Starting in the Primitive Eldarin of the 1930s, many initial consonants could have an s-prefix (PE18/43, 93). For s-prefixed voiced consonants, the resulting combinations [sr-, sl-, sj-, sw-, sm-, sn-] produced corresponding voiceless consonants in (Old) Sindarin: [r̥-, l̥-, j̊-, w̥-, m̥-, n̥-].

[sr-, sl-] > [r̥-, l̥-]: For Sindarin, Tolkien only explicitly discussed these sound changes for initial r, l:

LH represents this sound when voiceless (usually derived from initial sl-) ... RH represents a voiceless r (usually derived from older initial sr-) (LotR/1114).

There are numerous Sindarin examples of [sl-, sr-] > [r̥-, l̥-]:

• √S-LAS > S. lhaw “ears” (PE17/62).
• ✶(s)lōkō > S. lhûg “snake” (PE17/160).
• √SLON > S. lhôn “noise” (PE17/138).
• ✶srāban > S. rhovan “wild beast” (PE17/78).
• ✶srāwe > S. rhaw “flesh” (MR/350).
• ✶srugu > S. rhû “evil, wicked” (PE17/115, 170).

[sj-, sw-] > [j̊-, w̥-]: Most of the evidence for [sj-, sw-] > [j̊-, w̥-] is from the Noldorin sound changes of the 1930s. There are, however, a couple explicit examples of [sw-] > [w̥-] in Sindarin:

• √SWAW > S. hwâ and hwaewar “breeze[?]” (PE17/34).

One complication is that [j̊-] eventually became [h-] in both Sindarin and Noldorin. In the Comparative Tables from the 1930s Tolkien gave the development as primitive sj- > ON. ɧ- > χ- > N. h- (PE19/21). Tolkien described these sound changes more fully in his notes on the Old Noldorin use of the Feanorian Alphabet from the late 1930s:

[Old Noldorin] 31. 9 hyō hy [ɧ, later χ^i̯]; 32. o  hwai hw [ƕ, later χ^w̯] (chart on PE22/25).

Based on these examples, it seems that the actual development was likely primitive [sj-] to coarticulated [ɧ-] to palatalized [xʲ-], at which point the palatization was lost and the resulting initial [x-] became [h-] along with initial [x-] from other sources (such as primitive [kʰ-]). This line of development is supported by other quotes from the Feanorian Alphabet discussion:

The sound h did not exist in the older period. It redeveloped in ON (i) from d = kh > χ when it stood in absolute initial position in a word group; (ii) from older 9 = [ɧ] in the same position ... (PE22/29).

Signs of the change 9 [ɧ] > χ^ı̯ medially are seen in the confusion in late ON of medial dÔ = khy and 9, and in the occasional appearance of initial dÔ after the article `B [i] “the” for 9 (PE22/29).

There are number of Noldorin examples of sy- > h- in The Etymologies of the 1930s, unfortunately without any details on the intermediate changes:

• ᴹ✶syadā > N. hâdh vs. ᴹQ. hyando “cleaver” (Ety/SYAD).
• ᴹ✶syalmā > N. half “seashell” vs. ᴹQ. hyalma (Ety/SYAL).
• ᴹ✶syad-ta > N. hast “axe-stroke” (Ety/SYAD).

A quote from Appendix E of The Lord of the Rings shows that something like these Noldorin phonetic developments remained valid in Sindarin:

[Quenya] HY was usually derived from sy- and khy-; in both cases related Sindarin words show initial h, as in Q. Hyarmen ‘south’, S. Harad (LotR/1115).

Note that the words Q. Hyarmen and S. Harad are both derived from √KHYAR, so they are not themselves representatives of primitive sj- sound changes, but nevertheless hint at similar developments.

