P52 - The Faintest and Brightest Papyrus Star (with guest Dwayne Green)

The tiny fragment Papyrus 52 has gotten an inordinate amount of attention by being perhaps the earliest written copy of any text from a book of the New Testament (from John 18:31-33 and 18:37-38).   Catalogued as Papyrus Rylands Greek 457 and brought to to public attention by Colin H. Roberts in 1935, perhaps more facsimiles have been made of it than any other New Testament papyrus (one is offered at Credo Courses for example).  Roberts' analysis placed P52 in he first half of the 100s.

There is not enough to P52 to confidently proclaim its text Alexandrian but what is extant points in that direction.  Let's take a look at its text  (bold  = extant) 

Recto: from 18:31-33)

ΟΙ ΙΟΥΔΑΙΟΙ ΗΜΕ]ΙΝ ΟΥΚ ΕΞΕΣΤΙΝ ΑΠΟΚΤΕΙΝΑΙ

ΟΥΔΕΝΑ ΙΝΑ Ο Λ]ΟΓΟΣ ΤΟΥ ΙΗΣΟΥ ΠΛΗΡΩΘΗ ΟΝ ΕΙ-
ΠΕΝ ΣΗΜΑΙΝΩ]Ν ΠΟΙΩ ΘΑΝΑΤΩ ΗΜΕΛΛΕΝ ΑΠΟ-
ΘΝΗΣΚΕΙΝ ΙΣ]ΗΛΘΕΝ ΟΥΝ ΠΑΛΙΝ ΕΙΣ ΤΟ ΠΡΑΙΤΩ-
ΡΙΟΝ Ο Π]ΙΛΑΤΟΣ ΚΑΙ ΕΦΩΝΗΣΕΝ ΤΟΝ ΙΗΣΟΥΝ
ΚΑΙ ΕΙΠ]ΕΝ ΑΥΤΩ ΣΥ ΕΙ O ΒΑΣΙΛΕΥΣ ΤΩΝ ΙΟΥ-
ΔΑΙΩ]N

Verso:  (18:37-38)

ΒΑΣΙΛΕΥΣ ΕΙΜΙ ΕΓΩ ΕΙΣ TO[ΥΤΟ ΓΕΓΕΝΝΗΜΑΙ
ΚΑΙ ΕΛΗΛΥΘΑ ΕΙΣ ΤΟΝ ΚΟ[ΣΜΟΝ ΙΝΑ ΜΑΡΤΥ-
ΡΗΣΩ ΤΗ ΑΛΗΘΕΙΑ ΠΑΣ Ο ΩΝ [ΕΚ ΤΗΣ ΑΛΗΘΕI-
ΑΣ ΑΚΟΥΕΙ ΜΟΥ ΤΗΣ ΦΩΝΗΣ [ΛΕΓΕΙ ΑΥΤΩ
Ο ΠΙΛΑΤΟΣ ΤΙ ΕΣΤΙΝ ΑΛΗΘΕΙΑ Κ[ΑΙ ΤΟΥΤΟ
ΕΙΠΩΝ ΠΑΛΙΝ ΕΞΗΛΘΕΝ ΠΡΟΣ [ΤΟΥΣ Ι]ΟΥ-
ΔΑΙΟΥΣ ΚΑΙ ΛΕΓΕΙ ΑΥΤΟΙΣ ΕΓΩ ΟΥΔ[ΕΜ]ΙΑΝ
ΕΥΡΙΣΚΩ ΕΝ ΑΥΤΩ ΑΙΤΙΑΝ

As P52's Wikipedia listing says, "There appears insufficient room for the repeated phrase (ΕΙΣ ΤΟΥΤΟ) in the second line of the verso, and it is suggested that these words were inadvertently dropped through haplography."

Deviations from the Byzantine Text:

v. 31 - ημειν instead of ημιν

v. 32 - ισηλθεν instead of εισηλθεν

v. 33 - the word-order is Alexandria; παλιν preceding εις το πραιτώριον

The abundance of online material about P52 makes P104 seem neglected in comparison.  A few samples:

David Litwa Profiles P52 

Michael Marlowe Presents P52 

BiblicalTours - P52 with Dan Wallace

Recently P52 was mentioned on the Joe Rogan podcast in an interview with Wesley Huff of Apologetics Canada.  and Brent Nongbri has provided some helpful clarifications about some statements Wes made  which need improvement.  Nongbri has continued to profitably research P52.

