204. Compounds a topic of Syntax.

In western languages compounds are not considered a topic of Syntax. The fact of their being made up of two or more self-existent words — however important for the etymologist — has little or nothing to do with their employment in speaking or writing. In Sanskrit it must be otherwise. Keeping apart such compounds as have got any special meaning, which stamps them to unities (*1), there exists in that language an almost illimited freedom of expressing any kind of relations, grammatical or logical, by the way of compounding. Every moment the speaker and especially the writer of Sanskrit may have the opportunity of substituting compounds to the analytical mode of expression. For this reason, Sanskrit Syntax has to deal with compounds, as far as regards giving an account of the part they are acting in the phraseology and of the modes and ways how to employ them, whereas it is a topic of Sanskrit Etymology to expound their structure and their outer shape.

The three great classes of compounds, set up by vernacular grammar, are dvandva, tatprusha, bahuvrîhi. They include nearly all varieties as well of the simple compounds, which are made up of but two non-compound words, as of those, which are most intricate and of an immoderate length. Outside of them, there remains only the class of such compounds as are produced by putting together the preposition + the noun-case depending on it, as ākaṇṭham etc. (152) (*2); in most cases this fourth class coincides with the avyayîbhâva of vernacular grammar (*3).

(*1)
Such as himālaya when the name of the mountain, mahātman »noble,” bhūsura »brahman,” saptarṣayaḥ the well-known constellation, abalā »woman,” and the like.

(*2)
Whitney Sanskrit Grammar § 1310 calls them »prepositional compounds.”

(*3)
All compound adverbs, the former member of which is an indeclinable word, are comprehended by the general appellation avyayîbhâva (P. 2, 1, 5-16). Moreover this category contains some few kinds of compound adverbs, whose former member is a noun-case or an adj.; they are summed up by Pâṇini (2, 1, 17-21). — But compound adjectives are never styled avyayîbhâva, even if their former member be an indeclinable word. So for example when saying rāmaḥ sasītaḥ, we have a »prepositional compound” indeed, sasītaḥ being = saha sītayā, but not an avyayîbhâva.

204. 統語論のトピックとしての複合語

西洋諸言語では、複合語は統語論の1トピックとしては見なされていない。複合語が2つ以上の自立語から成るという事実は、語源学者には重要であるけれども、読み書きにおけるそれらの用法にはほとんど、あるいはまったく寄与しない。サンスクリットにおいては、決してそうではない。深く結合した特殊な意味を持つもの（*1）を除けば、サンスクリットは、複合法によって、文法的・論理的なあらゆる種類の関係をほぼ無制限に表現することができる。サンスクリットの話者、そしてとりわけ筆記者は、あらゆる瞬間において、分析的な表現方法の代わりに複合語を用いる機会がある。このため、複合語の構造や外形を解釈するのが語源学のトピックであるのに対し、サンスクリットのシンタックスは、複合語が句法の中でどのような役割を果たしているのかであるとか、その使用方法や様式を説明するかぎりで複合語を扱わなければならない。

土着文法家により設けられた複合語の大分類は、dvandva、tatpuruṣa、bahuvrīhiの3つである。これらは、単純な複合語、単に複合語でない2語で構成されたもの、至極複雑で法外に長いものなど、ほとんどすべての種類を含む。それらの他には、ākaṇṭham（152）（*2）などの、前置詞＋それに係る名詞-格を組み合わせることによって作られる類の複合語だけが残っている；ほとんどの場合、この4番目の分類は土着文法のavyayībhāvaに一致する（*3）。

（*1）
himālaya（ヒマラヤ山）、mahātman（高貴な）、bhūsura（ブラフマン神）、saptarṣayaḥ（リシの七星＝北斗七星）、abalā（女性）など。

（*2）
WhitneyのSanskrit Grammar §1310では、これを前置詞複合語（prepositional compounds）と呼ぶ。

（*3）
前分が不変化辞のあらゆる複合副詞は、一般名avyayībhāva（P. 2, 1, 5-16）によって理解される。さらに、このカテゴリは、前分が名詞-格や形容詞の複合副詞を含む；これらはPāṇiniが概説している（2, 1, 17-21）。けれども、複合形容詞は、その前分が不変化辞であったとしても、avyayībhāvaと呼ばれることはない。例えばrāmaḥ sasītaḥと言う場合、saha sītayāと等しいsasītaḥはまさしく「前置詞複合語」であるが、avyayībhāvaではない。

205. Dvandva.

The dvandva serves to express concatenation and addition. Two or more nouns linked to another by „and” may be united into a dvandva. {P. 2, 2, 29.} So instead of saying rāmo lakṣmaṇaś ca we may use the dvandva rāmalakṣmaṇau, instead of striyo bālāś ca vṛddhāś ca it may be said strībālavṛddhāḥ.

Rem. In the archaic dialect the freedom of making dvandvas was very little. At least in the vaidik writings dvandvas are almost bound to set formulae and do never consist of more than two members. Most of them are dvandvas of divinities, especially in the mantras, such as mitrāvaruṇā, indrāgnī. See Whitney, Sanskrit Grammar § 1255 and 1256.

205. dvandva

dvandvaは、連結と付加を表すことに資する。「…と」（and）で接続される2つ以上の名詞がdvandvaとなる。{P. 2, 2, 29.} よって、rāmo lakṣmaṇaś caの代わりにdvandvaのrāmalakṣmaṇauとするとか、striyo bālāś ca vṛddhāś caの代わりにstrībālavṛddhāḥとできる。

【補足】
古代方言においては、dvandvaを形成する際の自由度はかなり低かった。少なくとも、ヴェーダの作品においては、dvandvaはほとんど形式的でなくてはならず、3つ以上の要素で構成されることは決してない。とりわけマントラでは、ほとんどは神性のdvandvaである：mitrāvaruṇā、indrāgnī。WhitneyのSanskrit Grammar、§1255-1256をみよ。

206. Its gender and number.

The dvandva has the gender of its last member. Its number is determined by the real number of the persons or things, comprised by it. Panc. I, 4 ajātamṛtamūrkhebhyo mṛtājātau sutau varam (from the three classes of sons: 1 not born at all, 2 sons died, 3. blockheads, the first and second classes are to be preferred), ibid. p. 195 prāṇāntikaṃ sadaiva vāyasolūkānāṃ vairam (there always is a deadly hatred between crows and owls), Harshac. p. 28 nṛtyagotavāditreṣvabāhyāḥ (no strangers to dancing, singing and playing on instruments). As the number of the members is illimited, we may have such long dvandvas as f. i. Nala. 1, 28 vayaṃ hi devagandharvamānuṣoragarākṣasāndṛṣṭavantaḥ etc.

But if the dvandva is to represent a real unity or if not individuals but categories are linked together, it generally is á neuter and á singular. So it is said putrapautram (children and grandchildren), gavāśvam (kine and horses), ārāśastri (awl and knife).

Rem. 1. A full and exhaustive account on this subject is given by Pâṇini (2, 4, 2-17) and his commentators. They distinguish between those cases where the dvandva must be á singular and á neuter, those where it must not be so, and those in which it is allowed to use either idiom optionally. So, among others, the singular of the neuter is of necessity with dvandvas signifying parts either of the body or of musical instruments or of the army, as pāṇipādam, rāyikapādātam {P. 2, 4, 2.}, likewise if names of rivers and countries, when of different gender, are linked together, as gaṅgāśoṇam, kurukurukṣetram. {P. 2, 4, 7.} On the other hand, dvandvas made up of nouns denoting animate beings are not allowed to be put in the singular number, save a.) very small animals, as daṃśamaśakam {P. 2, 4, 8.}, b.) such as by their nature are living in eternal mutual enmity, as mārjāramūṣakam {P. 2, 4, 9.}, c.) classes of çûdras, not considered abject (*1), as takṣāyaskāram {P. 2, 4, 10.}, d.) some others as gavāśvam, putrapautram, śvacaṇḍālam, strīkumāram. {P. 2, 4, 11.} — Dvandvas of contrasting qualities or things are optionally put in the singular or in the dual, as sukhaduḥkham or -khe, śītoṣṇam or -ṣṇe. {P. 2, 4, 13.} And so on.

Rem. 2. It is forbidden to compound a genus with its species. See Pat, I, p. 252.

Rem. 3. Instead of the dvandva mātāpitarau the simple pitarau may be used {P. 1, 2, 70 and 71.} (so f. i. Ragh. 1, 1 jagataḥ pitarau vande pārvatīparameśvarau. Likewise śvaśurau is synonymous with śvaśrūśvaśurau, see f. i. Kathas. 58, 89. — Dvandvas = »brother and sister,” »son and daughter” are not used, here bhrātarau, putrau are of necessity. {P. 2, 1, 68.} — Cp. Latin soceri = socer et soclus, fratres = frater et soror.

Rem. 4. As to the order, to be followed in putting together the links of a dvandva, fixed rules cannot be given. {P. 2, 2, 32-34.} Yet it is common to put at the head either the themes ending in i or u, or those commencing by a vowel while ending in a, or the shortest.

(*1)
P. 2, 4, 10 śūdrāṇām aniravāsitānām. Kâç. yair bhukte saṃskāreṇāpi na śudhyati ta niravasitāḥ

206. dvandvaの性・数

dvandvaは、その最後の要素の性をもつ。数は、それが意味する人・物の実際の数により決まる。
Panc. I, 4 ajātamṛtamūrkhebhyo mṛtājātau sutau varam (from the three classes of sons: 1 not born at all, 2 sons died, 3. blockheads, the first and second classes are to be preferred)
ibid. p. 195 prāṇāntikaṃ sadaiva vāyasolūkānāṃ vairam (there always is a deadly hatred between crows and owls)
Harshac. p. 28 nṛtyagotavāditreṣvabāhyāḥ (no strangers to dancing, singing and playing on instruments)

要素の数は無制限である。以下のような長いdvandvaもある：
Nala. 1, 28 vayaṃ hi devagandharvamānuṣoragarākṣasāndṛṣṭavantaḥ

けれども、dvandvaが実際の結合を表す場合や個物ではなくカテゴリが連結される場合には、ふつうn.sg.である。よって例えばputrapautram（子と孫）、gavāśvam（牛馬）、ārāśastri（錐と短剣）など。

【補足1】
この主題に関する完全な説明はPāṇiniとその注釈者たちによって与えられている。彼らは、dvandvaがn.sg.である場合、n.sg.であってはならない場合、任意でどちらかを用いうる場合、を区別する。
n.sg.は、pāṇipādamやrāyikapādātamのように、体・楽器・軍隊のいずれかの部分を意味するdvandvaで必要とされる。{P. 2, 4, 2.} gaṅgāśoṇamやkurukurukṣetramのように、異なる文法的性をもつ河川や国の名前を連結して表す際も同様。{P. 2, 4, 7.}
一方で、生物を表す名詞から成るdvandvaは、以下の場合を除いてsg.で置くことが許されない：
a.) daṃśamaśakamのようなとても小さい動物 {P. 2, 4, 8.}
b.) mārjāramūṣakam（猫と鼠）のように、その性質上、永遠の闘争にあるもの {P. 2, 4, 9.}
c.) takṣāyaskāram（大工と鍛冶師）のような、卑しいものと見なされないシュードラの集団（*1） {P. 2, 4, 10.}
d.) gavāśvam、putrapautram、śvacaṇḍālam、strīkumāramなどのその他 {P. 2, 4, 11.}
—性質や物を対比するdvandvaは、sukhaduḥkham / -khe、śītoṣṇam / -ṣṇeのように、任意でsg.かdu.かに置かれる。{P. 2, 4, 13.} などなど。

【補足2】
〔上位カテゴリである〕動物の属を〔その下位カテゴリの〕種と複合することは許されない。Pat, I, p. 252をみよ。

【補足3】
dvandvaのmātāpitarauの代わりに、単なるpitarauを用いることができる（よって例えばRagh. 1, 1 jagataḥ pitarau vande pārvatīparameśvarauとなる）。{P. 1, 2, 70-71} 同様に、śvaśurauはśvaśrūśvaśurauと同義である。例としてはKathas. 58, 89をみよ。—「兄と妹」や「息子と娘」を表すdvandvaは用いられず、この場合にはbhrātarauやputrauが必要となる。{P. 2, 1, 68} —ラテン語のsoceri（＝socer et soclus）、fratres（＝frater et soror）と比較せよ。

【補足4】
順序に関しては、dvandvaを用いる際に従うべき固定された規則を与えることができない。けれども、iやuで終わるもの、語頭が母音で語尾がaのもの、〔複合される要素の中で〕最も短いもの、のいずれかがふつう先頭に置かれる。 {P. 2, 2, 32-34}

（*1）
P. 2, 4, 10 śūdrāṇām aniravāsitānām. Kâç. yair bhukte saṃskāreṇāpi na śudhyati ta niravasitāḥ

207. Its significance.

Besides its most common duty of expressing coordination, the dvandva is also available, if „and” connects persons or things standing in mutual relation with one another. Another species is the distributive dvandva.

Examples of the dvandva of relationship. — R. 3, 27, 10 sa saṃprahārastumulo rāmatriśirasostadā / saṃvabhūvātivalinoḥ siṃhakuñjarayor iva (then a wild battle began between Rama and Triçiras, both of extraordinary strength, as if between a lion and an elephant), Mâlav. I, p. 21 atra bhavataḥ kila mama ca samudrapalvalayor ivāntaram (forsooth, there is as great a difference between you and me, as there is between a pond and the ocean), Harshac. 5 nisargavirodhinī ceyaṃ payaḥ pāvakayor ivaikatra dharmakrodhayor vṛttiḥ (and like water and fire, so righteousness and auger, when meeting at the same place, by their proper nature combat each other).

Examples of distributive dvandva. — Mâlav. V, p. 137 tau pṛthagvāradākūle śiṣṭām uttaradakṣiṇe / naktaṃ dinaṃ vibhajyobhau śītoṣṇākiraṇāviva (let them rule severally the banks of the Varadâ, one the northern bank, but the other the southern, as moon and sun share their sway over night and day); Mudr. I, p. 19 dvitayaṃ] phalaṃ kopaprītyor dviṣati ca vibhaktaṃ suhṛdi ca (I have bestowed the double fruit of my wrath and my affection on foe and friend); Kathâs. 25, 229: Açokadatta by his utmost bravery has conquered a golden lotus and presents it to the king his master, who puts the precious flower in a silver vase; on that account the poet makes this comparison ubhau kalaśapadmau ca śuśubhāte sitāruṇau / yaśaḥ pratāpāviva tau bhūpālāśokadattayoḥ (and both the vase and tho flower shone, one white, the other red, as if they were that splendour and that glory combined, which adorned one tho king, the other Açokadatta).

207. dvandvaの意味

対等関係を表すという最も一般的なはたらきの他に、「…と」（and）が対立関係にある人・物をつなぐ場合にもdvandvaは用いられ得る。これ以外には配分的dvandvaがある。

・関係のdvandva
R. 3, 27, 10 sa saṃprahārastumulo rāmatriśirasostadā / saṃvabhūvātivalinoḥ siṃhakuñjarayor iva (then a wild battle began between Rama and Triçiras, both of extraordinary strength, as if between a lion and an elephant)
Mâlav. I, p. 21 atra bhavataḥ kila mama ca samudrapalvalayor ivāntaram (forsooth, there is as great a difference between you and me, as there is between a pond and the ocean)
Harshac. 5 nisargavirodhinī ceyaṃ payaḥ pāvakayor ivaikatra dharmakrodhayor vṛttiḥ (and like water and fire, so righteousness and auger, when meeting at the same place, by their proper nature combat each other)

・配分的dvandva
Mâlav. V, p. 137 tau pṛthagvāradākūle śiṣṭām uttaradakṣiṇe / naktaṃ dinaṃ vibhajyobhau śītoṣṇākiraṇāviva (let them rule severally the banks of the Varadâ, one the northern bank, but the other the southern, as moon and sun share their sway over night and day)
Mudr. I, p. 19 dvitayaṃ] phalaṃ kopaprītyor dviṣati ca vibhaktaṃ suhṛdi ca (I have bestowed the double fruit of my wrath and my affection on foe and friend)
Kathâs. 25, 229: Açokadatta by his utmost bravery has conquered a golden lotus and presents it to the king his master, who puts the precious flower in a silver vase; on that account the poet makes this comparison ubhau kalaśapadmau ca śuśubhāte sitāruṇau / yaśaḥ pratāpāviva tau bhūpālāśokadattayoḥ (and both the vase and tho flower shone, one white, the other red, as if they were that splendour and that glory combined, which adorned one tho king, the other Açokadatta)

208. Dvandvas of adjectives.

Dvandvas of adjectives are relatively seldom, if compared with the frequency of those made up of substantives.

So Kathâs. 25, 6 saṃprāpa so ’thi vindhyamahāṭavīṃ viveśa ca… tāṃ gahanāyatām (thick and long), Kumâras. 1, 35 vṛttānupūrve ca nātidīrghe jaṅghe (legs round, well-proportioned and not too long), Kathâs. 25, 229 sitāruṇau see 207, Mâlav. V, p. 137 uttaradakṣiṇe see 207. — Panc. I, 204 vyākīrṇakesarakarālamukhā mṛgendrāḥ (lions with dishevelled mane and frightful mouth) may be an instance of a dvandva of two adjectives, either of which is a compound itself.

208. 形容詞のdvandva

形容詞のdvandvaは、実名詞から成るdvandvaの頻度と比べると比較的まれである。

Kathâs. 25, 6 saṃprāpa so ’thi vindhyamahāṭavīṃ viveśa ca… tāṃ gahanāyatām (thick and long)
Kumâras. 1, 35 vṛttānupūrve ca nātidīrghe jaṅghe (legs round, well-proportioned and not too long)
Kathâs. 25, 229 sitāruṇau see 207
Mâlav. V, p. 137 uttaradakṣiṇe see 207
Panc. I, 204 vyākīrṇakesarakarālamukhā mṛgendrāḥ (lions with dishevelled mane and frightful mouth)

最後のものは2つの複合形容詞（vyākīrṇa-kesara＋karāla-mukha）から成るdvandvaの例である。

209.

Two kinds of compounds are reckoned by vernacular grammar among the tatpurushas, which by their meaning should rather be considered dvandvas:

Combination of participles: 1. kṛtākṛta and the like; 2. dṛṣṭanaṣṭa and the like.

1. Such as aśitānaśita (eatable and not eatable), kṛtākṛta (done and undone; wrought and unwrought), kṛtāpakṛta, gatapratyāgata. {P. 2, 1, 60 with vârtt.} Cp. Kathâs. 27, 1 namitonnamitena (by turns lowering and raising his head).

2. Those made up of two participles in -ta, the compounding of which declares the two actions being done immediately one after another. {P. 2, 1, 49.} The former in time is also the former member. Of the kind are dṛṣṭanaṣṭa (as soon lost as it is seen), snātānulipta (after bathing and anointing one’s self). Çâk. IV atha praviśati suptotthitaḥ śiṣyaḥ (enters a disciple, just arising from his couch), Ragh. 4, 37 kalamā iva te… utkhātapratiropitāḥ (like stalks of rice dug out and forth with replanted), Panc. I, 5 jātapretaḥ (died soon after birth), ibid. V, 7 vittavihīnāḥ] satataṃ jātavinaṣṭāḥ payasām iva budbudāḥ payasi. Kathâs. 29, 141 an illness (rogaḥ) is said to have been jñātāpanītaḥ »as soon driven out as its nature had been recognised.

209. tatpuruṣaに分類されるdvandva的複合語

土着文法家はtatpuruṣa中に、その意味からすればむしろdvandvaと見なされるべき、2種の複合語を認める。

分詞の複合語：1. kṛtākṛta等；2. dṛṣṭanaṣṭa等

1. aśitānaśita（食べられるものと食べられないもの）、kṛtākṛta（なされたものとなされていないもの）、kṛtāpakṛta、gatapratyāgata。{P. 2, 1, 60 with vârtt.} 以下と比較せよ：
Kathâs. 27, 1 namitonnamitena (by turns lowering and raising his head)

2. 2つの分詞-taから成り、2つの行為が次々に行われることを表すもの。{P. 2, 1, 49.} 時間的に前のものが前分となる。dṛṣṭanaṣṭa（見られるやいなや失われる）、snātānulipta（沐浴し自身に油を塗って）など。
Çâk. IV atha praviśati suptotthitaḥ śiṣyaḥ (enters a disciple, just arising from his couch)
Ragh. 4, 37 kalamā iva te… utkhātapratiropitāḥ (like stalks of rice dug out and forth with replanted)
Panc. I, 5 jātapretaḥ (died soon after birth) ibid. V, 7 vittavihīnāḥ] satataṃ jātavinaṣṭāḥ payasām iva budbudāḥ payasi
Kathâs. 29, 141 an illness (rogaḥ) is said to have been jñātāpanītaḥ »as soon driven out as its nature had been recognised.

210. Tatpurusha.

The tatpurusha serves to express in a condensed shape a noun — substantive or adjective — together with some other noun qualifying it, as tatpuruṣaḥ = tasya puruṣaḥ (his man), ahihataḥ = ahinā hataḥ (bitten by a serpent), navayauvanam = navaṃ yauvanam (the first youth). The noun qualifying is the former member of the tatpurusha; the noun qualified, which is at the same time the main element, its latter member (*1).

