After this chapter you will be able to say when a Sanskrit word is doubled, and what happens to it after the doubling. You will derive pacati pacati "he cooks again and again" and vṛkṣaṃ vṛkṣaṃ siñcati "he waters every tree". You will double pari "except" and upari "above", and a word of address such as cora cora "thief, thief!". You will see why the doubled eka gives ekaikam "each single one", with one case-ending, not two. You will see how paṭvī "skilful" becomes paṭupaṭvī "somewhat skilful", and how anyonyam and parasparam "each other" are built. The chapter ends with two ready-made forms, yathāyatham and dvandvam. It has twelve sūtras, SK 2139 to SK 2150, and it is the last chapter of the first half of the Kaumudī. The printed edition calls it dvirukta-prakriyā, "the section on the doubled".
The twelve rules are the opening rules of the eighth book of the Aṣṭādhyāyī, A 8.1.1 to A 8.1.15. Three rules of that stretch are taught elsewhere. A 8.1.2, which names the second member of a doubled pair āmreḍita, is SK 83 (chapter 3). A 8.1.3 belongs to the accent section, and A 8.1.6 to the Vedic section.
pacati → pacati pacati; vṛkṣam → vṛkṣaṃ vṛkṣamThis group sets up the whole chapter. SK 2139 is the heading: in place of a whole word, two word-forms come. SK 2140 is the first rule that uses it. A word is doubled when repetition of an action, or "every one" of a set, is to be made plain. So pacati "he cooks" becomes pacati pacati "he cooks again and again". And vṛkṣam "tree" (second case) becomes vṛkṣaṃ vṛkṣaṃ siñcati "he waters every tree".
Rule. This is an adhikāra, a heading. It says: "in place of the whole [word], two [word-forms come]". Its sense, "in place of the whole, two word-forms", is read into every rule to the end of the chapter, SK 2140 to SK 2150. Bhaṭṭoji writes only "taking this as the heading". One of Bhaṭṭoji's commentators supplies the rest: "the injunctions of doubling will be set out."
The name of the operation is dvirvacana, "doubling". Here it is a substitution. The word sarvasya "of the whole" names the thing replaced, the sthānin. The word dve "two" names the substitutes (ādeśa): two word-forms (śabdarūpe). So one word goes, and two identical word-forms take its place. Bhaṭṭoji's commentators both read the rule this way. One of them calls it the principal reading. Another view says the whole is simply pronounced twice; scholars discuss this. Bhaṭṭoji's commentators record that view and prefer the substitution reading.
Example.
pacati "he cooks" → the whole pada pacati is the thing replaced
→ two word-forms, pacati and pacati, come in its place
→ pacati pacati "he cooks again and again" (the sense comes from SK 2140).
Applies when / blocked by. Why "of the whole"? Bhaṭṭoji's commentators both answer. Without sarvasya, in a form like vṛkṣābhyām only the base vṛkṣa would be doubled. With it, the whole word, ending and all, is replaced.
One of Bhaṭṭoji's commentators gives two reasons why this is a heading and not a rule by itself. First, the rules after it name neither the operation nor the thing replaced, so they need it. Second, it is marked with the svarita accent, and a svarita-marked rule is a heading. The same commentator adds that what is carried forward is the heading's sense, "in place of the whole, two word-forms", not its bare wording. He cites SK 82 as evidence.
Watch out. "Two word-forms" does not mean two different words. The two forms are the same word. The second of them is the one SK 83 names āmreḍita.
Rule. A pada is doubled when nitya or vīpsā is to be made manifest. Nitya here is repetition: Bhaṭṭoji calls it ābhīkṣṇya, "again and again". Vīpsā is the speaker's wish to convey pervasion: that something holds of every member of a set. By the heading, two word-forms come in place of the whole pada.
The sense is already in the word itself. One of Bhaṭṭoji's commentators says the doubling only makes it manifest. The doubling is a signifier (dyotaka), not the bearer of the meaning.
Example.
Repetition:
pacati "he cooks" → doubled for repetition → pacati pacati "he cooks again and again".
bhuktvā "having eaten" → doubled for repetition → bhuktvā bhuktvā "having eaten again and again".
Pervasion:
vṛkṣam "tree" → doubled for pervasion → vṛkṣaṃ vṛkṣaṃ siñcati "he waters every tree".
grāmaḥ "village" → doubled for pervasion → grāmo grāmo ramaṇīyaḥ "every village is delightful".