P52 is notable for  being from a codex which was very probably a codex of only the Gospel of John. Robert's analysis supporting this is accessible online.

P52 has been featured on YouTube in simple profiles and in a 2024 lecture Steven Combs claimed that it supports the Textus Receptus.  Dwayne Green has offered some pushback in a video that can be viewed here.

TTotG:  Welcome, Dwayne Green!   Tell us:  does P52 support the Textus Receptus?

Dwayne Green

Dwayne Green:   No.   P52 does not support the Textus Receptus.  A couple weeks ago, Dr. Steven Combs under the banner of the King James Bible Research Council (KJBRC) had released a short video entitled “Textus Receptus Found in the Papyrus?”. Evidently this was clipped from a larger lecture on the Textus Receptus (TR) that was given at the KJBRC Regional Conference in 2023. Within the first 45 seconds of the short video Dr. Combs made a rather curious claim that P52, our oldest extant manuscript of the New Testament, actually supports the Received Text. In the words of Dr. Combs:

“Papyrus 52 . . . has one spelling difference and one place that I consider to be a scribal error. Because it was an omicron . . . that was misplaced and put in front of the word instead of behind it. Besides that it is a perfect match to the Textus Receptus.”

Curious indeed. It’s worth asking the questions: What spelling error is he talking about and what does he mean by ‘scribal error’ and is this really a simple mishap that would have otherwise left us with a perfect representative of the TR in our earliest papyri?

First, it’s worth noting that the TR in John 18:31-33 shares the same form alongside the Byzantine Textform, HF Majority, Antoniades, and Family 35. Though family 35 contains the textual variant εμελλεν instead of ημελλεν, however P52 is lacunose at this word and so from the perspective of this analysis, they are essentially the same. For the most part, the NA28 matches with the WH text and differs from the former group in one important way. In John 18:33, the critical editions read “Εισηλθεν ουν παλιν εις το πραιτωριον ο Πιλατος” whereas the former editions read “Εισηλθεν ουν εις το πραιτωριον παλιν ο Πιλατος”. Make a mental note of where παλιν and ο Πιλατος are in relation to each other as this will be very important later on in this analysis. Admittedly, from the perspective of our English translations, this is hardly discernible and regardless of where παλιν appears in the phrase, it has little consequence to our English understanding of how this verse is to be understood.

TTotG:  So the difference in meaning is zero - but there is an orthographic difference.  What are your thoughts about that?

Dwayne Green:  Dr. Combs claimed that P52 “has one spelling difference”. If you take a look at the image, I was able to find two spelling variants that are visible on the manuscripts itself. The first line likely reads “ΟΙΙΟΥΔΑΙΟΙΗΜΕ”. The final epsilon is not expected, rather what is expected is the third person pronoun in the dative form ΗΜΙΝ, it should be an iota instead. According to Swanson’s collations, there are no textual variants that match the final epsilon in this passage. So what’s the deal? I offer two possible explanations. 1) This is a simple case of itacism and it’s possible that the scribe had written down ΗΜΕΙΝ. This has occurred in at least one place in P66 (see John 17:21). Whether the scribe had copied what was in front of him or inserted his own unintentional variation would be impossible to determine. 2) This could be a case of a grammatical error if the scribe had written ΗΜΕΙΣ, however there are no variants elsewhere in the manuscript tradition that support this. Given these two options, itacism seems far more plausible than the grammatical error. We shall give Dr. Combs the benefit of the doubt in this case and suggest that itacism is the culprit and this is a simple spelling variation rather than a spelling error.

The second spelling error is found on the fourth line which reads ΘΝΗΣΚΕΙΝΙΣ (image 2). Though the text is faded, you can still make out the iota and the sigma. The expected word here is “Εισηλθεν” and the astute viewer will notice that the epsilon is missing. This is indeed true. Evidently, the dipthong ΕΙ may have sounded the same or similar as a single iota and this may have confused the scribe creating a spelling error. This is of course is assuming that the scribe did not correctly copy the exemplar before him and produced the error rather than copied it. So we’ll give this one to Dr. Combs as it is likely the ‘spelling error’ that he had referred to in the short video.

I therefore take no issues with his claim of a single spelling error.

