Skip to content

Chapter 5: Bhāskara I, Bhāskara II and the Classical Synthesis

Two astronomer-mathematicians called Bhāskara frame five centuries of Indian mathematics. Bhāskara I, writing in 629 CE, is the first great commentator on Āryabhaṭa and the author of a celebrated rational approximation to the sine. Bhāskara II (born 1114), often called Bhāskarācārya, wrote the Līlāvatī and the Bījagaṇita, the most widely studied mathematical textbooks of medieval India, and brought the algebra of indeterminate equations to its classical form with the cyclic method (cakravāla). Between them lies a period of steady development by authors such as Lalla, Govindasvāmin, Śrīdhara, Mahāvīra, Vaṭeśvara, Muñjāla, Śrīpati and Jayadeva. The chapter ends with the commentators and astronomers who carried Bhāskara II’s tradition into the eighteenth century (Plofker 2009; Datta and Singh 1935).

5.1 Bhāskara I

5.1.1 Life and works

Bhāskara I’s commentary on the Āryabhaṭīya uses the year 629 CE in its examples, and he also wrote two astronomical treatises in the tradition of Āryabhaṭa: the Mahābhāskarīya (“Great [treatise] of Bhāskara”) and the shorter Laghubhāskarīya. Where he lived is uncertain. He refers to places in Saurāṣṭra (Gujarat), including Valabhī, and also to the Aśmaka country; K. S. Shukla, who edited all three works, suggested that he worked in western India (Shukla 1976). His commentary, the oldest surviving prose commentary on a Sanskrit mathematical text, is discussed in Chapter 3; its explanations, diagrams and examples have been studied in detail by Agathe Keller (Keller 2006).

5.1.2 The rational approximation to the sine

In the Mahābhāskarīya (7.17–19) Bhāskara I gives a way of computing the sine of an arc without a table. For an arc of \(x\) degrees (\(0 \le x \le 180\)), his rule is equivalent to

\[R\sin x^\circ \approx R\,\frac{4x(180-x)}{40500 - x(180-x)}, \qquad\text{or, in radians,}\qquad \sin\theta \approx \frac{16\,\theta(\pi-\theta)}{5\pi^2 - 4\,\theta(\pi-\theta)} .\]

Mathematical Result: Accuracy of Bhāskara I’s formula

The formula gives the exact value at \(0^\circ\), \(30^\circ\), \(90^\circ\), \(150^\circ\) and \(180^\circ\). Over the whole range its largest absolute error is about \(0.0016\) (for \(R=1\)), so that with \(R = 3438\) the error is never more than about 6 minutes of arc. For example, at \(x = 45^\circ\) it gives \(\frac{4\cdot 45\cdot 135}{40500 - 6075} = \frac{24300}{34425} = 0.70588\), against \(\sin 45^\circ = 0.70711\). The relative error is largest near \(0^\circ\) and \(180^\circ\), where the sine itself is small.

Bhāskara I does not say how he found the rule. Several modern reconstructions have been proposed, for example by R. C. Gupta, who showed how it can be obtained by fitting a rational function to known values.

A simple way to arrive at the formula

Suppose one looks for an approximation of the form

\[\sin x^\circ \approx \frac{x(180 - x)}{A - B\,x(180-x)},\]

which, like the sine, vanishes at \(0^\circ\) and \(180^\circ\) and is symmetric about \(90^\circ\). Requiring the exact values at \(30^\circ\) and \(90^\circ\) gives two equations:

\[\frac{4500}{A - 4500B} = \frac12,\qquad \frac{8100}{A - 8100B} = 1,\]

that is, \(A - 4500B = 9000\) and \(A - 8100B = 8100\). Hence \(B = \tfrac14\) and \(A = 10125\), and the approximation becomes

\[\frac{x(180-x)}{10125 - \tfrac14 x(180-x)} = \frac{4x(180-x)}{40500 - x(180-x)},\]

which is exactly Bhāskara I’s rule. Whether he reasoned in this way is not known.

Bhāskara I’s rule was used, with variations, by later astronomers including Brahmagupta, who gave it in a form with the radius \(R\) built in (Plofker 2009; Datta and Singh 1935).

5.2 Between the Two Bhāskaras

5.2.1 Varāhamihira

Bhāskara I’s predecessor Varāhamihira (mid-sixth century), who probably worked near Ujjain, is best known for his works on divination and astrology, but two of his books matter for mathematics. The Pañcasiddhāntikā summarises five astronomical systems current in his time—the Paitāmaha, Vāsiṣṭha, Romaka, Pauliśa and Saura (Sūrya) siddhāntas—and so preserves early astronomy that would otherwise be lost. Its trigonometry includes a table of sines for a radius of 120 at intervals of \(3^\circ 45'\) and rules equivalent to \(\sin^2\theta + \cos^2\theta = 1\) and to the expression of the sine of half an arc through the versed sine of the whole. The Bṛhatsaṃhitā contains the combinatorics of perfumes and the magic square discussed in Chapter 8. Varāhamihira also argued against Āryabhaṭa’s rotating earth, on the ground that birds would not be able to return to their nests (Pingree 1981; Plofker 2009).

