It's often difficult to say "why" a sound change happened, so I'll focus on your other questions. Rhotacism in Latin happened via a series of sound changes. It only affected inherited *\*s* in specific environments. ##The reconstructed steps of rhotacism, and their conditions The first step towards rhotacism is believed to have been voicing of single "s" to the sound [z] in certain contexts: generally, between vowels. Phonologically, this change can be seen as an assimilation of the consonant to the surrounding voiced segments. This kind of voicing assimilation is common historically; for example, it also occured between Classical Latin and many Romance languages. Double *ss* remained unvoiced. In some related languages, such as Oscan, /z/ remained as such between vowels and did not undergo rhotacism. We know this because we have Oscan inscriptions written in the Latin alphabet that use the letter Z for this sound (Buck 1904). (However, my understanding is that we don't have evidence of the letter Z being used this way in Latin inscriptions: see the related question https://latin.stackexchange.com/questions/34/why-are-there-no-native-latin-words-with-a-z?lq=1) The next step of rhotacism (z > r) is a little less common across languages, but we do have some other examples. The process also applied historically in Germanic; one place where you can see the result in English is *forlo**r**n,* which is historically related to the verb *lo**s**e*. Rhotacism is thought to have occured in Old Latin, and to have been complete by the 4th century BCE. As a result of these changes, original intervocalic *\*s* ultimately became *r* in Classical Latin. In the ancestor of Latin, *\*s* seems to have also been voiced to [z] after the voiced resonant /r/, so another related sound change is original *\*rs* to Classical Latin *rr.* One word that shows this change is [*terra*][1] "earth," from Proto-Indo-European *\*ters*- (“dry”). The cluster *rs* that occurs in Classical Latin is a simplification of earlier *\*rtt* or *\*rts*; this is why most words with *rs* also have *rt* in some forms or in related words (such as [*ars*][2] "art," genitive *artis*). There also seems to have been voicing of *\*s* after the voiced lateral resonant /l/, but the end outcome of that cluster in Classical Latin was geminate *ll.* These two environments (between vowels and after *r*) are the main cases where Latin *r* regularly came from earlier *\*s.* There are also some words with r followed by a consonant where the *r* was originally intervocalic *\*s*, but the following vowel was lost (as in [*veternus*][3]). However, there also seem to be some words where *\*s* developed into *z* and then *r* directly before a voiced consonant; in fact, the voicing in these words seems to have occurred earlier than the intervocalic voicing. For example, Proto-Indo-European *-*sg*- seems to be the ancestor of *rg* in *[mergo][4]* "to immerse, engulf" and probably in *[virga][5]* "twig, switch, rod" (de Melo). This also applied before the semivowel *v* [w]; thanks to TKR for pointing this out in a [comment][6]. A strange example that does seem to have undergone rhotacization between Old and Classical Latin is the word *carmen,* which Lewis and Short say used to be *casmen.* ##In Classical Latin, the distribution of *r* and *s* often didn't conform to the conditions of rhotacism, which indicates that it was no longer an active sound change by that time Once the sound *r* became established in certain words, their derivatives might come to have it in other environments. For example, as Joonas Ilmavirta mentions, some words such as *honor* developed final *r* instead of *s* after a vowel; it's believed this is due to the presence of *r* in other related forms (like *honoris*). Single intervocalic *s* did exist in Classical Latin; we believe it was pronounced [s]. In many words, it corresponds to inherited *\*ss* which seems to have become shortened in some environments. Other words with intervocalic *s* have been explained as borrowings from other languages, such as Greek or Italic languages such as Oscan that did not undergo rhotacism. Whatever the reason, their presence indicates that rhotacism was not active as an automatic sound change in the Classical era (Gorman). Peck summarizes the change as follows: > In Latin rhotacism is a regular law, for though there are many > apparent exceptions to it, examination shows that in most of these > cases the s was not originally intervocalic or else that the word is > of (a) foreign or (b) late origin. [...] > > The unvoiced s first passes into the voiced s (z) and thence to > lingual r, for the position of the vocal organs in pronouncing z is > substantially the same as that required for r. References: - Buck, Carl Darling. 1904. [*A grammar of Oscan and Umbrian : with a collection of inscriptions and a glossary*][7] - de Melo, W. D. C. "[Faliscan: Phonology Part 2][8]" - Gorman, Kyle. "[Latin Rhotacism for Real][9]" - Peck, Harry Thurston. 1898. [*Harpers Dictionary of Classical Antiquities*][10] [1]: https://en.wiktionary.org/wiki/terra#Latin [2]: https://en.wiktionary.org/wiki/ars#Latin [3]: https://en.wiktionary.org/wiki/veternus [4]: https://en.wiktionary.org/wiki/mergo#Latin [5]: https://en.wiktionary.org/wiki/virga#Latin [6]: https://latin.stackexchange.com/questions/136/rhotacism-why#comment237_146 [7]: https://archive.org/stream/grammarofoscanum00buckuoft#page/74/mode/2up [8]: http://wolfgang.demelo.de/faliscan/5_phonology.pdf [9]: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.225.8217&rep=rep1&type=pdf [10]: http://www.perseus.tufts.edu/hopper/text?doc=Perseus%3Atext%3A1999.04.0062%3Aalphabetic%20letter%3DR%3Aentry%20group%3D4%3Aentry%3Drhotacism-harpers