When is there a U instead of an E in gerund(ive)?

Question

Tuomo Pekkanen's Latin grammar mentions (§52.3) that the -e- added to the present stem before -nd- in the gerund and gerundive (in the third and fourth conjugations) can be replaced with a -u-. For example, mittendus = mittundus and faciendus = faciundus. No more details are given.

When can I one replace the typical -e- with -u-? Is this specific to some era, author, or verbs? Would it be appropriate to occasionally make the replacement in any context?

I wanted to understand the -e-/-u- variations in gerund(ive)s better. The variations can also be across words, not only within single words; I consider both variations to be essentially the same phenomenon.

Answer 1

Weiss writes that

"The u-forms are characteristic of legal and archaizing style, e.g. pecuniae repetundae (the recovery of extorted money), and are found in the isolated forms secundus 'following' and rotundus 'round' (p. 444; emphasis mine - Alex B.).

Powell 2007/2011 adds that even though "the gerundive suffix in -undus, rather than -endus, has an old-fashioned flavor and is characteristic of legal Latin," "neither confined to it nor universal in it" (p. 482, note 21; emphasis mine - Alex B.).

As for specific authors, Leumann 1977 (p. 331) mentions first Plautus and Terence (Neue and Wagener write that the u-forms are gewöhnliche, i.e. usual/normal, in Plautus and Terence, p. 332); also Sallust and Gellius, cf.:

c) Für -endo- erscheint bei Verben der 3. und 4. Konjug. archaisierend oft -undo- (aus -ondo-), so meist Plt. Ter., dann Sall. Gell., auch inschr. Lex repett. (ausgenommen 65 tribuendei, also hinter u), D 620 veniundum est; archaisch wohl emundi und vendundi bei Cic.; alte Formel dat. iure dicundo; immer eundo- (wie ptc. eunt-) zu īre. Nur hinter u u̯ qu̯ scheint -undo- früh gemieden worden zu sein (doch beachte oben secundus): loquendi Plt., volvendus Enn., restituendos D 460, oben tribuendei. Danach ist vermutlich ursprüngliches -ondo- an ptc. -ent- zu -endo- angeglichen worden; bei dieser Annahme kann das -end- nicht auf idg. -n̥d- zurückgehen.

From the above quote from Leumann, we can also learn that in verbs whose stem ended in u, u̯, qu the u-forms were avoided in Latin - the only exception to this observation seems to be secundus.

For more examples, see pp. 331-340 in Neue and Wagener 1897, volume 3: Das Verbum

Answer 2

To clarify a little more.

In Gramatica Latina (latin grammar) by Santiago Segura:

Participe of passive future: It is also called verbal adjective in -NDUS and gerundive and is formed by adding to the present theme the suffix -ND-US, -ND-A, -ND-UM, sometimes by -e- (3rd and 4th conjugation): ama- ndus, -a, -um; dele-nd-us, -a, -um; leg-e-nd-us, -a, -um; cap-i-e-nd-dus, -a, -um; audi-e-nd-us, -a, -um. It is declined as a bonus, -a, -um.

In the archaic period the verbs of the 3rd and 4th conjugation offer forms in -undus (with vocalism o> u, instead of e): legundis, deferundo, querundai, repetunda. The analogical action of the present participle in -e-ns, -e-ntis made generalize the form in -endus, surviving scarce forms in -undus in the archaizing language of the law (repetundis) or as adjectives (oriundus, secundus and others).

Answer 3

Just to add on to Alex B's answer, though I can't offer as authoritative of sources:

Would it be appropriate to occasionally make the replacement in any context?

In the late Republican period, the answer seems to be yes, but it sounds a bit archaic. I'd compare it to putting the object before the verb in English ("speaking with the object the verb preceding"): people will understand it readily, and it's perfectly acceptable for poetic purposes, but it's not how people actually speak in day to day life.

Historically, based on corpus searches:

• Plautus and Terence tend to prefer -und- over -end-, though they use both
• Cato tends to prefer -end- over -und-, though he uses both
• Both Sallust and Lucretius chose to use -und- almost universally as a stylistic choice, and were both considered deliberately archaicizing—but not difficult to understand
• Caesar uses -und- very rarely, heavily preferring -end-
• Cicero uses -end- (almost?) exclusively
• Vergil never uses -und- as far as I can tell, except in fossilized secundum

So in the famous authors at least, it's clear that -und- was steadily dying out in the second century BCE, and was considered somewhat archaic during the first.

After that, during Imperial times, it was presumably still understood, but I haven't been able to find it used with any sort of frequency. I assume it was treated more like the verb endings -est and -eth in English: people will still unambiguously understand "whither goest thou?", but even in poetry it's considered extremely archaic, and is a stylistic affectation more than anything else.