A place for meteorology

Historians have recently sought to go beyond centre–periphery models of scientific practice and ‘diffusion’, which position colonial sciences as subservient, derivative and inferior to metropolitan sciences.Footnote ¹⁰ Alongside broader moves in imperial historiography, the primacy of the metropole has been rethought, and networks of colonial science have come to be read as polycentric, often transient, with relations not of dependency but of reciprocity and exchange.Footnote ¹¹ Richard Grove, for example, has emphasized the contributions of colonial scientists and savants to emerging understandings of atmospheric patterns and interconnections from the late eighteenth century onwards, while Gregory Cushman has emphasized the complex transnational and trans-imperial exchanges which shaped the development of tropical meteorology, both in the tropics and in the meteorological citadels of temperate capitals.Footnote ¹² Mauritius has provided its own grist to the revisionist mill, with William K. Storey's history of sugar cane botany providing further evidence of the existence of distinctive colonial cultures of scientific knowledge making, organized around local needs, attuned to local structures of authority and sociologies of trust, and productive of their own truths, many furnished with the ability to travel beyond the local contexts of their making.Footnote ¹³ Retooling the language of centricity, Lissa Roberts has recently described French Mauritius as a location which a number of actors sought to position as a ‘centre of accumulation’. Writing against McClellan and Regourd's arguments about the functioning of colonial science in the context of a centralizing French state,Footnote ¹⁴ Roberts argues that processes of centralization are best studied as multiple, contested and essentially polycentric. Efforts to make Mauritius a centre of accumulation of wealth, information and power played out through multiple connections, through alliances made and battles fought at a range of spatial scales. Centralization efforts made French Mauritius not only an object of intensified knowledge making itself, with island surveys of land and resources proliferating, but also an entrepôt for knowledges concerned with, and relevant to, wider processes and spaces – hydrography, navigation, botany, acclimatization, agricultural improvement. While this culture of centralization might have been closely associated with the structures and styles of French administration, the bilingual meteorological culture of the mid-nineteenth century reflects a deepening of this urge to position Mauritius as a centre for the accumulation of information and the dispersal of new knowledge.

Mauritius, dubbed the ‘star and key of the Indian Ocean’ in the mid-eighteenth century, was described in 1807 by the French colonial official Félix Renouard de Sainte-Croix as ‘a central geographical point between every other place in the world’.Footnote ¹⁵ This notion of geographical centrality continued under the British, who took control of the island in 1810 in a bid to secure their hegemony over the trading routes to India. Strategically important, Mauritius became a garrison colony, where the intention was never to promote widespread British settlement but simply to guarantee defence of a crucial maritime trading route.Footnote ¹⁶ For the next few decades, a British ruling and military class would coexist – often in great tension – with a landowning French agricultural elite, and from 1835 with freed African slaves and indentured Indian labourers. British trading rules meant that the island's own economic survival depended on the production of export commodities, and the island's reliance on sugar cane, thought to be the crop most resilient to tropical cyclones, expanded rapidly. From 4,148 tons in 1816, production reached 150,480 tons by 1862.Footnote ¹⁷

Lying within the tropics, Mauritius is cooled for most of the year by the south-east trades and enjoys reasonably reliable rainfall. Occasionally, however, the rains fail, while during the warm, humid summer months the island is periodically visited by violent cyclones. As such, study and speculation about the weather have long formed part of the island's scientific, economic and cultural life. The formal making of meteorological records on Mauritius can be dated to 1774, when Jean-Nicolas Céré, the gardener and agronomist who would become director of the Botanical Garden in 1775, began making daily weather observations.Footnote ¹⁸ Mariners plying the waters around the island and traversing the Indian Ocean had also by this time begun piecing together the patterns of oceanic and atmospheric circulation which both drove and arrested trade between the Mascarenes and Asia.Footnote ¹⁹ The first public observatory on the island was set up in 1832 by Colonial Lloyd, the surveyor general, at the harbourside in Port Louis, where daily weather observations were made alongside seemingly more functional work on magnetism and timekeeping. But despite the new regularity this afforded to Mauritian meteorology, the observatory rarely published its findings, thus limiting its impact in scientific, maritime and agricultural communities, and by the 1840s the observatory's equipment was considered sorely outdated. This was a situation which greatly irked the young Charles Meldrum, who in 1848 travelled to the island to take up the position of professor of mathematics at the Royal College. Meldrum had graduated with an MA from the Marischal College in Aberdeen in 1844, and went on to work on the educational staff of the Bombay Presidency before moving to Mauritius.Footnote ²⁰ Meldrum quickly became interested in the meteorology of the island, and in 1851 he secured the backing of fellow men of science, government officials, planters, mariners and military officers for the formation of a meteorological society, under the official patronage of the governor. The first president was colonial secretary and future governor of the Bahamas Charles J. Bayley. Lieutenant-Colonel Robe, deputy quartermaster general and former lieutenant governor of South Australia, served as vice president along with the naturalist François Liénard, while Meldrum and Henri Bousquet, the acting government observer, jointly took on the role of secretary.

