Evidence page · the western attestation
The oldest securely dated attestation of Indo-Aryan is not in India. It is in cuneiform, in a treaty between a Hittite king and a Mitanni king, around 1380 BCE — and it names four Rigvedic gods in the correct grouping.
| Source | Content |
|---|---|
| Šuppiluliuma–Šattiwaza treaty c. 1380 BCE, Hittite archive at Boğazköy |
Among the divine witnesses: Mi-it-ra-, Aru-na-, In-da-ra, Na-ša-at-ti-ya — Mitra, Varuṇa, Indra, and the Nāsatyas. These four appear as a set, which is how the Rigveda groups them |
| The Kikkuli text Hittite horse-training manual |
Numerals for training circuits: aika-, tera-, panza-, satta-, na- — one, three, five, seven, nine. Aika is decisive: Sanskrit eka, against Iranian aiva |
| Mitanni royal names | Tušratta, Artatama, Šuttarna, Šauššatar, Šattiwaza — throne names with Indo-Aryan elements, borne by rulers of a Hurrian-speaking state |
| Technical vocabulary | marya-nnu — the chariot-warrior elite, from Indo-Aryan marya, "young man." Horse colour terms including babru, parita, pinkara — Sanskrit babhru, palita, piṅgala |
Indo-Iranian split into an Indic branch and an Iranian branch, and the numeral "one" is one of the places the split shows. Sanskrit has eka; Avestan has aēva.
Kikkuli's word is aika. That is the Indic form, not the Iranian one. So this is not undifferentiated Indo-Iranian in Syria in the fourteenth century BCE. It is already the Indic side of the split.
The branch had separated before 1400 BCE, and one part of it was in northern Mesopotamia.
For a recognisably Indic language to be supplying royal names, technical vocabulary and a divine formula in Syria around 1400 BCE, Indo-Aryan must have differentiated from Iranian appreciably earlier and travelled a long way.
That is the constraint. It is not a date for the Rigveda.
Two things it does not establish:
1 · It does not show the Rigveda is older than conventionally dated. A language can be centuries old before a particular text is composed in it. The Mitanni material dates the language branch, not the corpus.
2 · It does not settle direction of movement. A group speaking Indo-Aryan in Syria is consistent with a westward branch from a shared homeland, and consistent with an eastward one. The attestation is a point on a map, not an arrow.
What it does do is put a floor under the differentiation date, and that floor is earlier than the conventional Rigvedic dating comfortably assumes. The tension between the two is real and is not resolved by preferring whichever suits an argument.
The Mitanni horse-colour vocabulary includes babru, parita and pinkara — Sanskrit babhru (brown), palita (grey), piṅgala (reddish-brown).
These are horse-colour terms in a horse-training manual. That is exactly the distribution the Rigveda shows: babhru occurs in 24 stanzas, applied to horses, to soma juice, to cattle — and to Rudra, who is its strongest divine association.
So the colour vocabulary travelled with the horses and the chariot technology. It is a technical register, and its appearance in Syria alongside marya and the training circuits shows what the package consisted of: horses, chariots, a warrior class, a colour terminology for the animals, and four gods.
Mitanni was a Hurrian-speaking state. Hurrian belongs to the Hurro-Urartian family, which has no established relationship to Indo-European or to any other family. The administration, the correspondence and the population were Hurrian.
What is Indo-Aryan is a thin, specific layer: throne names, the chariot-warrior term, horse-training vocabulary, and a divine formula in a treaty. Nothing suggests the population spoke Indo-Aryan.
A ruling stratum using a prestige language for names, gods and horses, over a population speaking something unrelated — the Indo-Aryan element in Mitanni is an elite register, not a demography.
That is directly relevant to how the same evidence is read for South Asia. Mitanni demonstrates that Indo-Aryan names, gods and technical vocabulary can appear in a society whose population is not Indo-Aryan-speaking — with no mass migration required, and with the substrate language continuing underneath.
It is a documented case of exactly the mechanism that is contested for India, and it happens to be the only case where both the elite register and the underlying language are attested in writing at the same time.