Applies when / blocked by. Doubling for repetition is found in two kinds of words, Bhaṭṭoji says. They are words ending in a verb-ending (tiṅanta), like pacati, and kṛt forms that are avyaya, like bhuktvā. Bhaṭṭoji's commentators add that it is found only there. Both give the reason. Repetition belongs to the action, and the action is the principal thing only in these words.
The word padasya "of a pada" is drawn back into this rule from the later heading SK 401. Bhaṭṭoji's commentators both say so. One of them gives the purpose. The doubling for repetition must not apply to a bare root, and it must not block the affix yaṅ. Yaṅ is an affix added to a root that itself expresses repetition or intensity of the action. It is taught with the verbs, in the second half of the Kaumudī.
Watch out. "Every tree" does not mean every tree on earth. Bhaṭṭoji's commentators both say that the set is limited by context: the trees of this garden, which are being watered. Where nothing limits it, the set is unlimited. One of them cites jāto jāto nidhanam upaiti "every one who is born comes to his end". Note also that the doubled form stays singular. Each member is taken one by one, so no plural appears.
pari vaṅgebhyaḥ → pari pari vaṅgebhyaḥThree rules double particular words in particular senses. Pari is doubled when it means "except". Upari "above", adhi "over" and adhas "below" are doubled when they mean "close by". And a word of address at the head of a sentence is doubled in five feelings, from envy to threat. The running example is pari vaṅgebhyo vṛṣṭo devaḥ "the rain-god has rained except on the Vaṅgas". With pari doubled it becomes pari pari vaṅgebhyo vṛṣṭo devaḥ.
Rule. Pari is doubled when it is used in the sense of exclusion (varjana), "except". By the heading, two word-forms come in place of the whole.
Example.
pari vaṅgebhyo vṛṣṭo devaḥ → pari in the sense "except" is doubled
→ pari pari vaṅgebhyo vṛṣṭo devaḥ "the rain-god has rained, except on the Vaṅgas".
Bhaṭṭoji glosses it: "leaving out (parihṛtya) the Vaṅgas". One of Bhaṭṭoji's commentators supplies the rain-god (parjanya) as the subject of vṛṣṭaḥ.
Applies when / blocked by. Bhaṭṭoji's commentators both explain the grammar of the example. Pari is a karmapravacanīya in the sense "except" by SK 596. So vaṅgebhyaḥ takes the fifth case by SK 598.
A vārttika follows: parer varjane vāvacanam, "for pari in the sense of exclusion, the doubling is to be stated as optional". So pari vaṅgebhyaḥ, undoubled, is also correct.
One of Bhaṭṭoji's commentators adds another condition, "not in a compound". So there is no doubling in the compound paritrigartaṃ vṛṣṭo devaḥ "it has rained except on Trigarta". Scholars discuss the exact wording of this vārttika.
Rule. Upari "above", adhi "over" and adhas "below" are doubled when proximity (sāmīpya), nearness, is to be understood.
Example.
upari grāmam → upari in the sense "close above" is doubled
→ uparyupari grāmam "in the region close above the village".
adhaḥ lokam → adhas in the sense "close below" is doubled
→ adho 'dho lokam "in the region close below the world".
adhi sukham → adhi in the sense "close upon" is doubled
→ adhyadhi sukham "sorrow at a time close upon happiness".
Applies when / blocked by. The nearness may be in space or in time. Bhaṭṭoji's commentators both say that uparyupari grāmam and adho 'dho lokam show space, and adhyadhi sukham shows time.
Why "when nearness is meant"? One of Bhaṭṭoji's commentators answers. In upari candramāḥ "the moon is above", only the relation of higher and lower is meant. The same holds in upari śiraso ghaṭaṃ dhārayati "he holds a pot above his head". So there is no doubling.
Rule. A word of address (āmantrita) that stands at the beginning of a sentence is doubled in five senses. They are envy (asūyā), approval (sammati), anger (kopa), censure (kutsana) and threat (bhartsana). By the heading, two word-forms come in place of the whole. One of Bhaṭṭoji's commentators supplies the words "two come" to complete the sūtra.
Example.
Envy: sundara "handsome one" → doubled → sundara sundara vṛthā te saundaryam "Handsome one, handsome one! Your beauty is worthless." One of Bhaṭṭoji's commentators says it is said by one who cannot bear another's beauty.
Approval: deva "lord" → doubled → deva deva vandyo 'si "Lord, lord! You are worthy of worship."
Anger: durvinīta "insolent one" → doubled → durvinīta durvinīta idānīṃ jñāsyasi "Insolent one, insolent one! Now you will know." One of Bhaṭṭoji's commentators supplies what he will know: the fruit of his insolence.
