rarely takes place; e. g. ánu rājati ṣṭúp (ix. 9618); tri sadhdstha (ill. 565); níḥ ṣadhásthāt[1] (v. 319); nū͏̂ ṣṭhirám (i. 6415); ádhi sṇúnā (ix. 9716).
The ṣ usually follows short words such as u, tu, nú, sú, hí, and occurs where there is a close syntactical connection of two words; e. g. rájaḥsu ṣidan (vii. 3416); váṃsu ṣídati (ix. 573); diví sán (vi. 26); diví sántu (v. 210); ṛcchánti ṣma (x. 1026). In no word, however, even when these conditions are fulfilled, is the change of initial s to ṣ invariably made.
a. In the later Saṃhitās, apart from passages adopted from the RV. this form of external Sandhi is very rare except in the combination ū͏̆ ṣú. Examples are ā͏̂d u ṣṭendm (AV. iv. 34); máhi ṣáÑd dyumdn ndmah (TS. m. 2. 82)[2]
58. The breathing h. — The sound h is, at the present day, pronounced as a breathing in India, and this was its character at the period when Greek and Indian words were interchanged, as is shown e. g. by ὤρα being reproduced by hōrā. It is already recognised as a breathing by the TPr. (ii. 9), which identifies it with the second element of voiced aspirates (g-h, d-h, b-h). This is borne out by the spelling ळह l-h (= ḍh) beside ळ ḷ (= ḍ). The TPr. further (ii. 47) assigns to it, on the authority of some, the same place of articulation as the following vowel, this being still characteristic of the pronunciation of h at the present day in India[3]. The breathing is, moreover, stated by the Prātiśākhyas[4] to have been voiced. This pronunciation is proved by the evidence of the Samhitās themselves; for h is here often derived from a voiced aspirate, e. g. hitá- from dhā- 'put'; it is occasionally replaced by a voiced aspirate, e. g. jaghā͏̂na from han- 'strike'; and in Sandhi initial h after a final mute regularly becomes a voiced aspirate, e. g. tád dhí for tád hí. It is in fact clear that whatever its origin (even when = IIr. źh)[5], h was always pronounced as a voiced breathing in the Sarnhitās.
As h cannot be final owing to its phonetic character[6], it is represented in that position by sounds connected with its origin[7]. It appears in combination with voiced sounds only; being preceded only by vowels, Anusvāra, or the semivowels r and l[8] (in Sandhi also by the nasals ṅ and n), and followed only by vowels, the nasals ṇ, n, m, or the semivowels y, r, l, v.
The breathing h as a rule represents a voiced aspirate, regularly a palatal aspirate, occasionally the dental dh and the labial bh. It usually represents a new palatal (= IIr. jh, Av. ȷ [9]), appearing beside gh, e. g. druhyú- : drógha-, as j beside g, e. g. ójiyas : ugrá-. But in many words it also stands (like j for ź) for the old palatal źh, the voiced aspirate of ś, being recognizable as such in the same way as j[10].
1 a. h represents the palatalization (= IIr.jh) of gh when, in cognate forms, gh (or g) is found before other sounds than s; e. g. hán-ti 'strikes' : ghn-ánti, jaghāna; árhati 'is worthy' : arghá- 'price'; rh-dhá- 'weak' : ragh-ú- 'light'; jámh-as- 'gait' :jáṅghā- 'leg'; dáhati 'burns' : dag-dhá- 'burnt'; dóhate 'milks' : dúghana- 'milking', dug-dhá- 'milked'; dudróha 'have injured' : drug-dhá-
'injured', drógha- 'injurious'; máṃhate 'presents' : maghá- 'gift'; míh- 'mist' :
- ↑ iḥ and uḥ produce the same effect as simple i and u, as they were originally pronounced as iṣ and uṣ (cp. Sandhi, p. 71, e 2); e. g. agní(ṣ) ṣṭave; yáju(ṣ) ṣkannám.
- ↑ 2 Cp. Wackernagel I. 207 b.
- ↑ Cp. Whitney on APr. I. 13 and TPr. II. 47.
- ↑ See RPr. I. 12; XII. 2; APr. I. 13.
- ↑ See below I b.
- ↑ See below, Sandhi, 66.
- ↑ Ibid, b 6, S.
- ↑ The combination lh is rare; it is found in upa-valh- (VS.) 'test by riddles'; vihálha- (AV.), of unknown meaning; and in the Kānva recension of the VS. as representing ḍh.
- ↑ Cp. 36. This h being related to gh as j is to g, it must represent IE. gh. This survives, with loss of the aspiration, in jahí (= IIr. jha-dhi), 2. sing. impv. of han- 'strike'.
- ↑ Cp. 41.