5.2.2 The later authors

The five centuries from Bhāskara I to Bhāskara II were productive, although many works are lost or known only from quotations. Table 5.1 lists the main authors. Śrīdhara and Mahāvīra are discussed in Chapter 10.

Table 5.1. Some mathematical astronomers between Bhāskara I and Bhāskara II
Author Date Main work and contributions
Lalla 8th c. (disputed) Śiṣyadhīvṛddhidatantra; astronomy in the Āryabhaṭa tradition
Govindasvāmin 9th c. Commentary on the Mahābhāskarīya; a more precise sine table
Śrīdhara c. 750 (uncertain) Pāṭīgaṇita, Triśatikā; a lost algebra with a rule for quadratic equations
Mahāvīra c. 850 Gaṇitasārasaṅgraha
Pṛthūdakasvāmin 864 Commentary on Brahmagupta
Vaṭeśvara 904 Vaṭeśvarasiddhānta; extensive trigonometry and interpolation
Muñjāla 932 Laghumānasa; rules for the instantaneous motion of the Moon
Āryabhaṭa II c. 950–1100 Mahāsiddhānta; algebra and the kuṭṭaka
Śrīpati c. 1039 Siddhāntaśekhara; Gaṇitatilaka (arithmetic)
Jayadeva before 1073 The cyclic method (cakravāla), known from quotations by Udayadivākara (1073)

5.3 Bhāskara II: Life and Works

Bhāskara II tells us that he was born in Śaka 1036 (1114 CE), that he completed the Siddhāntaśiromaṇi at the age of thirty-six (1150 CE), and that he lived at Vijjaḍaviḍa, a place in the Deccan whose site is uncertain; his father Maheśvara was an astronomer and his teacher. An inscription of 1207 at Patan (near Chalisgaon, Maharashtra) records that his grandson Caṅgadeva, court astronomer to a Yādava king, founded a school (maṭha) for the study of Bhāskara’s works. Later tradition associates him with Ujjain, but there is no contemporary evidence for this. His last datable work is the Karaṇakutūhala, a practical handbook with epoch 1183 (Plofker 2009; Pingree 1981).

Definition: Bhāskara II’s principal works

  • Līlāvatī: arithmetic and mensuration, in about 270 verses of rules and examples.

  • Bījagaṇita: algebra, including indeterminate equations.

  • Siddhāntaśiromaṇi (“Crest-jewel of the siddhāntas”, 1150): mathematical astronomy in two parts, Grahagaṇita (planetary computation) and Golādhyāya (the sphere), with Bhāskara’s own commentary (Vāsanābhāṣya).

  • Karaṇakutūhala (epoch 1183): a practical astronomical handbook.

The Līlāvatī and Bījagaṇita were treated in the manuscript tradition sometimes as the first two parts of the Siddhāntaśiromaṇi and sometimes as separate works.

5.4 The Līlāvatī

5.4.1 Content and style

The Līlāvatī follows the standard order of the arithmetic texts: tables of units of money, weight and length; the operations on whole numbers, fractions and zero; methods such as inversion and supposition (solving a problem by assuming a trial value and correcting it); the rule of three and its extensions; mixtures, interest and partnership; series; plane figures; excavations; piles; sawing; heaps of grain; shadows; the pulveriser; and the “net of numbers” (aṅkapāśa), that is, permutations and combinations (Chapter 8). Its examples are addressed to a girl—“O Līlāvatī”, “O fawn-eyed one”—and are written with a poet’s care. This made the book a classic: it was the standard arithmetic text in much of India until the nineteenth century, and it attracted many commentaries. Table 5.2 summarises its sections (Colebrooke 1817; Plofker 2009).

Two problems from the Līlāvatī tradition

The broken necklace. (Not in Bhāskara’s own text: an extra example from the commentary Manorañjana, which Colebrooke gives in a note to Līlāvatī §54.) In a lovers’ quarrel a necklace broke. A third of the pearls fell to the ground, a fifth onto the bed; the woman saved a sixth and her lover a tenth, and six pearls remained on the string. How many pearls were there?
Solution. If there were \(n\) pearls, \(n\left(1 - \tfrac{1}{3} - \tfrac{1}{5} - \tfrac{1}{6} - \tfrac{1}{10}\right) = \tfrac{n}{5} = 6\), so \(n = 30\).
The bees (Līlāvatī §54 in Colebrooke). A fifth of a swarm of bees settled on a kadamba flower and a third on a śilīndhra flower; three times the difference of these went to a kuṭaja flower; and one bee remained, hovering in the air. How many bees were there?
Solution. \(\tfrac{n}{5} + \tfrac{n}{3} + 3\left(\tfrac{n}{3} - \tfrac{n}{5}\right) + 1 = n\), that is \(\tfrac{14n}{15} + 1 = n\), so \(n = 15\).

