The Roll,the Codex,the Wax Tablet and the Synoptic Problem

Downing versus Goulder

A number of scholars have charged that the FH envisions a compositional procedure for Luke that is problematic to imagine on the terms of the ancient technology of writing—that is, that the difficulty of simultaneously handling two texts, moving through rolls, etc., causes certain aspects of the FH to fail from a lack of historical imagination. Downing, surveying the study of ancient compositional practices in general, notes a consensus among classical scholars that writers used only one source at a time. Any microstructural conflation that occurred along the way should be put down to the effects of memory. ² Downing also claims—ostensibly on the authority of the same scholars—that a writer seeking to ‘reconcile’ two sources first would have searched out their agreements (apparently meaning their verbatim agreements), and then used that common material as the core for a new account. According to Downing, the fact that Goulder’s Luke does not work in this fashion would make him unique among ancient writers. Goulder’s Luke, instead of comparing sources to isolate their agreements, has (according to Downing) done just the opposite: he has isolated the agreements, but in order to ‘unpick’ them rather than to preserve them.

There are three problems with Downing’s case against the FH, which I will discuss in varying detail: (1) his depiction of the physical difficulty of working with two separate rolls is a caricature; (2) he has misrepresented classical scholarship, as well as Goulder’s understanding of how Luke works with the texts of Mark and Matthew; ³ and (3) those elements of Downing’s case that are still valid after correcting for (1) and (2) are not binding on the FH in the abstract, but apply only to Goulder’s characterization of Luke’s motives (especially his belief that Luke is ‘reconciling’ Mark and Matthew)—a characterization widely rejected by fellow proponents of the FH. After discussing these aspects of Downing’s critique of Goulder, I will turn to some widely shared complaints about Goulder’s ‘reverse-scrolling’ thesis. Then I will discuss how our knowledge of ancient writing postures should impact the debate over the synoptic problem, and finally I will address the question of Luke’s writing medium.

Downing’s complaints against Goulder’s depiction of Luke’s procedure are often so pedantic that they do not pack any real force. He problematizes every small detail of Luke’s physical circumstance as a writer. He even problematizes the notion of bookmarking a roll (based on Goulder’s reference to Luke’s having ‘a marker at [Mk] 3.19’ [see Goulder 1989: 346]), citing the fact that we have no discussion within classical writings of such a procedure, and complaining that a lack of space between words disallowed any sort of mark to be made within the text itself. ⁴ But why would the use of scriptio continua matter? Downing seems to think that Goulder’s scheme of moving back and forth between rolls calls for a means of marking one’s exact place, rather than one’s page (or, strictly speaking, ‘column’). (This in itself is odd, given how rarely today’s readers feel the need to mark their place at a more exacting locus than the page.) Marking one’s page (or column), however, is about all that Goulder’s scheme requires. It should also be noted, however, that other writers apparently had room on their tables for two source rolls, and it is unnecessary to follow Goulder too literally when he refers to a roll being left temporarily ‘on the floor’. ⁵ Juvenal, for example, referred to Cicero’s Second Philippic as that which ‘you unroll next to the first’ (volveris a prima quae proxima [Satires 10.126]) (see Small 1997: 14). This part of Downing’s critique appears to be based on misunderstandings of some basic concepts.

It must be said that a reference (whether by Goulder or by Downing) to Luke’s ‘rolling up’ a roll before placing it on the floor is worded unfortunately, as readers might easily envision a roll being completely rewound to its beginning in the act of ‘rolling up’. Goulder’s reference to ‘rolling up’ a roll, however, more likely meant simply bringing its coils or spindles together—an act that preserved one’s place within the roll in a way roughly equivalent to the modern reader’s turning a book spine up on a desktop. Downing explicitly allows for this, although he is not necessarily convinced that it is what Goulder meant (1992: 18). As noted above, the level of place-marking that Downing requires (viz. exact location) is a level that has seldom obtained within the world’s print cultures, and is probably not one that Goulder implied. In spite of allowing for a more practical understanding of what ‘rolling up’ a roll means, Downing ridicules Goulder’s suggestion that Luke could pick up a roll from the floor and find the place where he left off as ‘apparently so simple an action’ (1992: 20). When understood according to what ‘rolling up a roll’ really means, however, it is a very simple action, as simple as opening a codex to a bookmarked page. ⁶ The suggestion that Luke alternates between Mark and Matthew hardly represents a case of ‘los[ing] all touch with first-century reality’, as Downing claims (1992: 20), and Luke really had no need to mark his place other than by simply rolling the spindles together. And things would become all the easier for Luke if we allow him as much table space as Juvenal assumes his readers to have.