Rem. The efficient elements of a tatpurusha are not of necessity self-existent words. The former part may be such a particle as su-, duḥ-, a-, see 218. The latter may be a kṛt, not otherwise used but in compounds, as -ja, -ga, -stha, -jña, -kṣa, -bhuj, -bhāj, -śaya, -kṛt, -kārin, -bhojin, -hārin etc. Many of these compounds have got a special meaning so as to make them indissoluble unities, as aṇḍaja »bird,” kumbhakāra »potter,” kiṃkara »servant.” Yet free compounding is also allowed. So Panc. I, 103 kauśeyaṃ kṛmijam may be analyzed into kau- kṛmerjātam, ibid. p. 28 prasādasaṃmukho naḥ svāmī vacanavaśagaś ca [= vacanavaśe gataḥ], Bhojapr. 2 loko ‘yaṃ māṃ sarvajñaṃ vakti, Panc. 41 samīpavirtini nagare [= samīpe vartamāne na-].

(*1)
Pat. I, p. 392 uttarapadārthapradhānastatpuruṣaḥ. In the same way the dvandva is styled ubhayapadārthapradhānaḥ, the bahuvrîhi anyapadārthapradhānaḥ, and the avyayîbhâva pūrvapadārthapradhānaḥ.

2. tatpuruṣa

tatpuruṣaは、名詞（実名詞・形容詞）とそれを修飾する別の名詞を簡約された形で表現するのに資する：tatpuruṣaḥ＝tasya puruṣaḥ（彼の男）、ahihataḥ＝ahinā hataḥ（蛇に噛まれた）、navayauvanam＝navaṃ yauvanam（最初の青年期）。修飾する名詞はtatpuruṣaの前分である；修飾される名詞は、同時に主要素であり、後分である（*1）。

【補足】
tatpuruṣaにおいて有効な要素は自立語である必要がない。前分はsu-、duḥ-、a-でありうる；218をみよ。後分は、本来複合語では用いられないkṛt接尾辞でありうる：-ja、-ga、-stha、-jña、-kṣa、-bhuj、-bhāj、-śaya、-kṛt、-kārin、-bhojin、-hārinなど。
これら複合語の多くは、分解されざる結合になるため、特別な意味をもつ：aṇḍaja（鳥）、kumbhakāra（ 陶工）、kiṃkara（従者）。
けれども、自由な複合も許される。よって、Panc. I, 103 kauśeyaṃ kṛmijamはkau- kṛmerjātamと分析できる。
ibid. p. 28 prasādasaṃmukho naḥ svāmī vacanavaśagaś ca [= vacanavaśe gataḥ]
Bhojapr. 2 loko ‘yaṃ māṃ sarvajñaṃ vakti
Panc. 41 samīpavirtini nagare [= samīpe vartamāne na-]

（*1）
Pat. I, p. 392 uttarapadārthapradhānas tatpuruṣaḥ
同様に、dvandvaはubhayapadārthapradhānaḥ、bahuvrīhiはanyapadārthapradhānaḥ、 avyayībhāvaはpūrvapadārthapradhānaḥと呼ばれる。

211.

The former member may be either = a noun-case (as in tatpuruṣaḥ, ahihataḥ), or = an adjective (as in navayauvanam). In the latter case, there exists grammatical concord between the two members; such tatpurushas bear the special appellation of karmadhâraya.

karmadhâraya

The faculty of combining adjectives with their substantives into karmadhârayas is theoretically almost unrestricted, but in practice not all possible combinations are used (*1). Most karmadhârayas are terms often recurring which either have got some special meaning, or are wont to be much employed though nothing impedes expressing them by the two elements severed. Of the former kind are such as paramātmā (the highest soul), yuvarājaḥ (heir apparent), of the latter such as kṛṣṇasarpaḥ (a black serpent), pakvānnam (cooked rice), and the great class of compounds, a full account of which is given by Pâṇini in the first adhyâya of his second ashtakam (see espec. the sûtras 49, 58, 61, 67 and 70), containing those, the former member of which is a pronoun as sarva, eka, apara, sva, and such adjectives as nava, purāṇa, mahā-, madhya, madhyam, sat (good) etc. To them we may add such words as nija, mukhya, pradhāna, and even such as begin by su-, duḥ- and the negation a-, as sujanaḥ (an honest man), durjanaḥ (a bad man), akovit (not skilled).

(*1)
Pâṇini's rule 2, 1, 57 viśeṣaṇaṃ viśeṣyeṇa bahulam plainly shows not all combinations of the kind to be allowed.

211. karmadhāraya

〔tatpuruṣaの〕前分は名詞-格（tatpuruṣaḥ、ahihataḥなど）あるいは形容詞（navayauvanamなど）である。後者の場合、2要素間に文法的一致（concord）が起こる；そのようなtatpuruṣaはkarmadhārayaという特別な名称をもつ。

karmadhāraya

形容詞と実名詞を組み合わせてkarmadhārayaにする能力は、理論的にはほとんど制限されていないが、実際には、すべての可能な組み合わせが使用されるわけではない（*1）。ほとんどのkarmadhārayaは、特別な意味をもつか、あるいは、2つの切断された要素で説明可能だがよく用いられるもの、であり、よく繰り返される。
前者の類はparamātmā（いと高き魂）、yuvarājaḥ（皇太子）、後者はkṛṣṇasarpaḥ（黒い蛇）、pakvānnam（調理された米）および複合語の大部分であり、その詳細はPāṇiniの2巻1節以下に述べられている（特に2, 1, 49、58、61、67、70をみよ）；これらスートラは、sarva、eka、apara、svaのような代名詞やnava、purāṇa、mahā-、madhya、madhyam、sat（良い）のような形容詞が前分のものを含む。 これらには以下の語が加えられよう：nija、mukhya、pradhāna、su-やduḥ-で始まるもの、否定辞a-：例えばsujanaḥ（誠実な男）、durjanaḥ（悪い男）、akovit（未熟な）など。

（*1）
P. 2, 1, 57 viśeṣaṇaṃ viśeṣyeṇa bahulamは、許される類のすべての組み合わせを明確に示しているわけではない。

212.

Yet there are instances enough of a freer employment. Panc. 327 kim evaṃ palāyyate ‘līkabhayena (why do you run away thus by a false fear?), Pat. I, p. 2 śaṣkaidhaḥ = śaṣkamedhaḥ, Panc. 30 anucitasthāna upaviṣṭaḥ, Mâlav. I, p. 3 vartamānakaveḥ kālidāsasya (of the living poet K.), Harshac. 6 śailūṣa iva vṛthā vahasi kṛtrimopaśamam (like an actor you are displaying in vain a fictitious tranquillity of mind), Bhoj. 28 prāktanakarmato dāridhyram anubhavati (in consequence of his deeds in a former existence he is now poor), Panc. 87 tvadīyavanam (your orders), Mâlav. I, p. 28 yasyāgamaḥ kevalajīvikāyai (to whom his learning serves only for a livelihood), Kathâs. 39, 131 dattvā varāśvam (— gave a best horse).

Upon the whole, such freer karmadhârayas are used in a greater extent in poetry, also when being themselves but a member of some large compound, as f. i. Panc. 37 anekasādhujanadattasūkṣmavastravikrayavaśāt (by selling fine clothes given to him by many pious people), in analyzing which we get anekasādhujanair [karm.] dattānāṃ sūkṣmavastrāṇāṃ [other karm.] vikrayasya vaśāt.

Rem. In the case of such words as pāpanāpitaḥ, there seems to exist a slight difference between the karmadh. and the analytical construction of the same purport; pāpanāpitaḥ is „a bad barber” who knows his art badly, but pāpo nāpitaḥ »a barber of a bad temper.” See P. 2, 1, 54.

212. 承前：その他の用法

けれども、自由な用法の例も多い。
Panc. 327 kim evaṃ palāyyate ‘līkabhayena (why do you run away thus by a false fear?)
Pat. I, p. 2 śaṣkaidhaḥ = śaṣkamedhaḥ Panc. 30 anucitasthāna upaviṣṭaḥ
Mâlav. I, p. 3 vartamānakaveḥ kālidāsasya (of the living poet K.)
Harshac. 6 śailūṣa iva vṛthā vahasi kṛtrimopaśamam (like an actor you are displaying in vain a fictitious tranquillity of mind)
Bhoj. 28 prāktanakarmato dāridhyram anubhavati (in consequence of his deeds in a former existence he is now poor)
Panc. 87 tvadīyavanam (your orders)
Mâlav. I, p. 28 yasyāgamaḥ kevalajīvikāyai (to whom his learning serves only for a livelihood)
Kathâs. 39, 131 dattvā varāśvam (— gave a best horse)

全体として、そのような自由なkarmadhārayaは、詩文においては、それ自体が何らかのより大きな複合語の要素となる場合にも、より多くの範囲で用いられている。
Panc. 37 anekasādhujanadattasūkṣmavastravikrayavaśāt (by selling fine clothes given to him by many pious people)

これを分析するとanekasādhujanair dattānāṃ sūkṣmavastrāṇāṃ vikrayasya vaśātが得られる。

【補足】
pāpanāpitaḥのような語の場合、karmadhārayaとそれと同じ趣旨の分析的構文との間にはわずかばかり差異がある；pāpanāpitaḥはその技術が下手だと知られているところの「下手な床屋」を意味するが、pāpo nāpitaḥは「気質の邪悪な床屋」を表す。P. 2, 1, 54をみよ（*訳注）。

（*訳注）
P. 2, 1, 53 kutsitāni kutsanaiḥ
P. 2, 1, 54 pāpāṇake kutsitaiḥ
軽蔑対象（kutsita）を指示する語は非難の言葉（kutsana）と複合してtatpuruṣaを形成する（2, 1, 53）。pāpaおよびaṇakaは軽蔑対象（kutsita）と複合してtatpuruṣaを形成する（2, 1, 54）。
これらのスートラで問題になっているのは（技術などの）「非-熟練」などの限定された範囲ではなく軽蔑（√kuts）なので、pāpanāpitaḥの訳語としてはむしろ「ヤブ医者」の「ヤブ」くらいの語勢が欲しいところである。その点、Charu Deva Shastriの”a wretched barber”（p. 125）はしっくりくる。

しかしながら、pāpanāpitaḥとpāpo nāpitaḥを意味の異なるものとして対比することについては同意できない。
P. 2, 1, 49 pūrvakālaikasarvajaratpurāṇanavakevalāḥ samānādhikaraṇena

特定の場合におけるtatpuruṣaの形成に関して、ここでは同格（samānādhikaraṇa）にある語を指定している。上記のスートラはどちらもP. 2, 1, 49の支配下にあるから、P. 2, 1, 54の規則により形成される複合語pāpanāpitaはpāpo nāpitaḥのように分析されることになる。
このことはKâç.からも読み取ることができる。P. 2, 1, 54に関連するスートラにP. 6, 2, 68 pāpaṃ ca śilpini（複合語において「職人」を表す語の前に位置するpāpaは、任意で、第1音節にudāttaアクセントを持つ）がある。これに対してKâç.は以下のように述べている：
Kâç. on P. 6, 2, 68 pāpāṇake kutsitaiḥ (2-1-54) iti pāpaśabdasya pratipadoktaḥ samānādhikaraṇasamāsaḥ iti ṣaṣṭhīsamāse na bhavati, pāpasya nāpitaḥ pāpanāpitaḥ iti
「〔ここで言及するのは前のスートラにおいて〕“pāpāṇake kutsitaiḥ”（2-1-54）とpāpaという語が明示的に述べられている同格複合語（＝pāpanāpitaḥ）である。第6格複合語（pāpanāpitaḥ←pāpasya nāpitaḥ）では起こらない」

以上のことから、ここで”a barber of a bad temper”に対応させるべきなのはpāpasya nāpitaḥ（悪人に属する床屋＝悪い床屋）とすべきであろう。

213.

We will insist on some species:

a.) such as are made up of a title + the noun of its bearer, as āryacāṇakyaḥ (Sir Câṇakya), amātyarākṣasaḥ (Minister Râxasa). So Utt. II, p. 30 kumāralakṣmaṇaḥ (Prince L.), Mâlav. I, p. 24 paṇḍitakauśikī khalu bhavatī (you are the learned Kauçikî, are not you ?), Mṛcch. III, p. 115 ahaṃ hi gaṇikāmadanikārtham akāryam anutiṣṭhāmi, but some lines after (p. 116) we read in inverse order madanikāgaṇikārthe, cp. Panc. 59 anena viṣṇujāmātṛprabhāveṇa sarve atravo nihatāḥ, here the proper noun viṣṇu is followed, not preceded by its epithet. In some cases the latter idiom seems to be the regular one, as vindhyācalaḥ (Mount Vindhya).

b.) those ending in -jana, the former part being a subst., as strījanaḥ, duhitṛjanaḥ, gurujanaḥ. Here jana has sometimes the power of a collective, sometimes it denotes the individual (19 R), strījanaḥ may be = »womankind,” »women” or even one »woman,” and so on;

c.) the type ardhapippalī (half a pepper). It is not allowed to say pippalyardham, but pippalyā ardham; when compounded it must be ardhapippalī (*1). {P. 2, 2, 1 and 2.} So f. i. Ragh. 7, 42 ardhamārga (halfway), Panc. 203 ardhapādaspṛṣṭabhūmiḥ (touching the earth with the half of his foot). The same applies to pūrva, apara, adhara, uttara, etc. when denoting: the fore-part, the part behind, the lower and upper part. Therefore it is said pūrvakāyaḥ (the fore-part of the body), pūrvāhṇaḥ (fore-noon), apararātraḥ (the latter part of the night), uttamāṅgam (head), prathamarātrī (the fore-night), and the like. So madhya in madhyāhnaḥ = madhyam ahnaḥ. We have here the same adjectival conception as in Latin summus mons, media urbs, Greek μἑση ἡ πόλις etc.

Rem. 1. ardha, like our »half,” is also compounded with a participle or some other adjective, as ardhoditaḥ sūryaḥ (the sun, half-risen), Panc. 9 ardhasphāṭito dārumayaḥ stambhaḥ.

Rem. 2. As to compounds, commencing by dvitīya, tṛtīya, caturtha or turya, when = »half, the third —, fourth part,” one may say as well dvitīyabhikṣā as bhikṣādvitīyam (half an alms) and the like. {P. 2, 2, 3.} The same may be stated of agra (top, edge, extremity), as it is said as well agranakha (the edge of a nail) as nakhāgra (see Petr. Dict. s. v. and the passages adduced there s. v. agra).

(*1)
But it is allowed to say pippalyardhaḥ (a portion of a pepper). Pat. I, 407 states ardha to be a neuter, when meaning »half,” but a masculine, when = »portion, part:” samapravibhāge napuṃsakaliṅgo ‘vayavavācī puṃlliṅgaḥ

213. 肩書き・-jana・ardha

いくつかの種類について強調しておく：

a.) āryacāṇakyaḥ（Cāṇakya殿）、amātyarākṣasaḥ（Rāxasa大臣）のような、肩書き＋その所有者から成るもの。
Utt. II, p. 30 kumāralakṣmaṇaḥ (Prince L.)
Mâlav. I, p. 24 paṇḍitakauśikī khalu bhavatī (you are the learned Kauçikî, are not you ?)
Mṛcch. III, p. 115 ahaṃ hi gaṇikāmadanikārtham akāryam anutiṣṭhāmi
数行語（p. 116）では逆順：madanikāgaṇikārthe

以下と比較せよ：
Panc. 59 anena viṣṇujāmātṛprabhāveṇa sarve atravo nihatāḥ

ここで固有名詞viṣṇuは、その肩書きに先行するのではなく後に続いている。vindhyācalaḥ（Vindhya山）のように、いくつかの場合では、後者のイディオムが通常のようである。

b.) strījanaḥ、duhitṛjanaḥ、gurujanaḥのような、前分が実名詞で-janaで終わるもの。ここでのjanaは集合的な効力をもったり、個物を表したりすることがある（19-補足）。strījanaḥは集合的な「婦人」（womankind, women）を指したり、個物としての「女性」（woman）を指したりする、など。

c.) ardhapippalī（半分の胡椒）のタイプ。pippalyardhamと言うのもpippalyā ardhamと言うのも許されない；複合する際にはardhapippalīでなくてはならない（*1）。{P. 2, 2, 1-2}
Ragh. 7, 42 ardhamārga (halfway)
Panc. 203 ardhapādaspṛṣṭabhūmiḥ (touching the earth with the half of his foot)

同じことが、ものの前側、後ろ側、下側、上側を表す場合のpūrva、apara、adhara、uttaraなどに適用される。したがって、pūrvakāyaḥ (the fore-part of the body)、pūrvāhṇaḥ (fore-noon)、apararātraḥ (the latter part of the night)、uttamāṅgam (head), prathamarātrī (the fore-night)などとなる。madhyāhnaḥのmadhyaはmadhyam ahnaḥと等しい。これにはラテン語のsummus monsやmedia urbs、古代ギリシャ語のμἑση ἡ πόλιςなどのような同じ形容詞的構文がある。

【補足1】
ardhaは、英語の”half”と同様に、分詞や他の形容詞とも複合する：ardhoditaḥ sūryaḥ（半分昇った太陽）。 Panc. 9 ardhasphāṭito dārumayaḥ stambhaḥ

【補足2】
dvitīya、tṛtīya、caturthaやturyaで始まり「n分の1」を表す場合の複合語については、dvitīyabhikṣāともbhikṣādvitīyam（半分の喜捨）ともできる。{P. 2, 2, 3.} agraについても同様で、agranakha (the edge of a nail)ともnakhāgraともできる（Petr. Dict.のagraの項をみよ）。

（*1）
けれども、pippalyardhaḥ（胡椒の一部分）と言うことは許される。Pat. I, 407は、ardhaは「半分」を表す場合にn.で、「一部分」を表す場合にm.であることを述べている：
samapravibhāge napuṃsakaliṅgo ‘vayavavācī puṃlliṅgaḥ

214. Compounds in -bhūta and the like, the former member of which is a predicate.

A proper species of tatpurshas has is made up by those, whose latter members is a verbal noun, the noun predicate of which is signified by the former member. The commonest instances of the kind are adjectives in -bhūta (being, making up, behaving as). Pat. I, 39 pramāṇabhūta ācāryaḥ (the teacher, who is [the pupil’s] authority), Daç. 176 tasyāham asmy udāharaṇabhūtā (I am an example thereof), M. 1, 5 āsīd idaṃ tamobhūtam. Moreover there is a class of much used compound verbs, whose former part is a noun, whereas the latter is the verb kṛ or bhū; they carry the conception of something transformed from one state into another. They will be dealt with when treating of the Syntax of the Verb; see 308.

Among other similar tatpurushas we notice a.) those ending in ukta, samāmnāta, mata, samākhyāta, kṛta etc. {P. 2, 1, 59 with Kâç’s comm.} Daç. 61 sa ca… ambālikāyām abalāratnasamākhyātāyām atimātrābhilāṣaḥ (he, being passionately in love with Ambâlikâ surnamed: the jewel of womankind); b.) those in -bruva (having but the name of) and -mānin and -manya (thinking one’s self —), as brāhmaṇabruvaḥ »one who claims himself a brahman” [on account of his birth, but who does not behave as such], paṇḍitamānī (wise in one’s own opinion), Atharvav. 15, 13, 6 avrātyo ‘vrātyabuvaḥ, R. 3, 21, 17 śūramānī na śūrastvam, Daç. 99 dhanyaṃmanyaḥ (*1). {P. 3, 2, 83.}

Rem. Somewhat different is the nature of those, the former part of which is not the predicate, but the predicate’s attribute, as uttānaśaya and avamūrdhaśaya given as examples by the comm. of vârtt. 3 on P. 3, 2, 15 and pūrvasara (going at the head) see P. 3, 2, 19.

(*1)
With them may in some degree be compared such tatpurushas as Kathâs. 9, 48 āmiṣaśaṅkayā (holding [her] for a piece of raw flesh), Hit. 93 vyāghrabuddhyā (taking [him] for a tiger). Here the former member is the predicative object of the verbal noun, which is the latter member.

