Skip to content

Chapter 10: The Arithmetic Books (Pāṭīgaṇita)

Alongside the astronomical treatises, whose mathematical chapters were followed in Chapters 3–5, India produced a long series of books devoted to arithmetic and mensuration alone. They are called pāṭīgaṇita, “calculation on the board”, after the board (pāṭī) covered with dust or sand on which the figures were written and erased. Chapter 2 described the Bakhshālī manuscript and the debate over its date. This chapter treats Śrīdhara and Mahāvīra in full and looks at the genre as a whole: how the books were organised, what they taught, how they solved problems, and how the tradition continued from the ninth century to the fourteenth. The applications that gave the genre its character—commerce, crafts, building and surveying—are the subject of Chapter 11 (Plofker 2009; Datta and Singh 1935).

10.1 The Genre of Pāṭīgaṇita

10.1.1 Names and scope

The names of the genre and its place among the branches of gaṇita were set out in Chapter 1. Besides pāṭīgaṇita, “board calculation”, the texts speak of dhūlīkarman, “dust-work”, after the same practice, and of vyaktagaṇita, “calculation with manifest quantities”, in contrast to algebra. Bhāskara II’s pairing of the Līlāvatī with the Bījagaṇita gives this division its classical form: everything that can be done with numbers alone belongs to the first, and everything that needs an unknown to the second. In practice the boundary was not sharp. The arithmetic books solve many problems that a modern reader would treat algebraically, using special rules for each type of problem in place of a general unknown.

Brahmagupta had described the mathematician’s competence as the mastery of twenty operations (parikarman) and eight determinations (vyavahāra) (Chapter 1). The arithmetic books follow this two-part plan. The operations are the procedures of pure calculation: the arithmetic of whole numbers and fractions, squares and roots, cubes and cube roots, the rule of three and its extensions. The determinations apply these procedures to classes of practical problems: mixtures (interest, partnership, alloys), series, plane figures, excavations, piles of bricks, sawing, heaps of grain and shadows. Between the two, most books place a section of “miscellaneous” problems (prakīrṇaka), in which an unknown quantity is to be found from given relations among its parts. The arrangement varied from author to author, but the basic scheme remained stable for more than a thousand years, and it explains why a merchant’s problem about the sale of cloth and an astronomer’s problem about the volume of a sphere could stand in the same book.

10.1.2 How the books are written

Like the astronomical texts, the arithmetic books state their rules in verse, usually in the āryā or anuṣṭubh metres, so that they could be memorised. A rule (sūtra or karaṇasūtra) is followed by one or more examples (udāharaṇa), also in verse. The examples are not decorative: they fix the meaning of a compressed rule by showing which quantity is which. A commentary, written by the author himself or by a later teacher, then sets down the data of each example (nyāsa), works through the computation and, often, checks the result. Much of what we know about the actual practice of calculation comes from these commentaries and from the solved examples of the Bakhshālī manuscript (Section 10.2) (Plofker 2009; Keller 2006).

The examples give the books much of their charm. They are set among merchants, travellers, kings and lovers, in gardens and temples, and the poets among the mathematicians—Mahāvīra and Bhāskara II above all—wrote them with literary care. They also give the historian a picture, partial and stylised, of the economic life the books served: the units of money and weight, the goods traded, the rates of interest, the wages of labourers and the prices of animals and of slaves.

10.1.3 Readers and purposes

The arithmetic books were written for several kinds of reader at once. Students of astronomy needed fluent calculation before they could approach the siddhāntas. Merchants, moneylenders, goldsmiths and officials needed proportion, interest and the reduction of mixed units. Builders and surveyors needed areas and volumes. The authors addressed all of these. Mahāvīra opens his book with a list of the activities in which calculation is required—worldly, Vedic and religious affairs, love and economics, music and drama, cookery and medicine, architecture and prosody, logic and grammar—in the passage that ends with the verse quoted at the head of Chapter 1 (Rangācārya 1912). Such statements are partly rhetorical, but the range of examples in the books bears them out.

The arithmetic books were also the first Indian mathematical texts that were not chapters of astronomical works. Their authors were not always astronomers: Mahāvīra’s surviving work is purely mathematical, and the later vernacular arithmetics (Chapter 13) were written by officials and poets. The genre thus shows mathematics becoming a subject of its own, taught and used outside the circle of the jyotiṣa families.

10.2 The Bakhshālī Manuscript as a Book of Problems

Chapter 2 (Section 2.4) described the discovery of the Bakhshālī manuscript, its notation, its rule for approximate square roots and the controversy over its date. Here it is considered as an example of what an arithmetic book looked like in use, since it is the only one to survive in an early physical copy with its worked solutions.

10.2.1 Form and layout

The manuscript contains rules, examples and solutions in a fixed order. A rule in verse is followed by an example; the data of the example are then written out in a table ruled into compartments, each holding a number or a fraction, with abbreviations marking the role of each entry; the computation follows, and the answer is often verified by substituting it back into the conditions of the problem. Fractions are written in vertical columns, and negative quantities are marked by a sign after the number (Chapter 2). The language is a non-standard Sanskrit with many Prakrit features, the kind of Sanskrit used by people who wrote it for practical purposes rather than as literature. G. R. Kaye’s edition (1927–33) published photographs of the leaves; Takao Hayashi’s edition (1995) reconstructed the order of the surviving material, much of which is damaged or out of sequence, and identified the rules and examples (Kaye 1927; Hayashi 1995).

