The Dot That Counted: Dravidian Script Architecture and the Origins of Zero
ETTUGE RESEARCH ESSAY
·
APRIL 2026
·
SPECULATIVE
Linking the puḷḷi (virama dot), śūnya etymology, and the K-127 Cambodia inscription
Abstract
The conventional narrative attributes the mathematical zero to Sanskrit philosophical vocabulary (śūnya — emptiness) as elaborated through Buddhist śūnyatā doctrine and formalised arithmetically by Brahmagupta (628 CE). This essay argues that the conceptual and graphic substrate for zero may be substantially older and Dravidian in origin: the puḷḷi (Tamil virama dot, attested from the 2nd century CE), the possible Dravidian etymology of śūnya itself, and the divergent treatment of positional zero between Tamil and Kannada numeral traditions all point toward a richer, pre-Buddhist infrastructure for the concept of marked absence.
The essay incorporates new epigraphic evidence from the Chandravalli inscription (345–365 CE) showing a puḷḷi-variety virama in early Kadamba script, examines the Ciaruteun inscription (c. 450 CE) as evidence for virama in Southeast Asia before K-127, and analyses the Khmer inscription K-127 (683 CE) through the lens of the Pallava-Kadamba script controversy. A central argument is that the divergence between Tamil and Kannada in their treatment of the explicit absence marker is rooted in phonological divergence — specifically Tamil's preservation of word-final bare consonants versus Kannada's euphonic -u epenthesis — and that this phonological divergence maps directly onto the differential adoption of positional zero.
§ 1
Introduction: The Standard Story and Its Gaps
The story of zero as commonly told runs: emptiness as Buddhist philosophy (śūnyatā, c. 3rd century BCE) → placeholder dot in calculation (śūnya-bindu, Bakhshali manuscript, 3rd–4th century CE) → arithmetic zero with defined rules (Brahmagupta, Brahmasphutasiddhānta, 628 CE) → stone inscription zero (Gwalior temple, 876 CE).
Amir Aczel's Finding Zero (2015) added a crucial chapter: the earlier Khmer inscription K-127 (683 CE), bearing a zero in positional notation predating the Gwalior zero by nearly two centuries. Aczel's quest — recovering the stone from Phnom Penh — dramatised the question of where the zero symbol originated before reaching Cambodia.
Three neglected questions
Does śūnya have a plausible Dravidian etymology that predates its Sanskrit philosophical elaboration?
Was the puḷḷi (virama dot) — an earlier, independent, visible mark for "absence in a structured positional slot" — a cognitive and graphic precursor to the zero symbol?
Does the divergence between Tamil and Kannada in their treatment of the explicit absence marker reflect a deeper structural difference rooted in phonological divergence — specifically Tamil's preservation of word-final bare consonants versus Kannada's euphonic -u epenthesis — and does this map directly onto the differential adoption of positional zero?
§ 2
The Etymology of Śūnya: Sanskrit or Dravidian?
2.1 The standard internal Sanskrit derivation
Sanskrit śūnya is conventionally derived from the root śvi / śū- (to swell, to be hollow), giving śū-na (hollow, swollen) → śūnya (empty, void). The proposed Proto-Indo-European cognate *ḱeu- (to swell) links to Greek κύω and English swollen — but this connection is weak and contested. Mayrhofer's authoritative Etymologisches Wörterbuch des Altindoarischen (1992–2001) flags words without clean Indo-European cognates as candidates for Dravidian substrate influence.
Criterion:Śūnya has no clean IE cognate family. No Greek, Latin, Avestan, or Germanic word clearly reflects *śū-na in the sense of "empty/void." This is precisely the criterion Caldwell (1856), Kittel (1873), and Emeneau (DEDR, 1984) use to identify Dravidian loans into Sanskrit.
2.2 The Dravidian cluster: suḻi, suḻiyam, sonne
Language
Word
Meaning
Note
Tamil
suḻi
whirl, curl, vortex, spiral
concrete, physical shape
Tamil
suḻiyam
zero (the numeral ௦)
named from the shape of the symbol — a curl
Kannada
ಸೊನ್ನೆsonne
zero (numeral only)
not used for existential nothingness
Telugu
sūnnu
zero
phonologically close to śūnya
The Tamil suḻiyam is remarkable: it names zero from the shape of the symbol (a curl/whirl) rather than from philosophical emptiness. This is a concrete, visual, physically motivated etymology — exactly what you would expect from a root that predates philosophical abstraction.
