As the heavily laden caravan wound its way through the snow-filled valleys and passes of southern Tibet towards Lhasa, a solitary Buddhist pilgrim, rosary in hand, could be observed toiling alongside the straining yaks. As he strode his companions could hear him repeating endlessly the sacred Tibetan mantra 'Om! Mane Padme Hum!' Sometimes he would draw a prayer-wheel from the folds of his thick sheepskin coat and rotate it with a flick of the wrist for hour after hour, filling the thin Tibetan air with prayer. Conscious of their pious companion's need for privacy, the Ladakhi caravan men avoided questioning or talking to him at such times.

It was the winter of 1865. The leader of the caravan, bringing goods from Ladakh to Lhasa, had agreed to allow this good natured pilgrim to accompany them on the final, three-month stage of their journey to the Tibetan capital. What they did not know, and never did discover, was that this was no Buddhist holy man. Had they suspected this, and troubled to count the beads of his rosary, they would have found that there were only one hundred of these, instead of the sacred one hundred and eight. Had they removed the top from his prayer-wheel while he slept they would have discovered, instead of the usual scroll bearing block-printed prayers, tiny pencilled figures and mysterious jottings in Urdu.

Once their suspicions had been aroused, they might have noticed that sometimes this holy man dropped behind the slow-moving caravan. Then he would remove surreptitiously from his sleeve, a small, curious-looking device made of metal and glass, through which he would peer hastily at some distant feature, afterwards scribbling a brief note which he would then conceal inside his prayer-wheel. At other times, after making quite sure he was not being observed, he would carefully remove from the top of his pilgrim's staff a thin glass object (which some of his companions might have recognised as a thermometer), and dip it fleetingly into a boiling kettle or cooking vessel. Again, he would note down the reading and hurriedly secrete this in his prayer-wheel.

Not merely was this traveller not a Buddhist, but neither was he a holy man. He was a Hindu - and worse, he was a British spy. Had his identity been discovered, he would undoubtedly have been killed on the spot. Not long afterwards he was to witness the public beheading of another traveller who had entered Lhasa without permission. But who was this man, and why was he prepared to take such terrible risks, and face appalling hardships, for his British masters?

At this period in the history of the Raj, ever since the Tibetans had closed their frontiers with India, knowledge of what was going on there was extremely meagre. To those British Indian officials entrusted with gathering political and other intelligence from their northern neighbour, it must have seemed at times rather like waiting for signals from outer space. What little intelligence frontier officials did manage to glean came from native traders who travelled regularly between India and Tibet. A certain amount more was passed on by the British Legation in Peking and by the handful of consuls and foreign missionaries living in western China. But for the latest movements of the advancing Tsarist armies in Central Asia, the intelligence chiefs in Calcutta and Whitehall were almost entirely dependent on the St Petersburg newspapers. To make matters worse, much of the information reaching them from these sources proved to be highly unreliable, especially that to do with Tibet, just three hundred miles north of Calcutta, then the capital of British India.

British official maps at this time show Tibet as one huge white blank, as though the whole area was obliterated by snow. At their headquarters in the hill station of Dehra Dun, the cartographers of the Survey of India preferred to ignore the positioning of towns and rivers shown on the old, Chinese-made pictorial maps of Tibet. Thus, in the 1860s, the locations of towns even as important as Lhasa and Shigatse, and - further afield in Chinese Turkestan - Yarkand and Kashgar, were known only to the nearest hundred miles or more. Much the same applied to roads and passes, mountains and rivers. But now, with the Russians advancing across the great empty spaces of Central Asia, suddenly there was a call for accurate maps of this vast political no-man's land to the north.

There was no easy solution, if there was one at all. To send through the passes young officers trained in map-making, however brave, willing and well disguised, would have been not only personally hazardous but politically so also. Already one celebrated traveller - William Moorcroft - had been murdered in the unpoliced approaches to Tibet. Others were to meet with a similar fate, including old Central Asian hands like Andrew Dalgleish, hacked to death by a giant Afghan on a lonely Karakoram pass, and George Hayward, about whose violent end Sir Henry Newbolt wrote the epic poem He Fell Among Thieves. Anyone brave - or foolish - enough to trespass in these badlands was regarded as fair game by the lawless tribes and brigands who lived there by rapine. And because there was little or no chance of those responsible ever being brought to justice, any such enterprise by adventurous Raj officials (and there was no shortage of volunteers) was expressly forbidden. Even if they had succeeded in evading the murderous tribesmen, there was very little likelihood of their being able to outwit the vigilant Tibetan border guards who waited at every pass or approach.

