Find the river: Discovering the Tsangpo-Brahmaputra in the age of empire

Abstract Despite the enormous size and economic and scientific significance of the Tsangpo-Brahmaputra River, questions of where and what it was generated successive waves of dispute from the mid-eighteenth to early twentieth centuries. Geographical discovery in the eastern Himalayan borderlands neither entailed the application of fixed theories and techniques, nor resulted from consistent flows of information along established channels. Europeans instead understood the region’s rivers in many different ways, influenced by sporadic deluges of data, competing forms of expertise, shifting imperatives of colonial political economy, unsettling encounters with various bodies of water, and heterogeneous Asian knowledge structures. Informants, infrastructures, and cosmologies of often-overlooked communities at imperial margins fundamentally reshaped European knowledge. Under these conditions, practitioners of spatial sciences came to thrive on the proliferation of models and objects of discovery rather than seeking definitive closure.


Introduction: Fluid discoveries
Morphing, massive, and-unusually-masculine in both its Tibetan and Indian names,1 the river flows east across the Tibetan plateau before tumbling southwards through a deep gorge on the eastern fringe of the Himalaya.By the time it performs another right angle to tumble westwards through Assam, it has transitioned from Tibet's Tsangpo to India's Brahmaputra. 2 In the mid-eighteenth century, however, all of this seemed different-to Europeans, at least.Away from the toponymically crowded coasts and plains of the subcontinent and towards its sparsely labelled fringes, the 'East Indies' sheet of a Paris-published atlas showed the 'Baramputri' river emanating from a vast lake, labelled 'Chemay' (Figure 1).Further north, above the colour-tinted peaks of the eastern Himalaya, a wiggly black line, appreciably thicker than that denoting the Brahmaputra, represented the Tsangpo, gushing eastwards to become Burma's Irrawaddy and discharge water from the high Tibetan plateau into the Bay of Bengal hundreds of miles east of the shared mouth of the Brahmaputra and Ganges.
It is now understood that Chemay was a chimera and the confidently plotted Tsangpo does not become the Irrawaddy-and has not for the past 20 million years. 3ut the work required to establish these topographical truths was every bit as convoluted and mutable as the landscape it sought to describe.The river was not discovered by Europeans in progressive increments over time nor in a revelatory moment.Where and what the river was generated waves of dispute for an unusually long period. 4To equate the Tsangpo-Brahmaputra's discovery with the Royal Geographical Society's (RGS) award of its prestigious Gold Medal to the British officer Frederick Bailey following his exploration of part of the river's course through the Himalaya in 1913 is not only to accept a reductive myth of imperial geography, it is also to overlook how the objects and discoveries of field sciences violated dichotomies of colonial and indigenous, and of culture and nature.Attempts to know riverine space on the outskirts of empire comprised unruly mixtures of European and Asian men, periodicals and institutions, visual instruments and representations, and the active presence of rivers and mountains.
This article outlines how assemblages of these elements constructed and conveyed knowledge in very different ways during three extended moments between the 1820s and 1910s. 5Each phase related to a burst of British military and economic expansion into a different portion of 'Zomia', the recently coined region of high mountains and deep valleys encompassing Assam, northern Burma, Tibet, and southwest China. 6The unsettled political dynamics of this space went hand-in-hand with unsettled knowledge of water and rock.In their lower courses the region's huge rivers became both invaluable energy sources for cash crop agriculture and crucial channels of communication, allowing states, including British India, to thrive.Their usefulness, coupled with astonishment at their vast scale, drew a range of imperial agents against the flow, upstream and uphill into terrain resistant to conventional ways of garnering knowledge, accruing profits, and entrenching sovereignty.These watercourses blurred boundaries between lowland and upland, confusing the categories of not only nineteenth-century geographers and bureaucrats but also twenty-first-century scholars prone to asserting the distinctiveness and detachment of Zomia's hills. 7ivers in and around the eastern Himalaya have agency not only in geological terms; they have been active and fickle presences in the processes that generated spatial knowledge.Rivers spoke to many of the humans who sought to understand and to use them; ascertaining what they had to say was a major preoccupation of European geographers and surveyors. 8Richard White's famous thesis regarding salmon fishing on the River Columbia-'precisely where the river expended its greatest energy, humans had to expend the most labor'-also provides a reasonable model of how revelatory experiences of particular portions of vast streams gave impetus to geographical discovery. 9Especially when unusually broad, frothing, or in spate, bodies of water presented as uncanny presences that eroded extant ways of knowing and allowed new ideas to bubble up.As Jessica Lehman recently suggested, sciences dealing with large bodies of water are 'shaped through affective encounters with material forces' that set 'productive limits' on what can be known.10I accordingly deploy an expanded range of watery metaphors, including 'deluge', 'wave', and 'gush', alongside the limited liquid lexicon of 'flow' and 'circulation' used without sufficient reflection in much global history. 11Heeding anthropologist Stefan Helmreich's call to 'think critically about [the] turn to water' in humanities scholarship, 12 this terminology better represents the uneven rhythms, patchy spatiality, and laborious processes of riverine discovery in Zomia.
The lingering notion that Europeans in the age of empire perceived the natural and nonhuman as passive objects to be comprehended, dominated, and exploited simply does not hold in the case of their encounters with great rivers in this part of Asia.Coupled with diverse epistemological frameworks, information sources, and political economic goals, vivid experiences of gigantic flows of water prompted varied ways of comprehending not only where rivers are, but what rivers are.In riverine geography, distinctions between epistemology and ontology were murky at best. 13The various men, maps, and measurements that claimed to discover the Tsangpo brought into being a range of rivers.This article extends insights from the 'ontological turn' in anthropology to historical actors, conceiving of different ways of describing and doing as 'not so much a matter of "seeing differently"' but rather 'a matter of seeing different things'. 14European geographers and their interlocutors in the liminal spaces of Asia posited and acted upon what, drawing on Annemarie Mol, we might think of as the 'river multiple'. 15Differing notions of what a river is, existed not only between European and Asian communities, but also between distinct groups of imperial personnel.