The Old Noldorin quotes above indicate there were also complex developments for initial [sw-], but analyzing the finer details of these sound changes in Sindarin is complicated by Tolkien’s revisions of the ultimate results of these developments: primitive sw- > f- or chw- in Early Noldorin and Noldorin (see below) versus hw- in Sindarin (see above). Purely for notational convenience, the phonetic rules in this lexicon represent these Old Sindarin and Old Noldorin sound changes as [sj-, sw-] > [j̊-, w̥-], though as noted above the actual developments were probably more complex.

[sm-, sn-] > [m̥-, n̥-]: Most of the evidence for [sm-, sn-] > [m̥-, n̥-] also dates back to Noldorin. The few Sindarin examples show s-prefixed nasals becoming simple nasals:

• ✶snagdē > S. naedh “wound” (PE19/91).

In the Comparative Tables from the 1930s Tolkien gave the development as primitive sm-, sn- > ON., N. m-, n- (PE19/21), but in The Etymologies there are many examples of these developing into voiceless hm-, hn- (see below). Furthermore, in the Quenya Outline of Phonology from the early 1950s, Tolkien described an analogous Quenya development of primitive [sm-, sn-] first to voiceless [m̥-, n̥-] which then later revoiced to [m-, n-]:

sm, sn. In PQ s was lost and only m, n appeared. The process by which s was lost was peculiar to the initial position. There s was not voiced, but unvoiced the following nasal. The combination sm̌, sň then proceeded with loosening of s and assimilation to the following consonant (as also seen in sk, st > [χχ], [þþ]) to long voiceless m̌m̌, ňň, later simplified to m̌, ň. This voiceless m̌, ň was still present in AQ and was represented by signs that may be transcribed hm, hn. These voiceless nasals, limited to initial position and not of very frequent occurrence, became normal m, n.

The occurrence of hm, hn in AQ is the chief source of evidence for initial sm, sn in CE; since all remaining Eldarin dialects (other than Nandorin) also lost s in these combinations, and being much later in record than AQ show no clear trace of it. In the Nandorin dialect of Ossiriand sm, sn persisted, but that language is imperfectly recorded. Initial sm, sn, becoming medial in old compounds, followed the medial development and so affords occasional evidence of the former presence of s (PE19/79).

Based on this quote it seems likely Sindarin went through similar developments, but there is not enough evidence to be certain. If it did, the change of voiceless [m̥-, n̥-] back to voiced [m-, n-] was early enough not to be reflected in the written language.

Conceptual Development: There was a fair amount of complexity at each conceptual stage for s-prefixed consonants. Each period is worth separate discussion.

Gnomish Developments (1910s): The s-prefixion of consonants does not seem to be part of primitive word formation in Tolkien’s earliest conception of Primitive Elvish, but as discussed by Roman Rausch in his Historical Phonology of Goldogrin, there are enough examples to deduce some general developmental patterns in Gnomish (HPG/§2.1, 2.5):

sl- > thl-:	ᴱ√SḶPḶ > G. thlib- “to sup, lap up, suck” vs. ᴱQ. sulp- (QL/84, GL/73)
	ᴱ√SḶTḶ > G. thlid- “to sort out, sift” vs. ᴱQ. silt- (QL/84, GL/73)
sr- > thr-:	ᴱ√SIRIPI > G. thripthon “stem” vs. ᴱQ. siripta (QL/84, GL/73)
sy- > h-:	ᴱ✶sı̯on > χ̑on > G. ho(n) “grandson of” (GL/49).
	ᴱ√SAYAPA > χı̯ap > G. hab(in) “shoe” vs. ᴱQ. hyapa (GL/47, GL/66, PE12/26).
sw- > fw-:	ᴱ√SUẈU > G. fwi- “fem. patronymic prefix” vs. ᴱQ. sui “daughter” (QL/87, GL/36)
	ᴱ✶suŏ́sā > G. fo “clan” (PE13/117); in this example, probably with [wo] > [o]
sm- > f-:	ᴱ√smaika- > G. faig “cruel” vs. ᴱQ. maika (GL/33)

For primitive [sl-, sr-], some examples also have variant forms beginning with fl-, fr-, such as:

• G. thripthon vs. fripthon (GL/36, GL/73).
• G. thlidhra vs. flidhra “willowy, graceful” (GL/35, 73); ᴱQ. cognate sild(r)a “slender” (QL/84).

