juno.date · Research

Vākya vs Thirukanitha: Tamil Nadu's Two Schools, Measured Against Each Other

Every Tamil family knows the two almanac camps exist. Nobody had published how often they actually disagree about a marriage. We computed all 729 cases.

Abstract. Tamil Nadu runs its sacred timekeeping on two rival mathematical traditions. The Vākya school computes from memorised Sanskrit verse-tables descended from medieval astronomy — the beloved Pambu Panchangam is its flagship. The Thirukanitha (drik) school computes from modern observational astronomy. Their tithi and star timings famously differ; temples split between them; families inherit an allegiance the way they inherit a deity. But matching is where the divide bites hardest — and how often the two schools actually disagree about a couple has never, to our knowledge, been published as a measurement. We ran all 729 nakshatra pairings through juno.date's implementations of both schools' ten-porutham systems. Results: the schools flip the overall pass/fail verdict for 22.4% of pairings — one couple in five. Only 2.4% of pairings receive identical porutham-by-porutham verdicts. Yet the disagreement is spectacularly uneven: the schools agree 100% on nadi and 98.5% on Mahendra, while disagreeing on yoni for 63.6% of pairings and gana for 59.7%. The pattern reveals what the divide truly is — not different astronomy, but different inherited matrices — and it reframes a familiar anxiety: if your two Tamil reports disagree, the cause is probably not an error but a five-hundred-year-old argument. We map it porutham by porutham, and preview the second half of this study: measuring the schools' positional timings against the sky itself.

1. Two almanacs on one shelf

Walk into any Tamil household that keeps a panchangam and you will find one of two books. The Pambu Panchangam and its cousins carry the Vākya tradition: their computations descend from versified astronomical tables — the 248 chandravākyas attributed to Vararuci, refined in the medieval Vākya-karaṇa — chanted, memorised, and applied by rule, a living fossil of pre-telescopic Indian mathematics that we profiled in its own article. The Thirukanitha almanacs compute the same quantities from modern ephemerides — the school's name is simply "precise calculation." The two books print different minutes for the same tithi, occasionally different days for the same festival, and — the subject of this study — different verdicts for the same couple.

The matching divide matters more than the festival divide for a simple reason: a festival date resolves publicly (the temple picks a day) while a match verdict lands privately, on one family, holding two contradictory reports. The usual questions follow: is one report wrong? Is one school "more accurate"? Those questions deserve numbers, and the input space — 27 girl-stars × 27 boy-stars — is small enough to answer exhaustively. So: every pairing, both schools, every porutham, counted.

2. The headline: one couple in five

Measure (722 pairings with both engines)Result
Overall verdict flips (pass ≥6/12 in one school, fail in the other)162 — 22.4%
Pass rate, Vākya school90.0%
Pass rate, Thirukanitha school77.0%
Mean absolute score difference1.40 poruthams (max observed: 5)
Pairings with identical total score131 — 18.1%
Pairings with identical verdicts on all twelve checks17 — 2.4%

Table 1 — The two Tamil schools across the full pairing space. Same inputs, same city, same day — different books.

Two numbers frame everything. First, the 22.4% verdict flip rate: between the two Tamil schools, one couple in five passes one tradition's bar and fails the other's — a within-Tamil-Nadu disagreement almost as large as the 24.4% North–South gap we measured in the porutham-vs-guna study. Second, the 2.4% identical-verdict rate: only 17 pairings of 722 receive the same yes/no on every single porutham from both schools. Total agreement is not the norm between the schools; it is the exception, occurring about once per 42 couples.

Note also the asymmetry of generosity: Vākya passes 90% of the space, Thirukanitha 77%. The flips are therefore mostly one-directional — the typical disagreement is a couple the Vākya book blesses and the Thirukanitha book questions, not the reverse. If your family's two reports disagree, the measured split is 129 Vākya-blesses flips against 33 the other way — four to one.

