Читать книгу The Codex Mendoza: new insights - Barbara E. Mundy - Страница 10

CHAPTER 3

Report on Repairs of 1985-6, Watermarks, and Collation of Codex Mendoza

(Oxford, Bodleian Library,

MS. Arch. Selden. A. 1)

B. C. Barker-Benfield Bodleian Library, University of Oxford

Introduction

The purpose of this section is to enrich the present study of the Codex Mendoza by making available this previously unpublished report on repairs, watermarks, and collation, composed between 1985 and 1999. Not intended as a full physical description of the manuscript, this section brings together various observations which emerged before, during, and after the repairs of 1985-6, about the physical nature of the entire manuscript volume, MS. Arch. Selden. A. 1, and in particular about its paper and collation-structure. This text also records the minor structural changes made during this time.

Concerns over the physical safety of the manuscript in 1985-6 were related to the impetus for a new facsimile edition, first raised in 1985 by the University of New Mexico Press. A contract for the facsimile was signed on October 2, 1985 and the main body of 5” x 4” color transparencies were made in October-November 1985 by the Bodleian Photographic Studio (the set may have incorporated some earlier transparencies from stock). In 1986, the University of New Mexico Press decided to withdraw from the project. However, another agreement was later signed (July 21, 1989) with the University of California Press, which finally published the facsimile to coincide with the Columbus Quincentenary celebrations of 1992 (Berdan and Anawalt 1992). For these purposes, an important point to note is that the plates of the 1992 facsimile were derived from transparencies which predated almost all the repairs of 1985-6.

This essay focuses on four main areas: the 1985-6 repairs; the watermarks identified in the body of the Codex Mendoza as well as in endpapers and in the monetary tables which, although not included in this facsimile, make up the second part of MS. Arch. Selden. A. 1; the numbering systems, quire signatures, foliations, and collation of the manuscript; and a reconstruction of its binding history. The section on watermarks was originally created by the author of the present report in 1990 to accompany his selection of beta-radiographs and was reproduced in the 1992 facsimile (Berdan and Anawalt 1992, 20–23). It is reprised here with further observations. The collation notes and charts below were made directly from the manuscript and independently from the material and chart published in the 1992 facsimile (13–20, especially fig. 9) by Wayne Ruwet, who had examined the manuscript in 1971; however, Wayne Ruwet and Bruce Barker-Benfield corresponded in 1985 over some aspects of the binding, watermarks, and flyleaves. Further observations about the implications of repairs, watermarks, and collation of the Codex Mendoza for the study of its making and circulation were made by the editor of the present facsimile in his doctoral dissertation (Gómez Tejada 2012) and form part of the discussion on materiality in the present volume.

The repairs of 1985-6

A brief survey of the manuscript’s condition made in July 1985 indicated that many edges were badly frayed and that folios 71-2 had finally been detached. Folios noted as requiring repair were folios 1, 2, 7, 13, 41, 47, 51, 55, 62, 67, 68, 70, and 71. Among these, folios 1 (badly torn, with old repairs causing further damage), 67, and 70 (with a small piece in danger of detaching) needed urgent attention. Accordingly, immediate repairs were carried out between July 11 and 18, 1985 by Linda Sutherland of the Bodleian Conservation Workshop. However, on this first occasion, only folios 67 (center fore-edge, end of river on recto) and 70 (edge of writing on recto at lower fore-edge) were repaired. The repairs were made with minimal tissue (Japanese tengujo) and dry paste (wheat and potato starch).

A more extensive program of paper repair was carried out in 1986 by Nancy Bell of the Bodleian Conservation Workshop in consultation with Christopher Clarkson and Bruce Barker-Benfield. The repairs were purposely kept to a minimum: the volume was not disbound and its sewing was left undisturbed. The bulk of the work consisted of reinforcing the fraying edges with tissue. However, the first and last quires of the Codex Mendoza (Part 1) required intervention at a somewhat deeper level to counteract the damage caused by old repairs and stubs and, at the end, to reattach folios 71-2; Christopher Clarkson took an active part in these repairs.