10.2.2 Problem types

The surviving problems cover much of the later arithmetic syllabus: linear equations in one and several unknowns, often of the “equal wealth” type; arithmetic progressions, including problems about travellers whose daily journeys increase in progression; problems of profit and loss and of purchase and sale; the rule of three; and quadratic problems leading to the extraction of square roots, where the square-root rule of Chapter 2 is put to work. Many problems are solved by the method of “false position” (supposition): a convenient value is assumed for the unknown, the conditions of the problem are applied to it, and the assumed value is corrected in proportion to the error. The method appears in the later Sanskrit books as iṣṭakarman, “operation with an assumed number”, and it was also standard in Egyptian, Chinese, Islamic and medieval European arithmetic (Hayashi 1995; Datta and Singh 1935).

Worked example: false position

A problem often cited from the manuscript runs: a merchant pays duty at three places; at the first he gives a third of his goods, at the second a quarter of what remains, at the third a fifth of what then remains; the total duty is 24. How much did he have at first? (Joseph 2011)
Solution by false position. Assume 60, which is divisible by 3, 4 and 5. The duties are then \(20\), \(\tfrac14 \times 40 = 10\) and \(\tfrac15 \times 30 = 6\), a total of 36, and 24 remains. The true duty is 24, not 36, so the true amount is \(60 \times \tfrac{24}{36} = 40\).
Check. From 40 the duties are \(13\tfrac13\), \(6\tfrac23\) and 4, which add up to 24; \(16\) remains.

The method works because the answer is proportional to the data: whatever the original amount, the duty is a fixed fraction of it, here \(1 - \tfrac23 \cdot \tfrac34 \cdot \tfrac45 = \tfrac35\). The same reasoning solves a second type of problem, in which several people’s gifts or contributions stand in given ratios. If A gives an amount, B twice as much, C three times as much as B and D four times as much as C, and the gifts total 132, then assuming that A gives 1 makes the total \(1 + 2 + 6 + 24 = 33\); the true value of A’s gift is therefore \(132/33 = 4\).

Worked example: a traveller in progression

One traveller goes 18 yojanas each day. A second starts at the same time and place, going 3 yojanas on the first day, 6 on the second, 9 on the third, and so on. When does the second overtake the first?
After \(n\) days the first has gone \(18n\) and the second \(3 + 6 + \cdots + 3n = \tfrac32 n(n+1)\). They are level when \(\tfrac32 n(n+1) = 18n\), that is, when \(n + 1 = 12\): after 11 days, when both have gone 198 yojanas.

The numbers here are chosen for illustration. The manuscript’s problems of this type lead in general to quadratic equations, solved by the rule for the number of terms of a progression (Chapter 3).

10.2.3 Verification

The habit of verification is the manuscript’s most striking feature. After a solution the scribe often writes the word for “check” (pratyaya) and recomputes the conditions of the problem from the answer, showing that they are satisfied. For approximate square roots, the check consists in squaring the approximation and comparing it with the original number, with the small excess noted. This practice—computing, then confirming by reversing the computation—is found throughout later Indian arithmetic, and it is one source of the tradition’s justifications of its rules (Chapter 1).

The dispute over the date (Section 2.4) does not affect this content, only its place in history. On Hayashi’s dating the work stands at the head of the tradition of arithmetic books, beside Bhāskara I’s commentary, and that is how it is read here.

10.3 Śrīdhara: The Pāṭīgaṇita and the Triśatikā

10.3.1 The author and his works

Śrīdhara is known only from his works and from references to them by later authors; his date, usually put around 750 CE, rests on his relation to Brahmagupta and to Mahāvīra, and dates as late as the tenth century have been proposed. The opening invocation of the Pāṭīgaṇita is addressed to Śiva, and K. S. Shukla, who edited it, took him to be a Śaiva; the claim that he was a Jain rests on weaker grounds. Two arithmetic works survive. The Pāṭīgaṇita is the larger; it survives in a single incomplete manuscript, with an anonymous commentary, and breaks off before the end of the determinations. The Pāṭīgaṇitasāra (“Essence of the Pāṭīgaṇita”), known as the Triśatikā because it has about three hundred verses, is an abridgement that covers the whole syllabus and was widely copied and commented on. A work on algebra is lost, but later authors quote from it, and Bhāskara II gives Śrīdhara’s rule for the quadratic equation (Chapter 5). Śrīdhara also states rules for zero: adding zero to a number or subtracting it leaves the number unchanged, and multiplying by zero gives zero (Shukla 1959; Plofker 2009; Datta and Singh 1935).