Table 5.2. The main sections of the Līlāvatī
Section Contents
Definitions (paribhāṣā) Units of money, weight, length and volume
Operations (parikarma) The eight operations on whole numbers; fractions; zero
Methods (prakīrṇaka) Inversion; supposition; the rule of concurrence (sum and difference); problems on squares; the rule of three and its extensions
Mixtures (miśra) Interest, partnership, purchase and sale, alloys of gold, combinations
Series (śreḍhī) Arithmetic and geometric progressions
Plane figures (kṣetra) Right triangles; triangles and quadrilaterals; the circle; the sphere
Excavations, piles, sawing, heaps Volumes of pits, piles of bricks, timber, heaps of grain
Shadows (chāyā) Problems of gnomons and lamps
Pulveriser (kuṭṭaka) Linear indeterminate equations
Net of numbers (aṅkapāśa) Permutations and combinations

5.4.2 Some notable rules

  • Zero. Bhāskara repeats the rules for adding, subtracting and multiplying with zero, and calls a quantity divided by zero khahara. He adds a curious rule: if a quantity is multiplied by zero but further operations remain, the zero multiplier should be kept, and if the result is then divided by zero, the quantity is unchanged. This handles expressions such as \((a \times 0)/0\) by “cancelling” the zeros—a procedure that is not valid as arithmetic but shows an attempt to keep track of zero symbolically.

  • Approximate square roots. To find the square root of a non-square number approximately, multiply it by a large square number, take the root of the product to the nearest integer, and divide by the root of the square multiplier.

  • Excavations. For a pit whose top and bottom are rectangles, the “exact” volume is found by adding the areas of the top and the bottom and the area obtained from the sums of the sides, dividing by six, and multiplying by the depth. For a pit with square ends of sides \(a\) and \(b\) and depth \(h\), this gives

    \[V = \frac{h}{6}\left[a^2 + b^2 + (a+b)^2\right] = \frac{h}{3}\left(a^2 + ab + b^2\right),\]

    which is the correct volume of the frustum of a square pyramid.

  • The sphere. In his own commentary on the Golādhyāya, Bhāskara justifies the rule for the surface of the sphere by imagining it cut into many narrow zones, computing the area of each from the sine table, and adding them up—an argument by summation of small parts that foreshadows the methods of the Kerala school.

  • Plane figures. Rules for triangles and quadrilaterals (including Brahmagupta’s formulas, with the warning that the “accurate” area of a quadrilateral is not determined by its sides alone unless further conditions are given), the circle with \(\pi \approx 3927/1250\) (and \(22/7\) for practical work), and the surface and volume of the sphere, correctly given as \(4\pi r^2\) and \(\tfrac{4}{3}\pi r^3\) in the form “the circumference times the diameter” and “the surface times the diameter, divided by six”.

5.4.3 More problems from the Līlāvatī

The geometrical problems of the Līlāvatī are among its most attractive. Most reduce to the rule of the right triangle.

The broken bamboo

A bamboo 32 cubits high was broken by the wind, and its tip touched the ground 16 cubits from its foot. At what height was it broken?
If the break is at height \(x\), the broken part has length \(32 - x\) and forms the hypotenuse of a right triangle with sides \(x\) and 16: \(x^2 + 16^2 = (32 - x)^2\), so \(256 = 1024 - 64x\) and \(x = 12\). Check: \(12^2 + 16^2 = 400 = 20^2\).

The peacock and the snake

A peacock sits on top of a pillar 9 cubits high. It sees a snake 27 cubits from the foot of the pillar, gliding towards its hole at the foot. The peacock swoops on the snake in a straight line, and they meet when they have travelled equal distances. How far from the hole do they meet?
If they meet \(x\) cubits from the pillar, the snake has travelled \(27 - x\) and the peacock \(\sqrt{x^2 + 81}\). Setting these equal: \((27 - x)^2 = x^2 + 81\), so \(729 - 54x = 81\) and \(x = 12\). Each has travelled 15 cubits.

The lotus

The tip of a lotus stood half a cubit above the water. Driven by the wind, it disappeared beneath the surface at a distance of 2 cubits from where it had stood. How deep is the water?
If the depth is \(x\), the stalk has length \(x + \tfrac12\), and \(x^2 + 2^2 = \left(x + \tfrac12\right)^2\), so \(4 = x + \tfrac14\) and \(x = 3\tfrac34\) cubits. A similar problem about a reed in a pond occurs in the Chinese Nine Chapters.

Two bamboos and the crossing strings

Two bamboos, 15 and 10 cubits high, stand upright some distance apart. Strings are stretched from the top of each to the foot of the other. At what height do the strings cross?
By similar triangles the height is \(\dfrac{15 \times 10}{15 + 10} = 6\) cubits, whatever the distance between the bamboos.

Two monkeys

Two monkeys sit at the top of a tree 100 cubits high. A well lies 200 cubits from its foot. One monkey climbs down and walks to the well; the other leaps straight up some distance and then descends in a straight line to the well. Both travel the same distance. How high did the second leap?
If the leap is \(x\), then \(x + \sqrt{(100 + x)^2 + 200^2} = 100 + 200\). Squaring after isolating the root: \((100+x)^2 + 40000 = (300 - x)^2\), so \(800x = 40000\) and \(x = 50\). The hypotenuse is \(\sqrt{150^2 + 200^2} = 250\), and \(50 + 250 = 300\).

Interest and the pulveriser

Interest. If the interest on 100 for a month is 5, what is the interest on 16 for a year? By the compound rule of three (“rule of five”), \(\dfrac{16 \times 12 \times 5}{100 \times 1} = 9\tfrac35\).
The pulveriser. Find a number which, multiplied by 221 and increased by 65, is exactly divisible by 195. Dividing through by the common factor 13 gives \(17x + 5 \equiv 0 \pmod{15}\), that is \(2x \equiv 10 \pmod{15}\), so \(x = 5\) (or \(5 + 15t\)). Check: \(221 \times 5 + 65 = 1170 = 195 \times 6\).

