Rule. The s at the end of a pada is replaced by ru. So is the final ṣ of a pada that ends in the word sajuṣ. ru is the sound r carrying the marker u; the marker is an it, also called an anubandha, and is never pronounced. What the marker is for comes out at SK 166.
Example.
śivas (the word śiva with its nominative-singular ending) → its final s stands at the end of a pada → s is replaced by ru → śivar(u), which we write śivar.
sajuṣ (nominative singular) → the pada ends in the word sajuṣ, so its final ṣ → ru → sajur(u); the finished nominative singular is sajūḥ.
sajuṣau, sajuṣaḥ → the ṣ is not at the end of a pada → no ru; the ṣ stays.
parama-sajūḥ → the pada ends in the word sajuṣ → ru applies here too, by the tadanta reading.
Applies when / blocked by. The rule sits under the adhikāra padasya, "of a pada", and sa in the sūtra means the sound s (the a is only for pronunciation). So the rule reads "of the pada ending in s, and of the pada ending in sajuṣ"; the substitute then lands on the last sound only. The rule is an apavāda to jaśtva. Without that status, śivas would come out as śivad: jaśtva is the earlier Tripādī rule, so by SK 12 the later ru-rule ought to be invisible to it. But the ru-rule has no case of its own outside jaśtva's territory — every pada-final s is a jhal that jaśtva would take. A rule with no scope of its own (niravakāśa) must be the apavāda of the rule that covers all its cases, and an apavāda is not treated as invisible; it applies by the sheer authority of its being stated. Bhaṭṭoji's commentators cite the Bhāṣya for this.
Watch out. The ru is not yet a sound you can hear. It is a marked r, and every rule in this chapter that says roḥ ("of the ru") is looking for this marked r and nothing else.
For the s at the end of a pada, and for the word sajuṣ, ru is [substituted]. [This is] an apavāda (exception) to jaśtva.
sasajuṣo ruḥ — The word-division (cheda) is "sasajuṣaḥ ru"; the [visarga] r is [explained by] elision of the r by "ro ri." The analysis (vigraha) is: "sa and sajuṣ = sasajuṣau [dvandva dual], of those two (tayoḥ)." In the operation ru, the u is an it-marker (anubandha); its result (fruit) will be stated just below. By "sa" the sound s (sakāra) is intended; the a [in "sa"] is for the sake of pronunciation. The governing adhikāra "padasya" is qualified by "s" and by the word "sajuṣ"; hence the tadanta-vidhi [applies]. Meaning: the pada that ends in s and [the pada] that ends in the word sajuṣ — for that, ru is [substituted]; and by "alo 'ntyasya" it [applies] to the final sound. He states the upshot with "for the s at the end of a pada."
On "and for the word sajuṣ": the meaning is, of the ṣ at the end of the pada that ends in the word sajuṣ. And so, in "sajuṣau, sajuṣaḥ," there is no ru-substitution for the ṣ, because it is not at the end of a pada. By the tadanta-vidhi "the pada that ends in the word sajuṣ," there is no under-application (avyāpti) in "parama-sajūḥ"; nor should under-application be suspected in "sajūḥ" [alone], since by vyapadeśivadbhāva it is [treated as] ending in that [word sajuṣ]. The two paribhāṣās (paribhāṣā) — "vyapadeśivadbhāvo 'prātipadikena" and "iti-grahaṇavatā prātipadikena tadantavidhir na" — have as their sphere [the maxim] "pratyayagrahaṇe yasmāt...", not the sphere of "yena vidhis tadantasya"; this is clear in the Bhāṣya on the sūtra "asamāse niṣkādibhyaḥ."
Objection: in "śivas," the s would become d by jaśtva [via "jhalāṃ jaśo 'nte"], since ru-substitution, though para (later) relative to jaśtva, is asiddha (regarded as not having taken effect) with respect to it. To this he replies, "[it is] an apavāda to jaśtva": and so, since ru-substitution has no [other] scope (niravakāśa), there is no asiddhatva — this is the idea. This is stated in the Bhāṣya: "in [the domain of] pūrvatrāsiddham there is no vipratiṣedha, because of the non-existence of the later [rule's having taken effect]," and "an apavāda [prevails] by the authority of [its explicit] statement (vacana-prāmāṇya)."
- textual The digital source duplicates one sentence ('sajuṣ-śabdāntaṃ yat padaṃ tadantasya ṣakārasya... pravṛtti') and garbles śivas as śivasi; the translation renders each substantive statement once, following the evident readings. Stray Vedic-accent marks serve as quotation delimiters.