Chapter 12: Handbooks, Tables and the Working Astronomer¶
12.1 From Treatise to Handbook¶
The great siddhāntas described in the preceding chapters were not, for the most part, the books from which almanacs, horoscopes and eclipse warnings were actually computed. A siddhānta counts its days from the beginning of the kalpa or of the Kali age, so that the astronomer who follows it must multiply a day-count of seven digits or more by revolution numbers of up to ten digits and divide by a number of civil days of ten digits, keeping the remainder. The work is long and errors are easy. From the seventh century onwards the practical astronomer worked instead with a handbook (karaṇa) or with tables (sāraṇī), and it was in these forms that most of the tradition reached most of its users (Pingree 1981; Plofker 2009).
Later writers classified astronomical works by the epoch from which they count: a siddhānta in the strict sense counts from the beginning of the kalpa, a tantra from the beginning of the current yuga, in practice usually the Kali age (Chapter 1), and a karaṇa from a recent date, usually in the author’s own lifetime. The recent epoch is the defining feature of the handbook, and most of its other features follow from it:
-
the handbook states the mean positions of the Sun, Moon, planets, apogees and nodes at its epoch (the initial values, called kṣepaka or dhruva), so that only the days elapsed since then need to be counted;
-
the long period relations of the siddhānta are replaced by small ratios or by motions per day, month or year, chosen to give nearly the same results over a few centuries;
-
the rules are shortened: Sine tables are reduced or replaced by approximations, iterations are cut to a single step, and the theoretical chapters on the sphere and on cosmography are omitted or reduced;
-
the contents are those that practice required: true positions, the elements of the almanac, eclipses, the visibility of the Moon and planets, conjunctions and the shadow of the gnomon.
The titles of the handbooks advertise their convenience. Khaṇḍakhādyaka names a sweetmeat made with sugar, presumably a promise that the subject will go down easily; Laghumānasa is “the short Mānasa”; Karaṇakutūhala means something like “the delight of computation”; and Grahalāghava, “lightness in the planets”, promises ease in computing them. Judging by the manuscripts recorded in Pingree’s Census, the most successful handbooks and tables were copied far more often than all but a few siddhāntas (Pingree 1970–1994). This chapter describes how they worked, how the astronomers who used them earned their living and handed on their knowledge, and how the tradition dealt with the slow drift of computation away from the sky.
12.2 The Khaṇḍakhādyaka¶
The oldest surviving handbook, and one of the most influential, is the Khaṇḍakhādyaka of Brahmagupta, whose epoch falls in Śaka 587 (665 CE). Brahmagupta was then sixty-seven, and thirty-seven years had passed since his Brāhmasphuṭasiddhānta (Chapter 9). The handbook has a surprising basis. It does not follow Brahmagupta’s own system but Āryabhaṭa’s midnight (ārdharātrika) system, the very school Brahmagupta had criticised at length in his siddhānta. The usual explanation is practical: the midnight system was the one in common use in the region where he worked, and a handbook was useful only if its results agreed with the almanacs and the eclipse times that its users already knew. Brahmagupta added a supplement, the Uttarakhaṇḍakhādyaka, in which he corrected some of the rules of the first part, partly in the direction of his own earlier work (Sengupta 1934; Pingree 1978).
The Khaṇḍakhādyaka shows how a handbook compresses a siddhānta. In a siddhānta the number of intercalary months that have accumulated is found by multiplying the elapsed solar months by the number of intercalary months in the yuga (1,593,336) and dividing by the number of solar months (51,840,000). Brahmagupta’s handbook does the same with small numbers. In effect it multiplies the elapsed solar months by 2 and divides by 65, with a small additive constant that fixes the epoch; in the same way it finds the omitted tithis by multiplying the elapsed tithis by 11 and dividing by 703. (The rules in the text are worded with their own multipliers and epoch constants, but these are the ratios they amount to.) The approximations are excellent:
the second value being the proportion of omitted tithis implied by the period relations of the Āryabhaṭa school. Both fractions are successive approximations of the kind produced by the continued-fraction methods of Indian arithmetic, though the handbook does not say how they were found. The first gains one intercalary month on the exact ratio only in about 2,500 years; with the epoch constant reset by each new handbook, the error never matters.
