Reportage

Cyclones, fake news and history

Science and searching the archive

IN THE DARK before dawn on 5 March 1899, half way up the eastern edge of Cape York Peninsula, five men camping on a sand ridge about forty feet above sea level and half a mile behind the beach found themselves waist deep in the ocean. Since midnight, the four Aboriginal troopers and a white officer of the Native Police had been huddled under a blanket as Australia’s deadliest cyclone, and one of the world’s fiercest, blew away their tents and killed or scattered their horses. Just after 5 am, as the eye passed north over Cape Melville and wrecked the pearling fleet anchored there, the ocean swept inland over the ridge, spoiling the officer’s watch.

The next day, the officer, Constable Jack Kenny, and his men began to walk back to Cooktown, where they arrived a week later to learn that the cyclone had drowned nearly 300 people. Kenny told his story to the manager of the Burns Philp store, Arthur Vidgen. By then, the disaster was international news and Vidgen wrote Kenny’s account in a letter to a friend in Brisbane, who showed the letter to a newspaper, which published an extract. Another newspaper used that as the basis for an article, and six months later that article was cut and pasted into an anonymous booklet. The rest, as they say, is history. Or, if you’re a historian, memory. Or for some scientists, data, because Kenny’s account has made its way from the anonymous booklet into the scientific literature and is now supporting evidence that the 1899 cyclone caused a world-record storm surge. But is it true?

One hundred and twenty years later, I’m standing on that same ridge south of Cape Melville. It’s taken me and Dan Gordon, a traditional owner of this country, a dusty day and a half by four-wheel drive from Cairns to reach the spot. We drove as far as we could and after stumbling through thick brittle scrub, Kenny’s ridge opens before us. It’s the end of the dry season and the heat rises from the fine grey sand. It’s quiet. In the wet season, the time of year when Kenny camped here, the scrub would be thicker, the nearby waterhole brimming, this oval of ridge-top cooled by a sea breeze crossing coastal swamps, and the wongai trees heavy with purple plums.

The crumpled satellite photo I’ve been carrying shows that this is a good campsite. It’s near fresh water and beside an ancient Aboriginal road of the Barrow Point people, a track patrolled in the late 1800s by the Native Police. I wasn’t really expecting to find any remaining physical evidence of Kenny’s camp or the wave that spoilt his watch, so I was confused at first by the seashells covering the entire flat top of the sand ridge, an area roughly the size of two football fields. Dan pointed to a large round stone. ‘People lived here, eh.’ Kenny’s story isn’t the only one written into this ridge. Science, history and culture meet here in a way I never imagined from satellite photos and digital archives. But without those tools, I’d never have found this place.

For the past five years, I’ve been working on a thesis at the University of Queensland, looking at the disaster caused by the 1899 cyclone that killed Kenny’s horses and destroyed Queensland’s pearling fleets. Part of the reason Australia’s deadliest cyclone isn’t well known today is that 96 per cent of the nearly 300 people who died weren’t Europeans; they were the ‘coloured’ crews of pearling luggers from more than twenty countries. And it occurred just before Federation, when white Australia was planning to send all those foreigners back to where they came from anyway. But that’s another story. The main reason I’m standing on this remote ridge is to solve a problem. Why is what we’re told about this disaster in history books and scientific journals so different to what happened here 120 years ago?

 

DISASTER SCIENCE IS booming. Journal publisher Elsevier, in its A Global Outlook on Disaster Science report in 2017, noted that 27,000 papers on disaster science had been published between 2012 and 2016, many of them about prevention and preparedness, stressing the need for good data to inform policy-making. Disasters are becoming more costly. Two years ago, one cyclone alone, Debbie, cost the insurance industry $1.67 billion, so there’s a big financial incentive to reduce the risks, to understand better how cyclones cause damage and how a changing climate might make things worse. In this area of science, scientists look backwards to look forwards and rely heavily on history to do so.

