Evidence page · comparative
Greek and Sanskrit are cousins with a shared inheritance that is thinner than either tradition admits — and both carry a large non-Indo-European substrate that the classical ideology erased. Europe canonised both, and reproduced the erasure twice.
The inheritance, at its real size
| Domain | Greek | Sanskrit |
|---|---|---|
| Sky-god | Zeus (voc. Zeu pater) | Dyaus (voc. Dyaus pitar) |
| Kinship | patēr, mātēr, phrātēr, thugatēr | pitṛ, mātṛ, bhrātṛ, duhitṛ |
| Numerals | treis, hekaton | trayas, śatam |
| Livestock, wheel | hippos, bous, kuklos | aśva, gau, cakra |
| Verbal system | Augment, reduplicated perfect, middle voice, aspect — deep structural agreement | |
| Poetic formula | kleos aphthiton | śravas akṣitam |
That last row is the strongest single piece of evidence for Indo-European poetics — the same formula, "undying fame," reconstructible in both. The inheritance is real. It is also mostly kinship, numbers, livestock and a sky-god.
Both classical languages are Indo-European skeletons carrying non-Indo-European flesh.
Greek's substrate is Aegean/Anatolian, probably related to Minoan and Eteocretan. Sanskrit's is South Asian — Dravidian, something Munda-adjacent, and probably at least one lost family.
These are not the same substrate, and phonological resemblance does not make them one. Clusters and geminates occur in unrelated families all over the world; Tamil having nt and Greek having -nth- is typology, not relationship. The Dravidian–Mediterranean hypothesis is not supported by regular sound correspondences and should not be leaned on.
What is genuinely parallel is the operation, not the material.
And the test that actually decides it
Sanskrit -nth- is routinely defended as inherited by citing panthā "path" against Latin pons and English find. Those cognates show nt, not nth. They establish the nasal and the stop and say nothing about the aspiration — which is the whole question.
The conclusion survives. The usual argument for it does not. The real test is below, and it is Iranian.
The correct account runs through laryngeals. Indo-Iranian voiceless aspirates — ph, th, kh — arise largely from a plain voiceless stop followed by the laryngeal *h₂.
| Form | Reconstruction | The independent check |
|---|---|---|
| pánthā- “path” | *pent- + laryngeal → Indo-Iranian *panth- | Avestan paθō — θ is the regular Avestan reflex of Indo-Iranian th |
| √manth- “to churn” | *ment-h₂- “to stir” | Lithuanian mentùris “stirrer”; Old Norse mǫndull |
| √grath-/granth- “to knot” | uncertain | No secure IE comparanda. Should not have been offered as an example of inheritance |
Avestan is the load-bearing evidence, not Latin. Because Avestan turns Indo-Iranian th into θ by regular rule, an Avestan θ proves the aspiration existed before Indic and Iranian separated — so it cannot be a South Asian contact effect. That is the argument I should have made.
Laryngeal sources do not account for every Sanskrit voiceless aspirate. Kuiper observed that a striking share of words containing them have no Indo-European etymology at all — and cluster in the same semantic fields as the rest of the substrate vocabulary.
So the position is split, and stating it as one thing was the mistake: some Sanskrit voiceless aspirates are inherited via laryngeals and confirmed by Avestan. Others sit in words with no IE parse and are substrate candidates. The feature does not settle anything by itself; the individual word does.
Retroflexion is not reconstructible for Proto-Indo-European, and Iranian does not have it. It is an Indo-Aryan development. Two mechanisms are documented internally: RUKI (s → ṣ after r, u, k, i) and nati (n → ṇ after r, ṛ, ṣ).
| Position | Argument |
|---|---|
| Contact Emeneau, Kuiper, Southworth | Retroflexion is a defining feature of South Asia as a linguistic area — shared by Indo-Aryan, Dravidian, Munda and Burushaski. Proto-Dravidian had it; Proto-Indo-European did not. And roughly 88 Rigvedic words show retroflexion that RUKI and nati do not explain. |
| Internal Hock, Deshpande | Retroflexion can arise from cluster simplification and RUKI extension without any contact. The conditioning environments were widespread enough to spread the feature natively. |
Genuinely contested, and it should be presented that way. What is not contested: retroflexion is absent from every other Indo-European branch and present across every language family of South Asia. Whatever produced it happened here.
