Bīja-Gaṇita: The Astronomical Seeds of Indian Algebra

A Question Nobody Thinks to Ask

Every student who has ever groaned through a chapter on "solving for $x$" has, without knowing it, brushed up against a fifteen-hundred-year-old Indian idea. In Hindi, Bengali, and most other Indian languages, the school subject you call algebra is called बीज-गणित: bīja-gaṇita. Break the word in two and something odd happens: gaṇita simply means "mathematics" or "calculation," the part you'd expect. But bīja means seed, as in the seed of a mango, or the seed you plant in a garden.

Why would an entire branch of mathematics, the one full of $x$'s and $y$'s, be named after something that grows in soil?

The answer runs through ancient astronomers predicting eclipses, a thousand years of "software updates" to astronomical formulas, and one textbook in which the origins of algebra turn out to be hiding inside a chapter on astronomy.

First, What Even Is Algebra?

A one-sentence refresher, since most of us stop using the word the moment we leave school. Arithmetic is mathematics with numbers you already know: $2 + 3 = 5$. Algebra is mathematics with a number you don't know yet, which you give a name (usually $x$) and then hunt down using logical rules, the way a detective identifies a suspect from clues. "I am thinking of a number; double it and add three, and you get eleven" becomes $2x + 3 = 11$, and algebra walks you, step by step, to $x = 4$.

$x$ is a value that is hidden: present in the problem, doing work, but not yet visible. That idea, a hidden quantity waiting to be revealed, is where our story about seeds begins.

Key idea

Ancient Indian mathematicians had a specific word for this hidden, not-yet-known quantity: avyakta (अव्यक्त), meaning "unmanifest" or "not yet made visible." Its opposite, vyakta (व्यक्त), meant "manifest" or "visible": the ordinary numbers of everyday arithmetic.

Two Names for Two Kinds of Mathematics

Indian mathematical tradition drew a clean line between two departments of mathematics, and gave each one its own name:

• Pāṭī-gaṇita / Vyakta-gaṇita: "mathematics of procedures" or "mathematics of the visible." This is arithmetic: adding, multiplying, land measurements, interest calculations, numbers you can already see and touch.
• Bīja-gaṇita / Avyakta-gaṇita: "mathematics of the seed" or "mathematics of the unmanifest." This is algebra: working with an unknown quantity before it has revealed its value.

The most famous book carrying the name is the Bījagaṇita of Bhāskara II, written in 1150 CE, one of the first books in the world devoted entirely to algebra as its own subject, five centuries before Europe treated algebra the same way. We meet Bhāskara again at the end of this article, since he is the one who finally puts into words why the seed metaphor fits. But to see where the metaphor came from, we first have to go five hundred years earlier, and up into the sky.

The Sky Was Ancient India's Supercomputer Problem

Long before anyone wrote a textbook called Bījagaṇita, Indian scholars were wrestling with a more urgent computational problem: predicting where the Sun, Moon, and planets would be on any given night, years or decades in advance. This mattered for calendars, agriculture, festivals timed to lunar phases, and for predicting eclipses.

The texts that did this job were called Siddhāntas (सिद्धान्त), literally "established conclusions." The most famous, the Sūrya Siddhānta ("Doctrine of the Sun"), is a manual in verse for computing celestial positions using cycles spanning millions of years.

Here's the catch: algebraic notation as we know it, the $x$'s and the equals signs, did not exist yet. There was no symbol for "multiply," no symbol for "the unknown." Yet the calculations these texts describe are unmistakably algebraic: take this quantity, multiply it by that constant, divide by this other number, add a correction, and out comes a planet's position. So how do you write an equation with no symbols? You write a poem.

What the Ancients Were Actually Tracking

Before we look at the verse that predicts a planet's exact position, it helps to know what "position" even meant to these astronomers, because the framework they used is not common knowledge today.

Watch the sky over a year and the Sun appears to trace a single path against the background stars, called the ecliptic. The Moon and the visible planets all wander close to this same narrow band. Ancient astronomers, needing a way to state where along this band an object was, divided the ecliptic into twelve equal 30° segments, called rāśis (zodiac signs). This is nothing more exotic than time zones on Earth: an arbitrary, agreed-upon way of turning a circle into numbered addresses, so that "the Moon is at 15° into the third sign" means exactly one thing to anyone doing the calculation.