In the first example Tolkien said the variant beginning with f- was dialectical, and in the second example he said it was not the “proper” form, so it seems these were not the normal phonetic developments. These Gnomish phonetic rules are complicated by the fact that the primitive voiceless spirant th (þ or [θ]) was also part of Tolkien’s conception of Primitive Elvish in the 1910s, and some of these initial primitive [sl-, sr-] might actually be [θl-, θr-]. For example, the root for ᴱQ. sild(r)a was given as both ᴱ√ÞḶÐḶ and ᴱ√SḶŘḶ in the Qenya Lexicon (QL/84), so G. thlidhra/flidhra could be from either primitive [sl-] or [θl-] (more likely the latter).

The change sy- > h- seems very like the Sindarin/Noldorin developments discussed above, and this particular sound change remained stable through all the conceptual periods (though no doubt some of the behind-the-scenes details changed). For sw- > fw-, Roman Rausch suggested that this form also likely passed through thw- [θw-] first (HPG/§2.5), as indicated by medial developments such as -thwi, which is the suffixal variant of prefix fwi-. The sound change sm- > f- for G. faig is odd, and Tolkien first wrote and then deleted the form {fwaig}, so perhaps he originally conceived of this it as a blending with sw-.

Early Noldorin Developments (1920s): There are fewer Early Noldorin examples of these sound changes for s-prefixed consonants, but there are enough to detect some conceptual shifts:

sl- > lh-:	ᴱ√SḶPḶ > ᴱN. lhif- “to drink” (PE13/148) vs. G. thlib- above.
	ᴱ✶sleiwa > ᴱN. lhui “pale” (PE13/149).
	ᴱ✶slingwé > ᴱN. lhiw “worm” (PE13/149).
sw- > f-:	ᴱ✶swadwé > ᴱN. fadhw “parchment” (PE13/148).
sm- > m-:	ᴱ✶smeigé > ᴱN. mui “crumb” (PE13/150).
sn- > n-:	ᴱ✶snóra > ᴱN. núr “muscle” (PE13/151).

It seems that [sl-] (and [sr-]?) became voiceless liquids in Early Noldorin, while [sm-, sn-] became simple nasals, perhaps passing through a voiceless phase though there is no direct evidence of this. The [sw-] development resembles Gnomish, and probably [sj-] > [h-] as well, given that the particular sound change remained the same in later Noldorin and Sindarin.

In the 1920s Tolkien also introduced a sound-change that was characteristic of the Noldorin stage of the language: ordinary initial [r-], [l-] were also unvoiced, not just those prefixed by [s]. There are numerous examples:

r- > rh-:	ᴱ✶roso- > ᴱN. rhó- “to arise” (PE13/152).
	ᴱ✶rotya- > ᴱN. rhoidia- “to let go” (PE13/152).
	ᴱ✶rotya- > ᴱON. rhoid > ᴱN. rhaid “rush(ing)” (PE13/165).
l- > lh-:	ᴱ✶leyé > ᴱN. lhai “folk” (PE13/148) vs. vs. G. lî (GL/54).
	ᴱ✶lept- > ᴱN. lhê “finger” (PE13/148) vs. vs. G. leptha (GL/53).
	ᴱ✶loktu > ᴱN. lhuth “magic, spell” (PE13/149).

There are still a fair number of Early Noldorin words with voiced initial l or r at this stage, however. The unvoiced lhuth “magic, spell” appeared beside voiced variant luith (PE13/149). There are other voiced examples, with and without voiceless variants, such as ᴱN. lung vs. lhung “heavy” (PE13/163) and ᴱN. rán “moon” (PE13/152) vs. earlier G. Rân and later N. Rhân, S. rawn (GL/64, Ety/RAN, VT48/7). Perhaps Tolkien was not fully committed to this phonetic rule in the 1920s, or perhaps these were dialectical variations. Roman Rausch noted these inconsistencies in his Historical Phonologies of Ilkorin, Telerin and Noldorin around 1923 (HPITN/§4.1.1).