5.4.4 The legend of Līlāvatī

The preface to the Persian translation of the Līlāvatī made for Akbar by his court poet Fayḍī (1587; Chapter 13) tells a story that has become famous: Bhāskara’s daughter Līlāvatī was destined by her horoscope to remain unmarried; her father set up a water-clock to mark the one auspicious moment for her wedding, but a pearl from her ornaments fell into it, the moment passed unnoticed, and Bhāskara consoled her by naming his book after her. The story appears nowhere in Bhāskara’s own works, and there is no evidence that Līlāvatī was a real person; the name may simply be an epithet (“playful, charming”) used for the addressee of the verses. The legend shows, however, the prestige of the book in the Persianate culture of the Mughal court.

5.5 The Bījagaṇita

The Bījagaṇita begins, like Brahmagupta’s chapter 18, with the rules for positive and negative quantities and zero, the operations on unknowns, and surds. It then treats the pulveriser, the square-nature equation, equations in one unknown (linear and quadratic), equations in several unknowns, and equations involving products of unknowns (bhāvita). Colebrooke’s translation of 1817 made it known in Europe; an English version made from Persian by Edward Strachey had appeared in 1813 (Colebrooke 1817).

5.5.1 Surds

Bhāskara calls a quantity whose square root is to be taken a karaṇī. His rules include the rule for adding two surds, which in modern notation is

\[\sqrt{a} + \sqrt{b} = \sqrt{a + b + 2\sqrt{ab}},\]

applicable when \(ab\) is a perfect square, so that the result is again a single surd; for example \(\sqrt{2} + \sqrt{8} = \sqrt{2 + 8 + 8} = \sqrt{18}\). He also gives rules for multiplying and dividing surds and for removing a surd from a denominator by multiplying by the “conjugate” expression, as in \(\dfrac{1}{\sqrt{a}+\sqrt{b}} = \dfrac{\sqrt{a}-\sqrt{b}}{a-b}\).

5.5.2 Quadratic equations and their two roots

For the quadratic equation, Bhāskara quotes the rule of Śrīdhara: multiply both sides by four times the coefficient of the square, add the square of the coefficient of the unknown, and take the square root.

Mathematical Result: Śrīdhara’s rule

Given \(ax^2 + bx = c\):

\[\begin{alignat*} {2} &\text{1. multiply both sides by $4a$:} &\qquad& 4a^2x^2 + 4abx = 4ac;\\ &\text{2. add $b^2$ to both sides:} && (2ax + b)^2 = 4ac + b^2;\\ &\text{3. take square roots:} && 2ax + b = \sqrt{4ac + b^2};\\ &\text{4. hence} && x = \frac{\sqrt{4ac + b^2} - b}{2a}. \end{alignat*}\]

This is completing the square, done so as to avoid fractions until the last step. Bhāskara knew that a quadratic equation may have two roots and discussed when both are acceptable. Quadratic problems had been solved long before in Babylon and by Euclid’s geometry, and al-Khwārizmī (c. 825) later sorted them into types with positive coefficients, each justified by a figure; the Indian rule is stated once for all cases, negative coefficients included (Plofker 2009; Katz 2009).

Two monkey problems

(1) The square of an eighth part of a troop of monkeys were playing in a grove, and twelve more were seen on a hill. How many monkeys were there?
Solution. \((x/8)^2 + 12 = x\), that is \(x^2 - 64x + 768 = 0\), with roots \(x = 48\) and \(x = 16\). Both are accepted.
(2) The fifth part of a troop, less three, squared, had gone into a cave, and one monkey was seen climbing a branch. How many were there?
Solution. \((x/5 - 3)^2 + 1 = x\), that is \(x^2 - 55x + 250 = 0\), with roots \(x = 50\) and \(x = 5\). Bhāskara rejects the second, because then the “fifth part less three” would be negative, and “people do not approve” of a negative quantity in such a case.

Bhāskara also solves some equations of higher degree that can be reduced to quadratics by adding suitable quantities to both sides. His example \(x^4 - 2x^2 - 400x = 9999\) is solved by adding \(4x^2 + 400x + 1\) to both sides, which turns the left side into \((x^2+1)^2\) and the right into \((2x+100)^2\); this gives \(x^2 + 1 = 2x + 100\) and \(x = 11\). Check: \(11^4 - 2\cdot 121 - 4400 = 14641 - 242 - 4400 = 9999\).

5.5.3 Rational solutions of the square-nature equation

Before turning to whole-number solutions, the Indian authors gave a simple rule for rational solutions of \(Nx^2 + 1 = y^2\), stated by Bhāskara II and related to rules in Brahmagupta: choose any number \(m\), and take

\[x = \frac{2m}{m^2 - N},\qquad y = \frac{m^2 + N}{m^2 - N}.\]

A short calculation shows that these always satisfy the equation. For \(N = 8\) and \(m = 3\) the rule gives \(x = 6\), \(y = 17\), which happen to be whole numbers (\(8 \cdot 36 + 1 = 289 = 17^2\)); for \(N = 13\) and \(m = 4\) it gives \(x = \tfrac83\), \(y = \tfrac{29}{3}\), which are rational but not whole. The difficult problem, solved by the cyclic method, is to find solutions in whole numbers.

