Anywhere from a day to a few hundred thousand years, and the wide end is narrower than usually reported. Salt crystallises out of a drying puddle overnight. The giant gypsum beams at Naica have never returned a measured age anywhere near the million years that circulates.
The famous million-year figure is a measurement across the crystal, not along it. Van Driessche and colleagues state the condition in the sentence that produces it: a gypsum crystal beam one metre in diameter, forming at 55 degrees, should have been growing for 0.99 million years, plus or minus 0.27. Diameter. The beams in the Cave of Crystals run to roughly eleven metres long and about one metre thick, and the paper is explicit that thickness is the dimension that governs, because the crystallographic direction they measured is the only one that sets beam thickness.
An eleven-metre beam is therefore not eleven million years old. The length grew by a different mechanism at a different rate, and the paper draws the line itself: it is not possible, the authors write, to offer a general age for all these crystals.
What the dating returned
Sanna, Forti and Lauritzen sampled the crystals and dated them by uranium-thorium, describing the work throughout as preliminary. The ages they measured run up to 191 thousand years, plus or minus 13. From growth rates they extrapolate further, and get two different answers: one deposition value gives about 245,000 years for the largest crystals, while on the rate data available to them the oldest crystals converge on something close to 300,000. An earlier absolute date by thermal ionisation mass spectrometry had put it at 600,000.
None of those numbers reaches a million. None is settled either. What they share is an order of magnitude, hundreds of thousands of years, reached by routes that disagree with each other and published as a preliminary assessment.
Sanna’s depositional rates of 0.56 to 1.22 millimetres per thousand years are described as in excellent agreement with tests on gypsum deposition run inside the deepest part of the mine, which is not the same body of work as the flow-cell measurements. Converting those flow-cell rates to the same units gives roughly 0.44 and 0.50 millimetres per thousand years, a little below the field range and not inside it.
The growth rate has a wrinkle of its own inside the same paper. The abstract gives the slowest measurable rate at 55 degrees as 1.4 hundred-thousandths of a nanometre per second. Table 1 and the results section give 1.6, and the age arithmetic in the paper uses 1.6. Both numbers are the authors’. Quoting the smaller one means quoting the abstract.
That rate is often reported as the slowest crystal growth in the world, which drops three qualifiers the authors kept. The rate is the slowest that has been directly measured. It is a normal growth rate, meaning growth perpendicular to a face rather than along one. And it came out of a laboratory flow cell running natural mine water over cleaved gypsum, not off anything in the cave.
Why there were so few crystals to begin with
The reason Naica produced beams instead of a crust is the same reason a granite has visible grains and a basalt does not, and it has almost nothing to do with patience.
Otálora and García-Ruiz put the question in its sharpest form. Why are there so few? Their answer is that the crystals nucleated at very low supersaturation, which they call the only way to keep the number of crystals low enough that each one can later grow very large. Size is a consequence of scarcity at the start. A fluid barely over the threshold starts almost nothing, and whatever does start inherits the entire supply.
Keeping a fluid barely over that threshold for a few hundred thousand years is the hard part, and Naica had a mechanism for it. Anhydrite and gypsum swap stability at a particular temperature, and a system held near that crossover dissolves one and deposits the other at a trickle. The authors call it an ion pump, and it is what kept calcium and sulfate arriving at a rate low enough to feed growth without triggering a fresh crop of nuclei.
The crossover temperature gets quoted more precisely than any source supports. Van Driessche’s paper uses 58 degrees; Otálora and García-Ruiz, with an author in common, use 54.5 and warn that the exact value is very elusive and variable in different works, flagging uncertainties of three and a half degrees and possibly more. Around 55 degrees is defensible, and a flat 58 is not.
The other end of the range
Salt and sugar crystallise in hours from a solution left to dry, small and imperfect, because a rapidly concentrating liquid starts far too many crystals at once for any of them to get anywhere. Gypsum does the fast version too. A desert rose is the same calcium sulfate as a Naica beam, grown in damp sand where evaporation drives it. A dinner-plate-sized one has sat on our shelves for years, its blades large enough and layered enough to open outward on every side instead of settling on a single best face. The formula is identical. The supply rates are several orders of magnitude apart.