Other parts of the work were equally practical. Its rule for interpolating in a table by second differences is described in Chapter 6. Its eclipse rules were used for centuries, and its commentators, among them Pṛthūdakasvāmin (864) and Bhaṭṭotpala (tenth century), wrote worked examples for students (Table 9.1). The book also travelled: an Arabic version made in Sind and known as the Zīj al-Arkand, from Sanskrit ahargaṇa, was among the first Indian astronomical works in Arabic (Chapter 16). P. C. Sengupta’s English translation (Sengupta 1934) remains the most accessible version.
12.3 Later Handbooks¶
Table 12.1 lists some of the principal handbooks. Their authors and epochs are well established; their schools and regions of use are given only where the texts or their manuscript tradition make them clear. Several of these authors have already appeared in Chapter 9; here the interest is in what their handbooks did.
| Handbook | Author | Epoch | Remarks |
|---|---|---|---|
| Khaṇḍakhādyaka | Brahmagupta | 665 | Ārdharātrika school; Arabic Zīj al-Arkand |
| Laghumānasa | Muñjāla | 932 | very short; an additional lunar correction |
| Rājamṛgāṅka | attributed to King Bhoja of Dhārā | 1042 | a handbook issued in a king’s name |
| Karaṇaprakāśa | Brahmadeva | 1092 | Ārya school |
| Bhāsvatī | Śatānanda, Purī | 1099 | Saura parameters of the Pañcasiddhāntikā; decimal fractions |
| Karaṇakutūhala | Bhāskara II | 1183 | Brāhma school; also called Brahmatulya |
| Grahalāghava | Gaṇeśa Daivajña, Nandigrāma | 1520 | cycles of 4016 days; no Sine table |
| Vākyakaraṇa | anonymous, Tamil country | c. 14th c. | vākya system (Chapter 13) |
| Karaṇapaddhati | Putumana Somayājī, Kerala | uncertain | derivation of the vākyas (Chapter 10) |
Three of them show the range of the genre.
The Bhāsvatī.
Śatānanda, writing at Purī in Odisha, based his handbook of 1099 on the old Sūryasiddhānta as summarised five centuries earlier in Varāhamihira’s Pañcasiddhāntikā (Chapter 5), not on the later text of that name. His innovation was arithmetical. Instead of sexagesimal fractions he expressed the motions in hundredths, so that most of the computation became a matter of multiplying, adding and shifting decimal places. The handbook was much copied and commented on, which shows that an archaic set of parameters could live on for centuries if it was packaged conveniently.
The Karaṇakutūhala.
Bhāskara II wrote his handbook in 1183, at the age of sixty-nine, thirty-three years after his Siddhāntaśiromaṇi. Its alternative title, Brahmatulya (“equal to the Brāhma [system]”), states its claim: it reproduces the results of the Brāhma school by easier methods. It was used for almanacs in parts of western and northern India.
The Grahalāghava.
Gaṇeśa Daivajña, son of Keśava of Nandigrāma on the western coast, wrote in 1520 the most radical simplification of all, and one of the most widely used handbooks. It has two famous devices: it does without a Sine table, replacing trigonometric functions by algebraic approximations, and it counts days in cycles (cakras) of 4016 days. The choice of cycle is ingenious. Four thousand and sixteen days are very nearly 136 synodic months (4016.2 days) and eleven sidereal years less about two days (11 × 365.26 = 4017.8 days). After one cycle, therefore, the mean Sun and the mean Moon each fall only a few degrees short of their starting positions, and the tithi has nearly returned to its starting value. The user divides the days elapsed since the epoch by 4016, multiplies the number of complete cycles by small corrections stated in the text (about \(-1.8^\circ\) for the Sun and \(-3.8^\circ\) for the Moon), and adds the motion for the remaining days, never handling a number of more than four or five digits. To take an illustration with the rates of the Sūryasiddhānta (the Grahalāghava’s own differ slightly), a count of 175,000 days from the epoch is 43 cycles and 2312 days. The Sun’s correction for one cycle is \(-1^\circ 49' 11”\), so for 43 cycles it is \(-78.25^\circ\); the motion for 2312 days is \(118.71^\circ\) after whole revolutions are removed; and the sum, \(40.46^\circ\), added to the Sun’s position at the epoch, is exactly what the long multiplication \(175{,}000 \times 0;59,8,10\) would give. The Grahalāghava remained the basis of many almanacs in Maharashtra, Gujarat and Karnataka into the twentieth century (Chapter 13).
