Two rival computations, one devout province — and the years when the goddess is welcomed on two different days.
Walk into a traditional Bengali household before Durga Puja and you may find the year's panjika — thick, maroon-covered, dense with tables — occupying the same shelf of honour a Tamil home reserves for its panchangam. Ask which panjika, and you have asked a question with lineage in it. For very many families the answer has been the same for four generations: the Gupta Press panjika, or its classic peer from P.M. Bagchi. For others — including, as is widely reported, institutions of the stature of the Ramakrishna Mission at Belur — the book is the Bisuddha Siddhanta panjika. Most of the year, the two books agree. And then comes a year when they do not: one panjika's Ashtami ends before sunrise and the other's after, and suddenly two neighbouring paras are performing Sandhi Puja at different hours, or an Ekadashi fast is kept on different days in two branches of one family.
Nothing in that scene is carelessness. It is a live, century-old scientific dispute conducted through published almanacs — possibly the most consequential ongoing computational disagreement in popular Hindu practice. Understanding it takes one ancient book, one reform movement, and a little orbital mechanics.
The Surya Siddhanta is one of classical India's canonical astronomical treatises — a Sanskrit work whose received form is roughly one and a half millennia old, setting out a complete computational astronomy: mean motions of the Sun, Moon and planets expressed as integer revolutions per cosmic age, epicyclic corrections, eclipse computation, and the almanac quantities built on them. As engineering for its era it is magnificent — its mean lunar month is within about two seconds of the modern value. Bengali almanac-making, like much of northern and eastern India's, descended through this siddhantic tradition: the panjika-maker's craft was to turn the treatise's rules into each year's tables.
The difficulty is the one every frozen model faces — the same one we describe for the Tamil vākya verses. The treatise's constants are fixed integers, consecrated by the text. The real Moon's orbit, meanwhile, evolves: its apogee precesses, perturbations accumulate, and small per-cycle errors compound over centuries. A mean month good to seconds still lets the Moon's position drift when its slower corrections are frozen; by the modern era, positions computed strictly by the received Surya Siddhanta stand displaced from the telescope's Moon by margins an almanac can feel — enough to move a tithi's ending by large fractions of an hour.
Around the turn of the twentieth century — the era when colonial-observatory astronomy was widely accessible to Indian scholars — a reform school arose in Bengal arguing a point that was, crucially, internal to the tradition: the shastras themselves repeatedly demand dṛk-tulya computation, agreement with the seen sky. An eclipse computed by the old constants that begins visibly late is not tradition, the reformers argued; it is error wearing tradition's clothes. The school's almanac took its name from its programme: Bisuddha Siddhanta — the purified computation — publishing tithi and nakshatra timings from corrected, observation-grade positions (in modern terms, drik astronomy: the same programme our engine implements with the JPL DE440 ephemeris).
Bengal's almanac public, however, did not simply migrate. The established panjikas held — by lineage, by liturgical continuity, by the entirely reasonable instinct that one's grandmother's Ashtami should be one's own. The result is the standing two-school equilibrium: the Surya-Siddhanta lineage (Gupta Press, Bagchi and their tradition) commanding the larger popular following, the Bisuddha school carrying the reform banner with distinguished institutional adherents — and the province long practiced at living with both.
Figure 1 — The amplification mechanism: tithi rules are keyed to sunrise, so a modest timing difference near dawn splits the date itself.
Why does a sub-hour computational difference produce headlines about "two Durga Puja dates"? Because of how the calendar assigns observances. For most vrata and puja purposes, a day "belongs" to the tithi prevailing at a reference moment — commonly sunrise, or a rite-specific window like the Sandhi juncture of Ashtami and Navami. Whenever the true tithi boundary falls near that reference moment, the two schools' computed boundaries — separated by their standing drift — can land on opposite sides of it. One panjika's Ashtami is spent by dawn; the other's still runs. The observance date flips by a whole day, though the underlying disagreement was only tens of minutes.