In a celebrated incident in 1849, the distinguished botanist Joseph Hooker and a friend had crossed briefly into Tibet from Sikkim on a plant-hunting expedition despite the tearful entreaties of the Tibetan guard commander (who, it is said, later paid for this lapse with his life) and also the protests of Sikkimese officials. The following month they attempted to bluster their way across again, but were turned back by Tibetan troops. On returning to Sikkim they were arrested and Hooker's companion, a Raj official, mistreated by having bamboo cords twisted round his wrists and being made to march with his hands bound to the tail of a mule. The perpetrators were punished for this, but it was the very sort of incident that the British authorities wished to avoid, and Hooker and his companion, Dr Archibald Campbell, were lucky to escape a severe reprimand.

Year after year Tibet thus continued to remain a no-go area for map-makers, and therefore a blank on everyone's charts. Then, in 1862, a young Royal Engineers officer attached to the Survey of India hit upon a brilliant solution. Why not, he asked, send native explorers, hand-picked for their intelligence and resourcefulness, and trained in clandestine surveying techniques? The officer, Captain Thomas George Montgomerie, explained his idea thus:

'When I was in Ladakh I noticed the natives of India passed freely backwards and forwards between Ladakh and Yarkand in Chinese Turkestan, and it consequently occurred to me that it might be possible to make the exploration by that means. If a sharp enough man could be found, he would have no difficulty in carrying a few small instruments amongst his merchandise, and with their aid good service might be rendered to geography.'

Montgomerie's chiefs agreed to let him put his idea to the test, confident perhaps that a native could always be disowned if it came to it, and that reprisals would hardly be called for in the event of his death. Nonetheless, in view of their anxiety not to upset their Asiatic neighbours, especially Manchu China, it was a surprising decision. For Yarkand, in the heart of Chinese Central Asia, was chosen as the target for this sensitive mission.

Montgomerie's first recruit was Mohamed-i-Hameed, a young Muslim clerk already trained in simple survey work. In the summer of 1863 he set out from Ladakh, the last outpost of British influence, and headed across the Karakoram passes towards Yarkand, an oasis town on the ancient Silk Road. Both he and Captain Montgomerie knew that he was taking his life in his hands, and that detection meant alnost certain death. Every effort, therefore, had been made to reduce to an absolute minimum the risk of discovery. The surveying instruments he carried were of the smallest possible size, designed and made specially in the Survey of India workshops. These were still early days, and Montgomerie and his colleagues had not yet begun to adapt Buddhist prayer-wheels and rosaries to clandestine ends. Chinese Turkestan, although only a stone's throw from Tibet, was in any case a Muslim region.

Mohamed reached Yarkand safely, residing there for six months, all the time taking furtive observations with his secret instruments, and keeping his ears open for any news of Russian activities in the region. Then, towards the end of his stay, he was warned by a friend that Chinese officials had become suspicious of his activities and were making enquiries about him. The only thing to do was to leave discreetly but fast, and this he did, hastily returning across the Karakoram towards Ladakh, and safety. But he never got there.

At first it was surmised that he and a travelling companion had been murdered, possibly on Chinese orders, but subsequent investigations proved that both travellers had, in fact, died of illness, possibly precipitated by the hardships of their journey through the grim passes of the Karakoram. Fortunately, however, the young explorer's carefully kept notes were recovered, and eventually delivered to Montgomerie. These proved to contain valuable topographical intelligence enabling Montgomerie, among other things, to fix the precise position and altitude of Yarkand and other nearby towns and villages. They also contained a brief report of Russian activities in this remote corner of the Chinese empire. It must have been with considerable satisfaction, albeit tinged with sadness at the fate of his agent, that Montgomerie was able to report to his superiors on the success of his first secret mission.

It was now agreed that this ingenious idea should be extended into Tibet. In anticipation of this, two Tibetan-speaking hillmen of British nationality had already been chosen for the task by Montgomerie and his immediate superior, Colonel James Walker, with the help of Major Etwall Smyth, a government education officer working in the frontier region. The two men were Nain Singh, aged thirty-three, the headmaster of a village school at Milam, lying 11,000 feet up in the Himalayas, and Mani Singh, his slightly older cousin. Both were experienced mountain travellers, having some years earlier accompanied a German expedition whose members had found them intelligent and resourceful.

