Another Story The making of the Downland Gridshell – Ten Years on

Back in 1996 Steve Johnson was a consultant at Edward Cullinan Architects when a small project for a rural museum came up at the offices weekly meeting. Much of the practice were caught up with large projects, so Johnson enthusiastically raised his hand, though not suspecting the six year journey it would set him, the practice, and a whole team of engineers, carpenters, and museum staff, on. Revisiting that six-year experience, Johnson has written this fulsome memoir about the making of Weald & Downland Open Air Museum, today, Cullinan Architects best known building

Photo: Stuart Keegan
Richard Harris, the timber-supremo Buro Happold structural engineer of the time, eloquently described a gridshell as "a shell with holes in it". Leave it to a highly-skilled structural engineer to describe something as enigmatic as a gridshell in such simple terms. Leave it to an architect to turn the story of the Downland Gridshell into a personal tome.

Apologies in Advance

Egg shells, sea shells, and tortoise shells are examples of natural shells. Man made shells can be baskets, boat hulls, and aeroplane fuselages. Whereas most of nature's shells are made from a single continuously curving material, man made shells tend to be made from a series of long, straight, and bendable materials - usually wood or bamboo woven or joined loosely together and then bent to form a shell. Shell structures are extremely strong for the relatively small amount of material that goes into them. In the case of a gridshell, the amount of material is minimal and this is what makes designing and making gridshell structures such an artistic and technical challenge.

Setting the Scene

Much has been written about the Downland Gridshell project. This is perhaps because the building was the first example of a permanent gridshell being built in the computer age. Previous gridshell objects and buildings from baskets to boats to yurts were created through handed-down skills. Even the impossibly large and complex Mannheim Horticultural Hall was designed using forgotten tools like slide rules and calculators. That was in the mid-seventies when computers took up entire buildings yet were still not capable of doing even slightly complex structural calculations. By the time of the Downland Gridshell, the engineers had hardware and software at their disposal capable of analysing tent structures under pure tension but not much more than that. By the end of the six-year project, the engineers had modified their software not just to analyse stiff fabrics and shells under load and in compression but they also developed a way of using computers to guide the building’s makers through a construction process by setting-out a sequential set of three-dimensional maps showing all of the forming steps from where the structure would begin as a flat matt to its completion as a whole single structure with curves to die for.

Photo: Buro Happold
The reason that this was needed was because, unlike any other structure, a timber gridshell could be built with an entire building’s structure being assembled lying flat like a sheet of fabric and then formed into a curvaceous shape strong enough to make a permanent building. Although this sounds simple, it’s not, and this is a description, from a personal perspective, of what it took to build the UK’s first gridshell building.

I was a reasonably young architect working with Edward Cullinan Architects, a well-respected firm with very broad experience working with whichever material they felt appropriate for a given project. As a first foray into complex timber construction, they had recently built a timber pole gridshell forming the roof over a student accommodation cabin called Westminster Lodge for John Makepeace’s developing Hooke Park woodland campus in Dorset . As with all of Hooke Park’s buildings, Buro Happold engineers were instrumental as, in every case, the buildings were setting an agenda of making use of local timber and exploring new ways to build with wood. One of the first buildings to go up at Hooke Park, the Prototype Building, interestingly used timber in tension. This was not normal. Buro Happold collaborated with the German architect and engineer, Frei Otto, who was known for his German Mannheim gridshell, which involved a group of London-based Ove Arup engineers who later went on to form Buro Happold.  This project was followed by the completion of a large round pole workshop built as a parabolic fanned rib structure done with the firm of architects Ahrends, Burton, and Koralek. Again, Frei Otto and Buro Happold were involved.

After Westminster Lodge, a masterplan was drawn up for the site and a second Hooke Park building was developed by team Cullinan-Happold and this would have been the first gridshell built since Mannheim. Sadly, the project was cancelled but, within a matter of months, the pair of firms collaborated on a competition-winning plan for a new workshop and artefact store for the Weald & Downland Open Air Museum near Chichester. The aim was to create a 21st century building making use of the same types of materials as used on the historic buildings displayed within the 45-acre museum campus. This project would carry-over some of the ideas from the aborted Hooke Park project.

The museum proposed funding the project mostly through the recently-created Heritage Lottery Fund which had yet to finance a new-build project as opposed to restoration and conservation work. A permanent gridshell had also never been built in the UK so the team had their work cut-out in terms of convincing the museum, potential funders, and ourselves that the project was technically feasible within the £1.3m budget set by the museum in 1996.

Within the Cullinan office, decisions had to be made as to who would run the project at a time when far larger and perhaps more likely projects were coming through the door. These included a masterplan and new buildings for the University of East London in the emerging London Royal Docklands, a new campus for the Faculty of Mathematics at Cambridge, and a masterplan for Bristol Harbourside. This left the gridshell project looking small and relatively unlikely in comparison. Even though I wasn’t part of the original gridshell competition team, from my point of view the project looked irresistible and the decision was made for the office to do the project principally as a one-man effort and see how it went. What had been realised was that the project would require a lot of research and development time, had a small budget (but very big for the museum) with 100% of the money needing raising, and, due to its innovative nature and relatively conservative location, would most likely become a planning battle from hell if the project got that far. What hadn’t been realised was that, in Chris Zeuner (, the museum director at the time, we had an extraordinarily motivated, savvy, and well-connected client who was backed-up by a committed museum staff including an assistant director, Richard Harris - not to be confused with the Buro Happold Richard Harris mentioned in the intro.

So, we had a strong and motivated client, one of the best firms of timber engineers in Europe, and an architect hungry to innovate with sticks. The next step would be to expand the team to include a quantity surveyor and, crucially, a carpentry team willing and able to take-on a project that was untried and riddled with risk. After a short competition, the Green Oak Carpentry Company joined the team. Without them and their boundless energy, creativity, and wit, we never would have had the wherewithal to prove that the project was possible. The plan was to use the expertise of each of the team members as part of the core design team and this included the client, the carpenters, and the quantity surveyors, Boxall Sayer to help design the building. There is no question in my mind that every member added an ingredient to the design salad that made the project happen. Aside from being an unprecedented design and construction challenge, the other major oddity of the project was the fact that, being publicly-funded, we had to go out to tender for a main contractor. Whoever this company would end up being, they would be responsible for a minority of the construction work as the already-appointed carpenters, one of their subcontractors, would carry-out the majority of the work. Not only that, the main contractor would carry the burden of liability for the project including the carpenters’ work. This would end-up leading to a shaky start for the project which will be gone into shortly.

