Count to ten in Romani, the language of the Roma people of Europe.
Ekh, duj, trin, štar, pandž, šov. If you grew up with Punjabi, Hindi or Urdu you have just recognized every one of those: ek, do, tin, chār, panj, chhe. Then something happens. Efta, oxto, enja. Those are Greek, close to what you would hear in an Athens market today. Then deš, and you are back in India.
Ten words, and they hold a map. Migration is not just the movement of DNA. It is the physical transmission of human ingenuity. This post is about what people carry with them, besides their genes, when they walk out of one continent and into another. The answer is written into the numbers.
Named after the wrong continent
When Roma families first appeared in Western Europe in the early 1400s, the people who met them guessed where they had come from, and guessed Egypt. Gypsy is Egyptian worn down by five centuries of use. The guess was wrong by about three thousand miles. Most Roma today do not use that word for themselves, and many hear it as a slur, so this post says Roma for the people and Romani for the language.
The language is what corrected the guess. In 1763 a Hungarian student in the Netherlands, István Vályi, fell into conversation with students from the Malabar coast of India and realized he could recognize their words from the Roma he knew at home. In 1782 a German scholar, Johann Rüdiger, sat down with a Romani speaker and a grammar of Hindustani and published the comparison. Nose is nakh in Romani and nāk in Hindi. Eye is jakh and ānkh. Water is pani in both. Black is kalo and kālā. The words for the body, for water, for family and for counting are the words a language keeps longest, and every one of them pointed east.
A language keeps a diary
A language does not only remember where it started. It picks up words wherever its speakers stop for a while, and it keeps them in the order it collected them, the way a cliff keeps its layers. The linguists who study Romani read it exactly that way.
The bottom layer is India: the numbers, the body, water, bread, the names for family. Above it is a thin Persian layer, and the kinds of words tell you what a life on the road needs. Baxt is luck, zor is strength, ambrol is a pear. Above that an Armenian layer, and it is about work: grast is a horse, bov is an oven, kotor is a piece of something. And above that a layer so thick that it changed the grammar as well as the vocabulary. Drom is a road, foro is a town, zumi is soup, kurko is a week, luludi is a flower, and seven, eight and nine are efta, oxto, enja. Nobody borrows their numbers from strangers. Numbers come from people you have traded with for a very long time, and the Roma lived inside the Greek-speaking Byzantine Empire, in what is now Turkey and the Balkans, for centuries.
Then the layers split. Above the Greek, each Romani dialect carries the language of wherever that family went next: Romanian, Hungarian, Serbian, German, Spanish, English. That is the fan at the top of the cliff.
What the genes add
Since 2012, genome-wide studies have read the same route out of DNA. Both of the large ones place the origin in north-west India. One, by Isabel Mendizabal and colleagues in 2012, put the founding of the Roma population at about 1,500 years ago. The other, by Priya Moorjani and colleagues in 2013, said 1,000 to 1,500 years ago, and dated the first mixing with European populations to about 850 years ago.
Here is the one piece of arithmetic in this post. Call a generation twenty-five years:
850 ÷ 25 = 34Thirty-four grandmothers in a row. The people who carried those numbers out of the Byzantine Empire and into Europe are the ancestors, thirty-four times over, of Roma children in Spain, Romania, Hungary and Britain today. And ekh, duj, trin are still the numbers of Punjab.
The Zutt, and the water buffalo
There is an older story, and it comes from Arabic chroniclers rather than from genes or grammar.
From the fifth century on, Arab and Persian writers mention a group they called the Zutt, living in the marshes of southern Iraq between Basra and Wasit. The word is Arabic for Jat, the herding people of the lower Indus in what is now Sindh and Punjab, and one chronicler explains that it is simply how the Sindhis said the name. The Zutt came with great herds of water buffalo, the animal you still see in those marshes today, and they farmed rice on salt-crusted, flooded land nobody else wanted. They knew water and animals the way river and desert people do, and by the ninth century they controlled the marsh channels well enough to tax the boats bound for Baghdad. In 834 the caliph sent an army against them, and after nine months of fighting in the reeds the Zutt surrendered. They were deported, tens of thousands of them, to the far northern edge of the empire, in Cilicia on the Byzantine frontier. Twenty years later, in 855, a Byzantine raid on the town of Anazarbus carried them off again, this time into Byzantium.