3. Where the schools agree perfectly — and where they argue

The averages conceal the study's sharpest finding. Disagreement is not spread evenly across the ten poruthams; it is concentrated with almost diagnostic precision:

PoruthamPairings where the schools' verdicts differRate
Nadi0 / 7220% — perfect agreement
Mahendra111.5%
Vedha547.5%
Vasya8211.4%
Dina10314.3%
Rajju10815.0%
Sthree Deergha11616.1%
Rasi13018.0%
Rasi Athipathi23732.8%
12th-house check30342.0%
Gana43159.7%
Yoni45963.6%

Table 2 — Per-porutham disagreement between the schools, over all 722 co-computed pairings.

Read from the top: the checks that are pure star arithmetic — nadi's three-class deal, Mahendra's fixed counts, Vedha's fixed pair list — agree perfectly or nearly so. The checks that depend on classification matrices — which animal each star is, which temperament class, how sign lords rank — disagree for a third to two-thirds of all couples. Yoni, the intimacy test, tops the argument at 63.6%: the two schools assign different animal pairings or different hostility rules often enough that a majority of couples get different yoni verdicts. Gana — deva, manushya, rakshasa — is barely behind at 59.7%.

💡 This table is the study's fingerprint, and it identifies the divide precisely. If the schools differed in astronomy — different star boundaries, different ayanamsas — the disagreement would fall on the position-sensitive checks (Dina's counts, Rajju's ropes). Instead those agree 85% of the time, while the position-independent classification tests explode. The vākya–thirukanitha matching divide is not about where the Moon is. It is about which lookup tables one's grandfather's almanac printed. The astronomy argument and the matrix argument are separate arguments — and matching families are almost always having the second one without knowing it.

4. One couple, dissected: the five-point flip

Abstractions convince slowly; a worked case convinces at once. The largest verdict swing in the whole space belongs to girl Shravana, boy Magha. Under the Vākya school this couple scores a glowing 10 of 12 — dina, gana, mahendra, sthree deergha, yoni, athipathi, vasya, rajju, vedha and the 12th-house check all pass; only rasi and nadi object. A Vākya-school astrologer would call the match near-excellent. The same two stars under the Thirukanitha school: 5 of 12, fail. Dina flips to fail. Gana flips. Yoni flips. Athipathi flips. Vasya flips. Same girl, same boy, same sky.

Walk the flips against Table 2 and the pattern holds perfectly: everything that flipped is a matrix-driven check — the temperament classes, the animal table, the lord rankings, the count remainders the two schools grade differently. Everything that held — rajju, vedha, nadi, mahendra, the structural arithmetic — is position-blind and school-proof. This one couple is the whole study in miniature: the two books are not measuring the couple differently, they are classifying them differently. (The five-point swing is extreme but not unique — U.Bhadrapada–Revati and Revati–Hasta swing five and four points by the same mechanism.)

The seventeen couples the schools agree on completely

At the opposite pole stand the 17 pairings — 2.4% — whose twelve verdicts match check for check across both books: among them Ashwini–P.Ashadha (10/12 in both), Bharani–U.Ashadha (10/12), Ashwini–U.Ashadha, Shatabhisha–P.Ashadha and Shatabhisha–U.Ashadha (9/12 each), U.Phalguni–Ashlesha (9/12), and — the only fully-agreed failure — Magha–Mrigashira at 5/12. A family blessed with one of these pairings can enjoy a rare privilege: a verdict no almanac argument can touch. Everyone else's verdict rests, at least in part, on which book the family's astrologer keeps.