Censure: dhānuṣka "archer" → doubled → dhānuṣka dhānuṣka vṛthā te dhanuḥ "Archer, archer! Your bow is worthless." One of Bhaṭṭoji's commentators says it reproaches one unfit for battle.
Threat: cora "thief" → doubled → cora cora ghātayiṣyāmi tvām "Thief, thief! I shall have you killed."
Applies when / blocked by. Threat usually comes from anger, and censure from envy. Why are they named apart? Bhaṭṭoji's commentators both answer. A teacher threatens a pupil without anger, and censures him without envy, wishing him well.
One of Bhaṭṭoji's commentators adds that the accent rules for an āmreḍita apply to these examples. They are svaritam āmreḍite… (SK 3622) and āmreḍitaṃ bhartsane (SK 3614), both in the accent section. He also notes that pluti is optional here and so not shown. Pluti is protraction of a vowel to three mātrās.
Watch out. The word of address must open the sentence. The doubling is for these five senses only.
ekam ekam → ekaikam; gatā gatā → gatagatāTwo rules make the doubled pair behave like a bahuvrīhi compound. This is an atideśa, a transfer: the pair is not a compound, but it is treated as one. The treatment makes the whole pair a single prātipadika, a noun stem. So the case-endings of both halves are deleted, one ending is added after the whole, and a feminine first half may become masculine. The running example is ekam ekam "one, one". Treated like a bahuvrīhi, it becomes ekaikam "each single one". Likewise gatā gatā "gone, gone" (feminine) becomes gatagatā.
Rule. The word eka "one", once doubled, is treated like a bahuvrīhi (bahuvrīhivat). Bhaṭṭoji names the result: deletion of the case-ending (sublopa) and treatment as masculine (puṃvadbhāva).
One of Bhaṭṭoji's commentators says the word dve "two" is carried forward. That is why Bhaṭṭoji writes "eka, when doubled". The other says, following the Bhāṣya, that this rule concerns only doubling for pervasion. The Bhāṣya is Patañjali's great commentary on the Aṣṭādhyāyī.
Example.
ekam ekam "one, one" (eka doubled for pervasion)
→ the whole pair is one prātipadika, because it is treated like a bahuvrīhi
→ the case-endings of both halves are deleted by luk → eka eka
→ the two halves join: a + e gives ai (vowel sandhi, chapter 3) → ekaika
→ one case-ending is added after the whole → ekaikam akṣaram "each single syllable".
Bhaṭṭoji gives the stem status, the deletion and the new ending. He leaves the vowel sandhi unstated.
The order matters. The pair first becomes a prātipadika. Only then can the endings be deleted, and only then can a new ending come after the whole. One of Bhaṭṭoji's commentators says the phrase "because it is a prātipadika" governs both steps, like a gem set in the middle of a string. Without the transfer, the pair would not be a prātipadika, and there could be no deletion. The other commentator adds that the new ending is singular only.
ekayā ekayā (feminine, third case; these steps are given by one of Bhaṭṭoji's commentators)
→ one prātipadika; both endings deleted
→ the first half is made masculine, ekā → eka
→ the third case again after the whole → ekaikayā āhutyā "with each single oblation".
Applies when / blocked by. In ekaikayā the masculine treatment of the first half changes nothing you can hear. The form is ekaikayā either way. Bhaṭṭoji says the difference shows in the avagraha. The avagraha is the division of a word into its parts in the pada-text, the word-by-word recitation of a Vedic text. Bhaṭṭoji's commentators both say it is eka-ekayā, not ekā-ekayā. One of them notes that the Taittirīya tradition divides it ekā ekayā.
Does the transferred bahuvrīhi lose the name sarvanāman? SK 222, na bahuvrīhau, withholds that name in a bahuvrīhi. Bhaṭṭoji reports the earlier authorities (prāñcaḥ). They said the word "bahuvrīhi" in SK 222 means the principal bahuvrīhi only. So a transferred bahuvrīhi keeps the name sarvanāman.
Bhaṭṭoji then gives his own view in two steps. In reality, the Bhāṣya rejects SK 222 altogether. And even if the sūtra is kept, it forbids the name only in the alaukika vigraha. That is the technical analysis made while the compound is being formed. It does not forbid the name in the finished bahuvrīhi. So the question does not even arise here.
ekaikasmai dehi "give to each one". One of Bhaṭṭoji's commentators explains. After both endings are deleted, the whole ends in eka, a word of the sarva-list. So the fourth-case ending ṅe is replaced by smai (SK 215).
Why does the sūtra say "bahuvrīhi" and not simply "compound"? Bhaṭṭoji's commentators both point to the accent. The transfer brings the accent of a bahuvrīhi, where the first member keeps its own accent (SK 3735, in the accent section).
