Dating Early Christian Papyri: Old and New Methods – Introduction

In 2016, as a member of the steering committee of the Society of Biblical Literature group ‘Archaeology of Religion in the Roman World’, I proposed to put together a panel on early Christian papyri and the problem of dating them. My aim was to discuss the state of the art of palaeographical and other methodologies used for assigning a date to papyrus manuscripts that do not contain any indication of their time of composition other than their material aspect.

My personal interest in the topic came from two direct experiences. First, since 2009 I have contributed to the curatorship and study of the Rylands papyrus collection at the University of Manchester, which allegedly holds the most ancient fragment of the New Testament ever found (P.Ryl. III 457 = P 52), besides a surprising number of other early Christian writings, such as the first copy of the Sub tuum praesidium (P.Ryl. III 470), which I will return to later, one of the earliest copies of the St Mark Liturgy (P.Ryl. III 465), and the earliest evidence of anti-Manichaean polemics in Egypt (P.Ryl. III 469). I found myself questioning the dates of some of these papyri, or at least the methodology used and reasons given for such date attributions, following a new wave of studies in papyrology that have rightly criticized a certain way of practising palaeography by past generations. Brent Nongbri, one of our contributors, has led the way in reconsidering the methodology followed by Roberts in dating our P 52, showing how his opinion has often been misused for various reasons in the years following publication. ¹ It took me many years to convince the Rylands Library to change the labels of the papyrus and provide a more critical approach to its dating. It is worth mentioning that the fragment in question is the only item in the Rylands extensive collection to be on permanent display due to the high cultural and religious meanings attached to it. This shows the complex and problematic relationship Europeans have with ancient manuscripts, Christian ones in particular.

The second reason which pushed me to organize the session was that I have been directly involved in recent attempts to use technology and scientific methods when deciphering and studying papyrus manuscripts; this led me to question the blind faith which academics and the wider public seem to have in research outcomes that result from scientific investigations. In my experience, these are useful but far from conclusive. I had the opportunity to employ radiocarbon analysis and multispectral imaging while editing P.Ryl.Greek Add. 1166, a Christian amulet with passages from the Bible and the liturgy, later re-used for a tax receipt (Mazza 2016). I realized on that occasion how misleading the way in which analysis results are communicated in publications can be, and how troublesome the cultural environment is in which science and technology seem to hold an unquestionable status in some fields and in the mind of the general public. The example of carbon dating is particularly interesting: too few people, including academics, are aware that this technology provides dates only for the papyrus and not for the ink used on it, and that the results are always given with an approximation of around a century. To give an example, the analysis performed on P.Ryl.Greek Add. 1166 showed that the piece of papyrus on which the text was written was produced between 574 and 660 with a margin of certainty of 95.4%, and between 608 and 650 with a margin of 68.2%. This has implications on the interpretation of the object and should be clearly explained in the edition of the manuscript; palaeographical analysis confirmed a date in that 95.4% range of time.

Besides using carbon dating, I also took part in a pilot project for developing new non-invasive technologies for reading papyri encapsulated in mummy masks and other cartonnage objects. ² The pilot project came as an answer to a rather serious practical preoccupation: in the last decade, especially in some American evangelical Christian quarters, the search for manuscripts bearing the first traces of the scriptures led to the destruction of ancient mummy masks, book bindings and other ancient objects without any concern for the many issues involved with such practices, let alone the serious unanswered questions about the provenance of the artefacts (Mazza 2014; Gibson et al. 2018: 1-2). To some academics and their audience the destruction of ancient artefacts seems to be a legitimate and unproblematic process, motivated by a cultural value system in which the search for the most ancient testimonies of Christian writings justifies even unorthodox and unscientific behaviours. In this case too, although I found the collaboration with scientists and imaging experts enriching and fruitful, I had to face problems of communication with specialists in other fields and experienced the limits of scientific disciplines, which tend to be wrongly thought of as more efficient than the humanities in discovering dates and other features of archaeological remains.