The phonological path suḻi → śūnya would require: s → ś (palatalisation, common in Sanskrit contact); uḻ → ū (loss of retroflex lateral, vowel lengthening); -i → -ya (Sanskrit nominalising suffix applied to borrowed root). This is entirely within the documented patterns of Dravidian-to-Sanskrit phonological adaptation.
Colloquial Kannada uses ēnilla (ಏನಿಲ್ಲ — "what-is-not") for everyday nothingness, while sonne is reserved exclusively for the numeral. Tamil likewise uses illai for existential negation and suḻiyam for the mathematical concept. This register separation — native Dravidian for "nothing exists," Sanskrit-derived or shape-coined for "the numeral" — suggests the mathematical zero arrived as a cultural import while the concept of absence was already fully native.
2.3 Chronological layering of śūnya
Phase
Source
Usage of śūnya
~1200–900 BCE
Atharvaveda
Concrete, mundane — "empty/deserted place, abandoned house." No philosophical weight.
~800–600 BCE
Early Upaniṣads
Pūrṇa/śūnya polarity crystallises. Bṛhadāraṇyaka 5.1.1 (~700 BCE): the mathematical content implies zero as identity element, though not yet symbolised.
~5th–3rd c. BCE
Pāli texts, Theravāda
Suññatā becomes a technical term. Nāgārjuna's Mādhyamaka (~2nd c. CE) develops śūnyatā into a complete philosophical system.
This layering is consistent with a word that entered Sanskrit from a Dravidian substrate at a mundane, concrete level, was then philosophically elaborated, and finally mathematised.
§ 3
The Puḷḷi Hypothesis: Absence as Scribal Architecture
3.1 The Tamil puḷḷi
The Tamil puḷḷi (புள்ளி, virama dot) is attested in inscriptions from the 2nd century CE (Mahadevan 2003). It marks "no vowel follows this consonant" — a visible dot encoding positional absence within a structured system.
Tamil grammar explicitly theorises the puḷḷi from the earliest period. The Tolkāppiyam treats the bare consonant (with puḷḷi) as a first-class grammatical entity — mey (dead body/consonant) — philosophically distinguished from uyir (soul/vowel). The Piṅkalantai nikaNTu (9th century CE) is the oldest Indian text to use the term virāma specifically for the "dead consonant with puḷḷi."
The word puḷḷi derives from pul- (small, insignificant, minimal) — cognate with Kannada hullu (ಹುಲ್ಲು, grass, low), showing the standard Kannada p- → h- shift. The puḷḷi is the smallest meaningful mark. This is also what cukke (ಚುಕ್ಕೆ) means in Kannada: both "dot/point" and "star" — the minimum perceptible mark.
3.2 The Chandravalli inscription: puḷḷi-variety in early Kadamba (345–365 CE)
The Chandravalli inscription of Kadamba Mayura Sharma (345–365 CE), written in Maharashtri Prakrit in an early Kadamba script stage close to Late Satavahana style, shows two distinct kinds of dots in its text (analysis: @pfkodial):
Dots above letters — anusvāra, marking nasalisation (equivalent to modern ಂ)
Dots to the side of letters — a puḷḷi-variety virama, marking vowel suppression
Critical discovery
The earliest Kadamba inscription already uses an explicit visible dot for both anusvāra and virama functions — the same graphic logic as Tamil-Brāhmī. The explicit dot virama was the original Southern Brāhmī default, not a Tamil-specific innovation. Tamil is the conservative branch that preserved this default; the Kannada-Telugu branch replaced it with conjunct stacking.
The Simhakatanjana inscription at Talagunda (conventionally dated 370 CE but argued on paleographic grounds to be considerably later — the letterforms show developed Early Chalukya features that cannot have evolved from Chandravalli's Satavahana-stage forms in only 15–20 years) confirms that the script evolution from early Kadamba to Chalukya-stage took substantially longer than conventional dating implies.
