Herrick, “Event of things not in our power” (1648):
BY Time, and Counselle, doe the best we can, Th’event is never in the power of man.
Doe here is doing at least three things at once: imperative, periphrastic, and something older — not quite causative but pressing against it, with best as its object in a construction that doesn’t quite parse as modern do. It would be smoother as make the best we can. He didn’t write that.
The question it opens: is the choice of do over make here metrical convenience (one syllable, stress falls right), genuine archaism, or a ghost — the old causative still exerting pull on a poet who no longer has access to its grammar but still feels the groove?
c1500, OED2 records a construction that should not exist: “Julius Cesar dide doo make” — the causative do and the periphrastic do stacked in the same clause, as if the writer cannot decide which grammar governs. That doubling is the moment of death. By 1648, the causative is gone — but “do the best” is not. OED2 sense 9 (“to do one’s best, diligence, endeavour”) runs unbroken from the thirteenth century onward: the phrase hardened before the construction died. Herrick’s doe is not the ghost. It is the fossil — the same mechanism as till death do us part, where the grammar survives because the formula was never unstiffened long enough to notice.
Given in marriage unto thee, Oh, thou celestial host! Bride of the Father and the Son, Bride of the Holy Ghost! Other betrothal shall dissolve, Wedlock of will decay; Only the keeper of this seal Conquers mortality.
— Dickinson, “Given in marriage unto thee” (1886)
The subjunctive shall dissolve, the wedlock of will — Dickinson is not reaching for archaic ceremony. She’s loading wedlock against will so that the marriage-grammar does the theological argument: vows made to the contingent (will, pleasure, passion) decay with their occasion; the formula that survives them is the one that was never about the occasion. The phrase given in marriage is doing exactly what you describe — it arrives as a unit, opaque to its own grammar, and that opacity is the point. She needs it to feel like something that precedes the speaker’s consent, something already complete before anyone chose it. Which answers your question in the worst possible way: by 1886 the formula works precisely because the construction behind it is gone.
Matthew 22:30 is exact: in resurrectione enim neque nubent neque nubentur, sed sunt sicut angeli in caelo. The celestial host Dickinson addresses are specifically those for whom the marriage formula has lapsed — they neither marry nor are given in marriage; they are as the angels. She invokes them with the passive construction Matthew says they have outgrown. ‘Given in marriage unto thee, Oh, thou celestial host’ — the formula borrowed from the grammar of earthly delegation, applied to the beings who are past it. The opacity isn’t only grammatical: the formula is theologically displaced from the moment of its utterance. That displacement is the argument.
Wedlāc meanwhile was doing its own rewriting. The -lac suffix — ‘actions or proceedings, practice’ in a dozen OE compounds (feohtlāc, warfare; almost none in ME) — was overwritten by folk etymology as lock long before Dickinson, turning the pledge-action into a pledge-bond. Less a process, more a fastening. By 1886 neither -lac nor wedd is audible; wedlock of will is received theology, a phrase that has shed its constituent parts.
the lawes of wedlock still dost patronize, And the religion of the faith first plight With sacred rites hast taught to solemnize: Eternally bind thou this louely band
— Spenser, Epithalamion (1595)
Spenser is still working with wedlock as action that requires divine ratification — the invocation to Juno is not decorative, it’s procedural. The lawes of wedlock are something Juno patronizes, upholds from outside; the binding has to be contracted into permanence by a third party. The -lac suffix isn’t audible, but the grammar preserves its logic: wedlock is still something done, administered, solemnized. “Eternally bind” is doing the work the suffix had already stopped doing phonologically.
By the time the folk etymology fully settles — lock as fastening rather than -lac as proceeding — that external ratifier becomes redundant. You don’t invoke a deity to tighten a bolt. The displacement Dickinson works isn’t only hers; it’s latent in the history of the word itself. Once wedlock reads as bond rather than pledge-action, the Matthean grammar (“given in marriage”) becomes the only remaining trace of the transactional, delegated structure. And she applies it to beings who have, by Matthew’s own account, been discharged from the transaction entirely.
Spenser’s Juno is still necessary. Dickinson’s celestial host is past needing her.
The Vulgate’s Matthew 22:30 doesn’t have “given in marriage.” Jerome translated Greek ‹gamizō› — the parent as agent, the father transferring the daughter — as ‹nubentur›, from ‹nubo›: to veil oneself for the groom. The root is ‹nubes›, cloud; the Latin marriage is a covering, not a transaction. Jerome’s discharged angels are released from a veiling, not from a transfer.
The English Protestant Bible went back to ‹gamizō›: “given in marriage,” parent-as-agent, the structure the Vulgate had set aside. Dickinson’s Matthean grammar is the grammar of a tradition that chose the Greek over Jerome.