Two decades ago, as the digitisation of historical data was taking off and the science of disaster prediction flourishing, Australian historian Paula Hamilton noted in the History Workshop Journal that disasters weren’t ‘usually regarded as the substance of scholarly work by historians’. That’s still the case. Relatively few historians are in the field, leaving scientists to mine the archives for historical data to learn those vital lessons from past earthquakes, fires, floods and cyclones. Geoscience Australia, in its 2007 report Natural Hazards in Australia: Identifying Risk Analysis Requirements, recommends including data from old newspapers to assess future risks: ‘For extreme events, newspaper records extending beyond the timescale of the Bureau of Meteorology’s database, which dates back to 1909, will help in the assessment,’ it suggests.

But how good are data-mining scientists at navigating the pitfalls of history? How do scientists test the information they find in old newspapers, or any historical document, to gauge its validity? It’s hard to generalise because there aren’t many published methodologies in the scientific literature. In one Canadian study that mined digital newspaper archives for information about a flood, scientists decided not to test the data ‘because [the study] assumes a degree of validity already exists… The effort of journalists to be accurate and precise means that there is often rich geo-spatial data embedded in the text.’[i]

My thesis on the 1899 cyclone includes examining its historiography, looking at what’s been written about the event. Here’s an example of the data collated by the Australian Disaster Resilience Knowledge Hub, an open-source database of Australian disasters that ‘informs policy, planning, decision-making and contemporary good practice in disaster resilience’. This is the entry for the 1899 cyclone, Australia’s deadliest, named Mahina:

On 4 March 1899 a category five cyclone hit Bathurst Bay with winds reaching 260 km per hour. A tsunami of 14.6 m swept inland for 5 km. A wave surge measuring 13 m at Ninian Bay adjacent to Barrow Point 30 km south of Bathurst Bay extended inland for 3–5 km. At Bathurst Bay, near Princess Charlotte Bay (Cape York) at least 307 crew members died from a pearling fleet of over 100 vessels plus other craft (with 152 sunk or wrecked, some found kilometres inland) as a result of the storm surge. Over 100 Aboriginals died while trying to help the shipwrecked. [ii]

Just to clarify this data, the eye of the cyclone crossed the coast on 5 March, the winds weren’t measured and there was no tsunami. There is a reliable report of a storm tide that may have been as high as thirteen metres, but that height was not measured and has not yet been corroborated. It wasn’t reported at Ninian Bay. The known dead are fewer than 300, fewer than sixty vessels sank or were wrecked, and none were found kilometres inland. There’s no evidence that 100 Aboriginal people drowned assisting men ashore, but there is evidence that a journalist invented this figure in 1971. The one piece of data that’s supported by reliable evidence is this: there were more than 100 vessels in the area at the time. The rest is the result of newspaper reports mangled by the telegraph, misunderstandings, rumours, exaggerations, people protecting reputations and the editing of information, subtly changing it, each time it passes from one book or journal to another.

This may seem like nitpicking, but scientists and disaster managers have spent time and money incorporating the same data from the same sources into studies and reports that assess future risk. My thesis on the 1899 cyclone represents a sample of one, and I don’t know if data from other disasters is in better or worse shape. But if we’re expanding our sample of cyclones back more than a century to learn what a rare but truly catastrophic cyclone looks like so we can plan for the next one, we may be in trouble if we haven’t done our history homework.

I’m not blameless in the spreading of bad data. I wrote a novel about this cyclone in 2008 and I’ve given talks about it to community groups and libraries over the years while I’ve continued to research it. As I’ve found better evidence, I’ve left the data I’ve discredited scattered like crumbs across the internet, and this is part of the problem. As the history of everything moves online, it becomes harder to correct. The Disaster Resilience Knowledge Hub has the problem of housing and curating a history of Australia’s disasters, and in fairness it does invite contributions, which it says will be ‘reviewed and verified before they are added to the collections’. For the 1899 cyclone, the entry cites sources including two government websites, a scientific paper from 2000, the 1984 Macquarie Book of Events, and an Australian Geographic article. It’s hard to know if the data in these secondary sources has been verified – and if so, verified how? The science for this cyclone has never been tethered to history.

