The Ghost Map

by Steven Johnson

  And so the miasma theory did not crumble immediately after the Broad Street outbreak, though its days were numbered. Eventually, Snow’s and Whitehead’s parallel investigations would be seen as the turning point in the battle against cholera. But it would take yet another outbreak—more than a decade later—for that narrative to take hold for good.

  It is not known if Sarah Lewis ever learned that the final days she spent tending to her daughter had triggered the most devastating outbreak in the history of London. If so, the weight of the news must have been unbearable, because the outbreak she had unwittingly set in motion eventually killed her husband as well. Thomas Lewis had fallen ill that Friday, September 8, within hours of the pump handle’s removal. He fought the disease much longer than most, surviving for eleven days. The young policeman finally succumbed on the nineteenth of September, leaving a childless widow alone in a ruined neighborhood. The outbreak had begun at 40 Broad Street, and it ended there as well.

  The timing of Thomas Lewis’ illness suggests one chilling alternative history. The Broad Street outbreak had subsided in part because the only viable route between the well and the neighborhood’s small intestines had run through the cesspool at 40 Broad. When baby Lewis died, the connection had died with it. But when her husband fell ill, Sarah Lewis began emptying the buckets of soiled water in the cesspool all over again. If Snow had not persuaded the Board of Governors to remove the handle when he did, the disease might have torn through the neighborhood all over again, the well water restocked with a fresh supply of V. cholerae. And so Snow’s intervention did not just help bring the outbreak to a close. It also prevented a second attack.

  Conclusion

  THE GHOST MAP

  IN THE FIRST FEW DAYS AFTER THE PUMP HANDLE’S REMOVAL, an engineer by the name of Edmund Cooper began examining the Broad Street epidemic on behalf of the Metropolitan Commission of Sewers. Rumors that sewer excavations had unearthed the decaying but still pestilent corpses from the plague burial grounds had been buzzing through the neighborhood. Even the newspapers had implicated the old pesthouse fields. (The Daily News had published a letter on September 7 accusing the sewer builders of unearthing an “immense quantity of human bones” during their excavations in the area.) With these scandalous accusations floating about, the Commission dispatched Cooper to investigate the claim. Cooper quickly arrived at the conclusion that the bodies of two-hundred-year-old plague victims posed little threat to the neighborhood, whether they had been disturbed by sewer construction or not. It was clear from the Weekly Returns—and from Cooper’s on-scene investigating—that the sewer construction had not likely played a role, given the geographic dispersal of deaths. But Cooper needed a way to represent these patterns in an intelligible manner that both the laypeople of the neighborhood and his supervisors might understand. So he created a map of the outbreak. He modified an existing plan of the neighborhood that showed the new sewer lines, adding visual codes to indicate both the location of cholera deaths and the site of the original plague pit. For each house that had suffered a loss, Cooper drew a black bar by the address, followed by a succession of thin lines indicating how many deaths had occurred at that address. In the northwest corner of the map, roughly centered over Little Marlborough Street, Cooper drew a circle inscribed with the words “Supposed Location of Plague Pit.” A quick glance at the map made it clear that the outbreak had been triggered by some other source: the deaths were concentrated several blocks to the southeast of the ancient burial site. Only a handful of deaths had occurred within Cooper’s circle, and the houses to the immediate south and east of the circle had been spared entirely. If some noxious effluvium had risen out of the plague pit, surely the residents living directly on top of the pit would have suffered the worst casualty rate.