Figures that have since moved
Large bodies of magma were long said to cool and crystallise over roughly a hundred thousand to a million years, and that number comes from a single-pulse conductive model. Glazner and colleagues ran that model themselves, getting the whole body below 750 degrees in about 500,000 years, and then argued from field data that real plutons do not behave that way: their evidence indicates a magmatic lifetime closer to ten million years for the Tuolumne Intrusive Suite, assembled in increments rather than emplaced at once.
Other rates that get quoted confidently have no publishable general value at all. The fastest stalagmite growth rate in the literature, one millimetre a year at Proserpine, is called exceptional by the team that measured it, which makes it the worst possible candidate for a typical figure. No general figure exists for how fast halite grows in the ground, or natural quartz, and a page that supplies one has supplied it from nowhere.
Growing one on purpose
Synthetic quartz is grown hydrothermally in steel pressure vessels called autoclaves, and the USGS gives the run length as 30 to 60 days for crystals to reach the desired size. Months, not weeks. Virtually all the quartz crystal used in electronics is cultured instead of mined, and lascas mining and processing in Arkansas, the raw feedstock operation the USGS tracks, ended in 1997.
Set a 30-to-60-day run against a few hundred thousand years and the ratio lands somewhere between two and four million to one, depending which end of each range you take, and none of it is a difference in the mineral. It is a difference in how hard the supply is pushed. An autoclave holds a steep gradient on purpose and keeps refreshing the feed. Naica did the opposite for a few hundred thousand years, holding the faintest gradient a solution can hold and topping it up by dissolving its own wall.
Formation time appears on no mineral data sheet anywhere, and the reason is that two pieces of one species can differ in it by six orders of magnitude without differing in a single measurable property. Gypsum’s entry in the Crystalance Mineral Library reads the same for a desert rose and for an eleven-metre beam. The clear and fibrous forms of it, and how to tell them apart, are treated under selenite.
For a tumbled stone with no locality and no matrix there is nothing left to date at all, which is why a published age exists for the Naica beams and not for the specimen on a shelf. The mine kept the cave, the water and the wall rock in one place, and a sampling team could reach every part of it.
Sources
- Alexander Van Driessche and others, “Ultraslow growth rates of giant gypsum crystals”, PNAS 108(38), 2011, for the growth rates at 55 degrees, the one-metre-diameter age calculation, thickness as the governing dimension, and the refusal of a general age.
- Laura Sanna, Paolo Forti and Stein-Erik Lauritzen, “Preliminary U/Th dating and the evolution of gypsum crystals in Naica caves”, Acta Carsologica 40(1), 2011, for depositional rates of 0.56 to 1.22 mm per thousand years, uranium-thorium ages to 191 thousand years, the extrapolated estimates of about 245,000 and close to 300,000 years, the earlier thermal-ionisation date of 600,000, and the authors’ own description of the work as preliminary.
- Fermin Otálora and Juan Manuel García-Ruiz, “Nucleation and growth of the Naica giant gypsum crystals”, Chemical Society Reviews 43, 2014, for columnar crystals up to eleven metres, beams over five tons, nucleation at very low supersaturation as the answer to why there are so few, the anhydrite-gypsum ion pump, and the crossover temperature as elusive and variable.
- Allen Glazner and others, “Are plutons assembled over millions of years by amalgamation from small magma chambers?”, GSA Today 14(4/5), 2004, for the conductive-cooling model result and for a magmatic lifetime closer to ten million years at Tuolumne.
- Stef Vansteenberge and others, “Reconstructing seasonality through stable-isotope and trace-element analyses of the Proserpine stalagmite, Han-sur-Lesse cave, Belgium”, Climate of the Past 16, 2020, for the Proserpine stalagmite’s mean growth rate of one millimetre a year, described by the authors as exceptionally high.
- Gemstones, Synthetic and Simulant, USGS and US Bureau of Mines SP-14-95, for autoclave runs of 30 to 60 days; and Mineral Commodity Summaries 2024, Quartz Crystal, USGS, for cultured crystal in electronics and for lascas mining and processing in Arkansas ending in 1997.