TTotG:   What about the scribal error?

Dwayne Green:  To understand how Dr. Combs came to the ‘scribal error’ conclusion we must first realize that he is looking at P52 from a TR perspective. If you take a look at the the following image, the line being shown reads ΡΙΟΝΟΠ. Dr. Combs understands the Π at the end of this line to be the first letter of the word ΠΑΛΙΝ. This is the underlying assumption that allows him to conclude that the omicron “was misplaced and put in front of the word instead of behind it”. Remember, the way the TR renders this passage is: “Εισηλθεν ουν εις το πραιτωριον παλιν ο Πιλατος”. So in order for Dr. Combs to justify the TR reading in P52 he must move the article Ο in front of ΠΑΛΙΝ in order to call it a scribal error. For Dr. combs P52 in the lacunose area should read ΡΙΟΝΟΠΑΛΙΝΠΙΛΑΤΟΣ (the red text represents a reconstruction of the lacunose area), but is this reconstruction plausible? Below I have provided three hurdles for Dr. Combs theory.

Hurdle 1: Grammar

Proper nouns such as ΠΙΛΑΤΟΣ are more often then not preceded by the article. However the word ΠΑΛΙΝ being an adverb virtually never has the article before it. A quick search in Logos Bible software results in 142 instance of ΠΑΛΙΝ in the Textus Receptus (presumably Scriveners edition) so I spot checked about 30 random instances and in no case does the article precede ΠΑΛΙΝ. Now Dr. Combs may understand this which is why he called it a ‘scribal error’, but the contention here is that the article Ο followed by Πιλατος would not only be the natural reading, but statistically based on the grammar, would be the more certain reading. Especially when we do not have a conclusion in search of the evidence.

Hurdle 2: Extant Textual Variants

Coming back to Dr. Combs' statement, his explanation is that P52 places the Ο before ΠΑΛΙΝ rather than ΠΙΛΑΤΟΣ resulting in the reading εις το πραιτωριον ο παλιν Πιλατος. This results in a textual variant that does not exist in any of the major collations. There’s plenty of support for the rendering of John 18:33 as found in the TR, such as Aleph, A, C², B, and P60 among others; the complete omission of ΠΑΛΙΝ both before εις το πραιτωριον and after is witnessed by at least two manuscripts according to Swanson: C^c and 33. Codex N and Codex Ψ comes the closest to Dr. Combs' suggestion with the reading ο Πιλατος Παλιν, but even in its closeness, it leaves his ‘scribal error’ theory wanting.

Given the current textual variants among our extant manuscripts it appears that the reading that Dr. Combs wants P52 to be DOES NOT EXIST. Put another way, in order for his theory to fit he must invent a brand new, never before seen textual variant.

Hurdle 3: A Perfect Match to the Critical Editions.

When you compare a reconstruction of the text of P52 with the Critical editions of the Greek New Testament, we find that it is essentially (minus the 2 spelling variants mentioned above) a match with the extant areas of the manuscript. Consider the image below, the highlighted characters represent where the critical editions match with P52. It is essentially Identical. We do not need to come up with convoluted reasons to make something else work. It just fits.

TTotG:  Any other observations about P52?

Dwayne Green:  There is one more matter that is worthy of our attention, although we do not consider it  a ‘hurdle’. That is the question of weather a reconstruction based on character counts per line can give us any helpful details to Dr. Combs claim. Can it make room for ΠΑΛΙΝ on the same line as ΠΙΛΑΤΟΣ or does it favor ΠΑΛΙΝ on the previous line. The short answer is that this methodology is inconclusive. There is tremendous debate surrounding the question of nomina sacra. Does the reconstruction of the name Ιησου in the second line and Ιησουν at the end of the fifth line appear in full form, or in nomina sacra? If we assume for the sake of Dr. Combs argument that the nomina sacra is employed here, it merely allows for the possibility that ΠΑΛΙΝ could fit, but its not decisive as it could still also fit in the line above it where the Critical texts presumes it would fit.

If on the other hand, it turns out that “Jesus” is written out in full form this would further demonstrate that ΠΑΛΙΝ could not be moved and share the same line as ΠΙΛΑΤΟΣ because removing it from its current place would drop the character count for that line far below the other lines represented by the manuscript. Of course, this makes a number of assumptions that must be made in the reconstruction to make sense of it and not everyone is convinced that this is a worthwhile endeavor.

