Does the two-word Greek “until” of Matthew 1:25 (heos hou) imply Mary bore other children?

Type: supporting argument (off-timeline) · Assessment: weak — the refinement scores one fair point (the classic rebuttal’s stock parallels really are the one-word heōs, not the two-word heōs hou of Matthew 1:25), but it does not carry the conclusion: the same two-word phrase is used without any change of state in the very Koine the Evangelist wrote (Septuagint Psalm 72:7, read exactly so on this verse by Chrysostom, whose own text of it is heōs hou) and inside the New-Testament sample itself (Matthew 18:34), and a fifteen-instance sample cannot establish an exceptionless rule against a ~5–14% base rate of non-terminating use. The grammar leaves Matthew 1:25 neutral, exactly as the one-word form does · advanced by Helvidius (in its modern heōs hou form by Eric Svendsen and James White)

The claim

— the thesis of Eric Svendsen, Who Is My Mother? (2001), as he states it.¹ This is the modern, linguistically-armoured form of the oldest Helvidian proof-text. Helvidius (c. 383) pressed Matthew 1:25 — Joseph “knew her not till (heōs hou) she had brought forth a son” — as implying that he knew her afterward. The classic reply (Jerome’s, weighed on the companion page “Do the infancy phrases imply siblings?”) was that Scripture uses “till” without implying a change after. Svendsen’s refinement concedes that the bare word “till” can do so, but argues that Matthew 1:25 does not use the bare word: it uses the two-word phrase heōs hou, and that this narrower construction, across the New Testament, always marks a terminus after which the state changes — so a change of state (Joseph knowing Mary) is implied here too.

Who advances it

The argument is a modern reconstruction, not a patristic one. Its scholarly form is the doctoral work of the Reformed New-Testament scholar Eric Svendsen (Who Is My Mother? The Role and Status of the Mother of Jesus in the New Testament and in Roman Catholicism, 2001, developed from his dissertation), and it was carried to a wide popular audience by James White of Alpha and Omega Ministries.² Both grant, against the older and cruder Helvidian handling, that the one-word heōs is genuinely neutral; the whole weight of the case rests on the claim that the two-word heōs hou is not. It therefore stands or falls on a corpus question — what the phrase heōs hou, specifically, does across its occurrences — and is best weighed as one.

The evidence for

The refinement begins from a real observation. Jerome’s stock counter-examples — Michal who “had no child unto the day of her death” (2 Samuel 6:23), the raven that flew “until the waters were dried up” (Genesis 8:7), the risen Christ “with you… unto the end of the world” (Matthew 28:20) — all render the one-word heōs (or heōs + noun), not the two-word heōs hou of Matthew 1:25. To that extent the classic rebuttal, as usually stated, does not meet the exact construction it is answering, and Svendsen is right to press the distinction.³

On top of that distinction Svendsen sets a corpus survey. The phrase heōs hou occurs seventeen times in the New Testament; two mean “while,” and of the remaining fifteen he judges that each marks a terminus at which the main clause ceases or changes — the leaven is hidden “till the whole was leavened” (Matthew 13:33), the vision is to be told no one “until the Son of man be risen” (Matthew 17:9), the conspirators will not eat “till they had killed Paul” (Acts 23:12), and so on — none, on his reading, meaning “until, and continuing after.” He confines the sample deliberately to Koine of roughly Matthew’s own period (c. 100 BC–AD 100). Read against that pattern, “he knew her not heōs hou she brought forth a son” falls into line: the not-knowing terminates at the birth, so Joseph knew her after — which coincides with the ordinary-language intuition that “he did not dine until she came” implies that he dined once she came. So stated, the argument is not frivolous.²

Adversarial assessment

What holds. The correction of the classic rebuttal is fair and worth conceding: the two-word phrase is not simply identical to the one-word one, and Jerome’s flagship parallels are indeed the bare heōs. It is also true that, across the New Testament, heōs hou most often does mark a terminus after which things change. An honest weighing starts by granting both points — and by granting that the stake is high, for if the terminative reading held it would touch the perpetual virginity directly (marital relations after the birth), not merely by the roundabout route of “firstborn” and “brethren.”

What is overstated. But four difficulties, taken together, keep the phrase neutral and sink the inference.

(1) The same two-word construction is used non-terminatively in contemporary Koine. The argument needs heōs hou — not merely heōs — to behave terminatively. Yet the Septuagint, roughly contemporary Koine and the very idiom the New-Testament authors imitate, uses heōs hou the other way: “in his days shall… abundance of peace heōs hou the moon be taken away” (Psalm 72:7, LXX) — where the peace plainly does not cease when the moon is removed but stands for perpetuity — and likewise “he shall not be afraid heōs hou he see his desire upon his enemies” (Psalm 112:8) and “the Lord will not cast off his people heōs hou righteousness return to judgment” (Psalm 94:14–15).⁴ The sharpest witness is a native Greek reading Matthew himself. John Chrysostom (c. 390), whose own text of Matthew 1:25 reads heōs hou, expounds that very phrase and denies precisely the inference the argument draws: “He has here used the word till, not that you should suspect that afterwards he did know her, but to inform you that before the birth the Virgin was wholly untouched by man.” He explains the usage — “it is usual in Scripture often to do this, and to use this expression without reference to limited times” — and glosses it by citing, among his parallels, Psalm 72:7’s “until the moon be no more,” which is itself heōs hou. A native speaker of the construction, on the disputed verse, saw in the two-word phrase no terminative rule whatever.⁵

