Two suppliers quote you white sesame. Both say 99% purity. One is four rupees a kilo cheaper, and the obvious conclusion — that the cheaper one is simply hungrier for the order — is wrong often enough that it is worth understanding why before you sign. The number is identical. The thing being measured is not.

The short version: “99% purity” is an admixture figure. It counts stones, chaff, stalk and dust by weight. It says nothing whatsoever about how many seeds in the lot are discoloured, immature or mould-marked — and those are the ones that show up after hulling.

Two machines, two entirely different jobs

Mechanical cleaning is a sequence of physical separations: a destoner, an aspirator pulling off dust and light chaff, gravity tables sorting by density, sieve decks sorting by size. It is fast and it is cheap. It also works on exactly one principle — that contamination differs from good seed in size, weight or aerodynamics.

Discoloured sesame is the same size and the same weight as sound sesame. No sieve on earth will catch it.

ContaminantMechanical cleanOptical sort
Stones, mud balls, metalRemovedRemoved
Chaff, stalk, dustRemovedRemoved
Undersized & shrivelled seedMostlyRemoved
Discoloured & off-colour seedPasses throughRemoved
Mould-marked & insect-damaged seedPasses throughRemoved
Same-size stone or glass fragmentPasses throughRemoved

Sortex cleaned sesame seeds are quoted at 99.95%, not 99%

This is the tell, and it is the most useful thing on this page. Around 99% purity — the familiar 99/1 notation, meaning 99% sound seed against 1% admixture — is roughly the practical ceiling of mechanical cleaning. Optical sorting starts where that ends, which is why genuine sortex cleaned sesame trades as 99.90, 99.95, 99.97 or 99.98, and why export contracts specify the decimals rather than rounding.

Where each process lands on the purity scale Machine cleaned 98.5 – 99.0% Sortex cleaned 99.90 – 99.98% 98.5% 99.5% 100% The gap between the two bands is colour, not weight.

A flat “99%” on a sortex claim is a contradiction worth querying.

The premium is yield loss, not machine time

Buyers assume the extra cost covers electricity and depreciation on an expensive machine. It mostly doesn’t. Run 100 tonnes of raw sesame through an optical sorter and 2 to 5 tonnes come out the reject chute — sound-weight seed that failed on colour, which must then be sold into a lower-value stream at a real discount. That rejected tonnage is the premium. The machine hours are almost a rounding error against it.

Which means the discount you were offered is not generosity. It is a lot that never lost its 3%.

What to put on the purchase order

  1. Specify purity to two decimals when you want sortex. “99%” invites a machine-cleaned delivery you cannot reject.
  2. Ask for admixture and defective-seed counts as separate lines. They are separate tests.
  3. Require a lot-specific analysis certificate, not a mill average from last quarter.
  4. State moisture alongside purity — a clean lot at 9% moisture will not stay clean.

Questions buyers ask us

Is sortex cleaning always worth paying for?

No. For oil extraction, colour is irrelevant — the crusher does not care what the seed looked like. Pay for optical sorting when appearance reaches the end consumer: bakery toppings, tahini, confectionery, retail packs.

Does sortex cleaning affect oil content?

Marginally, and upward. Removing shrivelled and immature seed lifts the average slightly. It is not a reason to buy sortex, and any supplier selling it as one is reaching.

Can a lot be double-sortexed?

Yes, and 99.97%-plus grades usually have been. Each pass costs another slice of yield, so ask what you are actually paying for rather than assuming more passes are better.

Buy the specification, not the adjective

“Sortex” has become a word suppliers attach to lots that have seen a gravity table and good intentions. It is not a certification and nobody polices it. The decimal place is the only part of that claim you can hold anyone to.

We publish purity, admixture, moisture and oil content against every lot in the Sain Agro catalogue, and we would rather lose an order on a spec than win one on a word. If a competing quote looks unusually sharp, ask for the defective-seed count — the answer is normally the whole explanation.

References & further reading