10.3.2 Fractions and their classes

Śrīdhara treats fractions systematically. After the operations on simple fractions he classifies the compound expressions met in practice into types (jāti), each with its own rule of reduction to a single fraction. The most common are:

  • bhāga: fractions to be added or subtracted, such as \(\tfrac12 + \tfrac13\);

  • prabhāga: a fraction of a fraction, such as \(\tfrac12\) of \(\tfrac34\);

  • bhāgabhāga: a quantity divided by a fraction, such as \(1 \div \tfrac34\);

  • bhāgānubandha: a fraction “increased by” a fraction of itself, such as a quarter increased by its own third;

  • bhāgāpavāha: a fraction “diminished by” a fraction of itself, such as a half diminished by its own fifth.

Mahāvīra adopts the same classes and adds a sixth, bhāgamātṛ, for expressions that combine several of the others. The classification reflects the way such quantities were spoken of in trade—“a quarter and a third of it”, “half less a fifth”—and the rules translate everyday phrases into exact arithmetic (Shukla 1959; Datta and Singh 1935).

Worked example: the classes of fractions

Association (bhāgānubandha). A quarter increased by its own third is \(\tfrac14 + \tfrac13 \cdot \tfrac14 = \tfrac14\left(1 + \tfrac13\right) = \tfrac13\).
Dissociation (bhāgāpavāha). A half diminished by its own fifth is \(\tfrac12\left(1 - \tfrac15\right) = \tfrac25\).
Fraction of a fraction (prabhāga). Two-thirds of three-quarters of five-sixths of 72 is \(72 \times \tfrac56 \times \tfrac34 \times \tfrac23 = 30\).
The rules are written so that the result is reached by multiplying numerators and denominators in a fixed order; the commentaries then reduce the result to lowest terms.

10.3.3 Proportion and the determinations

Śrīdhara’s statement of the rule of three, with its three terms—the “measure” (pramāṇa), its “fruit” (phala) and the “requisition” (icchā)—and his rules for the inverse rule and for proportions with five, seven, nine and eleven terms became the model for later books; they are described in Chapter 11. His determinations follow Brahmagupta’s eight: mixtures (including interest and the purchase and sale of goods), series, plane figures, excavations, piles, sawing, heaps of grain and shadows. In each he gives both “gross” and more accurate rules where the tradition had them, and in each the examples are drawn from practical life (Shukla 1959).

10.3.4 The volume of the sphere

One of Śrīdhara’s mensuration rules shows how the arithmetic books improved inherited results. Āryabhaṭa’s rule for the volume of a sphere—the area of a great circle multiplied by its own square root—is wrong (Chapter 3). Śrīdhara instead takes half the cube of the diameter and adds to it its own eighteenth part, that is \(V = \tfrac{19}{36}\,d^3\). Mahāvīra gives a “gross” value of \(\tfrac92 r^3\) and an “accurate” value nine-tenths of this. Bhāskara II, finally, gives the exact rule, the surface multiplied by the diameter and divided by six (Chapter 5). Table 10.1 compares them.

Table 10.1. Rules for the volume of a sphere of radius \(r\), as multiples of \(r^3\)
Author Rule (modern form) Coefficient Error
Āryabhaṭa (499) \(\pi r^2 \sqrt{\pi r^2}\) 5.568 \(+32.9\%\)
Śrīdhara (c. 750?) \(\tfrac{19}{36}d^3 = \tfrac{38}{9}r^3\) 4.222 \(+0.8\%\)
Mahāvīra, gross (c. 850) \(\tfrac92 r^3\) 4.500 \(+7.4\%\)
Mahāvīra, accurate \(\tfrac{9}{10}\cdot\tfrac92 r^3\) 4.050 \(-3.3\%\)
Bhāskara II (1150) \(\tfrac{1}{6} S d = \tfrac43 \pi r^3\) 4.189 exact

Śrīdhara’s coefficient is the best of the approximate rules; it corresponds to a value of \(\pi\) of \(\tfrac{19}{6} \approx 3.17\) in the exact formula \(\tfrac{\pi}{6}d^3\). How he arrived at it is not known.

10.4 Mahāvīra’s Gaṇitasārasaṅgraha

Mahāvīra, a Jain, wrote the Gaṇitasārasaṅgraha (“Compendium of the essence of calculation”) around 850 at Mānyakheṭa in Karnataka, at the court of the Rāṣṭrakūṭa king Amoghavarṣa I (r. 814–878), whom he praises at the beginning of the work. With more than a thousand verses of rules and examples it is the largest and most systematic of the early arithmetic books. Its nine chapters treat terminology (saṃjñā), the operations (parikarma), fractions (kalāsavarṇa), miscellaneous problems (prakīrṇaka), the rule of three (trairāśika), mixed problems (miśraka), plane figures (kṣetra), excavations (khāta) and shadows (chāyā). M. Rangācārya edited and translated it in 1912. This section follows the book through, chapter by chapter, to show the range and the method of a ninth-century arithmetic (Rangācārya 1912; Plofker 2009; Datta and Singh 1935).

10.4.1 Terminology and the scale of numbers

The first chapter (saṃjñā) defines the units of length, area, volume, time, grain, gold, silver and other metals that the problems use, and the rules for positive and negative quantities and zero. A negative number, Mahāvīra says, has no square root, “since it is not a square by nature”; and a number divided by zero remains unchanged, a rule that later authors did not follow (Table 7.2). The chapter also names the decimal places. Mahāvīra’s list runs to twenty-four places, from eka (units) through daśa (tens), śata (hundreds) and sahasra (thousands) to mahākṣobha, the place of \(10^{23}\); such lists, which differ from author to author in their higher names, show how naturally the place-value system was extended to very large numbers.

