Mitti Attar: How Kannauj Distils the Smell of Rain

There is a smell almost everyone can recognise and almost nobody can describe. It arrives a minute before the first heavy rain of the season, when the air over dry ground changes and a whole street seems to exhale at once. English has a word for it now. In Kannauj, a town in Uttar Pradesh that has been distilling fragrance for a very long time, it has a name, a season and a production line: mitti attar, the attar of baked earth.

The part that stopped me when I started reading about how it is made was the first step. To bottle the smell of rain, you begin by drying the ground out as far as it will go. Then you bake it.

An attar is distilled into oil, not into alcohol

This is the distinction everything else rests on. A modern perfume is aromatic material dissolved in alcohol. A traditional attar has no alcohol in it at all. The fragrant vapour is condensed directly into a receiving vessel already filled with a carrier oil, historically sandalwood, and the oil absorbs it.

That single choice shapes the whole workshop. You cannot simply run a still and collect a clear distillate at the other end. You have to keep a vessel of oil cool enough, for hour after hour, that vapour arriving from a boiling pot will give itself up to it rather than sit on top as water.

The apparatus is copper, bamboo, clay and a tank of water

Four things, none of them powered.

The deg is a wide copper pot, the still, charged with the raw material and water and closed with a domed lid. The bhatti is the clay furnace beneath it, fed with wood and dung cakes. The chonga is a bamboo pipe, wrapped along its length with twine and cloth, that carries vapour out of the deg. And the bhapka is the receiver: a long-necked copper vessel holding the base oil, sitting in a shallow tank of water.

There is no thermometer anywhere in that list, and no valve. The joins between deg, chonga and bhapka are sealed by hand with a paste of clay and cotton fibre, worked on wet and left to stiffen in the heat of the fire.

The earth is chosen, shaped and half-baked

For mitti attar the raw material is clay, taken from riverbanks and dried beds, cleaned of roots and grit, and formed into rough discs or crude little pots. These are fired in a small kiln, and only partly. Fully vitrified pottery has been cooked past the point of usefulness here; what is wanted is earth that has been driven dry and hot and no further.

Accounts differ on how much the firing matters and how hot it should be, and the people who make it well are not in a hurry to be precise in print. What is consistent is the sequence, and the sequence is the strange bit: the ground is baked until nothing wet is left in it, broken up, and then put back into water inside the deg. The rain is being reconstructed.

The round takes most of a day and can be lost in an hour

The fire is lit and kept deliberately low. Vapour rises, travels out through the bamboo, and condenses into the oil waiting in the bhapka, which is turned and rocked by hand so the new material folds into the base instead of collecting on it. The tank of water around the receiver is the only cooling there is, and it warms up as the day goes on, so it is changed.

A round runs roughly six to eight hours. Almost all of the skill is in the fire. Too hot and the charge scorches against the inside of the deg, and a burnt note is not something you can lift out later — the batch carries it to the end. Distillers watch the deg, and sponge it down with cool water when they think it is running away from them.

The seal is the other failure. If the clay-and-cotton paste cracks and vapour finds its way out at a joint, the loss is invisible and total, and the day is simply gone.

The next morning the water goes back in

Left overnight, the contents of the bhapka separate. The water — the hydrosol — is drawn off the top and, rather than being thrown away, is poured back into the deg to carry the next day's distillation. The oil keeps what it has taken up.

Then the whole thing is done again. And again. Published accounts of how many days this takes disagree sharply: some say around a fortnight for a small quantity, some say fifteen to twenty days, and for mitti attar in particular there are descriptions of the cycle being repeated daily for up to two months before the base is considered saturated. It plainly depends on the material, the season and whose workshop you are standing in, and I would treat any single confident number with suspicion.

The base oil is the expensive problem

Sandalwood oil is what makes a traditional attar behave the way it does, and it has become extremely scarce and extremely costly. The consequence is the one you would expect. A great deal of what is sold as attar now sits on a cheaper carrier — liquid paraffin, DPG, other synthetic esters — and the label rarely says which.

This is a familiar shape to anyone who follows Indian fragrance materials; it is almost exactly the story of halmaddi in incense, where the ingredient that gave a product its character became the first thing dropped when it got expensive. None of it makes the substitute version fraudulent on its own. It does mean that "attar" on a bottle tells you about a format, not about what is in it, much as the words on an incense label describe a category rather than a recipe.

The leather bottle that finishes the job

The last traditional step is a kuppi, a hand-formed bottle of camel hide. Attar is stored in it for weeks or months, and the leather is porous enough that the small amount of water still held in the oil sweats out through the wall and evaporates, while the oil itself stays. Old perfumers treated this as non-negotiable and are quoted saying an attar never kept in a kuppi was ruined. Very few workshops still use them.

The word is younger than the craft

In March 1964, two CSIRO scientists, Isabel Joy Bear and Richard Thomas, published a short paper in Nature on what they called argillaceous odour, and coined the word petrichor for it. They also did something worth remembering: they noted that perfumers in India had already captured this smell and absorbed it into sandalwood oil, and gave its name — matti ka attar. The naming came out of Australia. The bottling was already going on in Uttar Pradesh.

What the science settles is narrower than people assume. Geosmin, a compound produced by soil bacteria, is usually named as the molecule behind the smell of wet ground; Bear and Thomas were describing a yellowish oil accumulating in the pores of dry rock and soil. Whether a given mitti attar is carrying one, the other, or a mixture is not something I have found a good source willing to state plainly, and I am not going to fill that gap for you.

I have never worked a deg, never sealed a joint, and never made anything in this article. What is here is what distillers and published accounts describe, and they disagree with each other more than a tidy summary would suggest.

What stays with me

It is the reversal. Everything else in perfumery is an attempt to hold on to something living — a flower cut at dawn, a wood, a resin. This one waits for the ground to crack, takes the driest thing available, bakes the last of the moisture out of it, and gets back a smell that the whole subcontinent reads as water. Next time it hits you on the way to the gate, it is worth a second of attention. Somebody in Kannauj has spent two months trying to make that.