Rough breathing on ἕρξῃς

Question

Book 2 of the Iliad, line 364, reads:

εἰ δέ κεν ὣς ἕρξῃς καί τοι πείθωνται Ἀχαιοί,

Here ἕρξῃς is the second-person aorist subjunctive of ἔρδω. Some editions spell it with rough breathing (Rouse, wikisource), while others have smooth breathing (Benner, Anthon). Can anyone explain why there is even the possibility of rough breathing here, and how we would even know, given that Homer had to be transmitted orally, and there would be no effect on the meter? Is it a dialectical thing? Is there a whole nother form of the verb, ἕρδω, or would this just be something special about the subjunctive? There is nothing about this in the entries for ἔρδω in Beekes or the Cambridge Greek Lexicon. Cunliffe lists the form with smooth breathing. Could this be some sort of influence of the related verb ῥεζω?

Answer 1

The aspiration appears to be a variant attested in some Homeric manuscripts as well as occasionally in other authors. LSJ s.v. says:

Aspirated acc. to Sch. Ar.Ach.329, and so freq. in codd. of Hom., cf. Thgn.690, Epic. ap. Pl.Euthphr.12a

That is, a scholiast on Aristophanes' Acharnians says the verb is aspirated; it's frequently so spelled in manuscripts of Homer, and also in a line of Theognis and in a Homeric quote in Plato's Euthyphro.

On the rough breathing more generally, and how we might know if it occurred in a Homeric form:

Some Greek dialects were psilotic, that is, they lost the sound [h] in all positions. This is generally true of Ionic, at least in the Classical period. Whether or to what extent it was true of Homeric Greek -- which is largely but not wholly based on Ionic, but on the Ionic of an earlier period -- seems to be unclear. The text of Homer that we have passed through an Attic "recension" -- that is, it was written down in Athens (traditionally in the reign of the tyrant Peisistratus, 6th century BC, for performance in the Panathenaic Games) -- and this introduced occasional Attic dialect features into the poems. Breathings hadn't been invented yet at this point, though; these were introduced some three centuries later. When Hellenistic grammarians standardized the texts in that later period, they often wrote rough breathings in words that had them in Attic, but not consistently (and occasionally they also wrote them in words that the Attic dialect lacks).

The end result of this is that when a Homeric word has a rough breathing, it's hard to know whether this is due to the grammarians, the Athenian recitation tradition, or the "original" Homeric text. And this is of course complicated by the question of to what extent it even makes sense to talk about an original Homeric text.

On the specific word ἔρδω:

Etymologically, a rough breathing wouldn't necessarily be expected here, but it wouldn't be ruled out, either. As Cairnarvon mentions, the verb etymologically begins with digamma (it's cognate with English work). Initial digamma does not regularly yield [h-] in (non-psilotic) Greek, but there are sporadic instances where it does, for reasons that aren't well understood. This might conceivably be the case here, but it seems unlikely since in the known cases of *w > h-, the rough breathing spelling is consistent, not an occasional variant as in this case. (One could maybe come up with a phonological justification along the lines of *werdyō > *werdsō > *wersdō > *werhdō > *hwerdō > *herdō, using known sound changes, but I don't know of parallel cases and the relative chronology this assumes doesn't seem too secure.)

So there doesn't seem to be a good explanation for the rough breathing in ἔρδω. Chantraine (Grammaire Homérique 1:187-8) lists a number of words that show similar variation, sometimes etymologically justified, sometimes not. Of ἔρδω specifically he says:

Pour ἕρξας et ἕρδω où l'étymologie ne justifie nullement l'aspiration, l'esprit rude sert surtout à distinguer ἕρξα de ἕρδω, d'avec ἔρξα de ἐέργω. Mais nous n'avons à faire qu'à une artifice de grammariens alexandrins.

"For ἕρξας and ἕρδω, where the etymology in no way justifies the aspiration, the rough breathing serves mostly to distinguish between ἕρξα from ἕρδω and ἔρξα from ἐέργω. But this is nothing more than an artifice of Alexandrian grammarians."

That is, he thinks the grammarians introduced the rough breathing here to distinguish this form from a homophone -- obviously an ad hoc explanation, but there doesn't seem to be anything better.

Answer 2

Here's the most exact answer I can find for most cases of these freak aspirations. I can't find any other cases (where the dictionaries don't specifically point out analogical copying of the asper from a similarly shaped word) without this rule-set being the cause, except for one important case.

Initial digamma which isn't already aspirated (like *ͱϝέο οὗ, *ͱϝέοαὐτο͂ ἑωὐτοῦ) takes aspiration when in front of a non-low vowel, where that same non-low vowel is in front of:

1: an interaspiration, including from s-1 clusters with a resonant (these are usually lost elsewhere)

2: a rho before a mute [tenuis,] (ρκ, ρτ, ρπ) EDIT: This might include ρσ as a general rule.

3: a syllable-final primary sigma (debuccalized or not), or

4: any chain of syllables, not interrupted by a plosive consonant or low vowel, that fits the rules above, all the way to the initial. Edit: these syllables must all be barytone, or have been barytone when this phenomenon was in force.

By s-1 or primary sigma, I mean those from PIE before any assibilation of τ, yet also after the PIE dental dissimilations that produced the σ in *ϝοἶσθα from *ϝοἶδθα

This aspiration happened:

-before the expiration of Grassman's law, (so whether an aspirated mute can do this I'm not sure, and can't think of any words that would fit the necessary criteria.)

-after σ caused the devoicing of any plosives that come before them, (*γσ to ξ) and after ζ became σδ (assuming there was a stage where it was ever anything else).

-It must have happened right-during the 1st compensatory lengthening, because in Attic, the s+nasal clusters lengthen ο to produce ω, which blocks this transfer, instead of ου, which shouldn't (but I can't prove that).

So respectively to the conditions above:

1. *ϝεͱμα εἷμα (but *ϝοͱνα ὠνή; ω is too low)

2. ϝερκσηις ἕρξῃς (but *ϝεργϳω *ϝερσδω ἔρδω: ζ is voiced)

3. *ϝιστωρ ἵστωρ (but *ϝοισθα οἶσθα and *ϝεσθητ ἐσθής per Grassman's Law; ϝαστυ ἄστυ because the α is low; *ϝιδμεν ἴσμεν because σ is newly coined from analogy, and because the rest of the paradigm suppresses it).

4. *ϝεϝορτα ἑορτή, (and perhaps *ϝορϝος ὅρος (asper from the final sigma) if applicable)

This explanation has only one (rather huge) gap:

ὁράω, if it is from the root *wer-, must have received its asper from some sort of early analogy, which I would believe is from its own root, because although its group is aspirated, ὤρα resisted it. If they are all cognates, then it probably got the asper from ὅρος. But, if the word is cognate to latin 'servus', (or a syncretism of both roots) then ὤρα is not (fully) related, and both ὅρος and ὁράω are not part of this discussion.

The main source for this is something https://www.academia.edu/88618894/Towards_a_Reconstruction_of_the_Proto_Greek_Nominal_Morphology According to one Rok Kutner from U Ljubljana, called ἵστωρ-ἑόρτη rule.