10.4.2 The eight operations

The second chapter (parikarman) treats eight operations: multiplication, division, squaring, square root, cubing, cube root, the summation of series (saṅkalita) and the subtraction of a part of a series from the whole (vyutkalita). Addition and subtraction are not counted separately; they are taken for granted. Square and cube roots are extracted digit by digit, the digits of the number being marked off alternately as “odd” (viṣama) and “even” (sama) places for square roots, and in groups of three for cube roots, as in the method still taught in schools.

Among the multiplication exercises (2.10–17) are products whose results read like strings of beads, which Mahāvīra calls “necklaces”. Those he names so include

\[12345679 \times 9 = 111111111,\quad 14287143 \times 7 = 100010001,\quad 152207 \times 73 = 11111111,\]

and other exercises, such as \(139 \times 109 = 15151\), give similar patterns without the name. Such exercises gave practice in long multiplication and, at the same time, a check: an error anywhere spoils the pattern (Rangācārya 1912; Joseph 2011).

Worked example: summation and “subtraction” of series

Summation (saṅkalita). A progression with first term 4, common difference 3 and 10 terms has the sum \(\tfrac{10}{2}\left[2\cdot 4 + 9 \cdot 3\right] = 5 \times 35 = 175\).
Subtraction (vyutkalita). The sum of the last six terms of the same progression is the whole sum less the sum of the first four: the first four terms sum to \(\tfrac42[8 + 9] = 34\), so the last six sum to \(175 - 34 = 141\). Directly: \(16 + 19 + 22 + 25 + 28 + 31 = 141\).
Mahāvīra’s rules for vyutkalita let one find the sum of any run of terms in the middle of a progression without writing the terms out.
Geometric progression. The sum of \(n\) terms with first term \(a\) and ratio \(r\) is \(a(r^n - 1)/(r - 1)\). With \(a = 3\), \(r = 5\), \(n = 4\): \(3 \cdot \frac{625 - 1}{4} = 468\), and indeed \(3 + 15 + 75 + 375 = 468\).

10.4.3 Fractions

The third chapter (kalāsavarṇa, “reduction of fractions”) covers the six classes of compound fractions described above and an unusually full treatment of unit fractions—fractions with numerator one. One rule writes 1 as a sum of any chosen number \(n \ge 3\) of unit fractions,

\[1 = \frac{1}{2}+\frac{1}{3}+\frac{1}{3^2}+\cdots+\frac{1}{3^{n-2}}+\frac{1}{2\cdot 3^{n-2}},\]

so that with five terms \(1 = \frac12 + \frac13 + \frac19 + \frac1{27} + \frac1{54}\), since \(27 + 18 + 6 + 2 + 1 = 54\); another splits a unit fraction into two, \(\frac{1}{q} = \frac{1}{q+1} + \frac{1}{q(q+1)}\), as in \(\frac13 = \frac14 + \frac1{12}\). He gives further rules for writing a given fraction as a sum of unit fractions, for writing 1 as a sum of fractions whose numerators are given, and for related problems. Takanori Kusuba has analysed these rules and shown how they fit together. One of them splits a fraction into a unit fraction and a remainder, choosing an auxiliary number so that the division comes out exactly; applied repeatedly with the smallest such number, it gives the procedure later described in Europe by Fibonacci (Kusuba 2004; Rangācārya 1912).

Worked example: splitting a fraction into unit fractions

To write \(\tfrac27\) as a sum of unit fractions, take the largest unit fraction not larger than it: since \(7/2 = 3\tfrac12\), this is \(\tfrac14\). The remainder is \(\tfrac27 - \tfrac14 = \tfrac{8 - 7}{28} = \tfrac{1}{28}\), already a unit fraction. So \(\tfrac27 = \tfrac14 + \tfrac{1}{28}\).
For \(\tfrac{3}{7}\): the first unit fraction is \(\tfrac13\), and \(\tfrac37 - \tfrac13 = \tfrac{2}{21}\); for \(\tfrac{2}{21}\) the next is \(\tfrac{1}{11}\), and \(\tfrac{2}{21} - \tfrac{1}{11} = \tfrac{1}{231}\). So \(\tfrac37 = \tfrac13 + \tfrac1{11} + \tfrac1{231}\).

Why so much attention to unit fractions? The units of money and measure were divided into halves, quarters, sixteenths and other simple parts, and a fraction expressed as a sum of such parts could be paid or measured out directly. The interest may also have been partly recreational: the problems are ingenious, and some have no practical use at all.

10.4.4 Miscellaneous problems

The fourth chapter (prakīrṇaka) collects problems in which an unknown whole—a herd, a swarm, a sum of money—is to be found from statements about its parts. Mahāvīra sorts them into named types according to the form of the conditions: in the “fraction” type (bhāga) given fractions of the whole are removed and a known remainder is left; in the “remainder” type (śeṣa) the fractions are taken successively from what remains; in the “root” type (mūla) a multiple of the square root of the whole is removed as well; and there are further types combining these. Each type has its own rule. The “root” types lead to quadratic equations, and the rule for them is in effect the solution of a quadratic in the square root of the unknown.

