When Joseph Banks was dispatched overseas to hunt for exotic plants for George III's private gardens alongside the River Thames, he was doubtless proud of the exotic specimens from China, Africa and the West Indies with which he returned but he cannot have known the full fruits of his exploits. Two centuries later, as Kew Gardens celebrates its 250th anniversary, Banks's original seed packet has grown into the biggest in the world, and is housed at Wakehurst in West Sussex, the home of Kew's Millennium Seed Bank Project.

Between 2000 and 2010 the project has been given sufficient funding to bank 10 per cent of the world's flora, and by 2020 - if it can find more funding - it will have 25 per cent. But according to Paul Smith, head of Kew's seed conservation department who heads the seed bank, the project lacks most of the money to continue its work after 2010. We should all be worried.

Come on, you say, we really need only a few species. About 80 per cent of our carbohydrates come from a mere 12 species, and there are 1,400 seed banks around the world holding the vital 64 crop species. Do we need any more insurance than that?

You bet we do. Between 1970 and 2050 we shall have polished off 15 per cent of the world's flora by destroying its habitat, and Smith tells me that “global extinctions are running at 1,000 times the natural rate”. Suppose disease in intensive agriculture wipes out, say, our present strains of rice or wheat? And don't we need seed to produce future biofuel crops, and pharmaceuticals, and all the plants that help to regulate our environment?

We need all the insurance we can get. But here's the crunch: it's clear which domesticated crop species we should save but, as Smith says, “we haven't a clue what potential exists in all those other species”.

How, then, to decide which plants to save first, if money is tight? According to Smith it is straightforward: “We work to the three Es - plants that are endangered, endemic (they grow in only one place) or economic. But it's a rearguard action. So much is being lost as we speak.”

Which economic species would he save, then, if the funding position became desperate? “Well, Jatropha curcas is being used for biofuel but there are more than 100 species of Jatropha, some of which may be more appropriate for oil production on marginal lands because they are locally adapted, not invasive and don't require irrigation. Until we carry out this kind of research, we won't know.

“One of our major concerns is where such oil seeds are grown - they should not occupy primary agricultural land that could be used for growing food, or displace forest or woodland because this will result in greater CO2 emissions.”

And endemic? “Look at that!” he says. And there, in a nearby greenhouse at the seed bank, are pots of South African Faucaria tigrina, stocky clusters of succulent triangular leaves fringed with shark-like teeth and scattered with starbursts of yellow flower. Less than 100 exist in their native habitat and they are being grown here to provide seed to save, yet they wouldn't look out of place in a garden centre.

Endangered? “There's no easy answer. The snowdrop is a source of the drug galanthamine, one of the most important drugs yet in the fight against Alzheimer's disease. Twenty years ago we didn't know about this.” Meanwhile, potentially more useful snowdrop species are disappearing in Turkey as their habitat is lost.

“It was only 30 years ago that Catharanthus roseus, the Madagascar periwinkle, was found to contain a treatment for cancer, and it has improved the prospects of recovery from childhood leukaemia from 5 per cent to 95 per cent. It grew nowhere else on earth, and could so easily have been lost.”

Kathryn Bell, director of co-curricular activities at Ardingly College, West Sussex, regularly brings sixth formers to the seed bank and finds them extremely enthusiastic about its work. “The exam system includes appallingly little plant biology now,” she complains, “and the students find it fascinating. Some have even come back to do their own hands-on research here. The best lesson for them is to find out that there aren't answers to everything. Life isn't a textbook.”

As for economics, 50 per cent of the seed bank's funding comes from gambling, through all those £1 punts that feed the National Lottery. A hefty 42 per cent is found from trusts, foundations and corporate sources, and the Government magnanimously chips in 8 per cent.

It's not as if the Millennium Seed Bank Project is expensive in real terms. £100 million over the ten years would complete the banking of that projected 25 per cent of world species; £2,000 will save a species. But the lottery looks unlikely to provide any more money until the black hole of the Olympics has closed, and corporate funding has been shrunk by the recession even if, for now, trust funds remain forthcoming.

If the project seems expensive, remember this. It is setting out to collect everything. There may be many seed banks in the world, including large ones in China, South Korea, the US, Germany, Japan and India (Norway has one buried in the permafrost on the Svalbard peninsula, at -18C). But they have less ambitious objectives, saving only economic crops, or plants from their own latitude, or native species. Many seed banks are tiny. “Equipment and skills are poor,” says Smith, “and the result is poor practice.”

It is tempting to think of the British seed bank as a bunker, pulling in important seeds from all over the world and sitting there like a well-stocked larder. But in truth it is a hub rather than a hoard.

The greatest challenge is in research: to find out, for example, how climate affects seed behaviour, what makes seeds die, and how to ensure that species which are difficult to germinate after storage can be brought to fruition successfully. It is the work of molecular biologists, following up that of botanists collecting in the field.

When Smith takes me through the subterranean chambers of the seed bank, through several airlocked doors, it is aglow with fluorescent light. Paper bags of seed stand in plastic trays in the drying room, labelled with their species and origin - US, Royal Horticultural Society, Burkina Faso. They can stay here for months while they slowly dry to the optimum level for storage, and every percentage point reduction in moisture content will double the seeds' chances of germination after storage. This is why the seed bank is putting mini seed banks into schools, so children can learn to preserve seed successfully.

After drying, the seed is cleaned of its husks and a sample is x-rayed to check for insect infestation, then counted. A student in a white coat bends low over a bench, counting orchid seeds into a tiny glass jar, then adding a sachet of silica which will turn pink if the jar's seal is breached. Only the occasional species will stand abuse and still germinate, such as the sacred lotus seed collected in China that germinated on its herbarium paper when a bomb hit the Natural History Museum during the Blitz and exposed it to moisture for the first time in 170 years - or the single seed of a South African Leucospermum, found in the 200-year-old wallet of a Dutch merchant.

The jars of dried seed go into the cold store, an underground library full of sliding stacks of shelves, every one full of glass jars holding even tinier glass jars. The temperature is -20C but the air-conditioning system creates a further wind-chill.

In that room, the size of a corner shop, are seeds of 10 per cent of the world's plants. Beyond is another, larger room that one day, Smith hopes, will hold the rest. Empty shelves stretch away into the gloom.

Times recommends

• Live chat: Thrifty ways to update your winter wardrobe
• Is anybody out there really listening?
• Nutrition Q&A: Will eating late make me fat?