Newark Castle, Port Glasgow: a proto-modern roof of the late 16th century

Abstract

This article analyses the roofs over the north, east and west ranges of Newark Castle. Summarizing the results of a comprehensive research project undertaken in 2007–08, it discusses the roof carpentry of the castle in its historical and technical contexts and illuminates important aspects of building construction in 16th-century Scotland, in particular the problems associated with Scottish rafter single roofs, as well as proposing a construction sequence for the building of Newark. For the first time an important roof structure of central Scotland is discussed in detail, thereby focusing on a British region that has so far raised little interest in timber studies.

ABBREVIATIONS
HMSO	=	Her Majesty’s Stationery Office
RCAHMS	=	Royal Commission on the Ancient and Historical Monuments of Scotland

ABBREVIATIONS
HMSO	=	Her Majesty’s Stationery Office
RCAHMS	=	Royal Commission on the Ancient and Historical Monuments of Scotland

This paper would not have been possible without the help of many individuals and their support. Expressing my gratitude to Historic Scotland for generously funding this study, I would like to thank in particular Eric Ramsey and the custodians at Newark Castle for their readiness to provide unlimited access to the site, as well as the team at Kirkdale Archaeology — Gordon Ewart, Paul Fox, John Godbert, Sarah Hogg, Angus Mackintosh and Tom Whalley — for their contribution to this project. Special thanks go to Anne Crone for updating my knowledge on the use of pine and oak in 16th-century Scotland, to Scott Bain for familiarizing me with the roof of Skelmorlie Aisle, to William Bortrick for kindly offering information on the history of the Maxwell family and to Dennis Gallagher for his comments on the manuscript of this text.

The site archive for the work undertaken by Kirkdale Archaeology in 2007 and 2008 will be deposited with RCAHMS once remedial works to Newark Castle are completed.

Notes

1 For the definition of the term ‘Scottish Chateau’, see McKean 2001, v–vi.

2 This paper applies the terminology and orthography as suggested by Cordingley 1961, 77–8 and 111.

3 However, roof-features such as cruck roofs, open purlin roofs and solid stone barrel vaults, which were common in Scotland during the later Middle Ages, are not encountered at Newark Castle.

4 An archaeological investigation focusing on the roofs over Newark Castle was funded by Historic Scotland and undertaken in 2007 and 2008 by Kirkdale Archaeology. For its scope and results, see Kirkdale Archaeology 2007a; 2007b; 2008; and 2009. For the physical condition of timber used, see Demaus Building Diagnostics 2007, and Jatzlauk 2009. For geological aspects of the Newark roof, see Ruckley 2009. Due to the restricted accessibility, it was necessary to exclude the roof over the northern part of the east range. It is hoped that further investigations of this roof will be conducted at a later stage.

5 Browne 1881, 502. Raised in Victorian Glasgow, Browne was trained as an architect, with a firm grounding in the classics, following late 19th-century European architectural educational custom; for his career, see Mays 1992.

6 This idea was put forward first in John Sinclair’s Statistical Account; see Forrest 1973, 685–6.

7 MacGibbon & Ross 1887–92, vol. 3, 425–31; Metcalfe 1905, 164; Tabraham et al. 1996; Lewis 1998, 905–7; McKean 2001, 174–5; MacKechnie 2009, 79–115.

8 Lewis did not investigate the mansion itself, but excavated the area enclosed by the barmkin. Mac-Kechnie, on the other hand, refers to Newark Castle in the context of reappraising the life and career of Sir David Cunningham of Robertland.

9 Surviving Scottish Renaissance and post-Renaissance woodwork is normally confined to the ceilings, walls and roofs, but seldom consists of fitments. For early woodwork at Newark Castle, see Howard 1995, 93; Bath 2003, 227; Tabraham et al. 1996, 23.