Watch out. Only the first half becomes masculine. The second half keeps its feminine ending: eka-ekayā. One of Bhaṭṭoji's commentators gives the reason. The masculine treatment of SK 831 needs a following word that refers to the same thing. Only the first half has such a word after it.
Rule. When pain (ābādha) is to be made manifest, a word is doubled. By the word ca "and", the doubled word is also treated like a bahuvrīhi. Bhaṭṭoji states both: "two word-forms come, and it is like a bahuvrīhi." He glosses ābādha as pīḍā "pain".
Example.
gataḥ gataḥ "gone, gone" → one prātipadika by the bahuvrīhi treatment
→ both endings deleted by luk → gata gata
→ the ending after the whole → gatagataḥ.
Bhaṭṭoji says this is the cry of one tormented by separation. One of Bhaṭṭoji's commentators supplies the subject: "the time [spent] without the beloved". That is why the form is masculine: "[the time] is gone, gone!"
gatā gatā [priyā] "[the beloved] is gone, gone" → one prātipadika; both endings deleted
→ the first half becomes masculine, gatā → gata (SK 831)
→ gatagatā.
Applies when / blocked by. How can SK 831 apply? It needs a feminine later member (uttarapada) that refers to the same thing. Bhaṭṭoji's commentators both answer. The second gatā names the same woman as the first. And it counts as a later member through the transfer itself.
One of Bhaṭṭoji's commentators limits the transfer. It does not bring the samāsānta affixes, the closing affixes added to compounds. He gives the reason. Those affixes are restricted to compounds formed by the rules of the compound section. A pair only treated like a compound is not one of them. So the doubled forms are dhūrdhūḥ and panthāḥ panthāḥ, with no such affix.
Watch out. Our copy of this commentator reads priyā vinā kāla, with the nasal and the visarga lost. Another copy reads priyāṃ vinā kālaḥ "the time without the beloved", and the sense needs it. We follow that reading.
paṭvī paṭvī → paṭupaṭvīHere a second transfer begins. From SK 2146 onward, the doubled word is treated like a karmadhāraya. The fruits are three: the case-endings go, the first half may become masculine, and the whole takes the high accent on its last vowel. SK 2147 is the long rule of the chapter. It doubles a quality-word to mean "somewhat": paṭvī "skilful" (feminine) becomes paṭupaṭvī "somewhat skilful". Its vārttikas then add doubling for order, for alarm, for intense action, and for "each other". SK 2148 doubles priya and sukha to mean "with ease".
Rule. In the doublings that follow this rule, the operation is as for a karmadhāraya (karmadhārayavat). This is itself a heading. It covers SK 2147 to SK 2150.
A vārttika states the purpose: prayojanaṃ sublopapuṃvadbhāvāntodāttatvāni. "The purpose [of this]: deletion of the case-ending (sublopa), treatment as masculine (puṃvadbhāva), and final udātta (antodāttatva)." Antodātta means having the high accent on the last vowel. Bhaṭṭoji quotes it from the Bhāṣya, which has it in a slightly fuller wording.
Example.
The rule does nothing by itself. Its three fruits all show in the first example of SK 2147:
paṭvī paṭvī → case-endings deleted; first half masculine → paṭupaṭvī, with the high accent on its last vowel.
Applies when / blocked by. One of Bhaṭṭoji's commentators cites the Vṛtti, an older commentary on the Aṣṭādhyāyī. The heading alone would carry the transfer forward, so the words "in the following" are there only for clarity. The other commentator adds that the rule anudāttaṃ ca (SK 3670), though a heading, has no bearing on the final udātta. He follows the Bhāṣya in this.
Watch out. This transfer is a heading, not a one-rule transfer like SK 2144. It reaches every doubling to the end of the chapter, unless a later rule or vārttika says otherwise.
Rule. When similarity (prakāra) is to be conveyed, a guṇavacana, a word for a quality, is doubled. By the heading SK 2146, the doubled form is treated like a karmadhāraya. Bhaṭṭoji glosses prakāra as sādṛśya "similarity".
The word prakāra can also mean "difference". Bhaṭṭoji's commentators both say that "similarity" is fixed only by the tradition's explanation (vyākhyāna).
Example.
paṭvī paṭvī "skilful, skilful" (feminine)
→ treated like a karmadhāraya: one prātipadika, both endings deleted
→ the first half becomes masculine by SK 746, paṭvī → paṭu
→ the whole takes the high accent on its last vowel by samāsasya (SK 3734, in the accent section)
→ paṭupaṭvī "somewhat skilful (she)".