There were times when the Tsangpo-Brahmaputra seemed less surely known and less unitary.Like the objects of laboratory sciences discussed by Bruno Latour, no definitive moment of discovery rendered the river 'impervious to history'. 16aintaining particular versions of the Tsangpo-Brahmaputra was laborious and prone to failure.Rather than focusing exclusively on the factors that enable initial moments of discovery, this article investigates how claims of discovery are reinforced or eroded over time.Finding the river was 'a continuing event'. 17As Simon Schaffer reminds us, discovery is also 'made up' of many components that can only appear simple and singular through the work of retrospective narratives. 18In competing European versions of the Tsangpo-Brahmaputra, networks of personal trust, the growing authority of particular learned institutions (notably the Survey of India and the RGS), and the perceived credibility of particular modes of inscription and representation were all vital. 19However, these elements were also, to a substantial extent, negotiated through proposed discoveries rather than exercising a fixed influence.
Learned and interested communities in Tibet and the Indian subcontinent understood long before their European counterparts that the Tsangpo joined the Brahmaputra. 20What counted as geography in colonial India and Burma, and also metropolitan Britain and continental Europe, developed through encounters with and appropriations from various Asian cosmologies.Historians of geography have recently highlighted the essential role of non-European information and labour in imperial exploration and surveying. 21It nonetheless remains widely assumed that Europeans largely effaced traces of this reliance in rendering their findings for audiences in colonial hubs and imperial metropoles. 22In addition, non-Europeans are often supposed to have been confined to providing labour and, on occasion, information to populate frameworks set by Europeans. 23Even Kapil Raj's work on sciences as products of 'circulation' and 'encounter' in South Asia suggests that the active and 'varied roles' of 'indigenes' during the precolonial and early colonial periods gave way to circumscribed and heavily 'disciplined' roles in the age of high imperialism. 24n the case of the rivers of the eastern Himalaya, the work and thought of various Asian communities was vital not only to Europeans' acquisition of knowledge but also to the specific content and structure of that knowledge.Agents of empire and imperial geographers often mangled and derided Asian ways of comprehending water and earth, but their reliance on these traditions nonetheless left lasting residues on their written tracts, images, and data compilations.Terraqueous locales at the fringes of South, Central, and Southeast Asia-where land, water, and atmosphere mutually transformed through constant interplay 25 -enabled often-juxtaposed categories of 'indigenous' and 'colonial' knowledge to overspill and mingle. 26The concept of indigenous knowledge seems still more questionable when we consider Europeans' particular reliance on interlocutors who moved across mountains and valleys, in many cases after being displaced by imperial expansion. 27Various Asian ways of construing and constructing space fundamentally shaped European discoveries of the Tsangpo-Brahmaputra.
A spate of riverine histories of South Asia published during the past couple of decades has tended to posit that colonial knowledge of, and interventions in, India's water systems hewed to a single, sustained logic of control and exploitation. 28Dilip da Cunha, for example, suggests that British rule 'survived and thrived on keeping water contained with a line'. 29Building on but moving beyond this scholarship, I suggest that colonial scientific approaches to rivers were not limited to efforts to define, contain, and render them economically productive.'Untamed' portions of rivers could be valuable in an alternative way, as a source of scientific and imaginative intrigue.Above all, British understandings and uses of watercourses were multiple and mutable.Instead of 'inventing' singular rivers, as da Cunha has it, agents of empire constructed numerous Tsangpos-and each new or revisited river involved the denial of other rivers.'River colonialism' was not a unitary project but comprised myriad ways of seeing and acting. 30Individuals and institutions with varied expertise and methodologies, situated in varied locations, drawing on varied Asian knowledge structures, and influenced by varied imperatives of political economy, perceived very different river systems in Zomia.At no stage-including the present day-has there been a settled understanding of how these tumultuous flows of water could be definitively known.

Sourcing knowledge
During the mid-1820s, a British military venture to remove Burmese forces from Assam quickly turned into the full-scale annexation of the Brahmaputra valley. 31This was among the early acts of the East India Company-State's turn to focus on expansion at its external frontiers under the guise of security, having defeated its last major internal opponent in South Asia in 1818.The invasion saw unsettled politics mix with unsettled topography.32Among the skirmishes between Burmese and British forces that precipitated the invasion was a dispute over river islands in the Brahmaputra, which formed a section of the boundary between Assam and colonial Bengal. 33The Brahmaputra was the key means of wayfinding and travelling for British soldiers and administrators in Assam and, accordingly, the primary focus of their surveys and maps (Figure 2). 34Simultaneously, it was a fickle entity that agents of empire struggled to master.The main channel and its numerous feeders shifted course, flooded fields, and overwhelmed conventional strategies for settling and taxing agrarian land. 35The lack of river traffic and swathes of 'inhospitable jungle' along the banks often led to disorientation. 36Experienced against the flow, as the British did in approaching from the west, the Brahmaputra's formidable braided stream unravelled into elusive threads at the strategically sensitive eastern end of the valley (Figure 3).
Imperial interest in the rivers of this region was about more than administration and geopolitics.Senior officials in Calcutta tasked soldier-surveyors in the invading party with resolving a long-established geographical problem.The Tsangpo had first Source: British Library, London, India Office Records, H/678, map 10. © British Library Board.
became known in Europe in the 1720s through translations of the atlases commissioned by the Kangxi emperor, based on surveys jointly undertaken by Qing Chinese court personnel and French Jesuits. 37From his study in Paris, the geographer to the French court, Jean-Baptiste-Bourguignon D'Anville, concluded from these surveys that the Tsangpo joined the Irrawaddy. 38The advent of colonial rule in Bengal in the 1750s and 1760s took European enquiries on this region into the field.Encountering the Brahmaputra first-hand in 1765 was a revelatory experience for James Rennell, the leading geographer of early colonial India.Rennell pronounced himself 'surprized [sic]' that near its junction with the Ganges in northern Bengal the Brahmaputra was enormous-'rather larger than the Ganges'. 39Coupled with what he termed 'the positive assurances of the Assamers, "that their river came from the north-west, through the Bootan mountains"', Rennell opposed D'Anville and stated that the Tsangpo became the Brahmaputra (Figure 4). 40he approximately concurrent British invasion of Assam and establishment of key learned societies and journals in Europe and British India combined to reinvigorate the contest between Rennell and D'Anville's competing Tsangpos.The incumbent surveyor-general of India at the time of the British invasion of Assam, Valentine Blacker, was a firm supporter of Rennell's conjecture and instructed surveyors attached to the invading force to seek out relevant information.Assamese interlocutors suggested that the Brahmaputra's source was a pilgrimage site in the mountains east of Upper Assam, known as the 'Brahmakund'. 41This conjecture guided the first British surveys into the foothills at the northeastern corner of the Assam valley in early 1825, which seemed to confirm that Rennell and Blacker were mistaken and D'Anville was correct. 42The Survey of India map of June 1825 (Figure 3) stated plainly that 'the Brama Khoond' was 'the Source of the Brahmaputra'.