The two recruits were brought to Dehra Dun, where they were given two years' training in route survey and reconnaissance work. They were taught the use of sextant and compass, how to identify the stars and use them for fixing positions, and how to calculate altitudes by observing the boiling point of water. But that was no more than the Survey of India taught all its newly recruited native surveyors, or 'chain men' as they were known in the business. Nain and Mani Singh now entered what today would be called a spy-school, a world curiously familiar to readers of Kim, Kipling's story of a small boy caught up in the Great Game and eventually recruited into the Raj intelligence service. Here they were schooled in the ingenious techniques devised by Montgomerie and Walker for conducting clandestine surveys of territories which were hostile, politically sensitive, or belonged to other governments - or, in the case of Tibet, all three.

First they were trained by endless practice to take a pace which, whether they walked uphill, downhill or on the level, always remained the same - thirty-three inches in the case of Nain Singh. Next they learned how to keep an exact count of the number of such paces they took in a day, or between any two landmarks. This was done with the aid of a Buddhist rosary, which as we have noted normally comprises one hundred and eight beads. Eight of these were removed, leaving a mathematically convenient one hundred, but not a sufficient reduction to be noticeable. At every hundredth pace a bead was slipped. Each complete circuit of the rosary, therefore, represented ten thousand paces - five miles in the case of Nain Singh, who covered a mile in two thousand paces. Because the Buddhist rosary has attached to it two short secondary strings, each of ten smaller beads, these were used for recording every completed circuit of the rosary.

Nain and Mani Singh were trained in the use of cover stories and disguise, just as Kim was at Simla by the shadowy Lurgan Sahib. Their lives were to depend on just how convincingly they could play the part of holy man, Himalayan trader, or whatever the delicacy of the situation demanded. Their disguise would have to stand the test of months of travelling, in the closest intimacy with real traders and pilgrims. Like Kim, they also learned to forget their own names, and work under a number or a cryptonym. Thus Nain Singh became simply 'Number One', 'the Pundit' or 'the Chief Pundit'. His cousin was known in the Survey records as 'Pundit Number Two', 'the Second Pundit', or just 'G-M', which was arrived at by taking two of the letters of his name and reversing them. The word 'pundit', which suggests a man of certain learning, was to become the generic name by which they, and all subsequent recruits, were referred to. Their real names were not revealed until they were too old to make further secret journeys.

Not only was the Buddhist rosary ingeniously adapted to Montgomerie's purpose, but so were prayer-wheels. These were fitted with a secret catch which enabled the pundit to open the copper cylinder and insert or remove the scrolls of paper bearing his route notes and other intelligence. Later the workshops at Dehra Dun were to conceal compasses inside the wheels, so that a pundit could take bearings while pretending to be at prayer. Larger instruments like sextants were concealed in specially built false bottoms in the travelling chests which native travellers carried, while secret pockets were added to their clothing. Thermometers, for measuring altitude, were concealed in hollowed out staves, and mercury - necessary for setting an artificial horizon when taking sextant readings - was hidden in a sealed cowrie shell and poured into a pilgrim's bowl whenever needed.

For his novel Kim, which was written nearly forty years after Nain and Mani Singh graduated from the Dehra Dun spy school, Kipling borrowed at least two of his characters from Montgomerie's twilight world. The amiable Hurree Chunder Mookerjee - R. 17 on the Great Game payroll - is known to have been modelled on Sarat Chandra Das, one of the later pundits, while Colonel Creighton, who recruited the youthful Kim and masterminded his education, can only have been Montgomerie himself.

Their training now complete, it was time for Montgomerie's two proteges to show their paces. The route which he and Colonel Walker worked out was aimed to carry them some twelve hundred miles through Tibet to Lhasa and back. The two men, it was hoped, would return with a route survey which would enable Montgomerie to fill in a number of embarrassing blanks on the map of Tibet. These included the precise location and altitude of Lhasa, the route that the great southern caravan trail took from Lhasa westwards towards Gartok, and the course of the mysterious Tsangpo river as it flowed from west to east across Tibet. It was also hoped that the explorers would return with useful political intelligence. Despite the fact that they had spent nearly two years being schooled for it, it was a daunting task for the two hillmen - all for some twenty rupees a month and the promise of a larger reward if they were successful.

The journey would take them many months, every minute of which they would have to be on their guard against Tibetan suspicions. Every yard of the way would have to be paced, latitudes and altitudes calculated and carefully recorded. In the event, despite all Montgomerie's painstaking preparations, the two were to get off to a bad start.