In the masculine, paṭupaṭuḥ. Bhaṭṭoji gives its meaning: paṭusadṛśaḥ, "like one who is skilful". That is to say, īṣatpaṭuḥ, "somewhat skilful".
Both kinds of quality-word are taken here, Bhaṭṭoji says. One kind names a substance in which a quality is subordinate. The other names the quality alone.
śuklaśuklaṃ rūpam "a whitish colour": the quality alone.
śuklaśuklaḥ paṭaḥ "a whitish cloth": a substance with the quality.
One of Bhaṭṭoji's commentators says the word rūpam "colour" shows that the first example names the quality alone. The other says this sentence rejects the view that a guṇavacana names only a substance.
Applies when / blocked by. Why is the karmadhāraya transfer needed, after the bahuvrīhi transfer? The bahuvrīhi treatment would already give paṭupaṭvī, through SK 831. Bhaṭṭoji's commentators both answer. The karmadhāraya treatment brings SK 746, and SK 746 reaches words that SK 831 does not. It reaches even words with k before the final vowel, such as kārikā. One of them gives kālakakālikā.
Now come the vārttikas on this sūtra, in Bhaṭṭoji's order. Each adds a new sense in which a word is doubled.
Vārttika 1: succession in order. Ānupūrvye dve vācye: "when succession in order (ānupūrvya) is to be conveyed, two word-forms should be stated".
mūle "at the root" → doubled for order → mūle mūle sthūlaḥ "at the root, at the root, [it is] thick". Each part nearer the root is thicker than the part beyond it.
One of Bhaṭṭoji's commentators says there is no karmadhāraya treatment here, following the Bhāṣya's example. So the seventh-case ending of mūle stays in both halves. The other says why this is not pervasion. A bamboo stick has only one real root. Other parts are "root" only relative to the parts above them.
Vārttika 2: alarm. Saṃbhrameṇa pravṛttau yatheṣṭam anekadhā prayogo nyāyasiddhaḥ. "When someone acts in alarm (saṃbhrama), using the word many times, as often as one wishes, is established by reasoning."
sarpa "snake!" and budhyasva "wake up!" → repeated in alarm
→ sarpa sarpa, budhyasva budhyasva "Snake, snake! Wake up, wake up!"
→ or three times: sarpa sarpa sarpa, budhyasva budhyasva budhyasva.
One of Bhaṭṭoji's commentators explains the wording. Saṃbhrama is haste caused by fear and the like. The word anekadhā "many times" suspends the "two" of the heading. Yatheṣṭam "as one wishes" allows even a single use. And there is no karmadhāraya treatment, so the ending stays. The other commentator explains "by reasoning": use as many words as the hearer needs to grasp the meaning.
Vārttika 3: repetition or intensity of an action. Kriyāsamabhihāre ca: "and when kriyāsamabhihāra, repetition or intensity of an action, is to be conveyed, two word-forms come".
lunīhi "reap!" → doubled → lunīhi lunīhīty evāyaṃ lunāti "this man reaps in the manner 'Reap! Reap!'".
One of Bhaṭṭoji's commentators gives the grammar. Lunīhi is the imperative of lū "to cut, reap". The imperative is named loṭ; here it comes in the sense of repeated or intense action (SK 2825). The following lunāti is an anuprayoga, a finite verb of the same root added after it (SK 2827). Both rules are taught with the verbs, in the second half. The whole means intense reaping by a single agent.
Bhaṭṭoji then asks why the vārttika is needed. Doubling for repetition is already given by SK 2140. So the vārttika serves doubling in the sense of intensity (bhṛśa). Or else it lets the doubling signify repetition jointly with the loṭ. Bhaṭṭoji's commentators both draw the consequence. A yaṅ form such as pāpacyate "he cooks intensely" is not doubled. One of them gives the reason. Loṭ does not always mean repetition, so it needs the doubling beside it. Yaṅ always means it, and signifies it alone.
Vārttika 4: exchange of action. Karmavyatihāre sarvanāmno dve vācye, samāsavac ca bahulam. "When exchange of action (karmavyatihāra) is to be conveyed, two word-forms of a pronoun (sarvanāman) should be stated. And the doubled form is variously (bahulam) treated like a compound." Exchange of action means each does the act to the other.
Bhaṭṭoji's commentators both say the doubling itself is obligatory. The word bahulam goes only with the compound treatment. Bhaṭṭoji draws the result. Because of bahulam, doubled anya "other" and para "other" are not treated like a compound. Doubled itara "other" is always treated like one. One of Bhaṭṭoji's commentators adds that other pronouns are not doubled in this sense at all.