So essentially, as far as Dr. Combs theory is concerned and the application of reconstructing the lacunose area based on character counts, is dead in the water. It does not help us either way.

Conclusion

It is indeed a weird flex to invent a never before seen textual variant in order to substantiate the claim that P52 supports the TR; especially when all signs point to an exact match of the WH and the NA28 text. Add in the fact that grammatically speaking, the article often precedes proper nouns and not usually adverbs which only piles on to the arguments against this wishful thinking.

At the end of Dr. Combs brief analysis of P52 he states rather emphatically “Besides that, it is a perfect match to the Textus Receptus”. I can’t help but hear “Besides the differences from the TR, its a perfect match to the TR”. As a Byzantine Prioritist myself, it would be exceptionally welcomed to see a byzantine reading among the earliest manuscript, but we should also be careful not to make manuscripts say more than they do in an effort to impose our own conclusions on the evidence. We are not part of the “Oldest is best” club, and fortunately our New Testament text rests on much more than a small scrap of papyrus found in the Egyptian desert.

TTotG:  Thanks for sharing your thoughts, Dwayne.

 

Minuscule 750: The Byzantine Text with Pictures

          Minuscule 750 is a Greek copy of the four Gospels, written in a neat minuscule script, in dark ink, assigned to the 1100s.  The text is on 319 pages, written in one column per page, with 20 lines per page.  The text of 750 has been categorized as M27, an old form of the Byzantine Text.

          Matthew has a rectangular headpiece, with red, blue, and green pigment.  Later cursive notes are written in the outer margin in black and red.  The title appears to be written in gold.

        Small illustrations (sometimes slightly covering the text), appear in the outer margins of Matthew and Mark.  Matthew has about 35 margin-illustrations; Mark has about 10.  Very often the illustrations appear to have been scraped off the page.  Sometimes the shapes of scenes in damaged or non-extant pictures appear on the opposite page, allowing a view of the contours of the depicted scene.  Some of the illustrations feature writing describing the depicted scene.

          Only one illustration is extant in Luke:  on p. 322, opposite Luke 1:13.  Other pictures once occupied some pages after this; traces of pigment remain on p. 323.  On several other pages, rectangles are hued more lightly than the rest of the page; these are apparently the remains of places where a picture was once intended but never executed.  Occasionally a thinly drawn rectangle appears to signify where an illustration was once intended; such a frame appears on p. 518 (next to John 1:43).

        Eusebian Section-numbers begin to appear in the outer margin at Mt. 8:28, on p. 48, with #70 (although usually one would expect #70 to appear at Mt. 9:1).  Perhaps the section-numbers were not included before this point because the margin-illustrations made their inclusion difficult.

          Obeli in red often appear in the text at the beginning of a section.  It is not unusual to see a red cross (+) and a gap, blank, before the obelus, as if the addition of liturgical notices (αρχη and τελος) had been intended but never carried out except in an incomplete way (such as on p. 326).  A series of three crosses (+++) usually indicates the end of a chapter. Chapter-titles, written in gold, usually appear at the top of the page where the chapter begins, but titloi occasional appear in the lower margin.

          A later hand has occasionally added some liturgical annotations, featuring circles, which seem to be connected with an Eastertime rite (?):  Stasis #2 at Matthew 9:9, #3 at Matthew 14:1, #5 at Matthew 25:1, #2 at Mark 6:30, #2 at Luke 5:1, #3 at Luke 9:7, #4 at Luke 13:31, #5 at Luke 20:27, #2 at John 5:24, #3 at John 9:1, τελος at John 13:31.

 Here is a selective index to the manuscript:

 Mt. 1:1 – p. 7

Mt. 4:14-15a is written in the lower margin of p. 21, which was omitted in the main text due to parablepsis (Νεφθαλείμ, Νεφθαλείμ).