The Meteorological Society of Mauritius, the first of its kind in the British colonial world, took as its first aim the procurement of high-quality instruments with which observations could be made at Mauritius and its dependencies, such Rodrigues and the Seychelles. While the Admiralty in London had expressed their sympathy with the society's aims, a hoped-for grant was not immediately forthcoming, and the society would have to rely on its own funds. Although he had been appointed vice secretary, the founding of the society had understandably irked Bousquet, a Creole meteorologist ‘of unbounded enthusiasm’,Footnote ²¹ who perhaps interpreted it as ‘a reflection on his own inadequacy’ and initially resisted the request for hourly observations to be made at the Port Louis observatory.Footnote ²² The struggle for meteorological authority became a struggle over space. The society was permitted to rent rooms at the observatory from the government for its meetings and a nascent library, but the rent was set so high – amounting to about half the government's own contribution to the society's coffers – that they were forced to vacate them in 1853.Footnote ²³

Even if Mauritius was a ‘peculiarly’ and naturally adapted spot for meteorology, the new society had no natural right to the observatory. Support was again sought in London. In 1854 Edward Sabine expressed his enthusiasm for the new organization, having regretted that

The following year, after representations from the Colonial Office in London, where the value of the observatory sciences was slowly being acknowledged, the society was allowed back in. The Secretary of State commanded that the observatory buildings be made available to the society rent-free, while the Admiralty, recently convinced of the value of a systematic approach to maritime meteorology,Footnote ²⁵ also agreed to an annual subsidy of fifty pounds to help with publishing costs.Footnote ²⁶ With that, the instruments which had been ordered from London and Bombay two years earlier could finally be installed and put to use. Barometers were sent to Rodrigues and the Seychelles, and monthly reports received in return. Not long after the time when Port Louis could not seem to claim half an observatory, the town could briefly boast two, as the Royal Engineers added a node in their rapidly expanding network of imperial magnetic observatories some three hundred yards from the harbourside observatory, with weather observations made every six hours until 1858.

The society aimed from the outset at developing a wide membership structure, not just to bring in funds, but also to ensure that its meteorological truths could be ‘serviceable’ to all interested parties on the basis of their collaborative production. This was particularly significant on an island where, as Meldrum opined,

This was a moment when the institutionalization of meteorology as a distinct branch of physical science was starting to take hold in Europe, albeit amidst widespread disagreement as to its rigour, precision and practical utility.Footnote ²⁸ Many of meteorology's potential patrons showed little faith in the use-value of the science, with its low levels of predictive skill and lack of general physical theory, so sympathizers took every rhetorical opportunity to convince doubters that this was a science worth backing. Many of the early speeches and exchanges at the Meteorological Society meetings featured vaulting statements about the centrality of weather to everyday life, particularly in an island colony threatened regularly by drought, storms and climatically linked diseases, along with calls for the weather-minded polity to put some faith in a nascent science of the atmosphere. By 1853 the itinerant army surgeon Alexander Thom, who in 1845 had published An Inquiry into the Nature and Course of Storms in the Indian Ocean South of the Equator,Footnote ²⁹ had been elected the next president of the society, although his term was to be cut short by illness and a return to England, where he died the following year. Much like William Reid's earlier work in the West Indies, Thom's Inquiry used a number of ships’ logs to reconstruct the behaviour of recent storms, most notably one which struck Rodrigues in 1843. His work offered one of the clearest early statements of the theory that gyratory storms form through the meeting of opposing currents of air, but his target audience also included sailors in need of advice on how to anticipate the bearing of storms, and on how to position themselves to avoid their worst ravages.

‘Cyclonology’, as it was beginning to be called, was to be the chief area of interest for the society, and the ambition to expand an observational network on land and sea was geared towards building the foundational laws upon which could be based new ways of understanding, sensing and perhaps predicting the storms which were so much a part of the consciousness of an island society organized around agriculture and maritime trade. The society aimed to develop the approach pursued by Thom in the 1840s – compiling, tabulating and analysing observations made aboard ships calling at Mauritius and, in the process, it was hoped, helping to rationalize and standardize the making of these observations. The aim, ultimately, was to publish compiled, simultaneous observations not only for Mauritius itself, but for the entire southern Indian Ocean. In fact, earlier ambitions on the island had been to construct a global picture, but the weight of the required work meant that the Indian Ocean was a more practical aim.Footnote ³⁰ It was a task similar in scope to Maury's efforts to chart the winds of the North Atlantic, but with the crucial difference that

Around this time effort was being made in London and elsewhere to formalize the practice of meteorology at sea. While some within the Admiralty had been keen on promoting maritime meteorology since the early decades of the nineteenth century, it took the intervention of the Royal Society before barometers were formally issued to Royal Navy ships in 1843, with the updated Admiralty Instructions of 1844 requiring captains to make four observations daily.Footnote ³² Turning merchant ships into reliable observatories was, of course, a rather more difficult task, although officers on East India Company ships were often keen weather documenters.Footnote ³³ The efforts to turn both merchant and naval ships into ‘itinerant observatories’Footnote ³⁴ nonetheless reflect the wider juxtaposition between the idealized regularity of geophysical observation and the real-life instability of its social, political and material superstructures, which has been well documented in the burgeoning history of the observatory sciences.Footnote ³⁵ This instability was a common feature of early colonial meteorology, where the sporadic observations of explorers and missionaries had yet to be supplemented by centralized networks under governmental control.Footnote ³⁶ Mid-nineteenth-century weather data may have ‘followed the paths of British ships and traced a geography of storms that conformed to the contours of Britain's imperial interests’,Footnote ³⁷ but the metropolitan organs of state could not necessarily be relied upon for moral or financial support to budding colonial meteorologists. It is this background of a science still struggling for authority and recognition in London and other imperial centres which makes the Mauritian case so unique: here was a meteorological society with ambitions – if not yet the means – to develop a centralized structure, with public service at its centre, which pre-dated the founding, in 1854, of the Meteorological Department of the Board of Trade in London by three years. Indeed, it was in the year of the latter's founding that the three posts of government observer, officer-in-charge of the Royal Engineers’ observatory, and secretary of the Meteorological Society were given to one man, Lieutenant Fyers of the Royal Engineers. But it was Charles Meldrum who nonetheless remained the driving force behind the society's most ambitious projects – the William Reid of the southern hemisphere. Although Fyers succeeded in winning the society free access to the observatory following supportive interventions from London, it was not until Meldrum's re-election as secretary in 1859 that the society's newest instruments, procured again from London and from the Bombay Geographical Society, were installed and put to work.