3.3 The anusvāra: a parallel dot story
In the earliest Kadamba layer, the anusvāra dot marked specifically the nasalised or half-m of Maharashtri Prakrit sandhi: word-final -m before a consonant, prefixes like saṃ-, anuṃ-. This is pre-consonantal nasal assimilation — the phonological phenomenon where -m of saṃskṛta nasalises to ṅ before k, to n before t, etc. The extension to represent all homorganic nasals before stops came later.
This gives us two dots in early Kadamba doing related but distinct work:
Dot
Position
Function
Anusvāra
Above the letter
Nasalisation — phonological absence of full consonant articulation
Virama
To the side of letter
Vowel suppression — phonological absence of inherent vowel
Both mark a kind of phonological absence in a structured system. Both are, in the deepest sense, absence markers.
3.4 The structural isomorphism with zero
System
Mark
Meaning
Form
Abugida script
puḷḷi / virama dot
"Absence of vowel in this phonological slot"
·
Positional numeral
zero placeholder
"Absence of quantity in this positional slot"
০ ० ೦
Both encode "nothing here" as a first-class visible entity rather than leaving the slot blank. This is the key cognitive innovation — not the philosophical concept of emptiness (which many cultures had), but the scribal discipline of marking absence explicitly as a form of presence.
The Bakhshali manuscript zero (3rd–4th century CE) is a dot — graphically identical to the Southern Brāhmī virama dot tradition already operative in Kadamba and Tamil-Brāhmī by the 2nd–4th century CE.
§ 4
Phonological Divergence: Why Tamil Kept the Dot and Kannada Moved On
4.1 Tamil phonotactics and the puḷḷi solution
Tamil phonotactics severely restricts consonant clusters. Native Tamil words do not begin with clusters, and internal clusters are tightly constrained. The complex Sanskrit consonant clusters (śrī, kṣa, pra-, tra-, jña-) were alien to Tamil phonology. Rather than developing elaborate stacked conjunct forms (ottakṣara) to represent them, Tamil chose the minimal solution: the explicit puḷḷi on each consonant individually, marking "no vowel here." This left letterforms unchanged and avoided creating hundreds of new composite glyphs.
Kannada and Telugu carried heavier Sanskrit loanword loads requiring dense clusters. The ottakṣara (subjoined/stacked conjunct) system encoded these as single orthographic units, making the virama conceptually invisible — absorbed into the conjunct.
4.2 Word-final consonants: the original Dravidian state
Proto-Dravidian words ended in consonants freely. Both Tamil and Old Kannada inscriptions show word-final bare consonants. Both languages needed a mechanism for "consonant at word end, no following vowel." They chose differently:
Tamil: explicit puḷḷi on the word-final consonant — the consonant stands bare with its dot, a genuine phonological reality
Old Kannada: initially explicit dot virama (Chandravalli), then progressively shifted to the euphonic -u strategy
4.3 The euphonic -u: Kannada's phonological solution
Modern Kannada resolved most word-final bare consonants by adding a euphonic (epenthetic) -u. Sanskrit rājan → Kannada rājanu; native forms show -u where Proto-Dravidian had a bare consonant. This was a genuine phonological change — words ended in -u in spoken Kannada — not merely orthographic.
Tamil did not undergo this euphonic -u shift. Tamil preserved the word-final bare consonant as a genuine phonological reality. The puḷḷi therefore remained a live, productive, everyday element of Tamil script rather than a learned or marginal mark.
The zero consequence
A tradition in which the explicit absence-marker is daily and ordinary has a lower cognitive barrier to adopting a positional zero (another explicit absence-marker). A tradition in which the absence-marker has been progressively hidden — inside conjuncts, subsumed by euphonic vowels — is arriving at the explicit zero from a different starting point.