5.6 The Cyclic Method (Cakravāla)

Brahmagupta’s bhāvanā (Chapter 4) can produce solutions of \(Nx^2 + 1 = y^2\) once a suitable starting solution has been found, but it gives no guarantee that one can be found. The cyclic method supplies a systematic procedure. It is first known from verses of Jayadeva quoted by Udayadivākara in 1073, and was given its classical form by Bhāskara II in the Bījagaṇita (1150) (Plofker 2009; Selenius 1975).

Mathematical Result: The cakravāla

Suppose \(Nx^2 + k = y^2\) with whole numbers \(x\), \(y\), \(k\) (initially, for example, \(x = 1\) and \(y\) the integer nearest \(\sqrt{N}\)).

  1. Using the pulveriser, choose a positive integer \(m\) such that \((x m + y)\) is divisible by \(k\), and among such \(m\) choose the one for which \(|m^2 - N|\) is smallest.

  2. Form the new triple

    \[x' = \frac{xm + y}{|k|},\qquad y' = \frac{ym + Nx}{|k|},\qquad k' = \frac{m^2 - N}{k}.\]

    Then \(Nx'^2 + k' = y'^2\), and \(x'\), \(y'\), \(k'\) are again whole numbers.

  3. Repeat until \(k' = 1\). (If \(k'\) becomes \(-1\), \(\pm 2\) or \(\pm 4\), Brahmagupta’s rules can be used to finish at once.)

The step is a bhāvanā of the current solution with the trivial solution \((1, m)\) of \(N\cdot 1^2 + (m^2 - N) = m^2\), followed by division by \(k^2\); the choice of \(m\) makes the division exact and keeps the new \(k'\) small.

Scholarly Note: Why the new numbers are whole

Composing \((x, y, k)\) with \((1, m, m^2 - N)\) gives the triple \((xm + y,\ ym + Nx,\ k(m^2 - N))\). The rule chooses \(m\) so that \(k\) divides \(xm + y\). Then \(k\) also divides \(m^2 - N\): since \(y \equiv -xm \pmod{k}\) and \(y^2 - Nx^2 = k \equiv 0\), we have \(x^2(m^2 - N) \equiv 0 \pmod{k}\), and when \(x\) and \(k\) have no common factor this gives \(m^2 \equiv N \pmod{k}\). Finally \(ym + Nx \equiv -xm^2 + Nx = -x(m^2 - N) \equiv 0 \pmod{k}\). So all three numbers can be divided by \(k\) (and the last by \(k^2\) in total), leaving whole numbers. For \((5, 39, -4)\) with \(N = 61\) and \(m = 9\): \(45 + 39 = 84\), \(351 + 305 = 656\) and \(81 - 61 = 20\) are all divisible by 4.

Bhāskara did not prove that the process always ends with \(k = 1\); this was proved in modern times. In 1975 Clas-Olof Selenius showed that the method corresponds to a particular kind of continued-fraction expansion of \(\sqrt{N}\) and that it is, in a precise sense, an efficient one, often reaching the solution in fewer steps than the regular continued-fraction method of Euler and Lagrange (Selenius 1975).

The case \(61x^2 + 1 = y^2\)

Start from \(x=1\), \(y=8\): \(61 + 3 = 64\), so \(k = 3\). The cyclic method then runs as follows (\(m\) is the multiplier chosen at each step).

\(m\) \(x\) \(y\) \(k\)
— 1 8 3
7 5 39 \(-4\)
9 21 164 \(-5\)
6 58 453 5
9 195 1523 4
7 722 5639 \(-3\)
8 3805 29718 \(-1\)
8 60158 469849 \(-3\)
7 296985 2319527 4
9 1248098 9747957 5
6 3447309 26924344 \(-5\)
9 11590025 90520989 \(-4\)
7 42912791 335159612 3
8 226153980 1766319049 1

For instance, from \((1, 8, 3)\): \(m\) must make \(8 + m\) divisible by 3, and of the candidates \(4, 7, 10, \ldots\) the value \(m=7\) makes \(|m^2 - 61| = 12\) smallest; then \(x' = (7+8)/3 = 5\), \(y' = (61 + 56)/3 = 39\), \(k' = (49 - 61)/3 = -4\), and indeed \(61\cdot 25 - 4 = 1521 = 39^2\).
The final solution is \(x = 226{,}153{,}980\), \(y = 1{,}766{,}319{,}049\). Bhāskara II reaches it with fewer steps by combining the cyclic method with Brahmagupta’s special rules (for example at \(k = -4\), or by composing the solution with \(k=-1\) found at the seventh line with itself).