Worked example: a problem of the “root” type

Of a flock of peacocks, a quarter and twice the square root of the whole flock went to the hills, and 32 remained. How many peacocks were there? (Numbers chosen for illustration.)
Let the flock be \(x = y^2\). Then \(y^2 - \tfrac14 y^2 - 2y = 32\), that is \(\tfrac34 y^2 - 2y - 32 = 0\). The rule for this type amounts to

\[y = \frac{\tfrac{c}{2} + \sqrt{\left(\tfrac{c}{2}\right)^2 + (1-a)\,r}}{1 - a},\]

where \(a = \tfrac14\) is the fraction removed, \(c = 2\) the multiplier of the root and \(r = 32\) the remainder:

\[y = \frac{1 + \sqrt{1 + 24}}{3/4} = \frac{6}{3/4} = 8 .\]

The flock had \(x = 64\) peacocks.
Check. \(\tfrac14 \cdot 64 = 16\) and \(2\sqrt{64} = 16\) went away, and \(64 - 32 = 32\) remained.

The Līlāvatī’s problem of the bees (Chapter 5) belongs to the simplest, “fraction” type; Bhāskara II also has problems of the root type, which he solves by the same rule.

10.4.5 Proportion and mixtures

The fifth chapter treats the rule of three, direct and inverse, and the sixth, “mixed problems” (miśraka), is the longest in the book. It contains problems of interest, of the division of profits, of purchase and sale, a long section on mixtures of gold of different fineness, the rule for combinations illustrated with the six tastes (Chapter 8), and a section of “curious” problems solved with the pulveriser (vicitrakuṭṭīkāra). The commercial material is discussed in Chapter 11. The pulveriser problems include the type in which a given sum of money buys a given number of animals or birds of several kinds at different prices. This type is also well known in China: the Zhang Qiujian suanjing (fifth century) asks how 100 coins can buy 100 fowls, when a cock costs 5 coins, a hen 3, and three chicks cost 1.

Worked example: the “hundred fowls” type

With the Chinese numbers: if \(c\) cocks, \(h\) hens and \(k\) chicks are bought, then \(c + h + k = 100\) and \(5c + 3h + \tfrac13 k = 100\). Multiplying the second equation by 3 and subtracting the first gives \(14c + 8h = 200\), or \(7c + 4h = 100\). This is a linear indeterminate equation of the kind solved by the pulveriser. Its solutions in non-negative whole numbers are \(c = 0, 4, 8, 12\) with \(h = 25, 18, 11, 4\), and then \(k = 75, 78, 81, 84\). Check for \(c = 4\): \(4 + 18 + 78 = 100\) birds and \(20 + 54 + 26 = 100\) coins.

10.4.6 Plane figures, excavations and shadows

The seventh chapter (kṣetra) gives “practical” (vyāvahārika) and “accurate” (sūkṣma) rules for triangles, quadrilaterals, circles, segments of circles and a number of composite figures, among them the ring, the figure shaped like a drum and the “elongated circle” (āyatavṛtta). It also treats “generated” (janya) figures, triangles and quadrilaterals whose sides and areas are rational, constructed from rational right triangles in the manner of Brahmagupta (Chapter 4). For the elongated circle, an oval with length \(L\) and breadth \(B\), Mahāvīra’s more accurate rule for the perimeter is equivalent to

\[P \approx \sqrt{6B^2 + 4L^2}.\]

For a circle (\(L = B = d\)) it reduces to \(\sqrt{10}\,d\), the Jain value of the circumference (Chapter 2).

Worked example: the perimeter of an oval

Take an oval of length \(L = 10\) and breadth \(B = 6\). Mahāvīra’s rule gives \(P \approx \sqrt{6\cdot 36 + 4 \cdot 100} = \sqrt{616} \approx 24.82\). If the figure is an ellipse with these axes, its true perimeter is \(25.53\) (to two decimal places), so the rule is about \(2.8\%\) too small.
No simple exact formula for the perimeter of an ellipse exists; it requires an elliptic integral. One of the best simple approximations was published by Ramanujan in 1914 (Chapter 14): with semi-axes \(a = 5\), \(b = 3\), his formula \(\pi\left[3(a+b) - \sqrt{(3a+b)(a+3b)}\right]\) gives \(25.5270\), agreeing with the true value to four decimal places. The comparison is between two very different aims: Mahāvīra wanted a rule that a surveyor could apply, and it is not known whether he meant his oval to be an ellipse at all.

The eighth chapter (khāta) treats the volumes of excavations, including pits whose top and bottom differ, of spheres (Table 10.1) and of piles of bricks; the ninth (chāyā) treats shadows. Their rules are discussed with the other rules for builders in Chapter 11.