Constructing cyclones

The Meteorological Society's main advances were not to arise from the theoretically rarefied (but in practice rather troubled) spaces of the island's successive observatories. They were to arise instead from the even more unruly space of the ocean, criss-crossed by merchant vessels with their army of potential observers, some of whom embraced the role eagerly, albeit with varying levels of formality or regularity. From the outset the society had employed a clerk whose job it was to go aboard every ship entering Port Louis harbour, brandishing a letter from the secretary requesting permission to make a copy of the logbook and its weather observations. Meldrum would amass and study these copies, along with copies of logbooks deposited at the London Meteorological Department and subsequently despatched to Mauritius, tabulating into single tables the relatively synchronous observations of dozens of ships at different points in the southern Indian Ocean. This became very much Meldrum's project, with Bousquet seemingly unconvinced of the merit of studying the often unreliable observations of mariners.Footnote ³⁸ While Bousquet favoured the local knowledge of fishermen and engaged in the kind of qualitative recording of sensory perception which Lehmann has recently analysed in the archives of German colonial climatology,Footnote ³⁹ Meldrum pressed ahead with the project of enrolling and assembling the region's mariners into a dispersed, moving network of reliable meteorological observers. With Bousquet marginalized, the work followed Thom's model, while also according with the vision of Parisian meteorologist Urbain Le Verrier, who suggested that while the French concentrated on expanding their field of meteorological vision in the Atlantic, the British take care of the observations in the Indian and Pacific oceans, making use of the points and nodes of their imperial networks to create new centres of meteorological calculation.Footnote ⁴⁰ But while the Indo-Pacific might have been conceived as a ‘British’ meteorological region, the majority of the ships calling at Port Louis were voyagers of the Indian Ocean, particularly the England–India routes, and the society's more global ambitions were limited to their local ocean basin on account of the practical difficulties of reconstructing synoptic conditions from any farther afield.

The copying and the tabulating were taxing work in themselves; extracting meaning from it all was made all the more difficult by the lack of standard instruments and techniques used on board ships, the frequent contradictions between captains’ and masters’ logbooks, and the unwillingness of some mariners to open up their logbooks to any outside scrutiny at all. The latter were easily dismissed by the society, though – any captain whose character led him to so jealously guard his logbook would likely not fit the model of a well-disciplined and faithful weather observer anyway. The society nonetheless sought to make known the difficulties they were working with. In a concerted bid to attract more public and mariner support,Footnote ⁴¹ the society took to publishing much of the ship data in its rawest form, in order to illustrate to their interlocutors and potential allies just how difficult the work of standardization and tabulation could be. The aim was to ‘shew how difficult a task it is to trace out the general law of storms from data in which inconsistencies and inaccurate observations are mixed up with a few great truths’.Footnote ⁴² Although far from Latour's circulating immutable mobiles, with their flattening of the messiness and contingencies of scientific production into polished graphical forms, these messy tables served similar purposes: on the one hand claiming authority by illustrating the expanding spatial reach of the society across the ocean, and on the other illustrating the hard work and self-sacrifice of society members, particularly Meldrum, in working through the often chaotic raw data with which they were presented. It was an exercise in enrolling allies – allies who would offer this expanding network renewed support in return for promised serviceable truths, and allies who would more conscientiously uphold their side of the bargain through more careful observational efforts.

By 1859 Thom's vision of charts showing weather conditions ‘at the same time’ had been fulfilled, with the raw tables of data being translated into synoptic charts which, for the first time, gave a snapshot view of weather conditions across an area as wide as an ocean basin. This was new for meteorology, with previous charts or maps of a similar scale presenting only idealized models of the trade winds or general circulation, or covering smaller areas. Figure 1 shows the fruits of Meldrum's labours – the spiralling winds of two cyclones ‘raging furiously’, moving westward across the Indian Ocean towards Madagascar, each arrow a wind direction sensed simultaneously on board a numbered ship. Two cyclones are detected to the south of the Cape as well, while the Bay of Bengal and the South China Sea appear placid by comparison. This ability to compile and visualize synoptic data on such a scale aroused great interest in London and other meteorological metropoles, but the lack of printing facilities on Mauritius meant that in the immediate term, only pen-and-ink copies of the charts could be produced, and their circulation was initially limited.Footnote ⁴³ Yet the lessons of the charts could nonetheless be communicated locally to mariners through the society's meetings and its periodic publications. When Meldrum first presented his synoptic charts to the society he was not yet speaking of the holy grail of prediction, but rather of anticipation and planning based on well-ascertained data. Meldrum argued that his object was ‘not to advocate any theory or hypothesis, but to arrive at truth through just a careful examination of facts’.Footnote ⁴⁴

Figure 1. Two cyclones at noon of 13 February. From ‘Synoptic weather charts of the Indian Ocean for the month of February, 1861’, Meteorological Society of Mauritius. Information provided by the National Meteorological Library and Archive, Met Office, UK.