214. 前分が述語の複合語-bhūta

厳密なtatpuruṣa類は、後者の要素が動詞的名詞であり、その名詞述語が前者の要素によって示されているもので構成される。もっともありふれた類の用例は、-bhūta（…である、…を構成する、…として振る舞う）の形容詞である。
Pat. I, 39 pramāṇabhūta ācāryaḥ (the teacher, who is [the pupil’s] authority)
Daç. 176 tasyāham asmy udāharaṇabhūtā (I am an example thereof)
M. 1, 5 āsīd idaṃ tamobhūtam

さらに、前分が名詞で後分が動詞√kṛか√bhūで構成される、よく用いられる複合動詞の類もある；これらは何ものかがある状態から別の状態に変わる概念を伝達する。これらは動詞のシンタックスを扱う際に触れる；308をみよ。

他の類似のtatpuruṣa中、以下のものに言及しておく：
a.) ukta、samāmnāta、mata、samākhyāta、kṛta等で終わるもの{P. 2, 1, 59 with Kâç’s comm.}
Daç. 61 sa ca… ambālikāyām abalāratnasamākhyātāyām atimātrābhilāṣaḥ (he, being passionately in love with Ambâlikâ surnamed: the jewel of womankind)

b.) -bruva（…を騙る）、-mānin・-manya（自身を…と考える）で終わるもの：brāhmaṇabruvaḥ（〔本当はそうではないのに、生まれついて〕自身をバラモンと称する者）、paṇḍitamānī（賢明と自認する者）など{P. 3, 2, 83.}
Atharvav. 15, 13, 6 avrātyo ‘vrātyabuvaḥ
R. 3, 21, 17 śūramānī na śūrastvam
Daç. 99 dhanyaṃmanyaḥ （*1）

【補足】
これら〔複合語内の要素同士〕の性質は多少異なり、前分は述語ではなく、述語の属性である：vârtt. 3 on P. 3, 2, 15の例であるuttānaśayaやavamūrdhaśayapūrvasaraなど。P. 3, 2, 19をみよ。

（*1）
以下のようなtatpuruṣaとある程度比較できる： Kathâs. 9, 48 āmiṣaśaṅkayā (holding [her] for a piece of raw flesh)
Hit. 93 vyāghrabuddhyā (taking [him] for a tiger)

ここでは前分が、後分である動詞的名詞の述語的目的語である。

215. Tatpurusha, consisting of noun-case + noun.

Among such tatpurushas as are made up of a noun-case + the noun qualified by it, by far the most common are those, whose former part is to be periphrased by a genitive, as rājapuruṣaḥ = rājñaḥ puruṣaḥ, śatruvadhaḥ = śatror vadhaḥ or śatrvoḥ or śatrūṇām. As this type is met with on every page, it is useless to quote instances from literature. Another frequent type is that, represented by ahihataḥ = ahinā hataḥ. Panc. 118 gurujanānujñātaḥ = gurujanenānujñātaḥ, ibid. V, 93 rākṣasendragṛhītaḥ (seized by the prince of giants), Bhojapr. 7 so ’pi tadākarṇya vajrāhata iva bhūtāviṣṭa iva grahagrasta iva.

For the rest, any noun-case may become the former part of a tatpurusha, as māsakalyāṇaḥ = māsaṃ kalyāṇaḥ (happy for a month), mātṛsadṛśaḥ = mātrā or mātuḥ sadṛśaḥ (resembling his mother), yūpadāru = yūpāya dāru (wood for a sacrificial stake), vṛkabhayam = vṛkāt or vṛkebhyo bhayam (fear of a wolf or of wolves), sthālīpākaḥ = sthālyāṃ pākaḥ (cooking in a pot).

215. 名詞-格＋名詞から成るtatpuruṣa

〈名詞-格＋それを修飾する名詞〉から成るtatpuruṣaの中で最も一般的なのは、以下のようにgen.で言い換えられるものである：
rājapuruṣaḥ（＝rājñaḥ puruṣaḥ）、śatruvadhaḥ（śatror (śatrvoḥ; śatrūṇām) vadhaḥ）。どのページでも見られるこのタイプに関しては、文献から用例を引くまでもない。他の頻出するタイプは、ahihataḥ＝ahinā hataḥに代表されるものである。
Panc. 118 gurujanānujñātaḥ = gurujanenānujñātaḥ
ibid. V, 93 rākṣasendragṛhītaḥ (seized by the prince of giants)
Bhojapr. 7 so ’pi tadākarṇya vajrāhata iva bhūtāviṣṭa iva grahagrasta iva

その他、任意の名詞-格がtatpuruṣaの前分になりうる：māsakalyāṇaḥ＝māsaṃ kalyāṇaḥ (happy for a month)、mātṛsadṛśaḥ＝mātrā (mātuḥ) sadṛśaḥ (resembling his mother)、yūpadāru＝yūpāya dāru (wood for a sacrificial stake)、vṛkabhayam＝vṛkāt (vṛkebhyo) bhayam (fear of a wolf or of wolves)、sthālīpākaḥ＝sthālyāṃ pākaḥ (cooking in a pot)など。

216.

Yet, there are some restrictions. For this reason, Pâṇini when treating of compounds made up of a noun-case + noun, gives a detailed account of them. The summary of which runs in this way:

The former member is a genitive.

I. As a rule, any genitive may be compounded (shashthîsamâsa). {P. 2, 2, 8.} Some cases are excepted. Among others it is not allowed to use compounds, made up of a genitive + a participle or a gerund or a kṛtya or an infinitive, nor those consisting of a genitive + comparative or superlative or ordinal noun of number, nor such as where a genitive is compounded with some noun in -tṛ or -aka. {P. 2, 2, 11; 15} Therefore, such phrases as manuṣyāṇāṃ śūratamaḥ (the most heroic of men), chātrāṇāṃ pañcamaḥ (the fifth of the disciples), apāṃ sraṣṭāḥ (the creator of the waters), brāhmaṇasya kurvan or kṛtvā or karaṇīyam or kartum (doing etc. for the benefit of a brahman) are unfit for compounding. Partitive genitives are likewise excluded, nor is the dative like genitive (129), it seems, as a rule, fit for being compounded. — As little, so we learn from Pâṇini’s commentators(*1) an objective genitive in such cases, as vicitrā sūtrasya kṛtiḥ pāṇininā, here it is not allowed to say sūtrakṛtiḥ pāṇininā, since both the subject and the object of the action conveyed by the noun kṛtiḥ are expressed, for nothing impedes using the compound sūtrakṛtiḥ = sūtrasya kṛtiḥ, if the agent is not expressed (*2).

an accusative;

II. compounding the accusative is allowed: {P. 2, 1, 29.} (*3)

a.) when being one of time as māsakalyāṇaḥ. So R. 2, 71, 18 saptarātroṣitaḥ pathi, Mâlat. I, p. 14 bahudivasopacīyamānam iva manmathavyathāvikāram upalabdhavān asmi;

b.) with some participles in -ta, with active or intransitive meaning, {P. 2, 1, 24.} as grāmagataḥ (gone to the village), narakapatitaḥ (fallen to hell), kaṣṭaśritaḥ (come to hardship), sukhaprāptaḥ (*4) sim. {P. 2, 1, 26.} In practice, there are more. So f. i. the restriction of khaṭvārūḍha (P. 2, 1, 26) to a reproachful term does not imply the prohibition of compounding ārūḍha otherwise. See but Panc. 51 kareṇukārūḍhā, ibid. 30 gṛhāyātaḥ

an instrumental;

III. compounding the instrumental is allowed:

a.) if denoting the agent or instrument + some verbal noun, as ahihataḥ. {P. 2, l, 32.} The participles in -tavant are excepted, compounding ahinā + hatavān therefore not allowed. {P. 2, 1, 33.} — Some proverbial locutions are explicitly named by Pâṇini, as kākapeyā nadī, śvalehyaḥ kūpaḥ, but from this it should not be inferred that it is wholly forbidden to make up any other compound of instrumental + kṛtya. Panc. 327 tvadbhakṣyo ‘yaṃ mānuṣaḥ = tvayā bha-. Yet such compounds are not frequent.

b.) if the latter member is a word expressive of likeness, equality, superiority, want (*5), see 61 and 78. {P. 2, 1, 31.} Of the kind are such compounds as pitṛsamaḥ (equal to his father), mātṛsadṛśaḥ (resembling his mother), māsapūrvaḥ and māsāvaraḥ (earlier —, later by a month), M. 8, 217 alponaṃ karma = alpenonaṃ karma (v. a. work, almost finished), Panc. 23 svāmisadṛśā eva bhavanti bhṛtyāḥ;

c.) the instrum. + the words kalaha (quarrel), nipuṇa (clever); miśra (mixed), ślakṣṇa (lax). {P. 2, 1, 31.} Kâç. gives these examples: asikalahaḥ, vāṅnipuṇaḥ, guḍamiśra, ācāraślakṣṇaḥ. When extending the rule to all words of the same purport, as we may do (see above p. 92 N.[= 124, note 1]), the frequent compounds in -milita, -anvita, -yuta, -yukta, -upeta, etc. are included, also many of those, the former part of which is an instrumentalis partis (78).

d.) in the case of compound adjectives, the former part of which is an instrumental of causality, illustrative of the adjective it is joined to (*6). {P. 2, 1, 30.} So R. 3, 16, 13 niḥśvāsāndha ādarśaḥ (a mirror tarnished by exhalation), ibid. 3, 55, 20 vīryasamaḥ (equal by strength), ibid. 2, 118, 4 guṇaślāghyaḥ (praiseworthy by his qualities), Panc. I, 39 jātyādimahotsāhā narendrāḥ = jātyādinā mahotsāhā narendrāḥ, Kumâras. 3, 12 tapovīryamahatsu (men, great by their heroic penance). — Of the same kind are the compounds, made up of instrum. + arthin (76 R. 1), as Panc. 10 kiṃ bhavān āhārārthī kevalam eva (are you desirous of nothing but food?), Kathâs. 24, 176 etc.

e.) in the case of food dressed with some ingredient or by mixing two materials, as dadhyodanaḥ, guḍadhānāḥ. {P. 2, 1, 34 and 35.} Likewise Daç. 139 viṣānnam (empoisoned food).

a dative.

IV. the dative may be compounded:

a.) the dative of the aim in such cases as yūpadāru = yūpāya dāru, kuṇḍalahiraṇyam = kuṇḍalāya hiraṇyam; {P. 2, 1, 36.}

b.) the dative of the remote object with the words bali (offering to), hita (good for), sukha (pleasant for), rakṣita (kept, guarded for) and the like, {P. 2, 1, 36.} as viṣṇubaliḥ, gohitam, Panc. I, 47 sā sevā yā prabhuhitā, etc. — With them is named artha »purpose, aim, scope.” On the compounds in -artha we have treated 194.

an ablative.

V. the ablative:

a.) with words expressive of fear, as vṛkabhayam (fear of wolves), R. 3, 27, 20 vyādhatrastā mṛgāḥ (deer, afraid of the hunter); {P. 2, 1, 37.}

b.) with some participles, which signify a withdrawal (*7). Daç. 89 nṛtyotthitā sā prātiṣṭhata (she rose from the dance and went away), M. 2, 89 sāvitrīpatitaḥ (one who has forfeited the sâvitrî), R. 3, 25, 24 cicchidur bibhiduś caiva rāmabāṇā guṇacyutāḥ.

a locative.

VI. the locative:

a.) with such nouns as are construed with a locative of reference, as those of attachment, skill and the like (148). {P. 2, 1, 40.} R. 3, 19, 22 mānuṣau śastrasaṃpannau (two men, accomplished at arms), Panc. I, 18 bhāṇḍakrayavicakṣaṇāḥ (skilled in the commerce of merchandises);

b.) in some cases, when denoting a time or a place. Of the kind Pâṇini names compounds in -siddha (prepared, dressed), -śuṣka (dried), -pakva (cooked), -bandha, {P. 2, 1, 41.} also parts of the day or night + kṛta, {P. 2, 1, 45.} as sāṃkāśyasiddha (prepared in Sânkâçya), ātapaśuṣka (dried in the sun), pūrvāhṇakṛta (done in the forenoon);

c.) in some standing phrases and proverbial locutions, see P. 2, 1, 42-44; 46-48.

(*1)
See Kâç. on P. 2, 2, 14; Pat. I, p. 415, vârtt. 6. Pat. himself rejects the interpretation given there.

(*2)
The shashthîsamâsa is treated by Pâṇini in the second adhyâya of his 2nd book (2, 2, 8-17), some statements are also scattered in the third book, see f. i. 3, 3, 116. Additions ad corrections on them are of course made in the commentaries. But now and then the cavillations of the commentators have rather obscured the good understanding of some rules. So the Kâçikâ is wrong loosening sûtra 2, 2, 14 from its adhikâra, ktena and interpreting this rule — karmaṇi ca — as if it taught something concerning the objective genitive. Now, as the sûtra could in no way be explained so as to contain a prohibition of compounding any objective genitive whatever, as such compounds are very common indeed, the Kâçika was obliged to add a clause of its own ubhayaprāptau karmaṇīti ṣaṣṭhyā iti grahaṇam which statement certainly will be correct by itself, but not the smallest trace of which is to be found in Pâṇini. In fact, Pâṇini has here not thought of an objective genitive. When reading the sûtras 12, 13, 14 at a stretch and without prejudice, one sees plainly that karmaṇi of 14 qualifies ktena of 12. Sûtra 14 prohibits compounding a genitive + a participle in -ta with passive meaning. It is not allowed to say taddarśitam instead of tasya darśitam (shown to him), whereas Pâṇini allows it, when representing tena darśitam (shown by him), cp. 2, 1, 32.
The following sûtras 15 and 16 — tṛjakābhyāṃ kartari / kartari ca — afford a fair sample of absurd hairsplitting. In s. 15 Pâṇini had given a rule about the words in -tṛ and -aka when denoting the agent; with them a genitive cannot be compounded, save the few instances mentioned 2, 2, 9. Accordingly it is prohibited by Pâṇini to vajrabhartā instead of vajrasya bhartā (bearer of the thunderbolt) or odanapācakaḥ instead of odanasya pācakaḥ (one who cooks rice). But some schoolmaster, who commented on our great grammarian, discovered Pâṇini to have omitted some kind of words in -aka, which though not-denoting the agent are likewise forbidden to be compounded with a preceding genitive, as bhavataḥ śāyikā (your lying down) cp. P. 3, 3, 111. In order to make our sûtra comprise even them, the well-known yogavibhâga-expedient was taken recourse to, and our sûtra was split up in two. One made the discovery that the word kartari admitted of two acceptations, according to its being construed either with tṛjakābhyām or with the general adhikâra ṣaṣṭhī; in other terms, P. could mean either any genitive + agent in tṛ or aka or the subjective genitive + any noun tṛ or aka. By combining both and assigning to either an own sûtra the ἀπορία felt by the commentator found its λύσις. See but the artificial interpretation of both in the Kâçikâ. How Patañjali interpreted the rule we do not know, a comment of his on s. 15 and 16 being wanting; from vârtt. 2 on I, p. 415 it appears he was acquainted at least with s. 15.
In 2, 2, 11 it seems strange, that a special prohibition — that concerning the ordinal nouns of number — is enjoined immediately after the general one (2, 2, 10) which includes also that special case.

(*3)
The preceding sûtra 28 is too artfully interpreted by Pat. and Kâç. to have been interpreted well. It is likely, we have here again an instance of distortion by yogavibhâga. I am sure, Pâṇini himself has given but one rule kālā atyantasaṃyoge ca. Patañjali’s defence (I, p. 384) is not persuasive.

(*4)
I agree with Boetlingk and Whitney in explaining prāptajīvika and the like as bahuvrîhis. Pâṇini brings them under the tatpurushas, see 2, 2, 4. Inversely such compounds as surāpīta, māsajāta which P. 6, 2, 170 understands as bahuvrîhis, are to be recognised as tatpurushas.

(*5)
In Pâṇini’s text pūrvasadṛśasamonārtha- artha is of course to be construed with each of the members: pūrvārthi, sadṛśārtha, samārtha, ūṇārtha, cp. p. 92 N.

(*6)
Sûtra 2, 1, 30 is ill-handled by the commentaries. They expound tṛtīyā tatkṛtārthena guṇavacanena, as tatkṛtārthena were a dvandva = tatkṛtena + arthena. How they have come to this contorted interpretation I did not understand before perusing Patañjali; from him I have seen, that his very cavillations (I, 384 sq.) must have provoked it. Yet the aim of the author of our sûtra is unmistakable. He allows the instrumental to be compounded with any adjective (guṇavacanena), which has its justification by that instrumental: tatkṛtārthena = yasyārthas tayā tṛtīyayā kṛtaḥ syāt tena. So in niḥśvāsāndha ādarśaḥ the instance, I have quoted from the Râm., it is the exhalation that causes the mirror to be qualified a tarnished one.

(*7)
These compounds are not frequent alpaśaḥ (P. 2, 1, 38).

216. tatpuruṣaに係る諸制限

けれども、いくつかの制限も存在する。このため、名詞-格＋名詞から成る複合語を扱う際に、Pāṇiniはそれらを詳述している。その概要は以下の通り：

○前分がgen.のもの

I. 原則として、任意のgen.が複合されうる（ṣaṣthīsamāsa）。{P. 2, 2, 8.} gen.＋〈分詞 / 遊離分詞 / 未来受動分詞（kṛtya） / 不定詞〉、〈比較級 / 最上級 / 序数詞〉、〈名詞-tṛ / -aka〉など、いくつかの場合は例外である。{P. 2, 2, 11; 15} したがって、manuṣyāṇāṃ śūratamaḥ（男の中で最も英雄的な者）、chātrāṇāṃ pañcamaḥ（門弟のうち5番目の者）、apāṃ sraṣṭāḥ（水の創造者）、brāhmaṇasya kurvan (kṛtvā / karaṇīyam / kartum)（バラモンの利益のための行いetc.）は複合語に適さない。部分を表すgen.も同じく除外され、dat.的gen（129）も原則として複合語に適さないようである。—わずかながら、Pāṇiniの注釈者たち（*1）から、vicitrā sūtrasya kṛtiḥ pāṇinināのようなケースの目的語的gen.が知られる。ここでsūtrakṛtiḥ pāṇinināとすることはできない。というのも、両者はともに名詞kṛtiḥにより伝達される行為の主語と目的語が表されているからである。もし行為者が表現されていなかったなら、複合語sūtrakṛtiḥ（＝sūtrasya kṛtiḥ）を用いることができる（*2）。

○acc.

II. acc.を複合することができる（*3）。{P. 2, 1, 29.}

a.) māsakalyāṇaḥのような、時間を表す語と複合するもの
R. 2, 71, 18 saptarātroṣitaḥ pathi
Mâlat. I, p. 14 bahudivasopacīyamānam iva manmathavyathāvikāram upalabdhavān asmi

b.) grāmagataḥ（村に行った）、narakapatitaḥ（地獄に落ちた）、kaṣṭaśritaḥ（苦難に来たった）、sukhaprāptaḥ（*4）などのような、能動あるいは自動詞的意味を持つ分詞-taを伴うもの{P. 2, 1, 24; 26.}
実際にはもっと存在する。例えばkhaṭvārūḍha（P. 2, 1, 26）（*訳注1）による非難の言葉への制限は、他でārūḍhaを複合することの禁止を意味するものではない。以下をみよ：
Panc. 51 kareṇukārūḍhā
ibid. 30 gṛhāyātaḥ

○inst.

III. 以下の場合にはinst.を複合することができる：

a.) ahihataḥのような、行為者・用具＋動詞的名詞。{P. 2, l, 32.} 分詞-tavantは除外されるので、ahinā＋hatavānの複合は許されない。いくつかの決り文句はPāṇiniによって、kākapeyā nadī、śvalehyaḥ kūpaḥのように明らかに言及されているが、このことからinst.＋kṛtyaの複合語がすべて禁止されていると導くべきではない。{P. 2, 1, 33.}
Panc. 327 tvadbhakṣyo ‘yaṃ mānuṣaḥ = tvayā bha-

しかしながら、このような複合語はあまり見られない。

b.) 後分が類似、同等、優越、欠乏を表す語の場合（*5）。{P. 2, 1, 31.} 61と78をみよ。この種の複合語は以下の通り：pitṛsamaḥ (equal to his father)、mātṛsadṛśaḥ (resembling his mother)、māsapūrvaḥ and māsāvaraḥ (earlier —, later by a month)
M. 8, 217 alponaṃ karma = alpenonaṃ karma (v. a. work, almost finished)
Panc. 23 svāmisadṛśā eva bhavanti bhṛtyāḥ

c.) inst.＋kalaha（口論）、nipuṇa（賢い）、miśra（混ざった）、ślakṣṇa（なめらかな）。{P. 2, 1, 81.} Kâç.が例を与えている：asikalahaḥ、vāṅnipuṇaḥ、guḍamiśra、ācāraślakṣṇaḥ。124-注1で行ったように、この規則を同趣旨のあらゆる語に拡張する場合、頻出する複合語には-milita、-anvita、-yuta、-yukta、-upetaが含まれる。これらの多くは、前分が方向を表すinst.（78）である。

d.) 複合形容詞の前分が、それが結合している形容詞を説明する原因のinst.の場合（*6）。{P. 2, 1, 30.}
R. 3, 16, 13 niḥśvāsāndha ādarśaḥ (a mirror tarnished by exhalation)
ibid. 3, 55, 20 vīryasamaḥ (equal by strength)
ibid. 2, 118, 4 guṇaślāghyaḥ (praiseworthy by his qualities)
Panc. I, 39 jātyādimahotsāhā narendrāḥ = jātyādinā mahotsāhā narendrāḥ
Kumâras. 3, 12 tapovīryamahatsu (men, great by their heroic penance)

—inst.＋arthin（76-補足1）は同種である。
Panc. 10 kiṃ bhavān āhārārthī kevalam eva (are you desirous of nothing but food?)
Kathâs. 24, 176 etc.

e.) 何らかの材料と和えられた、あるいは2つの素材を混ぜた食べ物の場合：dadhyodanaḥ、guḍadhānāḥ。{P. 2, 1, 34; 35} 以下も同様：
Daç. 139 viṣānnam (empoisoned food)

○dat.