In 2009, crisis manager Ian Mitroff and statistician Abraham Silvers collaborated on a book called Dirty Rotten Strategies: How We Trick Ourselves and Others into Solving the Wrong Problems Precisely (Stanford University Press). They argue that businesses and governments tend to waste a lot of time and money solving the wrong problems. They term these ‘Type 3 Errors’: if we frame the question to a problem the wrong way, or start with false assumptions, what good is the answer? Understanding the 1899 cyclone seems like a job for science, but it’s also a problem of history. It’s been assumed that data found in the historical record and cited in the scientific literature, such as central pressure and wave height, have at some point been verified. They haven’t. They may have originally come from good sources – ships logs, letters from clerks, police reports and interviews with ships’ captains – but with each retelling, the data changed subtly. It’s what French historian Pierre Nora describes as memory’s ‘permanent evolution’ as history moves further away from ‘what is no longer’.[iii] It’s possible to follow this evolution to the point where the data enters scientific journals. It’s why primary sources are so important to historians. Governments and scientists using data based solely on memory risk solving the wrong problem, and doing it with a precision that makes the wrong answer look like the right one.

 

SO, DAN GORDON and I are standing on that ridge where Kenny camped 120 years before. The crest is surrounded by thick scrub, but I can see a higher ridge that’s just behind the beach, and there’s a tidal inlet nearby. Kenny was a duck shooter and carried a police issue Martini-Henry carbine. A crack shot with this gun might hit a target at half a mile, or about 800 metres. If I had Kenny’s experience, I could judge my distance from the beach and my height above the sea. We can define the wrong problem by making wrong assumptions, but let’s assume anyway, as some people have, that Kenny’s estimate of being forty feet above sea level is correct: exactly forty feet. If that’s the case, how high was the storm surge? One answer is forty feet converted to metres: 12.19. Three feet was added to Kenny’s estimate in a scientific paper in 1958, to account for the water up to his waist, and the storm surge became a world-record forty-three feet. Forty-three feet is 13.10 metres, and because Kenny was guessing, it’s been rounded down to 13 metres, but in the US that’s often converted back to feet, 42.65, which the World Meteorological Organisation has rounded down to forty-two feet. Today, depending on which source you read, the storm surge, or storm tide (which are different things) associated with Cyclone Mahina is 12.19, thirteen, 13.7, fourteen, and 14.6 metres. Sometimes, it’s a combination, as if the numbers came from different observers. I’ve heard scientists use the expression ‘garbage in, garbage out’. If the data fed into a model is wrong, the answer is wrong too. The precision of the answer in this case makes the original observation appear to be measured in a way that it wasn’t. The question has been poorly framed, but answered precisely. If those answers are then used to answer other questions about cyclone intensity, or the impact of coastal flooding from a similar cyclone, the answers could be dangerously misleading.

One scientist, whom I greatly respect, recently asked me, ‘What do historians do? I have no idea!’ It’s a good question. The paths of social and natural scientists don’t often cross. Even if they’re studying the same event, such as a cyclone, they’re usually looking at different aspects of it. I’m studying history, but don’t yet consider myself a historian. I’m not a scientist, but I’m interested in science. I’m a journalist by training, and my understanding is that scientists and historians, whatever problem they’re looking at, whether it’s the rise of fascism or the cellular changes that cause cancer, try to reach a hypothesis based on probability. The better the evidence, the greater the level of certainty, but there’s not really any ‘truth’ as such. Historians talk about the credibility of evidence. As Louis Gottschalk wrote in his 1945 essay ‘The Historian and the Historical Document’: ‘What is meant by credible is not that it happened, but that it is as close to what actually happened as we can learn from a critical examination of the best available sources.’[iv]

Today, a large number of sources for most historical topics can be found in digital archives, and to identity the most credible, particularly in old newspapers, I’ve developed my own methodology for testing them. It’s similar to the rules for evidence in law and based on the historical method. It’s not perfect, but I’m looking for credible witnesses and the store manager, Arthur Vidgen, the writer of the letter to whom Kenny described the storm tide, can be shown to be credible. As can Kenny. They are both identified; they can be placed together soon after the event; they’d both gone to the scene; and they had the ability to tell the truth and had no motive to lie. Both were professional record keepers in positions of responsibility, and the information that can be corroborated can be shown to be accurate. A chain of custody can be shown linking Kenny, Vidgen and the first newspaper to publish the letter (which was sent by steamship and avoided the telegraph – a mangler of newspaper articles from that era). Although the letter, its author and subject are credible, the data it contains still requires testing – and that’s a problem for science. No matter how credible Kenny was and how good his description, that storm wave can’t be measured again. It no longer exists. What can be measured, though, is the ridge on which Kenny stood when the sea came up to his waist.

