You want to fast tomorrow. The calendar on the temple wall says Ekādaśī.
But the day the calendar names and the day the vow attaches to are not the same object, and they can miss each other by a second. If the previous lunar day is still running ninety-six minutes before sunrise — if it holds on for one second inside that window — then tomorrow is not the day. The vow moves to the day after, which acquires a new name when it inherits the obligation.
That is the whole machinery in one paragraph, and it is worth slowing down over, because the tradition is not being fussy. It is being consistent about something it has already decided: the day is the object of the vow, and the day is an astronomical fact rather than a square on a grid. Get the object wrong and the observance is aimed at nothing.1
A tithi is a lunar day — a slice of the moon's motion relative to the sun. A solar day is a rotation of the earth. Neither divides the other cleanly, so a tithi drifts across sunrise, starting and ending at whatever hour it likes.
Which means every lunar observance has a boundary problem the solar calendar never has to face. You cannot just say the eleventh, because the eleventh may begin mid-afternoon and end before the following dawn, or straddle two sunrises.
The tradition's answer is a named ideal case and two named failure cases.
Śuddhā Ekādaśī — pure, also called pūrṇā, also called Harivāsara, the day of Hari. The tithi begins at least one aruṇodaya period — ninety-six minutes — before sunrise, and runs until the next sunrise with no overlap at either end.2
That's the clean case. The interesting rules are the dirty ones.
Here is the asymmetry, and it is the sharpest thing in the passage.
Pūrva-viddhā — pierced from before. The tenth tithi, Daśamī, extends even one second into the ninety-six-minute window before sunrise. The Ekādaśī is now contaminated at its head, and the vow is not observed on it at all. The following Dvādaśī becomes Mahādvādaśī — the great twelfth — and the fast transfers there.3
Para-viddhā — pierced from after. The overlap happens at the far end, with the Dvādaśī. And the ruling is flat: this does not present any problem. Observe on the Ekādaśī as normal.
So one second of the previous day invalidates; any amount of the following day does not. The impurity is directional.
Why? The book doesn't say. But the shape of the rule tells you what it thinks a day is: a day is defined by what it begins as, not by what it ends as. A trace of Daśamī at dawn means you would begin the fast on the tenth, and the fast is not the tenth's. A trace of Dvādaśī at the close means you finish the fast on the twelfth, which is already the day the fast is broken on.
That is a coherent principle, stated nowhere, doing all the work.
The full observance, as the 1993 commentary lays it out: celibacy the day before. On Harivāsara itself, fast even from water — nirambu-upavāsa — and stay awake through the night performing continuous bhajana. Celibacy again the next day, then break the fast at the appointed hour.
And one clarification that is more interesting than it looks:
Fasting from water means that one must also refrain from taking any
mahā-prasāda. Without doing so, one cannot be said to be observingnirambu-upavāsa.4
Everywhere else in this book, mahā-prasāda is not food. It is sanctified remnant, a substance whose honouring destroys anarthas. The tradition's normal grammar would let it through any door.
Here it is stopped at the door, explicitly, because someone clearly tried it. The vow is a vow about intake, and the sanctity of what you're taking in doesn't create an exemption.
For those who cannot do the full fast, an alternative is quoted from the Vāyu Purāṇa: naktaṁ haviṣyānnam — sacrificial food, at night.
Then the permitted items, and the ordering is the part to notice:
fruits, sesame, milk, water, clarified butter, the five articles derived from the cow, or simply air. Each of these is considered progressively better than the one preceding it.5
Read the list as a ranking and it terminates in nothing at all. Air is the best item on a list of things you may eat. The concession runs from solid food down to a placeholder for eating nothing — so even the relaxation is built as a ladder pointing back at the strict case.
Alongside it, a second list from the Mahābhārata: eight things that do not spoil a vow. Water, roots, fruit, milk, ghee — and then three that are not foods at all. The desire of a brāhmaṇa. The order of the spiritual master. Medicine.6
Those last three are the escape hatches, and they are social rather than dietary. Two of them are somebody else's instruction overriding your vow. The vow is invariable, nitya, and neglecting it is called a great transgression — and it yields to being told to eat.
Then the clause that has no analogue anywhere else in the sixty-four limbs.
If a
brāhmaṇawho is maintaining a sacrificial fire is incapable of observing fasting, he may appoint a son or anotherbrāhmaṇato observe the fast for him.7
Someone else's empty stomach discharges your obligation.
Hold the qualifications in view, because they are narrow: it is a brāhmaṇa, and specifically an āhitāgni, one maintaining the sacrificial fire — a man with a competing ritual duty that eating supports. The substitute must be a son or another brāhmaṇa. This is not a general licence.
But the principle is conceded regardless, and it is a large one. The vow is transferable. Which sits oddly beside everything else in the same limb, where the observance is prescribed for all four varṇas and both āśramas, for all women married or widowed, for the household together with wife and sons — that is, framed as a personal, universal, non-delegable duty right up until the page where it is delegated.
You are looking at two calendars that disagree about which day is Ekādaśī. One is your local temple's; one you found online. This is not a clerical error and it will happen again. They are computing the aruṇodaya overlap against different sunrise times for different longitudes, and the ninety-six-minute window is exactly where two nearby cities can land on opposite sides of a rule.
