or repair […]. The roof of this hall is thought to contain timbers from many different periods.’

11 For the date of the hall, see Leach, Arthur Francis, A History of Winchester College (London, 1899), pp. 133–34Google Scholar; Harvey, Perpendicular Style, pp. 61 and 277. The present roof is of 1819–20 by William Garbett, and is said to be a faithful replica of Herland's original; see Harvey, English Mediaeval Architects, p. 139; Emery, Anthony, Greater Medieval Houses of England and Wales, 1300–1500, 3 vols (Cambridge, 1996), III, p. 421Google Scholar; also Garbett's letters in Kirby, Thomas F., Annals of Winchester College (London, 1892), pp. 42–43 Google Scholar; and in Britton, John, The History and Antiquities of the See and Cathedral Church of Winchester (London, 1817), pp. 55–69 Google Scholar.

12 ‘Spur’ is the author's designation for these projections.

13 For a summary of the hall's historiography before 1987, see Courtenay and Mark, ‘Historiographic and Structural Study’, pp. 377–87.

14 Despite Courtenay and Mark's 1987 assertion that their study had ‘clearly resolved’ questions regarding the roof's structure (‘Historiographic and Structural Study’, p. 374), in 1995 Toby Morris et al. wrote that ‘there is as yet no definitive analysis’ (‘Finite Element Analysis’, p. 338). See also the vigorous exchange of views in the correspondence of Courtenay and Mark and also Heyman and Mainstone (as note 1), and the combative tone of Waddell's 1999 paper.

15 Courtenay and Mark, ‘Historiographic and Structural Study’, pp. 390 and 392; Toby Morris et al., ‘Finite Element Analysis’, pp. 340, 342 and 344; and Mainstone, ‘Structural Analysis’, p. 320. In order to ascertain the structural role of the various timbers, Courtenay and Mark used a 1:10 timber model, built by Yun Sheng Huang, fitted with electronic strain gauges.

16 In 1967, Heyman contended that the wall head provides the primary support for the roof and that the principal rafters are the main structural members. He added that hammer beams and hammer posts are ‘not load-bearing under a vertical dead load’, and hammer beams are ‘virtually unloaded’, while the arch rib and wall posts are only useful to alleviate lateral wind-loading; see Heyman, Jacques, ‘Westminster Hall Roof’, Proceedings of the Institution of Civil Engineers, 37 (1967), pp. 137–62CrossRefGoogle Scholar. He largely maintained this view in 1988 in his letter (as n. 1).

17 Beginning in 1913, the roof was surveyed and repaired by HM Office of Works. The ‘Schedules’ consist of detailed drawings executed to facilitate the work. A bound volume of the ‘Schedules’ is in the National Archives: ‘Works’ 29 / 3299, ‘Westminster Hall: Bound volume containing 38 drawings showing the roof trusses and the repairs carried out.’ Another copy is held in the Parliamentary Archives, Westminster. National Archives (‘Works’ 29 / 3419–3597) contains further drawings made by Baines's office during the restoration.

18 This author has worked briefly as a green oak carpenter; Mr Dalton is a qualified traditional carpenter, fully experienced in green oak construction.

19 Baines, Frank, Report […] on the Condition of the Roof Timbers of Westminster Hall (London, 1914)Google Scholar; Herbert Cescinsky and E. R. Gribble, Early English Furniture and Woodwork, 2 vols, (London, 1922), I, pp. 100–01.

20 Roofs of wider span existed in southern Europe, but of a different carpentry tradition; see Wilson, Christopher, ‘Rulers, Artificers and Shoppers: Richard II's Remodelling of Westminster Hall, 1393–99’, The Regal Image of Richard II and the Wilton Diptych, ed. Gordon, Dillian, Monnas, Lisa and Elam, Caroline (London, 1997), pp. 33–59 (p. 55, n. 98)Google Scholar, for the Palazzo della Ragione, Padua, and other wide-span roofs.

21 The author's measurements of the span of the hall: south end: 68 ft 6 inches (20.88 m); midpoint: 68 ft 3 inches (20.80 m); north end: 67 ft 5 inches (20.55 m). The span at each location was measured numerous times with a Bosch GLM 80 laser rangefinder at a height of approximately 1.2 m above the floor, with the shortest result recorded. The hall, therefore, tapers from south to north.