Downing’s exaggeration (1992: 21) of the difficulty of working with two rolls also does not help his case:

If Luke is doing his own writing, rather than dictating, he is balancing a springy new scroll on a board on his knees, together with a pen, and somewhere to hand are his scroll of Mark, and a sharpener, a pricker or other line-marker, ruler, dividers, sponge, bowl of water, pumice and ink (perhaps some sand?)… It is not the sort of circumstance to encourage the invention of a complex new way of composing a text from sources.

The most immediate question, perhaps, is whether such a scenario can problematize the FH without simultaneously problematizing the 2DH. ⁷ On the terms of the 2DH, Matthew was faced with much the same to-ing and fro-ing through his copy of Q. Is Streeter’s Matthew’s new roll any less ‘springy’, or is his writing kit any less handy, than that of Goulder’s Luke? The reader might insist that I am missing Downing’s point, as the payoff for problematizing the physical circumstance of Luke’s procedure is not based solely on the encumbering nature of that circumstance, but rather on combining it with the supposed need for Goulder’s Luke to invent ‘a complex new way of composing a text from sources’. But the fact of the matter is that Goulder’s Luke does not invent anything new at all—his procedure was in all respects in keeping with accepted ways of writing from sources in the ancient world. The compositional stage of Luke’s writing, moreover, was more likely done on wax tablets (rather than rolls), in which case most of the above-listed accessories were not immediately needed (see below).

Downing’s main complaint about Goulder’s Luke is that his way of using Mark and Matthew is (according to Downing) totally unlike the way other ancient writers combined overlapping sources. As classical scholars have shown, authors typically do not conflate their sources at a microstructural level, but rather follow one source at a time (Downing 1988, 1992), filling it out with material from other sources. This proceeding, Downing claims, is at odds with how Goulder’s Luke weaves his narrative. After (purportedly) describing how Goulder’s Luke works with his sources, Downing (1992: 35) concludes, ‘[W]e find an unquantifiable cluster of surreptitious hypotheses about an apparently fairly ordinary first-century writer’s ability and willingness to ignore simple and rational contemporary compositional conventions and scribal techniques and pioneer new, complex, and self-defeating ones all of his own’. This characterization of Goulder’s views is neither fair nor factually correct. As Ken Olson has shown (2004: 138-46), Goulder’s Luke does not conflate his sources at the microstructural level, as Downing alleges. In fact, he acts in accordance with the normal practice among ancient historians: he follows one source at a time, and allows his memory to help him conflate his sources. ⁸ Downing faults Goulder’s Luke for allegedly acting in a certain way, but it happens to be the way in which Goulder explicitly insists Luke did not act (see Goulder 1993).

Downing apparently understands Goulder’s use of the word ‘reconcile’ to imply that Luke looks for the common material (as determined by verbatim agreements) between Mark and Matthew, and uses that material as the basis for his own account (Downing 1992: 16; see Goulder 1989: 201-203, 207, 311, 328, 341, 407). Upon determining that Luke does not, in fact, preserve these areas of agreement verbatim, Downing concludes that Luke does not answer to Goulder’s description. This then becomes the basis for Downing’s contention that Goulder’s Luke must have done the opposite—viz. he must have compared Mark and Matthew word for word to find their most significant verbatim agreements, and then deliberately failed to preserve those agreements. As Downing rightly insists, a procedure of ‘unpicking’ shared agreements is at odds with the way ancient writers approached their sources. As Olson points out, however, the idea of comparing two sources word for word in order to preserve the verbatim agreements (which is how Downing thinks writers worked) is equally at odds with the practice of ancient writers. ⁹ Thus Downing essentially faults Goulder’s Luke for not doing what he (Downing) thinks ancient authors typically did, but which, in fact, they did not do, and which, in fact, Goulder does not say or imply that Luke did either. Olson writes,

Contra Downing, Farrer’s Luke appears to be behaving in a manner similar to that of the other ancient authors that Downing surveyed. Like Livy, Polybius, Dionysius, Josephus, and Plutarch, Luke just goes ahead and writes, mostly on the basis of a single prior text in front of him, rather than attempting to locate and reproduce the ‘common witness’ of his sources. ¹⁰