Vārttika 5: su for the first half. Asamāsavadbhāve pūrvapadasthasya supaḥ sur vaktavyaḥ. "Where there is no treatment like a compound, su should be stated as the substitute for a case-ending in the first member." So this vārttika concerns anya and para. One of Bhaṭṭoji's commentators says any case-ending of the first member, of any case, is replaced by su, the first-case singular ending.
anyam anyam → the first member's ending am is replaced by su → anya + su (the second anyam is untouched)
→ the s of su becomes ru (SK 162)
→ ru becomes u (SK 163) → anya + u
→ a + u gives o by guṇa (SK 69) → anyo
→ anyo + anyam → anyo 'nyam
→ anyo 'nyaṃ viprā namanti "the brahmins bow to one another".
One of Bhaṭṭoji's commentators gives these steps (su, ru, u, guṇa); Bhaṭṭoji gives only the result.
Likewise anyo 'nyau, anyo 'nyān, anyo 'nyena kṛtam "done by one another", anyo 'nyasmai dattam "given to one another". Only the second half changes its ending; the first half stays anyo.
anyo 'nyam and anyonyam are one form written two ways. Our text writes anyo 'nya- with the sign '. The printed edition writes anyonya- throughout.
Bhaṭṭoji cites the poet Māgha: anyo 'nyeṣāṃ puṣkarair āmṛśantaḥ "touching one another with the tips of their trunks". Bhaṭṭoji's citations answer an older view. One of his commentators states that view: this doubling is for the first-case singular only. The other describes it differently: su replaces only endings from the second case on, and in the singular only. Both say that the Bhāṣya's examples, and Māgha's usage, go against it. Scholars discuss this.
paraḥ param → parasparam "each other". Bhaṭṭoji says the visarga is replaced by s, because paraspara is in the kaskādi list (SK 144).
itaraḥ itaram → treated like a compound: both endings deleted, a fresh ending after the whole → itaretaram, itaretareṇa and so on. One of Bhaṭṭoji's commentators gives these steps.
Vārttika 6: ām in the feminine and neuter. Strīnapuṃsakayor uttarapadasthāyā vibhakter āmbhāvo vā vaktavyaḥ. "In the feminine and the neuter, ām should optionally be stated as substitute for the case-ending of the second member."
anyonyām or anyonyam; parasparām or parasparam; itaretarām or itaretaram.
ime brāhmaṇyau kule vā bhojayataḥ "these two brahmin women (or: these two families) feed each other". One of Bhaṭṭoji's commentators explains. This woman or family feeds the other, and the other feeds this one.
Bhaṭṭoji reports two views on the reach of ām.
The verse on bahula. Bhaṭṭoji then quotes a verse. One of his commentators calls it an older verse. It lists what the word bahula in Vārttika 4 achieves:
daladvaye ṭābabhāvaḥ klībe cādḍvirahaḥ svamoḥ | samāse sor aluk ceti siddhaṃ bāhulakāt trayam ||
"Absence of ṭāp in both halves; in the neuter, absence of adḍ for su and am; and non-deletion of su in a compound: these three are established from bāhulaka." Bāhulaka means the force of the word bahula.
1. In the feminine anyonyam and parasparam, no feminine ṭāp appears in either half.
2. In the neuter anyonyam and itaretaram, adḍ does not replace su and am.
3. When the doubled form enters a further compound, the su of the first half is not deleted. The name for non-deletion is aluk.
Why each of the three needs bahula. Bhaṭṭoji shows that each result would otherwise go wrong.
anyonyam and parasparam, ṭāp is due in both halves. Could the vārttika sarvanāmno vṛttimātre [puṃvadbhāvaḥ] remove it? That vārttika makes a pronoun masculine in any vṛtti. A vṛtti is a compound or similar derived formation. But anya and para are not treated like a compound. Nor is doubling itself a vṛtti. If it were, the masculine would wrongly come in yāṃ yāṃ priyaḥ praikṣata kātarākṣī sā sā "whichever woman the beloved looked at, she, the timid-eyed one…". One of Bhaṭṭoji's commentators lists the vṛttis as five: kṛt, taddhita, compound, ekaśeṣa, and the derived verb stems.anyonyam and itaretaram, adḍ is due by SK 315.anyonyasaṃsaktam ahastriyāmam "day and night joined to each other", deletion of the first half's su is due. So too in anyonyāśrayaḥ "mutual dependence", parasparākṣisādṛśyam "the likeness of each other's eyes", and aparasparaiḥ. One of Bhaṭṭoji's commentators analyses each of them as a compound, and in a compound the inner case-endings are deleted. (This is the rule of SK 650, chapter 17.)Bhaṭṭoji concludes: all this is settled by the force of bāhulaka. His authorities are two passages of the Bhāṣya. One is the Bhāṣya's example to this very vārttika. The other is the phrase anyonyasaṃśrayaṃ tv etad under striyām (SK 453).