The short route is easily followed. After the first step of the cyclic method the triple is \((x, y, k) = (5, 39, -4)\), with \(y\) odd, and Brahmagupta’s rule for \(k = -4\) (Chapter 4) gives the answer at once:

\[X = \frac{5\cdot 39\,(39^2+1)(39^2+3)}{2} = \frac{195 \times 1522 \times 1524}{2} = 226{,}153{,}980,\]
\[Y = (39^2 + 2)\left[\frac{(39^2+1)(39^2+3)}{2} - 1\right] = 1523 \times 1{,}159{,}763 = 1{,}766{,}319{,}049 .\]

Composing the seventh line of the table, \((3805, 29718, -1)\), with itself leads to the same result, since \(2\cdot 3805 \cdot 29718 = 226{,}153{,}980\).

5.7 Instantaneous Motion and the Sine Difference

In the Grahagaṇita of the Siddhāntaśiromaṇi, in the chapter on true positions, Bhāskara II distinguishes the ordinary daily motion of a planet, the difference between its positions on successive days, from its “instantaneous” motion (tātkālikī gati) at a given moment. To compute the latter he uses a rule for the change in the sine: when an arc changes by a small amount, the sine changes by that amount multiplied by the cosine and divided by the radius. In modern notation,

\[R\sin(\theta + \delta) - R\sin\theta \approx \delta \cdot \frac{R\cos\theta}{R}\quad\text{for small }\delta .\]

Rules of this kind for the Moon’s velocity had already been used by Muñjāla (932) and Āryabhaṭa II. Bhāskara also remarks that at the point where the correction to the mean position (the “equation”) is greatest, the instantaneous correction to the velocity is zero (Plofker 2009).

Scholarly Note: Is this the derivative?

These rules are often described as a discovery of the derivative or of the differential calculus. That goes too far. Bhāskara’s rule is the statement, for one function, that the rate of change of the sine is proportional to the cosine—which in modern terms is \(\tfrac{d}{d\theta}\sin\theta = \cos\theta\)—and he uses it for a specific astronomical purpose. He does not define a general notion of rate of change for arbitrary quantities, does not develop rules for differentiating functions in general, and does not connect the procedure with the finding of areas. It is a significant piece of reasoning about instantaneous change, and the Kerala astronomers later developed related ideas much further; whether their work as a whole deserves the name “calculus” is a debated question, set out in Chapter 6 (Section 6.6). Bhāskara’s own rule is best described in its own terms (Plofker 2009).

5.7.1 The “attractive power” of the earth

In the Golādhyāya, explaining why the earth does not fall and why objects on all sides of it do not fall off, Bhāskara II says that the earth has an attractive power (ākṛṣṭi-śakti) by which it draws heavy things towards itself, so that they appear to fall. Similar ideas are found earlier, for example in Brahmagupta. The statement is sometimes presented as a discovery of gravitation five centuries before Newton. It is a qualitative explanation of why things fall towards the earth, of a kind found in several premodern traditions; it contains no law relating the force to mass and distance, and it was not connected with the motions of the planets, which Indian astronomy explained by other means.

5.8 Trigonometry: The Jyotpatti

A short section of the Golādhyāya, the Jyotpatti (“Production of sines”), explains how sine tables are constructed. It gives the rules for the sine of the sum and of the difference of two arcs, which in Indian form, with sines measured in a circle of radius \(R\), read

\[R\sin(\alpha \pm \beta) = \frac{R\sin\alpha\cdot R\cos\beta \pm R\cos\alpha\cdot R\sin\beta}{R},\]

together with the rules for half-arcs and the exact values of the sines of \(36^\circ\) and \(18^\circ\), obtained from the regular pentagon and decagon. With the sum and difference rules, a table at intervals of one degree can be built up from a few values. These rules give a geometric derivation of what are now the addition formulas of trigonometry; their use for constructing tables is explained step by step (Plofker 2009; Van Brummelen 2009).

The Golādhyāya also contains the standard rules of Indian spherical astronomy, such as the rule for the declination of the Sun and the rule for the “ascensional difference” (cara) on which the length of daylight depends. Indian astronomers derived such rules by projections and similar triangles inside the sphere rather than by spherical triangles; a worked example of this kind of reasoning is given in Chapter 9 (Section 9.11.1), and the astronomy itself in Volume 12 of this series.

5.8.1 A table at intervals of one degree

The Jyotpatti turns the sum rule into a recursion that can be followed mechanically. With the traditional radius \(R = 3438\), Bhāskara takes the sine of one degree as 60. To pass from the sine of any whole number of degrees to the next, he diminishes the sine by its 6569th part and adds ten times the cosine divided by 573 (Bhāskara II 1861; Plofker 2009). Writing \(J_n = R\sin n^\circ\) and \(K_n = R\cos n^\circ\), the rule is

\[J_{n+1} = J_n - \frac{J_n}{6569} + \frac{10\,K_n}{573}.\]

The rule is the sum rule with \(\beta = 1^\circ\), \(J_{n+1} = J_n\cos 1^\circ + K_n\sin 1^\circ\), in which the two constants have been replaced by simple fractions. For the cosine of one degree it uses \(1 - \tfrac{1}{6569}\), where the true value is \(1 - \tfrac{1}{6566}\). For the sine it uses \(\tfrac{10}{573} = 0.0174520\), where the true value is \(0.0174524\). The divisor 573 is chosen with care: it is exactly a sixth of 3438, so that \(10R/573 = 60\). The value 60 for the first sine and the constant in the rule are therefore the same approximation, \(R\sin 1^\circ \approx 60\) (more exactly 60.001).