It should be noted that the language of ‘reconciling’ that Downing picks up is characteristic of Goulder, but not of most proponents of the FH. Thus Mark Goodacre distances himself from Goulder at this point: ‘the danger of even using the language of “harmonizing” is evident in F. Gerald Downing’s somewhat strident reaction to this’ (Goodacre 2002: 89 n. 27). Among Farrerians, it is more widely assumed that Luke’s view of Matthew is somewhat critical. The understanding of Luke’s designs inscribed in the title of Eric Franklin’s Luke: Interpreter of Paul, Critic of Matthew (1994) is more representative of how Farrerians think today. ¹¹

On Working Backward through Matthew

So far we have considered Downing’s criticisms of the notion of ‘rolling’ and ‘unrolling’ alternating source rolls, how (he thinks) Goulder imagines Luke to have worked with multiple sources, and the supposed idea that Luke searches out verbal agreements between Mark and Matthew in order to ‘unpick’ them. Although these criticisms are often cited by other defenders of the 2DH, only Downing and Derrenbacker have brought an ancient historiographical lens to focus on a critique of the FH from the perspective of the 2DH. ¹² When it comes to Goulder’s suggestion that Luke composed the middle section of his Gospel by working backward through a roll of Matthew, ¹³ however, Downing is joined in his harangue of Goulder by several others. There appears to be widespread skepticism that any Gospel writer would work backwards through a source, unless, of course, that writer were a ‘crank’ with some strong redactional motivation. Even fellow supporters of the FH have criticized Goulder’s view. Goodacre, for example, calls the idea of Luke scrolling backward through Matthew ‘the most implausible element in Goulder’s thesis on Luke’s use of Matthew’ (2002: 118 n. 23). The point of the present investigation is not to determine whether Goulder’s backward-scrolling idea is ultimately defensible on every ground—it is not ¹⁴ —but only to test whether it is as technically impracticable and historically unparalleled as Downing, Derrenbacker, et al. claim. As the point about reverse-scrolling is strictly academic (viz. intended to address the idea in the abstract), I will try to move through this matter directly. ¹⁵

The idea of reverse-scrolling is widely rejected as difficult and eccentric beyond belief. Derrenbacker (2005: 200) refers to this backward movement through Matthew as ‘peculiar, irregular, and unnecessarily difficult in a technical sense’, while Christopher Tuckett (2011: 40 n. 75) refers to it as ‘challenging in the extreme’. According to Derrenbacker (2002: 79), ‘It would be difficult…working consistently backward through a roll of Matthew, given that scrolls were designed for start-to-finish reading (not necessarily in one sitting), offering the reader sequential access forward (as opposed to random and reverse!).’ I can find no justification for Derrenbacker’s claim (2002: 79) that moving backward through a roll is not ‘technically feasible’, especially when the one using the scroll is concerned with elements of teaching rather than the order of a narrative. ¹⁶ Derrenbacker (2002: 81) refers to reverse-scrolling as a ‘more technically difficult procedure’ than rerolling a scroll and proceeding forward, and states that such a procedure ‘would appear to be quite peculiar in the ancient literary world’. Neither charge can be sustained. Reverse-scrolling is no more technically difficult than forward-scrolling, and neither is it as ‘peculiar in the ancient literary world’ as Derrenbacker thinks (see below).

Derrenbacker offers an analogy from ‘modern audio media’: Goulder’s scenario (Derrenbacker claims) ‘is comparable to recording the song sequence from…another cassette tape in reverse onto a blank audio cassette tape, but recording each individual song forward’ (2002: 80). As I show below, the parts of this analogy that give it the appearance of a technically cumbersome and messy operation do not, in fact, fit the analogy. There is nothing ‘peculiar, irregular, and unnecessarily difficult’ about the idea that Luke, having reached the beginning of Matthew’s Passion Week narrative, should work through a portion of Matthew in reverse fashion. In fact, Goulder (1984: 112) may be right in calling this the ‘obvious move’: given that a roll must be rolled or rerolled to move from end to end, working backward is a more natural procedure with a roll than with a codex.