10 This is somehow different in England, where successful 14th-century Master Carpenters such as Hugh Herland or William Hurley gained a considerable social status. For important English carpenters, see Harvey 1948. Mylne, however, has been able to chart the careers of a small number of Scottish medieval carpenters: see Mylne 1900.

11 Apted 1966; Dunbar 1999, 160–6; Bath 2003.

12 This is not only the case with the long gallery, which was probably covered with a combed ceiling, such as that over Earlshall and Pinkie House. It also applies for the great hall, which probably possessed a suspended ceiling structurally comparable to that at Queen Margaret’s Bedroom in Holyrood House. For the Holyrood ceiling, see Apted 1966, pl. 27.

13 For early Scottish timber architecture, see Peddie 1883/4, 465–76; Hume Brown 1891, 84; Dunbar 1956/7, 81–92; Stell 1972, 153–84; Hay 1973, 127–33; Hay 1976, 28–39; Walker 1976, 52–64; Stell 1977, 153–84; Stell 1992, 75–80; Davis et al. 2002; Smout et al. 2005, 80.

14 Mylne 1900, 288. See also Paton 1957, xxxv–xxxvii and xxxix for information on early Scottish wrights.

15 The analysis of structural timber offers conclusive evidence on dating issues for the following published buildings in Scotland: Darnaway Castle (Stell & Baillie 1993); 64–74 High Street Brechin (Crone et al. 2004); the great hall at Edinburgh Castle (Crone & Gallagher 2008) and the palace and great hall at Stirling Castle (Crone & Fawcett 1998).

16 In particular, this applies to Scotland, where timber import was common from the late Middle Ages onwards. For further information, see, for example, Balfour Paul 1900, 82; Crone & Fawcett 1998.

17 The history of the Finlawstoune-Maxwell family, as far as Newark Castle is concerned, has been summarized in Browne 1881, 494–5. For the genealogical history of the Maxwells of Newark, see <http://www.stirnet.com/genie/data/british/mm4ae/maxwell01.php> and <http://www.stirnet.com/genie/data/british/mm4ae/maxwell04.php#top> [accessed 15 January 2012]. I am grateful to William Bortrick for this information.

18 For the dendrochronological analysis of the roof-timber over the central part of the north range, see Crone 2008.

19 This approach is not unusual in Scotland, given the ‘patchwork-character’ of 16th-century Scottish castellated architecture. Rather than designing new houses, the Scottish nobility generally extended existing buildings. See Glendinning et al. 1996, 41 and McKean 2001, 15.

20 Tabraham et al. 1996, 14–17.

21 Unlike Brown, MacGibbon & Ross (1887–92) suggest separate construction stages for the gatehouse and the keep. Further to this, the gatehouse comprises two construction phases. Its top floor had been significantly altered before Patrick Maxwell commenced his building operations, although it is difficult to suggest the exact date for the second construction stage. Stell has hinted at the difficulties associated with assessing architectural structures of 15th-century origin, stating that many buildings can only indirectly be attributed to certain dates and builders (Stell 1977, 153–54, 161). The type of gun loops in the gatehouse, however, point to a possible late 15th-century date, thus confirming the documentary evidence: Tabraham et al. 1996, 16–17.

22 Tabraham et al. 1996, 4.

23 Lewis 1998, 908.

24 See Stell 1977, 157–9; Tabraham 1988 and Zeune 1992, 119–51.

25 While separate great halls are a prominent feature of the castles of Stirling, Edinburgh and Darnaway, the extent to which they existed at lesser castellated structures such as Newark remains an unsolved question; see Stell 1977, 157–9.

26 Tabraham et al. 1996, 4. James IV’s visit to Newark Castle in 1494 may be taken as an indication that Newark possessed an appropriate space, that is to say a separate great hall to accommodate the king.

27 For the building activities of the Scottish nobility during the late decades of the 16th century, see Glendinning et al. 1996, 29–70 and MacKechnie 1994.

