1 Whitehall, Walter Muir, Spanish Romanesque Architecture of the Eleventh Century (Oxford, 1941), p. 45 Google Scholar.

2 Puig i Cadafalch’s career was the subject of an exhibition, centred on a collection of rediscovered drawings, mounted in Barcelona in 1989. The comprehensive catalogue Josep Puig i Cadafalch: la arquitectura entre la casa y la ciudad was published by the Fundación Caja de Pensiones, Barcelona, and is in both Spanish and English.

3 Puig, Josep i Cadafalch, , de Falguera, Antoni, Goday, Josep i Casals, , L’Arquitectura Romànica a Catalunya, 3 vols (Barcelona, 1909, 1911, 1918; facsimile edition, 1983)Google Scholar.

4 Whitehill, Spanish Romanesque, p. 46.

5 Catalunya Romànica, Vol. XI: Bages (Barcelona, 1994). The drawings include a plan at gallery level and a longitudinal section (by A. Mazcuftan and F. Junyent), and an isometric projection (byj. A. Adell). The plan and projection have also been reproduced in a paper by Durliat, M., ‘La Catalogne et le premier art roman’, Bulletin Monumental, 147 (1989), pp. 209-38CrossRefGoogle Scholar and, recently, in Stalley, Roger, Early Medieval Architecture (Oxford, 1999), p. 135 Google Scholar.

6 Dimensions are from Junyent, Eduard, Rutas Románicas en Catalunya/I (Madrid, 1995), p. 82 Google Scholar.

7 Whitehill, Spanish Romanesque, p. 47.

8 For photographs of the exterior before restoration and reconstruction see Puig, i Cadafalch, et al., L’Arquitectura Romànica Vol. 2 Google Scholar, figures 77 and 78 (which shows considerable over-building of the north apse), and Whitehill, Spanish Romanesque, plates 11 and 12.

9 An estimate from drawings and photographs — Junyent, Rutas Románicas, gives the height of the nave vault as 18.80 m but gives no dimension for the transept.

10 Further evidence for the staged construction of the nave is apparent in its masonry. At the level where the giant order is simplified, a band of carefully cut stones, yellowish in colour, is carried across its stepped profile. Up to this level the wallwork is of larger blocks, fairly regularly coursed. Above it, and carrying into the barrel vault, the masonry is of smaller pieces, laid more randomly. Generally the stones are of a russet-colour that gives the interior of Sant Vicenç its distinctive, sombre glow. The stonework in the upper levels is darker and shows signs of discoloration, possibly a consequence of the military occupation.

11 Puig i Cadafalch’s adoption of wall arches for Sant Vicenç de Cardona may have been prompted by his familiarity with the church of Sant Ponç de Corbera, a few miles to the north-west of Barcelona. This building, for which there is no documentary evidence as to its date, has a single nave, barrel-vaulted and with transverse arches carried on pilasters. From pilaster to pilaster there are deeply articulated blind wall arches, so that ‘the engaged supports resemble halves of cruciform piers’ (Whitehill, Spanish Romanesque, p. 107). Whitehill illustrates the building with a sectional axonometric (Fig. 60).

12 After this paper was drafted my attention was drawn to an article by E. C. Fernie which recognizes the errors in Puig i Cadafalch’s plan and axonometric projection, and also postulates a change of plan in the building, in part along the lines advanced in this paper: Fernie, E. C., ‘St Vincent at Cardona and the Mediterranean Dimension of First Romanesque Architecture’ in Buckton, David and Heslop, T. A. (eds), Studies in Medieval Art and Architecture presented to Peter Lasko (Stroud, 1994), pp. 24–35 Google Scholar (n. 6).

13 It may be objected that if the opening were intended to illuminate the sanctuary it should be splayed towards the crossing rather than the nave. As it happens, there is also a splay (but smaller) towards the crossing (see Fig. 4). Since the opening is made above the springing of the dome, a wider splay on this side might have presented structural problems. The situation, represented here in the computer reconstruction of Figure 13b, is shown very well in the long-section reproduced in Catalunya Romànica. Given the tightness of the geometry (for a window above a supposed ‘roof’ but below the crown of the dome), the section adopted is a reasonable compromise for allowing illumination from the west to the crossing and the sanctuary. It is relevant that the windows in the apses, in the crypt and in the straight-sided section of the sanctuary all have external splays — as can be seen in Figure 2. Indeed this feature seems to be a hallmark of what is proposed here as the first phase of construction.

As it stands, the only lighting at the crossing is through the oculus and small openings made in the dome to the north and east. It is possible that the latter were inserted after the change of plan, and that the first intention was to have only the one, much larger, window to the west.

14 Puig, i Cadafalch, et al., L’Arquitectura Romànica, Vol. 2, p. 168 Google Scholar: ‘Avuy aquest espay està omplert de terra y reble ab motiu de les obres de fortificació modernes, lo que fa impossible coneixer ab certesa la disposició de la coberta dels colaterals.’

15 Sitjes, Xavier i Molins, , Esglésies Romàniques de Bages, Berguedà i Cardener, (Manresa, 1986), pp. 84–96 Google Scholar.