Moving backward through a roll is not an unknown practice. Annewies van den Hoek (1988: 152) has shown that Clement of Alexandria, in his use of Philo in Strom. 2.5.3–6.4, ‘evidently did not go forward through his scroll of De Posteritate but turned it backwards’. ¹⁷ Van den Hoek (1988: 150) presents the following ‘schematic overview’ of this backward movement:

Strom. 2.5.3 – Post. 18 (paraphrase with literal elements)

Strom. 2.5.4(5) – Post. 20 (reminiscence)

Strom. 2.6.1 – Post. 16 (biblical quotation)

Strom. 2.6.1 – Post. 14 (biblical quotation and continuation)

Strom. 2.6.2 – Post. 7 (quasi-literal citation)

Strom. 2.6.3-4 – Post. 5 (reminiscence)

Van den Hoek also notes three places where Clement follows the order of a work from Philo, but jumps back either at the beginning or in the middle of the order:

Van den Hoek (1988: 215-16) writes,

Occasionally Clement started his borrowing well into Philo’s treatise, then leaped back to the beginning and thereafter proceeded in a steady forward direction. The phenomenon has a psychological or even practical explanation; if one author wants to use another’s text, the book or scroll will be taken and leafed through. In the course of the search for a specific text, the adaptor may become so interested that not only the passage he had in mind but also other material that just struck his eye is taken over. The new discoveries can be developed more extensively than the original objective and may be recorded in a telegraphic style or in brief notes. This procedure is clearly taking place in the backward jumps of the Stromateis. ¹⁸

In light of Clement’s reverse-scrolling, complaints about the supposed uniqueness or technical difficulty of Luke’s working backward through a roll of Matthew really have little place. The technical difficulties of this procedure have been greatly exaggerated, and I see no reason to believe that any ancient writer would have considered it challenging at all, much less ‘challenging in the extreme’. When moving backward through a roll was the more convenient option, writers like Clement of Alexandria were open to doing exactly that, and they were also unencumbered by the technical aspects of the task.

Derrenbacker thinks he detects further problems in the fact that Goulder’s reconstructed proceeding is not a purely backward movement through Matthew. He notes two places within this backward movement where Luke actually moves forward, once about the space of a chapter. I fail to see, however, why this should represent a problem: Luke (in Goulder’s view) was reading a roll, not a ticker tape, and if he did move backward, he easily (and naturally) might have worked with an unrolled portion of text wide enough to account for these forward movements. ¹⁹ Contra Derrenbacker, for Goulder’s Luke to have his ‘eye…fall on’ Mt. 19.26 while consulting 18.23-25 need not involve any ‘forward turning’ of a roll. ²⁰ (Someone working backward through a scroll would presumably make macrostructurally backward leaps, but forward steps between these leaps.) ²¹ Just as van den Hoek (1988: 152) finds Post. 18-20 to ‘function as a unity’ within Clement’s backward movement through a roll of Philo, ²² so also Mt. 18.23–19.26 could ‘function as a unity’ (at least as far as the task of stepping backward is concerned). I see little resemblance between what Luke does, according to this scenario, and Derrenbacker’s analogy (2005: 197) of recording the songs found on a cassette tape in reverse order—surely rolling a roll to reveal the immediately preceding portion of text (which is exactly equal in task to rolling forward to the next portion of text) is not nearly as difficult as rewinding a tape to the beginning of the preceding song (which is far more bothersome than waiting for the next song to play). And I also fail to understand why Derrenbacker differentiates the possibility of working backward through a roll from the possibility of working backward through a codex. As far as I can see, the only significant difference between the two lies in the fact that a roll more forcibly makes its users respect the sequential nature of the text, but that sequentiality works both forward and backward, and is not binding in relation to the varying amount of text that might be open at any time.

As noted above, showing that reverse-scrolling is both historically and technically ‘feasible’ does not amount to a full rehabilitation of Goulder’s scenario. There is still the question of the representativeness of Goulder’s list of Luke–Matthew parallels, about which many questions arise (but which there is not now space to pursue). ²³ Other proponents of the FH are unconvinced by Goulder’s reverse-scrolling thesis, and I readily grant that they are probably right. ²⁴ Critics of Goulder’s work, however, would do well to familiarize themselves with van den Hoek’s detailed study of Clement’s moves through Philo before judging the options available to Luke.