5. How the divide got here — and its cousins across India

The two schools are the Tamil front of a subcontinental argument that has run for at least six centuries: should sacred time follow the received tables or the observable sky? The vākya corpus fossilises the received side — verse-encoded lunar tables whose constants were fixed when they matched the sky of their era. The drik ("visual") reform argued that computation must track observation; Parameshvara of Kerala systematised it in the fifteenth century, and Kerala's almanacs went drik centuries ago. Bengal fights the same battle under different banners — the Surya-Siddhanta loyalists of the Gupta Press panjika against the drik-reformed Bisuddha Siddhanta, a divide we profile separately — and Telugu and Kannada almanac culture carries its own version of the split. Tamil Nadu's distinction is that the old school remains genuinely dominant in popular devotion: the Pambu Panchangam's chanted tables still set festival and muhurtham time for millions, which is precisely why the matching divide measured here is not an antiquarian curiosity but a live fact about contemporary weddings.

What the matching data adds to this old story is a twist: the marriage matrices are not astronomical constants, so the drik reform never actually adjudicated them. When the schools' porutham tables diverged — through different manuscript lineages, different regional teachers, different printed almanac houses ⚖️ — the divergence rode along with the astronomy argument without being part of it. Table 2 is, in effect, the first separation of the two arguments: the astronomy-free rows at the top show what the schools still share; the matrix rows at the bottom show what drifted apart for reasons that have nothing to do with the sky.

6. What this means for a family holding two reports

First, the diagnosis is benign: two disagreeing Tamil reports are, with overwhelming probability, correctly computing two different inherited traditions — not exhibiting a bug, and not evidence that either astrologer is careless. Second, the disagreement has structure you can use. Check the top of Table 2: if the reports disagree about nadi, Mahendra or Vedha, something is wrong — those checks should be identical, so recheck the star inputs. If they disagree about yoni or gana, you are simply looking at the five-century argument, and no recomputation will settle it. Third, the shared verdicts carry extra weight: when both schools pass a couple — 72.4% of the space — the pass is robust to the whole divide; when both fail — 5.1% — the caution likewise stands on both books. The juno.date match page computes both schools side by side for every couple, labels them Vākya and Thirukanitha explicitly, and lets the shared verdicts speak first — which we believe is the only honest way to present Tamil matching online, and which, as far as we know, no other site does.

7. Which school should you trust?

The honest answer has two halves, and they point in different directions.

For matching: there is no measurement that can crown a winner, because the matrices are not claims about the sky — they are inherited classifications, and "which yoni table is correct" is a question about textual authority, not accuracy. Families should use the school their tradition follows; mixed families should read both and weight the shared core, exactly as with the North–South divide.

For positions and timings: here accuracy is measurable, and the schools genuinely differ. The vākya tables approximate the Moon's motion with medieval-era constants; five centuries of accumulated drift mean a vākya tithi or star boundary can sit hours from the observable sky — the Thirukanitha school's founding argument, and the reason the drik reform swept Kerala six hundred years ago (a story we tell in the Kerala almanac article). Quantifying that drift precisely — vākya star-boundary timings versus the DE440 sky, distribution of the error across the month and the year — is the second half of this study. Our vākya computation engine (the verse-table algorithm implemented directly, not scraped from any almanac) is in development; when it ships, this page will carry the measured drift tables alongside the matching map above. Watch this space — the number "how many hours is the Pambu Panchangam off today" deserves to be public, and it will be.

8. Questions this measurement raises

"How do I know which school my family follows?"

Ask which almanac the family (or its purohit) keeps. Pambu Panchangam and most of the older Madras almanac houses are Vākya; almanacs advertising "Thirukanitha" or "drik ganitha" on the cover are the reform school; several publishers now print both editions. Temple allegiances are often the deciding inheritance — many of Tamil Nadu's great temples, including Srirangam, keep vākya time for ritual by deliberate tradition — so a family's school is frequently its temple's school. If nobody remembers, that itself is useful information: the family's tradition is unlikely to hinge on the divide, and the shared-verdict reading (both schools side by side) fits naturally.

"Doesn't a 22% flip rate mean star matching is arbitrary?"

