Rid Us (Not) of the Temptation: A Note on the Text of Hebrews 11.37

In his 1989 article, Eldon J. Epp famously raised a series of incisive questions – indeed, one might almost say implicit complaints – concerning the impact of the papyri on modern textual criticism. He asks:

If Westcott-Hort did not utilize papyri in constructing their New Testament text, and if our own modern critical texts, in fact, are not significantly different from that of Westcott-Hort, then why are the papyri important after all? And have they really “come into their own” and been fully and appropriately utilized in the text-critical discipline? (1989: 285)

In over 30 years since, much has changed in the discipline, particularly in regard to methodological and editorial advancements: the development of the Coherence-Based Genealogical Method (CBGM), the resurgence of interest in early Christian manuscripts and material culture, and the steady flow of Editio Critica Maior volumes are just a few of the most important ones to be mentioned. But even by the time of Epp’s writing, the papyri had been instrumental in various changes to the ‘standard’ critical text. Though rarely introducing hitherto unknown readings, the papyri have given us an unprecedented glimpse into the earliest strata of textual transmission, thus often furnishing the earliest support for variants already known in other, often much later, streams of textual tradition. Such newly provided early support could then increase the critic’s confidence concerning variants whose external attestation might be less impressive, be it in age or extent.

One such noteworthy instance is at Heb. 11.37. Here, prior to the publication of P ⁴⁶ (P.Beatty II; LDAB 3011) all the earlier critical editions read either ἐπρίσθησαν ἐπειράσθησαν or ἐπειράσθησαν ἐπρίσθησαν as their main text, thus incorporating an experience of being ‘tested’ or ‘tempted’ into the extended list of ill treatments borne by the faithful. ¹ Even so, ever since Erasmus this passage has received a fair share of emendation due to a supposed ill-suitability of ἐπειράσθησαν, as we shall see below. On that reading, the two execution methods (ἐλιθάσθησαν and ἐπρίσθησαν) would follow each other immediately on the list, thus making for a smoother transition. Now, P ⁴⁶ provided (by far) the earliest manuscript support for the omission of ἐπειράσθησαν, a reading whose antiquity had only been established via patristic sources. And, indeed, beginning with the UBS Greek New Testament 3rd edn and Nestle-Aland 26th edn, the ‘standard’ text has shifted its wording to ἐπρίσθησαν. ² Nevertheless, the primacy of the shorter reading has not been established beyond reasonable doubt; indeed, some of the previous assertions concerning intrinsic and transcriptional probabilities are proving problematic, particularly in view of our current understanding of scribal behaviour. In what follows, then, we shall consider this textual problem more closely, offering an alternative explanation of the internal evidence.

Let us begin with the external evidence. ³ The Nestle-Aland 28th edn (henceforth: NA²⁸) presents the following variants at this point:

Evidently, the broadest and most diverse pool of witnesses attests the two longer readings. ⁴ On the one hand, ἐπειράσθησαν ἐπρίσθησαν is found in Codex Sinaiticus (GA 01; LDAB 3478), followed by a number of significant majuscules and minuscules, as well as the Harclean Syriac and one Bohairic manuscript. On the other hand, the external support for the variant ἐπρίσθησαν ἐπειράσθησαν is even more substantial: not only is it read by the Byzantine mainstream text, but also a rather diverse battery of early Greek and versional witnesses – the support of P ¹³ (P.Oxy. IV 657 + PSI XII 1292; LDAB 3018) being particularly significant. Importantly as well, the variant is attested by the entire Latin tradition, coupled with the Bohairic; besides, it appears to have been known to Origen.

Now, ἐπρίσθησαν, the current initial text, is read by the aforementioned P ⁴⁶, a twelfth-century minuscule 1241 (Mt Sinai, Saint Catherine’s Monastery MS Gr. 260), ⁵ Peshitta, Sahidic, ⁶ as well as Eusebius and (partly) Origen. ⁷ The support of P ⁴⁶ is significant, but ought not to be overrated: though by the sheer virtue of its antiquity this papyrus is a witness to an early form of text, ⁸ its scribe’s performance was marred by frequent errors at various levels, as we shall see below.