Noldorin Developments (1930s): The development for initial [r-, l-] to voiceless [r̥-, l̥-] in the Noldorin of the 1930s is well known. These sound changes were described explicitly by Tolkien in a couple places, such as his discussion of the use of the Feanorian Alphabet with Noldorin from the 1930s:

The voiceless character of absolute initial l, r was often undenoted in early Gondolic, and j 7 were used in all positions. Rarely j9 79 were used for the voiceless forms. Generally during the Gondolic period, and since the voiceless forms [sic] — chiefly initial, but also derived from contact of l, r + h medially — were written with the modified letters m u, named lhau, rhau, transcribed lh, rh (PE22/32).

These Noldorin developments of initial l-, r- to lh-, rh- are also described in the Comparative Tables of phonetic development from the 1930s (PE19/19). The Noldorin phonetic developments of s-prefixed consonants appear in the Comparative Tables as well (PE19/21), since by this point s-prefixion was a common feature of Primitive Elvish root formation:

• sm- > m-; sn- > n- (with lenited forms mh, v for m-).
• sl- > lh-; sr- > rh-
• sj- > ɧ- > χ- > h-; sw- > ƕ- > f- (with lenited forms χ for h- and chw for f-)

The sound changes in the Comparative Tables are basically the same as Early Noldorin, but revisions in the chart indicate some vacillation on Tolkien’s part. In particular, Tolkien first wrote: sl- > thl- (dialectical fl-) and sr- > thr- (PE19/21, note #37). This seems to be a throwback to Gnomish, and despite their deletion the attested forms in The Etymologies (composed shortly after the Comparative Tables) are more consistent with these Gnomish-like sound changes:

sl- > thl-/fl-:	ᴹ✶slaiwā > ON. thlaiwa > N. flaew “sick” (Ety/SLIW).
	ᴹ✶slindi > N. thlinn “fine, slender” (Ety/SLIN).
	ᴹ✶slingē > N. thling “spider, spider’s web” (Ety/SLIG).
sr- > thr-:	ᴺ√SRIP > N. thrib- “to scratch” (Ety/SRIP).
	ᴺ√SRUS > N. thross “whispering or rustling sound” (Ety/SLUS).

Perhaps Tolkien reintroduced these Gnomish sound changes into the Noldorin of the 1930s to better distinguish the derived results from those of primitive initial l-, r- (> N. lh-, rh-). Tolkien also revised the sound changes for sm-, sn- in The Etymologies into voiceless nasals:

sm- > hm-:	ᴹ✶smāgā > N. hmaw “soil, stain” (Ety/SMAG, EtyAC/SMAG).
	ᴹ✶smalu > ON. smalo > N. hmâl “pollen, yellow powder” (Ety/SMAL).
sm- > hm-:	ᴹ√SNEW > ON. sniuma > N. hniof or hnuif “noose, snare” (Ety/SNEW).

All the Noldorin words in The Etymologies with initial voiceless nasals had original but deleted forms with voiced nasals: {maw} >> hmaw, {mâl} >> hmâl, {nû} >> hniof/hnuif. As noted by Christopher Tolkien, J.R.R. Tolkien did not carry out the changes to voiceless nasals in the associated narrative texts: N. Melthinorn “Tree of Gold” appeared in the contemporaneous versions of the Silmarillion despite {malthen} >> (h)malthen “of gold” in The Etymologies (Ety/SMAL, EtyAC/SMAL), and some initial voiced nasals remained unchanged in The Etymologies as well. It seems the idea that voiceless initial nasals were possible in Noldorin was only a transient idea, but it does indicate that primitive [sm-, sn-] passed first through a voiceless phase as they did in Quenya, as discussed in the Sindarin phonetic developments above.