These two professional circumstances, working on the Rylands collection and multidisciplinary research, together with the direct experience of the power early Christian manuscripts still hold over their audience, convinced me that it was time to address the problem of dating early Christian manuscripts, from Egypt and potentially elsewhere, in a more sophisticated way than has been practised so far. The panel and the articles which resulted from it and are published in the following pages constitute a first attempt at solving a number of methodological contradictions and obstacles and answering some crucial questions: What are the limits of palaeography and how can we overcome them? Similarly, what are the limits of scientific methods such as radiocarbon analysis and Raman spectroscopy when applied not only to early Christian, but also to any other type of ancient manuscript? Finally, how can we communicate the complex and always partial results of multidisciplinary research on such manuscripts not only inside academia, but also, and perhaps more importantly, to the wider public?

In his contribution, Malcolm Choat traces the roots of papyrology against those of palaeography and shows how different preoccupations and research questions led to different approaches to the study of Greek handwriting within papyrology. He rightly recalls that ‘the subjectivity that lies at the heart of the system’ is often underplayed and students tend to have great confidence in the ability of papyrologists to attribute dates. This brings me back to one of the above-mentioned Christian Rylands papyri, P.Ryl. III 470, which is a good example of how uncritical reliance on past attempts at dating a papyrus can cause imprecision and misconceptions that still remain in current research.

The manuscript in question was purchased in Egypt by J. Rendell Harris in 1917 and published by C. Roberts in 1938. ³ It is a small fragment that presents many aspects of interest for its material shape and the text it carries. In his edition, Roberts proposed a date in the fourth century with a question mark due to the many doubts the papyrus raised. Roberts’s discussion in the introduction addressed both the handwriting and the material features of the object, as well as the textual content. Written along the fibre of a leaflet, the back of which is blank, was clearly a hymn or prayer to the Mother of God, as indicated by the vocative θεοτόκε at line 4. The shape of the capital letters – tall and compressed, adorned by decorative serifs – was briefly discussed and compared to the chancellery style of that used in a letter of the prefect Subatianus Aquilas, securely dated to 209 ce (SB I 4639). On the basis of this means of comparison, E. Lobel, with whom Roberts consulted, deemed that the papyrus could not be later than the third century. However, Roberts rightly warned the readers not to jump to conclusions: the papyrus mentioned Mary as mother of God, a title that appeared only later on, and there were many difficulties in dating this type of highly stylized handwriting. The first edition of this manuscript highlights some of the themes addressed by Choat and Nongbri in the following pages: palaeography developed hand in hand with the publication of papyri, and the use of palaeography as the only means for dating texts has always presented a challenge. Papyrologists answered as best they could by providing some analysis and discussion against securely dated items and consulting their peers as part of the amicitia papyrologorum, a phrase that started to be used in the 1930s and has since become the motto of the International Association of Papyrologists. Papyrologists rely on each other’s opinions and direct experience with large amounts of texts in order to give the best possible, but nonetheless partial and subject to change, answer to the question of attributing a date to a papyrus that does not bear one.

Right after its publication, the hymn was recognised as an early version of the first prayer to Mary as the Mother of God, the Sub tuum praesidium (Mercenier 1939). Debates on the date of the papyrus have not stopped since its publication and recognition of the Greek version of that famous prayer. P.Ryl. III 470 has been assigned to different centuries spanning from the third to the ninth. ⁴ Nonetheless, exactly as for P 52, early Christianity students have tended to choose information from editions in a rather selective and unsystematic way, and usually end up preferring the earliest date. For instance, in a recent monograph on the development of Marian devotion, S.J. Shoemaker dedicates a section to our P.Ryl. III 470, which opens with the following sentence:

In 1917 the John Rylands Library in Manchester, which houses one of the world’s most important collections of ancient papyri, added to its inventory a small fragment containing a prayer in Greek to the Virgin Mary that seems to date from the end of the third century. (Shoemaker 2016: 69)

As I have reported above and is clearly explained in the introduction to the edition, Roberts has never attributed that date to the papyrus. He gave an account of Lobel’s opinion and provided a very cautious estimate to the fourth century, stressing his many doubts including the comparative handwriting of the edict suggested by his colleague. Shoemaker, however, interprets our Rylands papyrus as proof that ‘at least some Christians in Egypt had begun to pray to the Virgin Mary and ask for her intercessions already by the end of the third century’ (Shoemaker 2016: 69). The author admits that there are debates surrounding the date of the fragment, but in light of considerations relating to the cult of Mary as attested in some passages of Origen, he concludes the following:

