Holistic grid: Designed in the late '60s but only now complete, this hospital in Mirano by Giancarlo De Carlo explores the notion of the grid as a flexible, neutral framework that sensitively orchestrates light and spatial relationships.

Few buildings designed in 1967 but not completed until the late '90s would command much interest, but Giancarlo De Carlo is among the most thoughtful and respected architects of his generation and despite the delay, several of his ideas have remained unrealized elsewhere.

The fashion in hospitals in 1967 was for huge deep-plan high-rise blocks, but instead De Carlo suggested a low courtyard plan with two inhabited storeys planned on a tartan grid. This neutral flexible framework allows ready access and servicing within which different kinds of functions and their architectural forms might be set. It seemed to De Carlo appropriate for a hospital, both to accommodate the internal changes of programme and to allow expansion by addition of departments. In the event, the flexibility was exploited by programme changes even before construction began, while budget limitations caused the building to contract rather than grow.

Rug-like, the initial and ideal plan shows a series of variously sized fields divided by regular strands, sometimes warp, sometimes weft. These strands become communication channels of various kinds -- wide or narrow, public or private, people or services -- which occur like the different coloured stripes of a tartan. In contrast with these strands, some of the fields in between can be left vacant, forming holes which bring light and air into the plan. They can be of different sizes and penetrate the volume to different depths. The grid as abstract system thus sets up a series of relations at various fixed sizes, which are habitable in a variety of ways.

In section, the essential idea was to organize everything on three layers. The ground floor has all the main treatment rooms, clinics and operating theatres. This largely artificially-lit internal world is heavily serviced with elaborate thresholds, rules of dress and procedure, and positive air pressure against infection: it is necessarily the most machinelike part. The upper floor, by contrast, carries the patient rooms and wards, and because only the roof is above, there is endless possibility of daylight, both vertically through rooflights in passages and sideways from the many courts. Everyone gets a view out and a glimpse of sky, even if many views are relatively short. Between the two inhabited floors runs a shallow servicing layer, an endless room for the pipes and wires. This seemingly extravagant arrangement allows any kind of service to be fed from above or below, and systems can be added and renewed with minimum disruption. A curiosity of the initial plan was that the upper floor was the more com plete of the two, setting the pattern for the whole. The ground floor with its various departments was then permitted to take over the space defined by the column grid.

The plan and model of the original competition entry are self-contained and evidently worked mainly from the inside out: an abstract system owing nothing to the site. De Carlo is normally the most site-sensitive of architects, but this place - close to the edge of the Venetian Lagoon - is very flat and lacks long views, while the older hospital buildings dividing the site from the main road offer little inspiration. Nonetheless, in developing the design he tamed his grid with two irregularities as added limbs: at the south-east corner a diagonal block for the day patients' and visitors' entrance plus cafeteria, on the west side a ramp and curved canopy for emergency vehicles. These two gestures tie the building to its context, the restaurant's corner placing and angular stance making it more liminal: a transitional space between. hospital and outside world. On its roof is a terrace for the doctors, set next to their library restroom, another breakaway from rhythmic discipline.

The corner placing also helps disguise one of the oddnesses of the building: that the main public route runs along the east side, rather than down the middle as Classical precedent would suggest. This reflects the anti-hierarchical and anti-axial nature of the '60s grid, and also the fact that the three spines running through the building serve different purposes: the east is public, the middle a medical corridor for doctors, nurses and patients on trolleys, the west a dirty corridor for services and waste only. Placed close to the service access, the latter is only ever seen by staff, and is definitely backstage. Thus moving from east to west there is a clear public/medical hierarchy. Moving from south to north at ground floor level is similarly hierarchical: the first block has day clinics to east and emergency to west, the middle one X-ray to east and operating theatres to west, and the end one more operating theatres and intensive care. The rationale is that as you move down the building you come to diff erent departments; the deeper you penetrate, the more intensively medical it becomes.

Passage through the building is marked rhythmically by views into the various courtyards, the two-storey ones being the main orientation points, with round windows in the passages as added articulation. Repeated as frequently as first planned, the courts might have become confusingly repetitive, but it is not a problem so far. On the outside the same rhythm appears as within, picked up by the ends of the long rooflights and the fire stairs. If the white tiled facades seem a little dated, they are at least preferable to those of most other hospitals of the period, and on the whole the building is much gentler than its rivals. Most memorable about it are the light-filled upper corridors and the intimate patient rooms which overlook plant-filled courtyards, an apt memory of the cloister.