Writing Postures in Late Antiquity

The question of the ancient scribe’s writing posture represents another point of contention in the debate over whether Farrer’s Luke answers to historical reality. Downing challenges both the distinctive aspects of Goulder’s understanding—specifically his references to Luke’s using a ‘table’, and to source rolls going ‘on the floor’ when not in immediate use—and the more general Farrerian idea of working with two sources. Derrenbacker has joined him in this line of critique. As Downing and Derrenbacker rightly emphasize, writers in antiquity did not use writing tables, but rather positioned their writing material on their knees.

Derrenbacker unfortunately fails to differentiate, however, between the availability of writing tables, and the availability of tables for one’s source texts. Scholars often note that an ancient scribe, when writing on a roll, wrote by balancing the roll on his/her knee. But they have sometimes extrapolated from this that the ancients were hindered in the use of multiple source texts, as if the nonuse of a writing table entails the nonuse of any sort of table, lectern, etc. for displaying one’s sources. Derrenbacker (2005: 4), for example, seems to interpret the writer’s supporting his/her writing media on a knee as evidence that tables were not used to support source texts (!):

[W]riting desks are often imagined as the surface upon which the evangelists worked. The later evangelist, whether it be Matthew and/or Luke, or Mark, has ‘in front of him’ his two sources, which he combines into one new written work. This is clearly an anachronistic conception of the Synoptic evangelists, given the earliest literary description and artistic depiction of writing desks comes several centuries after the writing of the Gospels. ²⁵

The question arises as to what (if anything) the unavailability of a ‘writing desk’ has to do with the question of a physical support for one’s sources. ²⁶

It is hardly the case that table surfaces were not used for arranging one’s sources. Kenneth Willis Clark’s suggestion about what went on upstairs from room L30 (the ‘scriptorium’) at Qumran remains the best explanation so far for how the long, narrow table-like structure found there was used. ²⁷ He argues against the suggestion that the table provided a writing surface, but his suggestion for how it was used is enlightening: ‘It contained the necessary equipment for writing such as the ink wells that have survived. It may be supposed that it held an exemplar also’ (Clark 1963: 70). ²⁸ This supposition about the surface holding an exemplar is fairly safe, in my view, as it is doubtful that the ancients would build a table of such dimensions (16 ft. × 16 in.) primarily for inkwells and knives. (Clark notes that this scenario is supported by Byzantine depictions of the evangelists using tables and lecterns to hold their exemplars.) In fact, one of the essays that Derrenbacker cites refers to the depiction of this scenario in Byzantine manuscripts: according to George M. Parássoglou (1979: 15), the evangelists ‘use ἀναλογεε̑α or ἀναλόγια to support…their exemplar’. This scenario is further supported by m. Kel. 24.6: ‘There are three kinds of חויסיסב: that which lies before a bed or before scriveners which is susceptible to midras-uncleanness; that of a side-table which is susceptible to corpse-uncleanness; and that of a cupboard which is not susceptible to any uncleanness’ (trans. Danby). Although Bruce Metzger follows Jastrow’s Dictionary in viewing a סיסב as a footstool (Metzger 1968: 135; see Jastrow 1971: 178-79), a meaning of ‘table’ is more likely in this case, as the סיסב ‘which lies before a bed’ likely refers to the sort of small tables that regularly accompanied ‘table-beds’ (stibadia), especially in Roman villas. ²⁹ A סיסב that lies ‘before scriveners’ is likely to be something similar, especially given that the Mishnah’s sentence construction implies taxonomic parity between it and the סיסב lying ‘before a bed’. ³⁰ Regardless of what the unusual table-like structure above room L30 at Qumran was—and here it must be admitted that many scholars have expressed their doubts about the matter—there is no dearth of evidence for the use of tables to support source texts.

We already saw above that Juvenal refers to Cicero’s Second Philippic as a volume that one ‘unroll[s] next to the first’. I take it that that does not mean that both rolls are open at the same time. It is more likely that Juvenal’s constructed reader, having finished reading the First Philippic, will have closed it by rolling its coils together—not rerolling it to the beginning—before unrolling the Second Philippic just to the right of the First. Goulder’s reference to one of Luke’s sources going ‘on the floor’ is therefore unfortunate, even if he perhaps did not mean it so literally. My point is that it would not require a large surface to accommodate two rolls, as opening one roll probably implied first closing the other. It bears repeating that ‘closing’ a roll did not imply losing one’s place. It is thus not difficult to imagine Luke working with two rolls or codices on his table simultaneously, but that does not mean that he had complete simultaneous access.