Greek φ θ χ were originally aspirated stops — but they come from PIE voiced aspirates *bʰ *dʰ *gʰ, devoiced. Sanskrit kept those as bh dh gh and built its voiceless aspirates separately.
So Greek and Sanskrit aspirates are not the same phenomenon, and their resemblance in a transcription table is misleading. Same inherited system, opposite outcomes.
Where the parallel between the two substrates actually lies
I dismissed the Greek/Tamil cluster resemblance as typology, and that was right. But I then failed to state the parallel that does hold — which is stronger, and methodological.
The marker is not the cluster. It is a suffix pattern with no Greek morphology behind it, appearing in a giveaway distribution:
Same test, same result. The marker is unconditioned retroflexion and unparseable word shapes, in the same categories:
Substrates announce themselves in flora, fauna, tools and place names — the vocabulary you cannot bring with you.
That is the parallel, and it is a parallel of method rather than of material. In both languages the substrate is identified by a phonological pattern that inherited rules fail to generate, appearing disproportionately in exactly the semantic fields an arriving population would have to borrow. Neither case requires the two substrates to be related, and neither supports it.
Tamil -nt- is native inherited morphology, not a substrate residue — it is a productive Dravidian cluster and carries grammatical function, including the past-tense marker. It appears throughout the core vocabulary, which is the opposite of a substrate distribution.
So the resemblance to Greek -nth- is not merely coincidental — it is structurally a different kind of thing. One is a borrowed suffix in plant names; the other is a grammatical affix in a language's own inflection. That distinction is worth more than the surface similarity, and it holds regardless of which conclusion one would prefer.
| Step | Greek | Sanskrit |
|---|---|---|
| Absorb the local | Pre-Greek deities, place-names, crafts and the sea taken into Greek without acknowledgement | Local deities given Sanskrit parentage — Murugan given Śiva as a father; Koṟṟavai becomes Durgā |
| Borrow the writing | Linear B adapted from Minoan Linear A; after four centuries of illiteracy, the alphabet taken from Phoenician, letter names and order intact | Brāhmī's origins debated; Kharoṣṭhī derives from Aramaic. Sanskrit's own oral canon is fixed long before it is written |
| Borrow the knowledge | Geometry from Egypt, astronomy from Babylon, astrology assembled in conquered Egypt — all credited by Greek sources themselves | Rebirth, karma and liberation from outside the Vedic fold — credited once, at Chāndogya 5.3.7, then absorbed |
| Declare a classical language | Attic Greek canonised; other dialects and the substrate demoted | saṃskṛta — literally "refined, perfected" — against prākṛta, "natural, of the source." The name is the claim |
| Name the excluded | barbaros — one whose speech sounds like bar-bar | mleccha — one who speaks indistinctly |
Both civilisations named outsiders by the sound of their speech. Barbaros and mleccha are independent coinages with an identical logic: the other is defined by being unintelligible. Not by appearance, territory or religion — by failing to speak correctly.
And in both cases the excluded had supplied a large part of the language doing the excluding.
Chronology · Greeks and Indians as subjects of one empire, two centuries before Alexander
This is the part routinely missing from accounts of Greek thought, and it is not speculative. It is Achaemenid administrative history.
| Date | Event |
|---|---|
| c. 740 BCE | Earliest attested Greek alphabetic writing — the Dipylon oinochoe and Nestor's cup. The script is Phoenician, adapted. |
| c. 750–700 | Homer and Hesiod. Greek literature begins essentially at the moment writing arrives — the entire written tradition sits downstream of a Near Eastern import. |
| c. 624–546 | Thales of Miletus. Dies at the moment Persia takes Ionia. He is the one figure who predates the shared state. |
| 547/546 | Cyrus conquers Lydia. The Ionian Greek cities pass under Persian rule. |
| c. 518 | Darius annexes Gandhāra, then Hindush (the Indus). Both are listed as satrapies in the Achaemenid inscriptions. |
| c. 515 | Scylax of Caryanda — a Greek from Caria, in Persian service — is sent by Darius down the Indus to the sea, and reports back. A documented Greek expedition to India two centuries before Alexander. |
| c. 570–475 | Pythagoras, Anaximenes, Heraclitus, Xenophanes. The flowering of Ionian philosophy happens entirely inside the Persian imperial system. |
| 480–479 | Indian troops fight in Xerxes' army in Greece. Herodotus describes them at Plataea — cotton clothing, cane bows, iron-tipped arrows. Greeks and Indians met on Greek soil, in the same war, a generation before Socrates. |
| c. 440s | Herodotus, himself born a Persian subject at Halicarnassus, devotes part of Book 3 to India and records that it paid the largest tribute of any satrapy. |
| 326–324 | Alexander in the Punjab. Pyrrho and the gymnosophists. By this point the two worlds have shared a state for nearly two hundred years. |
Ionia and Gandhāra were provinces of the same empire, with a common language of record, for roughly two centuries.
