Saffron has not set seed since the Bronze Age

A crocus flower gives up three stigmas. They sit inside the flower as three red filaments joined at the base, each one widening at the tip into a small trumpet, and someone has to open the flower, pinch them out, and set them to dry. To get a kilogram of dried saffron takes roughly 150,000 flowers and somewhere between 370 and 470 hours of that work. ¹

The part that rarely gets said out loud is that the plant doing this cannot make a seed. Crocus sativus is a triploid, carrying three copies of every chromosome instead of two, eight triplets in all, and at meiosis they have no clean way to pair off. ² The flowers are sterile. Every saffron field ever planted came from corms, the swollen underground stems the plant makes for itself, dug up and divided and carried by hand. ¹

Where it actually came from

That leaves a question people argued over for nearly a century: if saffron never crosses with anything, where did it begin, and out of what? ² The debate ran on area, on date, and on which wild species had contributed. ⁶ Iran's modern dominance made the obvious answer feel settled, and the name arriving in English through Arabic reinforced it.

Three lines of evidence closed on a different answer inside two years. In 2018 a group at the Leibniz Institute in Gatersleben sequenced multiple loci across the saffron crocus species group and found C. sativus sitting inside one wild species, Crocus cartwrightianus, with nothing to suggest any other species had contributed. ⁴ In 2019 the same group genotyped across the genome and placed 99.3 per cent of saffron's alleles within C. cartwrightianus, naming it the sole progenitor. The wild population most similar to the crop grows in the vicinity of Athens, and the crop appears to have arisen in Attica by combining two different genotypes of that single species. ³ A separate group painted the chromosomes with multi-colour probes, doing no sequencing at all, and read the same structure straight off the karyotype. ²

One qualification belongs here. In 2015 a Leicester group using retroelement markers had argued the opposite mechanism: an allotriploid built from two species rather than one, most likely C. cartwrightianus crossed with C. pallasii. ⁵ The disagreement is narrower than it sounds. It is about whether three chromosome sets came from one species or two, not about whether the plant is Greek, and the marker system is lower resolution than the genome-wide and cytogenetic work that followed it. By 2022 members of both 2019 teams were writing jointly that the artworks and the genetics converge on ancient Greece. ⁶ The fair sentence is that the evidence now points to Attica, which is what those authors say, rather than that the matter is closed.

One plant, eleven countries

What both camps agree on is more striking than what divides them. Both conclude it happened once. And both find the same thing afterwards. The Leicester group saw no polymorphisms at all among seventeen saffron accessions collected from Kashmir through Iran to Spain, in an assay that had no trouble separating the wild relatives run alongside them. ⁵ A Spanish group had already tested forty-three accessions from eleven countries against thirty RAPD primers, forty-eight ISSR primers and fifteen microsatellites, and recovered not one polymorphic band. ⁷

So the Iranian saffron and the Spanish saffron and the Kashmiri saffron on a shelf are not varieties. They are one clonal lineage, cut and replanted for something over three millennia. ⁶ Whatever differs between them was done to them by weather, soil and handling, because nothing else has had the chance to differ.

That turns out to be measurable, and it is the most quietly astonishing result in the literature. An Italian and Spanish team ran AFLP across Spanish germplasm and did find a little variation: 4.23 per cent of peaks polymorphic, twelve effectively distinct genotypes, the residue of somatic mutation piling up in a plant with no other route to change. Then they ran the methylation-sensitive version of the same assay on the same material and got 33.57 per cent polymorphic peaks and twenty-eight distinct epigenotypes, patterned by where the corms had been grown. ⁸ Roughly eight times more variation in the marks laid on the genome than in the genome itself.

The wall painting shows the wild parent

There is a Bronze Age settlement at Akrotiri on Thera, buried by a volcanic eruption and dug out with its wall paintings intact. In the building excavators label Xeste 3, the paintings show women gathering crocus flowers and carrying the stigmas to a seated female figure. Ferrence and Bendersky, reading the frescoes in 2004, identify the plant as Crocus cartwrightianus and argue that the seated figure is a divinity of healing shown with saffron as her attribute. ⁹

Take the theology as an argument, because that is what it is. The botany is the part that lines up. The species painted on that wall is the wild parent the genetics points to, growing in the region the genetics points to, being harvested by people who had already organised the work into stages. ⁹ Two disciplines that share no methods and no evidence arrived at the same corner of the Aegean.

What is actually in the thread

Saffron divides its labour unusually cleanly. Colour comes from the crocins, taste from picrocrocin, aroma from safranal. ¹⁰ The crocins are the odd ones: they are carotenoids, and carotenoids are meant to be fat-soluble, but here sugars are esterified onto both ends of a crocetin backbone, glucose and gentiobiose and neapolitanose, and the molecule dissolves in water instead. ¹¹ It also accumulates far beyond what a pigment normally does, reaching 10 per cent of the stigma's dry weight and stored in the vacuole, a water compartment an ordinary carotenoid could never enter. ¹² A few threads will colour a whole pot without any fat to carry them.

The aroma has the better story. Safranal is the aglycone of picrocrocin: the bitter compound with its sugar taken off. ¹¹ Drying does not create saffron's smell, it uncovers it, either enzymatically through an intermediate or directly by thermal degradation. ¹¹ The bitterness and the perfume are one compound at two stages, which is a useful thing to hold in mind while deciding how long to warm a pinch of it.

Forty-three per cent

Expensive, dried, and shaped like a thread is a poor combination. A 2024 study tested 104 market samples from sixteen countries by morphology, HPLC, HPTLC and DNA barcoding, and found forty-five of them, 43 per cent, adulterated. Barcoding turned up twenty different adulterant plant species, some of them unsafe to eat. ¹³ A 2026 systematic review puts global adulteration at 20 to 30 per cent with a huge regional spread, from 3.5 per cent in regulated EU markets to 60 per cent in India, with safflower, marigold and turmeric as the usual substitutes and synthetic dyes including Sudan compounds and auramine-O as the dangerous ones. ¹⁴

There is one honest tell, and it is chemical rather than visual. Picrocrocin has been found only in the genus Crocus, and no other spice reproduces the taste, so it works as a marker for the real thing. ¹¹ Reassuring in a laboratory. Of no use whatever at a market stall, which is the point.

Not in the statistics

One last oddity. FAO's crop production database carries 792 distinct items. Pepper has its own line, so do vanilla, cinnamon, cloves, nutmeg, anise and ginger. Saffron has none. It falls into item 723, a residual category called Other stimulant, spice and aromatic crops, n.e.c. ¹⁵ The most expensive spice in the world ¹² is statistically invisible in the main international record of what farms grow.

Which has a consequence for the figure everybody quotes. That Iran grows more than 90 per cent of the world's saffron comes from an FAO page describing a heritage-farming programme in Gonabad: undated, with no tonnage attached, and not a statistical release. ¹⁶ It is probably close to right. It is simply not measured the way the pepper figure is measured, and the best available number for world production, roughly 430 tons a year, comes to us second-hand too. ¹

None of that alters what to do with the threads. Warm them, steep them, use fewer than you were going to. But it changes what is in your hand. A flower that cannot set a seed has been kept alive for three thousand years purely because people kept deciding, season after season, to lift its corms and put them back in the ground. It cannot adapt and it cannot spread on its own, and on that arrangement alone it has travelled from Attica to Kashmir to Spain. The flowers open in autumn, in the morning, and they have to be picked the same day.