Image: Steve Johnson

Image: Steve Johnson

Two, not one, buildings

The Downland Gridshell is actually two buildings sharing the same site. The museum needed a conservation workshop where they could restore their tally of stored historic buildings and objects being stored in barns outside the museum at considerable expense. Preferably, much of this work would be brought to the museum and placed within view of the public as a way of better explaining what the museum did. In addition, the museum needed an artefact store to house the growing number of important objects and materials being collected by the museum. The decision was made to combine these two functions within a single project both of which were fundable by the Heritage Lottery Fund. The museum leased 45 acres of fields and woodland and, by that time, had restored and erected about forty buildings gathered from around the southeast of England. Where within this landscape the building would be placed was another big issue to be addressed.

It was characteristic of the museum’s bravery, that they were keen to see their new grand project built in a position where it could be seen. Clearly, the architects and engineers would have backed this idea had it not been for a planning authority which objected to such a large and unusual building being visible from the surrounding hills. In the end, a site was chosen on a parking terrace dug into a chalk hillside and screened with trees planted 15 years earlier by a young Singleton Museum later to become the Weald & Downland Open Air Museum. The chalk hillside worked to the project’s advantage compared with the relatively flat landscape of the rest of the museum. By digging the building into the chalk hill, the lower level Artefact Store would become an earth-sheltered space absent of damaging sun and, due to the sheltering earth outside the building’s walls, would require far less energy to heat and cool to protect the more delicate objects from rapid humidity and temperature change. To be crude, the Gridshell would sit on top of the Artefact Store like a bulbous timber lid over a masonry coffin. All of this would be hidden within the trees, albeit in the middle of a very attractive parking lot, and that placated the planners - at least during the early stages.

From left to right: Ted Cullinan - Edward Cullinan Architects (ECA) Michael Dickson - Buro Happold (BH)
Carol Costello - (ECA) Andrew Holloway - Green Oak Carpentry John Romer - (ECA) Analie Riches -
(ECA) James Rowe - (BH) Richard Harris - (BH) Chris Zeuner - Weald & Downland Open Air Museum
Steve Johnson - (ECA) (Photo: Buro Happold)
Having selected a site, we were off. To gain further confidence, and after a session of intense initial design, the team held a weekend of stick-bending at the museum in the full view of the public. The carpenters assembled a kit of parts including cleft sweet chestnut laths (long sticks), some round poles, sacks of plastic agricultural ties and screws, and bits of plywood. These would represent all of the components required to make a gridshell. In one day a fistful of architects (including Ted Cullinan), another fistful of structural engineers (including Michael Dickson of Mannheim fame), and a pair of carpenters assembled long laths, wove them into a grid, and tortured them to see how they would react. They were then clamped with plywood and screws to two round poles jointed to form an L. Voila - a timber gridshell! With that under our belts, the time came to get serious about finding the right timber and a main contractor.

E.A. Chiverton were a family-run Chichester-based company headed by an architect called Mike Wigmore. They had already done work at the museum so were familiar with the local procurement and approval systems. They charged into the project and managed to grapple well with the unorthodox way that the design team proposed to build the building. The Green Oak Carpentry Company were still a small and relatively young company and the prospect for Chivertons of taking liability for the carpenter’s majority share of the work proved hard to swallow - especially on such an untested construction challenge. At one design team meeting on site we nearly lost Chivertons when they realised how the building was to be built.

The original proposal was to copy Mannheim which was built with the flat structure being assembled on the floor and pushed up and formed from underneath at some risk to the crew and the timber. Chris Zeuner, the museum director asked the simple question: why don’t we fabricate the flat structure in the air and use gravity to lower the shell to the floor? The question was answered with the drawing-in of a German scaffolding company called PERI who made scaffolding that could be erected and manipulated by the carpenters, with a bit of training, and featured telescoping poles with metric scales with turnbuckles that could be infinitely adjusted with great accuracy. This system was normally used for concrete viaduct construction so the piddly weight of our timber didn’t even register.

The second issue that worried Chivertons was the method for forming the flat gridmatt into the three-domed final shape that had already been agreed by the core design team. Some uncertainty existed as to whether the flat matt should be formed into a continuous barrel vault and then pushed into the three domes. Or, alternatively, whether we should begin forming the three domes from the start? Chivertons demanded that we should develop a way of doing the latter or they would withdraw from the project. To achieve this in a controlled manner was an extremely difficult undertaking as the means to accurately plan and then track the movement of the shell in space throughout the forming process didn’t exist. Buro Happold said they would go away and think about it and this kept the main contractor onboard. Within days, the engineers returned with an idea - rather excitedly.

With slightly tweaked existing software, they proposed using the thousands of lath crossing points within the gridshell as spatial references. Going from flat to the final form could be divided into nine stages to be measured by the carpenters and engineers as they progressed. All of this information could be worked out on the computer and plotted on familiar Excel spreadsheets for use by the carpenters like sheet music given to musicians. In addition, points could be drawn and numbered on the Workshop floor’s thick curving edge and coded in colour to show very visibly where each lath end should end-up. This system may never have been developed had the main contractor not dug-in his heels to protect his company - and, in the end, us all. With a construction strategy in place, we then turned our attention to finding the right timber for the job.

By this time, the Heritage Lottery Fund were becoming convinced that we weren’t off our rockers and were beginning to stage-fund the project with strong provisos attached. The HLF were very keen on the ethos of the project to make use of local timber and local skilled cratfspeople so we were obliged to look for local timber producers capable of producing the quality of timber we needed to build a permanent gridshell. The structure would have to withstand the stressful forming process, carry the loads when built, and resist fire and insect attack. To help select a timber for the gridshell, Buro Happold and Green Oak enlisted the expertise and equipment of the University of Bath’s engineering department who seemed to take a perverse pleasure in testing a variety of timber species to breaking point. Timbers included ash, oak, sweet chestnut, spruce, larch, Douglas fir, and birch - all commonly found timber in the south of England. To our amazement, most of the timbers tested similarly for strength except one. Oak tested off of the strength scale and had the added advantage of giving warning before breaking as opposed to abruptly snapping as most of the other timbers tended to do. This was comforting knowledge to the people working on site. Making it yet more attractive, the heartwood of oak tastes awful to nasty beetles and is actually quite difficult to burn.