Arguments about Greek and Indian thought usually treat contact as beginning with Alexander in 326 BCE. It began with Cyrus and Darius, roughly two hundred years earlier, and it was not travel or trade but shared administration — one postal system, one imperial bureaucracy, one currency, troops levied from both ends and marched to the other.
This does not prove that any specific idea moved. It removes the objection that there was no channel. And it means the entire classical period of Greek philosophy — everyone after Thales — happened inside a political structure that also contained the northwest of the Indian subcontinent.
The vowel innovation is routinely credited to the Greeks without qualification.
Reassigning specific Phoenician consonant signs — aleph, he, yod, ayin, waw — to vowels is Greek in execution. But the idea of writing vowels was not new. Linear B, which the Aegean had used for 250 years, was a syllabary: it encoded vowels by definition. Ugaritic had already used three aleph signs distinguishing vowel quality in the 13th century BCE. And the Phrygian alphabet is contemporary with the Greek and also writes vowels — which of the two did it first is genuinely debated.
Accurate version: the specific sign-assignments are Greek or possibly Phrygian; the concept is older and Near Eastern; and the script itself is Phoenician. Adopting a foreign writing system is, in itself, evidence of the direction of cultural traffic at that moment.
Historiography · and what survives of the critique
Greek authors stated their debts. Herodotus says geometry came from Egypt. Thales is said to have learned it there. Pythagoras is placed in Egypt and Babylon. The founding texts of Greek astrology are attributed to Egyptians. The claim of originality was made later, and elsewhere.
Martin Bernal's Black Athena (1987) argued that an "Ancient Model" — in which Greeks acknowledged Egyptian and Phoenician debts — was replaced in nineteenth-century scholarship by an "Aryan Model" for racial reasons. Bernal's specific etymologies and many particulars were heavily criticised, notably by Mary Lefkowitz, and should not be leaned on.
But the substance was established from inside classics, independently. Walter Burkert's The Orientalizing Revolution and M. L. West's The East Face of Helicon document Near Eastern influence on Greek myth, literature and religion in detail, from mainstream philology. The historiographical observation survives even where Bernal's mechanism does not: nineteenth-century scholarship minimised debts the Greeks themselves had recorded.
The construction has a visible seam. The same century that made Athens the origin of Europe also made Sanskrit the proof of a shared Aryan pedigree — and then filed everything distinctively South Asian under decline from it. One operation, two provinces, and Sir William Jones announced the second from Calcutta in 1786.
The same century, the same institutions
Sir William Jones announced the Indo-European relationship in Calcutta in 1786, describing Sanskrit as more perfect than Greek and more copious than Latin. Within a lifetime, comparative philology had built a genuine science on regular sound correspondence — and a mythology on top of it.
In Greece: a Bronze Age palace culture with a four-hundred-year literacy gap and a borrowed alphabet became "the origin of Western civilisation." Its Egyptian and Near Eastern debts, which Greek authors state plainly, were minimised.
In India: Sanskrit's antiquity was used to establish a shared Aryan pedigree — and then everything distinctively South Asian was filed under decline from it. Retroflex phonology, the local deities, the substrate vocabulary, the whole rebirth complex: all read as degeneration rather than as evidence that the inheritance was thin.
The same move served opposite politics. In Europe it produced a self-flattering origin story. In India it was taken up by revivalists who accepted the frame and argued only about who was the truer Aryan. Almost nobody questioned that the Indo-European layer was the important one.
A word with an impeccable Indo-European etymology tells you where the word came from. It tells you nothing about where the thing came from. tapas is Latin tepeō; ascetic heat is not. śraddhā is Latin crēdō; the faculty that carries you to liberation is not.
The same holds in Greek. Inherited vocabulary carrying acquired content leaves no loanword trail — and etymology is the tool the field trusts most, which is precisely why this kind of borrowing has been invisible in both traditions for two centuries.