Watermarks

Part 1 (folios 1-71): Codex Mendoza

The original text block of the Codex Mendoza is made up of paper containing six watermark patterns, some of which can be further subdivided into their twin molds. Four of the six patterns (A-D) are of the same general design which Briquet (1968) classifies under the general heading “Homme,” subheading “Pélerin” (no. 7567-7603). Apart from the differences in the letters beneath, the pilgrims themselves are close in design, each with a wide-brimmed hat, pointed chin, and crutched staff. Briquet’s examples are mostly Italian, but Valls i Subirà (1980, 2:163–65, 232–35, no. 212–227) provides a number of Spanish examples from 1500 to 1597. Whatismore, the non-Spanish examples of “Pilgrim” papers reproduced by Briquet provide the closer parallels to the watermarks of the Codex Mendoza, but none are identical and the papers are no doubt of Spanish origin. The Christian iconography of the watermarks (pilgrims, crosses) may merit some reflection in the context of the Codex Mendoza.

Pattern A: “Pilgrim”

“Pilgrim” facing left (as seen from the wire side) in a circle; beneath: two(?) capital letters, of which the first is “B” and the second is illegible. The closest parallel in the two repertories is Briquet 7586 (Provence, 1568), where the letters are “B F”: but it is not very similar.

Figure 1. Specimen: folio 2 only (figure 1, beta-radiograph).

Pattern B: “Pilgrim”

“Pilgrim” facing left (as seen from the wire side), in a circle, with no letters or other additional designs outside the circle. There is no match in Valls i Subirà; it is fairly similar but by no means identical to Briquet 7570 (Milan, 1567).

Figure 2. Specimens: (no clear division into twin molds): folios 4 (figure 2, beta-radiograph), 8, 19

Pattern C: “Pilgrim”

“Pilgrim” facing left (as seen from the wire side), in a circle; beneath, three letters, “A M F.” The only comparable example in the two repertories is Briquet 7582 (Milan, 1570), with “A M F”; it is quite a good match, but not identical.

Specimens: folios 6, 9, 10, 11, 13, 14, 15, 21, 23, 24, 26, 29, 36, 37, 38, 39, 40, 41, 42, 45, 47, 48, 51, 52.

Among these, the clearer specimens provide examples of at least two molds, e.g.:

Figure 3. (i) Folios 6, 29, 36 (figure 4, beta-radiograph), 37 (beta-radiograph), 42, 48. Lower legs shapelier

Figure 4. (ii) Folios 9, 41, 52 (beta-radiograph). Lower legs straighter

But the two molds are not differentiated on the collation chart below, as some specimens are obscured by ink or paint and thus hard to distinguish.

Pattern D: “Pilgrim”

“Pilgrim” facing left (as seen from the wire side), in a circle: beneath, two letters, perhaps “I P”. No match for a pilgrim with these letters appears in Briquet or Valls i Subirà.

Specimen: folio 71 only (figure 6, beta-radiograph).

Pattern E: Cross (Latin), in pointed shield

On each side of the cross, within the shield, is a capital letter. The design makes sense from either direction, but as seen from the wire side the letters read “F”(?) on the left and “A” on the right. Valls i Subirà (1980, 2:2:129–32, 197–202, no 79–101) describes the shield pattern as “abundant throughout Spain,” and provides examples from 1495 to 1600. Briquet reproduces examples from Spain, Italy, and France (no. 5677-5704, “très nombreux”). If the design were reversed (e.g. read from the “felt” side), the closest parallel (but by no means identical) would be Briquet 5678, with the letters “A R” (Spain, 1576. Var. simil.: Rodez, 1579; Madrid, 1586).

Specimens in two readily distinguishable molds:

Figure 5. (i) Folios 56 (beta-radiograph), 63, 64 (figure 8, beta-radiograph)

Figure 6. (ii) Folios 59 (beta-radiograph), 60, 66, 68

Pattern F: Sphere with five-pointed star above

Similar, but not identical to Briquet 14013 (Angoulême, 1570). No examples of this pattern are reproduced by Valls i Subirà.