Watch out. Five places in this unit correct a slip in our text by the printed edition.
Two points of grammar are easily missed too. Not every doubling in this unit is compound-like: mūle mūle and sarpa sarpa keep their endings. And under Vārttika 4, anya and para are never compound-like, while itara always is.
Rule. Priya "dear" and sukha "easy" are optionally doubled when they denote absence of hardship (akṛcchra), ease. The doubled form is treated like a karmadhāraya, by the heading SK 2146.
Example.
priyeṇa priyeṇa → treated like a karmadhāraya: both third-case endings deleted by luk → priya priya
→ the same ending again after the whole → priyapriyeṇa dadāti.
Or, undoubled: priyeṇa dadāti.
Likewise sukhasukhena dadāti, or sukhena dadāti.
Bhaṭṭoji gives the meaning: "he gives even a very dear thing without effort".
Applies when / blocked by. One of Bhaṭṭoji's commentators glosses kṛcchra as hardship and akṛcchra as ease. He says the karmadhāraya treatment comes from the heading SK 2146. It does not come from the words "like a compound" in Vārttika 4 of SK 2147, since that is only a vārttika.
Watch out. The doubling is optional (anyatarasyām). Both priyapriyeṇa dadāti and priyeṇa dadāti are correct.
yathā → yathāyatham; dvi dvi → dvandvamThe chapter ends with two ready-made forms. Nipātana is the fixing of a form as correct simply by stating it in a sūtra. Whatever in the form no other rule accounts for is licensed by that statement. SK 2149 fixes yathāyatham "each according to its own nature". SK 2150 fixes dvandvam "in pairs, in secret", built from dvi "two" doubled.
Rule. In the sense of yathāsva, "each according to what is its own", the form yathāyatham is fixed by nipātana.
Bhaṭṭoji first explains yathāsvam. It is an avyayībhāva compound in the sense of pervasion. Its meaning is "whatever is one's self, and whatever belongs to oneself". In that sense two things are fixed by nipātana: the doubling of the word yathā, and neuter gender.
Example.
yathā → doubled by nipātana → yathā yathā
→ neuter gender, fixed by nipātana; so the final vowel is short (one of Bhaṭṭoji's commentators)
→ yathāyatham.
yathāyathaṃ jñātā "one who knows [things] each according to its own nature". Bhaṭṭoji gives the meaning as yathāsvabhāvam "according to its own nature", or yathātmīyam "according to what belongs to it".
Applies when / blocked by. Why must the doubling be fixed by nipātana? Why does SK 2140 not give it? One of Bhaṭṭoji's commentators answers. SK 2140 doubles a word only when the word itself names its object as qualified by totality. Otherwise sarva "all" in sarvo ghaṭaḥ "the whole pot" would be doubled too. Yathā does not work that way here. The editor of the printed edition says the same in a footnote.
Why the neuter? The same commentator explains. Being an avyaya, the form would otherwise have no gender at all. The neuter gives it a gender, and with it the short final vowel.
Why is yathāyatham in the second case with jñātā, not the sixth? Bhaṭṭoji's commentators both answer. Jñātā ends in the agent affix tṛn, so SK 627 bars the sixth case. One of them adds that with the affix tṛc it would be yathāyathasya jñātā.
Watch out. Sva can mean self, one's own, a kinsman, or wealth. One of Bhaṭṭoji's commentators says that here only the first two are meant, by the tradition's explanation.
Rule. The form dvandvam is fixed by nipātana in five senses. They are a secret (rahasya), a limit expressed (maryādāvacana), separation (vyutkramaṇa), the use of sacrificial vessels (yajñapātraprayoga), and manifestation (abhivyakti).
Bhaṭṭoji lists what nipātana fixes. The word dvi "two" is doubled. The final of the first member is replaced by am. The final of the second member is replaced by a. And the whole is neuter.
Example.
dvau dvau "two, two" → treated like a karmadhāraya by the heading SK 2146: both endings deleted → dvi dvi
→ am for the final i of the first member, by nipātana → dvam dvi
→ a for the final i of the second member, by nipātana → dvandva (the m becomes n before d; the sources do not state this step)
→ neuter, treated as singular; a case-ending comes after the whole
→ after a neuter stem in a, SK 309 replaces su (and am) with am → dvandvam.