IV. 以下の場合にはdat.を複合することができる：

a.) yūpadāru＝yūpāya dāru、kuṇḍalahiraṇyam＝kuṇḍalāya hiraṇyamのようなケースにおける、目的のdat.。{P. 2, 1, 36.}

b.) viṣṇubaliḥ、gohitamなどのような、bali（…に捧げる）、hita（…に適した）、sukha（…に快い）、rakṣita（守られた）などを伴う間接目的語のdat.：{P. 2, 1, 36.}
Panc. I, 47 sā sevā yā prabhuhitā

—これらはartha（目的、ねらい、機会）と呼ばれる。-arthaの複合語については194で扱った。

○abl.

V. 以下の場合にはabl.を複合することができる：

a.) vṛkabhayam (fear of wolves)のような「恐怖」を表す語を伴う場合。{P. 2, 1, 37.}
R. 3, 27, 20 vyādhatrastā mṛgāḥ (deer, afraid of the hunter)

b.) 撤退を表すいくつかの分詞を伴う場合（*7）。
Daç. 89 nṛtyotthitā sā prātiṣṭhata (she rose from the dance and went away)
M. 2, 89 sāvitrīpatitaḥ (one who has forfeited the sâvitrî)
R. 3, 25, 24 cicchidur bibhiduś caiva rāmabāṇā guṇacyutāḥ

○loc.

VI. 以下の場合にはloc.を複合することができる：

a.) 装備、技術などの名詞に係る、関係を表すloc.（148）。{P. 2, 1, 40.}
R. 3, 19, 22 mānuṣau śastrasaṃpannau (two men, accomplished at arms)
Panc. I, 18 bhāṇḍakrayavicakṣaṇāḥ (skilled in the commerce of merchandises)

b.) 時間・空間を表す場合における特定のケース。この類について、Pāṇiniは以下の複合語に言及している：-siddha（用意された、仕上げされた）、-śuṣka（乾いた）、-pakva（調理された）、-bandha、時間の区分名＋kṛta：sāṃkāśyasiddha（sāṃkāśyaで用意された）、ātapaśuṣka（日に乾いた）、pūrvāhṇakṛta（午前に為された）など。{P. 2, 1, 41; 45}

c.) いくつかの定型句と諺。P. 2, 1, 42-44; 46-48をみよ。

（*1）
Kâç. on P. 2, 2, 14; Pat. I, p. 415をみよ。Patañjali自身はそこでの解釈を却下している。

（*2）
ṣaṣthīsamāsaはP. 2, 2, 8-17で扱われており、いくつかの記述は第3巻に散見される（P. 3, 3, 116など）。それらの補足と訂正はもちろん諸注釈で行われている。しかしながら、時折、注釈者の揚げ足取りが、いくつかの規則の正しい理解を覆い隠してしまっている。

Kâç.は、P. 2, 2, 14 karmaṇi caをそのadhikāra（P. 2, 1, 22 tatpuruṣaḥ）から誤ってバラして、ktena（P. 2, 2, 12?）とこの規則の解釈を、目的語的gen.についての教示として間違って解釈している。 さて、スートラがあらゆる目的語的gen.の複合禁止を含むものとして説明できず、実際そのような複合語が非常に一般的であったので、Kâç.は独自の規則であるubhayaprāptau karmaṇīti ṣaṣṭhyā iti grahaṇamを追加せねばならなかった。 この規則は確かにそれ自体で正しいだろうが、Pāṇiniのスートラには一切見られないものである。実際、Pāṇiniはここで目的語的gen.を考慮していない。P. 2, 2, 12-14を、一度に、かつ先入観なしに読むならば、s. 14のkarmaṇiはs. 12のktenaに係ることが明らかに見て取られる。s. 14はgen.＋過去受動分詞-taの複合を禁止するのである。tasya darśitam（彼に見せられた）の代わりにtaddarśitamとすることは許されないが、tena darśitam（彼によって見せられた）を意味する場合にはそれが許される。P. 2, 1, 32と比較せよ。
続くs. 15とs. 16—tṛjakābhyāṃ kartari / kartari ca—は不合理な些事へのこだわりのよい例を提供している。s. 15で、Pāṇiniは、行為者を表す場合の-tṛおよび-akaの語に関する規則を与えている；P. 2, 2, 9に言及される数例を除いて、gen.はこれらと複合できない。したがって、Pāṇiniはvajrasya bhartā（雷電を運ぶ者）の代わりにvajrabhartāとするとか、odanasya pācakaḥ（米を料理する者）の代わりにodanapācakaḥとするのを禁止している。けれども、我らが大文法学者に注釈した一部の学匠（Kâç.の記述？）は、Pāṇiniが一部の-akaの語の類を省略しており、よって記述されていないけれども、行為者がそれに先行するgen.と複合することが同じく禁止されていることを発見した：bhavataḥ śāyikā（あなたの横臥）。P. 3, 3, 111と比較せよ。パーニニ・スートラをそのように構成するために、有名なyogavibhāga法が用いられ、スートラは2つに分割された。そこで、kartariの語が、tṛjakābhyāmかadhikāraのṣaṣṭhīかのいずれかと結びつけて解釈されることで、2つの解釈を許すことが発見された；換言すれば、Pāṇiniは、あらゆるgen.＋行為者を表す名詞-tṛ・-akaか、あるいは主語的gen.＋あらゆる名詞-tṛ・-akaか、のどちらかを意味している可能性があるのである。両方を組み合わせ、そして各々のスートラに当てはめることで、注釈者にそう感ぜられたところの難問はその解決策を見出した。Kâç.における両スートラの不自然な解釈をみよ。Patañjaliがこれをどのように解釈したかについては、s. 15-16についての注釈が欠けているために知られない；Kielhorn I, p. 415のvârtt. 2から、彼が少なくともs. 15に精通していたらしいことはわかる。
P. 2, 2, 11では、奇妙なことに、特別な禁則（序数詞に関するもの）が、その特別則をも含む一般則（P. 2, 2, 10）の直後に課されている。

（*3）
先行するs. 28は、PatañjaliとKâç.にあまりに巧妙に解釈するので、うまく解釈されていない。おそらく、ここにはまたyogavibhāgaによる曲解があるのだろう。Pāṇini自身はたった一つの規則kālā atyantasaṃyoge caを与えたのみであると私は確信している。Patañjaliの弁（I, p. 384）は説得的ではない。

（*4）
prāptajīvika等をbahuvrīhiとして説明することに関してはBoetlingkとWhitneyに同意する。Pāṇiniはこれらをtatpuruṣaに置く。P. 2, 2, 4をみよ。反対に、P. 6, 2, 170がbahuvrīhiとして理解するところのsurāpītaやmāsajātaはtatpuruṣaと見なされるべきである。

（*5）
Pāṇiniのテキストではpūrvasadṛśasamonārtha-となっている。arthaは当然ながら〔複合語の〕各要素と共に解釈されるべきである：pūrvārthi、sadṛśārtha、samārtha、ūṇārtha。124-注1と比較せよ。

（*6）
P. 2, 1, 30 tṛtīyā tatkṛtārthena guṇavacanenaは諸注釈者に誤って扱われている。彼らはこれを、tatkṛtārthenaがdvandva（tatkṛtena＋arthena）であるかのように解釈する。彼らがどのようにしてこの歪んだ解釈に至ったのかは、Patañjaliを精読する前には理解できなかった；私の見たところでは（I, 384以下）、まさしく彼の屁理屈がそれを引き起こしたに違いない。しかし、Pāṇiniのスートラのねらいは紛れもない。彼は、inst.が、そのinst.によって根拠付けられるところの形容詞と（guṇavacanena）複合することを許す：tatkṛtārthena＝yasyārthas tayā tṛtīyayā kṛtaḥ syāt tena。よって、上記の例R. 3, 16, 13 niḥśvāsāndha ādarśaḥ（呼気により曇った鏡）では、鏡を曇らせる原因は呼気（niḥśvāsa）である。

（*7）
これらの複合語は稀（alpaśaḥ）である（P. 2, 1, 38）。

（*訳注1）
Böhtlingk（1839 p. 62; 1887 p. 46）ではともに以下の通り：
P. 2, 1, 26 khaṭvā kṣepe
〔acc.の〕khaṭvāは、侮辱が表される場合（kṣepe）、分詞-taの名詞とtatpuruṣa複合語を形成する

スートラにはkhaṭvārūḍhaの文言が見えない。KāśikāとMahābhāṣyaでも同様であるが、注釈部分にてkhaṭvārūḍhaに言及している。KāśikāについてはSāstrī 1876, p. 111, ll. 19-24をみよ。ここでは分かりやすいMahābhāṣyaを引用しておく：

Mahābhāṣya (Kielhorn, I, p. 384, ll. 10-12):
kim udāharaṇam / khaṭvārūḍho jālmaḥ / kṣepe iti ucyate kaḥ kṣepo nāma / adhītya snātvā gurubhiḥ anujñātena khaṭvā āroḍhavyā / ya idānīm ato anyathā karoti sa ucyate khaṭvārūḍho ‘yam jālmaḥ / na ativratavān iti
〔Vārttikaでは〕何と言われているか？　“khaṭvārūḍhaḥ jālmaḥ kṣepe”と言われている。kṣepa（侮辱の場合）とはつまるところ何か？　勉強し・沐浴して・師の許しによって・寝台に上がる、このとき、〔慣習を破って〕それとは違った仕方で行う者が「寝台に上がった者」（khaṭvārūḍha）＝「馬鹿者」である。これがjālma（軽蔑される者）であって、〔ここでのārūḍhaは〕「…に登った」（ativratavat）〔を、すなわち、字義通りに「寝台に上がった者」を指すのでは〕ではない。

これに関する短いけれども有用な要約がCardona 1997の§301と309にある。

217. additions to the rules of Pâṇini thereabout.

This list of possible kinds of tatpurushas, made up of noun-case + noun, is however not complete, as will soon appear, if one undertakes to systematize the tatpurushas occurring in fact in some literary work (*1). So, among others, Pâṇini does not mention the abl. of comparison, compounded with anya and itara; the instrumental + words expressive of plenty; the accus. with the participial adjectives in -u. Then, many more participles, whose former part is some noun-case, are in common use, though not necessarily, if at all, implied by the foresaid rules.

1. abl. of comparison + anya, especially, if the former part be a pronoun, as Hit. 30 bhavadanyo mayā kaḥ suhṛtprāptavyaḥ, Panc. I, 12 trade is said to be the best means for making money, tadanyaḥ [sc. upāyaḥ] saṃśayātmakaḥ (any other but this is dangerous).

2. abl. of compar. + itara. — Such compounds are an elegant paraphrase, while calling something: the contrary of its opposite. So dakṣiṇetara = savya »left,” sulabhetara upāyaḥ (a difficult expedient), Daç. 175 tayā navavadhvā dveṣam alpetaraṃ babandha (he bore his newly married wife a heavy grudge).

3. instrum. + word expressive of plenty. Of the kind are those in -āṭhya, -pracura, -samākula etc. Panc. 319 ūrṇāpracuro ’yaṃ meṣaḥ, ibid. 7 siṃhavyāghrasamākule ’sminvane.

4. accus. + adjective in -u derived of a desiderative. Panc. 3 nāham arthalipsuḥ, M. 7, 197 yudhyeta jayaprepsurapetabhīḥ.

5. Instances of noun-cases + participles are manifold and often met with. First, such as where the former part represents a sociative instrumental, as Panc. I, 164 saṃmānasaṃyuktāḥ sevakāḥ, ibid. I, 229 bhāryā parasaṃgatā (a wife, who holds illicit intercourse with another). — Then, such as are expressive of separation (82), Panc. I, 35 sevāvivarjitaḥ (abstaining from attendance), ibid. p. 1 putrā vivekarahitāḥ (sons, deprived of discernment), ibid. I, 189 saṃmānavihīnaḥ, cp. 198. — Further locatives + sthita, gata (cp. 197), jāta, utpanna etc., as R. 3, 31, 2 janasthānasthitā rākṣasāḥ (the râxasas, staying at Janasthāna), Panc. I, 128 sadaivāpadgato rājā bhogyo bhavati mantriṇām (when being in distress, a king is always the prey of his ministers), ibid. I, 104 mūṣikā gṛhajātāpi hantavyā, Kathâs. 42, 149 aṅghripatitaḥ (fallen at their feet). — Or the former member is a dative or loc. of purpose: Panc, I, 125 rājyalābhodhyato vṛtraḥ (Vrtra, striving for obtaining the royalty); an accusative: ibid, p 37 saṃjīvakānuraktaḥ (attached to S.); an abl. of origin: ibid. p. 2 kulaprasūtasya (of one born from a respectable family); a loc. of reference: ibid. I, 15 goṣṭhikakarmaniyuktaḥ. And so on.

6. As to the compound adjectives, they may generally be said to be comprised by the rules of Pâṇini, as their former part is a genitive or may be accepted as such. Among them are to be noticed yogya, anurūpa, ucita, samartha, those of skill and ability as vicakṣaṇa, abhijña, kovit, then such as sadṛśa, sama (cp. 216, III b.) Panc. 17 rājā bhṛtyānuttamapadayogyān hīnādhamasthāne niyojayati, ibid. 21 tasya ca śabdānurūpeṇa parākrameṇa bhāvyam (his strength will be in proportion to his voice), ibid. 27 bhavān vacanapaṭur dṛśyate, ibid. 13 katham ahaṃ sevānabhijñaḥ. Even indubitable dative-like genitives are compounded with the adjectives, which they qualify. Panc. p. 1 rājā tān śāstravimukhānālovya (as the king understood they were averse to the çâstras —). Panc. p. 1 affords even this instance of a tatpur., made up of a dat. of interest + subst., when calling some king sakalārthikalpadrumaḥ (v. a. a blessing for all the indigent).

7. Compounds made up of a genitive + agent in -tṛ, though explicitly interdicted by Pâṇini, are in fact met with. Panc. I, 2 namo ’stu nayaśāstrakartṛbhyaḥ, ibid. p. 7 tasya maṅgalavṛṣabhau dhūrvoṭhārau sthitau (two splendid bulls drew his chariot).

8. Finally we may set up a category apart for such tatpurushas, the former part of which is a noun-case, doing more or less duty of an adverb. Panc. 21 na yuktaṃ svāminaḥ pūrvopārjitaṃ vanaṃ tyaktum, here pūrvopārjita = pūrvam upārjita »acquired before,” Kathâs. 29, 82 vyājasapraṇayair vāvyair vañcyate (— is deceived by words falsely kind), Panc. 63 snehasaṃbhāṣaḥ (a friendly discourse), Mhbh. 1, 152, 34 sukhasuptānvane bhrātṝn… na bodhayiṣyāmi (I will not awake my brothers who are sleeping quietly in the forest), Kathâs. 42, 149 paryāyāliṅgitaḥ (embraced by turns). In all but the first of these examples the former part is an instrumentalis modi, used almost as an adverb (77). Cp. the following paragraph.

(*1)
It would be indeed an interesting subject-matter for investigation to compare on a large scale these statements of Pâṇini with the facts offered us by tee extant Sanskrit literature.

217. Pāṇiniの規則への補足

しかし、何らかの文献に実際に出てくるtatpuruṣaを体系化しようとすればすぐにわかるように、名詞-格＋名詞で構成されるこの可能な類のtatpuruṣaのリストは、完全ではない（*1）。その他のうち、Pāṇiniは、anyaおよびitaraと複合する比較のabl.、inst.＋充足を表す語、分詞的形容詞-uを伴うacc.、には言及していない。次に、さらに多くの、一般的に用いられる、前分が名詞-格の分詞は、必ずしもそうではないが、あったとしても、前述の規則によって暗示されていない。

1. 比較のabl.＋anya、とりわけ前分が代名詞であるもの。
Hit. 30 bhavadanyo mayā kaḥ suhṛtprāptavyaḥ Panc. I, 12 trade is said to be the best means for making money, tadanyaḥ [sc. upāyaḥ] saṃśayātmakaḥ (any other but this is dangerous)

2. 比較のabl.＋itara。—このような複合語は、何かを呼ぶ場合のエレガントな言い換えである：ここでは前分の反対のものが指示される：dakṣiṇetara＝savya（左）、sulabhetara upāyaḥ（難しい手段）。
Daç. 175 tayā navavadhvā dveṣam alpetaraṃ babandha (he bore his newly married wife a heavy grudge)

3. inst.＋充足を表す語。この種のものは-āṭhya、-pracura、-samākulaなど。
Panc. 319 ūrṇāpracuro ’yaṃ meṣaḥ
ibid. 7 siṃhavyāghrasamākule ’sminvane

4. acc.＋願望法から派生する形容詞-u。
Panc. 3 nāham arthalipsuḥ
M. 7, 197 yudhyeta jayaprepsurapetabhīḥ

5. 名詞-格＋分詞の例は、種類・頻度ともに多い。
・前分が付帯のinst.を表すもの：
Panc. I, 164 saṃmānasaṃyuktāḥ sevakāḥ
ibid. I, 229 bhāryā parasaṃgatā (a wife, who holds illicit intercourse with another)

・分離を表すinst.（82）、198と比較せよ：
Panc. I, 35 sevāvivarjitaḥ (abstaining from attendance)
ibid. p. 1 putrā vivekarahitāḥ (sons, deprived of discernment)
ibid. I, 189 saṃmānavihīnaḥ

・loc.＋sthita、gata（197）、jāta、utpanna etc.：
R. 3, 31, 2 janasthānasthitā rākṣasāḥ (the râxasas, staying at Janasthāna)
Panc. I, 128 sadaivāpadgato rājā bhogyo bhavati mantriṇām (when being in distress, a king is always the prey of his ministers)
ibid. I, 104 mūṣikā gṛhajātāpi hantavyā
Kathâs. 42, 149 aṅghripatitaḥ (fallen at their feet)

・前分が目的のdat.かloc.：
Panc, I, 125 rājyalābhodhyato vṛtraḥ (Vrtra, striving for obtaining the royalty)

・acc.：
ibid, p 37 saṃjīvakānuraktaḥ (attached to S.)

・起源のabl.：
ibid. p. 2 kulaprasūtasya (of one born from a respectable family)

・関係のloc.：
ibid. I, 15 goṣṭhikakarmaniyuktaḥ
など。

6. 複合形容詞について、これらの前分はgen.であるか、あるいはそのように解釈されうるので、通常Pāṇiniの規則に含まれていると言える。これらのうち留意すべきなのは以下：yogya、anurūpa、ucita、samartha；およびvicakṣaṇa、abhijña、kovitのような技術・能力を表すもの；sadṛśa、sama（216-III-b）。
Panc. 17 rājā bhṛtyānuttamapadayogyān hīnādhamasthāne niyojayati
ibid. 21 tasya ca śabdānurūpeṇa parākrameṇa bhāvyam (his strength will be in proportion to his voice)
ibid. 27 bhavān vacanapaṭur dṛśyate
ibid. 13 katham ahaṃ sevānabhijñaḥ

疑う余地のないdat.的gen.でも、それが修飾する形容詞と複合する。
Panc. p. 1 rājā tān śāstravimukhānālovya (as the king understood they were averse to the çâstras —)
Panc. p. 1 affords even this instance of a tatpur., made up of a dat. of interest + subst., when calling some king sakalārthikalpadrumaḥ (v. a. a blessing for all the indigent)

7. gen.＋行為者を表す-tṛは、Pāṇiniにより明確に禁止されているが、実際には用いられる。
Panc. I, 2 namo ’stu nayaśāstrakartṛbhyaḥ
ibid. p. 7 tasya maṅgalavṛṣabhau dhūrvoṭhārau sthitau (two splendid bulls drew his chariot)

8. 前分が名詞-格で、多少なりとも副詞のはたらきをするもの。
Panc. 21 na yuktaṃ svāminaḥ pūrvopārjitaṃ vanaṃ tyaktum, here pūrvopārjita = pūrvam upārjita »acquired before,”
Kathâs. 29, 82 vyājasapraṇayair vāvyair vañcyate (— is deceived by words falsely kind)
Panc. 63 snehasaṃbhāṣaḥ (a friendly discourse)
Mhbh. 1, 152, 34 sukhasuptānvane bhrātṝn… na bodhayiṣyāmi (I will not awake my brothers who are sleeping quietly in the forest)
Kathâs. 42, 149 paryāyāliṅgitaḥ (embraced by turns)

これらの例のうち最初のものを除いては、前分はほぼ副詞として用いられる方法のinst.（instrumentalis modi）である（77）。以下のパラグラフと比較せよ。

（*1）
このようなPāṇiniの記述を、現存するサンスクリット文献から与えられる事実と大規模に比較することは、興味深い調査対象となるであろう。

218. The former member is an adverb or a particle.

The former part of a tatpurusha may also be an ad­verb or a particle. Panc. 59 prāgvṛttāntaḥ (the matter of late), Kathâs. 6, 165 prativasatipatākāḥ (flags, waving from every house), ibid. 25, 29 sā [viz. purī] ca me ‘vaśyagantavyā (and I have to go from necessity to that town), Kumâras, 3, 4 nitāntadīrghais tapobhiḥ (by very long penance).