One of the surveyors, Richard Wilcox, later couched this datum as a conspiracy of local mis-informants and environmental challenges-a fiction sustained because 'the Natives knew well that the boats of Bengal could not pass more than one day's journey beyond Sadiya', the easternmost British outpost in the Assam valley. 43In fact, for Assamese Hindus the Brahmakund was dedicated to an avatar of Vishnu and the essential source of the Brahmaputra.The influence of Hindu cosmology on British imperial geography lingered even after doubt was cast on the notion that the Brahmakund 41 Wilcox, 'Memoir', p. 316. 42Anon., 'The Burrampooter, or Brahmaputra', Asiatic Journal, vol.21, 1826, pp.52-53. 43Wilcox, 'Memoir', p. 316.branch was the largest tributary of the upper Brahmaputra. 44The first British surveyor to see the Brahmakund complained of 'so many discrepancies between the Hindu legends, and facts', but retained the idea that it was a spiritually significant feature, recounting for his European audience that the landscape contained 'gothic pinnacles and spires'. 45Like many later imperial geographers interested in the Tsangpo question, he drained the contents of Asian cosmologies and refilled them with material that also mixed the physical and metaphysical.We can understand this as a reversal of the flow of spiritual-spatial categories from European metropoles to South Asia that Sumathi Ramaswamy identifies in the case of the 'lost land of Lemuria' during the late nineteenth century. 46To Assamese Hindus of the 1820s, the Brahmaputra was defined by its flowing from the Brahmakund, and the residue of this ostensibly 'fabulous geography' took some time to be fully washed off of British conceptions of the river's source.
In 1827 and 1828, interested parties in Calcutta and London largely swung behind Rennell's hypothesis.Information from surveys of the Brahmaputra's northern tributaries appeared in the Calcutta Government Gazette and the London-based Asiatic Journal, which reprinted the Gazette's articles with a lag of around seven months.Print mattered in geographical discovery, 47 and these publications actively drummed up public intrigue in both colony and metropole in the geographical 'mystery' of these distant rivers.The Government Gazette even gave its own rendering of the importance of sensationalist narrative strategies to the search for river courses.'The progress of geographical discovery on our north-east frontier,' it told its readers in November 1826, 'has assimilated itself to the development of a well-wrought tale, in which expectation is kept alive by a succession of incidents, promising, yet retarding, the denouement, and disappointing expectation only to excite curiosity.' 48eanwhile, the British surveyors in Assam sought 'ocular and incontrovertible demonstration' of the Irrawaddy and Brahmaputra issuing from the eastern fringes of the Himalaya. 49They admitted, however, that the region's intractable terrain and inhabitants essentially ruled out the possibility of obtaining such evidence. 50Even when threats and inducements led locals to proffer information, topography often seemed to surveyors to become lively and malevolent.Wilcox felt that the 'rugged precipices' beyond the Brahmakund attacked his instruments-'derang[ing]' a thermometric barometer-while the 'crookedness' of the 'chasm of the Brahmaputra' itself occluded the view of the river he desired. 51As winding channels and steep-sided valleys seemed set on evading their gazes, the surveyors turned to other forms of data.
Multiple information sources were both a response to and driver of the multiplication of possible riverine sources. 52etrospective accounts of the surveys attempted to purify messy methods of discovery by implying that quantified discharge measurements of the Brahmaputra's tributaries were the decisive means of establishing the Tsangpo-Brahmaputra as a single object. 53In the 1832 'Memoir' of his work in Assam, Wilcox admitted: 'the question immediately occurs now why we did not take a [discharge measurement] of the river [Dihang]', by that stage the tributary deemed the most likely candidate for joining the Tsangpo and Brahmaputra.'We had not the means,' he continued, 'and the utility of providing them was not so obvious then.' 54 To surveyors in the field, the question of how to establish the Tsangpo's course seemed every bit as unsettled and elusive as the terraqueous landscape itself.
These men placed great value on tales and toponyms gleaned from the inhabitants of the uplands fringing Assam, a very different form of information to quantitative flow data.These communities did not share lowland Hindus' fixation on the Brahmakund, and the British also believed that they were in contact with Tibetans in possession of empirical knowledge of the Tsangpo's course.Wilcox's colleague John Neufville gleaned the single most influential piece of oral evidence, telling of 'a sudden and overwhelming flood [that] poured from the Dihong [sic.]… little more than half a century ago'. 55Rather than seeing it as an unusually severe example of the kind of destructive riverine agency that vexed imperial administrators in Assam, Neufville spun the episode as a vital geographical clue.'Various agricultural and household implements, elephant trappings, and numerous articles belonging to a race, evidently social and civilized, of pastoral and agricultural habits, were washed down in the stream', providing, he claimed, a material link between the Tibetan plateau and the Assam plains via the Dihang. 56Uplanders' oral narratives allowed a deluge of water and material artefacts to become a new stream of geographical knowledge.