22 Baines, Report on the Condition of the Roof Timbers, pp. 25–26. Green English oak weighs around 720 kg per cubic metre; see Peter Ross et al., Green Oak in Construction (High Wycombe, 2007), p. 24.

23 The largest cruck building in existence is Leigh Court Tithe Barn, Worcestershire (1344), which has a span of just 33 ft 6 inches (10.2 m): Alcock, N. W., Cruck Construction: an Introduction and Catalogue (London, 1981)Google Scholar, frontispiece, p. 27; Charles, F.W.B., The Conservation of Timber Buildings (Shaftesbury, 1995), pp. 48–50 Google Scholar.

24 Baines, Report on the Condition of the Roof Timbers, p. 27; Morris, Thomas, Brief Chapters on British Carpentry (London, 1871), p. 76;Google Scholar Yeomans, David, How Structures Work (Oxford, 2009), p. 111Google Scholar. From the author's own experience, a green oak arch-brace truss completed by Timber Framing And Conservation of Leicestershire in 2011, despite its full complement of pegged mortice and tenon joints, was only passed by the structural engineer as safe when it included three-way high tensile steel tie-rods.

25 Baines thought that the main ‘trussed’ purlin originally consisted of three timbers: two main structural timbers and a decorative cornice. He concluded that other timbers were later additions, Report on the Condition of the Roof Timbers, pp. 23 and 70. For the collar beam, see Report on the Condition of the Roof Timbers, pp. 19–20.

26 Herland certainly used loose tenons to bind the collar beams, but Baines is unclear as to whether these are dovetailed, Report on the Condition of the Roof Timbers, p. 19.

27 Illustrated in Hewett, Cecil, English Historic Carpentry, 2nd edn (Fresno, Ca., 1997), pp. 25–26 Google Scholar.

28 Épaud, Frédéric, De La Charpente romane à la charpente gothique en Normandie (Caen, 2007), p. 234Google Scholar.

29 McGrail, Seán, Ancient Boats and Ships (Princes Risborough, 2002), pp. 38 and 42–43 Google Scholar; Ancient Boats in North-West Europe (London, 1998), pp. 103 and 135.

30 Baines, Report on the Condition of the Roof Timbers, pp. 24 and, 29.

31 This form of tying joint is commonly found in late medieval English carpentry.

32 Baines, Report on the Condition of the Roof Timbers, pp. 23 and 70.

33 Ibid ., p. 30.

34 Courtenay and Mark, ‘Historiographic and Structural Study’, passim.

35 Toby Morris et al., ‘Finite Element Analysis’, pp. 336–47.

36 For the exterior of the hall of Winchester College see unnumbered plate in Winchester College: Sixth-Centenary Essays, ed. R. Custance (Oxford, 1982).

37 The first instances of northern European carpenters suspending tie beams via king posts from the apex date to the twelfth century. By 1220, and the construction of the nave roof of Notre Dame in Paris, carpenters were developing elaborate roof systems with ‘hanger’ king posts: see Hoffsummer, Patrick Roof Frames from the 11th to the 19th Century: Typology and Development in Northern France and Belgium (Turnhout, Belgium, 2009), pp. 168Google Scholar, 174, 183 and 185. See also Épaud, De La Charpente romane, pp. 185, 188 and 213.

38 Having consulted pictorial sources for English medieval carpentry in general, and for king-post roofs in particular, I can find no earlier example of a joggled king post than Herland's 1393 usage. Earlier European examples, however, can be found; see Épaud, De La Charpente romane, p. 189; Hoffsummer, Roof Frames from the 11th to the 19th Century, p. 174.

39 See Moxon, Joseph, Mechanick Exercises, or The Doctrine of Handy-Works (London, 1703), p. 145Google Scholar; Tredgold, Thomas, Elementary Principles of Carpentry, 3rd edn (London, 1840)Google Scholar, pl. V, fig. 49; pl. VI, fig. 51; pl. IX, fig. 62; pl. X, fig. 66; Newman, Rupert, Oak-Framed Buildings (Lewes, 2005), p. 49Google Scholar.

