Chapter 7: The Decimal Place-Value System and Zero¶
The numerals used throughout the world today—ten digits whose value depends on their position, with a sign for an empty place—were first developed in India. They reached the Islamic world by the eighth century and Europe from the tenth, and by the sixteenth century they had displaced the Roman numerals and the abacus in European commerce and science. They are usually called “Hindu–Arabic” numerals, a name that records both their origin and the route by which they travelled. This chapter examines the evidence for their development in India, the separate history of zero as a number, and their transmission (Plofker 2009; Hayashi 2003; Ifrah 2000; Menninger 1969).
7.1 Three Meanings of “Zero”¶
The history of zero is often confused because the word covers three different things:
-
A placeholder: a way of showing, in a positional notation, that a place is empty, as in \(305\), where the tens place is empty.
-
A written sign for that empty place, such as a dot or a small circle.
-
A number: a quantity that can be added, subtracted, multiplied and divided according to rules, and that can be the result of a calculation, as in \(5 - 5 = 0\).
The first two had been developed in other cultures. Babylonian astronomers of the Seleucid period (from about 300 BCE) used a sign to mark an empty place in their sexagesimal (base-60) numbers, though not at the end of a number; Greek astronomers, including Ptolemy, used a symbol for an empty place in sexagesimal fractions; and the Maya of Mesoamerica had a zero sign in their positional calendrical numbers. What is distinctive in India is the combination of a decimal place-value system, a sign for zero used in every position, and—from Brahmagupta onward—rules for operating with zero as a number.
7.2 Numerals Before Place Value¶
The earliest Indian numerals, found in Brāhmī inscriptions from the third century BCE onward, were not positional. In the Nānāghāṭ cave inscription (1st century BCE) and later Brāhmī and Kharoṣṭhī inscriptions, there are separate signs for 1 to 9, for each multiple of ten from 10 to 90, for 100 and 1000, and combinations for multiples of these. To write 345 one wrote the sign for 300, then the sign for 40, then the sign for 5. Such a system, like the Greek alphabetic numerals, needs no zero. Some of the forms of the units 4 to 9 in these inscriptions are recognisable ancestors of the later place-value digits (Ifrah 2000; Plofker 2009).
7.3 The Evidence for Place Value¶
7.3.1 Word-numerals¶
The first evidence for decimal place value in India comes not from written digits but from words. In the word-numeral (bhūtasaṃkhyā) system described in Chapter 1, digits are represented by words and listed from the units upward, and a word for “empty” (śūnya, kha, ambara) is used for an empty place. Wherever such word-numerals appear with a word for zero, place value is implied.
-
The Yavanajātaka of Sphujidhvaja, a Sanskrit verse adaptation of a Greek astrological work, uses word-numerals including śūnya. David Pingree dated it to 269/270 CE on the basis of a verse that he read as giving the date. In 2013, Bill Mak, using a newly identified manuscript, argued that the verse had been misread and that the text cannot be dated so precisely. The Yavanajātaka therefore no longer provides a secure third-century date for zero (Mak 2013).
-
The Lokavibhāga, a Digambara Jain cosmological text, states that it was copied in Śaka 380 (458 CE) and uses word-numerals with śūnya in a place-value way. The surviving Sanskrit text is a later version of an earlier Prakrit work, and how much of it goes back to 458 is debated.
-
Āryabhaṭa (499) names the decimal places, extracts square and cube roots by place-value methods, and designs his alphabetic notation on a decimal basis (Chapter 3).
-
Varāhamihira’s Pañcasiddhāntikā (mid-6th century) and the astronomers after him use word-numerals with zero routinely.
By the sixth century, then, the decimal place-value system with a concept of zero was in standard use among Indian astronomers.
7.3.2 The Bakhshālī manuscript¶
The Bakhshālī manuscript uses written place-value numerals with a dot for zero throughout. Its date is disputed (Section 2.4); on Hayashi’s dating it belongs with the evidence of the seventh century, and it does not, on present evidence, show zero in use in the third.