Figure 7. Specimen: folio 69 only (figure 10, beta-radiograph)

Early endpapers associated with Part 1

(front pastedown, folios i-ii, 72)

Thevet’s signature on the former flyleaf (which is now the front pastedown) confirms its sixteenth-century date (see further below). Similarly, the date “7 Sept. 1587” on folio ii verso provides a terminus post quem non. This firm dating evidence supports the conclusion from the physical data of pleats and sewing-holes in these leaves that they pre-date the present seventeenth-century English binding.

Pattern G: Fleur-de-lis, crowned

The fleur-de-lis and its crown are of equal size; at the point where they join, the letter “A” appears on the left and “B” on the right. No Spanish examples of the crowned fleur-de-lis are reproduced by Valls i Subirà or amongst Briquet’s specimens (no. 7223-7258); the closest parallels, with different initials, are Briquet 7249 (Poitiers, 1574, with “G L”) and Piccard XIII (Hauptstaatsarchiv Stuttgart 1983, vol. 13, no. 786), (Florence 1579, with “B G”).

Specimens are not quite identical, so they presumably represent the two molds:

Figure 8. Folio i (figure 11, beta-radiograph)

The front pastedown with Thevet’s signature is probably of the same paper; now that it is pasted down, no watermark can be discerned in it, but the chain-lines are approximately the same distance apart (approx. 27-8 mm.).

Figure 9. (ii) Folio ii (figure 12, beta-radiograph)

Pattern H: Three objects and a letter

This is a difficult pattern, since the objects are highly stylized and the letter could be read as an “M” (below the objects) or as a “W” (above the objects). Since the pattern is symmetrical, the direction of viewing whether from “mold” side or “felt” side makes no difference. The 1992 facsimile showed a difference in interpretation, with Bruce Barker-Benfield (vol. I, p. 23 and fig.) interpreting the patterns as “‘W’ over three swords or daggers” and Wayne Ruwet (vol. I, pp. 14 and 16 fig. 6) as “three hats over an M.” As hats, the objects have a high crown, flat brim, and two ties or tassels hanging down; as daggers, the objects have a short, blunt blade with a broad hilt.

Ruwet is very likely correct, since he was able to adduce a small group of similar patterns with three “hats” from Heawood (1950, nos. 2596–2600); they are associated with other letters (not “M”) which are, however, useful in confirming the viewing direction as allowing the objects to be read as hats rather than as daggers. Otherwise, however, their dates are too late to offer helpful parallels, even for the period of the English seventeenth-century binding: they are from English and Italian examples dated between 1685 and 1694, with one later example circa 1809. The earlier repertories reveal nothing as close as Heawood’s examples, though there was a pattern current in sixteenth-century Germany and Austria of three helmets — quite different in shape – within a shield (Briquet nos. 1111-1115, examples dated 1525 to 1580; Piccard IX. 2 [1980], nos. 284-365, examples of 1525 to 1593).

Figure 10. Specimen: folio 72 only (figure 13, beta-radiograph)

Part 2 (folios 73-84): Monetary tables

The nine tables are each written across a large folio sheet. The nine sheets were folded into bifolia and pasted together to form ten leaves, of which each of the inside eight consists of two half-sheets pasted together (see the collation chart below). Each bifolium contains a watermark in the center of one half but no countermark in the other. Two watermark patterns can be discerned; but, unfortunately, neither of the single outside leaves (folios 73, 82) happens to be the half-sheet with a watermark, and the paste-up job makes it impossible to make beta-radiographs or to observe the fine detail well enough to distinguish between molds.

Pattern I: Fleur-de-lis in crowned shield over WR monogram

This pattern was common throughout the seventeenth century, cf. Heawood no. 1660, 1721, 1721A, 1724, 1761-2, 1768-9, 1781.

Specimens: folios 74[a], 75[a], 75[b].

Pattern J: Crozier in crowned shield over the initials NCH/M

This pattern, less common, seems very similar to Heawood no. 1199 (“Schieland Records”, 1616).