Do not confuse the two *am*s. The first replaces the final of dvi and is fixed by nipātana. The second replaces the case-ending, by SK 309.
Bhaṭṭoji's commentators both give the karmadhāraya step. One of them reads the singular treatment, and so SK 309, from the word ca in Bhaṭṭoji's list.
Of the five senses, only a secret is what the word dvandva denotes (vācya), Bhaṭṭoji says. The other four are domains (viṣaya) in which it is used.
dvandvaṃ mantrayate "he takes counsel in secret". One of Bhaṭṭoji's commentators gives the thought behind it: "two by two, for so it is secret".ācaturaṃ hīme paśavo dvandvaṃ mithunīyanti "for these animals, up to the fourth [generation], desire to mate pair by pair". The mother mates with the son, and also with the grandson and great-grandson, taking the fourth as the limit.Ācaturam "up to the fourth" is an avyayībhāva by SK 667, with the closing affix ṭac (SK 677). Bhaṭṭoji's commentators both say so. One of them reads the ā as inclusive: "up to and including the fourth". Mithunīyanti "they desire mating" takes the affix kyac, which makes a verb meaning "to desire" something.
dvandvaṃ vyutkrāntāḥ "they have separated in pairs": they stand apart through their connection with two groups.dvandvaṃ yajñapātrāṇi prayunakti "he sets out the sacrificial vessels in pairs". One of Bhaṭṭoji's commentators names such vessels: the wooden sword (sphya) and the potsherds (kapāla).dvandvaṃ saṅkarṣaṇavāsudevau "Saṅkarṣaṇa and Vāsudeva are a pair". Bhaṭṭoji glosses: through their constant association (sāhacarya). One of his commentators adds that they are known in the world through it.Applies when / blocked by. Bhaṭṭoji adds a last line. By yogavibhāga, splitting the rule so that dvandvam stands as a rule of its own, the form is accepted elsewhere too. Bhaṭṭoji's commentators give examples.
dvandvāni sahate "he bears the pairs" (both commentators). One of them names the pairs: heat and cold, pleasure and pain, hunger and thirst.
dvandvaṃ yuddhaṃ pravartate "a fight goes on, two by two", with doubling for pervasion (the other commentator).
Watch out. Two places here follow the printed edition over our text.
Keep the senses apart as well. Dvandvam denotes only "in secret". In the other four senses the same form is used, but it does not name that sense.
| SK | What it does | Example |
|---|---|---|
| 2139 | Heading: in place of the whole word, two word-forms come | pacati → pacati pacati |
| 2140 | A pada is doubled for repetition or pervasion | pacati pacati; vṛkṣaṃ vṛkṣaṃ siñcati |
| 2141 | Pari "except" is doubled; optional by a vārttika | pari pari vaṅgebhyaḥ / pari vaṅgebhyaḥ |
| 2142 | Upari, adhi, adhas are doubled for nearness | uparyupari grāmam, adhyadhi sukham |
| 2143 | A sentence-opening word of address is doubled in five feelings | cora cora ghātayiṣyāmi tvām |
| 2144 | Doubled eka is treated like a bahuvrīhi: one stem, one ending, masculine first half | ekaikam, ekaikayā, ekaikasmai |
| 2145 | Doubling for pain, also like a bahuvrīhi | gatagataḥ, gatagatā |
| 2146 | Heading: the doublings that follow are like a karmadhāraya; vārttika on its purpose | paṭupaṭvī |
| 2147 | A quality-word is doubled for "somewhat"; vārttikas on order, alarm, intense action, "each other", ām; the verse on bahula | paṭupaṭvī; mūle mūle; sarpa sarpa; anyonyam, parasparam, itaretaram |
| 2148 | Priya, sukha optionally doubled for ease | priyapriyeṇa dadāti / priyeṇa dadāti |
| 2149 | Yathāyatham fixed by nipātana | yathāyathaṃ jñātā |
| 2150 | Dvandvam fixed by nipātana in five senses, and elsewhere by yogavibhāga | dvandvaṃ mantrayate |
With this chapter the first half of the Kaumudī, the pūrvārdha, comes to an end. In the printed edition the section closes with iti dvirukta-prakriyā, followed by the colophon of the first half. The first half has taught the sounds and their sandhi, the declension of nouns, the feminine, the cases, the compounds, the taddhitas, and now the doubled word. The second half, the uttarārdha, opens with the verbs (tiṅanta). Chapter 44 begins it with vartamāne laṭ (A 3.2.123, SK 2151), which gives the affix laṭ for present time.