Why would the prae­po­si­tion "per" ᴇᴠᴇʀ take an ab­la­tive in­stead of an ac­cu­sa­tive com­ple­ment?

Question

ᴘᴇʀ + ᴀʙʟ.: Bar­bar­ism, solœ­­cism, or di­a­chron­ic evo­lu­tion?

Lewis and Short clear­ly state that per is a prae­po­si­tion whose nor­mal com­ple­ment is in the ac­cusative. With­out hav­ing dol­ven too deeply in­to dusty cor­po­ra of lim­it­ed search ca­pa­bil­i­ty, all ex­am­ples I’ve so far chanced up­on, how­so­ev­er ar­bi­trar­i­ly scat­tered and there­fore ir­rep­re­sen­ta­tive­ly anec­do­tal, nonethe­less uni­ver­sal­ly em­ploy the ac­cusative for all of per’s com­ple­men­ta­tion.

How­ev­er, L&S fur­ther men­tion, al­be­it par­en­thet­i­cal­ly, that “by so­le­cism” per can some­times be found to take an ab­la­tive com­ple­ment, viz.:

per, prep. with acc. (by sole­cism with abl. PER QVO, =

Is the su­per­cit­ed “so­le­cism” say­ing that if you’re us­ing ab­la­tive, you have no rea­son to both­er with an ex­plic­it prae­po­si­tion in the first place?

In book 2 of The Æ­ne­id start­ing at line 210, Vir­gil has with/by blood and fire in the ab­la­tive with­out both­er­ing to dec­o­rate these with a ple­o­nas­tic prae­po­si­tion:

Fit so­ni­tus spu­man­te sa­lo; iam­que ar­va te­ne­bant,
ar­den­tis­que ocu­los suf­fec­ti san­gui­ne et ig­ni,
si­bi­la lam­be­bant lin­guis vi­bran­ti­bus ora.

I can read­i­ly un­earth un­count­ably many ex­am­ples of ven­er­at­ed writ­ers us­ing per nev­er with the ab­la­tive, on­ly with the ac­cusative alone. My read­ing of the OED’s first two sub­sens­es for so­le­cism sug­gest that they are equat­ing so­le­cism with cat­a­chre­sis:

1. a. An im­pro­pri­ety or ir­reg­u­lar­i­ty in speech or dic­tion; a vi­o­la­tion of the rules of gram­mar or syn­tax; prop­er­ly, a faul­ty con­cord.

   b. With­out ar­ti­cle: Vi­o­la­tion of the rules of con­cord in gram­mar or syn­tax; in­cor­rect or un­gram­mat­i­cal speech or dic­tion, or the use of this.

Shall I as­sume that giv­ing per an ab­la­tive com­ple­ment is to be con­sid­ered stig­mat­ic of one or an­oth­er sort of faulty gram­mar, wheth­er per­ceived to come from un­let­tered rus­tics who are pre­sumed to know no bet­ter, or as some de­spec­tive ne­ol­o­gism in­vent­ed by new­er gen­er­a­tions fur­ther along the Sis­y­phe­an pen­du­lum swing­ing from Clas­si­cal Lat­in’s high­ly syn­thet­ic mod­el to­wards the some­what more an­a­lyt­ic mod­el of Pro­to- and even­tu­al­ly Mod­ern Ro­mance?

Answer 1

The L&S entry is pretty clear, in my opinion. Per takes the accusative, but it has mistakenly been used with the ablative. It cites two examples from later inscriptions:

• Inscr. Miseni Repert. ex a. p. Chr. n. 159
• Inscr. Orell. 3300

After some tracking down, I found it in Campania tardoantica (284-604 d.C.). Here is a relevant image from pg. 283:

Answer 2

Francesco Cavalli called the mass he wrote in 1675 Missa pro defunctis per octo vocibus and that's still the name by which we know it. No doubt it should read "Missa pro defunctis octo vocibus", where octo vocibus is in the dative.