Specimens: folios 77[a], 78[a], 79[a], 80[a], 81[a], 81[b].

Lower endpapers of the seventeenth-century binding

Unlike those at the front, the lower endpapers clearly belong to the period of the present parchment-covered boards (seventeenth-century English work).

Pattern K: Bunch of grapes with countermark “A.GOUTON”

The countermark on the flyleaf (folio 85), running vertically between chain-lines with the name “A. GOUTON” in a narrow cartouche, is close to Heawood no. 2252 (London, 1607). The position of the “grapes” watermark as stuck down in the pastedown makes its pattern less clear to the eye and inaccessible to beta-radiography, but the tip of the bunch seems to terminate with a small cross, as in Heawood 2252.

Figure 11. Specimen: folio 85 (figure 14, beta-radiograph) + lower pastedown, conjoint

Numbering systems:

Quire signatures, foliation, etc.

Quire signatures

There are two types of quire signature, both of which appear in the top margin of the first leaf of the quire:

a.Near top edge, towards center. A small, neat sign written with a sharp-cut pen in dark ink, always shaped like a “7” (though clearly not intended to represent that numeral). The first appears on folio 11r (Quire II) and the last on folio 66r, the first leaf of Quire VIII. The sign is not visible on folio 1r.

Although of indeterminate date (in the absence of any palaeographically datable forms), it seems likely that this system of identical quire markings would have predated the numbered system described as (b) below. The sign at folio 21r (Quire III) is crossed out with a diagonal line (see below).

b.Top right-hand corner. An arabic numeral, plainly visible from “2” (folio 11r) to “7” (folio 56r) (none on folio 66r), written in ink and sometimes bracketed off with curved or straight lines. These were written before the folio numbers, since the latter are sometimes displaced to the left in order to allow room for them (e.g. folio 31r) and/or are separated from them by an additional diagonal line (e.g. folios 11r, 41r). The similar ink-colors, form of the numerals, and bracketing suggest that these quire numbers and the folio numbers were probably written by the same person, and, if so, without doubt, at the same time. All these quire signatures except the “2” on folio 11r have been crossed out with a diagonal ink stroke, possibly at the time of foliation; on folio 21r, the “7”-shaped quire sign is similarly crossed out.

There is no quire signature in the familiar position at folio 1r, though it is possible that a curving vertical stroke at the center top margin, crossed through with a long horizontal stroke, might have been intended as the “1” of this sequence.

Foliation

The still-used foliation of the Codex Mendoza consists of a sequence “1” to “71,” written boldly in dark ink at the top right-hand corner of each leaf; each number is marked off on the left and below with a curved line. It seems likely that the foliation was created in the seventeenth century, although it is difficult to be more precise. The fact that it does not extend into Part 2 is interesting but not decisive. The foliation is of particular importance for the order of leaves in Quire VI (folios 51-5), q.v., where the folio numbers of folios 53-5 show signs of alteration.

For some openings in the second and third sections of the Codex Mendoza, the folio number of the recto is repeated on the facing verso, in the same hand and usually with the familiar bracketing line. This happens especially at the beginning of the sections, presumably to show that the opening is to be read across as a unit of text and pictures, with “19” (on 18v) and “20” (on 19v) at the start of the tribute section, and “57” (56v), “58” (57v) and “59” (58v) at the start of the final section on the Aztec life. Whilst this could possibly mean that the foliator intended to number openings rather than folios (and maybe sometimes rather than always), the modern convention is to interpret and cite the numbers only as folio numbers (thus, the first opening of the third section is to be cited as “folios 56v-57r,” not “57-57”).

The folio numbers on the versos also occur in two openings in the middle of the tribute section, with “fo. 43.” (at folio 42v) and “fo. 44.” (43v): unusually with the word “f(oli)o” stated, but bracketed as usual. Finally, the last few openings of the section are also marked with “51” (on folio 50v), “52” (51v), “53” (52v), “54” (53v) and “55” (54v). These are entered correctly and without alteration. They may have been added there, by the seventeenth-century foliator in an attempt to arrange the disordered leaves (and corrections to the main folio numbers) in this area of the manuscript: see further at Quire VI.