A planet's position is one number, its longitude, measured around this 360° circle, divided into twelve 30° signs.

Once position is just a number on a 360° dial, a second idea becomes available: a planet's mean position is where it would be if it crawled around this circle at a perfectly constant speed, like a clock hand: trivial to calculate, since it's just (elapsed time) $\times$ (average daily motion). A planet's true position is where it actually appears against the stars tonight. These two numbers are almost never the same, because real planets speed up and slow down over the course of their orbit. The gap between mean and true position is exactly the correction the next section is about.

A Planet That Won't Keep a Steady Pace

The mean position is easy arithmetic, but it rests on a convenient fiction: that a planet crawls around its orbit at one unchanging speed, like a clock hand. Real planets don't, because a real orbit is an ellipse, not a circle, with Earth (approximately) near one focus rather than at the centre. A planet moves faster when it is closer to Earth and slower when it is farther away, so the same number of days always produces the same change in mean position, but not always the same change in true position.

An ellipse, with Earth near one focus rather than at the centre. The same stretch of time covers more ground near the close point of the orbit than near the far point. A mean position calculated at constant speed drifts away from the true position as a result.

Ancient Indian astronomers had observed the following pattern: a planet runs ahead of its mean position for part of its orbit and falls behind for the rest. Their fix was a correction, added or subtracted from the mean position, called the manda-phala (मन्दफल), the "slow-[planet's] result." It is computed from a small circle, the manda epicycle, whose geometry stands in for the same close-fast, far-slow pattern shown above.

Here is the verse that computes it: Sūrya Siddhānta, Chapter 2 (Spaṣṭādhikāra, "on true positions"), verse 39:

Sūrya Siddhānta 2.39

तद्गुणे भुजकोटिज्ये भगणांशविभाजिते।
तद्भुज्याफलं धनुर्मान्दं लिप्तादिकं फलम् ॥३९॥

Transliteration: tadguṇe bhujakoṭijye bhagaṇāṃśavibhājite, tadbhujyāphalaṃ dhanurmāndaṃ liptādikaṃ phalam.

Translation (with the commentary Gūḍhārthaprakāśikā's gloss folded in): "The bhuja-jyā and koṭi-jyā [the Rsines of the planet's anomaly from its epicycle's apex] are multiplied by that (the planet's own true epicycle-circumference) and divided by the degrees in a full circle. The result is the bhuja-phala and koṭi-phala. The arc produced from the bhuja-jyā of the manda anomaly is the manda-phala."

The commentary's own reasoning (upapatti), given right after the verse, is worth noting: the true divisor ought geometrically to be the manda-karṇa (the hypotenuse of the planet's epicycle triangle), but since that is always close to trijya (3438), the commentary explicitly allows substituting trijya instead, "because of its near-equality to trijya, admitted with only a small error." That is a fourteen-hundred-year-old author flagging his own first-order approximation.

To see exactly what the verse is computing, it helps to set it out the way a geometry textbook would: with labelled points, and one construction step at a time. First, one piece of vocabulary the verse leans on.

What is an Rsine?

Every Indian astronomical text works inside a single fixed circle: centre $O$, radius $R$. The value of $R$ is not $1$, as in the sine you learned in school; by convention it is set to $R = 3438$. Take any angle $\theta$ measured at $O$ from a reference point $A$ on the circle, and let $B$ be the point on the circle at that angle. Drop a perpendicular from $B$ to line $OA$, meeting it at $M$.

By definition, $BM = R\sin\theta$. Modern sine is the ratio $BM/OB$; Rsine is the length $BM$ itself, in a circle where the radius happens to be $3438$.

The segment $BM$ is the Rsine of $\theta$, written jyā or bhuja-jyā in the verse. Modern trigonometry divides $BM$ by the radius $OB$ to get a pure ratio, $\sin\theta$, that does not depend on how big the circle is. Indian astronomers instead fixed the radius once, at $3438$, and tabulated the length $BM$ directly. The specific choice $R = 3438$ is not arbitrary: it is (to the nearest whole number) the number of arc-minutes in one radian. That is what lets a length, $BM$, double as an angle, in arc-minutes, whenever $\theta$ is small, which is exactly the trick the five steps below rely on.