 

THE HEAT IS sapping and I’ve had to walk back to the four-wheel drive for more water. According to my satellite photo, the top of the ridge is eleven to thirteen metres above sea level. It’s not an accurate measurement, but Kenny’s observation does fit within it. I had planned to measure the height using trigonometry, but dense bush, mangroves, heat, lack of time and a fear of crocodiles makes that impossible. It’s a measurement for scientists anyway.

One reason I’m standing here is because of Jonathan Nott, Professor of Geoscience at James Cook University in Cairns, who studies extreme natural events and has long understood the pitfalls in using data from historical documents. In 2000, Nott co-authored a paper for the Australian Journal of Emergency Management titled ‘How high was the storm surge from Tropical Cyclone Mahina?’. Back then, before the existence of many digital archives, such as the National Library of Australia’s Trove newspaper database, the source for much of the published scientific data on the 1899 cyclone was an anonymously authored booklet, The Pearling Disaster, 1899: A Memorial, printed by the Outridge Publishing Company six months after the cyclone. The Outridge family had lost two of its members when the pearling fleets, anchored about forty kilometres from where Kenny was camping, sank. The booklet, based largely on newspaper reports, was given scientific credibility when, in 1958, Brisbane meteorologist Herbert E Whittingham described it as a ‘veritable mine of information’ and used it to reconstruct the disaster for the Australian Meteorological and Oceanographic Journal. Nott wanted physical evidence of marine debris to prove whether Kenny’s observation, repeated by the Outridge booklet, had any credibility, but he didn’t find any. Nor did any model of the cyclone from the data in the booklet predict a storm surge as high as Kenny had described.

Some years later, as a fellow at the State Library of Queensland’s John Oxley Library, I began to find data that predated and contradicted the Outridge booklet. The booklet had reported that William Field Porter, the captain of the Crest of the Wave – the only ship to survive the cyclone’s eye – recorded a low pressure of twenty-seven inches of mercury (27inHg, or 914hPa). But there were better sources, including Porter’s own letters, that said that he’d recorded 26inHg (about 880hPa). To give you an idea of what that means, when Cyclone Tracy smashed Darwin in 1974, its lowest pressure was recorded at 950hPa (about 28inHg).

On 6 March 1899, the day after the cyclone, Porter wrote to his parents describing his observations in detail, including the air pressure falling to twenty-six inches, much lower than the Outridge booklet described. Two days later, Porter gave the same information to Captain Hugh Craig, the pilot of a passing steamship. Porter was a respected ship’s captain with a state-of-the-art barometer, but Captain Craig was sceptical because, as he said, he’d ‘been in some of the worst typhoons ever experienced in the China seas; but he had never known the glass to fall lower than about 27.30’. By the time the Outridge booklet was published, Porter’s official reading had been rounded up to twenty-seven inches. His colleagues didn’t believe a cyclone’s central pressure could possibly fall as low as twenty-six, and his log was changed. The best evidence, however, is that Porter did record 26inHg, or 880hPa, and that although there are qualifications surrounding this reading, it represents the lowest recorded pressure of a southern hemisphere cyclone and one of the most powerful cyclones recorded to make landfall anywhere in the world. In a collaboration with meteorologist Jeffrey Callaghan and disaster-risk analyst Camilla Green, Nott and I searched further, and we published a paper in 2014 presenting the new data. When put into a model, a cyclone with such a low central pressure could, under certain circumstances, be shown to produce what Kenny described on his ridge. This is how both science and history are supposed to work; new information means adjusting theories. Old papers become outdated. The best evidence is that the cyclone was more severe than the record showed, and that has frightening implications for the Queensland coast.

