कालः — time, tithi, and writing a date

Why "for a month" and "in a month" are different words

2.3

English gets away with one preposition family — "for a month," "within a month," "at night" — and lets context sort out which relation is meant. Sanskrit does the sorting in the case ending itself, before the sentence is even finished, and the sorting rule is exact enough to state in one line each.

विभक्तिः	the relation	rule	what it looks like

All three read a कालशब्द the same grammatical way. None of them treats "a month" as an adverb bolted onto the sentence. Each treats it as an ordinary noun in an ordinary case — the case is simply chosen by how the time relates to the action, not by the word itself. मास takes द्वितीया in one sentence and तृतीया in the next; nothing about मास changes, only the relation being expressed.

The three rules, with the grammar's own examples

2.3.5 · 2.3.6 · 2.3.36

The examples below are not written for this page. They are the Kāśikā's own, quoted from its commentary on each sūtra — the same way this page pulls a paradigm, but for prose.

द्वितीयाDuration — an unbroken stretch of time

The rule

— 2.3.5 — a कालशब्द (a time-word) or an अध्वशब्द (a distance-word) takes द्वितीया when it names an अत्यन्तसंयोग: the action is in contact with the whole of that stretch, not just a point inside it. "He studies," continuously, for the entire month — not once, somewhere in it.

The same sūtra covers distance. अध्वन् — a road, a distance — takes the identical द्वितीया for the identical reason: क्रोशम् अधीते, "he studies for a krośa's walk," treats a stretch of road exactly the way मासम् treats a stretch of time. Time and distance are the same kind of thing to this rule: an extent, wholly covered.

तृतीयाTime within which — the action finishes inside it

The rule

— 2.3.6 — the same कालाध्वनोः अत्यन्तसंयोग relation from 2.3.5 is still in force by अनुवृत्ति, but now with one more condition: अपवर्ग, the action reaching its actual completion. Not "he spent a month at it" — "a month, and by the end of it, it was done."

The difference is not the noun, it is the participle. अधीते (present, "is studying") takes द्वितीया — the studying simply fills the month. अधीतः (past participle, "has been [fully] studied") takes तृतीया — the month is the container the completion happened inside. Same root, same कालशब्द, two case endings, because the two sentences claim two different things about what happened to the month.

सप्तमीThe point in time — an ordinary locative

The rule

— 2.3.36 — is not a time-specific rule at all. It is the general rule that gives सप्तमी to any अधिकरण, any locus an action occurs in — a mat, a village, a pot. The Kāśikā's own examples are all spatial:

Time is a locus exactly the way a mat is one. 2.3.36 does not single out कालशब्द the way 2.3.5 and 2.3.6 do by name — it does not need to. A moment is simply another kind of place an action can sit inside, so the identical rule produces the identical case when the locus happens to be temporal:

Every form above is C.paradigm(...)'s own सप्तमी cell — the rule does not change when the locus changes from cloth to calendar, and neither does the case it assigns.

Try the contrast yourself

Pick a relation. The example, the case and the sūtra all come from the same three rules above.

Dividing the day

दिनविभागाः · कालमानम्

Six named parts of a single अहोरात्र (day-night), and a chain of traditional units below them — most attested here as ordinary nouns, two not attested under any spelling tried.

प्रातः … रात्रिःThe six parts of a dayदिनविभागाः

मुहूर्तः · यामः · घटिका · निमेषःA traditional chain of units — not a Pāṇinian oneDharmaśāstra/jyotiṣa convention

None of these ratios comes from a sūtra — no rule anywhere in the Aṣṭādhyāyī's index counts kṣaṇas. But the words themselves are ordinary nouns, and several of their own corpus glosses happen to state the traditional ratio directly:

शब्दः	अर्थः (corpus gloss)

The week: a planet, plus वासर

सप्तवासराः

All seven weekday names are transparent: a planet's name, compounded with वासर, "day" — and the compound itself declines exactly like any other अ-stem masculine noun, taking सप्तमी for "on [that day]" the same way every other point-in-time above did.

ग्रहः	अर्थः (gloss)	वासरः	सप्तमी — "on that day"

पक्षः and तिथिः — the fortnight and the lunar day

शुक्ल-कृष्णपक्षौ

A synodic month splits into two पक्ष — a waxing half and a waning half — and each पक्ष into fifteen तिथि. This is where the corpus's coverage runs out fastest: six of the fifteen names are declinable nouns here, and nine are not attested under any spelling.