As for the second shorter reading ἐπειράσθησαν, no variant in this form, in fact, ever occurred in the transmission history. As it is, GA 0150 (Patmos, St. John the Theologian Monastery MS 61), which NA²⁸ cites for this variant, does not omit ἐπρίσθησαν but transposes it after ἐν φόνῳ μαχαίρης ἀπέθανον. At any rate, it is clear that the reading of 0150 derives, directly or indirectly, from a Vorlage with a longer reading, possibly ἐπρίσθησαν ἐπειράσθησαν, as it seems plausible that the scribe initially skipped over ἐπρίσθησαν by parablepsis and corrected himself in the course of copying. The Text und Textwert lists no further witnesses for the reading ἐπειράσθησαν – it does not list a reading in this form at all, only the above-mentioned singular transposition in 0150 ⁹ – and hence the support of some Vulgate manuscripts and Clement of Alexandria ought to be regarded as non-genetic. ¹⁰ Indeed, had not this reading entered the NA²⁸ apparatus by mistake (as seems probable), there would be no Greek variant to cite their support for. ¹¹

Even though the external evidence appears to lean more favourably towards the two longer readings, the value of the quantity of witnesses ought not to be overstated. More importantly, we need to establish which of the relevant variants best explains the origin of the others. Let us begin with ἐπρίσθησαν, the reading favoured by most modern text-critics. ¹² A strong case for this variant has been made by Günther Zuntz. His observations are worth quoting in full:

ἐπειράσθησαν is shown, both by its varying position and by its unsuitedness to the context, to be a corrupt dittography of the original ἐπρίσθησαν. The reference to the legend of the martyrdom of Isaiah was recognized already by Origen and Jerome; it leaves no room for the idea of ‘temptation’. There is cause for legitimate surprise in the all but general preservation of this scribal slip, which is in fact an indication of the very early origin of the corruption; besides one may well wonder at the persistent attempts of commentators to extract some meaning from it, some of them even trying to emend the text by conjectures based upon the ancient dittography. (1953: 47)

Zuntz’s argument is forceful, especially from the standpoint of intrinsic probabilities: a torturous death of being sown in two fits the context more intuitively than being tempted, and omitting ἐπειράσθησαν does smooth out the most immediate transition between ἐπρίσθησαν and ἐν φόνῳ μαχαίρης ἀπέθανον, which also denotes a violent form of death. Even so, the notion of ‘temptation’, particularly in the sense of trials and tribulations is not so foreign a concept as to be deemed unsuitable here – not least in the context of martyrdom. Moreover, Zuntz’s line of reasoning would have had greater gravity, had the only competing variant been ἐπειράσθησαν. But the two most widely attested readings combine ἐπειράσθησαν with ἐπρίσθησαν, a combination which, in fact, can be construed meaningfully in the immediate context, particularly since this list is not comprised of execution methods exclusively: besides the death by stoning, sword and sowing in two, there are references to mockery, whipping, bonds and imprisonment, walking in sheepskins and goatskins, being needy, afflicted and maltreated. ¹³ When we consider the passage in toto, the reference to being tempted appears less ‘unsuited to the context’ than Zuntz would seem to allow. ¹⁴ Exposure to trials of various kinds was in the very essence of the ‘flip side’ of living by faith, when faithfulness to God did not result in one’s overcoming the present unfavourable circumstances but rather suffering through, or even succumbing to, them, while holding fast to an eschatological hope (Heb 11.35, 39). ¹⁵

Where Zuntz’s argument fails to impress even more so is in his handling of transcriptional probabilities. For one thing, what we have here is not a dittography: while the two words do exhibit remarkable graphic similarity, they are at any rate noticeably different; even more so, postulating a corrupt ancient dittography seems rather far-fetched. ¹⁶ Typically, dittographies are just that – they repeat an exact sequence of letters due to parablepsis, or, more specifically, a scribal leap backwards. But this is not the case at Heb. 11.37: the elements that are identical here are the first two and the final six letters. Unless one is ready to explain the error as a dittography with a concomitant metathesis of ιρ and addition of ε and α – surely, an implausibly convoluted scenario – this line of argument ought to be dismissed. All in all, then, though ἐπειράσθησαν is, perhaps, not the most intuitively placed component on the list, its comparatively greater difficulty combined with the tenacity with which it was transmitted in divergent streams of the tradition ought to place the burden of proof on those who postulate the priority of its absence.