The manda construction

Now the construction the verse is describing.

$O$ is Earth. $C$, the mean position, is reached by turning through angle $\kappa$ (the manda-kendra) from reference ray $OA$. A small circle of radius $r$, centred at $C$, is the manda epicycle; $P$, the true position, lies on it, turned by that same angle $\kappa$ from ray $CO$ extended. $M$ is the foot of the perpendicular from $P$ to line $OC$.

1. Let $O$ be Earth, and let $\kappa$ (the manda-kendra) be the angle between the reference ray $OA$ and the planet's mean direction $OC$: that is, how far the planet currently is, in degrees, from the closest or farthest point in its orbit.
2. Draw a small circle centred at $C$, of radius $r$ (a fixed fraction of $R$, different for every planet, standing in for how far off-centre its orbit is). This is the manda epicycle. Mark $P$ on it, turned by the same angle $\kappa$ from ray $CO$ extended.
3. Drop a perpendicular from $P$ to line $OC$, meeting it at $M$. In right triangle $CMP$, $PM = r\sin\kappa$, the Rsine of $\kappa$ scaled down to the epicycle's own radius.
4. This length $PM$, expressed in the same $R = 3438$ units used for Rsines, is the phala: the correction, in arc-minutes. (This is exactly what the verse computes: the Rsine of $\kappa$, multiplied by the epicycle's size and divided by $360$, comes out to $PM$.)
5. Add the phala to the mean position $C$ on one side of the orbit, subtract it on the other, and the true position $P$ falls out. Line $OP$, not line $OC$, is where the planet actually appears.

Why does adding a small circle fix the problem?

The orbit is really an ellipse, with Earth near one focus, not a circle centred on Earth; the constant-speed circle is only a first sketch. The gap between that sketch and the true ellipse turns out, for a shape this close to circular, to itself trace a small circle: an ellipse bulges away from its best-fit circle twice per orbit, once at the close point and once at the far point, and a small circle spun at twice the mean rate traces exactly that back-and-forth wobble.

That small circle is the manda epicycle. Its radius $r$ is fixed by the orbit's eccentricity (roughly half the orbit's size times its eccentricity), and turning $P$ by the kendra angle $\kappa$ off the already-rotating direction $CO$ makes $P$ complete two full turns, relative to the fixed stars, for every one turn of $C$, exactly the doubled rate the wobble needs. This idea is very closely related to the process of estimating the curvature of a curve at a point by approximating it locally with a circle.

Try it yourself, in modern notation

This rule has exactly the shape of a formula still used today, the modern equation of the center:

$$\text{True longitude} \approx \text{Mean longitude} \pm \frac{r_{\text{epicycle}}}{R}\sin(\text{Mean anomaly})$$

Compare the first-order version still taught today, where $e$ is orbital eccentricity and $M$ the mean anomaly:

$$C \approx 2e\sin(M)$$

Swap "epicycle radius over trijya" for "$2e$" and the two formulas do the same job with different bookkeeping. As a sanity check: at mean anomaly $90^{\circ}$, the Rsine hits its maximum (3438), so the correction is as large as the epicycle allows, the biggest nudge the planet ever gets. At $0^{\circ}$ or $180^{\circ}$ (exactly at the close or far point), the Rsine is zero and the correction vanishes: the planet is exactly where the mean-motion clock said. That is the same reason a planet's real angular speed peaks and bottoms out at those same two points.

Source: Sūrya Siddhānta with the Gūḍhārthaprakāśikā commentary of Raṅganātha, Chapter 2, verse 39 (and its upapatti), from the Rashtriya Sanskrit Sansthan (Lucknow Campus) e-text edition, general editor Prof. Sarva Narayan Jha.

Ancient Indian astronomers did not lack algebra because they lacked algebraic thinking; they expressed it in sentences instead of symbols. Every verse in a Siddhānta is a tiny subroutine: take this number, do this operation, then that one, and the answer falls out.