Meldrum devoted much of this empirical energy to examination of the question of whether winds blow in circles around a low-pressure system, or in incurving spirals. In the 1840s Henry Piddington, a former merchant captain and, from 1844, curator of the Museum of Economic Geology in Calcutta, published a range of treatises on the law of storms, building upon the ideas of Atlantic observers William C. Redfield, based in New York, and William Reid. Redfield, who had influentially noted the circularity of hurricane winds in the 1830s, had nonetheless supposed that the winds must be more spiralling than circular, in common with other aerial or aqueous vortices. Practical restraints had nonetheless kept him from publicizing these particular thoughts, not least because a circular storm diagram proved easier to print than a spiral one: ‘the storm figure on my Chart of the storm of 1830, was directed to be engraved in spiral or involute lines, but this point was yielded for the convenience of the engraver’.Footnote ⁴⁵ In Calcutta, Piddington was moved to take up the study of tropical storms when he encountered Reid's An Attempt to Develop the Law of Storms by Means of Facts (1838), and set about the ‘tedious and sometimes repulsive’ work of acquiring ships’ logs from ‘the unwilling, the dilatory or the diffident’ sea captains plying the Indian Ocean and the Bay of Bengal.Footnote ⁴⁶ Piddington produced an influential series of articles and a number of editions of his Sailor's Horn Book for the Law of Storms, which, like Thom's 1845 work, made the theories of Redfield, Reid and others legible to those tasked with making consequential navigational decisions on their basis.

These books were exercises in the making and communicating of serviceable truths. Piddington is perhaps best known for introducing the term ‘cyclone’ for revolving tropical storms, although this term was intended to tidy up the loose semantics of hurricanes, typhoons and gales, rather than to precisely denote a spiralling rather than a circular motion. Piddington nonetheless was among the first to give new credence to the spiral hypothesis, and translations of his work by Bousquet saw his ideas disseminated to a wide audience in Mauritius. Piddington placed particular emphasis on the log of the Charles Heddle, which in 1845 was caught in a cyclone near Mauritius, and ‘scudded round and round for five successive days!’Footnote ⁴⁷ The logbook showed the ship to have gradually spiralled towards the centre of the storm, in a ‘natural experiment’ which appeared to bear out the spiral hypothesis, along with other emerging evidence, such as the peppering of ships near the centres of storms with land birds, insects and butterflies, which were surely ‘carried inwards by the incurving tendency of the winds … If the wind blew in true circles they would be scattered all over the body of the cyclone’.Footnote ⁴⁸ Yet Piddington's advice to sailors on how to judge the patterns of the winds, and how to best navigate them, remained based on the assumption of circular winds, and his ‘horn cards’ – wind diagrams printed onto transparent sheets (made from animal horn) to be read on top of maritime charts – followed an earlier circular design by Reid.Footnote ⁴⁹

Meldrum's expanding field of synoptic oceanic vision meant he could make stronger claims than those of the natural experiments of atmospheric flotsam and spiralling ships. He drew on observations of wind and pressure from dozens of different vessels taken at approximately the same time in different locations, vessels which were, like the mythical blind men and the elephant, each experiencing a different local part of a wider whole. By piecing together these distributed testimonies, Meldrum began to confidently assert that the winds blowing in a cyclonic system do so according to a spiral pattern, and frequently in an elliptical rather than a circular shape. At the September 1860 meeting of the society, Meldrum offered his detailed analysis of two storms from the preceding cyclone season. He argued, ‘Both storms were, apparently, a conflict between the trade and monsoon; but in neither of them did the wind blow in a circle, unless close to the centre. It was, however, evidently moving around a central point, or had a strong tendency to do so.’ He went on,

Meldrum's new empirical support for the emerging spiral hypothesis would, in the course of time, give sailors new laws by which to work out the position and possible track of a storm, and to avoid its worst ravages. The community of Mauritius cyclonologists were not as sanguine as Piddington about the significance of circular versus spiral theories for the instructions to be given to sailors. An episode in February 1860 seemed to bear out their concerns: forty-one vessels left the roadsteads of Réunion in anticipation of a storm, armed with sailing directions based on the idea of circular winds. Of the forty-one, three were lost without trace, three were wrecked on the coast of Madagascar, and twenty-five suffered serious damage and financial loss. Instructions based upon a spiral and elliptical hypothesis,Footnote ⁵¹ it was assumed, would have saved lives and property.Footnote ⁵² Meldrum began to achieve international renown for his work. He was elected a fellow of the British Meteorological Society in 1868, and during the 1870s his renown spread further, as scientists and mariners in different corners of the world recognized the significance of his work, which was described in one Wellington newspaper as one of ‘the most important contributions recently made to science’, yet ‘so deeply practical’.Footnote ⁵³ He was elected a fellow of the Royal Society in 1876, with nominations from such scientific luminaries as Edward Sabine, James Glaisher and Francis Galton. In the same year he was awarded an LL D from the University of Aberdeen. ‘Meldrum's rules’ for cyclone navigation became touchstones for sailors plying the southern Indian Ocean, their utility championed by the likes of Ralph Abercromby,Footnote ⁵⁴ and thereby dislodging the inertia of the circular hypothesis which seemed to owe as much to contemporary printing practices as to the dynamics of theoretical change. In 1885 he submitted to the British Association for the Advancement of Science a series of yearly charts of all known cyclones in the southern Indian Ocean from 1848 to 1885, an atlas which would become both a standard navigational reference and a landmark in the development of synoptic meteorology.Footnote ⁵⁵

A new law of stormsFootnote ⁵⁶ did not only offer new means for sailors to navigate their way safely through the cyclone season. The accumulation at Mauritius of reams of data was also furnishing serviceable truths which could guide the sensing of a distant storm from an island observatory:

‘From a careful analysis of the observations taken at sea’, Meldrum argued, ‘viewed in connection with observations taken at Port Louis, our knowledge of the Meteorology of the Indian Ocean is such that we can now tell with certainty not only when bad weather prevails at a distance, but almost with equal certainty, where it prevails’.Footnote ⁵⁸