7.3.3 Inscriptions¶
Inscriptions, which often carry dates, are the most reliable evidence for the use of written place-value numerals.
| Inscription | Date | Notes |
|---|---|---|
| Copper-plate grants from western India | 6th–7th c. (stated dates) | Several grants, such as one dated in year 346 of the Kalacuri era (594/5 CE), write the date in place-value numerals. Some are suspected of being later forgeries or recopies, and their evidence is treated with caution |
| Sambor-on-Mekong, Cambodia (K. 127) | Śaka 605 (683 CE) | Old Khmer inscription; the year written with a dot for zero |
| Kedukan Bukit, Palembang, Sumatra | Śaka 605 (682/3 CE) | Old Malay inscription of Śrīvijaya; place-value date with zero |
| Kota Kapur, Bangka, Indonesia | Śaka 608 (686 CE) | Śrīvijaya inscription; place-value date with zero |
| Caturbhuj temple, Gwalior | 876 CE (Vikrama 933) | Two inscriptions recording land and gifts to a temple; the numbers 270 and 50 written with a small circle for zero. The earliest undisputed example in India itself |
The Southeast Asian inscriptions are dated in the Śaka era of Indian origin and use numerals derived from Indian forms. Since the system must have been established in India before it was exported, they show that it was in use there by the seventh century, even though the earliest undisputed Indian inscription is two centuries later (Plofker 2009; Ifrah 2000).
7.3.4 Testimony from outside India¶
In 662 CE the Syrian bishop Severus Sebokht, writing at the monastery of Qenneshre on the Euphrates, praised the astronomy of the Indians and their methods of calculation, adding that “this computation is done by means of nine signs” (Ifrah 2000). In 718 the Chinese Jiuzhi li, compiled by the Indian astronomer Gautama Siddha at the Tang court, described Indian numerals and the use of a dot for an empty place (Chapter 12). Both show that the Indian system was known outside India by the seventh and early eighth centuries.
7.4 Zero as a Number¶
The decisive step from zero as a sign to zero as a number was taken in the arithmetic of Brahmagupta (628 CE), discussed in Chapter 4. Brahmagupta gave rules for adding, subtracting and multiplying with zero that agree with modern arithmetic, and attempted rules for division. His successors continued to work on the problem of division by zero.
| Author | Rule |
|---|---|
| Brahmagupta (628) | \(0 \div 0 = 0\); \(a \div 0\) is a quantity “with zero as divisor” (khaccheda), value not stated |
| Mahāvīra (c. 850) | \(a \div 0 = a\): a quantity divided by zero remains unchanged |
| Bhāskara II (1150) | \(a \div 0\) is khahara; it is not changed by adding or subtracting any quantity, “as there is no change in the infinite, unchanging [God] when worlds are created or destroyed” |
None of these is the modern position, in which division by zero is simply undefined, while limits such as \(a/x\) as \(x\) approaches \(0\) can grow without bound. Bhāskara’s description of \(a/0\) as an unchanging infinite quantity is the closest to later ideas, and his religious simile is one of the few places where a mathematical text links zero or infinity with theology. It is an analogy, not a mathematical theory of the infinite, and it should not be presented as an anticipation of the extended real line or of Cantor’s transfinite numbers (Plofker 2009; Colebrooke 1817).
7.5 Why India?¶
Why did a decimal place-value system with zero develop in India? The evidence does not allow a confident answer, and several factors have been proposed:
-
Decimal number words. Sanskrit, like other Indo-European languages, has number words built on powers of ten, and the Vedic lists of names for powers of ten (Chapter 2) show long familiarity with decimal structure.
-
Word-numerals. Writing numbers in verse as sequences of words, units first, is in effect a positional notation, and it requires a word for an empty place.
-
The dust board. Calculation on a board covered with sand or dust (pāṭī), where digits were written in columns and erased as the work proceeded, favours a notation in which each column holds a single digit.
-
Contacts with other systems. Indian astronomers learned from Hellenistic astronomy, which used sexagesimal place value with a zero symbol for fractions (Chapter 12). Chinese mathematicians calculated with counting rods arranged in decimal positions, with a blank for an empty place. Whether either influenced the Indian development is debated; there is no direct evidence of borrowing, and the Indian system differs from both.
It is often said that the Indian philosophical concept of emptiness (śūnyatā) in Buddhist Madhyamaka thought, or the Upaniṣadic idea of a formless absolute, prepared Indian culture to accept a number for “nothing”. The word śūnya is indeed shared, and the idea is attractive. But no early mathematical text connects the zero of calculation with these philosophical doctrines, and the philosophers did not write about the zero of arithmetic. The connection remains a suggestive hypothesis, not a demonstrated influence.