Among the particles several are noticed by Pâṇini, viz. svayam [2, 1, 25], sāmi [ibid. 27], the negation a- [2, 2, 6], īṣat [2, 2, 7], kim [2, 1, 64], ka, the particles styled »gati” and such particles as pra, ava, ā when meaning »a little,” su, duḥ, ati [2, 2, 18](*1) cp. 210 at the end. — So svayaṃmṛtaḥ (died of himself), sāmikṛta (half done), abrāhmaṇaḥ »no brahman” or »none but a brahman,” īṣadunnata (a little elevated), prapitāmahaḥ (a great-grandfather), sujanaḥ (a good man), durjanaḥ (a wicked man), etc.

(*1)
As to su, duḥ, īṣat, in sukara (easy to be done), duṣkara (hard to be done) and the like, see P. 3, 3, 126-130.

218. 前分が副詞・不変化辞のもの

tatpuruṣaの前分は副詞や不変化辞でもよい。
Panc. 59 prāgvṛttāntaḥ (the matter of late)
Kathâs. 6, 165 prativasatipatākāḥ (flags, waving from every house)
ibid. 25, 29 sā [viz. purī] ca me ‘vaśyagantavyā (and I have to go from necessity to that town)
Kumâras, 3, 4 nitāntadīrghais tapobhiḥ (by very long penance)

不変化辞のうちいくつかはPāṇiniにより言及されている。すなわち、svayam（2, 1, 25）、sāmi（ibid. 27）、否定辞a-（2, 2, 6）、īṣat（2, 2, 7）、kim（2, 1, 64）、ka、”gati”と呼ばれる不変化辞とpra、ava、「少し」を意味する場合のā、su、duḥ、ati（2, 2, 18）（*1）である。210末尾と比較せよ。—よって、svayaṃmṛtaḥ（自殺した）、sāmikṛta（半分終わった）、abrāhmaṇaḥ（バラモンでない / バラモン以外の何者でもない）、īṣadunnata（少し高い）、prapitāmahaḥ（曽祖父）、sujanaḥ（善人）、durjanaḥ（悪人）などとなる。

（*1）
sukara（行い易い）、duṣkara（行い難い）などにおけるsu、duḥ、īṣatについてはP. 3, 3, 126-130をみよ。

219. The former member is yathā or yāvat.

Some relative pronouns and adverbs are likewise fit for being compounded with some noun, especially yathā and yāvat. {P. 2, 1, 7 and 8.} Those beginning with yathā are the most common, they are either adverbs of the type yathākālam, yathāvayaḥ (according to time, — to age), or their second member is a participle in -ta as yathokta (as said) (*1). Examples: Mhbh. 1, 145, 16 nivartadhvaṃ yathāgṛham (go back, each to his own house), ibid. 1, 149, 1 yathāsaṃpratyayaṃ… preṣayāmāsa puruṣam (he sent a man, as was agreed before); — R. 3, 13, 25 yathopadiṣṭena pathā (on the way, as has been pointed out), Daç. 151 imam arthaṃ labdhakṣaṇo yathopapannair upāyaiḥ sādhayiṣyati (when having got the opportunity he will discharge this affair by such means, as are fit), Panc. 295 yathānirdiṣṭaḥ kṣapaṇakaḥ sahasā prādurbabhūva.

Examples of yāvat. — Panc. 276 itaḥ prabhṛti yāvajjīvaṃ mayātmā bhavate dattaḥ (from this day, I have given my own self to you for my whole lifetime), Kâç. on P. 2, 1, 8 yāvadamatraṃ brāhmaṇānām āmantrayasva (invite of the brahmans according to the number of the veasels), Panc. 54 yāvadvātsyāyanoktavidhinā (conformably to the rules, taught by Vâtsyâyana).

(*1)
Pâṇini (2, 1, 7) mentions only the former type.

219. 前分がyathā・yāvatのもの

いくつかの関係代名詞と副詞、とりわけyathāとyāvatは同様に名詞と複合しうる。{P. 2, 1, 7; 8} yathāで始まるものが最も一般的で、これらは yathākālamやyathāvayaḥ（時の経つにつれ、年を重ねるにつれ）のタイプの副詞か、yathokta（言ったように）のような分詞-taが後分のものか、のどちらかである（*1）。
Mhbh. 1, 145, 16 nivartadhvaṃ yathāgṛham (go back, each to his own house)
ibid. 1, 149, 1 yathāsaṃpratyayaṃ… preṣayāmāsa puruṣam (he sent a man, as was agreed before)
R. 3, 13, 25 yathopadiṣṭena pathā (on the way, as has been pointed out)
Daç. 151 imam arthaṃ labdhakṣaṇo yathopapannair upāyaiḥ sādhayiṣyati (when having got the opportunity he will discharge this affair by such means, as are fit)
Panc. 295 yathānirdiṣṭaḥ kṣapaṇakaḥ sahasā prādurbabhūva

yāvatの例：
Panc. 276 itaḥ prabhṛti yāvajjīvaṃ mayātmā bhavate dattaḥ (from this day, I have given my own self to you for my whole lifetime)
Kâç. on P. 2, 1, 8 yāvadamatraṃ brāhmaṇānām āmantrayasva (invite of the brahmans according to the number of the veasels)
Panc. 54 yāvadvātsyāyanoktavidhinā (conformably to the rules, taught by Vâtsyâyana)

（*1）
P. 2, 1, 7は前者のタイプにのみ言及する。

220. Comparison expressed by tatpurushas.

The tatpurusha serves also to express comparison. Such compounds are partly adjectives, partly substantives. The former are of the type ghanaśyāma (cloud-black), indranīla (sky-blue). The latter are made up of the thing’s real name + the image, under which it is represented, as puruṣavyāghraḥ, rājasiṃhaḥ, dordaṇḍaḥ, karakisalayaḥ.

Examples of the former type. Kâm. 3, 12 jalāntaś candracapalaṃ jīvanam (life I as fickle as the moon, that shines in the water), Daç. 174 himaśiśira- (as cold as ice), R. 3, 23, 1 gardabhāruṇaḥ (red-grey as the colour of an ass), Mhbh. 1, 152, 2 prāvṛḍjaladharaśyāmaḥ (as dark as a cloud in the rainy season), Mâlav. V, p. 122 parighagurubhidorbhiḥ, etc.

The latter type is adapted to signify either praise or blame. Generally the metaphors used are conventional ones. In this way a resolute, energetic character is called puruṣasiṃhaḥ, a beautiful face sukhapadmam, ānanāravindam sim., eloquent speech vāṅmadhu, heavy sorrow is by a standing comparison śokasāgaraḥ which ocean it is difficult or impossible to pass, and so on. In ancient literature this rather allegorical style is still employed with moderation and within certain limits. But the flowery compositions of medieval India are full of them so as not rarely to make the image appear an appendix wholly meaningless, if not to please the ear of the reader and to display the vaidagdhya of the author. The accumulation of such allegorical designations becomes tedious indeed, unless good taste direct their employment.

Sometimes the metaphor is worked out. Then we may have a set of homogeneous images, expressed by compounds. So Panc. I, 241 gopālena prajādhenor vittadugdhaṃ śanaiḥ śanaiḥ / grāhyam, here gopāla is »king” but at the same time it conveys the meaning of „cow-herd,” as go is = „cow” and = »earth,” »he must draw the vittadugdham (money-milk) of his prajādhenoḥ (subject-cow) by degrees” v. a. »a king must draw the money of his subjects by degrees, just as the cowherd draws the milk from his cow.” Mṛcch. IV, p. 138 iha sarvasvaphalinaḥ kulaputramahādrumāḥ / niṣphalatvamalaṃ yānti veśyāvihagabhakṣitāḥ (young gentlemen often come to poverty, being spoiled by courtesans, like great trees, the fruits of which are eaten out by birds). Kathâs. 29, 188 a faithful wife is thus compared to a warrior — her conjugal faith is her chariot, duty her charioteer, good behaviour her armour, wit her weapon bhartṛbhaktirathārūḍhāḥ śīlasaṃnāharakṣitāḥ / dharmasārathayaḥ sādhvyo jayanti matihetayaḥ.

Rem. 1. According to vernacular grammar, this class of compounds is to be considered a subdivision of the karmadhârayas, there being sâmânâdhikaranya between both members. This explication cannot be right, for it does not account for the inverse order of the two members; one should f. i. expect ratnastrī instead of strīratnam, as in the karmadhâraya the qualifying noun is of course put first. In fact, we have here no karmadhârayas, but shashtîsamâsas. The former member is a genitive, but it does not bear everywhere the same character. Sometimes it is a partitive one, as puruṣasiṃhaḥ = puruṣāṇāṃ (or puruṣeṣu) siṃhaḥ, rājāpasadaḥ, »an outcast among the kings,” cp. the compounds in uttama (best) and adhama (worst, lowest). Sometimes, too, it is a genitive of the kind represented by our „a jewel of a woman,” »a bell of a fellow,” Lat. scelus hominis; so gṛhabhūtiḥ (Mudr. III, p. 102) »a beauty of a house,” sahāyasaṃpat (Mudr. III, p. 121) v. a. ,excellent helpers,” bhāryācelam (P. 6, 2, 126) »a slut of a wife.” Not rarely both acceptations are alike probable; strīratnam f. i. may be as well = strīṣu ratnam »a jewel among women” as = »a jewel of a woman.”

Rem. 2. Pâṇini treats the said compounds severally, see 2, 1, 53; 56; 62; 66; 6, 2, 126 sq. — Note -kīṭa and -apasada expressing blame, and -ratna, -pāśa, -rūpa, signifying admiration.

220. tatpuruṣaで表現される比喩

tatpuruṣaは比喩を表現するのにも資する。そのような複合語は、一部は形容詞、一部は実名詞である。前者はghanaśyāma（雲の黒）、indranīla（空の青）の類である。後者は、〈物の実際の名前＋それを意味するイメージ〉から成る：puruṣavyāghraḥ、rājasiṃhaḥ、dordaṇḍaḥ、karakisalayaḥ。

前者の例：
Kâm. 3, 12 jalāntaś candracapalaṃ jīvanam (life I as fickle as the moon, that shines in the water)
Daç. 174 himaśiśira- (as cold as ice)
R. 3, 23, 1 gardabhāruṇaḥ (red-grey as the colour of an ass)
Mhbh. 1, 152, 2 prāvṛḍjaladharaśyāmaḥ (as dark as a cloud in the rainy season)
Mâlav. V, p. 122 parighagurubhidorbhiḥ

後者の類は、称賛か非難かのどちらかを表すのに適当である。ふつう直喩が常用される。この方法では、決然としてエネルギッシュな人物はpuruṣasiṃhaḥ、美貌はsukhapadmamやānanāravindamなど、雄弁はvāṅmadhu、深い悲しみは慣習的比喩により、越えがたい海のśokasāgaraḥ（火の海）と呼ばれる、など。古代の文献におけるこのやや寓話的な文体は、今でも適度に、一定の制限の中で用いられている。けれども、中世インドの美文体はこれらに満ちており、読者の耳を喜ばせたり、作者の知性（vaidagdhya）が示されたりしない場合でも、〔比喩に用いられる〕イメージが全く意味のない付加物と見えることは滅多にない。それを用いることでよい風情を演出されるのでなければ、そのような寓話的な名称が重なるのは実に退屈なものとなる。

時に、直喩が功を奏するすることがある。そのとき、複合語により表現される一続きの同質なイメージが見て取られる。
Panc. I, 241 gopālena prajādhenor vittadugdhaṃ śanaiḥ śanaiḥ / grāhyam

ここで、goは「牛」でも「地上」でもあるので、gopālaは「王」と同時に「牛飼い 」をも意味する。よって以下のようになる：「彼は、彼の〈臣民 / 牛〉の〈牛乳 / 金〉を、徐々に搾り取らねばならない」＝「王はその臣民の金を、牛飼いが彼の牛から牛乳を搾るように、徐々に搾り取らねばならない」。
Mṛcch. IV, p. 138 iha sarvasvaphalinaḥ kulaputramahādrumāḥ / niṣphalatvamalaṃ yānti veśyāvihagabhakṣitāḥ (young gentlemen often come to poverty, being spoiled by courtesans, like great trees, the fruits of which are eaten out by birds)
Kathâs. 29, 188 a faithful wife is thus compared to a warrior — her conjugal faith is her chariot, duty her charioteer, good behaviour her armour, wit her weapon bhartṛbhaktirathārūḍhāḥ śīlasaṃnāharakṣitāḥ / dharmasārathayaḥ sādhvyo jayanti matihetayaḥ

【補足1】
土着文法によれば、この類の複合語は、両要素の間にsāmānādhikaranyaがあるので、karmadhārayaの下位カテゴリと見なされるべきである、という。この説明は2要素間の逆順を考慮していないため、正しくはない；例えば、karmadhārayaでは修飾する名詞が最初に置かれるため、strīratnamではなくratnastrīが必要となる。事実、この場合にはkarmadhārayaはなく、ṣaṣthīsamāsaが用いられている。ṣaṣthīsamāsaの前分はgen.であるけれども、どこでも同じ性質をもつわけではない。時にpuruṣasiṃhaḥ＝puruṣāṇāṃ (puruṣeṣu) siṃhaḥ（人中の獅子）やrājāpasadaḥ（王の中の追放者）のような部分を表すgen.の場合がある；uttama（最高の）とadhama（最低の）の複合語と比較せよ。また、時にこのgen.は英語の”a bell of a fellow”や”a jewel of a woman”に代表される類のgen.でもある。ラテン語の”scelus hominis”の構文は以下に対応する；gṛhabhūtiḥ（Mudr. III, p. 102）»a beauty of a house”、sahāyasaṃpat（Mudr. III, p. 121） »excellent helpers”、bhāryācelam（P. 6, 2, 126） »a slut of a wife”。両者の意味が同じことも稀ではないだろう；例えばstrīratnamはstrīṣu ratnam＝「女性の中の宝石」＝「女性の宝石」と同等足り得る。

【補足2】
Pāṇiniは上記の複合をいくつか扱っている。2, 1, 53; 56; 62; 66、6, 2, 126以下をみよ。—侮辱を表す-kīṭaと-apasada、称賛を表す-ratna、-pāśa、-rūpaに注意せよ。

221. Tatpurushas made up of three or more themes.

Tatpurushas, made up of three or more stems, are always dissolvable into two members, either of which may be a compound itself. Mṛcch. p. 125 sukhaduḥkhasuhṛddhavān, here the a former part is a dvandva sukhaduḥkhayoḥ suhṛddhavān = sukhe ca duḥkhe ca. Panc. 323 matkaṇṭhasthitā [viz. ratnamālā], here the former part is a tatpurusha itself, matkaṇṭhasthitā being = matkaṇthe, that is mama kaṇthe, sthitā. — Panc. II, 153 strīvākyāṅkuśavikṣuṇṇāḥ (stirred by the sting-like words of a woman), here vikṣuṇṇa is the latter member, the former being a tatpurusha of comparison, the former member of which strīvākya is itself an ordinary shashthîsamâsa. — Mhbh. 11 155, 24 puṣpitadrumasānuṣu = puṣpitadruseṣu sānuṣu, here the former part of the tatpurusha is a bahuvrîhi. — Kâm. 2, 43 niyataviṣayavartī (one firmly attached to wordliness) is illustrative of the species of those, whose latter member is a compound, the analysis being niyataṃ viṣayeṣu vartamānaḥ.

221. 3つ以上の要素から成るtatpuruṣa

3つ以上の語幹から成るtatpuruṣaは、常に2つの複合語に分解しうる。
Mṛcch. p. 125 sukhaduḥkhasuhṛddhavān
ここで前分はdvandvaで、sukhaduḥkhayoḥ suhṛddhavān＝sukhe ca duḥkhe caとなる；
Panc. 323 matkaṇṭhasthitā [viz. ratnamālā]
ここで前分はtatpuruṣaで、matkaṇṭhasthitāはmatkaṇthe＝mama kaṇtheとsthitā；
Panc. II, 153 strīvākyāṅkuśavikṣuṇṇāḥ (stirred by the sting-like words of a woman)
ここでvikṣuṇṇaが後分で、前分は比喩のtatpuruṣa。その前分であるstrīvākya自体は通常のṣaṣthīsamāsa；
Mhbh. 11 155, 24 puṣpitadrumasānuṣu＝puṣpitadruseṣu sānuṣu
tatpuruṣaの前分はbahuvrīhi；
Kâm. 2, 43 niyataviṣayavartī (one firmly attached to wordliness)
後分が複合語の類。分析は以下のようになる：niyataṃ viṣayeṣu vartamānaḥ

222. Nature and characteristics of the bahuvrîhi.

The difference between the tatpurusha and the bahuvrîhi is an essential one. The former implies no more than is purported by its consistituent elements, but the bahuvrîhi always adds something tacitly understood, generally the conception of „having, possessing.” indraśatruḥ when tatpurusha = indrasya śatruḥ „Indra’s foe,” when bahuvrîhi it means „having Indra for foe, one whose foe is Indra;” sūryavarṇaḥ, when tatp. = sūryasya varṇaḥ „the colour of the sun,” when bahuvr. it denotes „one having the colour of the sun.” The bahuvrîhi, therefore, is invariably an adjective, referring to some substantive (*1). Pâṇini then is quite right, when he defines the bahuvrîhi as „a complex of elements serving to qualify some other word (*2).

Like other adjectives, the bahuvrîhis may be used as adverbs, when put in the accus. of the neuter singular, see 240.

(*1)
By this it is however not asserted, that a bahuvrîhi cannot be used as a substantive, but only this: when used so, they are to be just as any other adjective, that does duty of a subst. mahātmā when= »a noble-man” is to be compared with such a word as śuciḥ when meaning »an honest man” or bālaḥ when = »boy.”

(*2)
P. 2, 2, 23 sq. bahuvrīhir anekam anyapadārthe. According to that definition, Indian grammar does by no means make restriction as to the number of the elements out of which a bahuvrîhi is made up. This is distinctly expressed in a metrical rule of the Kâtantra (2, 5, 9)
syātāṃ yadi pade tu yadi vā syur bahūny api / tāny anyasya padasyārthe bahuvrīhiḥ etc.
Nevertheless, in analyzing even intricate bahuvrîhis it will appear that, logically, there are but two members — predicate and subject — either or both of which may be compounds themselves, even if it would not always be allowed to use such compounds by themselves as separate words. Pâṇini himself knows »a class of compounds only allowed for the sake of being used as the former part of other compounds” [P. 2, 1, 51].

222. bahuvrīhiの性質と特徴

tatpuruṣaとbahuvrīhiとの間の違いは重要である。前者はその構成要素によって主張されている以上のものを意味しないが、bahuvrīhiは常に、暗黙裡に理解される何か、一般的には「持つ」、「所有する」という概念を付加する。indraśatruḥは、tatpuruṣaだとindrasya śatruḥ（インドラの敵）であるが、bahuvrīhiだと「インドラを敵に持つ者」を意味する；sūryavarṇaḥは、tatpuruṣaだとsūryasya varṇaḥ（太陽の色）で、bahuvrīhiだと「太陽の色を持つ者」となる、など。したがって、bahuvrīhiは常に何らかの実名詞に係る形容詞である（*1）。そうすると、bahuvrīhiを「他の語を修飾することに資する要素の複合体」と定義したPāṇiniはとても正しい（*2）。

他の形容詞と同様に、bahuvrīhiは、n.sg.acc.で置かれる場合に副詞として用いられる。240をみよ。

（*1）
しかし、これによってbahuvrīhiを実名詞として用いることができないと主張されているわけではなく、これだけが主張されているのである：bahuvrīhiとして用いる場合、これら複合語はちょうど実名詞のはたらきをする他の形容詞のように見なされる。「貴い人」を表す場合のmahātmāは、「正直者」を意味する場合のśuciḥや「少年」を表す場合のbālaḥと比較すべきである。

（*2）
P. 2, 2, 23-24 bahuvrīhir anekam anyapadārthe（*訳注）
その定義によれば、インド文法学は、bahuvrīhiを構成する要素の数を制限しない。これはKâtantra（2, 5, 9）の韻律規則に明確に表されている：
syātāṃ yadi pade tu yadi vā syur bahūny api / tāny anyasya padasyārthe bahuvrīhiḥ etc.

それにもかかわらず、複雑なbahuvrīhiを分析すると、論理的には、述語と主語の2要素のみが存在し、そのような複合語がそれ自体で単語として用いられることがないとしても、そのどちらか、または両方が複合語でありうることがわかる。Pâṇini自身は「他の複合語の前分として用いられる場合にのみ許される複合語類」を知っている（P. 2, 1, 51）。

223.

Its members are to be called predicate and subject.

From a syntactic point of view, the bahuvrîhi, it may be made up of three, four or more elements, does contain but two members, virtually identical with the subject and the predicate of a full sentence, just as the tatpurusha represents a main noun with its attribute.