Alongside work on rainfall periodicity and links between climate, vegetation and health, Meldrum continued to refine his cyclonology, and began using his new-found laws to issue storm warnings to the press when a cyclone appeared to be in the vicinity. Following his appointment as government observer in 1861 he oversaw the long-desired move to a new observatory, the Royal Alfred Observatory (RAO), of which he became the director in 1874. From this new elevated, isolated location at Pamplemousses, away from the physical shelter and troubled history of the harbourside site, the meteorologist could see further, trusting the records of wind direction to be representative of a broader atmospheric field, while enjoying the order and serenity of a model observatory. It was at this site that the focus of Mauritian meteorology began to shift. The opening of the Suez Canal in 1869 and the coeval rise of steam-powered shipping meant that Mauritius was frequently bypassed by ships travelling between Europe and Asia. Although it had early been surmised that steamships, with their ability to maintain a steady course amidst strong winds, would offer new opportunities to observe ‘cross-sections’ of a storm (as opposed to the scudding and spiralling of ships under sail), this was not of much use if the information was not then deposited at Port Louis. The network of transient observers that Meldrum had encouraged began to slowly disperse, and the island's status as an archive and entrepôt of meteorological knowledge started to wane. By 1901, the year of Meldrum's death following an earlier return to Scotland, the efforts to produce daily synoptic charts had ceased, save for the occasional chart during the cyclone season to display storm tracks. The ship data, once collected with a view to ‘obtaining a correct representation of the atmospheric conditions over the Indian Ocean’, were ‘now well nigh useless’.Footnote ⁵⁹ With this we can say that Mauritian meteorology became more islandedFootnote ⁶⁰ – concerned less with the geophysics of an ocean basin, and more with the management of risk upon the island itself – a shift which was forced by changing geographies of oceanic trade, but one which was also enabled by the serviceable truths which had been built during Mauritius's time as a centre of meteorological calculation.

The arts of forecasting

Meldrum had furnished himself, and his colleagues and successors, with new practices for constructing and anticipating weather ‘at a distance’. Although the expansion of Mauritius's meteorological empire had proceeded through the establishment of telegraphic links to surrounding islands, this system provided little help in forecasting cyclones. In the British Isles, telegrams from the west coast meant that information about the weather could outrun the weather itself, and warnings could be issued about approaching storms.Footnote ⁶¹ In Mauritius, the techniques of sensing cyclones at a distance expanded the field of vision beyond that of the comparatively more limited telegraph system, which didn't connect the island to anywhere that could detect an approaching cyclone earlier than the RAO itself. The superior instruments at the RAO could detect a hurricane not much later than the observers at Rodrigues, who came into telegraph contact with the opening of the Durban–Perth segment of Britain's ‘All-Red’ line in 1901.Footnote ⁶² However, by closely observing the behaviour of the barometer and anemometer, and the clouds, and by comparing this behaviour with the laws being laid down by Meldrum and others, it seemed possible to determine the location and direction of a cyclone hundreds of miles away, and thus to determine the probability of its striking Mauritius. Prediction, rather than law building, thus became the chief means by which meteorology claimed a part in the public life of the colony.

The predictive skill of such techniques was severely tested by the devastating cyclone which struck the colony on 29 April 1892, leaving over 1,100 dead, two thousand wounded, one-third of Port Louis levelled and the colony's agro-industrial infrastructure in disarray (see Figure 2).Footnote ⁶³ In the preceding days a low barometer and veering wind had prompted Acting Governor Hubert Jerningham to enquire of the danger. With the barometer rising slightly on the 28th, no great risk was identified, although the harbourmaster was warned of a heavy sea. By the morning of the 29th conditions had become more ominous. The barometer was falling again, and the government's chief administrative officer was warned by telegram that ‘the depression was recurving and approaching the island’.Footnote ⁶⁴ The telegram never reached its destination, and soon communication across the whole island was disrupted as the storm, seemingly unanticipated by all but a nervous few at the observatory, on the harbour side and in the governor's mansion, started to bear down upon the island and wreak its havoc. But in spite of the carnage, the unusual timing and behaviour of the storm – appearing as it did several weeks after the usual end of the season and taking an unusual south-easterly track – brought the meteorologists a pardon from the governor. The disturbance, Jerningham later wrote, being ‘probably exceptional’, meant ‘it was excusable on the part of the meteorologists of the island to telegram on the evening of the 28th … [that] there was nothing to be apprehended’.Footnote ⁶⁵

Figure 2. ‘Port Louis after the hurricane of 1892’. Reproduced with permission from CO 1069/746, The National Archives, Kew

The 1892 storm, which lived long in the cultural memory of the colony, could have dealt a fatal blow to the authority of the island's meteorologists.Footnote ⁶⁶ As it was, though, a shared understanding of the ‘law of storms’ across the communities of knowledge makers and users, and crucially of the epistemological limits of such laws, meant that the imperfection of the meteorologist's craft was accepted and accounted for in the apportioning of blame. The laws, as serviceable truths, were taken as workable guides to anticipate and prepare for storms, their incertitude acknowledged and treated as an epistemological inevitability rather than a failing of the meteorologists themselves.

A few years later a young meteorologist named Albert Walter arrived in Mauritius following a short stint tabulating meteorological and astronomical observations at Greenwich Observatory.Footnote ⁶⁷ The twenty-year-old newcomer was assigned quarters at the observatory, in a small outbuilding in which the previous two European assistants had died of malaria, the disease having become endemic to the area during the construction of the observatory in the 1860s.Footnote ⁶⁸ The most urgent of his new tasks was familiarizing himself with the latest understandings of Indian Ocean meteorology, and particularly with Meldrum's work on the patterns and behaviour of cyclones. He spent weeks engrossed in Meldrum's charts, learning the laws which would enable him to practice the ‘art’, as he called it, of single-station forecasting.Footnote ⁶⁹ Good forecasts depended not only on good instruments, but on the particular expertise of the local forecaster. Forecasting cyclones was a combination of learned, relatively formal laws, and an ability to rapidly compute the relationship between those laws and the ever-changing responses of the meteorological instruments, together enabling the imaginative projection of a storm's location and track. The detection of a distant cyclone therefore meant battening down the hatches at the observatory, and the beginning of a long, lonely vigil by the side of the barometer.