In light of this evidence, and the initial suggestion that the palaeography reflects handwriting from the third century, the end of the third century seems to be a likely date for the production of this document, although one certainly should not rule out the possibility that it might be even slightly earlier. (Shoemaker 2016: 72)

The point in question here is not the correctness of the papyrus’s date but rather the methodology employed in fixing one. While there is no real attempt to provide a comparison with dated manuscripts, there is a plain acceptance of a shaky provisional date attributed to P.Ryl. III 470 almost a century ago. It is also worth noticing that the debate surrounding the date is very briefly mentioned in the section and considered only in a footnote in some depth (Shoemaker 2016: 71 with n. 18). Students of early Christianity seem to have too much confidence in the ability of renowned papyrologists to assign dates to undated manuscripts, including when the papyrologists themselves explain that their guess is far from firm, as in the case of Roberts’s first edition of the Rylands papyrus.

As Nongbri explains, the use of palaeography for dating Christian manuscripts is even more complex and problematic than this example shows. In order to fix a date for undated papyri, papyrologists try to find a securely dated document that presents a similar handwriting; they are largely working on the assumption that graphic similarity entails that the two documents are contemporary, as we have just seen in the comparison between P.Ryl. III 470 and the letter of Subatianus Aquilas. But this equation is far from certain: the activity of a scribe can span many decades; scribes could write in different styles on different occasions and even use, for instance, archaizing letters on purpose. Moreover, what does graphic similarity precisely mean, and how is the comparison carried out? The questions raised by Nongbri problematize the scientific methodology at the core of our practice; there are still many editions which provide too few explanations regarding an assigned date. Expertise based on academic prestige and competence – recalled by Choat with reference also to recent cases, namely the so-called first-century Mark fragment, which now has become P.Oxy. LXXXIII 5345 dated to the second or third century – is only partially justified; there must always be a place where the reasons for a date are provided: that is peer-reviewed publications.

The development of palaeography applied to papyri went hand in hand with the development of imaging technologies and the accumulation of published texts, which constitute the basis for the study of handwriting through comparisons. Yet, despite the establishment of palaeography as a field, the study of Greek hands remained firmly anchored to a humanistic way of considering the history of writing as a process with a beginning, a development and a slow decay. The use of the word ‘canon’ and the categorization of styles of writing are rightly recalled by Nongbri as part of a problematic approach to palaeography, which needs to be better clarified for the audience and practised with a greater awareness of the mechanics of scribal education and habits. ⁵

A question that was addressed in the course of the SBL panel concerns the classification and treatment of Christian papyri as a corpus separate from the rest of the papyrus evidence. Choat rightly criticized the approach taken by the title of the panel, pointing out that the separation of this specific material from the rest of the papyri produced in the same period and by the same scribes (or at least scribes trained in a similar environment) is in fact the main cause of distortions in early Christian scholarship on early Christian manuscripts. Nongbri’s article, however, shows that there is indeed something peculiar about Christian manuscript production, which is the extensive, almost exclusive, use of the codex format for the scriptures. As a consequence, none of the samples in question present the phenomenon of the re-use of one side of the roll, which instead occurs for non-Christian literature and provides some anchors for dating literary manuscripts. As Nongbri explains, the general shift from roll to codex in the third and fourth centuries complicates the task of dating literary handwriting further, as the pool of comparanda from reused scrolls clearly decreases and is problematic in other respects. To Choat’s warning not to isolate Christian material from the rest should be added the caution not to separate literary from documentary papyri in palaeographical analysis. As shown by recent and less recent scholarship, it is not infrequent to find papyrus documents written in literary hands and literature copied in documentary hands.