In 1903 a Dutch scholar, Michael Jan de Goeje, proposed that these people were the ancestors of the Roma. Out of the Indus, through Persia, into the caliphate, into Byzantium. It is the road the language records, with dates attached.
Where the evidence runs out
It is a beautiful fit and it does not quite close.
The language is the problem. If the Roma’s ancestors had spent three centuries in Arabic-speaking Iraq, Romani should carry a thick Arabic layer, the way it carries a thick Greek one. It carries almost none. Some scholars answer that Arabic was the language of rulers and the marsh people went on speaking something else. Others take the silence at face value and say the Roma were never there. The dates pull the same way. The genes allow an exit from India 1,500 years ago, which fits the Zutt. The grammar does not. Romani shares changes with the languages of northern India that happened after about the year 1000, so the linguists say its speakers cannot have left much before then.
So there may have been two migrations out of the same corner of India, four or five centuries apart, one recorded by chroniclers and one by grammar. Or one migration that the record only half describes. Nobody can currently say which. This is where an honest post has to stop. It is worth noticing what the disagreement is made of: three kinds of evidence, ink, DNA and grammar, each measuring a different thing and none of them wrong.
What traveled
What is not in doubt is what the Roma did when they arrived, because Europe wrote it down for six hundred years. They were coppersmiths, blacksmiths, horse dealers, musicians and basket makers. The largest Roma group in Eastern Europe is still called the Kalderash, from a Romanian word for a maker of cauldrons.
The coppersmiths’ trade is a piece of chemistry done in a village square. A copper pot is the best cooking pot there is, because copper carries heat faster than almost any other metal. But bare copper reacts with acid food, with tomatoes, wine, vinegar and fruit, and the green salts it makes are poisonous. The fix is to line the inside with tin, which does not react. Every year or two the tin wears through, and the pot has to be lined again. For centuries, the people who came through the village to do it were Roma.
The trick is in the temperatures. Tin melts at 232 °C. Copper melts at 1,085 °C. An open charcoal fire sits somewhere between 700 and 900. So a tinsmith can heat a copper pan until a stick of tin melts on contact and runs like water. He wipes the molten tin around the inside with a pad of rag or flax, and the pan itself never comes near softening. There are more than 800 degrees between the two metals, and the whole craft lives inside that gap.
The other half of the trick is a pinch of a white salt, sal ammoniac, scattered on the hot copper first. It strips the dull oxide off the surface so the tin wets the metal instead of beading up on it, the same job a plumber’s flux does. Nobody in that village square used the word flux or knew what an oxide was. They knew that without the white powder the tin rolled off, and that was enough to run a trade on for centuries.
Then they decorated it. A plain vessel got its pattern hammered in from the back, which is called repoussé, and sharpened from the front, which is called chasing. A water jug became something a family would pay more for. That is the whole of what this website is about, held in a single copper pot. The science is what makes the thing work. The art is what makes it wanted. The same hands did both.
Try it, with the numbers you already know
- Count to ten out loud in any language you speak at home. Then count in Romani: ekh, duj, trin, štar, pandž, šov, efta, oxto, enja, deš. Mark which ones you recognize. If your home language is Hindi, Punjabi or Urdu you will get six. If it is Greek you will get three, and they are not the same three.
- Find a copper pan, in a kitchen shop or a grandmother’s cupboard, and look inside. If the inside is silver rather than copper-colored, that is tin, and somebody put it there the way described above.
Where this comes from
- Reconstructing the population history of European Romani from genome-wide data — Mendizabal and colleagues, Current Biology 22, 2012
- Reconstructing Roma history from genome-wide data — Moorjani and colleagues, PLoS One 8, 2013
- Early Romani — the ROMANI Project at the University of Manchester, on the Persian, Armenian and Greek layers, with the example words used above
- Romani language — on Vályi, Rüdiger, the Greek numerals, and the linguistic dating of the departure from India
- Zutt — the Arabic chronicles, the marshes, the deportation of 835 and the raid of 855, and de Goeje’s hypothesis
- Hand re-tinning copper cookware — working tinsmiths on the sal ammoniac flux, the rag, and the temperatures
- The retinning process — East Coast Tinning, a shop that still lines copper by hand
This post began with reading Ayesa did on the Roma and the Jats of the Punjab. The two diagrams are original drawings made for it.