12.4 Counting the Days: A Worked Example¶
The first step of every computation, in siddhānta and handbook alike, is the ahargaṇa, the number of civil days from the epoch to the day required (Chapter 6). An almanac-maker did not start from a modern calendar date. He started from the date as his own calendar expressed it, a lunar month, a tithi and a year of an era, and converted it into days by the period relations of his system. The worked computation of Appendix A takes its ahargaṇa from modern tables; the example below shows how the same number is obtained by the rules of the Sūryasiddhānta itself (Burgess 1860).
Example 12.1 (The day-count for 1 January 2000 by the rules of the Sūryasiddhānta)
At midnight beginning 1 January 2000, Kali year 5100 had elapsed (Kali years \(\approx\) CE \(+ 3101\), the year beginning in March–April), and in the current lunar year eight months, Caitra to Kārttika, had elapsed. The current month was Mārgaśīrṣa (in the amānta reckoning), and its twenty-fifth tithi, the tenth of the dark fortnight, was running, so that 24 tithis of the month were complete. The Sūryasiddhānta gives, for a mahāyuga, 51,840,000 solar months, 1,593,336 intercalary months, 1,603,000,080 lunar days (tithis) and 1,577,917,828 civil days, so that 25,082,252 tithis are omitted.
-
Elapsed solar months: \(5100 \times 12 + 8 = 61{,}208\).
-
Intercalary months: \(61{,}208 \times 1{,}593{,}336 / 51{,}840{,}000 = 1881.27\); the integer, 1881, is kept.
-
Lunar months: \(61{,}208 + 1881 = 63{,}089\).
-
Elapsed tithis: \(63{,}089 \times 30 + 24 = 1{,}892{,}694\).
-
Omitted tithis: \(1{,}892{,}694 \times 25{,}082{,}252 / 1{,}603{,}000{,}080 = 29{,}615.11\); the integer, 29,615, is kept.
-
Civil days: \(1{,}892{,}694 - 29{,}615 = 1{,}863{,}079\).
This is exactly the value obtained in Appendix A from the Julian day numbers. Astronomers checked such a result against the weekday. The Kali epoch fell on a Friday; \(1{,}863{,}079 = 7 \times 266{,}154 + 1\), so the day is one weekday after a Friday, a Saturday, which is correct. Had the check failed, a day would have been added or subtracted, since rounding at steps 2 and 5 can leave the count one day out.
Mean positions.
With the ahargaṇa known, the mean longitude follows from Equation (6.2). For the Sun, \(1{,}863{,}079 \times 4{,}320{,}000 / 1{,}577{,}917{,}828 = 5100.7100\) revolutions; the fraction \(0.7100\) of a revolution is \(255.61^\circ\), or \(15^\circ 37'\) in Dhanus. For the Moon, the same computation with 57,753,336 revolutions gives \(68{,}190.5138\) revolutions and a mean longitude of \(184.98^\circ\). These agree with Table A.1. In practice the multiplication was done with the daily motions in sexagesimal form (0;59,8,10 for the Sun, 13;10,34,52 for the Moon; Table 6.2), keeping as many places as the desired accuracy required.
The handbook’s shortcut.
A handbook avoids the large numbers. Suppose an author had fixed his epoch at midnight beginning 17 April 1999, day 1,862,820 of the Kali era, a few hours after the mean Sun had entered Meṣa. He would compute once, and record in his book, the mean positions at that moment: for the Sun \(0^\circ 20'\), for the Moon \(12^\circ 18'\). A user wanting 1 January 2000 needs only the 259 days elapsed since the epoch. The Sun moves \(259 \times 0;59,8,10 = 255^\circ 16'\), so its mean longitude is \(0^\circ 20' + 255^\circ 16' = 255^\circ 36'\); the Moon moves \(259 \times 13;10,34,52 = 3412^\circ 41'\), which is \(172^\circ 41'\) after removing whole revolutions, so its mean longitude is \(12^\circ 18' + 172^\circ 41' = 184^\circ 59'\). The results are the same as before, but every number now has at most four digits before the fractional part. This is what a karaṇa offered: the same astronomy, at a fraction of the labour, valid for as long as its epoch values and rates stayed close to the sky.