Why the Formulas Kept Expiring

These verse-formulas were not eternally accurate. Every Siddhānta's calculations were built around specific numerical constants (how fast the Moon orbits, how long a year is, the epicycle-circumference numbers used above), and those constants were only as good as the observations available when the text was composed.

Over centuries, tiny errors compounded. A model tuned for accuracy around 1100 CE would drift further from the real sky as decades passed, much like a clock that gains a few seconds a day and is wildly wrong a year later.

A formula actually expiring: watch the arithmetic

The Sūrya Siddhānta states the length of the sidereal year as $365$ days, $6$ hours, $12$ minutes, $36.56$ seconds, precise to the second, stated with total confidence. The modern measured value is $365$ days, $6$ hours, $9$ minutes, $9.76$ seconds. The difference is about $3$ minutes $27$ seconds, too long, every single year.

That gap is invisible in any one year. But it does not cancel out; it accumulates. After $100$ years it has added up to under six hours. After $1{,}000$ years, it has added up to roughly $2$ days and $9$ hours, more than two full days between where the text's arithmetic says a given point in the year should fall and where it actually falls. A two-day error is not subtle: it is easily enough to put a predicted equinox, eclipse, or festival date visibly out of step with the sky. This is precisely the kind of drift bīja-saṃskāra existed to catch and correct, and precisely why an update like the Grahalāghava, issued roughly $900$ years after the Sūrya Siddhānta's figures were fixed, was not optional.

Indian astronomers had a name and a fix for this: bīja-saṃskāra (बीजसंस्कार), "seed correction." A bīja here is a small correction constant: a patch applied to an old formula's numbers to bring predictions back in line with fresh observations. Later astronomers, comparing the sky against old predictions, would recalculate these values and publish updated tables. One famous update is the Grahalāghava (1520 CE) by Gaṇeśa Daivajña, which re-issued the Sūrya Siddhānta's constants in corrected form.

An analogy for today's reader

Think of an old Siddhānta as software shipped once and never auto-updated. It works beautifully at launch, but the "operating environment," the actual sky, keeps moving in ways the original code didn't anticipate. A bīja-saṃskāra is the patch note: a small, targeted correction that keeps the old system usable for another few centuries.

This is also part of why so many different Siddhāntas exist. The sixth-century astronomer Varāhamihira wrote the Pañcasiddhāntikā ("the Five Siddhāntas"), surveying five earlier schools that were competing, and periodically correcting each other, in exactly this way.

So we already have one precise, technical sense of the word bīja: a small correction value, patched into an old formula. Keep that in mind: we are about to meet its second sense, in the very same book as some of these astronomical corrections.

The Book Where Algebra Was Born: Inside an Astronomy Text

In 628 CE, an astronomer named Brahmagupta completed the Brāhmasphuṭasiddhānta ("The Correctly Established Doctrine of Brahma"). Like the Sūrya Siddhānta, it is fundamentally a book about tracking the sky.

But buried inside it, in Chapter 18, titled Kuṭṭakādhyāya, is what historians of mathematics now consider the founding document of algebra as a formal branch of mathematics: rules for positive and negative numbers, rules for zero, a system for naming unknowns, methods for solving equations with one or several unknowns, and the general quadratic formula.

The world's first systematic algebra textbook is a chapter tucked inside a book about the stars. This is documented fact, not folklore, and it is the strongest answer to why "seed" and "algebra" ended up sharing a name: the astronomer managing all those bīja corrections is the same person who first wrote out the grammar of the unknown.

Naming the Unknown

Brahmagupta needed a word for "the unknown quantity you are solving for." He reached for an older term, yāvat-tāvat (यावत्तावत्), literally "as much as": some undetermined amount. He also used avyakta (unmanifest) and iṣṭa ("that which is sought").

Then he solved a problem any student of $x, y, z$ will recognize: how do you name a second or third unknown? His solution was colours. The word varṇa (वर्ण) conveniently means both "colour" and "letter of the alphabet," a pun that is exactly the point. The first unknown stayed yāvat-tāvat, abbreviated yā. Later commentators (starting with Prṭhūdakasvāmin, c. 860 CE) assigned the rest colour names, abbreviated the same way:

This is ancient India's version of $x, y, z, w$: a full notational system for juggling several unknowns, over a thousand years before European algebra settled on letters for the same purpose.

