There are three Quenya noun cases that have to do with the motion or location of the declined noun:

• The allative case indicating motion towards the noun: -nna.
• The ablative case indicating motion away from the noun: -llo.
• The locative case indicating location at/in/on the noun: -sse.

Tolkien often collectively labeled these the “adverbial suffixes”. As he described them in Common Eldarin: Noun Structure from the early 1950s:

Adverbial suffixes. Adverbial elements that formed virtual “cases” though not forms belonging to organized declensions in Eldarin. The most important of these, especially for the structure of the Quenya declensions were the following:

• 1) locative, adessive or inessive: sē̆, with “fortified” forms -sse, ste.
• 2) allative: dā̆, -ndā. Possibly also nā̆, -nna.
• 3) ablative: lō̆; “fortified” llō, ldō.

These were originally adverbial, incapable of indicating number, and not necessarily formed even from the same stem as the related noun. Thus the fortified forms added to stems yielding monosyllabic nouns always require a dissyllabic stem with ómataima: nenesse “in (the) water”. The inclusion of these elements in organized declensions with plural and dual forms is a special development of Quenya (PE21/79).

These three cases share a number of traits making them worth discussing as a group. As noted above, the suffixes were originally used to create adverbs and could only be used with a singular noun. For example, the allative suffix -nna (“towards”) was originally equivalent to the English suffix “-wards”, and so coanna = “housewards” and ambonna = “hillwards”. They eventually evolved into full inflexions in Quenya, usable with plural and dual nouns as well: coannar “to the houses”, ambonta “to the pair of hills”.

For the most part, these “adverbial” inflections are added directly to vocalic nouns, but it isn’t entirely clear how consonantal nouns are inflected. If the consonantal noun ended in the same letter, very often the suffix is assimilated to the end of the noun, so:

• Aman becomes Amanna “towards Aman” (VT49/26).
• menel becomes menello “from Heaven” (VT43/13).
• ainas becomes *ainasse “in the shrine” (presumably).

If the last consonant of the noun is different from the first consonant of the suffix, this kind of assimilation becomes complicated. Many of the examples have a “joining vowel” between the suffix and the noun, avoiding the need for assimilation:

• Elendil becomes Elendilenna “to Elendil” (PM/401).
• Ambar becomes Ambarello “from the World” (MS; Merin Sentence).
• Lórien (Lóriend-) becomes Lóriendesse “in Lórien” (RGEO/58).

In almost all attested examples of singular nouns in the 1950s and 60s, the joining vowel is -e-. However, with plural consonantal nouns the joining vowel is -i- instead:

• elen becomes elenillor “from the stars” (MC/222).
• mindon becomes mindoninnar “upon towers” (MC/222), beside the assimilated form mindonnar.
• [ᴹQ.] Koiviénen becomes Koivienenissen “at the waters of awakening” (VT27/7).

The joining vowel -i- is only used with plural consonantal noun. The allative plural of cirya is ciryannar “from ships”, and the locative plural of lasse is lassessen “on leaves”.

There are some special rules for the four cardinal directions númen, rómen, formen, hyarmen “west, east, north, south”. These words lose their final -n before a directional suffix is added. As indicated by Tolkien in a “Words Published” list from the late 1950s, they sometime lost their final -n before other noun cases as well, but this was not universal:

Cf. Róme or Rómen. Rómello but rómeno or rómeo, rómena or rómen, rómenwa or romeva (PE17/59).

Thus:

• formen becomes formenna “to the north” (VT49/26).
• rómen becomes rómello “from the east” (LotR/377).
• númen becomes númesse “in the west” (LR/72).

Other joining-vowel possibilities: Even though most examples show the joining vowel is -e- for consonantal nouns, it is possible the rules are more complex than this. As indicated in the quote from Common Eldarin: Noun Structure above, in Common Eldarin “the fortified forms added to stems yielding monosyllabic nouns always require a dissyllabic stem with ómataima: nenesse ‘in (the) water’ (PE21/79)”. Although this example also uses e, it is because the base vowel (ómataima) is also e, so the ancient locative for tāl “foot” would have been talasse (as mentioned on VT43/16) rather than talesse.

It is conceivable this system survived in Quenya as well, so that the joining vowel might match the base vowel of the noun. As evidence of this, the instrumental form of ambar (ambart-) is ambartanen “by doom” (S/223) rather than *ambartenen. Alternately, the vowel could be a preserved ancient vowel lost in the uninflected form. A third possibility is that the exact joining vowel that developed was conditioned by the suffix that followed, as it was in Early Qenya (see below). The inflectional system could even involve a mixture of these elements.

The reality is that we just don’t have enough information to reconstruct the exact system (or systems) Tolkien used in the 1950s and 60s. However, one plausible and relatively straightforward theory is that any ancient variations would have normalized into a common joining vowel via analogical leveling. For purposes of Neo-Quenya, I would stick with the joining vowel -e- in all the adverbial cases.

Also note that in some Quenya prayers from the 1950s, Tolkien experimented with a more complex set of assimilations for consonantal nouns. These are discussed in the entry on the assimilated locative.

Conceptual Development: It is well known that Quenya’s adverbial cases were originally inspired by Finnish. Tolkien admitted as much in notes for a letter written in 1964:

Finnish, which I came across when I first began to construct a “mythology” was a dominant influence, but that has been much reduced. It survives in some features: ... such as the inflexional endings -sse (rest at or in), -nna (movement to, towards), and -llo (movement from) (PE17/135).

For comparison, the equivalent Finnish case endings are allative -lle, ablative -lta and locative (innessive) -ssa, though the meanings are not entirely in alignment with Quenya.

Tolkien first mentioned the “adverbial suffixes” in the Early Qenya Grammar of the 1920s (PE14/46, 78), but this list also included partitive -inen (later used for the instrumental) and “adverb of manner” -ndon (which I label the similative case). In the 1920s the allative was -nta rather than -nna. In these early documents, the joining vowel depended on the suffix: -anta, -ullo, -isse, but “true” consonantal noun often used shorter forms with assimilation instead: -ta, -lo, -se (PE14/78). This was also the case in declension charts from later in the 1920s (PE16/113).

In the Declension of Nouns from the early 1930s, short suffixes with assimilations were the norm for nouns whose stem ended in a single consonant (PE21/20-22), but words with consonant clusters in the noun stem used a joining vowel: -unta, -ullo, -esse being the forms in this document (PE21/25-27). Later in the 1930s the allative became -nde and the joining vowels were -ande, -ullo, -esse. Thus in the 1920s and 30s, Tolkien used joining vowels depending on the suffix itself rather than the vowels of the declined noun, rather than the common joining vowel e typical of adverbial cases in the 50s and 60s.

For a more detailed discussion of their conceptual development, see the entries for the individual noun cases.