10.4.7 Reception

The Gaṇitasārasaṅgraha was copied and studied above all in the South, and it is from South Indian manuscripts that M. Rangācārya edited it in 1912. It was adapted into Telugu by Pāvulūri Mallana and was known to the Kannada mathematical writers (Chapter 13). In the North its place was taken, from the twelfth century, by the Līlāvatī. Mahāvīra’s book is also the main witness to the mathematical culture of the Jain community, but its methods are those of the general tradition, and its problems, though often framed with Jain imagery, are addressed to any reader who calculates (Rangācārya 1912; Plofker 2009).

10.5 Finding Unknowns Without Algebra

The arithmetic books solve many problems that we would solve with an equation, but they do so without writing down an unknown. Instead they teach a handful of general methods, each suited to a particular shape of problem. Four of them recur in every book from Āryabhaṭa to Nārāyaṇa: inversion, supposition, concurrence and the “dissimilar operation”. Together they cover most of the problems in one unknown that arise in practice, and they show how far a purely arithmetical approach could go.

10.5.1 Inversion

When a number has been subjected to a known chain of operations and the final result is given, the original number is found by “inversion” (viloma or vyastavidhi): one starts from the result and undoes the operations in reverse order, dividing where the problem multiplied, subtracting where it added, taking the square root where it squared, and so on. Āryabhaṭa states the method in a single verse of the Gaṇitapāda (Chapter 3). The only subtle point concerns fractional increases. If a number was increased by \(\tfrac{a}{b}\) of itself, it was multiplied by \(\tfrac{a+b}{b}\), and to undo this it must be diminished by \(\tfrac{a}{a+b}\) of itself; Bhāskara II gives exactly this rule, and the corresponding one for a fractional decrease.

Worked example: inversion in the Līlāvatī

Bhāskara II asks for the number which, multiplied by 3, increased by three-quarters of the product, divided by 7, diminished by a third of the quotient, multiplied by itself, diminished by 52, square-rooted, increased by 8 and divided by 10, gives 2 (Colebrooke 1817).
Inversion. Start from 2 and reverse each step: \(2 \times 10 = 20\); \(20 - 8 = 12\); \(12^2 = 144\); \(144 + 52 = 196\); \(\sqrt{196} = 14\); a diminution by a third is undone by an increase of a half, \(14 + 7 = 21\); \(21 \times 7 = 147\); an increase by three-quarters is undone by a diminution of three-sevenths, \(147 - 63 = 84\); \(84 \div 3 = 28\).
Check. \(28 \times 3 = 84\); \(84 + 63 = 147\); \(147 \div 7 = 21\); \(21 - 7 = 14\); \(14^2 = 196\); \(196 - 52 = 144\); \(\sqrt{144} = 12\); \(12 + 8 = 20\); \(20 \div 10 = 2\).

10.5.2 Supposition

The method of supposition (iṣṭakarman), already met in the Bakhshālī manuscript (Section 10.2), handles problems in which the conditions are all proportional to the unknown: a number multiplied, divided, increased by fractions of itself and so on, with a final result given. Any convenient number is assumed and put through the conditions; the true number is then the assumed number multiplied by the given result and divided by the result obtained from the assumption.

Worked example: supposition

A number is multiplied by 5; a third of the product is subtracted; the remainder is divided by 10; and a third, a half and a quarter of the original number are added. The result is 68. What is the number? (Colebrooke 1817)
Assume 3. Then \(3 \times 5 = 15\); \(15 - 5 = 10\); \(10 \div 10 = 1\); \(1 + 1 + \tfrac32 + \tfrac34 = 4\tfrac14\). The true result is 68, so the number is \(3 \times 68 \div 4\tfrac14 = 48\).
Check. \(48 \times 5 = 240\); \(240 - 80 = 160\); \(160 \div 10 = 16\); \(16 + 16 + 24 + 12 = 68\).

Supposition fails when the conditions include a constant term or a square, since the result is then no longer proportional to the unknown. The texts know this: problems with added constants are solved by inversion, and those with squares and roots by special rules such as Mahāvīra’s rules for the “root” type (Section 10.4).

10.5.3 Concurrence and the dissimilar operation

Two short rules complete the set. “Concurrence” (saṅkramaṇa) finds two numbers from their sum and their difference: half the sum of the two given quantities is the larger number, half their difference the smaller. The “dissimilar operation” (viṣamakarman) finds two numbers from the difference of their squares and their simple difference: dividing the first by the second gives their sum, and the problem is reduced to concurrence. Both rest on simple identities, the second on \(a^2 - b^2 = (a + b)(a - b)\).

Worked example: concurrence and the dissimilar operation

Concurrence. Two numbers have sum 101 and difference 25. The larger is \(\tfrac12(101 + 25) = 63\) and the smaller \(\tfrac12(101 - 25) = 38\).
Dissimilar operation. The squares of two numbers differ by 40, and the numbers themselves by 4. Their sum is \(40 \div 4 = 10\), so by concurrence they are \(\tfrac12(10 + 4) = 7\) and \(\tfrac12(10 - 4) = 3\). Check: \(49 - 9 = 40\).

These methods are local: each applies to one shape of problem. The algebra of Brahmagupta and Bhāskara II (Chapters 4 and 9) replaced them with a single procedure: name the unknown, write the conditions as an equation, and solve it. The arithmetic books nevertheless kept them, because they were quick, easy to remember and well suited to mental calculation.