60 miles – In all directions

For the gridshell, oak was our choice. For the rest of the building, almost all of the woods listed above were used somewhere in the project. Interestingly, the two Sussex counties are split tree-wise with East Sussex being abundant in sweet chestnut and West Sussex with oak. While we were scratching our heads at the museum in the west, another team were also scratching their heads in the east at the Woodland Enterprise Centre, Flimwell where another lath-structured building was in the process of being built. Their choice of timber was of course their plentiful coppiced sweet chestnut. We were both happy with each other’s decisions and happily traded notes throughout our respective projects. Where we in the west stumbled was over where to get our oak as our area had some of the most impressive oak stands in the UK. The trouble was that all of what was being offered was English oak with all of its eccentric swirling grain. This wasn’t good for building structures made from sticks as, when bent, the timber would break along its grain. We needed oak with the straightest possible grain and our carpenters knew where to find it. In France.

Photo: Steve Johnson
Within our Heritage Lottery Fund application we said that we would find all of our timber within the 60-mile swing of a compass from the museum. All we had to do was to swing the compass across the channel to include Normandy and Picardy which, amazingly is what previous carpenters had done over the past six centuries when they built many of the specimen buildings now located at the museum in French oak. It also helped that the pound was on the rise relative to the franc making our job of keeping within budget a bit easier. With French oak as a staple material our carpenters embarked on a programme of Medieval wood torture that made the University of Bath’s testing efforts look like an NHS procedure. Over a matter of weeks, small test models were built by the engineers while the carpenters got busy building full-scale mock-ups at the museum. It was an ugly sight seeing a beautiful section of gridshell being stretched, bent, and contorted on the very parking terrace that the building would be built. Despite their evil efforts, the oak never succumbed which was a delight to all.

As important as the choice of timber for the gridshell, was coming up with a way to join the four layers of lathwork into the grid. Mannheim was a simple grid with one lath laid over another lath oriented at 90 degrees. To make the tightly curving Downland Gridshell work, we needed to make the laths thinner to allow them to bend more tightly and then double the layers to make it strong again. We ended up with four alternating laths with two in one direction and two at 90 degrees. Fixing two laths only needed one hole with a bolt running through to allow the laths to scissor with each other once fixed. That is what makes it possible to form a gridshell. Adding a lath to the inside and outside of this meant that the laths would bend over different radiuses like the relationship of the inside and outside lanes of an athletics track. To join these would need slotted holes, which would make perfect breaking points when the timber was bent. We couldn’t take that risk especially knowing that the Mannheim project resulted in the breakage of 60 percent of its laths while being formed. Most breaks happened at bolt holes.

Photo: Stuart Keegan
I remember receiving a phone call from Andrew Holloway, at Green Oak, saying that he and his first mate, Stephen Corbett, had come up with an idea that could almost eliminate the need to drill through the laths. Their plan was to use a multi-layered sandwich of 100mm square steel plates with bolts run-through holes drilled through their four corners to avoid the laths. One stud at the centre would allow the central laths to be accurately located and held at their metre and half-metre grid centres. The outside laths would be left un-drilled but clamped loosely to allow the whole grid system to move when and where we needed it to. Buro Happold quickly analysed the “node clamps”, as we called them, to make sure that they would hold the shell once it was fully formed and the bolts were tightened to fix the gridshell. I tweaked the plate shapes turning them from simple squares to clover-leaf-like shapes. This was not a cosmetic decision as the revised shape meant that friction between the plates and laths would be reduced during the forming process as the plate edges would only be in contact with the lathes corners as opposed to along the whole lath face. Equally important was the crucial fact that, once the shell was formed and the clamps tightened to grip the laths, we had to make sure that, when the timber shrank with drying, the clamps’ grip wouldn’t be reduced. By making the squares into clover-leafs the corners could be torqued forcing the steel to bend like springs in anticipation of the timber shrinking. The plates would flex back to flat and maintain their grip even after the laths shrank. That was the idea anyway. In the end, the carpenters had to return to site some months later with a spanner, which was something of a surprise as Steve Corbett actually went out of his was and had the steel fabricators make full-scale node clamp mock-ups. He then stuck green oak laths in between, tightened the bolts, and roasted the lot in his kitchen oven overnight. It actually worked with the clamps and dried laths still tight the next morning. The things carpenters get up to in their kitchens.

That was our gridshell kit-of-parts gathered. Soon after, we received our order of French oak laths on the back of a single lorry from France. On first inspection, they were of outstanding quality - that was until Green Oak began removing the strapping binding them into tight bundles. Inside the external layers of near furniture-grade timber were large amounts of unacceptable timber. When questioned, the timber mill claimed that they hadn’t understood our specification and quickly rectified the problem by sending over an acceptable batch. The carpenters said that we had suffered an old trick reasonably common within the timber trade. For another use the faulty timber originally provided might have been usable but certainly not for our gridshell. In all, we had possession of about 12km of oak laths waiting to be jointed into the alarmingly long laths that would go into the gridshell. Green Oak and Buro Happold had become aware of a very special German timber-jointing machine that had miraculously been placed in the UK to show fabricators what the machine could do. What it could do was a lot of finger joints very, very quickly.

Finger jointing was new to the UK at that time and best described as two hands with bent and interlocked fingers with glue applied between to fix them together. Our Flimwell comrades, in particular a man named Nigel Braden (another architect-turned-timber-fanatic), had been working with a Swiss glue-maker called Collano and together they produced and tested a wood glue that could stick wet, flexible, green timber together; just what is needed to make gridshells. Buro Happold looked into the glue and Green Oak tested some joints, both finger and scarf, and decided that the glue was up to the job and the jointing machine people confirmed that the glue could be used within their jointing machine. Our next problem to overcome was that, although the machine was in the UK, it was all the way up in Newcastle. The beautiful French laths never left the back of its flat bed lorry and were sent north with Steve Corbett and a second carpenter to finger joint 12km of laths in just over a week. The now twelve metre lorry-length laths were returned to the museum for jointing to gridshell lengths of up to sixty-five metres long. This was done inside a polytunnel built as a semi-heated fabrication workshop next to a rather large and deep hole in the chalk where the museum had used its own staff to dig-out the Artefact Store hole giving the excavated chalk to a local farmer to surface a new farm track. The museum decided not to wait for the Heritage Lottery Fund to decide whether to fund the remainder of the project and simply got on with it.