Folios i-ii and 71-85 were foliated in pencil in Bodleian style, probably in the hand of Falconer Madan (late nineteenth/early twentieth centuries). Folios I and II were foliated by Bruce Barker-Benfield in 1986.

Item number (?)

Folio 1r also contains a “1.”, boldly written in ink at the lower edge of folio 1r, towards the right. This could be interpreted not as a quire signature, but as an item number of the seventeenth century or later. This number was probably not added prior to the time of the present binding, since it more or less matches in position with a “II” written at the lower edge of the first page of Part 2 on folio 73r. However, the latter’s numerals are smaller and the position not so far to the right of center, so this interpretation is not certain.

Collaboration, quire by quire

Conventions used in the following collation charts:

Watermark conventions

Part 1: Codex Mendoza with front endpapers

Front endpapers & Quire I

It seems clearest to present the complex evidence for these leaves in reverse chronological order, from the present visible evidence back to hypothetical reconstructions of their original arrangement. Accordingly, the discussion starts with a list of the changes made during the repairs of 1986 and graphic representations of the visible evidence in two charts, chart 1 “after” and chart 2 “before” the changes of 1986. Chart 3 restates the evidence of chart 2 to provide a hypothetical reconstruction of how the leaves were then attached during the period between the seventeenth-century binding and the repairs of 1986. Chart 4 focuses only the front endleaves and a hypothetical reconstruction of their state before the seventeenth-century binding, as evidence of a previous binding. While charts 5-6 present alternative reconstructions of the original structure of Quire I (omitting the endleaves).

However, it should be noted that even this series of charts offers probably only an over-simplified account of the complex history of these leaves over four and a half centuries. The intense interest which they have always attracted must have made them ever vulnerable to overhandling and, hence, to frequent running repairs over the years.

Front endpapers and Quire I:

Changes made in 1986

1.Folio i was released from the shoulder of the front pastedown to which it had been pasted.

2.Because of the build-up of paper and paste,1 the larger stubs (see charts below) were felt to be forming ridges and consequent weakening of the leaves against which they were pasted. They were released as follows:

–Stub (a) verso was released from stub (b) recto.

–Stub (b) verso was released from folio ii recto (stub [b] remains thickened with paste deposits on either side).

–Stub (c) verso was released from folio 1 recto.

–Stub (d) verso was released from folio 9 recto.

–Four strips of old repair paper at top and bottom of folios 1v and 2r (matching and presumably once conjoint to provide additional support for these leaves) were removed and discarded (these strips are visible in the Cooper Clark facsimile of 1938 and in the 1985 transparencies which were used for the facsimile of 1992).

3.Stub (c) was moved back one leaf from its position before folio 1 to a new position before folio ii, to prevent damage to folio 1 from its stiffened edge.

4.To prevent further pleating, two new flyleaves were added between folios i and ii:

Folio i was added between stub (a) and stub (b).

Folio ii was added between stub (b) and the relocated stub (c).

5.New repairing tissue had to be added across several gutters, e.g. between the front pastedown and folio i. These new joints do not imply that the adjoining leaves thus connected by the new tissue were necessarily originally conjoint.

Collation charts in reverse chronological order

Chart 1. State after the repairs of 1986: visible evidence only.

Chart 2. State before the repairs of 1986: visible evidence only

Chart 3. State before the repairs of 1986: hypothetical reconstruction (i.e. essentially as left by the seventeenth-century binder)

Notes on the stubs

1.Stub (a) and its mate (still firmly pasted to folio i verso), stub (c), and stub (d) and its mate (still pasted to folio 10v) are all of similar or perhaps the same laid paper. It seems slightly thicker than the papers of the main manuscript or of folios i-ii. Stubs (a) and (c) do not share the holes apparent at the gutters of folio i, stub (b), and folio ii (see below), and therefore were presumably added later. They may date from the time of the seventeenth-century rebinding.