28 For the custom of using horizontally arranged apartments in Scottish Renaissance architecture rather than vertically organized units, see Howard 1995, 79–86. For the disappearance of the great hall at the end of the Middle Ages, see Thompson 1995, 176–82. For typical design principles of Scottish medieval castellated architecture, see Zeune 1992.

29 Stell has remarked on the traditionally adopted Scottish solution, pointing out that early Scottish roof construction tends to employ the walls of gables or towers to meet individual roof compartments: Stell 1992, 79–80.

30 MacKechnie suggests Michelangelo’s Porta Pia in Rome as the model for some of the windows at Newark Castle, thus implying that Patrick Maxwell had access to up-to-date contemporary European architecture: MacKechnie 1994, 38. Du Cerceau’s bird’s-eye view of the Château of Bury shows an edifice characterized by four wings around a courtyard, with a roof using dormers and coned turrets as well as applying hip- and valley-rafters: see Bentley-Cranch 1986, 88. The castle of Anet and the Tuileries, both designed by Philibert de l’Orme and depicted by Du Cerceau, also possess valley rafters. For these buildings, see Ward 1909, pls xiv, xvii and xvix.

31 For a discussion of assembly marks recorded in Scottish roofs, see Hay 1976, 31; Crone et al. 2004, 155 and figs 3, 12, 13; Stell 2004, 21; Hanke 2004, 35; Newland 2007, 35.

32 Systematic analysis of photos held in the archive of the RCAHMS yields overwhelming evidence for the predominance of rafter single roofs in Scotland. A survey covering the counties of Lothian and Fife, for instance, has shown that more than 60% of the photographed roofs can be classified as rafter single roofs, while purlin roofs, solid stone structures, trussed roofs and cruck roofs account for the remaining 40%. For the typology of roofs in south-east Scotland, see Hanke 2006, 148, 197–202.

33 See Oldrieve 1916; Ruddock 1995; Hanke 2004; Crone et al. 2004.

34 For instance, the churches of St Martin, Sindelfingen (Binding 1991, 25), which possesses a roof constructed in 1132, and St Gangolf, Bamberg (Schuller 2004, 8) with a roof erected in 1185. England has late 12th-century roofs at Fyfield Hall (Walker 1999) and the Prior’s Lodge at Ely (Fletcher & Haslop 1969), while the Dutch roof at Raadhuissteeg 3, Zutphen (Janse 1989, 71) dates from the 13th century.

35 For early rafter single roofs of England, see Hewett 1980 and Fletcher & Haslop 1969. For early rafter single roofs of Germany, see Ostendorf 1908, Binding 1991, Eißing 1996 and Schuller 2004. For Dutch rafter single roofs, see Janse 1989; and for France, see Deneux 1927 and Hofsummer 2002.

36 Oldrieve 1916.

37 RCAHMS 1975, 108–9.

38 For Alloa Tower, see Ruddock 1995; for Bardowie Castle, see Stell 2004, 21; for Guthrie Collegiate Church and Tullibardine Chapel, see Fawcett 2002, 245–6.

39 For this roof, see Crone et al. 2004, 152–65.

40 Stell has mentioned a late 15th-century rafter single roof at Inverquharity Castle (Stell 2004, 21). However, no research has so far been undertaken on this roof. The same is true for the rafter single roofs over Claypott’s Castle, Dundee (also unpublished), which in all probability date from the years around 1569 (personal inspection). For the date of the Claypott roof, see Apted 1966, 107–8.

41 For instance, Gordonstoun School, Moray and the Cross Keys Inn at Peebles (both 17th-century) are characterized by tiebeam-less couples with extended ashlar posts. For Gordonstoun School, see RCAHMS, G85123PO; for Peebles, see RCAHMS, DC22928.

42 For a discussion of the ‘Scottish’ rafter foot, see in particular Ruddock 1995. Trusses with extended ashlar posts appear, for instance, in a late 12th- century building in Dijon and in the abbey church of Kerkrade (12th-century): Courtenay 1999, 108–11.