It is important to note that there was an earlier crucial effort to get the building through planning. The museum knew that it wasn’t going to be an easy ride having been warned that the local conservation board liked the project but certainly didn’t like our building. In addition, word was getting around that the planning case officers were siding with the conservation group. Chris Zeuner and I attended the planning committee meeting armed with brightly-coloured drawings and one of Buro Happold’s stunning models. One of the committee members took us aside and warned that the committee vote could swing either way and recommended that, if it began to look ugly, we should ask for a deferral of the decision and go away and regroup. An awful lot of work had gone into the project and time was flying past. Finger nails were being bitten so Chris replied that we were in it come what may. The head of planning stood up and gave a robust account of why the building should be refused. Our friendly committee member gave a brave account of how positive the project would be for the area if approved. We elected to refrain from giving a deposition and remained silent. A vote was taken and the project was approved unanimously by the committee. The planning advisors burst out of the chamber and went straight to the press declaring that the process had been an undemocratic act. I never quite understood the logic of this comment but was more than impressed following the completion of the building when some of the planning officers and the conservation board explained to the museum that they simply didn’t understand what was being proposed and actually admitted that they quite liked the finished building.

Three years in

1999. Hole in the ground, laths on site, and a near geological plateau of scaffolding and decks in place - work commenced in earnest amidst the wettest winter on record followed by a crippling foot-in-mouth outbreak. The museum entrances were matted with chemical quilts and all delivery lorries had to be scrubbed coming-and-going. Anyone visiting the site had to douse boots in buckets of disinfectant. The museum was in a quandary over what to do with their beautiful collection of rare breed pigs, sheep, and horses - all hugely popular with the public and now at great risk of infection. In the end, a decision was made to keep the animals where they were and the sun eventually broke through the clouds. Green Oak and Chivertons were soon in full flow. Even the scaffolding company had erected scaffold towers off the two ends of the rising building ready to film the gridshell process.

Photo: Steve Johnson
The assembly deck for the gridshell was connected to a higher parking terrace with a bridge giving the carpenters graceful access between their polytunnel and the deck 7 metres above what would become the gridshell floor. The triple hourglass shape of the floor and the two glue-laminated softwood and green oak parabolic gable end arches were barely visible through the forest of aluminum scaffolding. The polytunnel was a hive of fabrication and testing of laths being scarf-jointed and sent out the door to be laid-out over the awaiting scaffolding system designed by PERI and erected by Green Oak. One of the most wonderful and comical sights was the carpenter’s spray-painted curve laid out on the parking area road surface representing the tightest curve that the laths would have to withstand during the forming process. Each lath needed six people to carry it - not because of its weight - but to stop the incredibly long and flexible 35x50mm thick laths from dragging on the ground. One-by-one, the craftsmen munchkins followed the painted curve from the polytunnel around a bend up to the next terrace and onto the bridge to the deck safe in the knowledge that, if the lath could take being walked around this curve without breaking, it was fit for use in the building. The carpenters could occasionally be heard singing while doing this repetitive dance.

Each lath was numbered and meticulously laid onto the scaffold beams to match Buro Happold’s Excel spreadsheets. The laths were colour-coded with dye to remind the carpentry team of where the areas of the shell were that would be most highly-stressed. Unfortunately, these marks can’t be seen inside the completed building as they all face outwards. Over many days of laying-out, the grid matt was complete with four layers of pre-cut laths with loosely-fixed node clamps all in their exact positions hovering seven metres in the air.

Photo: Steve Johnson
To give an idea of what we were facing, as part of an exhibition at the museum, we laid a 35-metre length of our lath on an open hillside next to the site. We knew that a shin-breakingly stiff metre length of 35x50mm oak became, at 35-metres long, a virtual piece of rubber. Give one end a flick and a sine wave would travel down the thirty-five metre length to the opposite end like a flicked skipping rope. What shocked us about the assembled thirty-five metre wide by sixty-five metre long gridmatt was that this incomprehensibly elastic individual lathwork had been transformed from a rope into a two-dimensional sheet of timber weighing several tonnes and resting upon a scaffold forest. You could grab an edge and flick it and the whole grid surface reacted in liquid waves. Here was the entire structure of a reasonably large timber building suspended on aluminum poles and timber beams behaving like a living, reactive being. I couldn’t tell from the carpenter’s reaction whether this was a good thing? Nevertheless, the prospect of having an entire building lying flat and ready to be formed into a predetermined shape that would eventually house years of activity and events was very, very alluring.

Extremely sadly, before we had reached this point in the project, we had lost one of our crucial team partners. The museum’s director had very suddenly died of cancer. The Downland Gridshell was one of Chris Zeuner’s biggest dreams for the museum and, unexpectedly, he died midway through its construction. It was down to his brave wife Diana to crack a bottle of champaign on the awaiting gridmatt to begin the forming process.


The next day the carpenters set to work as grand sculptors. Beneath the forest of scaffolding was the mostly-completed shell of the Artefact Store with its reinforced concrete block retaining walls and Swedish glue-laminated beam roof. This roof was about to become the floor deck of the gridshell space. Frustratingly, when we began designing the project in 1996, there were half a dozen companies within the region capable of making our reasonably simple softwood beams. By the time we were ready to place an order, all of the fabricators on our list had been bought-up by Scandinavian or German manufacturers and shut-down and this caused us to go beyond our sixty-mile procurement radius. The floor beams were cut to length and centred at two metres running the ten metre span between the Artefact Store’s walls.  The ends of the beams cantilevered beyond the block walls to lengths required to form the beautiful triple hourglass shape that the floor needed to be to set the parameters of the eventual gridshell. I was amazed to see that the ends of these beams had been sculpted by the carpenters, using a special electric chain saw to match the exact curve of the gridshell form. The beam-ends would form an interface with a half metre deep twisting band of plywood that would be glued and screwed to the ends of the half metre deep beam ends. This formed a surface against which the gridshell laths would be lowered and then fixed. A second layer of matching plywood would eventually be run parallel with the first to the outside of the four layers of laths and fixed to the floor’s edge. As mentioned earlier, the second devices used to turn the flat grid structure into its final shape were the two vertical parabolic gable arches at the two ends of the building. The parabolas were made from softwood by the same Swedish company that provided the floor beams. The infill framing within the arches that were to provide bracing and to form the window and door openings were made by Green Oak from green oak using traditional oak-pegged joints. This is where things began to get very interesting as far as The Green Oak Carpentry Company were concerned.