  Cooper’s original layout would be copied and expanded in another map produced for the Board of Health investigation that included data from the more extensive survey that had been carried out that fall. Once again, the map exonerated the plague pit, though the committee ultimately included the sewer lines as a potential source of miasmatic poisoning in the area. Both maps were well-crafted specimens of the new art of dot mapping—that is, representing the spatial path of an epidemic by marking each case with dots (or bars) on a map. They were both attempts to tell the story of the Broad Street outbreak from the bird’s-eye view, to see the patterns of the disease as it erupted through the neighborhood. They were both superbly detailed: old and new sewer lines were documented with distinct markings; each gulley hole was represented by an icon on the map, along with ventilators and side entrances and the street number of every house in the parish. Even the neighborhood pumps were included. But as exacting as Cooper’s map was, it ultimately had too much detail to make sense of the story. The connection between the Broad Street pump and the surrounding deaths was lost under the sheer mass of data that Cooper had charted. For a map to explain the true cause behind the Broad Street outbreak, it needed to show less, not more.

  JOHN SNOW BEGAN WORKING ON HIS FIRST MAP OF THE Broad Street outbreak sometime in the early fall of 1854. Its initial form, which he shared publicly at a meeting of the Epidemiological Society in December, resembled Cooper’s survey, with two small modifications: Each death was represented by a thick black bar, which made the houses that had suffered significant deaths more vivid on the map. And the detail on the map was reduced, with everything but the basic street layout and the icons representing the thirteen public water pumps that served the greater Soho area eliminated. The visual impact of the map was striking. Because it represented a larger section of London—from Hanover Square in the west to Soho Square in the east, and all the way south to Piccadilly Circus—eleven of the pumps were shown to be entirely clear of local cholera cases. The Little Marlborough Street pump had a few black bars in its immediate vicinity, but they were nothing compared with the concentration of death around the Broad Street pump, black bars lining the nearby streets like solemn high-rises. Without a prominent icon for the Broad Street pump, the other dot maps of the epidemic had presented an amorphous shape, like a cloud hanging over western Soho. But when you emphasized the pumps in the image, the map took on a new clarity. Cholera wasn’t lingering over the neighborhood in a diffuse form. It was radiating out from a single point.

  In effect, Snow had given the death and darkness of the Broad Street outbreak a new kind of clarity. His first map has been rightly celebrated for its persuasiveness, and variations of it have been reproduced in countless textbooks on cartography, information design, and public health. A landmark 1911 textbook on epidemiology, Sedgwick’s Principles of Sanitary Science and Public Health, included a dozen pages on the Broad Street case and featured a revised version of the map prominently. Thanks to that continued attention, the map has become the defining symbol of the entire Broad Street outbreak. But its significance has been somewhat misunderstood. The black bars marking the ghosts of Soho were a striking visual element, but they were hardly Snow’s invention. Not only had dot maps been created to visualize previous cholera outbreaks, but at least one (Cooper’s) had already been created to document the Broad Street outbreak itself before Snow began work on his map. Part of what made Snow’s map groundbreaking was the fact that it wedded state-of-the-art information design to a scientifically valid theory of cholera’s transmission. It was not the mapmaking technique that mattered; it was the underlying science that the map revealed.

  Snow modified his original map for publication in two places—the Vestry Committee’s report and the second edition of his own monograph on cholera. Augmented by the new data on the outbreak that Whitehead and others had assembled, the second version of the map contained Snow’s most significant contribution to the field of disease mapping. (Ironically, it goes unmentioned in Edward Tufte’s extensive account of Snow’s mapmaking in Visual Explanations, which almost single-handedly elevated Snow’s work to the information-design canon.) After presenting to the Epidemiological Society, Snow had reali
zed that his original map was still vulnerable to a miasmatic interpretation. Perhaps the concentration of deaths around the Broad Street pump was merely evidence that the pump was releasing noxious fumes into the air. And so Snow realized he needed a way to represent graphically the foot-traffic activity around the pump that he had so painstakingly reconstructed. He needed to show lives, not just deaths; he needed to show the way the neighborhood was actually traversed by its residents.