The cascade of missives and maps out of Assam during the 1820s that made a splash in Calcutta and Europe was a product of more than the physical and political contours of the region: clashing theories also contributed a great deal of energy.The colonial surveyors' primary antagonist was Julius von Klaproth, a renowned German Orientalist and founder-editor of the journal Asiatisches Magazin. 57Klaproth approached riverine geography as a specialist on Central and East Asia with a keen interest in cartography, especially the Qing-Jesuit surveys. 58His interest in the Tsangpo sprang up in 1825, prompted by a combination of the British surveys in Assam, which he followed in the Asiatic Journal, and the founding of a French-language version of his periodical, Magasin Asiatique. 59In the first of only two issues before publication ceased, Klaproth vouched for the accuracy of Jesuit cartography and reinvigorated the Tsangpo-Irrawaddy theory.His field experience in Asia in the employ of the Russian imperial state and association with scientific explorer Alexander von Humboldt meant that Klaproth initially found favour in some journals in Calcutta and London, creating channels for his intervention to flow back to British surveyors on the fringes of Assam. 60From late 1826 onwards, they couched their attempted discovery of the Brahmaputra's sources as a nationalistic venture to discredit Klaproth and verify the conjectures of the Survey of India. 61fter colonial surveyors withdrew from the northeastern fringes of Assam in 1828, the deluge of information that fed the dispute slowed to a trickle, although not for lack of effort on the part of the Calcutta periodical press.Early in 1830, Gleanings in Science, established the previous year and aimed at a broad public audience, reported that Richard Wilcox's account of the Assam surveys was 'in the course of publication, and will, doubtless, ere long, put the finishing stroke to this controversy'. 62It took advantage of this brief window before discovery was projected to be complete to publish a 'sketch' commissioned by an unnamed 'gentleman' from 'a fine young Chinese priest' who had journeyed along the banks of the Tsangpo from Lhasa to Bengal (Figure 5). 63otwithstanding the correspondent's guarantee that his informant 'seemed ingenious and candid', 64 the Tsangpo-Brahmaputra's reduction to a wavy line accompanied by four scant toponyms met with less acclaim among European men of science than Wilcox's 155-page 'Memoir' to which Gleanings had alluded.Published in 1832, it concluded with a lengthy attempt to demolish Klaproth's logic and Chinese sources, 61 Anon., 'Review of books', Asiatic Journal, vol. 25, 1828, pp.472-477, here p. 473; Hodgson to Wilcox, 17 November 1826, quoted in Phillimore, Historical Records.Vol.III, p. 57; Henry Yule, commentary in Thomine D'Mazure, 'Memorandum on the countries between Thibet, Yunan and Burmah', Journal of the Asiatic Society of Bengal, vol. 30, 1861, pp.367-383, here p. 376. 62Anon., 'On the identity of the Sanpu and Irawadi rivers', Gleanings in Science, vol.2, 1830, pp.66-67, here p. 66. 63 Ibid. 64Ibid., pp.66-67.for which purpose Wilcox specially altered the Survey of India map of 1828 based on British surveys in Assam. 65Besides reducing the scale, he made only a single, highly significant, change (Figure 6).A thin line labelled 'M.Klaproth's river' joined the Irrawaddy, conspicuously separate from the authoritative, double line labelled 'Tsanpo R r ', which joined the Brahmaputra via the Dihang.Historian of cartography Brian Harley famously claimed that 'as much as guns as warships, maps have been the weapons of imperialism'. 66During the Tsangpo debates of the 1820s, maps were weapons less in processes of dispossession-which in northeast India tended instead to operate through written treaties and practices on the ground rather than authoritative cartographic images-and more in internecine disputes within European geography.This is not only a retrospective claim: some of the men embroiled in these debates recognized that maps of Asia's rivers were primarily powerful in expressing dissensus between geographers.Bemoaning the continued controversy over the Tsangpo's route, British administrator-scholar Henry Yule argued in 1882 that the 'beautiful French and German maps … founded on Klaproth's theory' had exercised 'a curious effect' on the question of river courses.French missionaries, he conjectured, carried these 'erroneous maps' to Tibet, where the images influenced the information that they communicated back to Europe. 67In British circles, however, Wilcox's victory was assured-at least for the following four decades-and maps played an important part in this.Klaproth's paper Tsangpo, although aesthetically appealing, admitted uncertainty as its blue-tinted channel gave way to an uncoloured double-track of dashed lines where the Qing-Jesuit data stopped in the eastern Tibetan plateau (Figure 7).Wilcox's reduction of 'Klaproth's river' to a frail line that petered out without joining a major river channel, let alone reaching the Bay of Bengal, exploited this hesitation while admitting no equivalent visual indeterminacy in his own Tsangpo.
This phase of finding the river was not defined by knowledge 'circulating' through a complete spatial circuit or 'flowing' consistently through already extant channels.Like the apparently unruly and fragmented river system they represented, texts and maps travelled various courses, and information gushed for brief periods before ebbing.Just as the substance of scientific dispute was a torrent of invective unleashed by the shifting terrains of colonial geopolitics and scholarly publications, so the stilling of the waters in the 1830s constituted discovery.

Data deluges
Through the middle decades of the nineteenth century, the Tsangpo-Brahmaputra rose and fell with monsoon rains and Himalayan meltwater, subtly reshaping its own course through mountains and valleys.For institutions of British imperial geography, though, it remained a static object, as established in the late 1820s and early 1830s.The imagined fixity of the Tsangpo-Brahmaputra was mostly passive rather than active, constituted by the absence of discussion rather than rehearsals of its upper course in print or on boats.An important factor in stabilizing discovery during this period was colonial Assam's tea industry, which advanced during the middle decades of the nineteenth century thanks to water from the Brahmaputra's feeders, leading to rapid demographic and ecological changes.The plantation economy shifted hydrological focus away from the mountain streams that fed the Brahmaputra and towards the watercourses in the valley that were harnessed to European capital. 68The river's main channel in lowland Assam became less mysterious and foreboding to the British as traffic increased, banks were stabilized, and ever more of the 'jungle' that had cloaked its floodplains was destroyed to make way for tea 'gardens'. 69he most significant challenge to the assumption of the river's stability prior to the late 1860s came from an alternative framing of the Brahmaputra as an unusually capricious geological agent, bearing sufficient power to spill over the divide between temporalities of natural and human history. 70In an 1863 paper to the Geological Society of London, the Bengal-based indigo planter James Fergusson conceptualized and depicted (Figure 8) the 'young and active, and roaming' lower Brahmaputra as conducting an ongoing 'invasion of Gangetic territory'. 71The lexicon of intentionality and sovereign power persisted as colonial geologists' focus on the Brahmaputra migrated upstream over subsequent decades.It reappeared early in the twentieth century in the conjecture that the Tsangpo had 'until quite recent times' streamed westwards across the Tibetan plateau, before being 'beheaded' and having its flow reversed by the Dihang. 