It means star matching is plural — which every tradition-keeper already knew, and which is not the same thing. Seventy-eight percent of the space gets the same verdict from two schools that have argued for centuries; the checks built on pure arithmetic agree almost perfectly; and the flips concentrate exactly where inherited classification differs, which is where reasonable traditions may differ. What the rate does forbid is false certainty: any tool that presents one school's 7/12 as the Tamil verdict — no school named, no alternative computed — is claiming a precision the tradition itself never possessed. Plurality honestly displayed is not arbitrariness; it is the actual shape of the inheritance.

"Why would the yoni animals differ between schools?"

The yoni assignment maps each nakshatra to an animal (and each animal to its allies and enemies), and the mapping travelled through manuscript lineages for centuries before print froze the variants ⚖️. Some stars carry different animals in different lineages; more often the animals agree but the hostility matrix differs — whether Tiger's enemy is Cow or Ox, whether Mongoose–Serpent is fatal or merely poor. Because a yoni verdict flips whenever either the animals or their relation flips, small matrix differences compound into the 63.6% disagreement measured — the largest of any porutham, on the check families ask about most.

"Does the divide affect muhurtham (wedding timing) too?"

Yes, and more visibly: muhurtham depends on tithi, nakshatra and lagna timings, which is where the schools' astronomy genuinely differs — a vākya star-boundary can sit hours from the drik one, so an auspicious window valid in one book may have already closed in the other. That is the positional half of this study, awaiting our vākya computation engine. The matching comparison above is immune to this (both schools agree what your star is on almost every birth; they argue about what it implies) — but wedding-day planning across the two books inherits the full timing drift, which is why mixed-school families so often hold the muhurtham argument and the matching argument at the same dinner table without realising they are different arguments.

9. A word of respect for both books

Nothing in these tables diminishes either school. The vākya corpus is one of the great achievements of memorised mathematics — an almanac computable by a scholar with no instruments, no paper tables, and no electricity, accurate to its era's observational limits, and chanted into the present by an unbroken teaching lineage. The Thirukanitha reform is the same civilisation updating its constants — the drik astronomers' insistence that the computed sky match the visible one is the exact instinct this research site runs on, and we consider ourselves, plainly, a Thirukanitha operation: juno.date computes positions from the JPL DE440 ephemeris, the modern descendant of the drik program. The measured map above is what respect for both looks like in practice: state each school's outputs faithfully, side by side, and let families see precisely where their two inheritances agree and where they argue. What we decline to do is what too many tools do silently — compute one school, label it "Tamil marriage matching," and leave one couple in five to discover the divide by anguish.

And a closing note on the article's title. "Accuracy, measured" is a promise in two parts. This first part measured the schools against each other, which is what a matching family actually experiences; it found the divide real, large, structured, and — once its anatomy is visible — far less frightening than a pair of contradictory reports feels at midnight. The second part will measure the vākya tables against the sky, where accuracy has a true answer rather than a plural one. Both halves belong to the same conviction: every number a family is asked to trust should be one somebody has actually checked, in public, with the working shown.

Method & sources. Engines: juno.date's implementations of the ten-porutham system under both schools, built from systematically harvested and cross-validated reference tables of each tradition's porutham matrices (the Vākya-school engine validated against a printed Vākya-school match report; see our source reviews). Input space: all 729 ordered star pairings at pada-1 Moon signs; 722 pairings have rows in both engines (7 lack a Thirukanitha table row and are excluded from comparisons). Verdict bar: 6 of 12 checks (the ten classical poruthams plus the sthree-deergha and 12th-house checks both schools carry). Per-porutham differences in Table 2 are counts of pairings where the two engines return opposite yes/no verdicts on that check. Positional drift of vākya timings is not measured here — that requires the vākya verse-table computation itself, which is in development; the matching comparison is deliberately astronomy-free, which is precisely what Table 2 demonstrates. School naming: "Vākya" and "Thirukanitha" are the traditional Tamil school names; no criticism of either school or of any almanac publisher is implied ⚖️. — juno.date Research