43 Courtenay 1999, 107.

44 Machine-sawn softwood boards gained popularity in the second half of the 18th century: Emerton 2000, 14. Assuming that split and adzed hardwood boards may have been used over higher-ranking medieval buildings, such boards possibly also featured over Newark Castle.

45 The distance was measured between the inner surfaces of the lateral walls.

46 For Tullibardine, see Fawcett 2002, 243–6.

47 This detail is typical of German medieval roofs, while post-medieval and early modern German roofs usually show pegs that are flush with the surface. There has been little systematic research on this issue in Scotland, and it is still difficult to employ the typology of nails and pegs as a tool for dating.

48 The distance was measured between the lateral faces of the rafters.

49 Trimmers also appear in the post-medieval roof at Brechin (Crone et al. 2004) and at Alloa Tower in Stirlingshire (Ruddock 1995, 299). However, it is impossible to say whether the 1480 version of the Brechin roof also employed trimmers. See RCAHMS FID/180/14.

50 See RCAHMS SC367340.

51 Much of the original timber used in the roof is in poor physical condition, and it is therefore difficult to identify holes and fixings. For a description of the condition of the timber, see Jatzlauk 2009.

52 Both roofs show assembly marks in the form of roman numerals. With the exception of truss 2 of the east range roof, which displays chiselled marks, all numerals have been cut using a saw. With few exceptions the trusses are positioned according to their numbering. Although some of the members have been assembled the wrong way (for example, ‘IIIX’ signifying ‘13’ rather than ‘XIII’), both roofs show a remarkably consistent pattern of assembly marks, showing that they had been assembled in one sequence. The numbering runs in both roofs from north to south, suggesting that the wrights started the erection process at the northern gable walls of the gatehouse and tower.

53 The typology of timber shown in Fig. 11 lists a significant number of redundant items in both roofs (Group 1: oak displaying redundant features, late 16th century or earlier). It is important, however, to emphasize that the gatehouse roof does not display any redundant mortise-holes. All redundant items relate to the vanished ceiling, thus suggesting that the wrights were using new material when they erected the gatehouse roof.

54 The northern part of the east range roof has not been included in the analysis due to inaccessibility. However, inspected from a distance through an aperture, all trusses appear to show the same structural composition identified over the southern part of the east range.

55 Browne 1881, 502; Tabraham et al. 1996, 2, 25; McKean 2001, 174.

56 Apted 1966, 27.

57 For the decorated timber panelling of these buildings, see Apted 1966, 67, 71.

58 Tool-marks visible on the underside of the southern rafters and their trimmers confirm the previous existence of a panelled decorative feature. Panelling of a supposedly similar kind can be found, for instance, at St Magdalen’s Chapel, Edinburgh, and at Culross Palace, both buildings erected c. 1600.

59 For the decorated timber panelling of Scottish Renaissance buildings, see Bath 2003 and Apted 1966.

60 For the painted timber decoration at Prestongrange and Pinkie House, see Bath 2003, 79–121.

61 While trimmers appear at buildings such as Alloa Tower and 339–43 High Street, Kirkcaldy (Ruddock 1995) and 68–74 High Street, Brechin (Crone et al. 2004), their extensive use at Newark Castle is unusual and clearly reflects Patrick Maxwell’s exceptional window and chimney architecture.

62 Ruddock describes a similar problem with regard to 339–43 High Street, Kirkcaldy, arguing that free lateral movement resulted in the failure of this roof construction: Ruddock 1995, 303–4.

63 This feature is visible, for instance, in the southern rafter of truss no. 13. Hexagonal nuts first appeared during the Industrial Revolution (Schuller, pers. comm.), so their employment at Newark points to a comparatively late date.

64 The building was ‘in weather-tight condition’ in 1881 (Browne 1881, 504). However, this statement does not prove that the re-sarking of the north range was completed by that time.