The day-count outside astronomy.
The ahargaṇa was familiar enough to serve as a date. In Kerala the number of days elapsed since the beginning of the Kali age (kalidina) was often recorded in a kaṭapayādi phrase (Chapter 1). The best-known example closes Nārāyaṇa Bhaṭṭatiri’s devotional poem Nārāyaṇīyam, whose last words, āyurārogyasaukhyam (“long life, health and happiness”), are traditionally read as the day of its completion. The syllables give the digits 0 (the initial vowel), 1 (yu), 2 (rā), 2 (ro), 1 (gya, whose last consonant is ya), 7 (sau) and 1 (khya); read from right to left they make 1,712,210. Counted from the Kali epoch used in the example above, that day falls in late 1586 CE. The phrase is at once a blessing, a date and a small exercise in the arithmetic of the almanac.
12.5 Tables¶
Tables (sāraṇī; the cells of a table are koṣṭhakas, “compartments”) went one step further than handbooks. A handbook gives rules to be applied; a table gives the results of applying them, so that the user enters with an argument, reads off a value, and interpolates. Sanskrit tables became common in the second millennium and survive in large numbers, mostly on paper, laid out in grids of sexagesimal numbers in Devanāgarī or other regional numerals (Montelle and Plofker 2018).
The main kinds are tables of mean motion (the motion of each body in so many years, months and days, to be added together); tables of the equations (phala) of the manda and śīghra corrections against the anomaly; tables for the elements of the almanac, which give the tithi, nakṣatra and yoga almost directly; and tables for eclipses and for the visibility of the Moon and planets. The Makaranda tables of Vārāṇasī (epoch 1478), based on the Sūryasiddhānta, became the standard basis of almanacs in much of northern and eastern India, and Gaṇeśa Daivajña, besides his handbook, produced tables for the tithis (the Tithicintāmaṇi, 1525). Many such sets were compiled by almanac-makers for their own use and never circulated widely. Pingree catalogued the Sanskrit tables in the collections of the United States and of England (Pingree 1968, 1973), and Montelle and Plofker (2018) have given the first general account of their types and construction.
An equation table shows both the convenience and the limits of the form. Table 12.2 gives, at intervals of \(15^\circ\) of anomaly, the equation of the centre of the Sun computed from the rules of the Sūryasiddhānta, whose epicycle for the Sun shrinks from \(14^\circ\) at the apsides to \(13^\circ 40'\) at the quadrants (Chapter 7).
| Anomaly | Epicycle | Equation |
|---|---|---|
| \(0^\circ\) | \(14^\circ 0'\) | 0;0,0 |
| \(15^\circ\) | \(13^\circ 55'\) | 0;34,23 |
| \(30^\circ\) | \(13^\circ 50'\) | 1;6,3 |
| \(45^\circ\) | \(13^\circ 46'\) | 1;32,57 |
| \(60^\circ\) | \(13^\circ 43'\) | 1;53,25 |
| \(75^\circ\) | \(13^\circ 41'\) | 2;6,12 |
| \(90^\circ\) | \(13^\circ 40'\) | 2;10,32 |
For an anomaly of \(50^\circ\) the user interpolates between the entries for \(45^\circ\) and \(60^\circ\). The difference is \(20'28”\), a third of which, \(6'49”\), added to \(1;32,57\) gives \(1;39,46\). The exact value is \(1;40,34\), so simple interpolation over so wide an interval is out by \(48”\), because the equation does not grow linearly. A table-maker could reduce the error by tabulating at intervals of a few degrees, or the user could apply a second-difference rule like Brahmagupta’s (Chapter 6). The choice between larger tables and better interpolation is one that every table-making tradition faced.