  To solve this problem, Snow drew upon a centuries-old mathematical tool that would later be termed the Voronoi diagram. (It is unlikely that Snow knew anything of the device’s history, though he was certainly the first to apply it to disease mapping.) A Voronoi diagram conventionally takes the shape of a two-dimensional field made up of points surrounded by “cells.” The cells define the region around each point that is closer to that particular point than any other point in the diagram. Imagine a football field with a point on each goal line. The Voronoi diagram of that field would be divided into two cells, the demarcation between them being at the fifty-yard line. If you stand anywhere on the field on the home-team side of the fifty-yard line, you are closer to the point on the home team’s goal line than you are to the point on the other goal line. Most Voronoi diagrams, of course, involve many points scattered about in unexpected ways, resulting in a honeycomb pattern of cells surrounding their local points.

  What Snow set out to do with his second map was to create a Voronoi diagram using the thirteen pumps as points. He would diagram a cell that showed the exact subsection of addresses on the map that were closer to the Broad Street pump than they were to any other pump. But these distances would have to be calculated on foot-traffic terms, not the abstract distances of Euclidean geometry. The cell was warped by the erratic arrangement of streets in Soho. Some addresses were closer to Broad Street as the crow flies, but if you actually paced the routes out by foot, winding your way through the crooked alleys and side streets of Soho, another pump turned out to be closer. It was, as the historian Tom Koch astutely notes, a map organized as much around time as around space: instead of measuring the exact distance between two points, it measured how long it took to walk from one point to another.

  And so the second version of the map—the one that made it into both Snow’s monograph and the Vestry report—included a slightly odd, wandering line that circumscribed the center of the outbreak, roughly in the shape of a square with five or six areas jutting out, like small peninsulas, into the surrounding neighborhood. This was the area encompassing all those residents for whom the quickest trip for water was to the Broad Street pump. Superimposed over the black bars that marked each death, the amorphous shape took on sudden clarity: each peninsula extended out to embrace another distinct cluster of deaths. Beyond the circumference of the cell, the black bars quickly disappeared. Snow’s visual case for his waterborne theory revolved around a striking correspondence between two shapes: the shape of the outbreak area itself, and the shape of best proximity to the Broad Street pump. If the cholera were somehow spreading as a miasmatic emission from the pump, the shape of the neighborhood deaths would have looked quite different: not a perfect circle, perhaps, since some houses would have been more vulnerable than others. But it certainly wouldn’t have followed so precisely the contours of street-level (i.e., foot-traffic) proximity to the Broad Street well. The miasma wouldn’t be influenced by the eccentricities of street layouts, after all, and it certainly wouldn’t be influenced by the location of other neighborhood pumps.

  And so the ghosts of the Broad Street outbreak were reassembled for one final portrait, reincarnated as black bars lining the streets of their devastated neighborhood. In dying, they had collectively made a pattern that itself pointed to a fundamental truth, though it took a trained hand to make that pattern visible. And yet, however elegant its design, the map’s immediate influence was far less dramatic than folklore has it. The map didn’t solve the mystery of the outbreak. It didn’t lead to the pump handle’s removal and thus bring an end to the epidemic. In fact, it failed to sell the Board of Health on the merits of the waterborne theory. Yet despite those reservations, Snow’s map deserves its iconic status. The case for the map’s importance rests on two primary branches: its originality and its influence.

  The originality of the map did not revolve around the decision to map an epidemic, or even the decision to encode deaths in bars etched across the street diagram. If there was a formal innovation, it was that wobbly circumference that framed the outbreak in the second version, the Voronoi diagram. But the real innovation lay in the data that generated that diagram, and in the investigation that compiled the data in the first place. Snow’s Broad Street map was a bird’s-eye view, but it was drawn from true street-level knowledge. The data that it sketched out in graphic form was a direct reflection of the ordinary lives of the ordinary people who made up the neighborhood. Any engineer could have crafted a dot map from William Farr’s Weekly Returns. But the Snow map drew on a deeper, more intimate, source: two Soho residents talking to their neighbors, walking the streets together, sharing information about their daily routines, and tracking down the long-departed émigrés. Neighborhood demographics had been projected onto maps before, of course, but invariably the projections involved the official interventions of the census takers or the Board of Health. Snow’s map—with Whitehead’s local knowledge animating it—was something else entirely: a neighborhood representing itself, turning its own patterns into a deeper truth by plotting them on a map. The map is a brilliant work of information design and epidemiology, no doubt. But it is also an emblem of a certain kind of community—the densely intertwined lives of a metropolitan neighborhood—an emblem that, paradoxically, was made possible by a savage attack on that community.