16 Sitjes argues for a third, early twelfth-century stage of construction, in which towers, façade and atrium (with gallery over), were added at the west end of the nave. However, his longitudinal section and plan, though they support his argument, are wrong in important and relevant respects. In the section, the west side of the westernmost giant order is shown without the outer step that is present throughout the structural system in the nave and, similarly, there is no stepped element against the inner face of the façade. This accords with Sitjes’ view that this western bay was not part of the same campaign of building as the lower parts of the nave. But, in fact, both stepped elements are present here and they can be easily seen, flanking the arched openings that give access to the gallery, to north and south. So it must be concluded that the western termination of Sant Vicenç was integral with and substantially built at the same time as the nave arcade; and if, as Sitjes accepts and there is no reason to doubt, the arcade was erected during the first campaign that included the complete eastern complex then it follows that the whole church as we now have it (except, perhaps the western towers) was laid out and constructed at least to a certain level before the consecration of 1040.

17 If there had been a lengthy gap between the two building campaigns it would not be possible to say whether the nave was then left open to the skies or whether it was given a temporary cover, but one interpretation of the distinctive band of masonry in the giant order (see note 10) is that these carefully cut stones were introduced as pad stones for seating the principal members of a timber roof.

18 Groin-vaulting is not alien to the first stage of building. It is used in the crypt and narthex.

19 E., C. Edson Armi, ‘Orders and Continuous Orders in Romanesque Architecture’, Journal of the Society of Architectural Historians, 34, No. 3 (October 1975), pp. 173-88Google Scholar.

20 Radding, Charles M. and Clark, William W., Medieval Architecture, Medieval Learning: Builders and Masters in the Age of Romanesque and Gothic (New Haven & London, 1992)Google Scholar.

21 Panofsky, Erwin, Gothic Architecture and Scholasticism (Cleveland and New York, 1957 Google Scholar; first edition 1951).

22 Radding and Clark, Medieval Architecture, p. 47.

23 Ibid., p. 14.

24 Because the contemporary building of the rich foundation of Santa Maria de Ripoll displays in its nave (now rebuilt) none of the sophisticated integration of Sant Vicenç de Cardona, Puig i Cadafalch concluded that the latter could not have been the work of local masons. He suggested that workmen were imported from northern Italy and looked there for his precedents. He cited as an example of like work the Ligurian church of San Paragono at Noli. It is true that in this small church, as in Sant Vicenç, the walls of the sanctuary and major apse are lined with niches and that the sanctuary is raised over a groin-vaulted crypt. The nave piers, too, have stepped articulations, but there is little evidence of the role they were intended to perform. San Paragorio features in Piemonte-Ligurie Roman (La Pierre-qui—Vire, 1979), with text by Duilio Citi.

Radding and Clark draw attention to the case of the cathedral at Speyer but since in the phase of that building contemporary with Sant Vicenç the giant order in the nave was not tied to a vaulting system it is not entirely relevant. For Speyer see the three-volume work by Kubach, Hans Erich and Haas, Walter, Der Dom zu Speyer (Munich, 1972)Google Scholar and Winterfeld, Dethard von, Palatinat Roman (La Pierre-qui-Vire, 1993)Google Scholar.

More recently, E. C. Fernie’s article ‘St Vincent at Cardona and the Mediterranean Dimension …’ revives Puig i Cadafalch’s original suggestions of (1) a Ligurian source for the design of Sant Vicenç and, more fundamentally, (2) a Byzantine influence over both the Ligurian and Catalonian monuments.

25 Pächt, Otto, ‘Art Historians and Art Critics VI — Alois Riegl’, Burlington Magazine 105, May 1963, p. 190 Google Scholar.

26 Dodds, Jerrilynn D., Architecture & Ideology in Early Medieval Spain, (University Park and London: Pennsylvania State University Press, 1989), p. 3 Google Scholar.

27 Puig, i Cadafalch, L’Arquitectura Romànica, Vol. 2, p. 476 Google Scholar, Fig. 405.

28 Catalunya Romànica, Vol. XI, Bages, p. 156.

29 Note that Puig i Cadafalch’s plan detail (Fig. 9b) is inaccurate in two respects: the face-widths in the articulation of the cruciform pier are much more differentiated than he shows (compare Fig. 9a) and he indicates, wrongly, that the outer order of the wall articulation and the outer order of the nave arcade are linked above by an arch (again, Fig. 9a more accurately represents the situation).

30 A combination of groin vaults with a three-stage modelling of the aisle wall, similar to the arrangement found in the narthex of Sant Vicenç, is to be found at San Paragorio at Noli, which Puig i Cadafalch proposed as a source for the Cardona building (see note 24 above).

31 It may be argued against this hypothesis that in none of the other Catalonian churches listed here is there a similar association of a triple articulation of the aisle wall with a half-barrel vault. But of these, only the late church at Gerri (consecrated I 149) has a fully-developed cruciform nave pier.

32 Durliat, in ‘La Catalogne et le premier art roman’ (pp. 225–26), sees a measure of improvisation in the vaulting system of the aisles, with decisions being taken ‘sur le tas’; he comments that ‘tous les problèmes posés par le système architectonique de la basilique voûtée n’étaient pas encore pleinement maîtrisés.’

33 Overall dimensions are taken from Junyent, Rutas Románicas (p. 82), but interpolations have been estimated from photographs. Note that in the sectional projections (as in Puig i Cadafalch’s corresponding diagram), the representation of the sanctuary does not take into account that it is raised over a crypt.

34 Panofsky, Gothic Architecture, pp. 50-51.

35 E. C. Fernie, ‘St Vincent at Cardona and the Mediterranean Dimension …’, p. 25.

36 Heyman, Jacques, The Stone Skeleton, Structural Engineering of Masonry Architecture (Cambridge, 1995), p. 51 CrossRefGoogle Scholar.

37 Fitchen, John, The Construction of Gothic Cathedrals (Chicago, 1981), pp. 46–47 Google Scholar.