12.6 The Jyotiṣī and the Profession¶
The people who computed with these books had many names. The Sanskrit gaṇaka (“calculator”), daivajña (“knower of fate”), sāṃvatsara (“man of the year”, that is, the calendar-maker) and jyotiṣī survive in regional forms and in family names: the Marathi and Gujarati jośī or Joshi is the jyotiṣī, and Tamil kaṇiyaṉ, “astrologer”, comes from gaṇaka. The word is old in Tamil: the Caṅkam poet Kaṇiyaṉ Pūṅkuṉṟaṉ bears it as part of his name. In Kerala the Kaṇiyār (Kaṇisan) community were hereditary astrologers, and in the Tamil country the Vaḷḷuvar (Chapter 13); the profession was thus not exclusively Brahmin, although the authors of Sanskrit astronomical texts almost always were (Chapter 1).
The working jyotiṣī lived from services, not from research. He computed and sold the year’s almanac, and in many regions read it aloud on New Year’s day, a custom still kept in the Telugu and Kannada country as pañcāṅga-śravaṇa, “hearing the almanac”. He fixed the auspicious times (muhūrtas) for weddings, journeys, house-building and the sowing of crops, cast and interpreted horoscopes, announced eclipses with their ritual times of beginning and end, and advised on the dates of festivals and fasts. A few served courts or temples for a stipend or a grant of land; most served a clientele of households. In the villages of the Deccan the jośī was often counted among the hereditary village servants (balutedār) who received a customary share of the harvest in return for their services, which included reading the almanac and fixing the times of household rites. From the later nineteenth century printed almanacs, issued by family firms and publishing houses, carried the computations of a few such families to very large numbers of readers (Chapter 13), and the village calendar-maker increasingly became a reader and interpreter of an almanac computed elsewhere. Astrology was therefore the economic base of astronomy, and the most technically accomplished astronomers wrote astrological works too (Chapter 15).
Skill and books passed within families. Chapter 9 has described the family of Bhāskara II and the school endowed for his descendants. The pattern can be followed in more detail in western India and Vārāṇasī in the sixteenth and seventeenth centuries, where the authors’ own statements allow genealogies to be reconstructed (Pingree 1981; Minkowski 2004). Keśava of Nandigrāma was the father and teacher of Gaṇeśa Daivajña. A family from Golagrāma in Maharashtra, descended from Divākara, who had studied with Gaṇeśa, produced authors over four generations, among them Nṛsiṃha (born 1586), who wrote a commentary on the Siddhāntaśiromaṇi, and his son Kamalākara. A rival family from Dadhigrāma, also in Maharashtra, included Raṅganātha, author of a commentary on the Sūryasiddhānta (1603), and his son Munīśvara. Both families settled in Vārāṇasī, and the controversy between Kamalākara and Munīśvara (Chapter 11) was in part a rivalry between lineages. Such families kept their own manuscripts, trained their sons with them, and wrote commentaries full of worked examples (udāharaṇa) for the next generation; several members of the Golagrāma family are credited with such example-commentaries on the handbooks.
The tradition had its own ideal of the practitioner. The second chapter of Varāhamihira’s Bṛhatsaṃhitā describes the sāṃvatsara a king should employ: of good family and conduct, truthful, learned in all three branches of the science, and able to answer questions on any part of it before a learned assembly; and it warns against the ignorant pretender, whose predictions do harm (Varāhamihira 1981–1982). The ideal combined mathematical competence with moral standing, because the practitioner’s computations governed other people’s rites and decisions.
In practice training was more modest. A pupil memorised a handbook, learned its rules on worked examples, and computed under supervision; mastery of a full siddhānta, of spherical astronomy and of the demonstrations (upapatti) was the mark of a learned astronomer rather than a requirement of the trade. Bhāskara II, at the opening of his Golādhyāya, warns that an astronomer who computes without knowing the demonstrations will neither win respect in learned assemblies nor be free of doubt himself; the warning was needed because many practitioners computed by rote.