10.6 Later Arithmetic Books

10.6.1 Śrīpati’s Gaṇitatilaka

Śrīpati, an astronomer of the eleventh century (his Dhīkoṭidakaraṇa has the epoch 1039), wrote an arithmetic, the Gaṇitatilaka (“Ornament of calculation”), as well as the astronomical Siddhāntaśekhara, whose chapters on arithmetic and algebra overlap with it. The Gaṇitatilaka survives incomplete. It is of interest partly for its clear and orderly rules and partly for its commentary, written in the thirteenth century by the Jain monk Siṃhatilaka Sūri, which supplies worked solutions; the book thus shows a Brahmin astronomer’s arithmetic being taught in a Jain setting. H. R. Kapadia edited the text with Siṃhatilaka’s commentary in 1937 (Kapadia 1937; Plofker 2009).

10.6.2 The Līlāvatī

Bhāskara II’s Līlāvatī (1150), described in Chapter 5, gathered the material of the earlier arithmetic books into a single, elegant text, and from the thirteenth century onward it displaced them over most of India. Its success shows the strengths of the genre: a fixed syllabus, rules short enough to memorise, examples that teach as they please, and commentaries that explain the reasons. Of the dozens of commentaries on it, those of Gaṇeśa Daivajña (1545) and of the Kerala school (Chapter 6) are the best known, and it was translated into Persian in 1587 (Chapter 13).

10.6.3 Nārāyaṇa’s Gaṇitakaumudī

The last major Sanskrit arithmetic book is the Gaṇitakaumudī (“Moonlight of calculation”) of Nārāyaṇa Paṇḍita, completed in 1356. Its fourteen chapters begin with the usual operations and determinations, in a more expanded form than the Līlāvatī, and end with two chapters of original work on combinatorics and on magic squares (Chapter 8). Between them come chapters on topics the earlier books had treated briefly or not at all, among them the square-nature equation and the decomposition of fractions. Nārāyaṇa also gives a rule for finding the factors of a number by writing it as the difference of two squares, the method later associated in Europe with Fermat (Plofker 2009; Kusuba 2004).

Worked example: factors from a difference of squares

To factorise \(N = 2021\): the smallest square larger than \(N\) is \(45^2 = 2025\), and \(2025 - 2021 = 4 = 2^2\). Hence

\[2021 = 45^2 - 2^2 = (45 - 2)(45 + 2) = 43 \times 47 .\]

If the first difference is not a square, one tries \(46^2 - N\), \(47^2 - N\) and so on. The method is quickest when \(N\) has two factors close to each other, since then \(N\) lies just below a square.

10.6.4 An overview

Table 10.2 lists the principal arithmetic books discussed in this volume. Those written in Prakrit and in the regional languages are described in Chapter 13.

Table 10.2. The principal arithmetic books (pāṭīgaṇita)
Work Author and date Remarks
Bakhshālī manuscript Anonymous; date disputed (c. 7th c. by Hayashi) Non-standard Sanskrit; rules with worked and verified solutions
Pāṭīgaṇita Śrīdhara, c. 750 (uncertain) Incomplete; with an anonymous commentary; ed. Shukla (1959)
Pāṭīgaṇitasāra (Triśatikā) Śrīdhara Abridgement in about 300 verses, covering the whole syllabus
Gaṇitasārasaṅgraha Mahāvīra, c. 850 Nine chapters, more than a thousand verses; adapted into Telugu
Gaṇitatilaka Śrīpati, 11th c. Incomplete; Jain commentary by Siṃhatilaka Sūri (13th c.)
Līlāvatī Bhāskara II, 1150 The standard textbook for some seven centuries; Persian translation 1587
Gaṇitasārakaumudī Ṭhakkura Pherū, c. 1318 In Apabhraṃśa; arithmetic of the Delhi Sultanate (Chapter 13)
Gaṇitakaumudī Nārāyaṇa Paṇḍita, 1356 Fourteen chapters, ending with combinatorics and magic squares

10.7 Shared Problems: The Arithmetic Books in Comparative Perspective

Many problems in the Indian arithmetic books have close relatives in other traditions. The broken bamboo whose tip touches the ground, the reed or lotus in a pond, the channels filling a cistern and the purchase of a hundred birds for a hundred coins all appear in the Chinese Nine Chapters on the Mathematical Art or its successors; problems in which given fractions of a whole and a known remainder determine the whole are found in Egyptian papyri and in the Greek epigrams collected in the Greek Anthology; and the “hundred animals” type appears in Latin in the collection of problems attributed to Alcuin of York (c. 800). Table 10.3 lists some of these parallels (Katz 2009; Plofker 2009; Joseph 2011).