4.4 The virāma cycle in Kannada
The Kannada virāma history follows a three-stage cycle:
Stage
Period
Virama handling
Early Kadamba
345 CE+
Explicit side-dot — original Southern Brāhmī default
Explicit virāma (halant ್ in Kannada) re-emerges as standardised diacritic for Sanskrit loanwords
This three-stage cycle — explicit → hidden → explicit re-emerges — is paralleled in the numeral system: additive Brāhmī numerals → transitional period → positional system with explicit zero. Both the script virama and the numeral zero are re-explicitisations of absence that arrive together in the 9th–10th century CE.
Terminology note: In Kannada, the sign ್ is called halant (ಹಲಂತ), not virāma. The term halanta derives from Pāṇini's grammatical vocabulary (hal = consonant, anta = end). Tamil uses puḷḷi; Malayalam uses candrakkalā; Hindi uses halant; Bengali uses hôshonto. This article uses virāma as the generic term and halant for the Kannada-specific sign.
§ 5
The Cambodia Connection: K-127 and the Pallava-Kadamba Debate
5.1 K-127 and its zero
Khmer inscription K-127, dated 683 CE (Śaka year 605), discovered at Sambor on the Mekong River, contains the earliest dated zero in stone. The numeral 605 is written with a positional zero in the hundreds place — unambiguous place-value notation. This antedates the Gwalior temple zero (876 CE) by nearly two centuries.
Inscription
Date
Location
Significance
Gudnapur inscription
~500 CE
Karnataka, Kadamba
First complete Kannada numeral set (1–9)
K-127, Sambor
683 CE
Cambodia (Mekong)
Earliest dated positional zero in stone
Gwalior temple
876 CE
Madhya Pradesh
First Indian stone inscription with positional zero
5.2 Virama in Southeast Asia before K-127: the Ciaruteun inscription
The Ciaruteun inscription from West Java, Indonesia (King Purnavarman of Tarumanagara, c. 450 CE), written in Pallava/Vengi script and Sanskrit, is one of the earliest Southeast Asian inscriptions with fully developed Sanskrit orthography — including conjunct consonants. Sanskrit requires extensive conjuncts (śrī, Viṣṇu-, Purnnavarman-), and their presence in a 5th-century Indonesian inscription confirms that the virama-conjunct architecture of Southern Brāhmī was fully operational in Southeast Asia at least two centuries before K-127.
The infrastructure argument
The zero did not arrive in a script-vacuum. The infrastructure for positional absence-marking (conjunct virama, ottakṣara-style stacking) was already present in Southeast Asia by 450 CE. K-127's positional zero is an extension of an already-embedded scribal architecture — not a sudden import.
5.3 The script attribution controversy
The script of K-127 is conventionally attributed to "Pallava Grantha." However, the attribution is contested on several grounds:
Epigrapher Arlo Griffiths prefers the term Late Southern Brāhmī for the cluster of related scripts, calling "Pallava" misleading.
The earliest Pallava inscriptions (Mayidavolu copper plates, 4th century CE) use Kadamba script. Grantha and Vatteluthu evolved from Kadamba, not independently.
The Śaka era used in K-127's dateline was introduced by the Chalukyas of Badami (Karnataka). The Pallavas never used the Śaka era — this is the strongest single marker of Karnataka-Deccan vs. Tamil Nadu origin.
The "box-headed alphabet" in inscriptions of King Bhadravarman of Champa (central Vietnam) is characteristic of the central and upper Deccan / Kannada-speaking areas, not Pallava territory.
A direct dynastic connection: King Kadaveśa Harivarma of Kambujadeśa (~730 CE) — the title Kadaveśa itself signals the Kadava/Kadamba connection — ruled Cambodia before Pallavamalla (Nandivarman II) came from Cambodia to claim the Pallava throne.
The ottakṣara connection: The Khmer script's conjunct consonant handling through subjoined forms — an architectural feature visible in modern Khmer fonts — is structurally identical to the Kadamba-Grantha ottakṣara tradition, not to the Tamil explicit-puḷḷi approach. This is the most direct paleographic argument for the Kadamba-Grantha lineage of K-127's script.