Photo: Steve Johnson
This company was made up of a range of carpenters with varying skills and interests. Some were young trainees while some were older trainees often coming from desk jobs and wanting to use their hands for a change. Others were extremely experienced carpenters with a passion for innovating with carpentry while others thrived off of traditional carpentry techniques and conservation work and liked nothing better than working on an old timber barn studying ancient joints and carpentry marks left behind by craftsmen many centuries earlier. In the case of the Downland Gridshell, we needed carpenters with all of these interests and skills and Green Oak split their workforce into two teams with one group taking on the gridshell and the other handling the more traditional work like the gable arch infills and, later on, the protective porch roofs over the ends of the building. The latter also produced the exposed Douglas fir door and window sub-frames that went into the Artefact Store and served as formers for the masons building the blockwork cavity walls. The walls were all a two-leaf construction of dense blockwork with a 100mm cavity that was fitted with reinforcing bars and later filled with concrete. With the frames standing in place, the masons laid their blocks around them and, once formed, the cavities were filled with concrete and the frames worked as stops keeping the concrete in place. Once the concrete had set, the timber frames stopped carrying the weight of the concrete and served simply as sub-frames to which the door and windows would be fixed. Chiverton’s team did such a good job of pouring the concrete that the frames only needed to be sanded and oiled to get to their finished state.

On top of the glulam floor beams was laid an 80mm thick tongue-and-grooved sub-floor of English spruce screwed to the beams to create a very strong and stiff floor that would both deal with the forces of the gridshell and provide a floor strong enough to take the weight of a loaded truck or forklift. Like the glue-lam beams supporting this deck, its thick edge was skillfully cut to the gridshell form using the same electric chainsaw used earlier. On top of this strong deck went a waterproofing layer, insulation, and a layer of cork to provide a bit of acoustic insulation between the Workshop floor and the Artefact Store below. The neighbouring West Dean Estate generously donated enough seasoned ash and oak from their own woodlands to finish the Workshop floor.

The process of forming the gridshell was the real nail-biter as the team really didn’t know how the grid would react to the controlled lowering and the removal of the scaffolding system beneath. The six carpenters, many of them boat builders from Portsmouth and Southampton, were conducted by Steve Corbett. Each telescoping scaffold upright was numbered and their metric scales and screw couplings allowed the carpenters to lower the system as set-out on their Excel spreadsheet guides. Remember that the strategy was to go from the flat gridmatt straight towards the three domes that would become the final building. They worked as a team making fine adjustments according to the charts listening to the timber as they went. The computer charts told them where they wanted to go with the shell but how quickly they moved was left to their judgement as craftsmen.

Photo: Steve Johnson
Photo: Steve Johnson
Thinking that the gridshell might need coaxing into its final position, Green Oak fixed three steel brackets at the centres of the three domes in case they had use ropes to pull the shell into position against the floor’s forming edges. These were never needed as gravity did all of the work. About halfway through the lowering process though, the sweep of the falling timber grid would conflict with the upper ground terrace which had to be excavated to allow the shell to pass by. Not only that, with the shell reaching its halfway mark came the realisation that the timber would have to be supported at awkward angles. To my and the scaffolding company’s delight, Green Oak made a series of steel hand-shaped grippers that could be attached to the ends of the scaffold poles with the poles angled out at varying angles to grasp and guide the outside edges of the shell into the awaiting sides of the floor edge below. Not only was the scaffolding system being used to support and guide the shell while shaping, it was now being used to help form the shell!

In all, it took six weeks to form the shell and the carpenters estimated that about 40% of their time was spent shifting scaffolding - something they were convinced they could improve upon if ever they were challenged with another gridshell. Once in place, the carpenters and the Buro Happold man-on-site, Oliver Kelly, set-out to survey the lowered shell. The steel node clamps that held the corners of the grid were used throughout the forming process as reference points to be measured back to the workshop floor as they progressed with the lowering. Photo 9.19 The team’s greatest fear at this point was that the shell might be so far out of shape that the timber would have to be manipulated to get it into its correct form. Trying to make major adjustments like that would have put serious strain on the timber and possibly have broken joints. What a relief when we learned that, at worst, the shell was out by only 50mm over a maximum span of fifteen metres across the floor and 8.5 metres in height. This sort of adjustment was easily dealt with with a bit of pushing and pulling before it was locked to the floor’s edge. Sitting back with the shell in its right shape, I had to appreciate the fact that, even with the computers and Excel guides, much of the forming process was done with great physical skill, attention, and fine eyes. Many of the carpenters working on the project were attracted by the drawings and models showing a building that was, in effect, an upturned boat. The crew’s boat-building skills has served their purpose. These same skills would prove useful a second time in a few months time.

The gridshell – Naked and dressed

Photo: Stephen Corbett GOCC
Photo: Guy Bagshaw GOCC
In the middle of the woodland on a former parking terrace now stood a bare timber gridshell. People were watching the process all along but now word was getting around and a fairly constant flow of onlookers were venturing through the trees to see this enormous basket. A common comment was: What a shame this beautiful thing had to be clad. Two of the carpenters even camped-out on site in order to take a series of flash-enabled photos of the naked gridshell.

One of the fascinating early lessons with this project came at a point when Buro Happold had built a first test model of the shell out of copper wire mesh. Half of this model was left as bare mesh while the other had a series of cross wires soldered across its surface working to triangulate the shell. When a finger pressed on the unbraced half, the shell sponged like a Christmas pudding. When done to the other half, it didn’t budge. With Mannheim, steel cables were fixed across the corners of the timber grids running in both directions like a X-ed boxes. With these cables, each grid became locked. Over an entire shell, the unbraced structure would be like a sponge. Braced, it became like rock. From the outset, the team was keen to reduce the metals from the system as much as possible and this included the cable bracing as used on Mannheim as well. Instead, we made a big intellectual leap and decided to make use of the same oak laths that the gridshell was made from. Rather than using cable that only works in tension, using stiff timber meant that bracing would only be necessary in one direction. If we were making a basket of a building, why not make it as basket-like as possible?