The second step

For \(n = 1\): \(J_1 = 60\), and the cosine, from the right-triangle rule, is \(K_1 = \sqrt{3438^2 - 60^2} = 3437.476\). Then

\[J_2 = 60 - \frac{60}{6569} + \frac{10 \times 3437.476}{573} = 60 - 0.009 + 59.991 = 119.982,\]

against the modern \(3438\sin 2^\circ = 119.984\).

Carried through all 89 steps, with the cosine at each step found from the sine by the right-triangle rule, the recursion yields a table that never differs from \(3438\sin n^\circ\) by more than about \(0.04\) (Table 5.3). Āryabhaṭa’s rule of sine differences (Chapter 3) uses the sines alone, with a slightly wrong constant, and its errors grow as the table is built. Bhāskara’s recursion uses both the sine and the cosine, with constants correct to about one part in 40,000, and its errors grow only slowly, remaining below about \(0.04\) throughout.

Table 5.3. Sines at intervals of one degree from the Jyotpatti recursion (\(R = 3438\), recomputed)
\(n\) Recursion \(3438\sin n^\circ\) Difference
1 60.000 60.001 \(-0.001\)
2 119.982 119.984 \(-0.003\)
5 299.635 299.641 \(-0.007\)
10 596.989 597.002 \(-0.013\)
30 1718.966 1719.000 \(-0.034\)
45 2430.993 2431.033 \(-0.040\)
60 2977.359 2977.395 \(-0.037\)
89 3437.475 3437.476 \(-0.001\)

A companion rule steps through the quadrant in the traditional intervals of \(3^\circ 45'\), using \(\tfrac{100}{1529}\) for the sine of the interval. By it the first sine is \(3438 \times 100/1529 = 224.85\). The 1861 translation adds, in a parenthesis that may be the translator’s, that the first sine is therefore not 225, as it is usually stated to be; the rule thus corrects the round value at the head of Āryabhaṭa’s table (Table 3.3), and the modern value is \(224.86\). The Jyotpatti also gives, beside the exact construction of the sine of \(36^\circ\), a simple approximation to it: the radius multiplied by 5878 and divided by 10,000. This is correct to four places (\(\sin 36^\circ = 0.58779\)) (Bhāskara II 1861).

5.9 After Bhāskara II: Sanskrit Mathematics to the Eighteenth Century

Bhāskara II’s books dominated Sanskrit mathematics in northern and western India so completely that most later work took the form of commentary on them, or of astronomy written within their framework. This has made the centuries after him look empty in many histories. They were not, although their achievements were of a different kind from his. The Kerala school, which also began from Bhāskara’s books, is the subject of Chapter 6, and Nārāyaṇa Paṇḍita’s Gaṇitakaumudī (1356), the last major Sanskrit arithmetic, of Chapters 8 and 10. This section follows the northern and western tradition from the sixteenth century to the eighteenth (Plofker 2009; Pingree 1981).

5.9.1 Commentary as a vehicle for mathematics

Dozens of commentaries were written on the Līlāvatī and the Bījagaṇita, and the most important of them did more than gloss the verses. Bhāskara had supplied explanations (vāsanā) for parts of his astronomy but few for his arithmetic and algebra, and it was the commentators who wrote out demonstrations (upapatti) of his rules systematically. Gaṇeśa Daivajña, of an astronomer family of Nandigrāma on the western coast, is best known for the Grahalāghava (epoch 1520), a handbook that computes planetary positions with simple algebraic approximations in place of a sine table; his Buddhivilāsinī on the Līlāvatī (1545) gives demonstrations of many of its rules. Sūryadāsa, son of the astronomer Jñānarāja, whose Siddhāntasundara is dated 1503, commented on both of Bhāskara’s mathematical books in the first half of the sixteenth century. Kṛṣṇa Daivajña, an astronomer associated with the Mughal court around 1600, wrote the Bījapallava on the Bījagaṇita, whose demonstrations of the rules of signs, the pulveriser and the square-nature equation are among the fullest in Sanskrit. M. D. Srinivas has drawn on these commentaries, among other sources, to show that Indian upapatti was a developed practice of justification, not an occasional afterthought (Srinivas 2005; Plofker 2009).

A commentary of this kind could carry new mathematics as well. The Kriyākramakarī on the Līlāvatī, begun by Śaṅkara Vāriyar and completed by Nārāyaṇa in Kerala in the mid-sixteenth century, contains much of the Kerala school’s work on infinite series, set out as comments on Bhāskara’s rules for the circle (Chapter 6).

5.9.2 The astronomers of Banaras

In the seventeenth century Banaras became the main centre of Sanskrit astronomy, and its astronomer families, several of them from Maharashtra, argued vigorously about how far Bhāskara’s authority should extend. Munīśvara, son of the astronomer Raṅganātha, wrote the Marīci (1630s), a long commentary on the Siddhāntaśiromaṇi, and an astronomical treatise of his own, the Siddhāntasārvabhauma (1646); he defended Bhāskara’s methods. Kamalākara, of a rival family, wrote the Siddhāntatattvaviveka (1658), which drew on Islamic astronomy and trigonometry and criticised Bhāskara on several points. Their dispute, conducted in Sanskrit verse and commentary, is one of the best-documented scientific controversies in the tradition, and it shows that the authority of a classic did not prevent argument over its content (Plofker 2009; Pingree 1981).