72lthough British geographers largely overlooked geologists' addition of a temporal dimension to the Brahmaputra's instability, a deluge of data and debates from the late 1870s rendered their own version of the river a little less discovered than it had been for the preceding 50 years.This was a tumultuous phase in British economic and military adventuring.The geographical debates of the 1820s related to the annexations and strategic concerns east of Bengal; those of the later period intertwined with commercial and territorial expansion to the north of India and Burma.Growing security concerns for British India's highly profitable tea frontier in Assam prompted the dispatch of survey parties to the region in the 1860s and 1870s. 73Instead of the narrow field of vision of the route surveys of Wilcox and his colleagues in the 1820s, these larger trigonometrical surveys took long-range theodolite sightings from prominent hilltops. 74This alternative visual technology allowed for renewed doubts about the Tsangpo-Brahmaputra connection.'After looking north into the mountain region from the two highest peaks', the senior surveyor was 'impressed with this idea' that the Subansiri, and not the Dihang further to the east, connected the Tsangpo to the Brahmaputra. 75ismissing river discharge data obtained 50 years earlier as unreliable, surveyors of the 1870s conducted their own flow measurements of the Brahmaputra's four major northern feeders. 76As with many other attempts to know capricious and changeable bodies of water, these men found that attempts to achieve precision quickly gave way to efforts merely to 'estimate truth'. 77Their supposedly authoritative devices for measuring stream velocity failed to survive either the journey to, or riverine conditions in, Assam.Instead, they had to resort to a hastily assembled instrumental repertoire, including Assamese canoes and wooden discs and tin tubes topped with cotton wool (Figures 9a and b). 78After adjusting the measurements on the basis of nearby residents' information on the rivers' annual fluctuations and then feeding the data into an avowedly 'clumsy' mathematical formula, 79 the surveyors pronounced the Dihang's discharge to be approximately triple that of the Subansiri. 80On this basis they ruled out the possibility that the latter was the continuation of Tibet's Tsangpo.However, the surveyor-general of India John Walker was concerned that surveyors had not travelled far into the uplands owing to fears of clashes with local inhabitants and claimed that these streams of numbers did not establish discovery. 81What he termed the 'main question' of whether the Tsangpo entered the Irrawaddy or the Brahmaputra remained unresolved in the absence of a full survey of the river course. 82s surveyors in Assam worked on discharge measurements, another version of questioning and rediscovering the river was underway in southern Burma.A midranking government hydraulic engineer working on embankment projects in the Irrawaddy delta, Robert Gordon, produced a lengthy report in 1879.83 His core contention appeared on the opening page: 'the Irrawaddy is continuous with the Sanpo of Thibet'.84 Gordon drew extensively on accounts of British traders and soldiers who during the preceding two decades had sought to extract teak, minerals, and metals from Upper Burma and to tap commercial routes into southwestern China.85 T. T. Cooper, one the most prominent of these trade advocates used 'Chinese maps' to argue for the identity of the Tsangpo and the Irrawaddy, and thereby to naturalize the flow of goods from the continental interior to the Bay of Bengal (Figure 10).86 Gordon emphasized that Cooper and 'all who have visited the upper part of the Irrawaddy' had experienced 'astonishment' at the river's size and lack of tributaries.87 This evidence corroborated Gordon's own engagement with the floods and erosive power of the lower Irrawaddy.From his vantage points by, on, and sometimes in the river, the Irrawaddy's vastness was striking, evoking a distant source.This echoed Rennell's experience of the Brahmaputra a century earlier: witnessed in person, Himalayan rivers appeared to many British observers so gigantic and potent that they must emanate from deep in the heart of the continent.In such cases, what Johannes Fabian terms 'ecstatic' moments in the field precipitated data-intensive discovery claims.88 Gordon couched his Tsangpo-Irrawaddy in direct opposition to the Survey of India establishment that had sustained and bolstered Rennell's Tsangpo-Brahmaputra. Latching onto the Survey's recent reopening of the Tsangpo question, he contrasted the complex array of feeders into the Brahmaputra in Upper Assam with the simplicity and singularity of the Irrawaddy's main channel.By considering each of the Subansiri, Dihang, Dibang, and Brahmakund as potential connectors of Tsangpo and Brahmaputra, British Indian surveyors had, Gordon charged, confusingly placed 'four Sanpos in the field of conjecture'.89 He also foregrounded his hydraulic expertise, proclaiming the accuracy and comprehensiveness of his year-round discharge measurements at various locations on the Irrawaddy relative to the hasty and improvised efforts of surveyors in Assam.90 Gordon's floats, flow graphs, and discharge basin maps all arose from Burma's specific version of the colonial efforts across the subcontinent to 'tame' river systems, which became more ambitious and destructive during the middle decades of the nineteenth century.91 Gordon aimed to master the Irrawaddy's 'yearly inundat[ions]' in the service of 'reclaiming a large tract of fertile ground' for rice cultivation.92 At the same time, he was sharply aware of the enormous challenge of making the Irrawaddy fit for colonial purposes, fearing potential 'disaster' and large-scale human suffering if the embankments proved unequal to surging water.93 This acute sense of riverine agency encouraged imaginative leaps northwards, far beyond the gauge dials and earthworks that lay in his quotidian field of view and into what he termed 'the vexed question of the geography of its upper sources'. 94The fraught terraqueous commitments of colonial Burma's political economy in the 1870s made possible Gordon's Tsangpo-Irrawaddy just as the fluvial entanglements of the colonial invaders of Assam in the 1820s allowed for the Survey of India's Tsangpo-Brahmaputra.
These divergent forms of imperial possession and extraction did not, however, wholly determine riverine geographies.In contrast to the Survey of India, Gordon placed his work within a global framework of the latest attempts to comprehend and control large river systems, especially work on the Mississippi by United States Army engineers. 95He also deployed a spatial logic that had not been in play among surveyors of the 1820s: watersheds and drainage basins. 96This distinct framing of how the Tsangpo question might be answered was also a distinct framing of what a river is.The discharge of the river in isolation signified little; the river channel was not the significant object.Instead, the outflow of water had to be considered in tandem with inputs: the area that fed the river and the rainfall that fed that area.Drawing on the rapid increase in data production, accumulation, and processing with the advent of the Indian Meteorological Department in 1875, 97 Gordon conceptualized rivers as complex systems comprising hydrography, climate, and topography.On this basis, he 89 Gordon, Report, part I, p. 51. 90 Ibid., pp. 2, 48-49. 91Ibid., pp.39-57; D'Souza, Drowned. 92Gordon, Report, part III, pp.1-2. 93Ibid., p. 6. 94 Ibid., pp.7-8. 95   mapped hypothetical drainage basins (Figure 11) and extrapolated rainfall measurements (Figure 12) that could account for the discharge of the Brahmaputra's tributaries without having them connect to the Tsangpo.
