To the ends of the Earth

Minus 50C; winds gusting to 70 knots; dreams about popping outside to check on the dogs: "Shan't be long." Then waking and seeing the vestigial glow of the Aurora Australis through the small window of the modular bedsit, and the triple layer of bubble-wrap beyond the double glazing. All very trying. But nothing that a bit of innovative architecture couldn't handle.

Minus 50C; winds gusting to 70 knots; dreams about popping outside to check on the dogs: "Shan't be long." Then waking and seeing the vestigial glow of the Aurora Australis through the small window of the modular bedsit, and the triple layer of bubble-wrap beyond the double glazing. All very trying. But nothing that a bit of innovative architecture couldn't handle.

Unfortunately, that has never quite been on the agenda in Antarctica. The headquarters of the British Antarctic Survey on the 150m-thick Brunt Ice Shelf has always been sturdy, functional and serviceable - think Alistair MacLean's fictional and rather ad hoc Ice Station Zebra.

Despite the cutting-edge research practised by the scientists there - this is where the hole in the ozone layer was discovered - brute, elemental forces have been met with brutish architecture. That, currently, means a glorified oil-rig on stilts jammed down into the pack ice like a giant, 100m-long steel crampon, whose prongs - the squeamish should look away now - sit on top of what the scientists refer to as "the onion" - a huge bolus of human waste matter buried in the ice.

This particular frontier will soon seem rather less crude. In January, representatives from three design teams will set out for Antarctica to compete for the arduous and unforgiving privilege of creating Halley VI, a post-Millennial survey base. It's a challenge without much obvious kudos. Icon on the ice? Doesn't quite ring true as a heading, does it? Perhaps something racy for the front cover of Fortean Times or the Ozone Fontanel Quarterly might suit the project better: scientists in bubble-wrap heaven.

No, the key thing about the shortlisted survey stations is that they must perform, protect and endure for up to a decade, and cost no more than £14m. And what's so fascinating about the proposals is that they have forced their designers to go right back to basics - to strip away anything that does not completely satisfy that trinity of specifications. The long-listed design teams involved - culled from 86 entries worldwide - were combinations of big-name architects and engineers. The shortlisted trio are: Buro Happold and Lifschutz Davidson; Hopkins Architects and Expedition Engineering; and FaberMaunsell with Hugh Broughton Architects. Among those who didn't make the cut were Richard Rogers Partnership.

Foster Associates, a practice whose fearsome technical expertise might seem to have been tailor-made to triumph, are not in the mix. Nor is the brilliant ex-Fosterian Ken Shuttleworth, now practising solo, whose hamlet of pyramids fell at the first judgemental hurdle; however well they might have performed as internal environments, would they have minimised the crucial problem of wind vortices and subsequent snow build-ups? The environment needed the architectural equivalent of a stripped-down, "slippery" Formula One racer rather than a hi-tech version of the necropolis at Giza.

Happold, Hopkins and FaberMaunsell have produced solutions whose apparent similarities disguise quite different internal approaches. They share one thing, though: the appearance of the proposed stations has resurrected the Walking Cities architropes of the Archigram collective that so hyperventilated designers in the Sixties and Seventies, and which have since become classic examples of once futuristic "impossibilities" that are suddenly, and almost quaintly, à la mode.

They all look the part, in terms of the stilted Walking Cities template. But only one, the Hopkins entry, seeks to deliver the key Archigram fantasy: their station will actually move. It's designed to shuffle along on steel legs shod with giant snow-shoes.

It's not a gimmick. Over the decades, the British Antarctic Survey's research stations have been sited less than 10 kilometres from the edge of the retreating ice-sheet. One eventually tipped into the sea, and another was dragged to a safer position by a helicopter flying a highly dangerous mission. Yet another, a buried habitat designed by the respected engineer, Stephen Whitby, was gradually deformed by the pressure of the slow-moving ice to such an extent that its inhabitants sometimes heard rivets pinging loose in the dead of night. It must have been rather like being in a submarine diving too deep into the Mariana Trench, the creak of its hull presaging doom.

Hopkins's idea is intriguing because it plainly recognises the implacable physical strictures of Antarctic existence, but also wants to add something distinct to the sheer ingenuity of the experience. Their design is a combination of a Big Idea that depends on highly refined structural and wind science - via computerised fluid dynamics - and the simplest possible arrangement and construction of the individual living and work spaces that would make up the research station's two 800-ton units. The inhabitants will live and work in standard shipping containers, fitted out before shipment and slotted into their final configurations on site like so many Duplo blocks.

It's strangely encouraging to report that the shuffling legs of the two buildings have been demonstrated both theoretically and practically. Satisfied with their drawings and putative power-drive systems, the engineers made a Lego Technic model of the station whose electric walking mechanisms worked exactly as the real ones would. The legs look elegant and a bit skimpy, but there's no doubt they will tote their loads.

The buildings' outer layer will be familiar to anybody who's visited the Eden Project in Cornwall or the space museum at Leicester: it's composed of EFTE "pillows", which form a translucent duvet around the building's perimeter.

But it's the legs that draw the eye again and again. Hopkins has shown increasingly deft touches in its designs of strut structures - the timber spindle-trees holding up the roof of their recently completed Norwich Cathedral extension were superbly svelte - but the final form of the survey-station legs are of such critical importance in terms of high-velocity snowdrift that they have plugged into the expertise of the Canadian extreme-weather design specialists, RWDI. The form of every steel profile, every flange and every joint matters greatly.

All the shortlisted designs are intent on minimising environmental impact. Hopkins appears to have come up with something special: an ice spine, like a long thin hummock running between its buildings, which with the right supporting technology could be used to generate hydrogen. And that, in combination with solar and wind power, could eventually deliver a zero carbon output environment.

And the competition? It's pretty fierce. The survey station designed by Buro Happold and Lifschutz Davidson features a series of two-tier linear "craft" standing on adjustable, stainless steel legs, linked by bridges and with an airbag around each foot to remove packed snow. The two-decked craft are formed with ribcages of curved beams with outer layers clad in a translucent skin. There are even small balconies - a rather eerie touch, this, because it seems to have more to do with apartments overlooking the Thames than impending frostbite or snow-blindness.

FaberMaunsell and Hugh Broughton Architects have gone for a series of tuggable bug-structures, again on stilts. Set out and interconnected on the ice-shelf, they would resemble a pattern of glinting bacilli. This is not to belittle their vision. While Hopkins and Buro Happold have delivered stations comprised of two and three structures respectively, it may be that FaberMaunsell's 11-pod solution would suit the various needs of the British Antarctic Survey better. On the other hand, it would surely produce more complex snowdrift effects - and that has significant implications in terms of snow management on the site.

What will the representatives of the three design teams make of the conditions when they're shuttled out to Antarctica next month during the eight-week safe-weather window? Will they still think their designs are up to it? And what will the researchers make of them?

One thing is certain: the three pairs of practices are facing a quite extraordinary design opportunity, and one that will set a new standard of accommodation - and existence - in these brutal wastes. The architects and engineers involved have, in effect, found themselves back in their third-or fourth-year classes at university or school of architecture, faced with a problem they know little or nothing about.

The quietly enjoyed thrills of these three ventures into the unknown must have provided timely reminders that, at its best, architecture should always be the result of fresh, exploratory thought. In the case of the British Antarctic Survey's forthcoming Halley VI research station, two kinds of exploration will coalesce on the Brunt Ice Shelf. For the first time, the boys from the white stuff will be sitting pretty.