Mt. 5:1 – p. 24 (ε in margin)

Mt. 7:1 – p. 38

Mt. 8:2 – p. 43

Mt. 8:16 – p. 46

Mt. 9:2 – p. 49 (ιγ in margin)

Mt. 10:1 – p. 56

Mt. 11:1 – p. 63 (κ in the margin)

Mt. 12:22 – p. 71 (κβ in the margin)  

Mt. 14:1 – p. 87 (κε in the margin)

Mt. 17:1 – p. 104 (λδ in the margin)

Mt. 20:1 – p. 122 (μβ in the margin)

Mt. 22:2 – p. 137 (να in the margin)

Mt. 25:1 – p. 157 (νθ in the margin)

Mt. 27:1 – p. 177 (τιζ in the margin)

Mt. 27:58 – p. 186 (ξη in the margin)

 

Chapter-list for Mark begins on p. 192

Full-page picture of Mark, with red footstool – p. 196

Mk. 1:1 – p. 197, with blue and gold headpiece, red and blue initial

Mk. 2:1 – p. 204 (κ in the margin)

Mk. 4:3 – p. 213 (θ in the margin)

Mk. 5:1 – p. 221 (ια in the margin)

Mk. 7:1 – p. 238 (ιη in the margin)

Mk. 9:2 – p. 251 (κε in the margin)

Jesus greeting Mary Magdalene in Mark 16:9,
with imprint on the opposite page.

Mk. 10:2 – p. 260 (κη in the margin)

Mk. 12:1 – p. 274 (λϛ in the margin)

Mk. 14:3 – p. 289 (μδ in the margin)

Mk. 15:48 – p. 306 (μη in the margin)

Mk 16:9 – p. 309

 

Chapter-list for Luke begins on p. 312

Full-page picture of Luke, with red footstool – p. 319.  Luke has written the first word of his Gospel.

Lk. 1:1 – p. 320, with red, blue and green rectngular headpiece.  Title in gold.

Lk. 2:1 – p. 331 (α in the margin)

Lk. 4:1 – p. 345 (ζ in the margin)

Lk. 6:6 – p. 361/362  (ιε in the margin)

Lk. 8:16 – p. 382 (In Luke 8:16, the words αλλ’ επι λυχνίας επιτιθησιν are not in the text.  In the outer margin, the words αλ επι λοιχνηας επι τηθησει are provided – perhaps recollected from memory – and there is a mark in the text showing where they belong.)

Full-page picture of Luke.

Lk. 9:1 – p. 390 (κζ in the margin)

Lk. 11:1 – p. 409 (λη in the margin)

Lk. 12:1 – p. 419 (μδ in the margin)

Lk. 15:3 – p. 441 (νϛ in the margin)

Lk. 18:1 – p. 456 (ξα in the margin)

Lk 20:1 – p. 470 (ξθ in the margin)

Lk. 22:24 – p. 485 (οζ in the margin, +++ in the text)

Lk. 23:27 – p. 496 (π in the margin)

Lk. 24:18 – p. 503 (πγ in the margin)

 

Chapter-list for John begins on p. 510

(No portrait of John)

Jn. 1:1 – p. 512.  Rectangular blue headpiece is topped with two birds.  Red and blue initial with hand-stem.

John 2:1 – p. 519 (α in the margin)

5:1 – p. 536

Jn. 6:5 – p. 544 (μη in the margin and +++ in the text)

Jn. 7:37 – p. 559

Jn. 7:53 – p. 561 (The pericope adulterae is included in the text.)

Jn. 11:1 – p. 582 (ια in the margin)

Jn. 12:3 – p. 591

Jn. 13:5 – p. 598 (+++ in the text)

Jn. 13:31 – p. 602 (+++ in the text)

Jn. 15:26 – p. 611 (last line of text)

Jn. 18:1 – p. 621

(beginning on p. 624, the parchment has been affected by mildew, which obscures some of the text on p. 629)

Jn. 19:38 – p. 633 (faint ιη in the margin and +++ in the text)

Jn. 20:11 – p. 636 (αρχ. on first line in the text)

Jn. 21:1 – p. 639 (αρχ.)

The text of John 21 ends on p. 644.  After the subscription and a squiggly horizontal line, there is the word “Ερμενια” followed by a few lines written in faint red ink; some of this text is obscured by mildew.

         A cross-like symbol appears occasionally, apparently written in pencil, in the margin of MS 750.

          Some idea of the quality of the text of 750 may be gained by a comparison of its text of John 18:1-11 and the same passage in Codex Sinaiticus.  In this comparison, I will compare the text of the main scribe of א (pre-correction), and the standard of comparison will be the text of the 27^th edition of the Nestle-Aland Novum Testamentum Graece.  Abbreviations such as kai-compendia and sacred name contractions will not be counted as variants.  Transpositions will be noted but not counted if no change is made to the amount of letters.