The earlier phonetic development of [sy-] > [h-] was also well represented in The Etymologies (see the Sindarin discussion above), but Tolkien made some revisions to the development of [sw-] from the Early Noldorin of the 1920s:

sw- > chw-:	ᴹ✶swanda > N. chwann “sponge, fungus” (Ety/SWAD).
	ᴹ√SWES > ON. hwesta > N. chwest “puff, breath, breeze” (Ety/SWES).
	ᴹ√SWIN > N. chwîn “giddyness, faintness” (Ety/SWIN).

These sound changes were likewise described by Tolkien in his discussion of the use of the Feanorian Alphabet for Old Noldorin and Noldorin from the 1930s:

9 o  already invented before in other applications of the Feanorian syllabary were regularly employed in ON for the voiceless j [ɧ] transcribed hy and the voiceless w [ƕ] transcribed hw, derived from PQ sy, sw. Initially [ɧ] endured until the end of the ON period. On phonetic changes affecting these sounds and consequently their spelling, see below (PE22/26).

For [ɧ-], the relevant quotes appear in the Sindarin section above, and show [ɧ-] > [xʲ-] > [x-] > [h-]. As for [ƕ-]:

The sound [ƕ] became [χw]. This probably occurred in all positions — possibly already in late ON. At any rate ON [hw] o and [khw] dé, dn (derived only from PQ khm) appear generally in all Exilic dialects as χw, spelt normally c. But in N.W. dialects (Hithlum and Gondolin) — either by a preservation of a special initial pronunciation, or by weakening in absolute initial position similar to that of χ > h, the sound was a voiceless w [ƕ] still in absolute initial position. This was naturally written o , called hwae — or owing to its absence from other dialects and from post-Gondolic, chwae gondoliðren “Gondolic chwae”. Since it was also normal as an initial sound in other Beleriandic languages, it was also called chwae feleriandren “Beleriandic chwae”. In late Gondolic and post-G[ondolic] Exilic it became chw, and so fell out of use in Noldorin (PE22/34).

It seems that ON. ƕ- (probably [ʍ]) generally became N. χw- [xʷ-], though voiceless w- was preserved for a while in some dialects. It also seems based on the examples above that the unvoicing of the s-prefixed consonants took place in the Old Noldorin period (with the possible exception of nasals), and likely the same is true of Sindarin as well.

Summary: As is well-known, in Sindarin Tolkien abandoned the idea l-, r- > N. lh-, rh-, which means he no longer needed distinct developments for sl-, sr-, so these sound changes reverted to their Early Noldorin developments: sl-, sr- > lh-, rh-. Voiceless initial nasals vanished from the language (they were probably only a brief-lived idea anyway). Tolkien adopted the sound change sw- > hw-, as shown by the Sindarin example above, and by the fact that chw- was no longer a valid initial combination in Sindarin.

To summarize conceptual changes of these phonetic developments for s-prefixed consonants in the various time periods, along with their analogous changes in Welsh:

Primitive Forms:	sl-	sr-	sm-	sn-	sy-	sw-	l-	r-
Gnomish (1910s):	thl-**	thr-**	f-	?	h-	fw-	l-	r-
Early Noldorin (1920s):	lh-	rh-*	m-	n-	h-*	f-	lh-	rh-
Noldorin (1930s):	thl-**	thr-	hm-	hn-	h-	chw-	lh-	rh-
Sindarin (1950s):	lh-	rh-	m-*	n-	h-	hw-	l-	r-
Welsh:	ll-	rh-	m-	n-		chw-	ll-	rh-

Developments marked * are unattested but probable. Developments marked ** have dialectical variations fl- or fr-. Note that Welsh ll- is also a type of voiceless initial l- [ɬ]. For Welsh developments see WGHP/§94-95; those given here are only the most common developments.

GS/§4.228 @@@ FIXME maybe earlier?