This also explains the split's rhythm: in most years, boundaries fall safely away from the reference moments and the two books agree everywhere that matters; households forget the divide exists. Then a year arrives with two or three tight boundaries, the books diverge on a marquee date, newspapers write their perennial "why two dates for the puja?" piece — and the divide is rediscovered. The disagreement is constant; only its visibility oscillates.
| Gupta Press / Bagchi school | Bisuddha Siddhanta school | |
|---|---|---|
| Computational base | Received Surya Siddhanta constants | Corrected, observation-grade (drik) astronomy |
| Claim to authority | Continuity of the almanac lineage; the tradition as practised | The shastra's own dṛk-tulya injunction; the sky as referee |
| Tithi/nakshatra timings | Drift from the observable sky by substantial fractions of an hour | Match the computed/observable sky |
| Festival dates | Agree in most years; split when a boundary falls near a reference moment | |
| Cultural footprint | The larger popular following | Reform lineage; distinguished institutional adherents (Belur Math is widely reported among them) |
| Nearest Tamil analogue | Vākya panchangam | Thirukanitha (drik) panchangam |
Table 1 — The two schools at a glance. The last row is the deep point: Bengal and Tamil Nadu are having the same argument in two languages.
Set this beside our profile of the Tamil vākya tradition and the symmetry is striking. In both provinces, a pre-telescopic computational system of genuine brilliance — verse-encoded lunar tables in the south, siddhantic integer constants in the east — powered the almanac for centuries. In both, a reform school armed with modern astronomy invoked the tradition's own demand for sky-agreement. In both, the popular almanac culture substantially stayed with the ancestral computation, for reasons that are about identity and continuity rather than ignorance. And in both, the practical consequence is identical: the same festival, two dates, in the same city — with each date perfectly correct inside its own system.
Kerala completes the triptych from the other direction: its printed calendar culture went effectively drik long ago (a story we take up in our Kerala calendar methods piece), showing that the reform equilibrium is also livable. India's almanac map is thus not a map of accuracy but a map of settlements — different provinces concluding the ancestors-vs-sky negotiation on different terms. Any engine that serves all of India must therefore do what no single printed almanac does: compute both honestly and label everything.
Three practical translations of the above. First: if your family panjika and any drik-based app (ours included) disagree about a tithi's end by up to an hour-odd, neither has malfunctioned — you are looking at the schools' standing drift, and for a family observance the family's book is the right authority, because that is what the observance means in your lineage. Second: if two apps or two printed panjikas give different dates for a vrata, identify each one's school before assuming error — the answer is usually one line deep in the almanac's front matter. Third: for astronomical questions — the Moon's actual nakshatra at a birth, a muhurta's actual lagna — drik computation is the correct instrument, which is why our jātakam engine is drik to the arc-second and shows its longitudes so any school's follower can verify and translate. Our planned Bengali panjika pages will carry the same transparency: drik timings as the computational base, with the school context stated rather than hidden — because in Bengal, telling a user which panjika a number agrees with is not a footnote, it is the number's meaning.
Fairness to the old school requires one technical concession its critics often omit. The siddhantic tradition was never as frozen as the caricature suggests: almanac-makers across the centuries applied bīja ("seed") corrections — small emendations to the canonical constants, sanctioned within the tradition itself, precisely to keep computations serviceable as drift accumulated. A working panjika of the received school is therefore not raw sixth-century arithmetic; it is the treatise plus its lineage's accumulated corrections. This matters for two reasons. First, it explains why the old-school panjika is far closer to the sky than a naive reading of the Surya Siddhanta would be — the lineage has been quietly course-correcting all along, which is its own kind of empiricism. Second, it reframes the dispute honestly: the argument is not "corrections versus no corrections" but how far correction may go while the almanac remains itself — periodic bīja patches within the canonical framework, or the reformers' full replacement of the framework with observation-grade positions. Seen this way, Bengal's two panjikas are two points on a continuum of loyalty, not opposites — and the reader who understands bīja will find both camps more reasonable than their pamphlets suggest.