Although innovative hydrological theories and technologies were integral to his argument, Gordon recognized that discovering the Tsangpo's route entailed winning over geographers rather than irrigation engineers.In turn, this required relating his contentions to older sources-both informational and riverine.Although an increasingly sharp divide between 'armchair' and 'field' geographers was emerging in London around this time, 98 incorporating interpretations of past texts and maps alongside first-hand observations remained essential to establishing credibility in colonial South Asia.Gordon cartographically compared the river courses in the eastern Himalaya postulated by each significant past contributor to Tsangpo geography (Figure 13).He also sought to rehabilitate Qing Chinese and French Jesuit accounts. 99When criticized by key players in British imperial geography (by this stage concentrated at the RGS rather than distributed across various institutions and journals as in the 1820s), Gordon calibrated his contentions with upland communities' toponyms just as Wilcox had. 100 In his paper to the RGS in 1885 Gordon added further information from Alaga, a Burmese man trained by a British surveyor and dispatched to track the upper Irrawaddy. 101The map accompanying the published version of this paper prominently stated that the Tsangpo's course was rendered 'according to native authorities'. 102ordon's marriage of quantification and oral testimony harked back to British attempts to discover the Tsangpo's course in the 1820s.It nevertheless failed to advance his cause among the gatekeepers at the RGS.He was undoubtedly swimming against the stream: the key authority on the subject in the Society and respondent to his paper was John Walker, erstwhile surveyor-general of India and overseer of the very surveys in 1870s Assam that Gordon critiqued. 103Although his key claims failed to gain traction, Gordon's intervention altered the framing of the Tsangpo question.His platform at the RGS forced the Survey of India to take discharge basins and climate seriously when they theorized river routes.While hewing to a distinct notion of discovery (as discussed below), Walker felt compelled to make the case that heavy precipitation in Burma caused by 'vast exhalations of moisture' from the Indian Ocean could account for the Irrawaddy's substantial volume. 104ordon also contributed to the Survey of India's reorientation of southern Asia's riverine geography towards the Tibetan plateau.Prohibited from entering Tibet, Europeans' narratives filled a void of reliable information with visions of the region as immensely significant in spiritual, spatial, and ethnographic terms. 105Like their predecessors' critical uptake of Hindu ideas of the Brahmakund, British imperial agents in the late nineteenth century disparaged Tibet's centrality within some Buddhist cosmologies while retaining the notion that this was an extraordinary space.Gordon's proposal that the orography of Tibet structured the hydrography and climate of central and southern Asia (Figure 14), constituting 'the great permanent centre of the monsoon', was part of this broader trend. 106He also contended that 'the great [Tibetan] plateau itself can be made to yield indications of its [climatic] action by the great rivers issuing from it and entering British territory'. 107For Gordon, then, the Irrawaddy's fluctuations spoke of natural processes in Tibet that would otherwise have remained inaudible.
Walker understood the capacity of the rivers on the southern side of the Himalayan range to convey information about their upper courses to the north of the range in directly opposing terms to Gordon: they did not communicate nearly enough.He accordingly believed that the answer to the Tsangpo question would come from 'native explorers' trained by the Survey of India to glean geographical information in regions of burgeoning imperial rivalry with Russia and China. 108The journeys of these men beyond British India's frontiers were even more shrouded in colonial mythmaking than most acts of imperial exploration.109This is especially the case for Kinthup, the Sikkimese man who undertook an expedition from 1880 to 1884 that focused on the Tsangpo's course through the eastern Himalaya.Kinthup was illiterate and his account of the river was delivered verbally to a Survey of India translator two years after he returned to colonial territory.His transcribed and translated recollections were then refracted again when 'compiled' by a senior British Survey of India official. 110Certain features of this multi-authored account became iconic through subsequent retellings, including in recent public-facing histories. 111Particularly celebrated was Kinthup's attempt to float cut logs down the Tsangpo with the intention that they would be collected from the Brahmaputra in Assam, their passage neatly proving the identity of the two rivers. 112This attempt to use the flow of water to facilitate the flow of geographical information failed: the logs were not picked up. 113However, the information that Kinthup provided to the Survey of India in 1886 bolstered the theory that the Dihang connected the river of the Tibetan plateau with that in the Assam valley (Figure 15). 114mong leading figures and institutions of British imperial geography, doubts persisted about Kinthup's trustworthiness on account of his race, limited training, and lack of records rendered in the field.As late as 1913, the president of the RGS opined that Kinthup's 'character for veracity still lies in the womb of the future'. 115The Tibetan government's violent reprisals when it learnt in the late 1880s that some of its subjects had assisted British India's 'native explorers' led the Survey of India to stop its programme of clandestine journeys into Tibet. 116Against this backdrop, the written and visual products of Kinthup's travels were not a means of closure but instead opened a new Tsangpo question.The narrative of his journey stated that around the point that the river enters the Himalayan range it 'falls over a cliff called Sinji-Chogyal from a height of about 150 feet.There is a big lake at the foot of the falls where rainbows are always observable.' 117 Appearing in the map as 'Falls of the Sangpo', this feature became a point of intense focus for imperial surveyors and geographers over the subsequent decades, as discussed in the next section.Having once again washed away the Tsangpo-Irrawaddy from the high places of British imperial geography, the deluge of information and conjectures on the Tsangpo during the 1870s and 1880s abruptly slowed to an intermittent trickle.In its wake the Tsangpo-Brahmaputra remained less than fully discovered.