The predicate precedes, the subject is the latter member.

And, as within the tatpurusha the attribute is put at the head and the main noun behind (210), so within the bahuvrîhi the predicate precedes, the subject is the latter member. When analyzing f. i. the bahuvrîhi mahāvīryaḥ we get the clause yasya vīryaṃ mahat „he whose strength is great,” similarly sūryavarṇaḥ = yasyavarṇaḥ sūryasyeva „he whose colour is like the sun’s,” bhūtalanyastalocanaḥ is an epithet of somebody, whose eyes are fixed on the earth yasya bhūtale nyaste locane. In these examples, the words vīrya, varṇa, locana are the subjects within the bahuvrîhis, that which precedes them being the predicates.

Types of bahuvrîhis.

In treating of the tatpurushas we have distinguished between l. the karmadhârayas, 2. those the former member of which represents some noun-case, 3. where it is a particle. In an analogous way we may speak of three types of bahuvrîhis: a.) those, where there is grammatical concord between subject and predicate, b.) such, whose predicate is a noun-case, c.) such, where it is a particle.

Type a. — Here the predicate is mostly an adjective or a participle, as indraśatruḥ „having Indra for foe.” When adjective, the bahuvrîhi has generally the worth of Latin gen. or abl. qualitatis, or abl. modi. So vīrovyūḍhoraskaḥ — heros lato pectore, Panc. 62 saraḥ svalpatoyam = lacus exiguae aquae. When participle, the bahuvrîhi not rarely concurs with the gerund, the absolute locative and the like. It may as well be said tyaktanagaro ‘raṇyāyāhaṃ prātiṣṭhe as nagaraṃ tyaktvā or tyakte nagare- (I left the town and set out for the forest).

Those belonging to type b.) are such as aśvamukhaḥ (having a horse’s face), Panc. 71 saṃjīvako yuṣmatpādānām upari drohabuddhiḥ (Sanj. meditates of doing harm to Your Majesty), here drohabuddhiḥ = yasya drohāya (or drohe) buddhiḥ „whose mind is to do harm.”

Those belonging to type c.) are such as adhomukha (having one’s face cast down), antaḥsattva (pregnant), evaṃprabhāvo rājā (a king of such a power). Very common are those, commencing by a-, su-, duḥ-, as aputraḥ (having no sons), suputraḥ (having a good son or good sons).

Type a.) and c.) are much more common than type b.).

223. 承前

○bahuvrīhiの要素は主語・述語と呼ばれる

統語論的観点からすれば、bahuvrīhiは3つ以上の要素で構成され得る。そして、tatpuruṣaがその属性を持つ語の主名詞を表すのと同じように、文全体の主語・述語と事実上等しい、2つの要素を含む。

○述語が先行し、主語は後分になる

また、tatpuruṣaでは属性が頭に、主名詞が後ろに置かれるように（210）、bahuvrīhiでは述語が先行し、主語は後分になる。例えばbahuvrīhiのmahāvīryaḥを分析してみると、yasya vīryaṃ mahat（その強さが優れているところの彼）となる。同様に、sūryavarṇaḥはyasyavarṇaḥ sūryasyeva（その色が太陽に似ているところの彼）。地面を凝視する者を形容してbhūtalanyastalocanaḥ＝yasya bhūtale nyaste locaneなど。これらの例でのvīrya、varṇa、locanaは述語に先行する、bahuvrīhiにおける主語である。

○bahuvrīhiの種類

tatpuruṣaを扱うにあたって、1. karmadhāraya、2. 前分が名詞-格のもの、3. 前分が不変化辞のもの、を区別した。同様に、bahuvrīhiの3区分について述べる：a.) 主語と述語の間に性・数・格の一致があるもの、b.) 述語が名詞-格のもの、c.) 述語が不変化辞のもの、の3つである。

a.) 述語はほとんど形容詞か分詞である：indraśatruḥ（インドラを敵として持つ）。形容詞の場合、bahuvrīhiはふつう、ラテン語におけるgen.・性質を表すabl.・仕方を表すabl.の性質を持つ。よって、vīrovyūḍhoraskaḥはラテン語だと”heros lato pectore”、Panc. 62 saraḥ svalpatoyamは”lacus exiguae aquae”、のように対応する。分詞の場合、bahuvrīhiは、遊離分詞（gerund）・絶対処格（abs. loc.）等と一致することが稀ではない。なので、tyaktanagaro ‘raṇyāyāhaṃ prātiṣṭhe（私は街を出て森に向かった）はnagaraṃ tyaktvāやtyakte nagare-でも同じい。

b.) aśvamukhaḥ（馬の顔を持つ）など。
Panc. 71 saṃjīvako yuṣmatpādānām upari drohabuddhiḥ (Sanj. meditates of doing harm to Your Majesty)
このdrohabuddhiḥはyasya drohāya (drohe) buddhiḥ（傷つける心）と等しい。

c.) adhomukha（下に向いた顔を持つ）、antaḥsattva（妊婦）、evaṃprabhāvo rājā（そのような力の王）など。頻繁に用いられるのはa-、su-、duḥ-で始まるもの：aputraḥ（息子を持たない）、suputraḥ（良い息子を持つ）。

a.)とc.)はb.)よりも一般的である。

224.

Examples of bahuvrîhis. Type a.) Nala. 1, 5 tathaivāsīd vidarbheṣu bhīmo bhīmaparākramaḥ. R. 3, 16, 11 mṛdusūryāḥ paṭuśītāḥ… divasā bhānti sāṃpratam (now the days are appearing with a mild sun, m11ch fog and a sharp cold). Hit. 90 asau pāpāśayaḥ (he is of a wicked disposition). Panc. 150 tena [sc. kroḍena]… daṃṣṭrāgreṇa pāṭitodaraḥ pulindo gatāsurbhūtale ‘patat (the Pulinda fell down lifeless on the earth, having his belly split up by the edge of the teeth of the boar). Mâlav. I, p. 14 rājā / anuvācitalekham amātyaṃ vilobya (as the minister has finished his lecture of the letter, the king looks on him). Panc. 71 saṃdahyamānahṛdayaḥ svayam evābhyāgato vaktum (as it stung my mind, I myself am come to tell you of it).

Nothing impedes, of course, both the subject and the predicate being concordant substantives. Bhoj. 17 rājānaḥ kośabalā eva vijayino nānye (no other kings are successful in their wars but those, whose power is a treasury), Panc. 185 tatra vāyasarājo ’nekakākaparivāraḥ prativasati sva [the bahuvr. = yasyāneke kākāḥ parivāraḥ »whose attendance are many crows”], Daç, 82 aryavaryaḥ kuberadattanāmā (a great merchant, whose name is Kub.), R. 3, 19, 22 mānuṣau cīrakṛṣṇājināmbarau.

Rem. 1. In such bahuvrîhis, as have an adjective behind, that adjective does duty of a subst. So f. i. Çâk. I abhirūpabhūyiṣṭhā pariṣat (the assembly is for the greater part made up of distinguished people) here the bahuvr. = yasyā abhirūpā bhūyiṣṭhāḥ »most of which are abhirūpāḥ,” Mâlat. I, p. 2 uditabhūyiṣṭha eṣa bhagavāntapanaḥ »the sun has almost risen,” R. 2, 40, 17 sītātṛtīyān ārūḍhān dṛṣṭvā (— saw them mounted, having Sîtâ as the third), R. 3, 55, 15 sahasram ekaṃ mama kāryapuraḥsaram (thousand men whose main object [puruḥsaram] it is to carry out my orders).

Rem. 2. A proper kind of bahuvrîhis are such as asikara (having a sword in one’s hand), aśrukaṇṭha (sobbing, liter. »one having tears in his throat”) . In analyzing them, the latter part turns out a locative, for asikaraḥ = yasyāsiḥ kare sa and aśrukaṇaṭhaḥ = yasyāśrūṇi kaṇṭhe sa. For the rest, we have here no exception to the general rule on the arrangement of the two members of a bahuvrîhi. It is asi and aśru which are predicated, not kare nor kaṇṭhe, for the intention is to say not that Mr. so and so has a hand or a throat, but what it is, he keeps within (*1). R. 3, 51, 9 Sîtâ has the epithet bāṣpalocanā »having tears in her eyes.” Comp. Whitney § 1303.

Rem. 3. In some bahuvrîhis the order of the members is optional. {P. 2, 3, 37.}{P. 2, 2, 37.} One may say promiscuously āhitāgniḥ and agnyāhitaḥ (one who keeps the holy fires), jātaputraḥ or putrajātaḥ (one having children). Of the kind are dantajāta or jātadanta, see M. 5, 58 with Kull., guḍapriya and priyaguḍa (fond of sweetmeats) (*2), R. 2, 119, 5 munayaḥ kalaśodyatāḥ (hermits with uplifted pitchers). Participles in -ta must be put behind, if the predicate be a weapon, therefore asyudyataḥ (with uplifted sword), see vârtt. on P. 2, 2, 36.

Rem. 4. The type a.) of the bahuvrîhi in its outer form is often identical with a karmadhâraya, for the discrepancies in the accentuation are not heeded. In practice, one avoids to use as bahuvrîhis such compounds as are wont to be karmadhârayas, as sarvajana, maharṣi, kalarava, and inversely such as mahābāhu, bahuśruta, kṛtakārya will not have to be otherwise accepted than as bahuvrîhis (*3). Yet, it often is only the context which will enable the reader how to accept a given compound.

(*1)
Cp. vârtt. 4 of Pat. on P. 2, 2, 36.

(*2)
As to priya, Pat. vârtt. 2 on P. 2, 2, 85 teaches the option. But it seems better to explain guḍapriya as being a tatpursha, because 1st priya may be not only = »beloved,” but also = »loving,” see Petr. Dict. IV, p. 1161 s. v. 1 c), 2ly as Pâṇini somewhere else P. 6, 2, 15 sq. mentions some tatpurusha, in -prati. The same may apply to some of the participles in -ta, if not to all. Since pīta may sometimes have an active signification and sometimes a passive one, it is plain we are allowed to compound as well the tatpurusha kṣīrapītaḥ = lac potus, as the bahuvr. pītakṣīraḥ = lacte poto. Comp. what has been said p. 157 N. 2. [216-(*4)]
Panc. 283 affords a specimen of a kind of compounds, in which two types are confounded: kācic chṛgālikā māṃsapiṇḍagṛhītavadanā, here the author seems to have blended promiscuously two bahuvrîhis māṃsapiṇḍavadanā and gṛhītamāṃsapiṇḍā, either of which would have sufficed. Comp. Hariv. 5814 cakrodyakara.

(*3)
See Cappeller Vâmana’s Stilregeln: Kâvyasamaya 7 and 8.

224. 承前：a.)の例文

a.)
Nala. 1, 5 tathaivāsīd vidarbheṣu bhīmo bhīmaparākramaḥ
R. 3, 16, 11 mṛdusūryāḥ paṭuśītāḥ… divasā bhānti sāṃpratam (now the days are appearing with a mild sun, m11ch fog and a sharp cold)
Hit. 90 asau pāpāśayaḥ (he is of a wicked disposition)
Panc. 150 tena [sc. kroḍena]… daṃṣṭrāgreṇa pāṭitodaraḥ pulindo gatāsurbhūtale ‘patat (the Pulinda fell down lifeless on the earth, having his belly split up by the edge of the teeth of the boar)
Mâlav. I, p. 14 rājā / anuvācitalekham amātyaṃ vilobya (as the minister has finished his lecture of the letter, the king looks on him)
Panc. 71 saṃdahyamānahṛdayaḥ svayam evābhyāgato vaktum (as it stung my mind, I myself am come to tell you of it)

もちろん、主語と述語の両方が実名詞に一致することを妨げるものは何もない。
Bhoj. 17 rājānaḥ kośabalā eva vijayino nānye (no other kings are successful in their wars but those, whose power is a treasury)
Panc. 185 tatra vāyasarājo ’nekakākaparivāraḥ prativasati sva [bv. = yasyāneke kākāḥ parivāraḥ] »whose attendance are many crows”
Daç, 82 aryavaryaḥ kuberadattanāmā (a great merchant, whose name is Kub.)
R. 3, 19, 22 mānuṣau cīrakṛṣṇājināmbarau

【補足1】
形容詞が後ろに来るbahuvrīhiでは、その形容詞が実名詞のはたらきをする。
Çâk. I abhirūpabhūyiṣṭhā pariṣat (the assembly is for the greater part made up of distinguished people) here the bahuvr. = yasyā abhirūpā bhūyiṣṭhāḥ »most of which are abhirūpāḥ”
Mâlat. I, p. 2 uditabhūyiṣṭha eṣa bhagavāntapanaḥ »the sun has almost risen”
R. 2, 40, 17 sītātṛtīyān ārūḍhān dṛṣṭvā (— saw them mounted, having Sîtâ as the third)
R. 3, 55, 15 sahasram ekaṃ mama kāryapuraḥsaram (thousand men whose main object [puruḥsaram] it is to carry out my orders)

【補足2】
厳密なbahuvrīhiはasikara（その手に剣を持った）、aśrukaṇṭha（その喉に涙を持った＝すすり泣く）の類である。これらを分析すると、asikaraḥはyasyāsiḥ kare sa、aśrukaṇaṭhaḥはyasyāśrūṇi kaṇṭhe saとなる。その他、bahuvrīhiの2要素の順番に関する一般則への例外はない。〔それによって指示される人・物が〕手や喉を持っていることではなく、それが何であるかが意図されるので、述語なのはasiとaśruであってkareやkaṇṭheではない（*1）。
R. 3, 51, 9 Sîtâ has the epithet bāṣpalocanā »having tears in her eyes”

Whitney §1303を参照せよ。

【補足3】
一部のbahuvrīhiでは、要素の順番が選択的である。{P. 2, 2, 37} 乱雑にāhitāgniḥやagnyāhitaḥ（聖なる炎を守る者）、jātaputraḥやputrajātaḥ（子供を持つ者）などとできる。この類にはM. 5, 58 with Kull. dantajāta / jātadanta、guḍapriya / priyaguḍa（砂糖を好む）（*2）、R. 2, 119, 5 munayaḥ kalaśodyatāḥ（掲げられた水差しを持つ隠者）など。述語が武器を表す場合、分詞-taは後ろに置かれねばならない。よってasyudyataḥ（掲げられた剣を持つ）となる。vârtt. on P. 2, 2, 36をみよ。

【補足4】
a.)のbahuvrīhiは、その外形においては、アクセントの違いに注意が払われないために、しばしばkarmadhārayaと同じである。実際には、sarvajana、maharṣi、kalaravaのようなkarmadhārayaではない複合語をbahuvrīhiとして用いることは避けられ、逆にmahābāhu、bahuśruta、kṛtakāryaのような複合語はbahuvrīhiとして見なされる。

（*1）
vârtt. 4 of Pat. on P. 2, 2, 36をみよ。

（*2）
priyaについて、Pat. vârtt. 2 on P. 2, 2, 35は選択的なものであることを説いている。けれども、tatpuruṣaとしてのguḍapriyaと説明するのがベターと思われる。というのも、第1に、priyaは「愛しい」だけを意味するのではなく、「愛情のこもった」（etwas liebend）をも表す；Petr. Dict. IV, p. 1161の該当項1-c)をみよ。第2に、別の所（P. 6, 2, 15以下）でPāṇiniは-pratiのtatpuruṣaに言及しているからである。同じことが、全てではないにしても、分詞-taのものにも適用される。pītaは能動的な意味を持つこともあれば受動的な意味を持つこともあるので、tatpuruṣaのkṣīrapītaḥ（＝lac potus）をbahuvrīhiのpītakṣīraḥ（＝lacte poto）と同じように複合することが許されているのは明白である。216-注4と比較せよ。
Panc. 283は2つのタイプが混ざった類の複合語を提示する： kācic chṛgālikā māṃsapiṇḍagṛhītavadanā

ここでは、作者は、どちらかだけで十分なmāṃsapiṇḍavadanāとgṛhītamāṃsapiṇḍāの2つの複合語を乱雑に混合したらしい。Hariv. 5814 cakrodyakaraと比較せよ。

（*3）
Cappeller 1880, 7-8をみよ。

225.

Type b.). Panc. 24 kadācid damanako ’yam ubhayavetano bhūtvā mamopari duṣṭabuddhiḥ syāt, here ubhayavetana is the epithet of one »who accepts wages from both parties,” Kathâs. 72, 186 haṃsadvandvarūpaṃ kiṃcit siddhadvayam (two siddhas, who bore the shape of flamingos), Mâlav. I, p. 24 kauśikī yativeṣā (K. wearing the dress of an ascetic).

Rem. Comparison is sometimes expressed by them, as R. 3, 69, 43 vau yuvāṃ vṛṣabhaskandhī (who are you, whose shoulders are like those of a bull ?). R. 5, 17, 10 Râxasawomen bear the epithet gajoṣṭrahayapādāḥ (with the feet of elephants, camels and horses). — But also by type a.) as rājīvalocanaḥ (having lotus-eyes), candrānanaḥ (moon-faced).

225. b.)の例文

b.)
Panc. 24 kadācid damanako ’yam ubhayavetano bhūtvā mamopari duṣṭabuddhiḥ syāt（ubhayavetana＝「両方の派閥から賃金を受け取る者」）
Kathâs. 72, 186 haṃsadvandvarūpaṃ kiṃcit siddhadvayam (two siddhas, who bore the shape of flamingos)
Mâlav. I, p. 24 kauśikī yativeṣā (K. wearing the dress of an ascetic)

【補足】
これらは時に比較を表す。
R. 3, 69, 43 vau yuvāṃ vṛṣabhaskandhī (who are you, whose shoulders are like those of a bull ?)
R. 5, 17, 10 Râxasawomen bear the epithet gajoṣṭrahayapādāḥ (with the feet of elephants, camels and horses)

a.)によっても表される。
rājīvalocanaḥ (having lotus-eyes)
candrānanaḥ (moon-faced)

225*.

Type c.) Prabodh. V, p. 103 devī mām evamavasthaṃ na samāśvasayati (the queen does not comfort me, who am in such a state), Panc. I, 137 antaḥsāraiḥ… mantribhir dhāryate rājyam (a kingdom is upheld by pithy ministers), Kumâras. 3, 14 the gods bear the epithet uccairdviṣaḥ (whose adversaries are mighty). — Apart from the very common employment of a, su, duḥ as the predicate in bahuvrîhis, several particles and prepositions may be used so, as ut, niḥ, vi, ati, ava etc., as nīrasa (sapless), udañcaliḥ »one with folded hands,” Ragh. 2, 74 puram utpatākam (a flagged town), Daç. 137 udāyudhaḥ (with uplifted weapon), Harsha. 9 tarur vipallavaḥ (a leafless tree), Bhoj, 8 khaḍgam apakośaṃ kṛtvā jagāma.

Rem. Compounding with sa-, saha- and the like has the same power, as English -ful, similarly English -less is expressed by compounds, beginning with a-, niḥ-, vi-, vigata-, vīta- etc. (*1).

(*1)
Patañjali enumerates also different species of bahuvrîhis: a.) those, the members of which are samânâdhikaraṇa as citragu, b.) the former part of which is an avyaya, as nīcairmukha, c.) whose former part is a locative or something compared as kaṇṭhekāla, uṣṭramukha, d.) where it is a gen. generis or a gen. materiae as keśacūḍa (with one's hair tied up, liter. »[bearing] a knot of hair”), suvārṇālaṅkāra (wearing golden ornaments), e.) whose former part is one of the gaṇa prādayaḥ [P. 1, 4, 58] as praparṇa (unleaved), f.) negation + noun, as aputra. In the case of c.) an ellipsis is stated of a middle element, kaṇṭhekāla representing kaṇṭhesyaḥ kālo ’sya and uṣṭramukha being = uṣṭramukham iva mukhamasya. As to e.) and f.) Patañjali states the option between saying in full prapatitaparṇo and the abridgement praparṇo vṛkṣaḥ, likewise between avidyamānaputraḥ and aputraḥ. Cp. f. i. Daç. 35 udgatabāṣpalocanaṃ tamapṛccham with ibid. 176 sā… udbāṣpovāca; here the full udgata- locana and the short udbāṣpa are synonymous.