65 While recent dendrochronological analysis has shown that some of the timbers used over the north range were felled in 1598 (Crone 2008, 3), it is more difficult to date the dismantling and reassembling of the roof. In the mid 19th century the castle was in a ruinous condition (Barr 1845). Around 1880, however, it seems to have been ‘watertight’, although there was still space for ‘some much-needed renovation’ (Browne 1881, 504). Given that Newark Castle went into public ownership in 1909, the reassembling and re-slating of the roof is likely to have taken place in the following years.

66 The southern rafter and the lower collar beam at truss no. 13 have been replaced by using new material, and the sole-piece and ashlar post at the northern rafter of truss no. 17 are missing. Some of the solepieces in the southern wall had been found in a state of decay and were subsequently removed. This was not, however, a deliberate alteration but reflected the poor condition of the fabric.

67 Some of the sarking boards expose circular sawkerbs and hence should be attributed to the 19th or 20th centuries. Other boards have dowel holes that appear in a similar form in the reused rafters.

68 For the structural composition of these roofs, see Hay 1976.

69 European influence is visible in the classical detailing of the windows and the entrance gate. However, there is little reliable evidence that the wright responsible for the roof(s) took his inspiration directly from examples found in France or Italy. It is more reasonable to suggest that he was using indigenous methodology in order to solve a new architectural problem. It should be stressed that solutions achieved by Scottish roof technology of the 15th century differ markedly from the architecture of contemporary continental Europe. Neither the plain rafter single roof, nor the merging of barrel vault and roof structure, as seen for example at Corstorphine Parish Church, can be found in comparable abundance on the Continent. For the source of the classical detailing at Newark Castle, see McKean 2001, 175; for the conservatism of medieval Scottish roof architecture, see Hanke 2006, 2.

70 Crone 2008, 7–8. Dendrochronologically dated items are shown under Group IIIa in Fig. 11.

71 In the past, little attention was paid to the use of oak in late 16th-century Scottish domestic architecture, but recent studies have established its use in a number of Scottish buildings of the time. I am grateful to Anne Crone for information on the employment of imported oak in 16th-century Scotland, for example, at Fenton Tower (c. 1577) and Dunbar Town House (c. 1530).

72 Anne Crone, pers. comm. See also Crone & Mills 2002.

73 For the provenance of the oak used over the north range at Newark Castle, see Crone 2008, 8.

74 For a detailed discussion of the timber used, see Kirkdale Archaeology 2008.

75 Crone 2008, 6.

76 In his study on Alloa Tower, Ruddock refers to Baillie’s assumption that Scotland had a regeneration of oak forests during the Middle Ages (Ruddock 1995, 297). It would be interesting to establish whether the timber used at Newark Castle belongs to such a population of trees.

77 Hay 1976, 31; Crone et al. 2004, 155, figs 3, 12, 13; Hanke 2004, 35.

78 The same kind of indication system was used in early written documents. The compilers of The Accounts of the Masters of Work, for instance, were also using roman numerals. See Paton 1957.

79 Davis 1954, 43.

80 For a description and depiction of tools used by 17th-century carpenters, see Montgomery 1970, 117–30, pl. 8.

81 The assembly marks visible in Scottish roofs erected from the 17th century onwards are almost always cut rather than scratched.

82 For the dated stones, see MacGibbon & Ross 1887–92, Vol. 2, 431; Browne 1881, 495; Forrest 1973, 685–6. For the entrance inscription, see Tabraham et al. 1996, 6–7.

83 The architectural elements attributable to Patrick Maxwell employ the same type of stone and are clearly distinguishable from the late 16th-century fabric employed in other parts of the castle. For a provisional geological evaluation of stone types at Newark Castle, see Ruckley 2009.