By contrast, from the perspective of transcriptional probabilities, the omission of ἐπειράσθησαν may be explained relatively easily. The earliest and most important manuscript that attests ἐπρίσθησαν is P ⁴⁶. While the antiquity and text-historical importance of this manuscript bolsters the external evidence for this reading considerably, P ⁴⁶’s known proclivities should make us wary of placing undue weight on its testimony. Indeed, in his extensive analysis of scribal habits, James R. Royse observes that the ‘scribe [of P ⁴⁶] has a very marked tendency to omit portions of the text, most often only one word but longer phrases also. Some of these are due to scribal leaps, but most seem to have arisen from simple oversight or carelessness’ (2008: 358). Elsewhere he notes that ‘almost all of the longer omissions and quite a few of the shorter omissions may be attributed to scribal leaps’ (283), such that omission by leaps was ‘one of the scribe’s pervasive tendencies’ (289). ¹⁷ In view of this, regarding Heb. 11.37 Royse states: ‘the evidence of P ⁴⁶ must be weakened once we see its tendency to omit by homoeoteleuton’ (291). Incidentally, the same doubt that looms over P ⁴⁶ may be extended to 1241 – and, indeed, to other late manuscripts not cited in NA²⁸ – as a leap of this type could easily have happened on numerous occasions independently, so that there seems little reason to regard their agreement as genetic. ¹⁸ Now, this is not to say that it must have been the scribe of P ⁴⁶ that made the omission. Indeed, the omission may have occurred quite early on, and was subsequently carried through the transmission. What the study of scribal habits demonstrates, however, is that the ubiquity of omission by parablepsis – particularly, but not solely, in the early period, and well beyond the New Testament tradition ¹⁹ – makes the shorter reading a less compelling option.

As regards the mechanics of the omission, it would seem that the reading ἐπρίσθησαν most likely arose by omission of ἐπειράσθησαν that, originally, preceded it; thus, we have a scribal leap occasioned by homoeoteleuton (ἐλιθάσθησαν ἐπειράσθησαν ἐπρίσθησαν). Alternatively, the scribe may have begun writing ἐπρίσθησαν but, having written επ, either lost his place in the exemplar or paused to re-ink his pen, ²⁰ and skipped over to the -σθησαν of the following ἐπειράσθησαν. Either way, the origin of this reading is best explained on the basis of one of the two longer readings.

This leaves us with the two longer variants. And here it seems much more difficult to reach a clear-cut decision, as the arguments concerning internal evidence may proceed both ways. For instance, Bernhard Weiss observes that ἐπειράσθησαν was placed after ἐπρίσθησαν so as to bring the latter into greater proximity with ἐλιθάσθησαν, so that ‘these two gruesome modes of death would not be separated by the altogether foreign ἐπειράσθησαν’ ²¹ (1896: 136). Even so, the scribes need not have proceeded in such an exegetically attuned manner. In fact, it is just as likely that the words were swapped due to a correction of an incipient leap – if not more. ²² And, contra Weiss, one might further argue that, actually, it was the author himself who grouped the semantically relevant words together, with the order getting reversed inadvertently by scribal error. In my view, it is the latter explanation that fits better with the well-documented tendencies of our scribes – hence, I would suggest that ἐπρίσθησαν ἐπειράσθησαν is the initial text at this point of variation.

Naturally, in due course, proper genealogical investigations will have to be made in order to substantiate (or repudiate) the present proposal. That line of inquiry, however, will only be possible once we have full electronic transcriptions of all the relevant witnesses as well as the local stemmata for the entire text of Hebrews. It is hoped that, at this particular passage, the present study will be of some use towards that latter end. At a more general level, I hope to have demonstrated that, notwithstanding the value of the early papyri, their support eo ipso does not warrant priority of the variants they might attest. Rather, careful consideration of transcriptional probabilities, informed by the recent advances in the study of scribal proclivities, ought to take precedence in establishing local genealogical relationships over external factors, including the antiquity of our papyri. For it is precisely in the study of scribal behaviour that one can attain adequate ‘knowledge of documents’ and so proceed to make ‘final judgments upon readings’, as Hort’s dictum continues to remind us. ²³