Bengal's drik argument did not stay in Bengal. When independent India convened the Calendar Reform Committee (1952–55) under Meghnad Saha — a Bengali physicist steeped in exactly this controversy — the committee's almanac programme was, in effect, the Bisuddha position nationalised: the resulting Rashtriya Panchang computes by modern ephemeris with the Lahiri ayanamsa (the standardisation story we tell in our ayanamsa article), and the committee's report is bracingly direct about almanacs that lag the sky. Yet sixty years on, the received-school panjikas still command Bengal's larger public — a standing demonstration that in calendrics, state standardisation proposes and the household disposes. For engine-builders the lesson is the one this series keeps arriving at: authority in this domain is plural — text, lineage, state, and sky each issue their own rulings — and software that pretends otherwise merely hides its chosen master. Ours is the sky, declared plainly, with every longitude printed so the other masters' followers can translate.
The Bengali panjika is famously a civilisation in paperback: alongside tithi tables sit the year's marriage dates and their lagna prescriptions, the vrata calendar with its rules, shraddha reckonings, rice-feeding and thread-ceremony dates, temple festival lists, even railway timetables and advertisement pages in the classic editions — the almanac as household operating system. That breadth is why panjika loyalty runs so deep, and why the computational-school question hides so well: a family relates to the whole book, not to the siddhanta note on page two. Any digital panjika worth building must respect that breadth — timings alone are a skeleton; the observance layer is the flesh.
Astronomically, the Bisuddha (drik) school matches the observable sky by construction — that is its entire programme. Liturgically, "accurate" means "what my tradition prescribes", and for the majority lineage that is the old-school computation. The two questions have different right answers, and conflating them is the source of most quarrels on the topic.
Nineteenth-century Calcutta was simultaneously the capital of traditional panjika publishing and of Indian engagement with observatory astronomy — the one city where both authorities lived on the same street. The received-school panjikas consolidated their mass readership in that print boom; the reform school crystallised from the astronomy side of the same milieu a generation later; and the readership, offered both, kept both. Bengal's two panjikas are, in that sense, the Calcutta enlightenment arguing with itself annually, in paperback — which is also why the dispute has stayed civil and productive for a century.
It varies by element and epoch, which is why we resist quoting one folkloric number. The honest statement: fractions of an hour on lunar quantities, with day-level festival splits when boundaries are tight. A measured comparison of old-siddhantic constants against JPL DE440 follows the same method as our vākya measurement study, and Bengal's case is on that study's roadmap.
The tradition's own instrument is the family's kula-purohit: observances follow the officiating lineage's book, and a household with two lineages typically follows each side's panjika for that side's hereditary observances — the same pluralism the province has practised for a century, scaled to one kitchen. Where a single shared date must be chosen (a joint puja, a child's rice-feeding), stating the two books' timings side by side and letting the elders choose knowingly beats any algorithmic tie-break we could offer — and presenting exactly that comparison is what a well-made digital panjika is for.
Usually yes — marquee-date splits between panjikas trace to exactly the boundary-near-reference mechanism of Section 4. (Inter-regional date differences — Bengal vs North India — are more often the amanta/purnimanta month-reckoning difference, a separate topic in our panchangam series.)
That is usually the month-reckoning difference, not the school difference: Bengal's religious months run amanta (ending at new moon) in some usages and the solar Bengali calendar governs others, while the Hindi belt reckons purnimanta — so the "same" tithi can belong to differently named months and festivals anchored to month-plus-tithi diverge. School drift moves timings by minutes; month conventions move names by a fortnight. Diagnose which of the two you are looking at before assigning blame.
Both have their own venerable panji/panjika traditions (the Odia panji lineage is a rich topic we intend to treat separately) with their own received-vs-reform histories; the two-school lens of this article travels well, but the specific institutional map differs by province, and we avoid asserting details we have not verified.
A koshthi computed today by any competent drik engine is unaffected by the panjika split — the split concerns almanac timings, not ephemeris positions. Where it does touch personal astrology is historical: an old koshthi computed strictly by siddhantic tables may carry the era's positional drift, worth knowing when comparing it against a modern recomputation — gently, and with the respect owed to the document.