225*. c.)の例文

c.)
Prabodh. V, p. 103 devī mām evamavasthaṃ na samāśvasayati (the queen does not comfort me, who am in such a state)
Panc. I, 137 antaḥsāraiḥ… mantribhir dhāryate rājyam (a kingdom is upheld by pithy ministers)
Kumâras. 3, 14 the gods bear the epithet uccairdviṣaḥ (whose adversaries are mighty)

—bahuvrīhiの述語としてはa、su、duḥがごく一般に用いられるが、それ以外にはut, niḥ, vi, ati, avaなどのいくつかの不変化辞と前置詞がそのように用いられる：nīrasa（樹液のない）、udañcaliḥ（畳んだ腕を持つ者）。
Ragh. 2, 74 puram utpatākam (a flagged town)
Daç. 137 udāyudhaḥ (with uplifted weapon)
Harsha. 9 tarur vipallavaḥ (a leafless tree)
Bhoj, 8 khaḍgam apakośaṃ kṛtvā jagāma

【補足】
sa-やsaha-などを伴う複合語は、英語の”-ful”と同じ効果を持つ。同様に、”-less”はa-、niḥ-、vi-、vigata-、vīta-などで始まる複合語で表される（*1）。

（*1）
Patañjaliも異なる種類のbahuvrīhiを列挙する：
a.) その要素がsamānādhikaraṇaであるもの：citragu
b.) 前分がavyayaのもの：nīcairmukha
c.) 前分がloc.あるいは比較対象のもの：kaṇṭhekāla、uṣṭramukha
d.) 前分が起源のgen.か素材のgen.のもの：keśacūḍa（結わえた髪の）、suvārṇālaṅkāra（金の装飾を身に着けた）
e.) 前分がgaṇaのprādayaḥ（P. 1, 4, 58）のもの：praparṇa（葉の落ちた）
f.) 否定辞＋名詞：aputra

c.)の場合、kaṇṭhesyaḥ kālo ’syaがkaṇṭhekāla、uṣṭramukham iva mukhamasyaがuṣṭramukhaとなるなど、中間の要素が省略される。e.)とf.)については、Patañjaliは、 省略のないprapatitaparṇoと省略形のpraparṇo vṛkṣaḥの任意選択を記述している。avidyamānaputraḥとaputraḥも同様。Daç. 35 udgatabāṣpalocanaṃ tamapṛcchamとibid. 176 sā… udbāṣpovācaを比較せよ；ここで、省略のないudgata- locanaと短縮形のudbāṣpaは同義である。

226. Large bahuvrîhis.

As bahuvrîhis of three and more members are excessively frequent, we will adduce some instances of them.

1. the subject is a compound. Panc. 322 tyaktāhārakriyaḥ (abstaining from taking food) here the subject of the bahuvr. is āhārakriyā, a tatpurusha; Vâr. Yog. 1, 8 kṣapitadhanarāṣṭro bhavati (he loses his wealth and his kingdom) here the subj. is a dvandva dhanarāṣṭre »wealth and kingdom;” Daç. 78 a Jaina monk is thus qualified prakīrṇamalapaṅkaḥ prabalakeśaluñcanavyathaḥ prakṛṣṭatamakṣutpipāsādiduḥkhaḥ (v. a. covered with dust and mud, enduring a heavy pain by pulling out his hair, suffering very much from hunger, thirst and the like) here the subjects of the three bahuvr. are respectively the dvandva malapaṅkau, the tatp. keśaluñcanavyathā and the tatp. kṣutpipāsādiduḥkham and of them the two tatpurushas are themselves made up of more than two themes, as it is the compound keśaluñcana which qualifies vyathā, and similarly the compound kṣutpipāsādi, which is the cause of duḥkham.

2. the predicate is a compound. — Instances of this category are very often met with, especially such bahuvrîhis as exhibit this type: qualifying noun-case + adjective or participle + substantive. Panc. 42 taddhartā madavihvalāṅgaḥ = madena vihvalānyaṅgāni yasya sa; Kathâs. 72, 180 prāpārkasaṃtaptasikatāṃ marubhūmim, here the bahuvr. is to be analyzed arkeṇa saṃtaptāḥ sikatā yasyāṃ sā [marubhūmiḥ]. But also other types, as: Mudr. III, p. 124 vultures (mṛdhrāḥ) have the epithet dīrghaniṣkampapakṣāḥ, here the analysis is yeṣāṃ dīrghā niṣkampāśca pakṣāḥ, the predicate, therefore, is an adjective-dvandva (208). Çâk. VII a curse is said to be aṅgulīyakadaeśanāvasānaḥ = yasyāṅgulīyakasya darśanamavasānaṃ sa śāpaḥ.

3. both subject and predicate are compounds, Kâd. I, p. 46 tatra vasrabdhaviracitakulāyasahasrāṇi… śukaśakunikulāni prativasanti sma (crowds of parrots and other birds were dwelling there, building confidently thousand(s) of nests), here visrabdhaviracita is the predicate and kulāyasahasra the subject of the bahuvrlhi, the analysis of which is of course yeṣāṃ visrabdhaṃ viracitāni [-taṃ] kulāyānāṃ sahasrāṇi [-srama].

226. 長大なbahuvrīhi

3つ以上の要素から成るbahuvrīhiは非常に頻繁である。これにはいくつかの用例を加えられよう。

1. 主語が複合語であるもの
Panc. 322 tyaktāhārakriyaḥ (abstaining from taking food)：Bv.の主語がTp.のāhārakriyā
Vâr. Yog. 1, 8 kṣapitadhanarāṣṭro bhavati (he loses his wealth and his kingdom)：主語がdvandvaのdhanarāṣṭre »wealth and kingdom”
Daç. 78 a Jaina monk is thus qualified prakīrṇamalapaṅkaḥ prabalakeśaluñcanavyathaḥ prakṛṣṭatamakṣutpipāsādiduḥkhaḥ (v. a. covered with dust and mud, enduring a heavy pain by pulling out his hair, suffering very much from hunger, thirst and the like)

最後のものには3つのBv.があり、主語はそれぞれdvandvaのmalapaṅkau、Tp.のkeśaluñcanavyathā、Tp.の kṣutpipāsādiduḥkham。Tp.の2つはそれ自体が2つ以上の主題から成るものである；keśa-luñcanaはvyathāを修飾し、同様にkṣut-pipāsa-ādiはduḥkhamの原因である。

2. 述語が複合語のもの
このカテゴリの例はよく見られる。とりわけ、修飾詞の名詞-格＋形容詞・分詞＋実名詞のタイプのBv.。
Panc. 42 taddhartā madavihvalāṅgaḥ = madena vihvalānyaṅgāni yasya sa
Kathâs. 72, 180 prāpārkasaṃtaptasikatāṃ marubhūmim = arkeṇa saṃtaptāḥ sikatā yasyāṃ sā [marubhūmiḥ]

他の形式もある：
Mudr. III, p. 124 vultures (mṛdhrāḥ) have the epithet dīrghaniṣkampapakṣāḥ = yeṣāṃ dīrghā niṣkampāśca pakṣāḥ （述語は形容詞のdvandva。208をみよ）
Çâk. VII a curse is said to be aṅgulīyakadaeśanāvasānaḥ = yasyāṅgulīyakasya darśanamavasānaṃ sa śāpaḥ

3. 主語・述語がともに複合語のもの
Kâd. I, p. 46 tatra vasrabdhaviracitakulāyasahasrāṇi… śukaśakunikulāni prativasanti sma (crowds of parrots and other birds were dwelling there, building confidently thousand(s) of nests)

visrabdhaviracitaが述語でkulāyasahasraが主語。分析は以下の通り：yeṣāṃ visrabdhaṃ viracitāni [-taṃ] kulāyānāṃ sahasrāṇi [-srama]

227.

In the case of non-compound words, adjectives carrying the notion of having, possessing, as is taught in Sanskrit etymology, may be made by putting some derivative suffixes as -mant, -vant, -maya, -ita, -in etc. to the substantive, as agrimant (fiery), putrin (having a son) and the like. Of these suffixes, -in is very common (*1). Kathâs. 24, 9 kirīṭī kuṇḍalī divyaḥ khaḍgī cāvātarat pumān (a divine person descended wearing a diadem, earrings and a sword), cp. R. 3, 50, 21. {P. 5, 2, 128.} They may also be put to dvandvas. Mhbh. 1, 126, 21 jaṭājinī (wearing tresses and a deer-skin), Pat. I, p. 1 sāmsnā-lāṅgūla-kakuda-khura-viṣāṇināṃ saṃpratyayo bhavati sa [go] śabdaḥ.

Now, sometimes, these suffixes are added even then, if they are in no ways necessary for the understanding. So R. 3, 15, 11 padmaiḥ surabhigandhibhiḥ… ramyā padminī (a pond charming by its sweet scented lotuses), likewise Panc. 53 the weaver, who bas assumed the attributes of Vishṇu, is said to be viṣṇucihnitaḥ = viṣṇoś cihnair anvitaḥ, Bhoj. 2 a brahman is said sakalavidyācāturyavān = sakalāya vidyāsu cāturyeṇa yuktaḥ. In these cases no suffix was required, for the babuvrîhis surabhigandha, viṣṇucihna, sakalavidyācāturya would be quite regular and plain. Compare Panc. I, 46 suvarṇapuṣpitā pṛthvī = suvarṇapuṣpā pṛthvī. This rather pleonastic idiom is especially used in some standing compounds. Grammarians teach and practice confirms -in being readily added to compounds in -śobhā, -śālā, -mālā, -dharma, -śīla, -varṇa. {P. 5, 2, 132.}

(*1)
See P. 5, 2, 115; 116; 128 with the vârttikas on them.

227. 複合語に付加される所有接尾辞-in類

複合語でない語の場合、所有の概念を伝達する形容詞は、サンスクリット語源学で説かれるように、-mant、-vant、-maya、-ita、-inなどのような派生接尾辞を実名詞に付すことで作られる：agrimant（火の付いた）、putrin（息子を持つ）等。これら接尾辞のうち-inがごく一般的である（*1）。{P. 5, 2, 128.}
Kathâs. 24, 9 kirīṭī kuṇḍalī divyaḥ khaḍgī cāvātarat pumān (a divine person descended wearing a diadem, earrings and a sword)

R. 3, 50, 21と比較せよ。 これらはdvandvaにも付されうる。
Mhbh. 1, 126, 21 jaṭājinī (wearing tresses and a deer-skin)
Pat. I, p. 1 sāmsnā-lāṅgūla-kakuda-khura-viṣāṇināṃ saṃpratyayo bhavati sa [go] śabdaḥ

さて、時にこれら接尾辞は、理解するのに全く必要ない場合にも付加されていることがある。
R. 3, 15, 11 padmaiḥ surabhigandhibhiḥ… ramyā padminī (a pond charming by its sweet scented lotuses)
Panc. 53 the weaver, who bas assumed the attributes of Vishṇu, is said to be viṣṇucihnitaḥ = viṣṇoś cihnair anvitaḥ
Bhoj. 2 a brahman is said sakalavidyācāturyavān = sakalāya vidyāsu cāturyeṇa yuktaḥ

これらの場合には、babuvrīhiのsurabhigandha、viṣṇucihna、sakalavidyācāturyaが決り文句で分かりやすいために、接尾辞が必要ない。以下と比較せよ：
Panc. I, 46 suvarṇapuṣpitā pṛthvī = suvarṇapuṣpā pṛthvī

このかなり冗長なイディオムは、特に一部の定型句の複合語で用いられる。文法学者らの教示と実践は、-śobhā、-śālā、-mālā、-dharma、-śīla、-varṇaの複合語に-inが容易に付加されることを追認する。{P. 5, 2, 132}

（*1）
P. 5, 2, 115; 116; 128とそのvārttikaをみよ。

228. Typical compounds.

So I call such compounds whose latter element is almost used in a typical sense, which is more or less remote from their primitive meaning. By them the great importance of compounding for Sanskrit composition appears best. Such among them, as are fit for periphrasing case-relations, have already been dealt with in Chapter IX, especially 188-196. Of the others the most remarkable are: 1. those in ādi, -ādya, -ādika, -prabhṛti which are expressive of „and so on,” 2. those in -pūrvam and -puraḥsaram which may serve to make adverbs of manner, 3. those in -pūrva =„formerly —,” 4. those in -mātram, which does duty of a limitative particle, vāṅmātram being nearly the same as vāgeva, 5. those in -kalpa, -deśya, -deśīya, -prāya to express „nearly, almost,” 6. those in -rūpa, if rūpa may be rendered by „namely,” 7. those in -antaram when having the worth of „some” or „other.” And so on.

228. 典型的複合語

後ろの要素が、原義から多少離れた典型的な意味で用いられている複合語を、私は典型的複合語と呼ぶ。これらにより、サンスクリットの作文法にとっての複合語の重要性が最もよく見て取られるのである。これらのうち、格関係を迂言するのに適したものは第9章、とりわけ188-196で既に扱った。その他のもので注目すべきなのは以下である：
1. 「…など」を表すādi、-ādya、-ādika、-prabhṛti
2. 方法を表す副詞を作る-pūrvam・-puraḥsaram
3. 「以前…」を表す-pūrva
4. 限定の不変化辞のはたらきをする-mātram（vāṅmātram＝vāg eva）
5. 「大体、ほとんど」を表す-kalpa、-deśya、-deśīya、-prāya
6. rūpaが「すなわち」を表す場合の-rūpa
7. 「或る」「他の」の価値を持つ-antaram
等々。

229.

To give a fuller account of them, we will treat of them separately.

1. -ādi etc.

1. Those in -ādi, -ādika, -ādya, -prabhṛti are babuvrîhis, meaning properly „the beginning of which is —,” as is still plain f. i. in M. 1, 50 etadantās tu gatayo brahmādyāḥ (the existences, at the head of which stands Brahmâ, end here). Commonly they are expressive of »etc.; and the like.” Panc. 8 tasyaurdhvadehikakriyā vṛṣotsargādikāḥ sarvāścakāra (be performed in his honour all the funeral rites, viz. the vṛshotsarga etc.), Hit. 123 suvarṇavastrādi prasādadānaṃ kriyatām (bestow on them presents of the king’s favour, gold, clothes and so on), Panc. 62 sarve matsyakacchapaprabhṛtayastaṃ papracchuḥ (all [aquatic animals], fishes, tortoises etc. —). In these examples the compounds are adjectives, but often they are used as substantives too [see note 1 on p. 165 (222)], as Bhoj. 64 aham api brāhmaṇaputraḥ / tvāmatra śayānaṃ vīkṣya kamaṇḍalūpavītādibhir brāhmaṇaṃ jñātvā bhavadāstarāsanna evāhaṃ prasuptaḥ, Panc. 27 adyaprabhṛti prasādanigrahādikaṃ tvayaiva kāryam (from to-day it is you by whom favour, punishment, etc. are to be administered.)

2. -pūrva, -puraḥsara

2. When adjectives, those in -pūrva and -puraḥsara may have the same purport as those in -ādi etc. Panc. 20 vyāghradvīpivṛkapuraḥsarāḥ sarve (all of them, tigers, panters, wolves and the rest). When adverbs, the latter member is almost meaningless: prītipūrvaṃ bhāṣate = saprīti bhāṣate, Hit. 7 tasya bahumānapuraḥsaraṃ putrān samarpitavān (respectfully he gave over his sons to him).

3. bhūtapūrva and the like.

3. Those in pūrva may also signify »having been formerly so and so,” but now being so no more: āḍhyapūrvaḥ „one who once has been rich,” bhūtapūrvaḥ (of old). N. 1, 13 na kvacid dṛṣṭapūrvā (never seen before), Çâk, VI satyamūḍhapūrvā tatrabhavatī rahasi śakuntalā (indeed, I once have wedded secretly the Lady Çak.) (*1).

4. -mātram

4. a.) Compounds in -mātram are bahuvrîhis, used as substantives of the neuter, and properly have the meaning »the exact measure (mātrā) of which is —.” Yet, as a rule they are used as if their latter member were some limitative particle and -mātram may be translated by »but, only.” Prabodh. I, p. 13 kiṃ vadantīmātram etat (it is but a vain rumour), Panc. 192 ko’pi te vāṅmātreṇāpi sahāya tvaṃ na kariṣyati (nobody will make you his friend only on account of your voice), R. 3, 71, 22 nāmamātraṃ tu jānāmi na rūpaṃ tasya rakṣasaḥ. This translation, however, does not suit all instances. Sometimes -mātram signifies, that the whole class is meant, not single individuals belonging to it. Pat. I, p. 242 brāhmaṇo na hantavyaḥ surā na peyeti brāhmaṇamātraṃ na hanyate surāmātraṃ ca na pīyate (since it is said: »one may not hurt a brahman, nor drink strong liquor”, one does not hurt anybody, that is named brahman, nor drinks anything; to which the appellation »strong liquor” is applicable). Comm. on R. 2, 12, 100 kaikeyīdaurjanyaṃ dṛṣṭvā rājā śokena strīmātraṃ nindati (the king seeing the foul conduct of K., by his sorrow chides the whole feminine sex).

-jātam

— In this meaning -jātam is almost synonymous, as Daç. 22 cauḍopanayanādisaṃskārajātam alabhta (he obtained the cauda, the upanayana, in short the whole set of sacraments), Bhoj. 62 akhileṣvapikośeṣu yadvittajātam asti tatsarvaṃ devena kavibhyo dattam.

b) Adjectt. in -mātra.

b.) -mātra is also put to participles; then it is an adjective and signifies »as soon as —.” Panc. III, 3 jātamātraṃ śatruṃ praśamaṃ nayet (one must abate a foe, as soon as he has arisen), ibid. p. 58 bhagavatā nārāyaṇena smṛtamātro vainateyaḥ saṃprāptaḥ, Kathâs. 36, 111, etc.

Rem. The adjectival employment of those in -mātra is however not restricted to the case that the former member is a participle. See but Panc. II, 95 nāmamātrā na siddhau hi dhanahīnāḥ… narāḥ (poor people do but bear the name of men, as they are of no use whatsoever).

5. -kalpa, -deśīya, -prāya.

5. »Almost, nearly, like” is signified by -kalpa, -dezya, -deśīya, -prāya, which have almost got the nature of pure formal suffixes, and, indeed, the former three are taught as such by Pâṇini (5, 3, 67). Of them, those in -kalpa and -prāya are the most frequent. R. 3, 16, 39 vākyāny amṛtakalpāni (speech like ambrosia), Kumâras. 3, 14 kāryaṃ tvayā naḥ pratipannakalpam (you have nearly engaged yourself to do our affair), Kathâs. 6, 51 chandogaḥ kaścid ity ukto viṭaprāyeṇa kenacit (some Sâmavedin was thus addressed by somebody like a rake), Mâlat. IX, p. 149 evaṃ paryavasitaprāyaiva naḥ pratyāśā (in this manner all my hope is almost gone), Panc. 202 pakvaśāliprāyaṃ deśaṃ gataḥ (gone to a country, where a good deal of the paddy was ripe), Daç. 78 vipralambhaprāyam idam adharmavartma (this way of unrighteousneas, full of deception). Those in -prāya are, indeed, bahuvrîhis, to be analyzed thus: »the greater part of which is —,” just as those in -kalpa properly are = »the manner or mode of which is —.”

6. -rūpa.

6. Those in -rūpa are likewise adjectives. As -rūpa means not only »shape, form” in general, but also »a beautiful shape, a beauty,” so the bahuvrîhis ending in it admit of either acceptation. Pâṇini (5, 3, 66) mentions the latter, when teaching such compounds as brāhmaṇarūpaḥ, pacatirūpam to be praise-denoting. (*2) But, in practice, those in -rūpa are not often met with in this meaning, by far oftener they are employed for the sake of qualifying some general kind by describing its species. Then we may often translate them by means of »f. i.; viz.” Instances are chiefly found in commentaries and the like. Sây. on Ait. Br. 2, 37, 1 [p. 272 ed. Aufrecht] rathabhaṅgarūpo vyāmohaḥ.

7. -avadhi.

7. Bahuvrîhis in -avadhi (limit) may be synonymous of yāvat in both of its acceptations (169 with Rem.). See f. i. Kathâs. 4, 100 pratyāgamāvadhi (till I shall have come back); 52, 146.

8. -viśeṣa.

8. Tatpurushas in -viśeṣa, when meaning »excellent species,” are expressive of something »first-rate, excellent.” See f. i. Ragh. 2, 7.

9. -antaram.

9. Tatpurushas in -antaram are often to be rendered by »some” or »other.” Properly speaking, antaram, viśeṣaḥ and bhedaḥ mean »variety, species,” and as a »variety of something” is »something different”, the transition of meaning may easily be accounted for. — Mâlav. III, p. 60 pūrvasmādavasthāntaram upārūṭhā tatrabhavatī (v. a. the lady has changed her former attitude to another), Panc. I, 132 śubhāśubhaphalaṃ sadyo nṛpād daivād bhavāntare (the fruit of good and evil deeds comes instantly, when from the king, but in some other existence, when from Destiny), ibid. p. 83 laghunām api durjanānāṃ madhye vastuṃ na śakyate / upāyāntaraṃ vidhāya te nūnaṃ ghnanti (one cannot dwell among wicked people, for they will hurt you by some means or other). The proper meaning of -antaram is not rarely transparent, as in the example quoted first. Likewise Panc. 248 bhartānyamārgāntareṇāgatya, Panc. 205 yuvayorbhāṣyāntaraṃ samyaṅ na śṛṇomi (I do not hear distinctly, of what kind of things you speak).

10. -vidhiḥ.

10. Tatpurushas in -vidhiḥ may denote, that the action spoken of is done »in due form.” Panc. I, 335 pradānavidhinā… saṃprāpyate phalam, Daç. 80 tadagāre ‘tyudārambhyavahāravidhimakaravam (in his house I gave a stylish dinner).

This list may be easily enlarged.