7.6 Calculation on the Dust Board¶
Indian arithmetic texts describe operations designed for an erasable surface. In multiplication by the “door-junction” method (kapāṭasandhi), the multiplicand and multiplier are written one above the other, like the two leaves of a door, and the partial products are written in and digits erased as the multiplier is moved along. Division and root extraction likewise overwrite the working figures. When these methods were adopted in the Islamic world, they were first used on a dust board (takht); in the tenth century Abū’l-Ḥasan al-Uqlīdisī (952) adapted them for pen and paper, noting that the dust board was associated with street astrologers. European “scratch” methods of division, in which digits are crossed out, descend from these erasure methods. Indian texts also describe several methods of multiplication, including one equivalent to the lattice method later popular in Europe (Plofker 2009; Datta and Singh 1935).
Several ways to multiply: \(135 \times 12\)
Bhāskara II’s Līlāvatī illustrates multiplication with the product of 135 and 12, worked in several ways.
-
Digit by digit: \(135 \times 1 = 135\) (in the tens place) and \(135 \times 2 = 270\); adding with the proper place values, \(1350 + 270 = 1620\).
-
By parts of the multiplier: \(12 = 8 + 4\), so \(135\times 8 + 135 \times 4 = 1080 + 540 = 1620\).
-
By factors of the multiplier: \(12 = 3 \times 4\), so \((135 \times 3) \times 4 = 405 \times 4 = 1620\).
-
By a difference: \(12 = 10 + 2\), or \(12 = 20 - 8\): \(135 \times 20 - 135 \times 8 = 2700 - 1080 = 1620\).
The variety of methods reflects an arithmetic in which the place-value system was fully understood and could be manipulated flexibly.
7.7 Fractions¶
Indian arithmetic wrote fractions with the numerator above the denominator, without a dividing line; a mixed number was written with the whole number above both, so that \(2\tfrac{3}{4}\) appeared as the column 2, 3, 4. The Bakhshālī manuscript and the arithmetic texts use this notation. The texts classify the operations needed to reduce complicated fractional expressions to simple fractions into several standard types (jāti): a fraction of a whole number, a fraction of a fraction, a whole number increased or diminished by a fraction, and so on. Common denominators were found by multiplying denominators or, in later texts, by using the least common multiple. When the notation passed to the Islamic world, Arabic writers kept the vertical arrangement, and in North Africa a horizontal bar was added between numerator and denominator by the twelfth century; from there the bar reached Europe.
Worked example: a fraction of a fraction
What is half of a third of three-quarters of a dramma, in cowries?
\(\tfrac12 \times \tfrac13 \times \tfrac34 = \tfrac{3}{24} = \tfrac18\) of a dramma. Since a dramma is \(16 \times 4 \times 20 = 1280\) cowries (Chapter 1), the answer is 160 cowries.
7.8 Why Place Value Matters¶
The advantage of the place-value system is easiest to see by comparison. In Roman numerals, \(1648\) is MDCXLVIII and \(37\) is XXXVII. To multiply them one must either convert them to some other representation or work on a counting board with pebbles; the numerals themselves give no help, because the value of a symbol does not depend on its position. In the place-value system, by contrast, multiplication reduces to a table of products of single digits (from \(1 \times 1\) to \(9 \times 9\)) and the rule for shifting by places:
Every step uses only the multiplication table, addition with carrying, and the shift of a digit one place to the left. The same is true of division, of roots, and of operations with fractions. The written numbers are the working, so that a calculation can be checked and taught. This is why the place-value system, once available, displaced the abacus and the counting board wherever it went, and why it made possible the large computations of later astronomy and science.
The system also changes how numbers can be thought about. Because any number, however large, is written with the same ten symbols, the distinction between “small” numbers that can be written and “large” ones that must be described in words disappears. The Indian lists of named powers of ten—and the names eka, daśa, śata, … for the places, which Āryabhaṭa and later authors give—show the transition: the names of powers of ten became the names of places in a written number.
7.9 Reading Dates: Eras in Indian Texts and Inscriptions¶
Indian texts and inscriptions date events in several eras, and the conversion of their dates is a small exercise in place-value arithmetic.