5.9.3 New material from outside

The same period brought mathematics from outside into Sanskrit. Nityānanda translated the Persian Zīj-i Shāhjahānī as the Siddhāntasindhu, and at Jai Singh’s court Jagannātha Samrāṭ produced the Rekhāgaṇita, a Sanskrit version of Euclid’s Elements from the Arabic recension of Naṣīr al-Dīn al-Ṭūsī (Chapter 12). Euclid’s deductive geometry thus became available in Sanskrit only in the eighteenth century, and it had little effect on the way Sanskrit mathematics was written. Astronomical tables in Sanskrit, many of them computed for the handbooks of this period, survive in large numbers; their arithmetic is studied by Clemency Montelle and Kim Plofker (Montelle and Plofker 2018).

5.9.4 What the period achieved

These centuries produced few new theorems, and the tradition did not develop the general algebraic notation or the analytic methods that were transforming European mathematics at the same time. Their achievement lay in consolidation and in justification: the classical rules were explained, demonstrated and taught; astronomical computation was made easier; and the Sanskrit tradition engaged, critically and selectively, with Islamic astronomy and Greek geometry. Much of this literature is still unedited, and David Pingree’s Census records large numbers of manuscripts of commentaries and handbooks from these centuries, many of which have never been studied (Pingree 1970). English translations by Edward Strachey (Bija Ganita, 1813, from the Persian), John Taylor (Lilawati, 1816) and H. T. Colebrooke (1817) introduced Bhāskara’s work to European readers just as the Sanskrit tradition of teaching it was beginning to give way to the colonial school (Chapters 13 and 14) (Colebrooke 1817).

5.10 Assessment

Bhāskara I and Bhāskara II represent two kinds of achievement. Bhāskara I was an interpreter and teacher: his commentary preserved and explained Āryabhaṭa’s work, and his sine approximation is an elegant piece of original mathematics. Bhāskara II was a synthesiser: he gathered the arithmetic, algebra and astronomy of five centuries into works of great clarity and charm, corrected and extended them, and gave the cyclic method its classical form. His textbooks shaped the teaching of mathematics in India for seven centuries. The limits of his work are those of a synthesiser. He gave the cyclic method its classical form but did not show that it always ends; his treatment of division by zero is a symbolic convention rather than a theory; and his rule for instantaneous motion is applied to one astronomical problem, not developed into a general method. What he did more than any other author was to fix the form in which Indian mathematics was learned, and the commentaries on his books carried that form into the eighteenth century (Plofker 2009; Datta and Singh 1935).

Exercises

  1. Compute \(\sin 60^\circ\) by Bhāskara I’s formula and compare with the true value.

  2. A bamboo 18 cubits high is broken by the wind, and its tip touches the ground 6 cubits from its foot. At what height is it broken?

  3. A peacock on a pillar 12 cubits high sees a snake 36 cubits away making for its hole at the foot of the pillar. If both travel equal distances, where do they meet?

  4. A lotus stands 1 cubit above the water; pushed by the wind, it submerges 3 cubits away. How deep is the water?

  5. Two bamboos 12 and 6 cubits high are joined by strings from the top of each to the foot of the other. At what height do the strings cross?

  6. Solve \(23x^2 + 1 = y^2\), starting from \(23\cdot 1^2 + 2 = 5^2\).

  7. Two monkeys sit on a tree 30 cubits high; a well is 60 cubits from its foot. One goes down and walks to the well; the other leaps up and goes straight to the well. If the distances are equal, how high was the leap?

  8. Simplify \(\sqrt{3} + \sqrt{12}\) by Bhāskara’s rule for adding surds.

Answers. (1) \(\frac{4\cdot 60\cdot 120}{40500 - 7200} = 0.86486\); true \(0.86603\). (2) \(x^2 + 36 = (18 - x)^2\) gives \(x = 8\). (3) 16 cubits from the pillar. (4) \(x^2 + 9 = (x+1)^2\) gives 4 cubits. (5) \(12 \cdot 6/18 = 4\) cubits. (6) Composing \((1, 5, 2)\) with itself gives \((10, 48, 4)\); halving gives \((5, 24)\), and \(23 \cdot 25 + 1 = 576 = 24^2\). (7) \(x = hd/(2h + d) = 30 \cdot 60/(2 \cdot 30 + 60) = 15\) cubits (Chapter 9). (8) \(\sqrt{3 + 12 + 2\cdot 6} = \sqrt{27}\).

Scholarly Note: Further reading

Shukla’s editions of Bhāskara I (1976) and Keller’s study of his commentary (2006); Colebrooke’s translation (1817) of the Līlāvatī and Bījagaṇita, and the modern translation of the Līlāvatī with notes by Patwardhan, Naimpally and Singh (2001); Selenius (1975) on the cyclic method; Plofker (2009), chapters 5–6, and chapters 8–9 on the later commentators and astronomers; Srinivas (2005) on the commentators’ demonstrations (Shukla 1976; Keller 2006; Colebrooke 1817; Patwardhan et al. 2001; Selenius 1975; Plofker 2009; Srinivas 2005).