When designing the steel node clamps, we included a threaded bolt welded to the centre of the outside of the outermost plate. This allowed us to run our bracing laths over the gridshell after it had been formed with holes to match the node clamp bolts. Over the lower two-thirds of the shell, bracing laths were run horizontally from end-to-end. Over the upper third, laths were run across the shell. Although being able to run all of the laths in the same direction, it felt intuitively right to run them in both directions. Once mounted it looked right as well. Furthermore, and probably most importantly, we knew that the cladding system had to be fixed to the shell in some way and these laths provided the answer to how this could be done - the gridshell was telling us how it wanted to be dressed.

Ted Cullinan was determined from the outset that he would not allow our building to be another stocking-clad building like the Mannheim building. As a permanent building we needed to clad the building in materials that would last while doing several tasks. The hardest parts of a shell structure to protect are the upper low-pitched areas where water is most likely to get into the building so we wanted a watertight membrane at the top of the building. We also needed to make sure that plenty of sunlight would enter the space without cooking its inhabitants. Ted’s initial sketches showed a cross section that looked a bit like an armadillo with a flat board balanced on its back. Originally, the museum’s carpenters said that they wanted the gridshell workshop to be a covered yard with plenty of air-flow to keep them cool and their timber dry. These are the reasons why the building ended up with its characteristic flat ribbon roof, polycarbonate clerestories, and three bands of vertical timber boarding. Photo 9.17 The flat rolling roof runs parallel with the top of the centre of the shell protecting the low pitched tops of the domes and allowing access to the roof for installing the clerestories and maintaining the building. Insulated polycarbonate panels set in aluminum bars were used to make the clerestories. The bars were fixed to the node clamp points and the bendable polycarbonate panels were measured and cut on site with the entire system installed by two glazers in two weeks. One hanged from the ribbon roof mounts like a harnessed mountain climber on a rope. He measured the opening between the glazing bars and shouted the dimensions to his mate on the ground while dropping a rope to him. The ground man cut the individual panel shapes with a skill saw on a table, ran a rope around the panel and the hanging man pulled them up, and fixed them in place in minutes before scurrying over to the next section working his way down the length of the two sides of the building.

Photo: Steve Johnson
The installation of the all-important clerestories could only have been done after the ribbon roof was in place. This roof required a support system that could form the gap between the top of the curvaceous gridshell and the underside of the ribbon roof. Not to get too mathematical, this meant a transition from the geodesic structure of the of the gridshell to the parallel linear structure of the ribbon roof. I like explaining this as it sounds impressive. We had to devise a system to do this and, yet again, the gridshell gave us our answer. The node clamps became the locations of a series of five small round timber spacing pillars fanning in rows across the tops of the top gridshell bracing laths like standing hairs. Their lengths varied to maintain the flatness of the roof running across the building and connecting into straight roof joists also running across the building. These posts were mounted radially along the shell in both directions which meant that there were only five standard pillar lengths to make the entire roof. Once the roof joists were fixed to the pillars, the carpenters then added a series of flat timber battens to the tops of the joists to fix their positions over the length of the building. Tho thin bendable layers of plywood deck with a waterproof membrane and insulation created the rolling roof deck. The finished surface of the roof was a new cement-based liquid membrane called Roofkrete. This created a strong and permanent working surface that would allow foot traffic and, uncannily, would move and flex with the timber structure below - as it surely would. All timber buildings move.

Below the polycarbonate clerestory bands were mounted three horizontal belts of western red cedar vertically-mounted boards. Again, the horizontal bracing laths running the length of the building told us where the timber boards would stop and start. We knew the maximum lengths of boarding we could get from English woodlands and mounted a series of test boards down the outside of the shell to make sure that they could follow the shell’s curves while remaining as straight as Ted’s sketches had shown them. Trying to clad a building as curvaceous as the Downland Gridshell was difficult especially using the straight materials that Ted wanted. To make matters more difficult, the museum and engineers rightly put forward the requirement that, if the museum were to raise the sorts of money required to build this building, they should at least end up with a sealed and insulated building rather than a simple covered carpentry shed. Although this made perfect sense, it meant that we had to find membranes and insulation materials that could follow the double curvature and make sure that the building was reasonably airtight with no increase in the project budget.

Photo: Stuart Keegan

Photo: Stuart Keegan

Luckily for us, Roofkrete, a French insulation product new to the British market had come to our attention. This was an ultra thin multi-layered foil and fibre blanket that could be stapled to the outside of the bracing laths and vertically hung over the shell. Before this could be applied, however, we had to line the entire shell with a reflective membrane to help with insulation values while and to give us the required fire protection to prevent a possible fire from traveling down the length of the building. In addition to these wonder properties, the foil provided a shimmering internal surface within the workshop that bounces light from the clerestories throughout the space. This is enhanced by the shell’s curves that refract the light rather than simply reflecting it in a linear way.

On a normal project, Green Oak’s carpenters would make a building frame, deliver it to site, erect it, and then move on to the next project leaving the main contractor to clad and finish the building afterwards. Not so with this project. Instead, their skills were applied to the timber cladding and everything behind it. The glazing, joinery, rendering, and roofing was all done by Chiverton’s subcontractors and it was wonderful watching specific craftsmen carrying-out the work that they were best at. In the case of the timber cladding, Green Oak developed a method of applying horizontal battens to the outside face of the vertical boards and then using them as sawing guides. They trimmed the belts of board ends to the exact horizontal length in single runs down the entire length of the building. This is what makes the building’s cladding sing and silenced the critics who said that the clad gridshell would never be as beautiful as the naked one.