Sinaiticus:  John 18:1-11

1 – א has αυτοις instead of τοις (+2)

1 – א has του instead of των (+2, -2)

1 – א has Κεδρου instead of Κεδρων (+2, -2)

2 – א has χιμαρρου instead of χειμαρρου (-1)

3 – א does not have εκει (εκι added in margin) (-4)

4 – א has δε instead of ουν (+2, -3)

4 – א has εξελθων instead of εξηλθεν (+1, -1)

4 – א has ειπεν instead of λεγει (+4, -4)

5 – א has ΙΣ after αυτοις (+2)

5 – א has ιστηκει instead of ειστηκει (-1)

6 – א does not have αυτοις after ειπεν (added above the line) (-6)

6 – א has επεσαν instead of επεσον (+1, -1)

7 – א transposes to αυτους επηρωτησεν

8 – no variations

9 – no variations

10 – א has επεσεν instead of επαισεν (+1, -2)

10 – א transposes to δουλον του αρχιερεως

11 – no variations

Taking the text of א “as is,” pre-correction, it has 17 non-original letters, and is missing 27 original letters, for a total of 44 letters’ worth of corruption. 

Removing orthographic readings and transpositions:

1 – א has αυτοις instead of τοις (+2)

1 – א has του instead of των (+2, -2)

1 – א has Κεδρου instead of Κεδρων (+2, -2)

3 – א does not have εκει (εκι added in margin) (-4)

4 – א has δε instead of ουν (+2, -3)

4 – א has εξελθων instead of εξηλθεν (+1, -1)

4 – א has ειπεν instead of λεγει (+4, -4)

5 – א has ΙΣ after αυτοις (+2)

6 – א does not have αυτοις after ειπεν (added above the line) (-6)

6 – א has επεσαν instead of επεσον (+1, -1)

11 – no variations

Removing orthographic variants from the picture yields a total of 16 non-original letters present, and 23 original letters absent, for a total of 39 letters’ worth of corruption. 

Now, let’s look at the text of John 18:1-11 in 750.  Comparing 750’s text to the text of the Robinson-Pierpont Byzantine Textform (2005), I found very few disagreements in John 18:1-11:

 ● v. 2:  750 has και after συνηχθη.  (This και is noted in the Byz margin, and this και is included in the Hodges-Farstad Majority Text (1982).)

● v. 7:  750’s scribe initially seems to have written αυτοις instead of αυτους, but upon close examination, this appears to have been a glitch involving the ink on the pen; the final stroke of the υ is visible.  I think.)

● v. 11:  750 does not have σου.

 Comparing the text of John 18:1-11 in RP2005 to the text of NA27, we see the following differences (based on the footnotes in RP2005): 

1 – Byz has ὁ before ΙΣ (+1)

1 – Byz has των instead of του (+2, -2)

2 – Byz has ὁ before ΙΣ (+1)

3 – Byz has εκ των before Φαρισαίων (+5)

4 – Byz does not have και after εξηλθεν (-3)

4 – Byz has ειπεν instead of λεγει (+4, -4)

5 – Byz has ὁ ΙΣ after αυτοις (+3)

6 – Byz has οτι after αυτοις (+3)

6 – Byz has επεσον instead of επεσαν (+1, -1)

7 – Byz transposes to αυτους επηρωτησεν

7 – Byz has ειπον instead of ειπαν (+1, -1)

10 – Byz has ωτιον instead of ωτάριον (-2)

11 – Byz has σου (+3)

          Thus, the difference between John 18:1-11 in the Robinson-Pierpont Byzantine Textform and the Nestle-Aland Novum Testamentum Graece is 24 non-original letters that appear in RP2005 but not in NA27, and 13 letters that appear in NA27 but not in RP2005 – for a total difference of 37 letters.

          The difference between NA27 and the text in minuscule 750 must be slightly increased by three due to the και in v. 2, but it must also be decreased by three due to 750’s non-inclusion of σου in verse 11.  So, the text of John 18:1-11 written by the copyist of 750 is slightly more accurate than the text written by the copyist of א before correction.

          Minuscule 750 can be viewed page by page at CSNTM and via the Gallica website.