On one evening in February 1902 Walter and Thomas Claxton, the director, left the observatory and headed home. Owing to the site's malarial tendencies, all the observatory staff had been excused to live up-country. But that night the barometer was falling, and reports had arrived from Rodrigues of a cyclone approaching there, albeit heading south, and thus likely to miss Mauritius by a wide margin. As in April 1892, an early telegram was issued predicting no danger for the island. However, by 10 p.m. the barometer was falling rapidly at Mauritius and Claxton was banging on Walter's front door, as it was clear the storm had changed track and was heading for the island. With no transport available, there began a six-hour epic of the two meteorologists trudging through the night, through heavy rain and gale-force winds, drenched to the skin and dodging falling trees, traversing the ten miles to Port Louis so they could take a train up to the observatory. When they finally arrived in the capital early the next morning no train driver would leave town for fear of being blown off a bridge, and by 9 a.m. all telegraphic communication was down.

Subsequent reports from ships suggested that the Rodigues cyclone had indeed continued south, but in doing so had masked a second approaching cyclone. The official report of the RAO argued that forecasting this second, disguised cyclone would have been impossible, but Walter, not lacking in confidence in his new-found forecasting abilities, disagreed, and argued that a proper study of the instruments of the observatory could have disentangled the two signals.Footnote ⁷⁰ This incident saw the governor enact a law that either the director or his assistant must be in attendance at the observatory at all times during the cyclone season.Footnote ⁷¹ For Walter, now assistant director, these new rules of responsible meteorological conduct meant that most of the next few Christmases were spent in watchful isolation at Pamplemousses. During the very next cyclone season, however, Claxton was thrown out of his cariole while travelling – in some dereliction of his duty, Walter would suggest – from the observatory to Port Louis. He was perhaps only saved from the worst effects of a serious head wound by the fact that the accident occurred outside the house of a prominent doctor on the island, who took charge of his treatment. Laid up for several months in the capital, Claxton was forced to pass on responsibility to Walter for a cyclone season which saw four storms pass close to the island. Walter later recalled,

Throughout the second half of the nineteenth century, meteorologists had been divided on the origins and functioning of cyclonic storms, with consequences for how the possibility of prediction was conceived. Thermal or convective theories, which saw the release of latent heat through the condensation of water vapour as the chief source of kinetic energy in cyclones, slowly came to marginalize earlier, dynamic or mechanical theories which saw the existence of conflicting currents in the wind field as the chief source of motion. James Pollard Espy's early convective theory of the 1830s required the thermodynamic advances of the 1850s and 1860s before it could be enumerated, and later studies of surface temperatures and cloud and precipitation patterns appeared to support the notion of convection being the key source of kinetic energy. Convective theory, rooted as it was in the key theoretical concerns of mid-nineteenth-century physics, would go on to secure reasonably widespread acceptance in meteorological communities in North America and Europe, following Espy's sparring on the matter with Redfield.Footnote ⁷³ However, disagreements remained, and were heightened as observations began to be made of the upper atmosphere, with temperature discrepancies above cyclonic systems and competing views of the independence of cyclones from the general circulation posing challenges to a unified convective theory.Footnote ⁷⁴ Mechanical theories thus continued to propagate. Hippolyte Marié-Davy, head of the Paris Observatory's troubled storm-warning service in the 1860s and later director of the Montsouris Observatory,Footnote ⁷⁵ posited cyclonic storms, whether of tropical or extra-tropical varieties, as large rotating aerial disks whose centrifugal forces caused the outflow of air, the diminution of pressure in the centre, and thus the inflow of air at lower and higher altitudes. Similar ideas were put forward by the Swiss-born Marc Dechevrens, director of the Jesuit observatory at Zikawei in modern-day Shanghai from 1877 to 1887, and later the founder of an observatory at the Maison Saint Louis on the island of Jersey. Dechevrens found support for his ‘whirl-generator’ hypothesis through empirical studies at the Puy de Dôme and the Pic du Midi, posing direct challenges to convective theories.Footnote ⁷⁶ The respected astronomer Hervé Faye, a tracer of comets and theorist of solar radiation, also published profusely on cyclone theory, positing a model whereby cyclones formed in the upper strata of the atmosphere, with kinetic energy ‘propagated downward by descending motion accompanied by gyration around a vertical axis’.Footnote ⁷⁷