Photo: Steve Johnson
Photo: Stuart Keegan
While the cladding work was being done, another set of carpenters had to deal with the two porched ends of the building. The ribbon roof was designed to continue beyond the gabled end arches to form the building’s porches. Green oak joists were fixed running from the tops of the parabolic arches outwards to be picked-up by oak beams running across the outer ends of the porch roofs. These beams were in-turn supported on two sets of two splayed oak angle braces and large 200mm diameter laminated oak parabolic arches leaning away from the gable ends at 60 degrees. These arches would form the gateways into the two ends of the building. They would also support the three belts of timber boarding and their support laths as they traveled past the insulated walls of the workshop to meet the leaning arches so protecting the porches. This was very much in-keeping with the Cullinan tradition of oversailing cladding to express the material layering on the outsides of their buildings.

One major problem that none of the design team wanted to deal with early on was the actual shape of the leaning parabolic porch arches. We wanted to continue the downward curvature of the building’s roof and walls over the porches and tried to work-out the shape of this difficult leaning arch. All of my drawings were hand drawn and we thought it better anyway to wait until the late stages when most of the cladding was in place to try to work out the arch shapes on site. I was happy with this and, as our confidence in our carpenters was building by the day, we knew it could be done. Green Oak were still slightly fuming from having to go to the Swedes for our softwood glue-lams. Had they had the manpower at the time, they could have made them themselves but perhaps not as cheaply as the Scandinavians or Germans with their massive and technically-advanced industrial plants and limitless pine forests. Nevertheless, the leaning oak arches provided the carpenters with an opportunity to go well beyond expectations and this part of the story still makes me smile.

Photo: Steve Johnson
Photo: Steve Johnson
Leaning round-sectioned oak parabolic laminated arches exposed to the elements? No problem. We’ll get some dry French oak boards, hire a local mast-maker, and make them in the Artefact Store. Oh yeah - and just how do you propose to do that? Answer: we’ll use the magic PERI telescoping poles still on site to build a laminating machine of course. Within days of being given the go-ahead, the carpenters had measured the ends of the building and made a plywood template of what they estimated would be the right arch shape. They mounted it onto the eastern end of the building at 60 degrees awaiting an approval. In addition, they had extended the wall cladding laths out to meet the arch so that we could judge the curvatures. We held a grand pow-wow and agreed the shape. Shortly after, Green Oak invented and built a PERI-based forming machine within the Artefact Store between the space’s central timber posts.

This was truly a beautiful thing with its orange and silver PERI poles suspended horizontally on wooden blocks over the floor. They radiated to form a parabola in-waiting for the many layers of thin French Oak boards 250mm wide and many metres long to be laid into the machine. Special weatherproof glue was applied to each layer and clamped by the machine until the glue set when another layer was added. This carried on until they had a 250x250mm square-sectioned parabolic arch looking quite enormous within the Artefact Store. The mast-maker scribed an octagon in each end of the arch and, armed with a saw, turned the square arch into an eight-sided shape. He then scribed a twelve-sided shape and cut the arch again. This was done several times until he had a thirty-two-sided arch which, to the human eye, appeared to be perfectly round. This procedure, of course, had to be done twice and once the two leaning arches were in position, the Downland Gridshell’s cladding could be completed. The two gable ends of the building were glazed in transparent glass, which allowed for beautiful views out into the surrounding woodland. Very large top-hung central doors could be fully opened to make the building seem as if it were open-ended.

Fit-Out time

Photo: Steve Johnson
With the shell of the building being complete, the museum got on with the fit-out of the two levels. The workshop had always been planned to have two internal heated workshops placed at the east end of the building. These would be used as an office and a classroom when the weather outside got very cold. The main workshop was designed to be unheated, or as an intermittently-heated space. When the carpenters placed orders for the thick tongue-and-groove floor decking, they included a 7% wastage factor for boards that might have been deemed unusable for whatever reason. It turned out that there was virtually no wastage and we were left with enough spare 3.6 metre long 80x150mm spruce boards to build the frames for the internal workshops. The T&G interfaces worked well to build guardrails around the roofs of the single-storey internal workshops as the museum were keen to use these roof decks to store timber. The workshops were direct-glazed to the timber posts so the whole system used to build the internal workshops was simple and cost very little. Other than the internal workshops, the fit-out of the workshop level was mostly a matter of trying to install the mechanical and electrical services in an elegant way. This was done with the assistance of a very skilled Buro Happold services engineer called Doug King. The lighting was problematic as there was some discussion over whether the pendulum lights should be hung level or whether they should follow the underside of the gridshell. We chose the latter option, which gave a sense of fun to the interior. Electrical conduits and boxes were mounted to the ever-helpful node clamps. Disappointingly, the engineering analysis showed that the required proportions of the workshop floor were too long and narrow to provide natural ventilation. We didn’t want to use roof vents so electric fans were installed within the upper gables. The architect head of E.A. Chivertons, Mike Wigmore, volunteered to design the steel fan mounting brackets and did a fine job of it as well. Sadly, concerns that the central bay of the workshop wouldn’t properly ventilate led to the installation of a series of fan units hung from the gridshell.

The Artefact Store fit-out was more complicated as this is where the museum’s artefacts would have to be stored and protected. It is a surprisingly beautiful and robust space with two-thirds holding roller racking packed with an amazing amount of fascinating objects ranging from agricultural equipment to tools to textiles. The other third of the lower level holds a reception space, toilets, and some of the largest staff offices at the museum. Being mostly an earth-sheltered enclosure, the internal atmosphere is reasonably stable. The most difficult condition to deal with is the sudden influx of a large number of visitors on a cold rainy day when doors are held open overcoming the mechanical equipment. To keep running costs down, a series of three 600mm diameter concrete tubes were laid in the chalk with one end opening under the cantilever of the gridshell to the rear of the building. They would run down the outside of the rear wall, under the floor for 10 metres and open into the plant room to the north side of the building. Air making its way through these three pipes would pre-heat or pre-cool to ground temperature before going into the building. If help was required to raise or lower the temperature of the air, a single domestic-sized gas boiler would do the job as well as provide enough heat to run the Artefact Store’s underfloor heating system. Lighting was low-energy fluorescent strips throughout. All quite industrial-looking with beige block walls, a green-painted screed floor, timber ceiling, and exposed structure. Flat galvanised steel ceiling-mounted air ducts wind their way through the store and glulaminated timber posts run down the centre of the space like soldiers at attention. I have visited the museum on several occasions with architecture students from the Architectural Association and the Bartlett and am always very pleased to see their pleasant reactions to the serene Artefact Store. Some prefer it to the workshop above.