This is perhaps the best place to ask whether dictation played a roll in Luke’s writing process. If Paul had an amanuensis, we should at least consider the possibility that Luke had one as well. Luke might have had a servant dictate his source rolls to him, or, alternatively, he might have read his source rolls and dictated his own improved version of these sources to a servant. Tuckett (2007: 189) writes, in a review of Derrenbacker’s book:

At times Derrenbacker is evidently considering the practice of scribes, their posture, furniture etc. At other times, he is clearly thinking of authors, composing texts. When it comes to the Gospels, a key question might be whether the evangelists themselves (the ‘authors’) were also the (physical) scribes of their own work. (I am grateful to my colleague Dr Andrew Gregory for highlighting to me the importance of this question.) Did they write their own texts themselves? Or did they dictate to others? If the latter were the case, the physical difficulties of scribes accessing more than one text simultaneously as they wrote might be less relevant.

Tuckett’s ‘key question’ is indeed key, especially given that that is how much (if not most) writing was done in antiquity. ³¹ It is thus possible that Luke dictated his Gospel. On the other hand, the improvements that he makes are often sophisticated, and may have involved a lot of erasing and rewriting. After so much rewriting, it would have been easier for Luke to handle the stylus for himself (see the advice of Quintilian, cited by Winsbury [below]). It also would have made it easier for Luke to remain mindful of how much space was left on his tablets if he did his own writing.

The Medium of Composition

In connection with the question of writing posture, it makes a real difference whether one imagines an evangelist writing on a roll, a codex, or—to name the scenario that almost all previous discussions have excluded—a wax tablet. There is no question that the technology of book production represents an important context for understanding the creation of the Gospels, but the attempts to apply this area of study so far have turned too often on a set of unrealistic and factually groundless claims.

The increased awareness of how the technological aspects of book production impact our understanding of the Gospels is owed mostly to an increased interest in literacy in the ancient world. In its latest phase, however, it also has something to do with an appreciation for the possibility that some of the evangelists’ written sources were in codex rather than roll form. ³² For example, Derrenbacker (2005: 253) finds it necessary, given all Matthew’s to-ing and fro-ing through Q, to posit that Matthew knew Q in codex form. ³³ Scholars are so used to thinking of first-century writings in the form of rolls rather than codices that they might greet Derrenbacker’s suggestion as a besetting qualification, but there is no a priori reason to reject the possibility of a codex in a first-century context. ³⁴

There is, however, one striking piece of evidence that Luke was almost certainly ‘published’ in roll form: the NA ²⁷ text of Luke’s Gospel contains 95,972 letters, while the text of Acts contains 95,838. The difference in length between the two is less than two-tenths of a percent of the length of either book. This strongly suggests that Luke worked with identical space constraints for both works. This could point to a predetermined number of quires in an identical pair of codices, but it makes it more likely that Luke wrote on rolls. ³⁵ This implies nothing, of course, about the form of Luke’s sources.

Whether the earliest Christian writings circulated in roll or codex form, however, is mostly immaterial for the question of how those writings were composed. One major flaw in Downing’s and Derrenbacker’s discussions of the evangelists’ compositional practices is that they omit an important step found in most literary writing of the period, in that they fail to consider the use of the wax tablet as a rough-draft medium for the writing of a Gospel. ³⁶ Although this omission has been typical for NT scholarship, its significance looms for the matters at hand, and there can be little question that Luke, given the opportunity, would have used wax tablets to compose his work. ³⁷ Rex Winsbury (2009: 103) writes,

[O]ne must not ignore the use of practical objects like wax tablets in the process of literary composition, at least in the early gestation stage of a new work. The poet Catullus, admittedly writing before the recitatio came into fashion, refers to his pugillaria and codicilli (i.e. notebooks) that a lady-friend will not return to him, and to the verses written the day before with Licinius in meis tabellis—‘on my (wax-covered) tablets’. But it is Quintilian, writing well into the age of the recitatio, who provides an interesting discussion of the relative merits of writing privately on wax tablets, and of dictating to an amanuensis secretary. Acknowledging the ‘delights’ (deliciae) of dictating to a secretary, Quintilian nevertheless insists on the virtues of writing with a stilus pen on wax, in that it can be rubbed out easily with the pen (or rather, the flat end of it—Horace refers to creating poetry by lots of rubbing out, multa litura) and you are not up against the faults of the amanuensis (secretary), who may be slow, both in writing and in understanding, or to the contrary, get impatient. Moreover, he says, writing on a wax tablet is a private affair, and you can make gestures or pull faces or strike yourself, all of which others watching might find ridiculous (slaves presumably did not count). ³⁸

