This noun case is one that Tolkien (probably) abandoned in his later writings. In the Early Qenya Grammar (EQG) he called it the “Manner” case (PE14/46, 78), but in declension tables from the late 1920s through 1930s he often labeled it the “Adverb” case (PE16/113; PE21/4, 53) and in a couple places he called it the “Comparative” case (PE21/42, 46). Elsewhere, though, Tolkien used the term “adverbial cases” to collectively refer to the allative, ablative, locative cases. I use the term “similative” (as in “something that is similar to another thing”) as the label for this noun case to avoid confusion with other grammatical functions, but Tolkien himself never used this term.

This case was used to form “an adverb of manner - a Q. formation, it is also employed as an indeclinable adj. = like” (PE21/3, item #10). It was formed with the suffix -ndon, such as:

• ᴱQ. kiryandon “like or as a ship” (PE14/46; PE21/69; PE17/74).
• ᴱQ. kulundon “like gold” (PE14/46).
• ᴱQ. wilwarindon “as/like a butterfly” (MC/213, 220).

The noun case appears in several phrases from the 1920s and 30s, most notably in early versions of the Markirya poem:

• ᴱQ. valkane wilwarindon “vague as a butterfly” (MC/213).
• ᴱQ. tinwelindon talalínen “on wings like stars” (MC/213).

This noun case continued to appear in declension charts and descriptions of Quenya noun cases through the 1930s and 40s, and there is even a mention of it in notes on word lists for The Lord of the Rings written around or shortly before 1960, but at this point Tolkien abandoned the noun case:

like gold, in gold fashion = laurendon ... abandon -ndon. this is an agental suffix ? kiryando = sailor (PE17/58).

However, this is not the last mention of -ndon; it appears again in some notes from 1966-67:

Cf. also the ending -ndo, -ndon, as in kiryando(n) “like a ship” (PE17/74).

Thus it is conceivable it continued to be valid into the late 1960s. However, given its absence from the noun declensions in the contemporaneous Plotz letter or the 1960s version of the Markirya poem, I think it is likelier Tolkien had mostly given up on this noun case.

Conceptual Development: This noun case appeared in the Early Qenya Grammar (EQG) from the 1920s along with several examples: a kálie kulundon “it shone like gold”; kiryandon “like a ship” (PE14/46). With consonantal nouns the suffix became -indon with the i derived from consonantal plurals (PE14/47, 78). In the manuscript version there was a single (archaic) shorter variant -don (PE14/47) but in the typescript version, there were two short forms: -son, -don which were often used with “true consonantal words” (PE14/78).

The suffix -ndon was retained in a declension chart later in the 1920s except that the long consonantal form became -anon alongside short -don (PE16/113). In this chart Tolkien explicitly gave plural forms, which were simply -ndon added the vocalic or consonantal plural (PE16/115). He also gave a “rare & irreg[ural]” dual form -ndos (PE16/114).

These singular and plural forms reappeared in the Declension of Nouns from the early 1930s, but the dual did not (PE21/4-6, 25). For many consonantal nouns, the non-assimilated form was -undon rather than -anon: nēn(un)don (PE21/23), malanon vs. malundon (PE21/24), etc. In the Declension of Nouns Tolkien also gave quite a few examples of the short suffix -don assimilated to consonantal nouns, but this only differed from -don in a handful of examples:

• kas → kardon.
• sūt → sūtanon, †sutson.
• yāt (yak-) → yakanon, yakundon, †yakson.
• ambor (ambos-) → amborundon, †ambusson.
• qelet → qeletundon, †qelitson.
• kelut → kelutundon, †kelutson.
• filet (filek-) → filekundon, †filikson.
• arat (arak-) → arukundon [sic.], †arakson.

Aside from the first example which is just the regular assimilation sd > zd > rd, these all seem to represent an archaic variant suffix †-son as mentioned in the typescript version of EQG (PE14/78). This noun case shows little change after this, except that in a chart from later in the 1930s, the consonantal singular suffix -indon was restored (PE21/52), replacing -anon from the late 1920s and from the Declension of Nouns. This case ending continued to be mentioned in discussions of Quenya noun cases through the 1940s:

Q. also developed a comparative adverb or case ending in -ndon (of obscure origin): “like, as”; so kiryandon “like or as a ship”. This remained incapable of [?indicating] number (PE21/69, Notes for Qenya Declensions).

As noted above, Tolkien ultimately abandoned this noun case in the 1950s (PE17/58).

The full set of conceptual developments is given in the table below, using the version numbers for the declension charts from PE16 and PE21, with EQG for the Early Qenya Grammar and LQ for Late Quenya forms (in Plotz and elsewhere). The Entu, Ensi, Enta Declension (EEED) is inserted between versions 2 and 3, and the Bodleian Declension (BD) between versions 5c and 6, as discussed in the entry on noun cases.

V	Similatives	Sg.	Du.	Pl.
EQG	vocalic	-ndon
EQG	consonantal	-indon
EEED	vocalic	-ndon	-ndos
EEED	consonantal	-anon	-anos
3	vocalic	-ndon	-ndos	-lindon
3	consonantal	-anon		-indon
4	vocalic	-ndon		-lindon
4	consonantal	-anon/-undon		-indon
5a	vocalic	-ndon	-ndont	-lindon
5b	vocalic	-ndon	-ndont	-lindon
5c	vocalic	-ndon		-lindon
5c	consonantal	-indon		-indon
BD	vocalic	-ndon
6	vocalic	-ndon
LQ	vocalic	-ndon

Neo-Quenya: I recommend against using this noun case in Neo-Quenya. Its function can be replaced by the preposition ve “as, like, similar, after the manner [of]”.