2.There is no corresponding stub to provide a clue as to how stub (c) fitted into the structure. Its function could have been to help reinforce the join between folios 1 and 2 and/or to hold folios 1 and 2 in place.

3.The strip of paper which forms stub (a) and its mate is made up of two shorter strips pasted together end to end: the join occurs about two thirds of the way down. The chain-lines of both strips are horizontal in stub (a), but vertical in stubs (c) and (d).

The front endleaves before the seventeenth-century rebinding: hypothetical reconstruction

The front pastedown has no visible watermark, but is probably of the same paper as folios i-ii and could be conjoint with folio ii. The prominent vertical pleats at the center of the pastedown exactly match those of folios i-ii, showing that, like them, it was once a free-standing flyleaf before it was used as the pastedown in the seventeenth-century binding. Although the texture of stub (b) is thickened and coarsened by deposits of paste on both sides, it still seems to be conjoined at the top with folio i. Since the pastedown carries the signature of André Thevet (d. 1592), these three leaves and stub must date, at the latest, from the manuscript’s period in France during the sixteenth century and before the manuscript’s voyage to London in 1587, as implied by the dated inscription on folio ii verso.

It has been argued by Frank Lestringant (1991, 38–39)2 that Thevet’s signature on the pastedown belongs not with his dated signatures of 1553 (folios 1r, 71v, with Latinized form “Thevetus”), but rather to a later group (front pastedown and folios 2r, 70v, French form of name, bolder script). This later group is undated, but on folio 2r it is accompanied by the title Cosmographe du Roi, which Thevet assumed, according to Lestringant, only from 1568/9. If this is correct (in spite of the slight suspicion that the signature and title on folio 2r might not be uniformly written), the signature on the present front pastedown might have been written after 1568. This does not provide a similar terminus ante quem non for the creation of that leaf itself, though a rebinding at some later stage of Thevet’s ownership would offer a motive for him to have rewritten his signature there. Despite all these uncertainties, it is tempting to very tentatively suggest that the earlier binding of the Codex Mendoza, as now represented only by the earlier endleaves, was made in the twenty years or so between 1568 and 1587. This would be entirely in line with the French watermarks of the 1570s which are indicated above as (non-exact) parallels with the watermarks of folios i-ii (Pattern G).

At their gutters, folios i, stub (b), and folio ii (but not stub [a] or stub [c]) have holes which do not seem to match the seventeenth-century sewing and which presumably predate it. This evidence is confusing, but these holes were perhaps caused both by earlier damage and by an earlier sewing-structure. Similar, though less severe damage appears in the equivalent blank leaf which follows the end of the Codex Mendoza (folio 72), which may therefore be presumed to belong with this group of early endpapers.

Chart 4. The front endleaves before the seventeenth-century rebinding: hypothetical reconstruction

Original structure of Quire I: alternative reconstructions

Folio 9 can never have been conjoint either with folio 2 or with folio 10, since all three leaves carry watermarks and the directions of wire and felt sides do not fit. These types of evidence would allow folio 1 to have been conjoint either with folio 2 or with folio 10.

Charts 5-6. Original structure of Quire I: alternative reconstructions

Conclusion

Reconstruction B is the more economical solution: it assumes that a late decision was made to remake the original second leaf with a leaf of “pilgrim” paper carrying a slightly different watermark. The present folio 2 with the frontispiece is the only leaf that breaks the regular pattern of mold/felt sides. This disturbance meant that special arrangements had to be made to keep folio 9 in place after the loss of its original mate.

In Reconstruction A, the regularity of the quire was disrupted from the start, for unexplained reasons: folios 1-2 formed a separate pair of leaves (perhaps a conjoint bifolium) and folios 9-10 were two singletons which could never have been conjoint. A factor in favor of this version is that the texture of folio 1, somewhat coarse and thick, seems to match that of folio 2 better than that of folio 10; however, this effect could have been caused by the additional wear and tear to the first two leaves.

