top of page


June 2020

Victoria: - Did you invent the pattern?

Mark: - I don’t know. I don’t recall one like it, but no doubt it’s been done before!


This pattern for a stone floor in a loggia is new in the sense it was neither based on a specific precedent, nor developed by hunting for examples to imitate or vary. It riffs off past experience and encounters, with following general principles of Roman floor patterns. The main field or body of the pattern does this in two respects. First is the ad quadratum rotation of squares. Second is the alternation of units, rather as white and black squares alternate on a chequerboard. 

Design Evolution


1) Idea 


Having manifested itself spontaneously, the idea of a stone carpet attracted us because the eastern associations of carpets seemed to suit the palm capitals bounding this particular space, a type which finds antecedents in ancient Egypt and Assyria before becoming employed in a minority of Greek and then Roman architecture. It happens that the client is fond of carpets and Islamic ornament (which had a Roman underpinning), and we have a fondness for the theories of Gottfried Semper. Though critical of the common notion that Doric ornament were born of petrifying timber detailing, Semper highlighted examples of textile patterns becoming petrified, as in the case of an Assyrian relief in the British Museum [Semper / Mallgrave, p. 139; see also p. 306]. There are plenty of mosaics from the Roman, late Roman and early Islamic periods that recall carpets in one way or other, including the wonderful Umayyad Palace of Hisham in Jericho, where several different patterns jostle for attention like carpets and kelims in a bazaar. We were also curious to experiment with different types of finishes (rough versus polished), as well as the inherent contradiction of an outside carpet and one cut in marble, just as was Roman opus sectile. 

2) Premises


Once the carpet idea manifested itself, the basics of the design quickly became clear. The dominant portion would be the main field, wrapping around a rectangular centrepiece while itself being wrapped around by a border. Rather than follow the typical rhomboid geometry so often used for kelims, the field was set out on a square grid so as to conform with Roman marble floors, and also to simplify cutting. We chose the number of modules so as to balance visually the effects of repetition and the legibility of the module. We opted straightaway for a border, a motif of both Islamic carpets and architectural crenellations, including from the Assyrian period. The use of a square-based pattern dictated 45 degree symmetry at the corners, and so necessarily a centrepiece with a narrower aspect ratio than the carpet as a whole, creating an agreeable tension. Wishing to avoid treading on a figurative composition while privileging abstraction we chose to use the marble as the ‘picture’. 


initial sketch


scheme for the fabrication of an opus sectile pavement

"Modern architects often like to create the impression of coming up with their own ideas by virtue of autochthonous inspiration – perhaps out of the blue or conceptual logic, while doing their best to cover the traces of existing precedent since this might open them up to the heinous charge of copying. We like to create our own designs too, but we also like interrogating ourselves honestly as to where the ideas came from, and to document the extent they were derived from precedent, knowingly or unconsciously. It’s the quality of the outcome and of the interaction between tradition and novelty that excites us." 

3) Design Development


While there are countless variations of Roman floors based on square grids, the choice to use ad quadratum and offset squares came instinctively, since both geometrical concepts go back to Mesopotamian times. Later they migrated West into Greek and then Roman work, and later still they migrated back East into Islamic patterns (while ad quadratum continued to dominate medieval design in Europe). 


The progressive rotation of squares brings notable practical advantages, ensuring efficient production and use of materials since smaller pieces are made by bi-secting bigger pieces, with next to no wastage [MWJ 2000, pp.90-91 Add fig.?]. The offset pattern on the other hand seems to have been Apollodorus of Damascus’s favourite, being used in both the Forum of Trajan and the Pantheon, in combination with the other key principle of alternation. The floor of the Pantheon does this, although here one type of square contains an offset square, the other type roundels. These further sub-divide into two kinds, discs of porphyry alternating with discs of grey quarries (also from Egypt, this time from Mons Claudianus). [Williams 1997; MWJ 2000]


At first we experimented with subdividing the squares by just one rotation and one offset. This created unsatisfactory adjacencies with more than two colours, so we fixed on two rotations and two offsets. Next we decided to make the offset squares smaller by √2, since this produced exactly the same size of smaller square in all cases. (It was then that Victoria asked “Did you invent this?”) 

VSD AutoCAD Floor.png

Next we played with grid dimensions, the size of the carpet relative to the space that contains it, the nature of the jointing around its perimeter and the rhythm of the crenellation / tassles (which were made to align with every other square so as to ‘vibrate’). This and the colours were changed and adjusted by trial and error according to the effect on the eye. We were aiming for many different ways to read the pattern. Fine tuning involved thinning the framing bands, while testing the development and the surface treatment of the different materials in draft renders. In order to make the pattern more ‘carpet-like’ we use a matt, honed, finish for the carpet pattern, whilst the floor around has a reflectively polished finish. 


Finally, we experimented with a few subtleties. One was to differentiate the bands into two slightly contrasting colours so as to prompt the eye to zoom around as if in a labyrinth. Another was to translate the same design, more or less, in mosaic, which if gridded could recall the weft and warp of woven textile. Then Jakub suggested basketwork allusions instead, which we incorporated despite the latent contradiction in combining this on the diagonal with weaving orthogonally. It makes for an intriguing contradiction to add to the inherent contradiction in the idea of an stone carpet – and it looks good.





Balmelle, C. et al 1985 / 2002 > Le décor géométrique de la mosaïque romaine: Répertoire graphique et descriptive.

Blake, Marion E. 1930 > 'The Pavements of the Roman Buildings of the Republic and Early Empire', MAAR 8: 9-159.


Dunbabin, Katherine 1999 > Mosaics of the Greek and Roman World.


De Franceschini, Marina 1991 >Villa Adriana, Mosaici, Pavimenti, Edifici.


Guidobaldi, F. 1985 > 'Pavimenti in opus sectile di Roma e dell’area romana. Proposte per una classificazione e criteri di datazione', in P. Pensabene, ed. Marmi Antichi: Problemi d’Impiego, di Restauro e d’Identificazione: 171-233.


Semper, Gottfried 2003 > Style in the Technical and Tectonic Arts, trans. Harry F. Mallgrave of Der Stil…[1861-3].

Williams, Kim 1997 > Italian Pavements: Patterns in Space.


Wilson Jones, Mark 2000 > Principles of Roman Architecture


bottom of page