Maintaining mystery
During 1870s and 1880s Tsangpo theories invoked different rivers to those in the 1820s, replacing fluvial channels with a complex hydrological system.Over the following decades, another new object of discovery came into focus.As in previous iterations of the Tsangpo question, this version lay at the confluence of debates over what counted as geographical knowledge, new horizons of British colonial expansion, and input from specific Asian cosmologies.During the decades either side of the turn of the twentieth century, geography as an exploratory endeavour had become a victim of its own supposed success.'Geography triumphant', as Joseph Conrad termed the era in which the major features of the earth's surface had been established, was a time of anxiety as much as of celebration. 118Discovering the Tsangpo was bound up with the desire among many involved in British imperial geography for exploration to retain its capacity for substantial breakthroughs.Europeans counted Tibet among the world's last places of 'mystery'-ripe for, but resistant to, knowledge gathering. 119The British invasion of Tibet in 1903-1904 precipitated a flood of objects and information, which, along with the tumult induced by Chinese expansion into eastern Tibet over the following decade, set the stage for a revised version of riverine discovery. 120he older question of whether the Tsangpo joined the Brahmaputra resurfaced occasionally after the turn of the twentieth century.Important here was that wouldbe contributors to the hydrography of frontier India and Burma still felt that their objects of study were unusually capricious.Like predecessors from Rennell to Gordon, the surveyor L. A. Bethell found that encountering rivers in person could swamp expectations generated by paper rivers in previously consulted texts and maps.While with the Tibet invasion party in 1903, Bethell 'got to know the river [Tsangpo] rather well' and was 'struck' by the fact that 'it was a very big river indeed, and wherever it came out in India it must be a thing well worth looking at'. 121His subsequent experience of the Dihang in northern Assam fell short of his expectations to such an extent that he refused to believe it linked the Tsangpo and Brahmaputra.Bethell's justification of these doubts intermingled the quantitative logic of hydrological systems and claims of extraordinary fluvial agency: 'Rivers do marvellous things in that part of the world.They twist and screw themselves between barriers into unexpected places.' 122 In keeping with 'geography triumphant' more broadly, Bethell also saw value in overturning an established discovery.'If the Dihong and the Tsang-po are not one,' he told the RGS, 'then it remains to us to find out where the Tsang-po does go to.That is a problem worth having in these days when unsolved geographical questions are rare.' 123ike Gordon 30 years earlier, Bethell received short shrift from the gatekeepers of imperial geography.They had by this stage largely turned towards the new Tsangpo question.During the interstitial period between the route surveys of 'native explorers' in the 1880s and the invasion of 1903-1904, various agents of empire continued to seek corroborating evidence of Kinthup's falls.Tibetan traders and monks in and around the colonial hill station of Darjeeling were central to this knowledge-gathering effort. 124This new channel of information through Sikkim combined with the new object of riverine discovery-an extraordinary topographical feature rather than an entire hydrological system-to allow Tibetan images and texts a role in British geographical debates.Among this new evidence was a sketch by a man who resided in the vicinity of the falls (Figure 16), collected (and probably commissioned) by Darjeeling-based surgeon Laurence Waddell. 125addell positioned the image as pertinent both to imperial geography and to his vitriolic critique of Tibetan Buddhism as analogous to Catholicism in the venality of its monastic elite. 126Forwarding it to the RGS in 1895, he parsed the depiction of a divine figure in the rock behind the falls as representing the tyranny and opportunism of 'local Lamas', who told 'awestruck pilgrims' that the carving was 'a king-devil' that they had 'placed under a spell'. 127As discussed below, British imperial personnel with a less sceptical attitude towards Tibetan Buddhism soon produced alternatives to this fundamentally inaccurate reading of the falls' spiritual significance.Even those British geographers who considered the Tibetan sketch and Kinthup's narrative to be valid contributions to discussions of the Tsangpo tended to believe that true discovery of the falls required in-situ visual verification by a European.They perceived that seeking this single sublime view had the advantage of extending 'heroic' geography into the 127 Waddell, 'Falls', p. 258.twentieth century.In this vein, the retired British Indian frontier surveyor and later president of the RGS Thomas Holdich proclaimed in 1913 that 'we may congratulate ourselves that there is at least one little corner in this far-off north-east frontier of ours about which a little mystery still hangs, for surely the unexplored portions of the globe now are getting all too narrow for our practical explorers'. 128hile geographical discovery was strictly defined along racial lines in London lecture theatres, practices in Asian borderlands were more complex.Tibetan modes of spatial configuration and discovery provided a crucial basis for British exploration in the eastern Himalaya during the 1910s and 1920s.In 1913 two officers in the Indian Army, Frederick Bailey and Henry Morshead, split off from a military party surveying the Dibang tributary of the Brahmaputra.They headed for Mipi, a village on the Dibang's upper reaches inhabited by Tibetan and Bhutanese communities. 129These migrants had established themselves in Mipi as it lay in the region of Pemak ö , a beyul (hidden land) ordained as a place of refuge by Guru Rinpoche, the originator of Tibetan Buddhism in the eighth century ce. 130The spatial logic of the beyul configured Pemak ö as the terrestrial body of the goddess Vajravārāhī.Topographical features corresponded to the deity's bodily features or cakras: the Tsangpo was her spine, while her womb-Pemak ö 's inner, most secret sanctuary-lay in forest-clad mountains south of the Tsangpo gorge. 131As well as fleeing Qing imperial expansion into eastern Tibet during the preceding decades, some residents of Mipi were also actively engaged in 'opening' Vajravārāhī's cakras. 132Their tantric Buddhist version of spatial discovery became a vital element in colonial attempts to know the Tsangpo-Brahmaputra.
Having resided in Tibet as a trade agent following the 1903 British invasion, Bailey took a considerably more sympathetic approach to Tibetan Buddhism than Waddell.Before venturing into Pemak ö , he read the first European book on the region (published only a year before), 133 so already understood it as 'a sort of "Promised Land"'. 134Buddhists in Mipi were critical to Bailey's exploration, constituting, he later recalled, 'my chance of getting through to Tibet from the Assamese side'. 135Bailey's month-long stay in the village fulfilled his expectations.He relied on the village's Tibetan headman Gyamtso (Figure 17) for information, letters of recommendation to local potentates, and the recruitment of three guides, including a hunter and a monk, and ten porters for the journey northwards to the area of the falls. 136The monks of Mipi , performed what Bailey termed a 'special service' to ensure the party's safe passage as it started along a route previously used by Tibetan migrants. 137Gyamtso's knowledge of the region, Bailey recounted, 'provided an excellent justification for the exploration … [because] I was not going to launch into the blue, but merely to follow certain routes which I 137 BL, MSS Eur F157/208, ff.9-10.had already outlined'. 