12.7 Manuscripts and Collections¶
The whole of this literature survives in manuscript copies, and its history is in large part the history of copying (Chapter 1). Astronomical manuscripts have two features that make them unusual. First, their colophons are often precisely dated by the copyist in the terms of the calendar itself: an era year, a lunar month and fortnight, a tithi, a weekday and sometimes a nakṣatra. Because these elements are redundant, the date can be checked by computation, and the tables of Sewell and Dikshit (1896) were designed partly for verifying such dates in inscriptions and manuscripts. Second, they are full of numbers. A copyist who misreads a digit in a table or a word-numeral in a verse leaves no grammatical trace of his error; commentaries, with their worked examples, are therefore an essential control on the text, and modern editors have often had to restore parameters by recomputing them. Manuscripts were written in the scripts of their regions, Devanāgarī in the north and west, Śāradā in Kashmir, the Bengali and Oriya scripts in the east, and Grantha, Malayalam, Telugu and Kannada scripts in the south, and the numbers in tables were written with the regional forms of the decimal digits. A table copied from one script into another was therefore exposed to a further kind of error.
Example 12.2 (Checking a colophon date (a hypothetical case))
Suppose a colophon is dated on the first tithi of the bright half of Caitra in Śaka year 1500, a Sunday. With the rules of Example 12.1, Kali year 4679 (Śaka \(+\,3179\)) has elapsed and no months of the new year: \(4679 \times 12 = 56{,}148\) solar months; \(56{,}148 \times 1{,}593{,}336 / 51{,}840{,}000 = 1725.7\), so 1725 intercalary months; \(57{,}873\) lunar months and \(1{,}736{,}190\) tithis; \(1{,}736{,}190 \times 25{,}082{,}252 / 1{,}603{,}000{,}080 = 27{,}166.3\), so 27,166 omitted tithis; and \(1{,}709{,}024\) civil days. Since \(1{,}709{,}024 = 7 \times 244{,}146 + 2\), the day falls two weekdays after a Friday, on a Sunday, and the colophon is consistent. The day is 9 March 1578 in the Julian calendar, the day after a new moon. Had the colophon named a Wednesday, the historian would ask whether the scribe counted the year as current rather than elapsed, used another reckoning of the months, or simply erred. The tables of Sewell and Dikshit (1896) make such checks routine.
Manuscripts were kept by families, by temples and monasteries, and by rulers. The Jain libraries (bhaṇḍāras) of Gujarat and Rajasthan preserved astronomical and astrological works of all communities; the library assembled for Sawai Jai Singh’s astronomers at Jaipur survives and has been catalogued (Pingree 2003); and many Kerala texts were preserved on palm leaf in the households of the families who wrote them, until collectors and editors such as K. V. Sarma brought many of them to light (Sarma 1972). In the nineteenth and twentieth centuries large numbers of manuscripts entered public libraries in India and Europe through purchase, colonial searches for Sanskrit manuscripts, and gifts. Today the largest holdings of astronomical manuscripts are divided among public libraries and research institutes in Vārāṇasī, Pune, Jaipur, Bikaner, Thiruvananthapuram, Chennai and Kolkata, among the Jain libraries, and among libraries in Britain, Germany and the United States; many remain in private hands. The National Mission for Manuscripts, set up by the Government of India in 2003, has been surveying and cataloguing them.
The indispensable guide to this scattered material is David Pingree’s Census of the Exact Sciences in Sanskrit (Pingree 1970–1994). Arranged by author, it records for each the works attributed to him, the manuscripts of each work in the catalogues of collections throughout the world, and the editions. Five volumes appeared between 1970 and 1994; the series was not completed before Pingree’s death in 2005. With his separate catalogues of tables and of the Jaipur collection, it showed for the first time the true size of the literature, and in particular how much of it consists of handbooks, tables, commentaries and astrological works rather than the few famous siddhāntas.
12.8 Computation, Observation and Correction¶
A handbook fixed a set of epoch values and rates. Like every set of parameters, it drifted slowly away from the sky, and its users found out when an eclipse began at the wrong time or a planet rose earlier than computed. How the tradition responded to such failures reveals how it understood its own authority.