So we may suppose that some writers did use wax tablets and did start the process of composition in private, perhaps depending on what sort of work it was, and on the quality of their eyesight—Quintilian admits that wax tablets are not always easy to read.

In introducing the different writing media, Derrenbacker (2005: 33) mentions the wax tablet, and he admits that it was sometimes ‘intended for initial drafts of various publications’, ³⁹ but he unfortunately allows this admission to drop from view when actually considering how Luke wrote his Gospel. The evangelists almost certainly would have composed their Gospels with the aid of these tablets. ⁴⁰ This would have allowed them to refine their structure, phrasing and word choice with nearly as much ease as writers in the twenty-first century enjoy. Only after a complete set of tablets had been filled up would the new Gospel be ‘published’ by being transferred onto a roll or codex. ⁴¹

Winsbury’s words (quoted above) represent a break in how scholars usually fit the wax tablet into the world of writing. The main discussion of wax tablets has centered on the uses impressed upon us by archaeology—viz. receipts and other permanent records. The chief purpose of the wax tablet, however, is not revealed by the archaeological record. There is little reason, after all, for permanent records of the sort found at Pompeii, Herculaneum, Vindonissa, etc. ⁴² to be written in wax rather than directly on wood. In fact, tablets have been found with recesses for wax, but with the wax missing and the writing made directly on the wood (see Marichal 1992: 171). The wax must have served mainly for an altogether different type of writing. The school exercises so widely cited cannot possibly explain the frequency with which wooden tablets featured recesses for wax, and school use cannot explain why elaborate polyptychs (some with as many as ten leaves) were created with recesses for wax. Everything points to the wax tablet’s leading purpose being its use as an aid to literary composition. ⁴³ This escapes the archaeological record, of course, because these tablets would have been erased after the ‘publication’ of the work in question.

Pliny the Younger tells us about his uncle’s practice of composing with the assistance of a stenographer recording his dictation on pugillares (Epistles 3.5.14-15). Horace wrote that if one frequently inverted the stylus (viz. used it as an eraser), then one’s writing would be more worthwhile to read (Sat. 1.10.72-73). As Gamble (1995: 268-69 n. 36) observes, such a statement can really only refer to the writing of literary drafts. Jocelyn Penny Small (1997: 153-54) shows the extent of the evidence for writing on wax tablets in the classical Greek period, citing vase paintings that depict the practice, statues of ‘scribes’ with diptychs, references to Plato’s practice (cf. Diogenes Laertius 3.37) and to that of Plato’s contemporary, Dionysius the Elder (cf. Lucian, Adversus indoctum et libros multos ementem, 4). In Lucian’s account, the tablets belonging to the scribe statues might represent record-keeping rather than literary writing, but Small’s other references clearly have to do with writing in a more literary way. She also traces this practice into the succeeding centuries, citing Callimachus’s self-described practice of placing a writing-tablet on his knees at the very outset of writing (Small 1997: 155 [cf. Callimachus, Aetia 21-22]). ⁴⁴ Tiziano Dorandi examines numerous accounts of how works were written and reduces their commonality to two steps: the gathering and reworking of material on pugillares, and the ἔκδοσις of the final form. Another step occasionally intervened, in which one improved the writing with the aid of ὑπομνηματικά (Dorandi 1991: 32-33). (A stenographer was often used at either the first step or the intervening step.) The wax tablet, in fact, represented the basic outfitting of anyone with literary pretensions. P. Kellis 63.17-21, for example, counts the acquisition of a tablet as a logical gift for the newly grapho-literate: ‘Send a well-proportioned 10-page notebook of city quality for your brother Ision. For he has become a writer of Greek and thoroughly Attic reader…’ (translation from Whitehorne 1994: 278). ⁴⁵ By all accounts, the wax tablet was an important part of just about all literary writing. ⁴⁶