  As for influence, it’s pretty to think of John Snow unveiling the map before the Epidemiological Society to amazed and thunderous applause, and to glowing reviews in The Lancet the next week. But that’s not how it happened. Its persuasiveness seems obvious to us now, living as we do outside the constraints of the miasma paradigm. But when it first began circulating in late 1854 and early 1855, its impact was far from dramatic. Snow himself seems to have thought that his South London Water Works study would ultimately be the centerpiece of his argument, the Broad Street map merely a piece of supporting evidence, a sideshow.

  The tide of scientific opinion would eventually turn in Snow’s favor, and when it turned, the Broad Street map grew in stature. Most accounts of the outbreak reproduced the map in some fashion—so often, in fact, that copies of copies began appearing in textbooks, described erroneously as original reproductions. (Most of them lacked the critical Voronoi diagrams.) As the waterborne theory of cholera became increasingly accepted, the map was regularly invoked as a shorthand explanation of the science behind the theory. It was easier to point to those black bars emanating ominously from the pump than it was to explain the whole idea of microorganisms invisible to the human eye. The map may not have had the impact on its immediate audience that Snow would have liked, but something about it reverberated in the culture. Like the cholera itself, it had a certain quality that made people inclined to reproduce it, and through that reproduction, the map spread the waterborne theory more broadly. In the long run, the map was a triumph of marketing as much as empirical science. It helped a good idea find a wide audience.

  SNOW’S MAP MAY HAVE HAD A CRUCIAL SHORT-TERM IMPACT as well, though this is closer to an inference than an empirical fact. We know that Henry Whitehead’s interest in the waterborne theory turned decisively after Snow gave him a copy of his revised cholera monograph in the late winter of 1855. That monograph contained the second edition of Snow’s map. It’s entirely possible that seeing all those deaths radiating out from the Broad Street pump played a role in changing the curate’s mind. He had spent more time than anyone working through the intimate details of those lives and deaths—first attending the sick as a clergyman, then investigating the outbreak as an amateur de
tective. It must have been a revelation to see all that data rendered from above for the first time.

  Persuading an assistant curate of the merits of the waterborne theory might seem like a minor accomplishment. But Whitehead’s investigations in 1855 were ultimately as decisive as Snow’s in solving the Broad Street mystery. His “conversion experience” reading Snow’s monograph set him off in search of the index case, eventually leading him to baby Lewis. The discovery of baby Lewis led to York’s excavation of the pump, which confirmed a direct connection between the pump and the cesspool at 40 Broad.

  It’s conjecture, of course, but it’s nonetheless entirely reasonable to assume that without the Reverend Whitehead’s contributions the Vestry Committee would have never blamed the outbreak on the Broad Street pump. Without an index case and an unequivocal link to the well water, without the support of one of the neighborhood’s most beloved characters, it would have been so much easier for the Vestry Committee to equivocate, to blame the outbreak on the neighborhood’s generally pitiful sanitary standards—in the streets and in the houses, in the water and in the air. It would have been so much easier for the Vestry Committee to fall back on the miasmatic haze of the Board of Health report. But the final compilation of evidence had been too overwhelming for such stock explanations. When you combined Snow’s original data with Whitehead’s more exhaustive investigation, when you factored in the index case and the decaying brickwork, the conclusion was inevitable: the pump was the source of the outbreak.