Table 10.3. Some problem types shared by the Indian arithmetic books and other traditions
Problem type Indian examples Elsewhere
False position Bakhshālī manuscript; iṣṭakarman in the Līlāvatī Rhind papyrus (Egypt, c. 1550 BCE); “excess and deficit” in the Nine Chapters
Fractions of a whole with a remainder Mahāvīra’s prakīrṇaka; the Līlāvatī’s bees Greek Anthology, book 14; medieval European collections
Broken bamboo; reed in a pond Līlāvatī Nine Chapters, chapter 9
Channels filling a cistern Mahāvīra; Līlāvatī Nine Chapters; Greek Anthology
A hundred birds for a hundred coins Mahāvīra’s “curious pulveriser” Zhang Qiujian (5th c.); Alcuin (c. 800); Abū Kāmil
Travellers in progression Bakhshālī manuscript; Mahāvīra Nine Chapters; Fibonacci’s Liber Abaci

Such parallels raise the question of transmission, and it cannot usually be answered. Some problems were natural to any society that traded, built and taxed: a merchant anywhere might ask how long several channels take to fill a tank. Others are so specific in their form—the hundred birds, the broken bamboo with its particular geometry—that independent invention seems less likely than travel along the trade routes, by word of mouth among merchants as much as in books. But the routes, the dates and even the directions of travel are unknown, and the earliest surviving version of a problem is not necessarily the original. Historians of recreational mathematics have therefore learned to speak of a shared stock of problems circulating across Eurasia, to which India contributed and from which it drew, rather than of borrowing by one tradition from another.

What was distinctive in the Indian books was not the problems themselves but their treatment. Each problem type was tied to a named rule stated in general form, the rules were arranged in a fixed syllabus, and the examples were composed as poetry to be remembered. When Fibonacci’s Liber Abaci (1202) brought the Indian numerals to Italian merchants, it also brought many problems of the kinds described in this chapter, received through Arabic sources; how many of them were Indian in origin cannot now be determined.

10.8 Assessment

The arithmetic books are the part of Indian mathematics that touched the largest number of people. Their rules for fractions, proportion, interest and mensuration were learned far more widely than the sine table or the cyclic method, and many of their procedures survived in the village schools into the nineteenth century (Chapter 13). Their methods were local rather than general: a separate rule for each type of problem, where a modern text would use one equation. But the classification of problem types was itself a form of analysis, and the best of the books—Mahāvīra’s above all—show a real interest in mathematical structure for its own sake, in unit fractions, combinations, series and “necklace” numbers. The verification of results, documented from the Bakhshālī manuscript onward, shows that the authors cared not only that a rule worked but that its answers could be checked.

Exercises

  1. Solve by false position: a man gives away a fifth of his money, then a quarter of the remainder, then a third of what is left, and has 12 left. How much had he at first?

  2. Reduce to a single fraction: a third increased by its own half, and the result diminished by its own quarter.

  3. Write \(\tfrac{5}{6}\) and \(\tfrac{4}{13}\) as sums of unit fractions by repeatedly taking the largest unit fraction that is not too large.

  4. Using Śrīdhara’s rule, find the volume of a sphere of diameter 6, and compare it with the exact value.

  5. Of a swarm of bees, a third and three times the square root of the swarm flew away, and 27 remained. How many bees were there?

  6. Factorise 5767 by writing it as a difference of two squares.

  7. Verify that \(27994681 \times 441 = 12345654321\), and explain why Mahāvīra might call such a product a “necklace”.

  8. Write 1 as a sum of four unit fractions using Mahāvīra’s rule.

Answers. (1) Assume 60: after the gifts \(60 \to 48 \to 36 \to 24\), so 24 would remain; the true remainder is 12, so the amount was \(60 \times 12/24 = 30\). (2) \(\tfrac13 \cdot \tfrac32 = \tfrac12\), then \(\tfrac12 \cdot \tfrac34 = \tfrac38\). (3) \(\tfrac56 = \tfrac12 + \tfrac13\); \(\tfrac{4}{13} = \tfrac14 + \tfrac{1}{18} + \tfrac{1}{468}\). (4) \(\tfrac{19}{36} \cdot 216 = 114\); exact value \(36\pi \approx 113.10\). (5) With \(x = y^2\): \(\tfrac23 y^2 - 3y - 27 = 0\), so by the rule \(y = \bigl(\tfrac32 + \sqrt{\tfrac94 + 18}\bigr)\big/\tfrac23 = (1.5 + 4.5) \times 1.5 = 9\), and the swarm had 81 bees. Check: \(27 + 27 = 54\) flew away and 27 remained. (6) \(76^2 = 5776\) and \(5776 - 5767 = 9 = 3^2\), so \(5767 = 73 \times 79\). (7) The product reads the same forwards and backwards, rising and falling like the beads of a necklace. (8) \(\frac12 + \frac13 + \frac19 + \frac1{18}\).

Scholarly Note: Further reading

Datta and Singh (1935–38), Part I, on arithmetic; Plofker (2009), chapters 4 and 6; Shukla (1959) for Śrīdhara’s Pāṭīgaṇita; Rangācārya (1912) for the Gaṇitasārasaṅgraha; Hayashi (1995) and Kaye (1927–33) for the Bakhshālī manuscript, with Plofker et al. (2017) on its dating; Kusuba (2004) on unit fractions; Kapadia (1937) for the Gaṇitatilaka (Datta and Singh 1935; Plofker 2009; Shukla 1959; Rangācārya 1912; Hayashi 1995; Kaye 1927; Plofker et al. 2017; Kusuba 2004; Kapadia 1937).