40 Baines, Report on the Condition of the Roof Timbers, p. 32.

41 Courtenay and Mark, ‘Historiographic and Structural Study’, p. 381.

42 Baines called these outer timbers ‘ribs’. As the whole of this composite structure is usually referred to as a ‘rib,’ I have decided to designate these timbers ‘laminates’.

43 Viollet-le-Duc, Eugène in his Dictionnaire raisonne de l'architecture (Paris, 1854–68)Google Scholar considered the arched rib to be a rigid monolithic structure; see Courtenay and Mark's comments on his appraisal in ‘Historiographic and Structural Study’, p. 380.

44 Baines, Report on the Condition of the Roof Timbers, p. 19.

45 The curved members of Yun Sheng's model were ‘formed by steam bending’, Courtenay and Mark, ‘Historiographic and Structural Study’, p. 388. Yun Sheng called his model ‘a reasonable close facsimile of the prototype’ (‘A Technological Reconstruction’, p. 12). He also concluded that the ‘great arch rib […] acts as a continuous member to convey much of the vertical dead load and horizontal thrust’ (p. 15, emphasis added).

46 The laminates in Yun Sheng's model clearly do not narrow at the intersection with the hammer post and beam. Courtenay and Mark (‘Historiographic and Structural Study’, p. 388) describe the outer laminates as being ‘continuous across the intersections.’

47 Courtenay and Mark concluded that ‘It is mainly the rigidity of the continuous arch that allows it to transfer much of the roof loading to the corbels’; ‘Historiographic and Structural Study’, p. 390 (also 392). The study by Toby Morris et al. concurred: ‘axial or compressive forces in the main members run primarily through the extremely large and stiff great arches to the corbel stones’, and ‘the extremely stiff arch is responsible for discharging most of the roof's loads to the corbel stones’ [emphasis added]; ‘Finite Element Analysis’, pp. 342 and 344. See also, Yun Sheng Huang, ‘A Technological Reconstruction’, p. 15.

48 Peter Ross et al, Green Oak in Construction, pp. 24–25, 54–55.

49 Toby Morris et al., ‘Finite Element Analysis’, p. 340.

50 Ibid ., pp. 340 and 342.

51 Courtenay and Mark, ‘Historiographic and Structural Study’, p. 387.

52 Notching the joint thus may also have had the intention of allowing the moulding to run uninterrupted through the timbers, but Baines's drawing shows the projection to be greater than any depth of moulding: see National Archives ‘Works’ 29/3299/29.a.

53 Waddell, ‘The Design of the Westminster Hall Roof’, p. 63.

54 Ellis, George, Modern Practical Carpentry, 1st edn (London, 1906), p. 97Google Scholar. Such structural intent was less elegantly alluded to by Thomas Morris (Brief Chapters, p. 50).

55 Baines, Report on the Condition of the Roof Timbers, pp. 18 and 30.

56 Courtenay and Mark, ‘Historiographic and Structural Study’, p. 384.

57 Baines, Report on the Condition of the Roof Timbers, pp. 30, 32; Waddell, ‘The Design of the Westminster Hall Roof’, pp. 56 and 61.

58 The ratio of hammer-beam length to hammer-brace length at the Pilgrims’ Hall is approximately 6:7. At Westminster it is 4:3.

59 See the comments of William Harvey (‘Westminster Hall and the Woodman’, pp. 220–21).

60 Courtenay and Mark, ‘Historiographic and Structural Study’, pp. 387–93; Toby-Morris et al., ‘Finite Element Analysis’, passim

61 Waddell, ‘The Design of the Westminster Hall Roof’, p. 49. Examples of hammer-beam roofs sans corbels: Norfolk churches: Hockwold, Holme Hale, South Acre; Suffolk churches: Bacton, Bramford, Cotton, Fressingfield, Ipswich St Mary at Quay, Newbourn, Roughham, Wetherden, Worlingworth. This list is by no means comprehensive.

62 Salzman, L. F., Building in England Down to 1540 (Oxford, 1952), pp. 212, 215, and 218–19Google Scholar.

63 Howard, F. E. and Crossley, F. H., English Church Woodwork (London, 1917), p. 19Google Scholar.

64 From the author's personal observation.

65 Sydney Smirke, ‘Remarks on the Architectural History of Westminster Hall’ (including ‘Second Letter’), Archaeologia XXVI (1836), pp. 417–18. For C. H. Smith, see, Thomas Morris, ‘Glances at the Structural Principle of the Roof of Westminster Hall’, Royal Institute of British Architects Proceedings, 1st series, 3 (1849–50), p. 7 (203).