Svendsen’s answer is to rule this evidence out of court: Septuagint and patristic usage, he says, is “irrelevant” to fixing the sense of the phrase in Matthew’s own day, since to read the wider corpus into the New Testament would commit the “semantic obsolescence” fallacy that D. A. Carson warns against. But the principle is misapplied. Semantic obsolescence forbids reading an earlier, obsolete sense forward across a demonstrated shift in meaning; no such shift has been shown here. The Septuagint is contemporary Koine, Chrysostom is later Koine, and to make the exclusion work the claimant must suppose that heōs hou meant “until (possibly continuing)” in the Septuagint, then narrowed to “until (never continuing)” across just the fifteen New-Testament verses, then broadened back again by Chrysostom — a swing for which there is no evidence and against which the burden lies. Absent that showing, the contemporary usage is not irrelevant; it is the best control available, and it says the phrase can continue.⁶

(2) A non-terminating case lies inside the New-Testament sample itself. In the parable of the unforgiving servant the master “delivered him to the tormentors, till (heōs hou) he should pay all that was due” (Matthew 18:34). The debt — ten thousand talents — is unpayable, so the imprisonment never ends: Calvin reads the verse as pointing “not temporal death… but eternal death,” and Wesley likewise took it of a punishment never discharged. There the terminus is never reached and the main clause does not cease — a non-terminating heōs hou among the very fifteen.⁷ Two further instances are at least soft: Calvin flags the difficulty of “until the day dawn” (2 Peter 1:19), extending the “darkness” that needs the prophetic lamp “to the whole course of life”; and Paul “kept in custody heōs hou I send him to Caesar” (Acts 25:21) reads naturally as custody that continued after the sending — Svendsen must gloss “kept” as “kept here,” a possible but strained reading — so “each of the fifteen terminates” is not even secure on its own ground.⁸

(3) The sample is too small to license the rule. Set even the counter-examples aside. The Septuagint uses heōs hou on the order of eighty-five times, of which a real share — by Svendsen’s own tabulation, seven or eight — do not terminate. Project that base rate onto only fifteen New-Testament instances and one expects one or two non-terminating uses; so even a clean fourteen-of-fifteen would leave Matthew 1:25 sitting comfortably as the statistical exception — and Matthew 1:25 cannot be counted among the confirming fifteen without begging the very question at issue. The move “the other fourteen terminate, therefore this one does” is precisely the base-rate error.⁹

(4) At its very best it would fall short of children. Grant the terminative reading and it yields, at most, that Joseph and Mary had marital relations after the birth. That would tell against the perpetual virginity — but it is a step short of the Helvidian thesis this argument is enlisted for, that the “brethren” were Mary’s own later-born sons. Richard Bauckham — no Helvidian — states the limit from the side least friendly to the doctrine’s deniers: Matthew’s words “cannot be pressed to imply that Joseph and Mary had sexual relations after the birth of Jesus, but even if they could it need not follow that they had children.” And the phrase’s plain function in Matthew is to secure the virginal conception (Matthew 1:23), not to report a later marriage.¹⁰

Conceded even by a defender. That the grammar leaves the verse neutral is granted from the side that most wants the perpetual virginity to be true. Basil of Caesarea (c. 377), preaching on this very verse, allows outright that Matthew 1:25 could be read as implying that Mary “did not subsequently refrain from normal conjugal relations,” and that such a conclusion “would not have affected the teaching of our religion at all” — then rests the ever-virginity not on the grammar of “till” but on the consensus of “the lovers of Christ.” A committed holder of the doctrine thus treats the two-word “until” as grammatically open, exactly the neutrality argued here; the doctrine, if it stands, stands on other ground than this proof-text.¹¹

Assessment

Weak. As a proof that the Greek of Matthew 1:25 implies Mary bore other children — or even that she and Joseph later had relations — the argument does not hold. It earns one fair correction: the classic parallels are the one-word heōs, and the companion page on the infancy language notes as much. But the conclusion does not follow. The two-word heōs hou is used with no change of state in the very Koine the Evangelist wrote — Psalm 72:7, read exactly so on this verse by a native-Greek father whose own text of it is heōs hou — and with no change of state inside the New-Testament sample itself (Matthew 18:34), and a fifteen-instance sample cannot establish an exceptionless rule against a base rate of non-terminating use that predicts one or two exceptions in a set that size. So the grammar of Matthew 1:25 is neutral, exactly as the one-word form is: the phrase denies a union before the birth and is silent on what followed. The heōs hou refinement adds sophistication to the old Helvidian “until” proof-text without adding probative force; whatever weight the Helvidian reading carries must still come from the plain “brethren,” not from the two-word “until.” The overreach to flag is the a-priori dismissal of the Septuagint and the fathers as “irrelevant”: that is a valid linguistic principle deployed where its precondition — a demonstrated shift in the phrase’s meaning — has not been met.

Sources