The project entered Cullinan’s office in 1996 as a quirky if not unlikely timber project. By the time the building had been designed, mostly funded, and built, six years had passed and, in 2002, it opened to the public following a herculean effort by the museum to relocate their conservation staff and all of their collections from the various storage locations around the area. The project ran extremely smoothly considering the risks involved, the planning challenge, the funding challenge, and the death of Chris Zeuner mid-construction. Shortly before the building opened, Glenn Howell’s Architects visited us with their prospective client in tow to see a completed gridshell building. They went on to win the prestigious competition to build the Saville Gardens gridshell visitor centre using Buro Happold and the Green Oak Carpentry Company. Satisfyingly, the Duchy had more than adequate timber within their Windsor Great Park woodland to build this large and beautiful structure.

Edward Cullinan Architects have yet to be invited to produce another gridshell. They and several other architects and engineers have produced other types of timber grid structures like lamellas and dia-grids across the UK. I was so affected by the Downland Gridshell experience that I established The Architecture Ensemble to work specifically with timber. Early on, I was asked by Anthony Penrose, the son of artist and surrealist art proponent Sir Roland Penrose and the photographer, Lee Miller, to design a gridshell multi-purpose art gallery and teaching centre for Farley Farm at Chiddingly, East Sussex. Both Buro Happold and the Green Oak Carpentry Company joined us in the effort but the project was stopped stone-dead by the local planning committee. First designs for workshops at the Woodland Enterprise Centre at Flimwell showed pre-fabricated gridshell roofs vaulting between glulaminated ring beams. Even this system was judged to be too expensive and a replacement cruck frame alternative went on to be built. Buro Happold and Green Oak took part in the first stages of this work.

Photo: Stuart Keegan
My feelings are that, as beautiful as gridshell buildings are, they are extremely difficult and expensive to design and engineer at reasonably large scales - especially as one-off projects. Trying to clad a gridshell is also a difficult and costly procedure. Having said this, the carpenters estimated that the Downland Gridshell used between one-tenth and one-twelfth the amount of timber (trees) that an equivalent-sized building would have consumed as a timber framed structure. This is a resource-saving that must be taken into consideration in a time of ever-depleting natural resources. Personally, I find that there is a musicality found in the lines and patterns found within gridshell structures that don’t exist within the much heavier lamella and dia-grids.

The Downland Gridshell is a pure expression of the load patterns within the shell. The varying grids running through the structure are an exposition of what happens when structures are taken to their absolute extreme of material efficiency. Simply working on such a rare project I found that, as it developed along its long gestation process, it drew interesting people towards the project and caused extraordinary events to happen. Below are examples of the ups and downs that I witnessed during the process and I’m sure the Weald & Downland Open Air Museum, Buro Happold, The Green Oak Carpentry Company, E.A. Chivertons, and Boxall Sayer could turn this list into volumes of experiences and thoughts.

Postscript: Downs and Ups

The Downs

The death of Chris Zeuner caused a huge upset to the project with the loss of an extremely adventurous, experienced, resourceful, energetic, and masterful human being and friend. The project’s loss was dwarfed compared to what the British museums and galleries industry lost - his and his wife’s impressive influence went far beyond the museum. He would be proud to see how the museum has developed under the directorship of Richard Harris and, now, Richard Pailthorpe since his death. The project would never have happened without him.

Foot and mouth and a lot of rain.

A roof leak.

Having to add rainwater pipes to the completed building.

Seeing the theft from the site of a young mastmaker’s inherited tools. He disappeared from the site for some time but eventually returned to finish the job.

Missing-out on grabbing the 2002 Stirling Prize for Ted Cullinan.

The Ups

Being trusted with the job in the first place.

Experiencing a rare and unstoppable team spirit and the maintaining of long-lasting friendships with some hugely inspiring people.

Making real a very unlikely project.

Seeing the museum develop and prosper over the ten years since the completion of the project.

Helping to add an important building to the list of achievements of Edward Cullinan Architects.

Watching an enormous stack of sticks being turned into an enchanting architectural space.

Delving deep into the magical world of timber and carpentry.

Watching timber become a staple UK building material again.

Anecdotes and Observations

Standing in front of Buro Happold’s structural model on display at the museum, an old and elegant man approached me and said how pleased he was that architects and engineers had taken to the idea of building timber gridshells. He could see that I was puzzled so admitted that he was once a young engineer working with Barnes Wallace on the design of the first Wellington bombers. These enormous planes were built from spruce gridshells as part of an industrial process. It made me realise that what we were attempting to do was relatively trivial.

Within the museum’s woodland up from the site, I caught a Frenchman, one of Green Oak’s carpenters, looking slightly guilty. He had heard about the project and decided to travel between France and West Sussex to take part in the fun. When at the museum, he would live in an old caravan on site and travel back to France occasionally “to make proper money”. He had purchased a 400 year-old hand adze head at a French market on a recent trip and was looking amongst the museum’s trees for a coiled branch that he could turn into a handle. He found his branch and, by the end of his lunch break, had fashioned it into a beautiful coiling handle to wrap around his forearm and wedged it into his 4 century-old adze head.

A carpenter-written sign hanging on the site’s Heras fencing (facing at the public) saying “please feed the inmates - it encourages them”.

An old beer keg-sized wooden mallet lying on the scaffolding with the name “The Big Thumper” written across it. 

Walking out of one of the buildings deep within the museum after a team meeting. It was well into the evening and the sun had set some time ago. The place was so dark that we had to feel our way back to the museum gates.

From the terrace of the Baltic Exchange in Gateshead we watched a lone ship pass beneath a recently-built and awesomely-tilting Wilkinson Eyre Millennium Bridge. The crowd of architects roared and the gridshell team gulped beginning to suspect that the Stirling Prize was lost.

Visiting the museum year-after-year seeing it’s in-house carpenters, Roger Champion and Joe Thompson, working on a different project every time I visit.

Meeting carpenters from all across the south of England who have been trained at the museum within the Downland Gridshell.

Steve Johnson
 (, project architect on the Weald & Downland Open Air Museum Gridshell, today runs The Architecture Ensemble from his studio right in the heart of London