Marius Comper RO

Āryabhaṭīya · Gītikāpāda 10 · c. AD 499

Aryabhata’s 24 half-chords, c. AD 499: a sine table held in one verse

In one verse of the Āryabhaṭīya, traditionally dated to 499, Aryabhata wrote a quarter-circle table of sines as 24 words. Each word spells a number in letters: the step by which the half-chord grows as the arc opens by another 3¾°, on a circle of radius 3,438. Pick a word, see how its letters make the number, and watch the half-chord meet the circle.

· kalārdhajyāḥ

Half-chord, summed
3,438 × today’s sine
Difference
Compare with
A quarter circle of radius 3,438. Each tick is a step of 3¾°; each vertical line is a half-chord built by adding the verse’s numbers. Drag around the arc or pick a word.
For each of the 24 steps, the summed half-chord minus 3,438 times today’s sine. All 24 sit inside the band of one unit.

Numbers spelled in letters

Near the start of the book, Aryabhata explains his way of writing numbers in letters. Each consonant has a value: k to m are 1 to 25, and y, r, l, v, ś, ṣ, s, h are 30 to 100 in tens. The vowel it carries multiplies it: a by 1, i by 100, u by 10,000, and so on. So makhi is ma, 25, plus khi, 2 × 100: 225. A cluster of consonants shares one vowel: hasjha is ha, 100, plus s, 90, and jha, 9: 199. Long and short vowels count alike.

The tenth verse of the first part is 24 such words and ends kalārdhajyāḥ, half-chords in minutes of arc. Decoded, they are 225, 224, 222, 219 and so on down to 22 and 7. The numbers on this page come from W. E. Clark’s 1930 translation, the words from H. Kern’s 1874 edition of the text. Decoded letter by letter, 23 of Kern’s 24 words give Clark’s numbers. The ninth, as Kern prints it, is svaki, 250; other manuscripts read skaki, which decodes to 191, the number the commentary Kern printed gives. This page uses skaki and shows svaki beside it: 59 units too many would lengthen every later half-chord by 59.

A circle measured in its own minutes

The numbers are steps. Add them one after another and each running total approximates 3,438 times the sine of the next multiple of 3¾°: the half-chord, the line from the end of the arc straight down to the diameter. A full circle is 360 degrees, or 21,600 minutes of arc, and on Aryabhata’s scale one unit of length is about one minute of arc along the rim. The 24 steps add up to exactly 3,438. Elsewhere the book gives a circle of diameter 20,000 a circumference of about 62,832; with that ratio, a rim of 21,600 has a radius of 3,437.7, so 3,438 is the nearest whole unit.

Against 3,438 times the modern sine, all 24 running totals are within one unit; the largest gap is about 0.70, at 67½°. Twenty-two of the 24 steps equal the nearest whole number to the modern difference; the other two, at 22½° and 30°, are each one unit off, in opposite directions, so the totals never stray.

ArcWordStepHalf-chord3,438 × sineDifference
3.75°makhi225225224.86+0.14
7.5°bhakhi224449448.75+0.25
11.25°phakhi222671670.72+0.28
15°dhakhi219890889.82+0.18
18.75°ṇakhi21511051,105.11−0.11
22.5°ñakhi21013151,315.67−0.67
26.25°ṅakhi20515201,520.59−0.59
30°hasjha19917191,719.00+0.00
33.75°skaki19119101,910.05−0.05
37.5°kiṣga18320932,092.92+0.08
41.25°śghaki17422672,266.83+0.17
45°kighva16424312,431.03−0.03
48.75°ghlaki15425852,584.83+0.17
52.5°kigra14327282,727.55+0.45
56.25°hakya13128592,858.59+0.41
60°dhāhā11929782,977.40+0.60
63.75°sta10630843,083.45+0.55
67.5°sga9331773,176.30+0.70
71.25°śjha7932563,255.55+0.45
75°ṅva6533213,320.85+0.15
78.75°lka5133723,371.94+0.06
82.5°pta3734093,408.59+0.41
86.25°pha2234313,430.64+0.36
90°cha734383,438.00+0.00

Words as printed in Kern’s 1874 edition, except word 9: Kern prints svaki; the table uses skaki, which other manuscripts read and which gives 191, the number in the commentary. Numbers as in Clark’s 1930 translation; the decoding is the page’s own, by the rule given near the start of the book.

Half-chords, not chords

Ptolemy’s tables tabulated the full chord across an arc, on a circle of radius 60. The Indian tables gave half of the chord of the doubled arc, which is the sine. Read as full chords on the same circle of radius 3,438, Aryabhata’s sums miss by about 1,424 units at 90°; read with steps of 3° instead of 3¾°, by hundreds. Of the readings tested here, only the half-chord, every 3¾°, fits.

The scheme was shared. Varāhamihira’s Pañcasiddhāntikā, from the sixth century, tabulates 24 values at 3¾° on a radius of 120, and the surviving Sūryasiddhānta, in a recension dated to around 800, gives the same 24 sums on 3,438. Aryabhata’s verse holds all 24 differences in one stanza. The word itself travelled too: Sanskrit jīvā, a synonym of jyā, became Arabic jība; written without vowels, it was apparently read as jayb, a fold or pocket, and translated into Latin as sinus, the English sine.

What the verse cannot say

The book also gives procedures for producing sines, one geometric and one arithmetic. Read with fractions kept, the arithmetic rule ends about 63 units below the verse at 90°; rounding each quotient, as Clark’s worked example does, about 70. Scholars have proposed readings that fit better. This page shows the plain reading for comparison and does not use it as a check. Which procedure produced these particular whole numbers remains uncertain.

Share the page

The link opens the verse at the sixth word.

Sources

Every figure on this page is traced in the project’s source notes.

  1. Walter Eugene Clark, The Āryabhaṭīya of Āryabhaṭa, Chicago, 1930, text (Gītikāpāda 2 and 10, p. 19; Gaṇitapāda 10–12, pp. 28–29; Kālakriyāpāda 10, pp. 54–55).
  2. H. Kern, The Aryabhatiya, with the commentary Bhatadipika of Paramadiçvara, Leiden, 1874, pp. 16–17, facsimile.
  3. R. C. Gupta, article on the fractional parts of Āryabhaṭa’s sines, Indian Journal of History of Science 6, 1971, p. 52, PDF (the readings of word 9).
  4. Kim Plofker, Mathematics in India, Princeton, 2009, p. 71 (Sūryasiddhānta), p. 257 (jīvā → jība → jayb → sinus), p. 317 (dating).
  5. Ptolemy, Almagest I.11, table of chords, translation.