शुक्लपक्षः · कृष्णपक्षःThe two halves — and a gloss that confirms them by accidentपक्षः — पञ्चदशदिनानि

पक्ष itself is glossed in this corpus as "" — fifteen days, among other senses — which is exactly a fortnight. Better still, the two adjectives that name the halves are independently glossed with the fortnight sense built in: शुक्ल ("bright, white") is glossed "," which names पूर्वपक्षः — the earlier fortnight — as one of its senses; कृष्ण ("dark") is glossed "," naming अपरपक्षः — the later fortnight. Neither gloss was written with this page in mind; both happen to state exactly the fact this page needs.

The compound itself — शुक्लपक्षः, कृष्णपक्षः — is not an attested headword here, so it is built the same way वासर's compounds were: the adjective's bare stem in front of पक्ष's own declension. शुक्लपक्षे, कृष्णपक्षे — सप्तमी, the point-in-time case again.

प्रतिपद् … चतुर्दशीFifteen names, six of them attestedतिथयः

The first four names and the fifteenth are ordinary nouns — प्रतिपद् is a consonant stem, द्वितीया/तृतीया/चतुर्थी/पूर्णिमा/अमावास्या are आ- or ई-stems, exactly the declension families page 5 already taught. द्वितीया and चतुर्थी are the same homographs page 10 met as ordinals — चतुर्थी's own gloss there already read "completing four, [also] a particular tithi," which is this page's fact, met early.

#	नाम	सप्तमी	अर्थः (gloss)

The twelve months

मासाः · नक्षत्राधारः

Every one of the twelve month-names is a full, attested masculine noun — the best-covered table on this page. Each is traditionally named for the nakṣatra the moon is full in during that month; that naming convention is not a Pāṇinian rule, so it is stated here, not pulled — but where the nakṣatra's own name is independently in this corpus, its gloss is shown too, and read carefully rather than assumed to agree.

मासः	सप्तमी	अर्थः	नक्षत्राधारः

Writing a date

पत्रकः · Try it

A colophon dates a manuscript by stringing five elements together — era, year, month, पक्ष, तिथि, and often the वासर — almost all of them in सप्तमी, because a date is exactly what 2.3.36 already covers: a point in time is a locus, and every element that narrows down which point is another locus nested inside it.

इति सम्पूर्णम्One worked examplea colophon's own grammar

A real colophon might read, spelled out in full rather than abbreviated:

"In the Vikrama era, year 1550, in the month of Śrāvaṇa, in the bright fortnight, on the tithi Tṛtīyā, on Thursday." Read it left to right and every word but the numeral is a सप्तमी this page has already pulled: विक्रमसंवत्सरे (काल, general), श्रावणे (मास), शुक्लपक्षे (पक्ष), तृतीयायां तिथौ (तिथि), गुरुवासरे (वासर). The year itself is written as a plain positional numeral — exactly page 10's अङ्काः, not a declining word at all, which is why it needs no case ending of its own.

Or the year can be words, not digits. A poet who wants the date inside a metrical verse cannot use digits — a verse cannot scan around "१५५०." That is exactly the job page 10's भूतसङ्ख्या decoder does: pick nouns whose traditional count spells the year, read right to left. A colophon that says वेदाब्धिचन्द्रवसुसंख्ये शरदि विक्रमे is naming the same four digits page 10 taught how to decode, not composing new poetry — go run it through that page's decoder now that you know what it is for.

वर्षम्

मासः

पक्षः

तिथिः

वासरः

every element — सप्तमी

Which year: the two eras

विक्रमसंवत् · शकसंवत्

Manuscript colophons almost never give a Common-Era year — they give a year in one of two running eras, and the offset between them and CE has to be learned, not derived. Neither offset is grammar; both are stated here as historical convention.

विक्रम itself is not glossed as an era. This corpus's own entry for विक्रम reads "" — great valour, a stride — the word's ordinary sense, not a reference to King Vikramāditya or a calendar. The eponymous era is history layered onto an ordinary word, the same way the tithi names द्वितीया and चतुर्थी carry a second, unrelated sense each. संवत् and शक are not in this corpus under any spelling tried at all — they are conventional abbreviations, not declining words this page can pull.

Everything this page checked, attested or not

Coverage

Group	Stem	Liṅga checked	Status

Where this page's grammar comes from. The three case-rules, their sūtra text and their worked examples are the corpus's own — the examples are the Kāśikā's, quoted rather than composed, via C.layer('kashika', ref). Every declined form — months, the six attested tithis, the seven weekday planets, the traditional time-units — is C.paradigm(...), checked live against C.has(...) at build time. Not pulled, and marked wherever it appears: the calendar conventions themselves (which noun names which count of days, which planet governs which weekday, the nakṣatra a month is named for), the traditional time-unit ratios not stated in a corpus gloss, the two eras and their CE offsets, and the positional-digit year in the date builder — none of these are Pāṇinian facts, and none is claimed to be one.

Check yourself

अभ्यासः

Six questions. Every answer names the rule that settles it.