Once we admit the likelihood that Luke used wax tablets in composing his Gospel, additional dimensions to the task of composing a Gospel immediately appear, some of which might serve as confirming proofs. To compose a text the length of Luke (or of Matthew, Mark or John) with a single complement of wax tablets would take a rather large stack of tablets. If Luke wrote his Gospel from a fairly cosmopolitan location, it is possible that he could have bought or borrowed enough tablets to do the job in one pass, but it is also possible that he simply wrote his Gospel in multiple segments—viz. that he wrote until he ran out of room on his tablets, stopped and refined his wax writing, then had it transcribed to a more permanent medium—after which he erased his tablets and wrote the next portion of text. When one looks at Luke’s Gospel through the lens of this scenario, the usefulness of such an explanation becomes immediately apparent, as it is widely recognized that (discounting the infancy narrative) Luke consists of three fairly equal sections: 3.1–9.50; 9.51–18.14; 18.15–24.53. These three sections might very well correspond to three successive uses of the wax tablets. Luke’s ‘great omission’ of Mk 6.45–8.26 lends support to this scenario: these Markan pericopes would have appeared near the end of Luke’s first section. It has long been suggested that Luke skipped over this part of Mark’s text in order to allow his completed Gospel to fit onto a standard-size roll. ⁴⁷ But if he needed to make room at the level of the book-roll, we must marvel at his foresight, as presumably not many writers could have calculated how much room they needed to make at such an early point in the narrative. It is more likely, in my opinion, that Luke was nearing the end of his supply of wax tablets, and that he recognized that he had to skip over a portion of Mark in order to fit all of the pre-Passion Week Markan material into a complete section. That would explain why the ‘great omission’ appears near the end of Luke’s first section. ⁴⁸

The wax-tablet model, in fact, goes a long way toward explaining how a Farrerian Luke was able to rearrange the so-called travel-narrative material so freely. Once Luke had recorded this material on a set of tablets, he easily could have rearranged these tablets in any order desired. It would have been easy to rearrange the material marked out for the travel narrative—as easy, in fact, as virtually any other aspect of writing a Gospel. Viewed in this light, it appears that far too much has been made of the supposed difficulty of Luke’s rearrangement of Matthew. Accounting for the wax-tablet stage in the writing of the Gospels thus makes a significant difference when it comes to determining what an evangelist could have accomplished in a first-century context. It presents more reasons to dismiss Downing’s caricature of Goulder’s Luke balancing a ‘springy’ roll on his knee. As it turns out, the evangelists did not live in as different a world of ‘word-processing’ conventions as Downing imagined. In fact, when considering the compositional conveniences of the wax tablet (qua an erasable and rearrangeable medium), the evangelists, in some respects, perhaps had more in common with today’s computer-assisted writers than they did with writers of our grandparents’ generation.

Conclusion

The delving of synoptic source critics into the question of ancient compositional practices is unquestionably a step in the right direction. No theory of Gospel origins can stand if it is out of touch with how first- and second-century writings were composed, and source criticism cannot move forward without due attention to such matters. Recent attempts to view the synoptic problem through the lens of writing technology, however, have tended to operate with a flawed understanding of some basic aspects of the writing process. Perhaps the most egregious problem is the lack of attention to the wax tablet. Given that a wax-tablet stage helps explain the threefold structure of Luke’s Gospel, and shows how easily Luke could have rearranged material in the travel narrative, any failure to include it can only lead to a seriously attenuated view of Luke’s working options.

The failure to include the wax tablet among the basic stages of writing is not the only respect in which scholarship has presented a problematic account. Downing, in particular, has misrepresented numerous aspects of how the ancients wrote, from how one worked with multiple sources, to the possibility of moving both forward and backward in a source. (Derrenbacker’s recent study represents a distinct advance over Downing, but it still leaves a number of problems unresolved, and often misstated.) All of these problems affect his analysis of the synoptic problem. When Downing (2000: 14) refers to Goulder’s understanding of the Gospels’ literary interrelationship as having no ‘basis in the known or discoverable conventions and practicalities of literary composition at that time’, he goes much further than our knowledge of ancient compositional practices will allow, and sometimes in a totally wrong direction. ⁴⁹ Farrer’s and Goulder’s shared vision of Luke is completely at home in the first-century world, as are those aspects distinct to Goulder. ⁵⁰