66 The process of framing these joints in our reconstruction revealed the utility that this arrangement must have had during the early stages of erection. Erecting fully morticed, braced post and beam carpentry is always awkward. The brace has to be lodged in a mortice, usually of the vertical timber, for the horizontal timber then to be placed on top. Aligning multiple mortice and tenon joints is never an easy procedure, and requires many hands. When attempting this at Westminster, Herland was dealing with massive timbers, and a slip might prove fatal. It was in part to facilitate less arduous construction that Herland adapted the lower mortices of the core and the curved strut by leaving them open at their wall end. This refinement, allied with the level-shouldered spurs, makes the erection process easier. The post and beam can be assembled, and then eased apart slightly for the upper tenons to be fitted to the soffit of the hammer beam. The lower, open-ended, mortices can then simply slide onto the tenons of the wall post, with little further exertion.

67 See n. 56.

68 Thomas Morris, ‘Glances’, p. 4. Baines calls the core at this point a ‘wood column’, Report on the Condition of the Roof Timbers, p. 32.

69 Ed Levin, ‘Hammer-beam Roofs I’, Timber Framing, 48 (June 1998), pp. 15–16; Levin, ‘Hammer-beam Roofs II’, Timber Framing, 49 (September 1998), pp. 10–11.

70 Colvin, King's Works, I, p. 529.

71 Courtenay, ‘New Archaeological Source’, p. 105. It should also be remembered that the upper courses of the masonry wall were new, and that the long curing time of lime-based mortars, over months or even years, meant that the upper portions of the walls would have poor resistance to any thrust in this massive roof. As a further proviso, the spurs may have been inserted to maintain the lower wall post position longitudinally, i.e. to prevent the wall post swinging around the fulcrum of the hammer beam, especially during erection. Unfortunately, as ever with Herland's roof, nothing is explicit. Trusses 2 and 10 have no cavities behind the wall post, and have short, stubby spurs on one wall post only. The absence on truss 2 can be explained: the width of Westminster Hall narrows from south to north by more than a foot (305 mm). Widths: South: 68 ft 6in. (20.88 m); midpoint: 68 ft 3inches (20.80 m); north: 67 ft 5 inches (20.55 m). The gap between wall post and wall increases from north to south corresponding with the increasing width of the hall. It appears, therefore, that Herland carpentered all the frames to the same size. Truss 2, at the narrower end of the hall, is oversized, and so Herland embedded the eastern post into the masonry. Consequently there was no need for cavities and substantial spurs. The lack of cavities and substantial spurs on truss 10, where the hall is approaching its widest point, is difficult to explain. The observation that Herland carpentered all the frames to the same width casts doubt on suggestions that the gaps between wall post and wall were Herland's prescient attempt to prevent rot by providing ventilation (Courtenay, ‘New Archaeological Source’, pp. 103 and 107). The most northerly units, where the hall is at its narrowest, are either embedded or tight against the wall – so Herland was evidently not concerned about rot here. Rather, the majority of these enormous units are carpentered undersize for constructional reasons. It is easier to erect the frame, fill any gaps with timber packing (probably folding wedges), and cover them with moulding, than to erect the frame, realise it is too big, dismantle it, hack out substantial portions of masonry, and re-erect it. On frames of this scale Herland would have been happy with a gap of a few inches at either end to allow for adjustment during erection. Given, therefore, the scale and the idiosyncrasies of the masonry, it would have been impractical and probably impossible for Herland to carpenter all the frames to be a sliding, snug fit.

72 Quoting Shakespeare (Macbeth), Baines concluded that ‘the introduction of the great curved rib in the original design was doubtless designed to “make assurance doubly sure”’; Report on the Condition of the Roof Timbers, p. 32.

73 Ibid., p. 32.

74 Toby Morris et al. state that the rib is ‘nearly two foot square in section’ (‘Finite Element Analysis’, p. 338.) Before moulding, the rib is just a little over half that section. Combined, the three structural sections measure 24½ × 12 inches (622 × 305 mm).