Equations were laid out on the page in tabular form, with no equals sign at all; terms of the same "kind" were written one beneath the other. Here is how they would write $10x - 8 = x^2 + 1$:

yā va  0    yā 10    rū 8̇
yā va  1    yā  0    rū  1

(yā va = square of the unknown; yā = the unknown; rū = plain constant, from rūpa; a dot over a number marks it negative.) Reading down each column achieves exactly what our equals sign achieves today.

The Rules of the Unknown

Brahmagupta also states the rules for combining unknowns, remarkably close to how a modern textbook would phrase them:

Brāhmasphuṭasiddhānta 18.41–42

अव्यक्त वर्गघनवर्गवर्गपञ्चगत षड्गतादीनां। तुल्यानां संकलितव्यवकलिते पृथगतुल्यानाम्॥
सदृशद्विवधो वर्गस्त्र्यादिवधस्तद्गतोऽन्यजातिवधः।
अन्योन्यवर्णघातो भावितकः पूर्ववच्छेषम् ॥

Transliteration: avyaktavargaghanavargavargapañcagataṣaḍgatādīnāṃ tulyānāṃ saṃkalitavyavakalite pṛthagatulyānām. sadṛśadvivadho vargastryādivadhastadgato'nyajātivadhaḥ. anyonyavarṇaghāto bhāvitakaḥ pūrvavaccheṣam.

Translation: "For the unknowns, their squares, cubes, fourth powers, fifth powers, sixth powers, and so on: addition and subtraction are performed [directly] for like terms; for unlike terms, they are simply stated separately. The product of two like unknowns is a square; the product of three or more like unknowns is a power of that designation. The product of different unknowns is called bhāvitaka."

In the symbols you know: $x + x = 2x$, but $x + y$ just stays as $x + y$; $x \cdot x = x^2$; and $x \cdot y$, the product of two different unknowns, gets its own name, bhāvitaka. This is the algebra of combining variables that every high schooler learns today, expressed 1,400 years ago as a Sanskrit verse.

And Then, the Quadratic Formula

The single most striking moment in the chapter is verse 44, where Brahmagupta states, for completely general coefficients, the formula for solving any quadratic equation of the form $ax^2 + bx = c$:

Brāhmasphuṭasiddhānta 18.44

वर्गचतुर्गुणितानां रूपाणां मध्यवर्गसहितानाम्।
मूलं मध्येनोनं वर्गद्विगुणोद्धृतं मध्यः ॥

Transliteration: varga-caturguṇitānāṃ rūpāṇāṃ madhyavargasahitānām, mūlam madhyenonaṃ vargadviguṇoddhṛtaṃ madhyaḥ.

Translation: "The absolute quantity [$c$], multiplied by four times the coefficient of the square of the unknown [$a$], increased by the square of the coefficient of the middle term [$b$]; the square root of that result, diminished by the coefficient of the middle term [$b$], and divided by twice the coefficient of the square of the unknown [$2a$], is the value of the unknown."

In modern notation:

$$x = \frac{\sqrt{4ac + b^2} - b}{2a}$$

Historians regard him as the earliest known author to state this formula for arbitrary coefficients, part of why mathematician David Mumford has called him "the key person in the creation of algebra as we know it."

So what did Brahmagupta call his own subject?

Brahmagupta never uses the word bījagaṇita for his chapter. He calls it kuṭṭaka ("the pulverizer"), after one of its key techniques. The name bīja-gaṇita, "seed mathematics," comes later, from the next character in our story.

The Poet Who Finally Called It "Seed"

Five hundred years after Brahmagupta, in 1150 CE, Bhāskara II (also called Bhāskarācārya, "Bhāskara the Teacher") wrote a dedicated, standalone book on algebra and gave it the name that has stuck ever since: Bījagaṇita.