84 Tabraham refers to ‘one of the corner turrets in Patrick Maxwell’s new mansion … which obviously existed in 1584’ (Tabraham et al. 1996, 7). Most interestingly, the north range does not have a consistent pattern of walls. Indeed, some of the walls are considerably thicker than others — a phenomenon which points to the adaptation of existing structures rather than building from scratch. The material used also consists of an awkwardly mix of grey and red sandstone, with a concentration of red stone around the windows, turrets and dormers.

85 This would usually take place during the construction process on site in order to remove disturbing features.

86 <http://www.stirnet.com/genie/data/british/mm4ae/maxwell04.php#top> [accessed 15 January 2012].

87 Crone 2008, 7.

88 For a cross-section of the roof over the great hall at Bardowie Castle, see RCAHMS STD/60/5. For the date of Bardowie, see RCAHMS 1963, 254.

89 MacKechnie attempts to associate Sir David Cunningham of Robertland with the evolution of architectural design in a group of buildings in south-west Scotland during the late 16th and early 17th century, including Newark Castle: MacKechnie 2009, 84–7.

90 The original roofs over Stirling Palace, for example, did not comprise any hip and valley rafters; see Hanke 2007, 16.

91 See, for example, the south-east corners of Killo- chan Castle and Claypotts Castle: MacGibbon & Ross 1887–92, Vol. 2, 145 and 210.

92 This does not mean that Scottish medieval carpentry did not know of the application of longitudinal members. Cruck buildings, as well as the medieval purlin roofs over the great halls at Darnaway Castle, Stirling Castle and Edinburgh Castle, for example, employ longitudinal members. However, rafter single structures feature over the vast majority of buildings, and purlin roofs are an exception rather than the rule. See Hanke 2007, 11–12.

93 Newark Castle employs corbels for the support of joists in the first floors of gatehouse and east range. For the poor structural relation of centre-purlins and walls in the eastern bedchambers, see Jatzlauk 2009, 133.

94 This applies in particular to the joists, which span the western part of the eastern bedchamber and function as a substructure for the combed ceiling. The joists sit on two supporting timbers which are crudely nailed to the rafters. The entire construction is primitive, and there is no doubt that it was concealed by a ceiling. The eastern valley rafter, as well as one of the struts which braces one of the centre-purlins against the wall, bears marks that confirm the previous existence of a combed structure. For details related to the vanished ceiling architecture over the central part of the north range, see Figs 23, 24, 25 and 26.

95 The marks visible on these beams are not related to the assembly process of the 1598 roof, but appear on reused material.

96 Comparable examples in Scotland include the long galleries at Pinkie House (Apted 1966, 32, pl. 20) and Earlshall (Apted 1966, 33, pl. 32), and the painted timber ceilings at Skelmorie Aisle (Apted 1966, 23, pl. 24) and Culross Palace (Apted 1966, 30, pl. 28). Tudor England knew similar spaces (Coope 1986, 43–72). The profile of the galleries at Beaudesert (Tipping 1922, 99, pl. 116) and Candover (Tipping 1922, 172, pl. 202) closely resembles the long galleries at Pinkie and Earlshall.

97 In 1916, Innocent pointed to the rarity of hipped roofs in the British Isles: ‘In the Teutonic and Slavic lands of the Continent, such roofs [hipped roofs] are usual and evidently of early origin, but there appears to be no evidence for their age in this country. In the North they seem to be associated with the Renaissance, for in Scotland a hipped roof used to be known as an Italian roof. The rafters, which run up the intersection of the planes of a hipped roof are called hip rafters and in the older examples in the south of England they do not meet at the ridge, but finish below it, leaving a small gablet …’ (Innocent 1971, 90). The earliest literary use of the terms ‘hip’ and ‘valley’ appears to have been by Moxon in 1677. He gave the names ‘principal rafter’ and ‘sleeper’ to the hip.

98 The apertures inserted into the parapet of the keep closely resemble the windows in the north range. For dating, see Tabraham et al. 1996, 12.

This paper is published with financial assistance from Historic Scotland