The conflict between largely francophone mechanical theories and the convective theories which found greater favour among anglophone meteorologists was played out, appropriately, in the public and bilingual meteorological culture of Mauritius. In February 1902 the French-language daily newspaper Le cernéen, which regularly carried criticisms of the RAO meteorologists, published a piece which put Faye's theories, recently compiled in his Nouvelle étude sur les tempêtes, cyclones, trombes ou tornados,Footnote ⁷⁸ alongside the ideas of William Ferrel, who at that moment was perhaps the paradigmatic convective theorist. Albert Walter, writing in the Proceedings of the Meteorological Society, found the symmetrical treatment of these bodies of work deeply unwarranted.Footnote ⁷⁹ Walter and Claxton were both committed to convective theories, and Walter railed, with uncharacteristic directness and large doses of sarcasm, against Faye's apparently circular reasoning about the role of upper air currents in directing storms, and thus of the possibility of identifying the currents by tracking the storms themselves. Walter drew attention too to Faye's contradiction of the spiral hypothesis with his insistence that cyclonic winds are regulièrement circulaire. While Meldrum's apparent confirmation of the spiral hypothesis had become a serviceable truth for mariners and forecasters alike, meteorological theorists were still divided, with Napier Shaw – director of the Meteorological Office in London from 1905 to 1920 – recollecting in 1926 that ‘even within our own official experience we have received protests against Meldrum's representation of incurved spiral paths as being erroneous’.Footnote ⁸⁰ But Walter saved most of his ire for a passage in the Nouvelle étude which compared terrestrial atmospheric cyclones to the cyclonic behaviour of sunspots, particularly in how they apparently functioned as descending gyratory currents. In some ways this was Faye claiming the terrestrial atmosphere for the theoretical and analogical reasoning of astronomers; students of the heavenly bodies, with their superior grasp of physical theory, could lower their eyes to Earth, ready to interpret the more mundane phenomena of fluid motion which animated the atmosphere. But Walter, while admitting great empirical deficiencies in collective understandings of the tropical atmosphere, could nonetheless point to a growing armoury of local observational data which appeared to support a modified version of dominant convective theories. Cloud observations at the RAO contradicted Faye's arguments about gyratory action originating in the cirrus stratum, while the amassed studies of cyclone tracks seemed to contradict emerging ideas about the continual circulation of new masses of air through cyclonic storms, indicating their imbrication with the general circulation.Footnote ⁸¹ Walter instead sided more with Ferrel, arguing that southern-hemisphere tropical cyclones were quite independent of any anticyclonic action and of the general circulation writ large – ‘our cyclones are quite distinct in themselves and quite independent of any anticyclonic action’.Footnote ⁸² This theoretical stance – cyclonic storms with ‘definite boundaries’Footnote ⁸³ – furnished the Mauritius meteorologists with discrete entities which could be tracked, understood and perhaps predicted without the observational apparatus which would have furnished a picture of the wider synoptic situation. Viewed from Mauritius at the turn of the century, a tropical cyclone was a distinct object, an island in itself, roaming around an otherwise blank and, crucially, now largely unobserved oceanic space according to empirically derived rules whose regularity permitted a novel form of prediction.

The resulting predictive efforts were, however, not without their deficiencies, and were built around relationships of trust and authority which came under frequent strain. The seemingly troubled relationship between Claxton and Walter (‘Claxton was not too fond of me’, Walter recalledFootnote ⁸⁴) continued to fracture during subsequent cyclone seasons. Walter had been taking on extra work compiling and analysing sugar cane statistics, while Claxton ‘did all he could to make life difficult by insisting on the first and last train each day for attendance at the Observatory’.Footnote ⁸⁵ On 10 January 1910 Claxton and Walter had disagreed over the significance of a falling barometer, and Claxton, less concerned about the prognosis than his assistant, headed home, leaving Walter to observe an incoming storm without the directorial authority to issue a warning and keep ships in the harbour. A worried telephone call from the harbourmaster informing Walter that the Loodiana was setting sail despite worsening conditions led to frantic attempts to reach Claxton at home. But with no official word forthcoming from the director, the harbourmaster was powerless to prevent the departure. The Loodiana was lost in the storm, along with 176 passengers and crew.Footnote ⁸⁶ Not long after the official inquest, the directorship of the Hong Kong Observatory fell open. The Colonial Office wanted Walter to apply, but the colonial secretary on Mauritius seemingly stopped the proposal reaching Walter, and put Claxton forward instead. Claxton left Mauritius in May 1911, with Walter taking over as director of the Royal Alfred.Footnote ⁸⁷

By this point Walter had established himself as a prominent figure in island society. In 1910 he published The Sugar Industry of Mauritius: A Study in Correlation, the product of a number of years’ work examining meteorological and agricultural records, the latter gleaned from the island's largely French planter establishment. Walter had married into a prominent French landowning family, easing his making of new links among the island's francophone elite. His book made a number of claims about how cane yields were affected by climatic conditions, but his most ambitious arguments concerned the effects of cyclones. He suggested that observations of cyclone intensity at opposite ends of the island allowed him to infer, through techniques of statistical interpolation, the intensity of the cyclone at all points in between. ‘Isoptherms’, lines of equal devastation, could be drawn across the map of the island, allowing the presumed effects of a transiting storm to be down-scaled to the level of an individual plantation. Walter proposed his statistical machinations as the basis for a cyclone insurance scheme, whereby the true value of lost crops could be inferred not from the unreliable and often tardy testimony of insurance field inspectors, but from the statistical manipulation of meteorological data.

Walter was working on his statistics during a period of acute capital shortage on Mauritius. The plantation economy had frequently struggled through periods of low sugar prices and insufficient credit. Between 1892 and 1909, steadily falling prices, along with a damaging series of droughts, cyclones and disease outbreaks, hurt the prospects for outside capital investment,Footnote ⁸⁸ and Walter's scheme aimed to smooth the inter-annual cycle of capital circulation, which saw capital lying dormant between December and April, while profits were frequently consumed by the repair of storm-damaged buildings and equipment.Footnote ⁸⁹ Walter's proposals received the support of the planter and financial communities, but it was not until after the Second World War that a cyclone insurance scheme based on his calculations was instituted.Footnote ⁹⁰ By that time Walter had long since moved on to Nairobi, where he led the establishment of the British East African Meteorological Service. Working across huge swathes of territory, he regretted the loss of his island laboratory where total vision and freedom from ‘the effect of extraneous outside influences’ had allowed him to trace, across a wide range of phenomena, ‘the relation of cause and effect’ in seemingly precise terms.Footnote ⁹¹ Despite ongoing efforts to develop new infrastructures for field-based agricultural meteorology, it was not until his involvement in the post-war development of Britain's ‘groundnut scheme’ in Tanganyika, with the possibility of intensive monitoring of climate and crops, that Walter could at last regain something of the laboratory vision he felt himself to have attained in Mauritius.