Some of the standard texts claimed more than human authority. The Sūryasiddhānta presents itself as the teaching of the Sun, revealed at the end of the Kṛta age, and Brahmagupta presented his siddhānta as a corrected restoration of the system of Brahmā (Burgess 1860; Pingree 1978). If the parameters were revealed, a discrepancy with the sky was a problem for the observer, not for the text. The Sūryasiddhānta itself offers one way out near its beginning: the science is the same as the one the Sun first taught, and the differences arise only from the passing of the ages. The way most widely taken was the bīja (Chapter 6): a small change in the revolution numbers, added to the text rather than replacing it, which brought its results back into line while leaving the authority of the root text intact. From about 1500 the Sūryasiddhānta was generally used with such a bīja, and Sewell and Dikshit (1896) took this into account in their tables.
The handbook-makers were the people most exposed to the sky, and several of them claimed to have adjusted their parameters by observation. In Kerala the Parahita system added a correction proportional to the years elapsed since a fixed date in the Śaka era, and Parameśvara rebuilt the parameters after decades of eclipse observations (Chapter 10). In western India Keśava of Nandigrāma, in his Grahakautuka (1496), stated that he had corrected his parameters to agree with what was seen, and his son built such corrections into the Grahalāghava. The test was almost always the same: eclipses, conjunctions of planets with each other and with bright stars, and the first and last visibility of the planets, the events that a careful practitioner could time with a water clock and a gnomon.
The justification of such corrections was argued explicitly by Nīlakaṇṭha Somayājī, for whom all parameters rest on observation and must be revised from time to time (Chapter 10), and was contested in the seventeenth-century debates of Vārāṇasī (Chapter 11). Against it stood a view that the times required for ritual are those computed by the authoritative text, whatever the sky may show, since the ritual’s efficacy is unseen. Both positions were reasoned, and both had defenders among learned astronomers (Minkowski 2002). The same disagreement returned in the nineteenth and twentieth centuries, now between almanacs computed from modern ephemerides and those computed from the traditional texts (Chapter 13). The reformers, too, often worked in the old genre. Veṅkaṭeśa Bāpūjī Ketkar (1854–1930), one of the leading advocates of dṛk almanacs in western India, set out modern parameters and methods in Sanskrit verse handbooks of his own, so that the karaṇa form outlived the parameters for which it had been invented.
12.9 Handbooks in a Wider Perspective¶
Every pre-modern astronomy that computed a calendar had to turn its theory into a practical instrument. Ptolemy followed his Almagest with the Handy Tables (Neugebauer 1975); the Islamic astronomers organised their practical astronomy in the zīj, a book of tables with rules for their use; Chinese official calendars were issued as computational procedures. The Indian handbook shares the aim of all these. Its distinctive features are two. First, its primary form was a set of rules in verse, meant to be memorised and applied with small arithmetic, rather than a set of tables; tables became common later, alongside the rules. Second, its authors sought simplification by arithmetic as much as by tabulation: the small ratios of the Khaṇḍakhādyaka, the decimal fractions of the Bhāsvatī and the cycles and algebraic approximations of the Grahalāghava replace tables by rules that are easy to carry in the head. The working astronomer’s mathematical skill, as much as his books, was the instrument of his trade.
The handbook was also the form in which Indian astronomy travelled most readily. The Arabic Zīj al-Arkand was built on the Khaṇḍakhādyaka (Section 12.2); the calendrical astronomy of the Kālacakratantra, still used in Tibet, is of the handbook type (Section 9.8); and the luni-solar calendars of Myanmar, Thailand, Laos and Cambodia, counted from an era beginning in 638 CE, were computed by rules derived from Indian handbooks (Chapter 16). A set of rules with a recent epoch, small numbers and no need for the full theory behind them could be learned, carried and adapted far more easily than a siddhānta. In that portability lies much of the historical importance of the karaṇa.
Further Reading¶
Pingree (1981) surveys the handbooks and tables genre by genre, and Pingree (1978) gives their parameters and schools. Sengupta (1934) translates the Khaṇḍakhādyaka. On tables, Montelle and Plofker (2018) is the standard account, with Pingree (1968, 1973) for the manuscripts. Pingree (1970–1994) is the guide to manuscripts and authors, and Pingree (2003) describes the Jaipur collection. On the reasons astronomers gave for their parameters and corrections, see Minkowski (2002); on the families of Vārāṇasī, Minkowski (2004). Sewell and Dikshit (1896) explain the calendar computations that the almanac-makers performed.