Kept by Secrecy

The stock whose keepers know exactly where it is and will not say: a tree valued by everyone, findable by almost no one, because the catalogue entry is itself the threat

R838’s wild cousin was kept by indifference — a margin nobody improved because nobody cared. Turn that inside out. Here the plant is cared about intensely, by name, in parliament and in newspapers, and it is kept by the opposite act: a small circle of people who know precisely where it grows and have agreed, for thirty years, not to tell. The keeping is not an omission. It is a withheld fact, renewed every time a journalist is flown in with the view blocked, every time a photograph has its location tags stripped, every time a database answers a request for this species with nothing at all.

A fossil that was breathing. In September 1994 a New South Wales parks officer, David Noble, abseiling into a remote sandstone gorge in Wollemi National Park, roughly 150 kilometres northwest of Sydney, found trees he did not recognise — dark, bubbly bark, fern-like foliage. They were Wollemi pines, Wollemia nobilis, a genus known until then only from fossils tens of millions of years old and assumed long extinct. The discovery was announced to the world. The address was not. That split — publish the species, classify the site — was the conservation plan from the start, and it has never been relaxed.

How little there is to hide, and how completely it is hidden. The recent federal recovery planning counts the entire wild population at 45 mature individuals and 46 seedlings, in four small stands within one canyon system; older summaries say simply fewer than 100 mature trees. New South Wales sorts sensitive species into tiers of disclosure, and the Wollemi pine sits in the strictest one: no records provided at all, because even a general locality may threaten a taxon this exposed to exploitation and disease. The general machinery of open biodiversity data makes the same judgement at scale — platforms such as iNaturalist automatically obscure the coordinates of threatened species, replacing a true point with a random one inside a cell of roughly 500 square kilometres. For most species the map is blurred. For this one, the map is blank.

What the secret is keeping out. Not, mainly, collectors — though disturbing the trees has carried a fine reported at A$220,000. The sharpest threat rides in on a boot. Phytophthora cinnamomi, a soil-borne water mould that rots roots and has no cure, can arrive in a gram of infected mud. Authorised visitors sterilise boots and gear; even so, the pathogen was detected among the wild trees in 2005, believed carried in by unauthorised visitors who had found the site anyway. Read that date carefully: the secrecy did not fail on paper. It failed the one way secrecy can — somebody learned the fact, and acted on it, and the fact once acted on cannot be unlearned. A fence can be repaired. A location, once loose, stays loose.

The paradox the keepers chose on purpose. Against the hidden grove stands a deliberate flood. From 2006 the pine was propagated for public sale — hundreds of thousands of plants prepared, licensed nurseries growing it on several continents, botanic gardens worldwide holding a shared living collection, translocated stands planted elsewhere in the park. The reasoning was plain: make the tree easy to buy and there is less reason to go and take one; spread copies everywhere and the loss of the canyon, if it comes, is not the loss of the species. So the organism was made common and the address was kept rare. Secrecy here does not hoard the thing. It hoards only the one fact that would let the world’s affection arrive in person, with mud on its boots.

The secret’s own emergency. In the 2019–20 fire season the grove lay in the path of a vast fire, and keeping it meant breaking the seal in a controlled way: specialist crews were winched into the canyon, told where to go, and set to work with irrigation lines and water-bombing aircraft while the coordinates stayed unpublished. Most of the wild population survived. That operation is the secrecy’s floor made visible. A secret held by nobody could not have mounted it; a secret held by everybody would not have needed to — the grove would already have had its 2005 many times over. The keepers must be numerous enough to act, few enough not to leak, and replaceable enough that the fact outlives each of them. That is a narrow corridor, and it has to be walked every year, not once.

The toy — schematic, stated plainly. No empirical parameter: suppose the location is known to n people, and in any one year each knower has a small fixed chance q of the fact escaping through them — a tagged photo, a careless direction, a notebook left open. Take q = 1% per knower per year. The secret survives a year only if every knower keeps it, so survival to year t is (1 − q)n·t, and the median lifetime of the secret — the year by which half of such secrets are out — is ln(0.5) / (n · ln(1 − q)). The numbers below are chosen to show the shape, not measured from any ranger station.

Run it and the shape is unforgiving in one direction only. Tripling the circle does not triple the risk in any gentle sense — it divides the secret’s expected life by three, because every knower is an independent door and the doors do not guard each other. Three knowers give the secret a median life of about 23 years; thirty give it about 2.3. And yet the corridor’s other wall is just as real, and the toy cannot show it: at n = 1 the secret does not leak, it vanishes, the first time its one keeper retires, falls ill or simply misremembers which fork of the canyon. R835’s Fogbank was exactly a secret that died that way — kept so close its own makers could not readmit themselves. So this keeping’s floor is neither R836’s percentage nor R837’s keeper-count of planters. It is a count of knowers held inside a corridor: above the number that can still mount a rescue, below the number that cannot keep a fact.

So here is the ninth shape. The accession is kept by being withdrawn, the landrace by being used, the wild cousin by being left — and the pine by being found and then unfindable: the only keeping in this run whose enemy is knowledge itself, held off not by distance or worthlessness but by a discipline renewed in every boot-wash and every blank database row. Its strangest property is that success looks like nothing. No crowd at the gate is the evidence it worked. Thirty years of nobody arriving is the whole monument, and it can be dismantled by one person, once, saying where.