Now (atha), for what purpose is the ā-sound (ākāra) made tapara (marked with a following consonant)?

The making of ā tapara is done.

What is the purpose (prayojana)?

For the sake of homogeneous sounds (savarṇa).

So that — by [the maxim] ‘a sound with a following t denotes those of its own duration (taparaḥ tatkālasya)’ — [only] the sounds of that same duration (tatkāla) may be taken.

Of which ones?

Of the udātta, the anudātta, and the svarita [varieties].

And what would be the reason [they] should not [be taken]?

The udātta and the rest (udāttādi) are differentiators (bhedaka).

But how is it known that the udātta and the rest are differentiators?

For thus it is seen in the world (loka):

One who, when an udātta is to be made, makes an anudātta — the schoolteacher (khaṇḍikopādhyāya) gives him a slap [saying,] ‘You are doing something else!’ — thus.

This is a [real] purpose.

Then what [more]? — thus.

It must [also] be stated: ‘because the quality (guṇa) is a differentiator (bhedaka)’.

What is the purpose?

Nasality (ānunāsikya), namely, is a quality (guṇa).

So that [the rule] may apply even to what is distinguished by it (tadbhinna) [i.e. by nasality].

And what would be the reason [it] should not [apply]?

Because the quality (guṇa) is a differentiator (bhedaka).

Qualities (guṇa) are differentiators (bhedaka).

But how is it known that qualities are differentiators?

For thus it is seen in the world:

This is one single substance (ātman), namely water (udaka).

Through difference of its quality (guṇa-bheda) there arises otherness (anyatva): ‘this cold [water] is one thing, this hot [water] another’ — thus.

But surely, sir (bhoḥ), qualities are also seen [to be] non-differentiators (abhedaka)!

That is, for instance:

Devadatta — whether shaven-headed (muṇḍin), or matted-haired (jaṭin), or wearing a topknot (śikhin) — does not give up his own name (ākhyā).

Likewise: a child (bāla), a youth (yuvan), an old man (vṛddha); a young calf (vatsa), a trainable bullock (damya), a full-grown ox (balīvarda) — thus.

Both these have been said of qualities: [that they are] differentiators, [that they are] non-differentiators — thus.

But which is the proper [view] (nyāyya) here?

Just ‘qualities are non-differentiators (abhedaka)’ is the proper [view].

Whence this?

Because he (the teacher) makes the mention of udātta in [the sūtra] ‘asthi-dadhi-sakthy-akṣṇām anaṅ udāttaḥ’ (the anaṅ [substitute] of asthi, dadhi, sakthi, akṣi is udātta).

If qualities were differentiators, he would [simply] have pronounced [it] as udātta [and no separate word ‘udātta’ would be needed].

If, then, qualities are non-differentiators, then whatever is stated [as applying] to a [stem] beginning with anudātta and ending in udātta (anudāttādi, antodātta) would [wrongly] extend also to one beginning with svarita and ending in svarita (svaritādi, svaritānta).

This is not a fault.

A quality that is expressly resorted to (āśrīyamāṇa) [as a condition in a rule] does become a differentiator (bhedaka).

That is, for instance:

‘One should seize a white [victim]’ (śuklam ālabheta).

‘One should seize a black [victim]’ (kṛṣṇam ālabheta).

There, one who, when a white [victim] is to be seized, seizes a black one — by him, indeed, what was enjoined (yathokta) is not done.

The letter t (takāra), then, is for the sake of removing doubt (asandeha).

When ‘aic’ is [merely] stated, there would be doubt.

Is it these two, ai and au (aicau), [alone], or is the ā-sound (ākāra) also indicated here? — thus [would be the doubt].

This is a mere doubt (sandeha-mātra).

And in all doubts this [maxim] stands by: ‘the specific determination [comes] from the [teacher's] explanation (vyākhyāna); for a rule is not annulled (alakṣaṇa) merely by [there being] doubt’ — thus.

We shall explain [it as] ‘the taking of the three (trayāṇām grahaṇam)’ — thus.

For elsewhere too, in doubts of this kind (evañjātīyaka), he (the teacher) makes no effort at all [to remove them].

That is, for instance:

‘auṭ aṃśasoḥ’ — thus. This, then, is the purpose: by proximity (āntarya), that the substitutes (ādeśa) of three-mora and four-mora originals (sthānin) should not [themselves] be three-mora and four-mora — thus: khaṭvā + indraḥ → khaṭvendraḥ; khaṭvā + udakam → khaṭvodakam; khaṭvā + īṣā → khaṭveṣā; khaṭvā + ūḍhā → khaṭvoḍhā; khaṭvā + elakā → khaṭvailakā; khaṭvā + odanaḥ → khaṭvaudanaḥ; khaṭvā + aitikāyanaḥ → khaṭvaitikāyanaḥ; khaṭvā + aupagavaḥ → khaṭvaupagavaḥ — thus.

Now, even when the t is made, why exactly do the substitutes of three-mora and four-mora originals (sthānin) not become three-mora and four-mora?

Because of the restriction (niyama) ‘a [sound] with a following t denotes [those] of its own duration (taparaḥ tatkālasya)’.

But surely (nanu), a ‘tapara’ is that after which t [follows] (taḥ paraḥ yasmāt) — that is this ‘tapara’.

‘No,’ he says.

‘Tapara’ [means] also ‘that which is subsequent even to a t’ (tāt api paraḥ) — thus.

If ‘tapara’ [means] also ‘that which comes even after a t’, then [the affix of the rule] ‘ṝdoḥ ap’ would apply only here:

yavaḥ, stavaḥ,

and would not apply here: ‘lavaḥ, pavaḥ’ — thus.

This is not a t (takāra).

What then [is it]?

A d (dakāra).

What is the purpose in [its being] a d (dakāra)?

Then what is the purpose in [its being] a t (takāra)?

If the t is for removing doubt (asandeha), so is the d as well.

Or if the t is for ease of pronunciation (mukha-sukha), so is the d as well.