The Bakhshali manuscript: earliest known dot-symbol for zero, carbon-dated to as early as the 3rd century CE

# The Bakhshali manuscript: earliest known dot-symbol for zero, carbon-dated to as early as the 3rd century CE

In September 2017, the Bodleian Library at Oxford published the results of a radiocarbon dating performed on 70 folios of birch bark in their collection. The folios — a mathematical manuscript discovered in 1881 near the village of Bakhshali in modern Pakistan — contained arithmetic problems with worked solutions in Sanskrit.

What made the result a news story was a single symbol that appears throughout the manuscript: **the dot**. The Bakhshali scribes used it as a placeholder in calculations — the same role that the digit 0 plays in modern arithmetic. The carbon dating placed the earliest folios between **224 and 383 CE**.

That's about **500 years earlier** than the previously-accepted earliest documented use of place-value zero (the 9th-century Gwalior temple inscription).

Hoernle, who first studied the manuscript in 1888, had already noticed the dot's dual role:

> The dot is also used for another purpose, namely as one of the ten > fundamental figures of the decimal system of notation > > — A. F. R. Hoernle, *On the Bakhshali Manuscript* (1888)

That sentence, written in 1888, was a careful inference from the manuscript's content. Hoernle saw the dot acting *as a number* — not just a marker, but a participant in arithmetic. Carbon dating in 2017 confirmed what Hoernle's textual analysis had implied: the Bakhshali scribes were performing place-value arithmetic with zero in the 3rd-4th century CE.

What the dot was doing in the manuscript

Two roles, sometimes confusingly overlapping:

**Role 1 — placeholder for an unknown quantity.** When the manuscript poses a problem like "a man has [unknown] coins; he spends 3, then doubles what's left; he has 8; how many did he start with?", the unknown is written with a dot above a number indicating its position in the calculation. This is the algebraic *x* role.

**Role 2 — the digit zero.** In place-value notation, the dot marks a position with no value. The number 305 was written as three digits with a dot in the tens position: *3 · 5*. This is the *arithmetic 0* role.

Both roles are essential. The first is the conceptual move that makes algebra possible (treat an unknown as a symbol manipulable like a number). The second is the place-value move that makes large-number arithmetic feasible (1,000,000 is just six digits, not a thousand times a thousand).

Modern notation separated the two: we use *x* for unknowns and 0 for the zero digit. The Bakhshali scribes were operating in a system where both roles shared a symbol, but did so unambiguously by context. That's a more compact notation than modern math — at the cost of requiring context-aware reading.

The 2017 dating was a real surprise

Before 2017, the consensus among historians of mathematics was that place-value zero **as a number in arithmetic operation** was first documented in:

- The 9th-century Gwalior inscription (876 CE), which uses the open circle (0) for zero in a date. - 7th-century mathematical texts from China and India (without firm carbon dating).

The Bakhshali manuscript was *known* to be older — Hoernle had argued for a 7th-8th century origin based on textual analysis — but the physical manuscript hadn't been carbon-dated.

When the Bodleian finally did the dating in 2017, the results were unexpected: **three distinct date ranges** for different folios (224-383 CE, 680-779 CE, 885-993 CE). The manuscript is a **composite** — some folios are early, others are later copies or additions. The dot-zero appears on folios from the earliest range, which pulled the documented zero-symbol earliest evidence back by ~500 years.

Some scholars (including Kim Plofker, the modern authority on Indian mathematics) have been careful about how to interpret the result. The earliest folios show the symbol in use; we don't know if those folios were copies of even earlier originals. The 3rd-4th century date is a *lower bound* on the use of the technique.

But "lower bound" of 3rd-4th century is still extraordinary. It places operational place-value zero firmly in late-antique Indian mathematics, contemporary with the late Roman Empire, well before any other tradition documents the same step.

How the dot became the circle

The geographical journey from dot to modern zero:

1. **Indian texts, ~3rd century CE onward.** Dot as zero in arithmetic. 2. **~7th-9th century CE.** Indian astronomers continue using the dot or evolving variants (small circle, partially-filled shapes). 3. **9th century CE.** Arabic mathematicians, working from translated Indian sources, adopt the symbol. al-Khwarizmi's *Algebra* (~830 CE) uses it in the same way. Arabic calls the symbol *ṣifr* — directly transliterating *śūnya*. 4. **12th century CE.** Latin translations of al-Khwarizmi reach Europe. The Arabic *ṣifr* becomes Latin *cifra* and Italian *zefiro*, contracted to *zero*. 5. **13th-17th century.** European mathematicians adopt the open circle (0) for zero. The dot persists in some specialized notations but the circle wins for general arithmetic.

The English word "cipher" and the English word "zero" both descend from the same Indian original via Arabic intermediaries. They split in the early modern era to mean slightly different things ("cipher" came to mean "encoded message," "zero" stayed as the digit) but their etymological history is identical.

What this doesn't prove

Honest qualifications:

**The 2017 dating doesn't prove India invented zero first.** It establishes the earliest *documented* evidence we have. It's possible that other traditions had similar techniques that simply weren't preserved on durable media. The Bakhshali manuscript survives because it was written on birch bark (durable, dry) and was buried/sealed (protected from decay). Most pre-modern manuscripts of comparable age didn't survive.

**It doesn't prove the symbol was invented in the 3rd century.** The folios are dated; the technique they document could be older. The Bakhshali scribes appear to be working in an established arithmetic tradition; the dot-zero is in use, not in development.

**It doesn't say anything about the *concept* of zero.** Zero as a *placeholder symbol* in arithmetic is what the manuscript documents. Zero as a *number with arithmetic rules* (addition, subtraction, multiplication, division) is Brahmagupta's 628 CE contribution — documented in a separate text, ~250 years later.

What the manuscript does prove: **late-antique Indian mathematicians were performing decimal place-value arithmetic with zero, at scale, in worked calculations, in a tradition robust enough to leave a written record we can carbon-date 1,700 years later.**

That's not a small thing. It's the operational floor of all modern arithmetic, in physical form, sitting in a glass case in Oxford's Bodleian Library.

---

Sources

- [On the Bakhshali Manuscript, A. F. R. Hoernle, 1888](https://archive.org/details/onbakshalimanusc00hoeruoft) — Hoernle's introductory chapter, cited above. - Bodleian Libraries (2017). "Carbon dating finds Bakhshali manuscript contains oldest recorded origins of the symbol 'zero'." Press release, 14 September 2017. - Plofker, K. (2009). *Mathematics in India*. Princeton University Press, ch. 4 — pre-Bakhshali-dating discussion of Indian place- value notation. - Hayashi, T. (1995). *The Bakhshali Manuscript: An Ancient Indian Mathematical Treatise.* Egbert Forsten. — modern technical edition of the manuscript.

Related claims

- [Brahmagupta's arithmetic of zero (628 CE)](/c/d10352e6-dc8c-58ee-bbbf-9b1489efdc9e) — same tradition, ~250 years after the Bakhshali manuscript; formalizes zero as a number with arithmetic rules rather than just a place-value symbol. - [Aryabhata's pi approximation](/c/0b862684-d325-5002-b054-169bd2253ef9) — same Indian mathematical tradition, working with mature decimal arithmetic that the Bakhshali manuscript shows in calculation form.