-
The Kali era, used in astronomy, begins in 3102 BCE. Āryabhaṭa’s statement that 3600 years of the Kali age had elapsed (Chapter 3) corresponds to \(3600 - 3101 = 499\) CE (there being no year zero between 1 BCE and 1 CE).
-
The Śaka era begins in 78 CE. The Sambor inscription’s year 605 is \(605 + 78 = 683\) CE, and Brahmagupta’s Śaka 550 is 628 CE.
-
The Vikrama era begins in 57 BCE. The Gwalior inscription’s year 933 corresponds to \(933 - 57 = 876\) CE.
These conversions are approximate to within a year, because Indian years begin at different points from the Western calendar year. They show, however, why accurate dates in inscriptions are so valuable for the history of the numerals: an inscription that gives its date in place-value numerals, in a known era, fixes a point at which the system was in use.
7.10 Transmission to the Islamic World and Europe¶
7.10.1 The Islamic world¶
Indian numerals and methods of calculation were known in Baghdad by the late eighth century, probably arriving with the astronomical texts translated at that time (Chapter 12). Around 825 Muḥammad ibn Mūsā al-Khwārizmī wrote a book on Indian reckoning. Its Arabic original is lost, but a twelfth-century Latin version, beginning Dixit Algorizmi (“Al-Khwārizmī said”), survives; from the Latinised form of his name comes the word “algorithm”. Other Arabic arithmetics followed, among them al-Uqlīdisī’s Kitāb al-fuṣūl fī al-ḥisāb al-hindī (952). Al-Uqlīdisī’s book is also notable for being among the earliest known works to use decimal fractions, marking the units place with a sign so that digits to its right could represent tenths, hundredths and so on—an extension of the Indian place-value principle that Indian texts themselves had not made. The forms of the digits developed differently in the east of the Islamic world (the “Arabic–Indic” digits still used in the Middle East) and in the west, in North Africa and Spain, where the so-called ghubār (“dust”) forms gave rise to the European digits (Plofker 2009; Ifrah 2000).
In the eastern Arabic–Indic digits still used in much of the Arab world, zero is written as a dot—the older Indian form—while the digit five is written as a small circle; in the western forms, and hence in the European digits, zero became a circle.
7.10.2 The chessboard problem¶
Al-Bīrūnī, in a treatise on Indian numbers and elsewhere, discussed the famous problem of the chessboard: if one grain of wheat is placed on the first square, two on the second, four on the third, and so on, doubling each time, how many grains are there on all sixty-four squares? The answer is
which al-Bīrūnī computed and checked by several methods. The problem, attached in later Persian and Arabic literature to legends about the Indian origin of chess, shows how the Indian numerals made such enormous calculations routine.
7.10.3 Europe¶
-
976: The Codex Vigilanus, copied in northern Spain, contains the earliest known European example of Hindu–Arabic numerals, the digits 1 to 9, without zero.
-
c. 1000: Gerbert of Aurillac (later Pope Sylvester II) used counters marked with these digits on an abacus.
-
12th century: Latin translations of Arabic arithmetics, including the version of al-Khwārizmī, introduced the full system with zero to scholars.
-
1202: Leonardo of Pisa (Fibonacci), who had learned the system as a boy in North Africa, wrote the Liber Abaci. It opens by introducing the nine figures of the Indians together with the sign 0, called zephirum in Arabic, with which any number can be written. The book did much to spread the numerals among Italian merchants.
-
1299: The statutes of the Arte del Cambio, the bankers’ guild of Florence, forbade the use of the new numerals in account books, apparently because they were easier to falsify; amounts had to be written out in words or Roman numerals.
-
1478–1494: Printed arithmetic books (the Treviso Arithmetic, 1478; Luca Pacioli’s Summa, 1494) taught calculation with the new numerals, and printing standardised the forms of the digits.
By the sixteenth century the Hindu–Arabic numerals were general in European mathematics and commerce, although Roman numerals survived in some uses for much longer.
7.10.4 The words “zero” and “cipher”¶
The Sanskrit śūnya, “empty”, was translated into Arabic as ṣifr, which also means “empty”. Latin writers turned ṣifr into zephirum (as in Fibonacci) and cifra. From zephirum, through Italian zefiro and zero, comes the English “zero”; from cifra comes “cipher”, which first meant zero, then any digit, and later a secret code.