The storm tide is the most dangerous part of a cyclone, and so the more we know about this worst-case event, the better prepared we are for something similar. Nott and I have had a long interest in this cyclone. We’ve been talking about it and swapping theories for more than a decade, since I published The Devil’s Eye (HarperCollins), my 2008 novel based on the disaster. That sort of collaboration seems to be unusual. The benefits of collaboration between disciplines are talked about, but I don’t think it happens often enough.

Now, we’re asking again, how high was that 1899 storm surge? Nott has forensically searched the coast for physical evidence, sea shells or sediment, high enough to account for what Kenny described. I’ve been forensically searching archives. We’re looking for the same thing, but from different perspectives – and so I shouldn’t have been surprised when, after pinpointing Kenny’s campsite on a map and then stepping onto the sand ridge, I found what Nott had been looking for: sea shells.

Finding this site would not have been possible without history and science acting together.

 

THE HISTORICAL RECORD that scientists have been interrogating is a minefield of misinformation. There’s a report repeated in the scientific literature of porpoises being found fifty feet, or about 15 metres, up a cliff on Flinders Island, near where the pearling fleets sank. The first reference to porpoises up a cliff is in the Brisbane Courier newspaper on 14 March 1899: ‘Thirteen porpoises were found 15ft up the cliff at Flinders’. But it entered the scientific literature in 1959 as, ‘At Flinders Island searchers found thirteen dead porpoises fifty feet up a cliff hurled there by wind and waves.’ This was written by journalist Norman Bartlett in his 1954 book, The Pearl Seekers, and it was Bartlett’s report that was cited five years later by meteorologist Herbie Whittingham, who linked it directly to Kenny’s storm tide. The two events happened fifty kilometres apart and the original newspaper report, in any case, was based on an unsourced rumour.

We’re all prone to making assumptions that lead to mistakes: every document, including this one, contains errors and biases. And I think there’s a role for historians in sorting out the history of disasters. More recent cyclones, certainly since satellites have been around, are well recorded and I don’t assume that the problems I’ve seen in the scientific literature associated with the 1899 disaster are repeated for every other historical cyclone. But if we’re mining archives for examples of the disasters we could face on the basis of the rare catastrophic ones that might become more frequent, it’s a question that needs to be asked.

When Dan Gordon and I stepped from the Northern Road of the Barrow Point people and onto Kenny’s campsite, we found the top of that long flat ridge covered in shells, and that raised questions that science and archaeology can help history answer. Were the shells placed here, or were they washed up by a wave? Collaboration can better answer the questions than either science or history alone. We’re worried about the impacts of a changing climate, and we need to get the history right to get the science right. Thanks to science, we know that the 1899 cyclone occurred in the middle of an unusually long period of El Niño, an event that usually produces fewer, but more severe, cyclones. Climate change will mean more of the same conditions. Are we prepared for another cyclone like the one in 1899? I don’t think we can be if we’re not more careful with the historic record. I hope science will prove me wrong.

 

References

[i] Yzaguirre A; Smit M; Warren R. 2016. ‘Newspaper Archives Text Mining = Rich Sources of Historical Geo-spatial Data’, IOP Conference Series: Earth and Environmental Science, 34, No. 1.

[ii] Australian Government, Attorney General's Department, ‘Cyclone Mahina’, Australian Disaster Resilience Knowledge Hub, Australian Institute of Disaster Resilience, https://knowledge.aidr.org.au/resources/cyclone-cyclone-mahina-cape-york-queensland.

[iii] Nora P; Roudebush M. (trans.), 1989. ‘Between Memory and History: Les Lieux de mémoire’, Representations, 26, 8.

[iv] Gottschalk L. 1945. ‘The Historian and the Historical Document’, The Use of Personal Documents in History, Anthropology, and Sociology, prepared for the Committee on Appraisal of Research, Gottschalk L. et al (eds), New York: Social Science Research Council, 1945, 46.

 

Get the latest essay, memoir, reportage, fiction, poetry and more.

Subscribe to Griffith Review or purchase single editions here.

Griffith Review