The picture at the bottom is taken from Introduction To Sanskrit by Egenes Thomas, in which we see that, if a vowel different from a or aa precedes the visarga, and if the initial lettre of the following word is r, then the visarge disappears and the preceding i or u is lengthened.
But, from A Sanskrit Grammar For Students by MacDonell, A we find the statement that, in the above situation, the visarga should be changed to r, as r is a soft consonant.
Therefore I am confused as to which rule is correct, and post here to know what others think about it.
This is my first post here; tell me if any inappropriate points occur, thanks in advance.

enter image description here


I think the confusion comes from using prepausal forms as the starting point. Both /s/ and /r/ become visarga pre-pausally, so Macdonnell treats /r/ as an 'etymological' exception. You get /punar api/ → [punar api/ but /devas api/ → [devo 'pi]. If the preceding vowel is /a, ā/ and the following segment is voiced, then /r/ remains [r] (except before /r/ where it is deleted). But /as/ is treated the same as /o/ (you could say that is becomes o, then is further changed to [a] with hiatus before a vowel, except before /a/). And /ās/ becomes [ā] before any voiced segment so /aśvās vahanti/ → aśvā vahanti (but /ār/ is unchanged, viz. ahār dāmnā). Whitney's treatment in his grammar is clearer, IMO.


To clarify, "visarga" is not an underlying segment of Sanskrit, is is phonetically voiceless [h] transcribed , and derives from /r,s/. Theoretically, there could be two distinct underlying forms, /grāmayos rāmas/ or /grāmayor rāmas/. In either case, the surface outcome would be [grāmayo rāmaḥ].

The standard reference grammar of Sanskrit is Sanskrit Grammar, by William Dwight Whitney.

  • Thanks for explaning some rules, but you didn't cover the case I am asking: for example, what would this imaginary combination be after the sandhi rules: /grāmayoh. rāmah./, where h. designates the visarga? Also, might I know the name of Whitney's grammatical treatment? Thanks again.
    – awllower
    Apr 19 '15 at 4:05
  • Thanks a lot for clarifying the doubts, and thanks for the link.
    – awllower
    Apr 19 '15 at 15:09
  • Be warned that Whitney's Sanskrit Grammar is very opinionated and full of idiosyncrasies. It diverges from the native grammatical tradition in many many ways. May 2 '15 at 19:01
  • 1
    True, but when one hopes to learn the sandhi rules, it's kind of hard to make sense of the Aṣṭādhyāyī.
    – user6726
    May 2 '15 at 22:13

Most English books on sanskrit attempt to rephrase Panini's grammar rules and in doing so, the conciseness of the rule is lost. An additional side-effect is that it takes several sentences to explain the application of rules. This is the natural outcome of all translations - I am not finding fault with them.

Your question about the elongation of अ/इ/उ when र् is elided (or made to disappear) is governed by two rules: 8-3-14 रो रि and 6-3-111 ढ्रलोपे पूर्वस्य दीर्घोऽणः

Take for example हरिः राजते meaning Hari is resplendent. The derivation always starts with

हरिस् + राजते

The suffix (or प्रत्ययः) for the nominative singular (प्रथमा-विभक्तिः एकवचनम्) is सुँ The उँ in सुँ gets elided due to 1-3-2 उपदेशेऽजनुनासिक इत् and 1-3-9 तस्य लोपः Continuing with the derivation,

हरिरुँ + राजते 

Any स् at the end of a पदम् (word, defined by 1-4-14 सुप्तिङन्तं पदम्) is replaced with रुँ by the rule 8-2-66 ससजुषो रुः

As before, the उँ in रुँ drops and we now have

हरिर् + राजते

Now 8-3-14 रो रि mandates elision (called लोपः) to a र् as another र् follows. This results in

हरि + राजते

Finally, whenever the elision of ढ् or र् happens, the previous अण् (i.e अ इ उ) gets elongated (called दीर्घः) So, after performing sandhi, the resulting word appears as

हरी राजते

The second rule you mentioned from MacDonell isn't in conflict with the above. 8-3-15 खरवसानयोर्विसर्जनीयः asks to replace the र् at the end of a पदम् with विसर्गः as any letter from खर् (ख फ छ ठ थ च ट त क प) follows. As you can see, र् is not in this list of letters. An example of this rule would be as follows:

रामर् + चिनोति (meaning Rama collects)

Note that the र् at the end of राम comes from the application of 8-2-66 ससजुषो रुः and two more rules, as explained above.

रामः + चिनोति

Further to this rule, the विसर्गः is replaced with स् by 8-3-34 विसर्जनीयस्य सः giving rise to

रामस् + चिनोति

Finally, the स् is replaced with स् due to 8-4-40 स्तोः श्चुना श्चुः yielding

  • Thanks for such an expounded clarification; it is quite surprising to find that the rules listed in the introductory books are in fact composites of some more basic rules. Thanks again.
    – awllower
    Apr 22 '15 at 23:46
  • 2
    If you are interested in sanskrit grammar, I highly recommend following avg-sanskrit.org - there are copious examples that demonstrate application of rules (and also video recordings of classes). You can leave us a note in the comments section if you have questions.
    – linuxfan
    Apr 23 '15 at 0:02
  • OK, I am trying to see what I can find in that site. Thanks for the recommendation. Best regards here. :)
    – awllower
    Apr 23 '15 at 0:07
  • Why do you have so many chandrabindus in your answer (e.g. सुँ instead of सु)? Generally in Devanagari the chandrabindu is used to indicate nasalization, but I get the impression you're using it for some other (what?) purpose. May 3 '15 at 16:39
  • 1
    Yes the chandrabindu means anunasika i.e nazalised vowel. Many of the pratyayas (suffixes) in Panini's grammar have nazalised vowels. This allows us to remove them by उपदेशेऽजनुसासिक इत् Hence, although the pratyaya is सुँ, what remains after removing the nazalised अच् is स् Similarly in the case of रुँ etc.
    – linuxfan
    May 4 '15 at 17:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.