The tradition's own instruction, if you follow the logic of the passage rather than hunting for a verse: the boundary is astronomical and local, so the local computation is the one that governs where you actually are.
Second scene. You have committed to the water fast, it is four in the afternoon, and someone hands you mahā-prasāda from the altar. Every instinct trained by the rest of this book says accept it — the remnant is not ordinary food, honouring it destroys obstructions, refusing it looks like ingratitude.
The text has already ruled on precisely this, and the ruling is no. Which is worth registering as a piece of practical theology: when two goods collide, this tradition does not average them. It names which frame is operative and lets the other one lose cleanly.
Third. You physically cannot complete the fast — illness, work, pregnancy, medication. The passage does not treat that as failure. It supplies a graded fallback ending in air, and it names medicine among the eight things that do not spoil a vow.
What it does not supply is a general right to send someone in your place. The substitution clause is for a fire-maintaining brāhmaṇa, and I would not stretch it beyond that — stretching it is my invention, not the book's.
Put the two rulings side by side and the picture is strange.
Mahā-prasāda — the tradition's own sanctified substance, honoured in limb thirty-six as a destroyer of obstructions — breaks the fast. A different person entirely, eating and drinking normally on your behalf while you eat and drink normally too — does not.
So the vow is strict about substance and loose about who is performing it.
That is only coherent if the fast is not primarily an act of the body. If it were about the body's abstinence, a proxy would be absurd and a spoonful of sanctified remnant would be trivial. Reverse it: if the fast is a claim registered on a particular day, then who registers it is negotiable in an emergency and what breaks it is a technical question about intake, on which sanctity has no bearing.
Stated at the strength the evidence carries: this is one passage in one commentary, and the reading is mine rather than the book's. [n=1 OBSERVED] — I am reporting an inconsistency in the shape of the rules, not naming a principle the tradition holds.
Strongest support. Every rule above is quoted with a citation in the source's own apparatus — Hari-bhakti-vilāsa 12.97, 12.82, 12.100, 12.47, 12.3, each in turn citing the Vāyu Purāṇa, the Mahābhārata or the Viṣṇu-dharmottara. The tithi definitions are the commentary's own exposition rather than a quoted verse.
Tension one. The vow is called nitya, invariable, and neglect of it a great transgression. Three of the eight non-spoiling items are a brāhmaṇa's wish, the guru's order and medicine — two of which are other people's instructions. An invariable vow with an obedience override is a conditional vow with extra steps, and the text does not notice.
Tension two. Universal personal obligation on one page, transferable to a son or another brāhmaṇa on the next. The narrow qualification does not remove the concession; it just fences it.
Open question. Is the directional contamination rule — head-impurity fatal, tail-impurity harmless — argued anywhere in the tradition, or is it a received computational convention with the reasoning lost? Nothing in the range read (Text 4, limb 9) offers a rationale, and I will not claim there is none elsewhere.
The vault's Bryant material presents the devotional calendar as a support structure — the festivals and fast days that give a practitioner's year its shape and keep the practice from dissolving into private preference.
Read this limb and the calendar acquires an entirely different character. It becomes a jurisdictional instrument: a set of rules for deciding which of two candidate days carries an obligation, with named contamination states, a transfer procedure, a substitution clause and eight enumerated exemptions.
That is the grammar of law, not of rhythm. And it converges with what the rest of this ingest keeps finding — that this book's characteristic move is to take an inherited devotional item and specify it until it has failure conditions. The same instinct that gave bhakti a definition tested against hired assassins gives the fast a rule for a one-second overlap.
A deadline whose force comes entirely from your not being able to move it. Set this against Manufactured Deadlines. That page's mechanism is Greene's: believing you have unlimited time diffuses attention, and a hard immovable cutoff jolts a mind into a gear it will not otherwise find — so the recommendation is to manufacture cutoffs deliberately, since the effect comes from the constraint rather than from where the constraint came from.
The Ekādaśī machinery is that claim's mirror image, and the mirror is instructive. This deadline is not manufactured. It is astronomical, it arrives twice a month whether or not anyone is ready, and the practitioner has no authorship over it at all.
Here's what the pair produces. Greene's manufactured deadline has a hidden weakness he doesn't dwell on: you know you set it, so you can unset it, and the intensity leaks out through that knowledge. An inherited deadline can't be unset — but it pays for that with a boundary problem, because a cutoff nobody controls is also a cutoff nobody can clarify. Hence ninety-six minutes, pūrva-viddhā, Mahādvādaśī: an entire arbitration apparatus that a self-imposed deadline never needs, because a self-imposed deadline is whatever you say it is.
So the trade is legible in both directions. Authorship over a deadline buys you clarity and costs you force; inheriting one buys you force and costs you clarity — and the cost shows up as procedural machinery, which is exactly what this limb is made of.
One handshake here, not two. The substitute-faster clause has obvious pull toward delegation and proxy performance elsewhere in the vault, but the source gives it three lines and one qualification, which is not enough to build a second connection on without inventing the argument.
Sharpest implication. The tradition treats an inherited, uncontrollable deadline as more binding than a chosen one — and pays for that with an arbitration apparatus, a transfer rule and an exemption list. The apparatus is not decadence creeping into a simple practice; it is the unavoidable cost of anchoring an obligation to something you do not control.
Generative questions.