138Far from suggesting that he was venturing into terra incognita, then, Bailey actively acknowledged that he followed established paths through the beyul.Gyamtso also provided Bailey with the first indication that the Tsangpo's turn southwards in the region of the falls was not an abrupt right angle as the Survey of India's rendering of Kinthup's information had it, but an extended, looping bend during which the river cut a deep gorge between high mountains. 139Bailey hastily rendered a sketch map based on this information before leaving Mipi (Figure 18), rightly considering it an improvement on the Survey's depiction of the area. 140ibetan Buddhist geography structured Bailey and Morshead's entire itinerary, linking villages and monasteries constructed by émigrés engaged in opening the beyul. 141Bailey was both aware and content that this was the case.'The idea never occurred to us,' he later wrote, 'that any expedition should be routed and highly organized.There was so much to discover that once we left Mipi wherever we went and whatever we saw was important.' 142He and Morshead relied on cross-referencing the numerical survey data they generated with Pemak ö 's narrative and orally recounted topography.Their ability to navigate and map the region rested on sightings of Gyala Peri and Namcha Barwa, the enormous mountains flanking the Tsangpo gorge, by their Tibetan collaborators, who, Bailey noted, 'supposed them to be the breasts of [Vajravārāhī]'. 143ibetan and British knowledge became further entangled as the party approached the Tsangpo gorge.Bailey had already heard hints that the falls were not the 150-foot marvels reported in Kinthup's Survey of India narrative: 'an old man who knew something of the Falls' claimed 'the river fell about 30 feet (he said twice the height of a walnut tree which was near the house)'. 144Upon reaching the gorge, Bailey and Morshead confirmed that the falls were in a small tributary rather than the main stream of the Tsangpo.Contrary to Waddell's pejorative description, the carving behind the torrent was of Shinje Chogyal, the lord of the Underworld, marking a gateway for travellers into Pemak ö . 145 The gigantic falls fantasized about in RGS lecture theatres were, then, an excessively literal translation of the feature's significance within the beyul into the aesthetics of exploration-based geography.Bailey and Morshead's subsequent journey into the inner sanctum of the gorge, which, Bailey thought, 'might perhaps conceal a fall rivalling the Niagara or Victoria Falls in grandeur', 146 relied wholly on devotional infrastructure.They used the pilgrim path along the river's right bank skirting Namcha Barwa, obtaining guides and porters by threatening violent punishments in order to tap into 'ula', a form of labour demand developed to aid Buddhist travellers. 147Bailey also benefitted from a chance encounter with a monk on pilgrimage from Pemak ö to 147 Ibid., p. 149; BL, MSS Eur F157/208, f. 105.Lhasa proceeding in the opposite direction, using the path the monk's retinue had cut through thick foliage on steep slopes to progress eastwards deeper into the gorge. 148lthough they left approximately 45 miles of the Tsangpo unexplored, Bailey and Morshead photographed falls 30 feet high in the main stream of the river and reported the vast scale of the gorge.These 'discoveries' were premised not on the region's pristine isolation, but its deep incorporation into Buddhist cosmology.British knowledge of the gorge ran along channels carved by Tibetan conceptions of beyul space.In addition, the material and human infrastructures of pilgrimage provided the crucial first stage in the winding and laborious passage of information into representations-maps (Figure 19), photographic prints, numerical data, and texts-that the surveying establishment in India and major scientific societies of Europe and North America considered sufficiently authoritative to declare that the river had been found.Bailey's receipt of the RGS's Gold Medal in 1916 provided official closure to the discovery of the Tsangpo.
What did discovery mean to the main players in this episode of British imperial geography?Bailey was insistent that he had merely verified Kinthup's work. 149Moving through Pemak ö , he meticulously calibrated his own observations and Tibetan oral testimony with Kinthup's information. 150After returning to British India and seeking out Kinthup in Sikkim, Bailey was satisfied that the notion of a large waterfall in the course of the Tsangpo was a translation error rather than Kinthup's fabrication. 151esponses to Bailey's paper to the RGS in June 1914 detailing his survey were in keeping with the broader ambivalence of 'geography triumphant'.Discovery was an anticlimax.The president of the Society glumly opined that 'travel tends to dispel romance', in this case by disproving in narrative and photographic forms 'the legend of an enormous waterfall clouded in rainbow haze'. 152An attempt to reinstate an element of this sublime fiction can be discerned in a dispute between Bailey and the editors of the Society's Journal over a photograph in the published version of his paper (Figure 20).After seeing the draft version of the article, Bailey complained about the excision of the bottom portion of this image showing 'the part of the river clean of spray [i.e., below the falls, which] gives a much better idea of it'.The editors seemingly ignored Bailey's request to reinstate the complete photograph-a particular instance, perhaps, of the desire to hint at a grander vista and thereby leave discovery incomplete. 153

Conclusion: The river mutable
Although most imperial surveyors and geographers understood Bailey and Morshead's journey to have answered the Tsangpo questions for good, alternative renderings of 150

Figure 2 .
Figure 2. Portion of Survey of India, 'Geographical plan of the N.E.Frontier of Bengal with part of Assam' (March 1824).Source: British Library, London, India Office Records, H/678, map 3. © British Library Board.

Figure 8 .
Figure 8. James Fergusson, 'Map of the Rivers of Bengal' (1863).Source: Quarterly Journal of the Geological Society of London, vol.19, 1863, facing p. 321.Reproduced by kind permission of the Syndics of Cambridge University Library.

Figures 9 a
Figures 9 a and b. H. J. Harman, sketches of instruments used to measure stream velocity (1879).Source: Journal of the Asiatic Society of Bengal, vol.48, no. 2, 1879, p. 6. Reproduced by kind permission of the Syndics of Cambridge University Library.

Figure 10 .
Figure 10.Portion of T. T. Cooper, 'Map of the head waters of the Kin Char Kiang, Lan Tsan Kiang, Now Kiang, and Great River of Tibet' (1869).Source: Proceedings of the Royal Geographical Society of London, vol.13, no. 5, 1868-1869, facing p. 392.Reproduced by kind permission of the Syndics of Cambridge University Library.

Figure 11 .
Figure 11.Portion of Robert Gordon, 'Sketch map showing the probable course of the Sanpo of Tibet to the Irawadi of Burma according to native authorities' (1885).Source: Proceedings of the Royal Geographical Society and Monthly Record of Geography, vol. 7, no. 5, 1885, facing p. 352.Reproduced by kind permission of the Syndics of Cambridge University Library.

Figure 13 .
Figure 13.'Maps illustrating Mr R. Gordon's paper' (1885).Source: Proceedings of the Royal Geographical Society and Monthly Record of Geography, vol. 7, no. 5, 1885, facing p. 352.Reproduced by kind permission of the Syndics of Cambridge University Library.

Figure 17 .
Figure 17.Frederick Bailey, 'Gyamtso-Headman of Tibetans at Mipi' (1913).Source: British Library, Photo 1083/36(45).© British Library Board.monastery, established as part of the 'opening' of the beyul of Pemak ö, performed what Bailey termed a 'special service' to ensure the party's safe passage as it started along a route previously used by Tibetan migrants. 137Gyamtso's knowledge of the region, Bailey recounted, 'provided an excellent justification for the exploration … [because] I was not going to launch into the blue, but merely to follow certain routes which I