The cyclonic sublime

Making knowledge about cyclones in colonial Mauritius was always about finding imaginative ways of filling in the gaps between sparse, often unreliable instrumental observations, whether that be the retrospective construction of cyclone tracks and behaviour from loosely coordinated ship observations, or Walter's inferences about the damage wrought by cyclones between the points where the cyclone, and its effects, could be directly observed. The practice of single-station forecasting pushed the challenges of ‘remote sensing’ even further, and in a curious passage in his memoirs Walter related the emotional, imaginative work which lay behind this palimpsest of observation, projection and correction. In the preface to a passage entitled ‘The Cyclone … A meteorological fantasy … or an objective reality?’ he appealed to future keepers of his memoirs to either retain this section within the memoirs of his ‘official life’ or relegate it to the more informal autobiography he planned to write for his family. Their decision was to be based on whether Walter's animist reading of the character of cyclones was in the future deemed to be nearer fantasy or reality. When compiling his notes in the 1960s, Walter appeared to be leaning towards the latter, noting the new trend for naming individual storms as affirmation that others shared his appreciation for the individual character and essence of discrete storms.Footnote ⁹²

In the passage, made up of diary entries and text written during his time in Mauritius, Walter's otherwise businesslike tone erupts into a riot of metaphor and allusion as he describes the process of communing with this ‘deadly creature of Nature’. From the moment a disturbance is first registered by the barometer, he writes, ‘I felt that there was a special genie of the storm with which my own consciousness seemed to be in contact’. The first feeling was of ‘uneasy anticipation accompanied by the perception of peculiar moaning in the gentle gusts of the wind which to me was always the prelude to a storm’. It was this perception which sent Walter to his post at the observatory to begin his lonely vigil with barometer and anemometer. From there he would observe the development of high cirrus clouds ‘streaming out from the centre of the storm’ like ‘a host of fairy chariots mounted by the gremlins of the storm’. The roar of the sea would start to sound an ‘ominous accompaniment … to the great atmospheric orchestra’, and as he ‘followed each slight change in the wind and pressure, summing up almost unconsciously, the probabilities of their significance, there always seemed to be a directing consciousness of control saying: “not this way but that”’.

These themes of consciousness and control recur throughout the passage. He describes the storm as a ‘living organism’ being created and ‘built up by some active intelligence of which I myself became consciously a part’. As the storm neared, he ‘never felt terrified’ but, he argues, ‘I … felt that I was part of it and in control’. Once the centre of the storm had passed over the island and the observatory, he writes, ‘there grew a sense of achievement, of finality and satisfaction, as the muted sounds of the storm played their andante movement’. As the eye moved off and the drama increased again there was nothing more to be done by way of forecasting, and so Walter could revel in the extreme aesthetics of thunder and lightning, wind and rain.Footnote ⁹³

While the passage might be dismissed as irrelevant to the real business of cyclone prediction, the terminology Walter uses – ‘glamour’, ‘romance’, ‘poetry’, the force of ‘imagination’ – speaks in important ways both to our understanding of scientific observation, and to the moral economy of forecasting. Lorraine Daston has written of how practices of observation are shaped by pedagogy, experience and habit into the ‘all-at-once-ness’ of scientific perception, what field naturalists refer to as ‘jizz’ – the ‘sure, swift, silent’ movement from flashes of perception to reliable knowledge, ‘without pause for mental analysis’.Footnote ⁹⁴ We might wonder whether Walter's cyclonic metaphors – fairy chariots ridden by gremlins, a cascading orchestra of multiple sections – were not just means of retrospectively colouring his story, but played a more active role in producing pattern and order in the sensory experience of a storm. We can understand this method of sensing cyclones, which is not unlike the sensory engagements of Henri Bousquet half a century earlier, as a process of gestalt perception – a sensory perception of synthesis, through the recognition of form, pattern and the coalescence of key attributes of a storm which is either heading out to sea or coming this way.

Turn-of-the-century cyclone forecasting in Mauritius was quite clearly reliant on tacit and ‘personal knowledge’ in Polanyi's sense,Footnote ⁹⁵ as illustrated by the economy of responsibility which had grown around it. In some sense, then, this was little different to the broader nineteenth-century politics of weather forecasting and storm warning, where public debates about the authority of meteorology were widespread, at least in Western Europe and North America, along with the sometimes tragic politics of forecasters’ personal reliability and responsibility.Footnote ⁹⁶ But what is unique in Mauritius is the combination of a well-developed culture of public meteorological debate and an almost complete lack of the infrastructure for seeing further, such as storm forecasters in Great Britain, the US and the West Indies by that time had, for example. Cables from surrounding islands were not much help, as by the time a storm had been sensed there it would be too late to start the process of observation, prediction and warning on Mauritius. The consequent reliance on what we might call an individual's cyclonic jizz, and the social pressure brought to bear on the forecaster, are reflected in Walter's rendering of the sudden freedom of responsibility when the eye of the storm arrives. With the storm squarely overhead, no more forecasting is required, and despite the destruction undoubtedly being caused to the neighbourhood, the gestalt perception of the storm can switch from anxious anticipation to aesthetic pleasure in something which, in Walter's rendering, is very much akin to the sublime.Footnote ⁹⁷