So far, I have discussed the pitfalls and perils of a methodology – palaeography within papyrology – that belongs to the humanities. Are science and technology any better? Can they help us overcome the problems outlined so far? In his article, Nongbri shows some confidence in, but also preoccupation with, carbon dating; I have discussed in the opening the limits of this procedure and the need to improve the way results are communicated to the audience. Carbon dating is also a destructive analysis, since a fragment from the manuscript under study needs to be sampled and processed; in Manchester, for instance, I suspect the technique will be never allowed on P 52 and other important papyri, unless the size of the samples needed will significantly decrease. The following contribution of Goler et al. deals with a potentially revolutionary and non-destructive technology that has started to be tested on papyri and other manuscripts only recently: Raman spectroscopy. In simple terms, this is a non-invasive imaging technology which focuses on the ink rather than the ‘paper’. This technique is based on the assumption that ancient inks are made from different elements, the molecules of which vibrate in various ways; through advanced imaging techniques, scientists are able to measure those vibrations and decompose the ink. Technically speaking, this procedure enables the analysis of the components of inks rather than their dating. However, the assumption of the article and similar experiments in Raman spectroscopy is that components and their mixing must have changed over time, and by sampling a consistent number of them, it will be possible in the future to attribute a date to manuscript inks on the basis of pattern recognition.

To some extent, Raman spectroscopy is what palaeography was for Mabillon in the early modern period: a tool for detecting forgeries, rather than assigning dates. ⁶ In fact the Columbia–NYU–ISAW project team that present some of their results here was brought together for analyses carried out while researching a papyrus that we now know is a modern forgery, the so-called Jesus’s Wife fragment. This Coptic papyrus has been under the spotlight since September 2012, when Karen King presented it at the International Congress of Coptic Studies in Rome. A scholarly debate about whether the fragment was genuine or not has kept papyrologists, coptologists, historians of early Christianity, as well as scientists and imaging technicians busy for almost four years. The nature of the text transmitted also attracted wide and sustained media attention, as has always happened with Christian texts.

The Jesus’s Wife fragment went through carbon dating analysis, and the result was a calibrated age between 648 and 800 ce (95.4 %) (Hodgins 2014; Tuross 2014); ink analysis provided a result in line with carbon black pigments (lamp black) found in a ‘wide variety of manuscripts including many dating from the early centuries of the Christian era’ (Yardley and Hagadorn 2014: 164). These results might seem puzzling, since if taken without caution they might prompt the false idea that the manuscript was indeed genuine. However, upon closer inspection, both analyses were, and still are, completely inconclusive if isolated from other considerations pertaining to the humanities: palaeography, textual philology and collection history, the last of which was indeed the conclusive piece of evidence for establishing without any doubt that the papyrus was in fact a forgery. ⁷

Carbon dating analysis proved that the forger of the Jesus’s Wife fragment used a genuine piece of papyrus to write the text; anyone working in papyrus collections is aware that blank papyri and papyri with wide sections of unwritten texts have been found (and might still be found) in Egypt. The use of genuine ancient papyrus to fabricate forgeries is a well-known phenomenon. ⁸ As for the ink, as mentioned above, the brief report appended to the publication of the Jesus’s Wife papyrus confirmed that it was mainly based on carbon black pigments, and the spectra obtained did not present anomalies compared to the compositions of similar samples from other manuscripts at the centre of previous investigations. Nonetheless, it is widely known that forgers are aware of ancient ink composition and are able to produce it; the problem with Raman spectroscopy is that too few papyri have been sampled and there is no reliable set of data so far. Although a statistical model can offer some solution to this problematic status of the question, as Goler et al. are proposing here, the way forward is indeed the collection of a consistent pool of data.

The solution of the Jesus’s Wife papyrus’s case is a wonderful and successful example of multidisciplinary research (planned and unplanned), in which the combination of various methods led to a final verdict of forgery. This kind of multidisciplinary approach and collaboration has been rightly deemed as the way forward in a recent study on the authenticity of fragments of Dead Sea Scrolls recently emerged from the antiquities market:

This study seeks to implement a multidisciplinary approach of investigation using empirical methods alongside historical interpretative methodologies, intuitions and aesthetic evaluation, which may allow for credible multi-contextual reasoning. (Davis et al. 2017: 196)

The SBL panel and the articles published here were gathered together with the same awareness that multidisciplinary research is the only possible path toward solving the many and complex questions connected to the study of ancient papyrus manuscripts, never forgetting that our conclusions are always partial and transitory because of the true nature of academic research.