In both Sindarin and Noldorin, the long vowels [ē] and [ō] became [ī] and [ū]. Where [ē] and [ō] occur in Sindarin and Noldorin words, they are generally the result of the lengthening of short vowels in monosyllables. The same sound changes also occurred in the earliest versions of the language (Gnomish and Early Noldorin), and similar changes occurred in Welsh (WGHC/§59i), though in Welsh the sound change [ō] > [ū] only occurred in final syllables. Tolkien described these changes several times over his life:

ê gave î ... ô [gave] û (Gnome Vowels from the 1910s, PE15/13).

In the earliest recorded Noldorin, [ē, ǣ, ā, ǭ, ō] remained distinct though changed to [ī, ei, ǭ, ou, ū] (first version of Tengwesta Qenderinwa [TQ1] from the 1930s, PE18/46).

In the earliest Noldorin ē, ę̄, ā, ǭ, ō appear as ī, ę̄, ǭ, ǭ, ū (second version of Tengwesta Qenderinwa [TQ2] circa 1950, PE18/96).

C.E. ei, oi > [S.] ē > ī ... ō > ū (diphthong chart associated with Tolkien’s discussion of Hands, Fingers and Numerals from the late 1960s, VT48/7).

Examples are plentiful. Here are a few clear examples from each major period:

• ᴱ✶tē > G. tî “straight” (GL/71).
• ᴱ✶t’lḗpe > ᴱN. tlî “butter” (PE13/154).
• ᴹ✶nēthē > N. nîth “youth” (Ety/NETH).
• ✶dērā > S. dîr “hard, difficult” (PE17/154).
• ᴱ✶Noronōr- > G. Dronūrin “Messenger of the Gods” (GL/31).
• ᴱ✶kópa > ᴱN. cû “bay, cove” (PE13/141).
• ᴹ✶b’rōnā > ON. brūna > N. brûn “old” (Ety/BORÓN).
• ✶χīþilōmē > hithlũṽ > S. Hithlũ “Hithlum” (PE17/133).

Conceptual Development: This seems to be an exceptionally stable and well documented phonetic rule, established very early and maintained thereafter. In Noldorin, however, it seems this change took place at a much more ancient stage than in Sindarin, in the Old Noldorin period before [ɣ] vanished. For example:

• ᴹ✶b’rōnā > ON. brūna > N. brûn “old” (Ety/BORÓN).
• ᴹ✶teʒē > N. tê (Ety/TEƷ).

In the second example, if the ʒ [ɣ] vanished first, the result would have been [teɣē] > [tē] > [tī]. Likewise, this change took place before short [a], [o], [u] became [e], [œ], [y] preceding [i] in Noldorin, since there are several examples where the agental suffix -dîr (from ✶-ndēr) caused i-affection:

• N. feredir “hunter” = fara- + dîr (Ety/SPAR).
• N. ceredir “doer, maker” = car(a)- + dîr (Ety/DER).

However, in Sindarin, suffixal -dir or -nir never causes i-affection:

• S. dagnir “slayer” = dag- + (n)dîr (PE17/97).
• S. Thorondir “*Eagle-man” = thoron + dîr (LotR/1039).

The same is true of suffixal -hîr “lord” (from ✶-khēr):

• S. Barahir “*Fiery-lord” (S/148).
• S. rochir “knight” = (Let/282).

Note that there are also Noldorin examples where i-affection does not occur, such as N. rhandir “wanderer” vs. S. randir “wanderer”. These Noldorin examples can be explained as late or reformed compounds, rebuilt from their constituent elements. In theory, the same might be true of the Sindarin compounds as well. But since i-affection occurs in no Sindarin examples, including those like dagnir and rochir that are likely ancient in origin, it is probably safer to assume that [ē] > [ī] simply occurred after i-affection in Sindarin.

As further evidence that the Sindarin sound change occurred well after the Old Sindarin period, in the 1960s Tolkien vacillated on the singular form of têw “letter” (< teŋwa), waffling between têw (PE17/44) and tîw (PE22/149). This vacillation makes sense if [ē] > [ī] occurred around same time that [ŋ] vanished with compensatory lengthening ([eŋ] > [ē]), and Tolkien couldn’t quite make up his mind which happened first. This makes the most sense if all these phonetic developments were part of the Sindarin rather than Old Sindarin period.

Hat tip to Elaran for pointing out the difference in i-affection between Noldorin and Sindarin in these cases.

HPG/§1.1