Tamil kings, not Tamil script
Cambodia was substantially influenced by Tamil-speaking rulers and merchants. But what was transmitted was the Grantha/Kadamba tradition for writing Sanskrit — not the Tamil vernacular script. Tamil kings carried a Sanskrit-writing system to Southeast Asia. K-127 is written in Sanskrit in a script with ottakṣara-style conjuncts; its zero belongs to the Karnataka-Deccan scribal tradition.
§ 6
Synthesis: A Dravidian Archaeology of Zero
Assembling the evidence, we can sketch a speculative but coherent genealogy:
~1200 BCE – 500 BCE · Conceptual substrate
The word enters Sanskrit
Śūnya enters Sanskrit from a Dravidian *suḻ-/sul- root (circular, hollow, spiral) at the mundane level in the Atharvaveda period. The pūrṇa/śūnya polarity crystallises in the Bṛhadāraṇyaka Upaniṣad (~700 BCE), providing the conceptual infrastructure for zero a millennium before its mathematical formalisation.
~2nd c. CE – 365 CE · Scribal infrastructure
Two dots in Southern Brāhmī: anusvāra and virama
Tamil-Brāhmī preserves the explicit dot virama (puḷḷi) from the 2nd century CE. The Chandravalli inscription (345–365 CE) confirms the same two-dot architecture — anusvāra above, virama-variety to the side — in the earliest Kadamba layer. The explicit dot for positional absence was the Southern Brāhmī default across the entire script family.
~300 CE – 628 CE · Mathematical transition
Dot migrates from phonology to arithmetic
The dot appears in the Bakhshali manuscript as śūnya-sthāna ("empty place"), functioning as a placeholder. Brahmagupta (628 CE) formalises arithmetic rules for zero but uses the word śūnya, not yet a hollow-circle symbol. The zero symbol — hollow circle, suḻi shape — evolves from the dot: dot → hollow dot → circle.
~450 CE – 683 CE · Southeast Asia
Ciaruteun to K-127: virama architecture precedes zero
The Ciaruteun inscription (c. 450 CE) establishes that the Southern Brāhmī conjunct tradition was already operational in Island Southeast Asia two centuries before K-127. K-127 (683 CE) uses positional zero in the Kadamba-Grantha-Chalukya scribal tradition, confirmed by the Śaka era dateline. The zero was already conventional — transmitted by Brahmin scholars in the Karnataka-Deccan network.
~9th–10th century CE · Tamil/Kannada divergence
Kannada adopts zero; Tamil retains additive system
Tamil retains its additive numeral system — because the puḷḷi was already live, daily, and phonologically motivated by real word-final bare consonants; the explicit absence-marker needed no reinforcement. Kannada adopts positional zero contemporaneously with the explicit halant (್) being standardised. Both restore explicit absence-marking to a tradition that had hidden it. The phonological cause (euphonic -u eliminating word-final bare consonants) and the mathematical effect (zero adoption) are two facets of the same Kannada drift away from explicit absence-marking — and back toward it, differently.
The verse asserts that taking the full from the full leaves the full unchanged. This is structurally identical to the statement that ∞ − ∞ = ∞ — the infinite as an identity element under subtraction of itself.
What the verse does not say is: "the empty from the full leaves the full." But it implies it: if pūrṇa minus pūrṇa equals pūrṇa, then zero is the only number consistent with this arithmetic. The pūrṇa/śūnya polarity was the conceptual container that eventually had to be formalised arithmetically — and when it was, it produced the Hindu-Arabic numeral system that the world uses today.
The speculative conclusion
If śūnya is of Dravidian origin — from *suḻ-, the hollow at the centre of the vortex — then the mathematical zero is not only Indian in origin but specifically Dravidian in its conceptual seed, philosophically elaborated by Buddhism and Sanskrit metaphysics, graphically grounded in the original Southern Brāhmī dot, and mathematically formalised in the Karnataka-Deccan scribal tradition that transmitted it to Cambodia.
§ 8
Open Questions and Further Research
This essay has been speculative throughout. The following questions require philological and paleographic investigation before any of the hypotheses above can be treated as established.
DEDR entry for pul-/puḷ-: Does Burrow & Emeneau's Dravidian Etymological Dictionary contain entries that link puḷḷi (dot), puḷi (tiger, spotted), and pul (insignificant/small) to a reconstructed Proto-Dravidian root? The semantic coherence (minimal mark = spot on an animal = smallest unit) needs philological confirmation.