He opens it with an invocation, a devotional verse as was customary, that turns out to be a deliberate, layered pun connecting mathematics, philosophy, and theology:

Bījagaṇita, Invocation

उत्पादकं यत्प्रवदन्ति बुद्धेरधिष्ठितं सत्पुरुषेण साङ्ख्याः।
व्यक्तस्य कृत्स्नस्य तदेकबीजं अव्यक्तमीशं गणितं च वन्दे ॥

Transliteration: utpādakaṃ yat pravadanti buddher adhiṣṭhitaṃ satpuruṣeṇa sāṅkhyāḥ
vyaktasya kṛtsnasya tadeka bījaṃ avyaktamīśaṃ gaṇitaṃ ca vande

"I salute that avyakta-gaṇita [Algebra], which the wise declare to be the producer of mathematical intellect, and which is the sole bīja (seed) of all vyakta-gaṇita [ordinary, manifest mathematics]."

Bhāskara is saying, directly: algebra, the mathematics of the not-yet-known, is the seed out of which all of ordinary, visible mathematics grows. An unknown quantity is a seed in exactly the sense a real seed is: small, easy to overlook, and yet containing, folded up inside it, everything that will eventually unfold. Solving an equation is watching that seed germinate into an answer.

So, Why Is Algebra Called "Bīja-Gaṇita"?

Two distinct but closely related uses of the word bīja run through this thousand-year tradition:

1. In astronomy, a bīja (as in bīja-saṃskāra) is a small correction constant, patched into an aging formula to keep it accurate.
2. In algebra, bīja is the unknown quantity itself: the small, hidden value that, once solved for, "grows into" the full answer.

It would be an overreach to claim algebra is named after astronomical corrections in some direct, one-step act of borrowing; the record does not quite show that. What it does show is something better: both senses of "seed" come from the same intellectual world, the same instinct to reach for the metaphor of a small essential value determining a much larger outcome, and, in Brahmagupta's case, literally the same book. Algebra grew, seed-like, out of the demands of astronomical calculation, inside a chapter of a Siddhānta, five centuries before anyone gave it a name of its own. When that name finally arrived, its coiner reached, quite deliberately, for the same word the sky-watchers had already been using for a hundred years.

Why an Old Word Still Matters

The next time you are asked to "solve for $x$," you are participating in an idea with a long, specific history: a hidden quantity, called a seed, first tamed by astronomers keeping pace with a drifting sky, later named by a poet-mathematician who thought in puns. The word travelled from India through the Arab world (picking up its English name from the Arabic al-jabr) and into every classroom on Earth. But the seed at the center of it all is still exactly the one Brahmagupta and Bhāskara described: an unknown, waiting to be solved.

---

A note on sources. This article draws on Amartya Kumar Dutta's "Mathematics in India, Part 6: The Foundations of Algebra" (Bhāvanā, April 2023); H. T. Colebrooke's 1817 English translation of the Bījagaṇita and the Kuṭṭakādhyāya of the Brāhmasphuṭasiddhānta; the GRETIL (Göttingen Register of Electronic Texts in Indian Languages) edition of the Brāhmasphuṭasiddhānta's mathematical chapters, input by Takao Hayashi from S. Dvivedin's 1902 Benares edition, for verse-exact readings of Chapter 18; B. Datta and A. N. Singh's History of Hindu Mathematics; and the Sūrya Siddhānta with the Gūḍhārthaprakāśikā commentary of Raṅganātha, in the Rashtriya Sanskrit Sansthan (Lucknow Campus) e-text edition (Chapter 2, verse 39, for the manda-phala rule). Verse numbering follows the standard chapter divisions used in these editions; minor variations exist across manuscripts and commentaries.

Glossary

• bīja (बीज) — seed; also, a small correction constant in astronomy, or the unknown quantity in algebra.
• gaṇita (गणित) — mathematics, calculation.
• avyakta (अव्यक्त) — unmanifest, unknown.
• vyakta (व्यक्त) — manifest, known/visible.
• varṇa (वर्ण) — colour; also, letter of the alphabet, used to name unknown quantities.
• yāvat-tāvat (यावत्तावत्) — "as much as"; the term for the first unknown quantity.
• samīkaraṇa (समीकरण) — equation, literally "making equal."
• Siddhānta (सिद्धान्त) — an established astronomical treatise giving methods for computing planetary positions.
• rāśi (राशि) — one of the twelve 30° divisions of the ecliptic used to state a planet's longitude.
• bīja-saṃskāra (बीजसंस्कार) — "seed correction"; a periodic update to an astronomical formula's constants.