7.10.5 Within India¶
The spread of the place-value numerals within India was itself gradual. The modern scripts of India—Devanagari, Bengali, Gujarati, Gurmukhi, Oriya, Telugu, Kannada, Malayalam and others—all have their own forms of the ten digits, descended from the same Brāhmī ancestors. Tamil is an instructive exception: traditional Tamil numerals, like the early Brāhmī ones, were not positional, and had separate signs for ten, a hundred and a thousand; a sign for zero and positional writing came into use in Tamil only in modern times, under the influence of the other systems. Different regions and communities thus adopted the new system at different times, and older notations survived for a long time in accounts and documents.
7.10.6 Southeast Asia¶
The system also spread eastward. The numerals of the Khmer, Thai, Lao, Burmese, Javanese and Balinese scripts all descend from Indian forms, and Southeast Asian inscriptions provide some of the earliest dated examples of zero (Table 7.1).
7.11 Comparison of Numeral Systems¶
Table 7.3 sets the Indian system beside some of the others.
| System | Base | Positional? | Zero |
|---|---|---|---|
| Egyptian hieroglyphic | 10 | No | None needed |
| Babylonian | 60 | Yes | Placeholder sign in Seleucid period, not used at the end of a number |
| Greek alphabetic | 10 | No | None needed (Greek astronomers used a zero symbol in sexagesimal fractions) |
| Roman | 10 | No | None |
| Chinese counting rods | 10 | Yes | Blank space; written circle from the 13th century |
| Maya (calendrical) | 20 (modified) | Yes | Shell-shaped sign |
| Indian | 10 | Yes | Dot or circle in all positions; zero also treated as a number |
The Indian system combined the advantages of the others: a small set of symbols, a base that matched the spoken number words, a sign for zero usable in any position, and methods of calculation adapted to it. This combination, rather than any single element, explains why it spread so widely.
7.12 Assessment¶
The development of the decimal place-value system with zero was gradual. Decimal number words are Vedic; word-numerals with zero were in use among astronomers by the fifth or sixth century; written place-value numerals with zero are attested in dated inscriptions from the seventh century (in Southeast Asia) and the ninth (in India), and in the Bakhshālī manuscript, whose date is disputed; and zero was treated as a number with its own rules by Brahmagupta in 628. The system spread to the Islamic world and through it to Europe, and eastward to Southeast Asia and, briefly, China. It is one of India’s most important contributions to the world’s mathematics. Two questions remain open. The first is the date at which written place-value numerals came into use in India itself, which turns on the Bakhshālī manuscript and the suspect copper plates; the second is why the system arose where it did, for which the dust board and the word-numerals are better candidates than any philosophical doctrine (Plofker 2009; Ifrah 2000).
Exercises¶
-
Convert to the Common Era: Śaka 1036; Vikrama 1000; Kali 4000.
-
Read the word-numeral rūpa-kha-netra-agni, digits from the units upward.
-
Multiply 246 by 15 by three of the methods of the Līlāvatī: by digits, by parts (\(15 = 10 + 5\)) and by factors (\(15 = 3\times 5\)).
-
Write 1648 and 37 in Roman numerals, and explain why their product cannot be computed directly in that notation.
-
How many grains are there on the first ten squares of the chessboard in the doubling problem?
Answers. (1) 1114 CE (Bhāskara II’s year of birth); 943 CE; 899 CE. (2) The digits 1, 0, 2, 3 give 3201. (3) \(3690\) in each case. (4) MDCXLVIII and XXXVII. (5) \(2^{10} - 1 = 1023\).
Scholarly Note: Further reading
Plofker (2009), chapter 3, gives a careful account of the evidence for place value and zero in India; Ifrah (2000) and Menninger (1969) survey numeral systems worldwide, though Ifrah’s treatment of Indian material should be used with caution on dates; Hayashi (1995) and Plofker et al. (2017) discuss the Bakhshālī manuscript; Mak (2013) re-examines the date of the Yavanajātaka (Plofker 2009; Ifrah 2000; Menninger 1969; Hayashi 1995; Plofker et al. 2017; Mak 2013).