Śūnya in Mayrhofer: Does Mayrhofer's EWAia entry for śūnya flag the absence of IE cognates and mention any Dravidian parallel? If so, this would be the authoritative acknowledgement the hypothesis needs.
K-127 numeral paleography: A detailed comparative analysis of K-127's zero glyph against Kadamba (Gudnapur, ~500 CE), Grantha, and Tamil-Brāhmī numeral forms would clarify the script lineage of the world's oldest dated zero.
Chandravalli dot survey: Full paleographic documentation of all side-dot virama instances in Chandravalli, with comparison to Tamil-Brāhmī puḷḷi positioning, would establish whether these are independent parallel developments or a single shared Southern Brāhmī practice.
Anusvāra extension timeline: When did the anusvāra in Kadamba tradition extend from -m specifically to all homorganic nasals? Did Tamil-Brāhmī influence this, or was it internal?
Euphonic -u dating: When did the euphonic -u epenthesis become systematic in spoken Kannada? If this correlates with the fading of the explicit virama in inscriptions, it provides direct phonological evidence for the script-change mechanism proposed in §4.
Tamil astronomical texts: Did Tamil astronomers (post-Āryabhaṭīya, 499 CE) ever use the puḷḷi in numerical contexts as a zero placeholder? The answer would clarify whether the dot was ever consciously identified with zero in Tamil scribal practice.
Ciaruteun inscription (~450 CE) — King Purnavarman, Tarumanagara, West Java. Pallava/Vengi script, Sanskrit. Earliest SE Asian inscription with developed Sanskrit conjunct orthography.
K-127 (Khmer inscription, 683 CE) — Sambor, Cambodia. Earliest dated zero in stone. Published in Corpus des inscriptions du Cambodge (Cœdès & Damais). See also Aczel (2015).
Pillai, Karan Damodaram (2023). "The Hybrid Origin of Brāhmī Script from Aramaic, Phoenician and Greek Letters." Indialogs 10. DOI: 10.5565/rev/indialogs.213
Salomon, Richard (1998). Indian Epigraphy. Oxford University Press.
Griffiths, Arlo — advocates "Late Southern Brāhmī" over "Pallava Grantha" as script designation.
Burnell, A.C. (1874). Elements of South Indian Palaeography. London.
Sarkar, Himansu Bhanu (1969). "The Pallava Script and Southeast Asia." Bijdragen tot de Taal-, Land- en Volkenkunde 125.
Wisseman Christie, Jan (1998). "The Diffusion of Hinduism and Buddhism to Southeast Asia." Journal of Southeast Asian Studies.
Virāma vs. halant:Virāma (Sanskrit: "cessation") is the generic pan-Indic term. Language-specific terms: Tamil puḷḷi, Malayalam candrakkalā, Kannada halant (ಹಲಂತ), Hindi halant, Bengali hôshonto. The term halanta derives from Pāṇini's grammatical vocabulary (hal = consonant, anta = end), describing consonant-final syllables — secondarily applied to the orthographic diacritic. In Kannada specifically the diacritic ್ is called halant, not virāma. This article uses virāma as the generic term.
Eke romanisation (used throughout ettuge): capital letters mark long vowels (A=ā, I=ī, U=ū, E=ē, O=ō); uppercase retroflex consonants (T, D, N, L); S = palatal sibilant. See Eke reference.
Tadbhava / tatsama: A tatsama is a Sanskrit word used unchanged in a Dravidian language; a tadbhava is phonologically adapted. The hypothesis here inverts the usual assumption: sonne (Kannada) and suḻiyam (Tamil) would not be tadbhavas of Sanskrit śūnya but cognates of a common pre-Sanskrit root.
Ottakṣara: Subjoined conjunct consonant form used in Grantha, Kannada, Telugu, and Khmer. Encodes a consonant cluster as a single composite glyph, hiding the virama inside. Architecturally distinct from Tamil's explicit puḷḷi approach, which leaves each consonant visually distinct and marks absence of vowel explicitly.