(*1)
One is wont to analyze bhūtapūrvaḥ by pūrvaṃ bhūtaḥ and so on, see f. i. Kâç. on P. 5, 3, 53, but that analysis does not give a satisfactory account of the nature of the compound. If the adverb pūrvaṃ were compounded with the noun bhūta, one would expect pūrvabhūta in the same way, as f. i. M. 9, 267 pūrvataskarāḥ »people who have formerly been thieves.”

(*2)
Pâṇini speaks of -rūpa as of a taddhita. Kâç. when commenting on our sûtra shows -rūpa to be used to signify the highest pitch of a quality, as vṛṣalarūpo’yaṃ palāṇḍunā surāṃ pibati. Blame, inversely, is expressed by compounds in -pāśa (P. 5, 3, 47), as in this verse of Bhojapr. (p. 7) svāmy ukte yo na yatate sa bhṛtyo bhṛyapāśakaḥ (the attendant, who does not exert himself, when ordered hy his master, is a bad attendaut).

229. 承前

これらを分けて説明する。

1. -ādi etc.

1. -ādi、-ādika、-ādya、-prabhṛtiは、正確には「…で始まる」を意味するbabuvrīhiである。
M. 1, 50 etadantās tu gatayo brahmādyāḥ (the existences, at the head of which stands Brahmâ, end here)

ふつうこれらは「…など」を表す。
Panc. 8 tasyaurdhvadehikakriyā vṛṣotsargādikāḥ sarvāścakāra (be performed in his honour all the funeral rites, viz. the vṛshotsarga etc.)
Hit. 123 suvarṇavastrādi prasādadānaṃ kriyatām (bestow on them presents of the king’s favour, gold, clothes and so on)
Panc. 62 sarve matsyakacchapaprabhṛtayastaṃ papracchuḥ (all [aquatic animals], fishes, tortoises etc. —)

これらの例での複合語は形容詞であるが、しばしばこれらは実名詞としても用いられる（222-注1をみよ）。
Bhoj. 64 aham api brāhmaṇaputraḥ / tvāmatra śayānaṃ vīkṣya kamaṇḍalūpavītādibhir brāhmaṇaṃ jñātvā bhavadāstarāsanna evāhaṃ prasuptaḥ
Panc. 27 adyaprabhṛti prasādanigrahādikaṃ tvayaiva kāryam (from to-day it is you by whom favour, punishment, etc. are to be administered)

2. -pūrva・-puraḥsara

2. 形容詞の場合、-pūrvaおよび-puraḥsaraの複合語は-ādi等で始まるものと同じはたらきをする。
Panc. 20 vyāghradvīpivṛkapuraḥsarāḥ sarve (all of them, tigers, panters, wolves and the rest)

副詞の場合、後分はほとんど意味を持たない：prītipūrvaṃ bhāṣate = saprīti bhāṣate。
Hit. 7 tasya bahumānapuraḥsaraṃ putrān samarpitavān (respectfully he gave over his sons to him)

3. bhūtapūrva等

3. pūrvaの複合語は「以前は…であった」がそうではなくなったものを表す：āḍhyapūrvaḥ（かつて富んでいた者）、bhūtapūrvaḥ（老いた）。
N. 1, 13 na kvacid dṛṣṭapūrvā (never seen before)
Çâk, VI satyamūḍhapūrvā tatrabhavatī rahasi śakuntalā (indeed, I once have wedded secretly the Lady Çak.)（*1）

4. -mātram

4. a.) n.の実名詞として用いられる-mātramの複合語は、正確には「…の正確な長さ（mātrā）は…」を表す。けれども、原則として、これらはその後分が何らかの限定不変化辞であるものとして用いられ、-mātramは「…だけ、…のみ」で翻訳される。
Prabodh. I, p. 13 kiṃ vadantīmātram etat (it is but a vain rumour)
Panc. 192 ko’pi te vāṅmātreṇāpi sahāya tvaṃ na kariṣyati (nobody will make you his friend only on account of your voice)
R. 3, 71, 22 nāmamātraṃ tu jānāmi na rūpaṃ tasya rakṣasaḥ

しかしながら、この翻訳はすべての用例に適するわけではない。-mātramは、あるクラスに属する個物ではなく、クラス全体を表すことがある。
Pat. I, p. 242 brāhmaṇo na hantavyaḥ surā na peyeti brāhmaṇamātraṃ na hanyate surāmātraṃ ca na pīyate (since it is said: »one may not hurt a brahman, nor drink strong liquor”, one does not hurt anybody, that is named brahman, nor drinks anything; to which the appellation »strong liquor” is applicable)
Comm. on R. 2, 12, 100 kaikeyīdaurjanyaṃ dṛṣṭvā rājā śokena strīmātraṃ nindati (the king seeing the foul conduct of K., by his sorrow chides the whole feminine sex)

-jātam

—この意味だと-jātamがほとんど同義語である。
Daç. 22 cauḍopanayanādisaṃskārajātam alabhta (he obtained the cauda, the upanayana, in short the whole set of sacraments)
Bhoj. 62 akhileṣvapikośeṣu yadvittajātam asti tatsarvaṃ devena kavibhyo dattam.

b.) -mātraの形容詞

b.) -mātraは分詞にも付加されうる；そのとき、複合語は形容詞であり「…するやいなや」を表す。
Panc. III, 3 jātamātraṃ śatruṃ praśamaṃ nayet (one must abate a foe, as soon as he has arisen)
ibid. p. 58 bhagavatā nārāyaṇena smṛtamātro vainateyaḥ saṃprāptaḥ
Kathâs. 36, 111, etc.

【補足】
-mātra複合語の形容詞的用法は、しかしながら、前分が分詞の場合に限定されない。
Panc. II, 95 nāmamātrā na siddhau hi dhanahīnāḥ… narāḥ (poor people do but bear the name of men, as they are of no use whatsoever)-

5. -kalpa・-deśīya・-prāya

5. 「ほとんど、大体」は、純粋で正式な接尾辞の性質をほとんど持つ-kalpa、-dezya、-deśīya、-prāyaによっても表される。後ろ3つはPāṇiniが説いている（5, 3, 67）。これらのうち、-kalpaと-prāyaが最も頻出する。
R. 3, 16, 39 vākyāny amṛtakalpāni (speech like ambrosia)
Kumâras. 3, 14 kāryaṃ tvayā naḥ pratipannakalpam (you have nearly engaged yourself to do our affair)
Kathâs. 6, 51 chandogaḥ kaścid ity ukto viṭaprāyeṇa kenacit (some Sâmavedin was thus addressed by somebody like a rake)
Mâlat. IX, p. 149 evaṃ paryavasitaprāyaiva naḥ pratyāśā (in this manner all my hope is almost gone)
Panc. 202 pakvaśāliprāyaṃ deśaṃ gataḥ (gone to a country, where a good deal of the paddy was ripe)
Daç. 78 vipralambhaprāyam idam adharmavartma (this way of unrighteousneas, full of deception)

-kalpaが正確には「…の仕方は…」であるように、-prāyaは以下のように分析される：「…の大部分は…」

6. -rūpa

6. -rūpaは同じく形容詞である。これはふつう「形、外形」だけでなく「美貌、美」も意味するので、これで終わるbahuvrīhiはどちらの解釈をも許す。Pāṇiniは、brāhmaṇarūpaḥ、pacatirūpamのような称賛を表す複合語を教示する際に後者に言及している（5, 3, 66）（*2）。けれども、実際には、-rūpaの複合語は この意味ではあまり用いられておらず、〔下位カテゴリの〕種（species）を記述することで〔その上位にある〕普遍的な種類（kind）を修飾するために使われることが多い。その場合、しばしば「例えば、すなわち」で翻訳される。用例は主に注釈書などで見られる。
Sây. on Ait. Br. 2, 37, 1 [p. 272 ed. Aufrecht] rathabhaṅgarūpo vyāmohaḥ.

7. -avadhi

7. -avadhi（期限）で終わるbahuvrīhiは、その解釈においてyāvatと同義語である（169、補足）。
Kathâs. 4, 100 pratyāgamāvadhi (till I shall have come back)
ibid. 52, 146.

8. -viśeṣa

8. -viśeṣaのtatpuruṣaは、「優れた類の」を表す場合、「第一等の、…に優れた」を表す。例としてはRagh. 2, 7をみよ。

9. -antaram

9. -antaramのtatpuruṣaはしばしば「或る」や「他の」で翻訳される。正確に言うと、antaram、viśeṣaḥおよびbhedaḥは「種類、種」を表し、そして「某の種類」は「種々の某」となる、というように意味の変遷は容易に説明できる。
Mâlav. III, p. 60 pūrvasmādavasthāntaram upārūṭhā tatrabhavatī (v. a. the lady has changed her former attitude to another)
Panc. I, 132 śubhāśubhaphalaṃ sadyo nṛpād daivād bhavāntare (the fruit of good and evil deeds comes instantly, when from the king, but in some other existence, when from Destiny)
ibid. p. 83 laghunām api durjanānāṃ madhye vastuṃ na śakyate / upāyāntaraṃ vidhāya te nūnaṃ ghnanti (one cannot dwell among wicked people, for they will hurt you by some means or other)

最初の例のように、-antaramの正確な意味が明らかなことは稀ではない。同様に：
Panc. 248 bhartānyamārgāntareṇāgatya
Panc. 205 yuvayorbhāṣyāntaraṃ samyaṅ na śṛṇomi (I do not hear distinctly, of what kind of things you speak)

10. -vidhiḥ

10. -vidhiḥのtatpuruṣaは、それについて話される行為が「正式な形で」終わっていることを表す。
Panc. I, 335 pradānavidhinā… saṃprāpyate phalam
Daç. 80 tadagāre ‘tyudārambhyavahāravidhimakaravam (in his house I gave a stylish dinner)

このリストは容易に増広しうる。

（*1）
bhūtapūrvaḥをpūrvaṃ bhūtaḥ…で分析することはできない。Kâç. on P. 5, 3, 53をみよ。そのような分析は、複合語の性質について十分な説明を与えるものではない。副詞pūrvaṃがbhūtaと複合しているとしても、以下と同様にpūrvabhūtaとなるであろう：
M. 9, 267 pūrvataskarāḥ »people who have formerly been thieves.”

（*2）
Pāṇiniはtaddhita接尾辞の-rūpaに言及している。Kâç.はそのスートラ（5, 3, 66）に注釈する際、性質の最上なることを表すために用いられる-rūpaを提示している：vṛṣalarūpo ’yaṃ palāṇḍunā surāṃ pibati。反対に、誹謗は-pāśaで表される（P. 5, 3, 47）。
Bhojapr. (p. 7) svāmy ukte yo na yatate sa bhṛtyo bhṛyapāśakaḥ (the attendant, who does not exert himself, when ordered hy his master, is a bad attendaut)

230. Intricate compounds.

Any Sanskrit compound belongs to one of the great classes mentioned before. Now, as not only the members of a compound but even their constituent elements may he compounds themselves, hence arises an almost unlimited freedom of enlarging compounds by taking up into them all sorts of nouns or adverbs serving to qualify the whole of the compound or part of it. In this way, very large and very intricate compounds are available, and in fact they often occur, albeit that the field of combinations and images is in some degree limited by conventional usage and by the examples of the best authors. For the rest the frequency and the nature of those intricate and bulky compounds will much depend on the style of the literary work. It requires, therefore, a good deal of training to catch forthwith the purport of many an intricate compound.

A few instances will suffice. Kâdamb. I, p. 15 the king, it is said, saw a lady kupitaharahutāśanadahyamānamadanadhūmamalinīkṛtāmiva ratim »who was like Rati, stained by the smoke of Kâma burning by the fire of angry Çiva,” for when analyzing the complex, we get kupitasya harasya hutāśanena dahyamānasya madanasya yo dhūmo ’bhūt tena malinīkṛtām, apparently a tatpurusha, the former member of which is also a tatpurusha the former member of which is also a tatp. and so on. Now a bahuvrîhi. In the same Kâd. (p. 39) a forest araṇyam bears the epithet daśarathasutaniśitaśaranikaranipātanihatarajanīcarabalabahularudhirasiktamūlam (where the roots [of the trees] had been moistened by the abundant blood of the army of the Râxasas killed by the shots of the crowd of sharp arrows [discharged] by the son of Daçaratha), here mūlam is the subj. of the bahuvrîhi, the preceding complex being its predicate, an intricate tatpurusha, as it is thus to be analyzed daśarathasya sutasya niśitānāṃ śarāṇāṃ nikarasya nipātair yan nihataṃ rajanīcarāṇāṃ balaṃ tasya bahulena rudhireṇa siktam. This whole clause is comprehended within one compound. And so often.

230. 複雑な複合語

サンスクリットの複合語はすべて先述の大分類のどれかに属する。複合語の構成要素だけでなく、その構成要素自体も複合語でありうるために、複合語の全体または一部を修飾するためのあらゆる種類の名詞や副詞を複合語の中に取り込んで、複合語を無制限に自由に拡張することができる。このようにして長大で複雑な複合語が使用可能であり、組み合わせやイメージの領域は従来の使用法や模範的作者の用例によってある程度制限されているにもかかわらず、これは実際にはしばしば発生する。それらの複雑で扱いにくい複合語の頻度や性質は、文学作品の文体に大きく依存する。したがって、多くの複雑な複合語の趣旨をすぐに理解するには、かなりの修練が必要とされる。

数例で十分であろう。
Kâdamb. I, p. 15 the king, it is said, saw a lady kupitaharahutāśanadahyamānamadanadhūmamalinīkṛtāmiva ratim »who was like Rati, stained by the smoke of Kâma burning by the fire of angry Çiva”

分析すると、見たところはTp.で、前分もTp.、その前分も…（略）のkupitasya harasya hutāśanena dahyamānasya madanasya yo dhūmo ’bhūt tena malinīkṛtāmが得られる。bahuvrīhiは以下：
Kâd. (p. 39) a forest araṇyam bears the epithet daśarathasutaniśitaśaranikaranipātanihatarajanīcarabalabahularudhirasiktamūlam (where the roots [of the trees] had been moistened by the abundant blood of the army of the Râxasas killed by the shots of the crowd of sharp arrows [discharged] by the son of Daçaratha)

ここでmūlamはbahuvrīhiの主語、その前の複合語はその述語の複雑なTp.である。よって以下のように分析される：daśarathasya sutasya niśitānāṃ śarāṇāṃ nikarasya nipātair yan nihataṃ rajanīcarāṇāṃ balaṃ tasya bahulena rudhireṇa siktam。この節全体が一つの複合語に内包されている。

231. Case-nouns standing outside the compound, that to be construed with it.

Case-nouns standing outside the compound are very often to be construed with it or with one of its members. This is but consistent with the whole spirit, which pervades Sanskrit composition. A great liberty is left to the speaker to prefer either a rather synthetical or a rather analytical mode of expression. He has the opportunity of enlarging compounds by making enter within them any noun or adverb serving to qualify the whole or any part of it. But on the other hand nothing compels him to do so. The qualifying noun may as well be a self-existent word having its own noun-case. (*1) So Çâk. V himavato girer upatyakāraṇyavāsinaḥ (dwelling in the forests on the slope of mount Himavân), here himavato gireḥ stands outside the compound as to its form, but belongs to it by its meaning, as it qualifies the member upatyakā. Panc. 42 a weaver returns home to his wife, prāg eva karṇaparaṃparayā tasyāḥ śrutāpavādaḥ »having heard evil report on her account;” when using a mere analytical expression, the author of the Pancatantra would have said tasyā apavādaṃ śrutvā or śrute tasyā apavāde etc., when a mere synthetical one śrutatadapavādaḥ, but he has here availed himself of a mixed idiom. — Mâlav. V, p. 140 sagaraḥ pautreṇa pratyāhṛtāśvaḥ (S., who had the horse brought back by his grand-son); Kumâras. I, 37 etāvatā nanv anumeyaśobhi… [asyā] aṅkam, here etāvatā qualifies anumeya the former member of the compound; Mahâv. I, p. 6 prakṛtyā puṇyalakṣmīkau kāveta, here prakṛtyā also is intimately connected with the compound. These few examples will suffice, as the idiom is met with on almost every page of Sanskrit.

(*1)
I wonder, what reasons may have induced Whitney (§ 1316) to speak of this idiom as something irregular. On the contrary, nothing can be more regular.

231. 複合語に係る名詞-格

複合語の外にある名詞-格は頻繁に、その複合語か、あるいはその要素の一つに係る。これはサンスクリットの作文法に浸透している精神と全く調和している。合成的な表現法を好むか、分析的な表現法を好むかは、話者に大きな自由が委ねられている。話者は、複合語全体あるいはその一部分を修飾することに資する任意の名詞・副詞を挿入することで、複合語を拡張できる。しかし一方で、それは強制されない。 修飾する名詞は独自の名詞-格を持つ自立した単語である場合もある（*1）。
Çâk. V himavato girer upatyakāraṇyavāsinaḥ (dwelling in the forests on the slope of mount Himavân)

ここでhimavato gireḥは、その形式に関して複合語の外にあるが、これはupatyakāを修飾するので、意味としては複合語に属する。
Panc. 42 a weaver returns home to his wife, prāg eva karṇaparaṃparayā tasyāḥ śrutāpavādaḥ »having heard evil report on her account”

単に分析的表現を用いるなら、Pancatantraの作者はtasyā apavādaṃ śrutvāやśrute tasyā apavāde等と言っている。合成的表現を用いるならśrutatadapavādaḥとなるが、作者は〔両者が〕混合したイディオムを用いている。
Mâlav. V, p. 140 sagaraḥ pautreṇa pratyāhṛtāśvaḥ (S., who had the horse brought back by his grand-son)
Kumâras. I, 37 etāvatā nanv anumeyaśobhi… [asyā] aṅkam

ここでetāvatāは複合語の前分のanumeyaを修飾する。
Mahâv. I, p. 6 prakṛtyā puṇyalakṣmīkau kāveta

ここではprakṛtyāも複合語と複雑に接続されている。これらはサンスクリット文献の至る所で見られるので、数例で十分である。

（*1）
Whitney（§1316）がこのイディオムを不規則なものとして言及する理由は何であろうか？ むしろこれほど規則的なものはない。

232.

By this equivalence, and to a certain extent also, promiscuousness of analytical and synthetical expression it is also explained that there must be some freedom in using the so called figure of ellipsis even in compounds. Nala. 1, 13 the beauty of Damayantî is said to surpass that of all other women, even of time past, na… tādṛgrūpavatī kvacit… dṛṣṭapūrvāthavā śrutā, here śruta is of course = śrutapūrvā. By a similar abridgment Malâv. V, p. 137 moon and sun are named śītoṣṇakiraṇau »the hot- and the cold-rayed (*1).”

(*1)
A striking example is afforded by R. 3, 20, 12, if I am right reading there rākṣasās te caturdaśa / ūcur vācaṃ susaṃkruddhā brahmaghnāḥ śūlapāṇayaḥ / saṃraktanayanā ghorā rāmaṃ saṃraktalocanaṃ / paruṣā madhurābhāṣaṃ dṛṣṭādṛṣṭaparākramam; the Bomb. edition has hṛṣṭādṛṣṭapa-. There is antithesis between the paruṣāḥ [in full parākramāḥ] Râxasas and Râma madhurābhāṣaḥ, and likewise between them dṛṣṭa[parākramāḥ] and Râma who was adṛṣṭaparākramaḥ. The saṃdhir ârshaḥ dṛṣṭā[ḥ] [a]dṛṣṭaparākramam is admitted in the Râmâyana, see f. i. 2, 51, 8; 74, 13; 3, 64, 23.

232. 複合語における要素省略

この同等性によって、そしてある程度までの分析的・合成的表現の乱雑性によって、複合語においてさえ、いわゆる省略形を用いることにある程度の自由がなければならないことも説明される。
Nala. 1, 13 the beauty of Damayantî is said to surpass that of all other women, even of time past, na… tādṛgrūpavatī kvacit… dṛṣṭapūrvāthavā śrutā

ここでのśrutaはもちろんśrutapūrvāである。同様の省略は以下など：
Malâv. V, p. 137 moon and sun are named śītoṣṇakiraṇau »the hot- and the cold-rayed” （*1）

（*1）
〔複合語に適用される省略の〕好例はR. 3, 20, 12。私の読みが正しければrākṣasās te caturdaśa / ūcur vācaṃ susaṃkruddhā brahmaghnāḥ śūlapāṇayaḥ / saṃraktanayanā ghorā rāmaṃ saṃraktalocanaṃ / paruṣā madhurābhāṣaṃ dṛṣṭādṛṣṭaparākramamである（ボンベイ版ではhṛṣṭādṛṣṭapa-）。

paruṣāḥ [parākramāḥ]のRāxasaらとRāma（madhurābhāṣaḥ）、また同様にdṛṣṭa[parākramāḥ]とadṛṣṭaparākramaḥであるRāmaとの間には対照法がある。〔ヴェーダのsaṃhitā読みのように解釈されるべき〕作者によるsandhi（saṃdhir ārṣaḥ）であるdṛṣṭā[ḥ] [a]dṛṣṭaparākramamがRāmāyaṇa中に認められる。ibid. 2, 51, 8、2, 74, 13、3, 64, 23をみよ。