Chart 7. Quire II

Chart 8. Quire III

Chart 9. Quire IV

Chart 10. Quire V

Note on Quire V: The sequence of “mold” / “felt” sides of the paper is broken by the bifolium 43/48, where the two leaves are folded in reverse of the normal folding. However, since there is no sign of textual disturbance between folios 42v and 43r, 43v and 44r, 47v and 48r, and 48v and 49r, this reversal presumably happened accidentally when the quire of blank paper was formed before use.

Quire VI

Apart from the edge repairs, this quire was left undisturbed during the repairs of 1986. Every leaf is lined at the gutter with a stub of paper similar to that used for the stubs (a), (c), and (d) in the front endleaves and Quire I, dating perhaps from the time of the seventeenth-century rebinding. The stubs here are all still firmly pasted down, and it is impossible to tell how those attached to folios 51, 52, and 54 connect and fit into the sewing structure.

Folio 55 has long been misplaced between folios 52 and 53 (Cooper Clark notes the disturbance, vol. I p. 85 n. 1, but artificially restores folio 55 to its correct position in his facsimile of 1938, vol. III; the same policy was followed in the 1992 facsimile). Folios 55 and 53 are artificially joined by a fold of what is thought to be the seventeenth-century repairing paper to form the central bifolium of the quire made up with a thread at the gutter.

In the seventeenth-century ink foliation on the rectos of folios 55, 53, and 54, each second digit has been altered. When viewed under the video spectral comparator, it is clear that the present “53” has been altered from “54,” but the first readings of the other two numbers are not clear. As noted above, under “Foliation,” the versos in this area of the manuscript (as sometimes elsewhere) are numbered with the folio number of the facing page, presumably to show that the opening was to be read as a unit with text-facing pictures. The verso foliations here are all entered correctly for the textual content, with “53” (on 52v), nothing (55v), “54” (53v) and “55” (54v). All the folio numbers, both recto (original and corrections) and verso, seem to be by the same seventeenth-century hand. The foliation evidence is difficult to evaluate, but the alterations at least confirm some disturbance in the order already present at the time of the seventeenth-century foliation.

Chart 11. Quire VI, visible evidence

The above sequence (unlike that of chart 12) in fact fits the order suggested by the original foliation, as stated above in square brackets. In theory, foliation in this order might have taken place either before or after the disordering of the leaves. However, the most natural explanation would be that the foliator originally numbered the leaves on the rectos as he found them (perhaps already disordered). Then, after realizing the disorder, he corrected the numbers on the three rectos and added the correct equivalents on the relevant versos as well, though the misplaced folio 55 itself was left between 52 and 53 where it still remains.

Quire VI, hypothetical reconstruction of Purcha’s order (1625 or before)

Chart 12. Quire VI [cont.]

The sequence of woodcuts of the Codex Mendoza in Purchas his Pilgrimes, Part 3 (London 1625, Bodleian shelfmark Lister E 55), pp. 1100-1101, shows pictures derived incompletely from these pages in the following order: (p. 1100) Codex Mendoza folios 52r, 53r, and (p. 1101) folios 55r and 54r. These woodcuts are the basis of the reconstruction in the chart below. The text, in English translation, fits the pictures, an order which could not have been achieved without some silent editorial reordering.

Unfortunately, the chronological relationships of Purchas’s edition, the strips of repairing paper, the seventeenth-century ink foliation, and the present seventeenth -century binding are not clear. Chart 12 may put too much weight on Purchas’s order, which takes only selected elements from each picture; nevertheless, all his other woodcuts are in the correct order. His evidence seems at least sufficient to confirm that the order of Quire VI was already disturbed by the time the materials for the 1625 edition were prepared (1625 at the latest). If chart 12 genuinely reflects the physical order of the leaves in Purchas’s time, it should be sufficient to date both the seventeenth-century binding and the foliation later than that time.

Since (apart from any artificial rejoinings) all five surviving leaves of the quire are singletons which lost their conjoint leaves long ago, their order when loose may not have been obvious. The positions of folios 55 and 54 might originally have been reversed by someone who did not realize that each page of text was to be read with its facing pictures; for example, the presence of the closing title of the second section at the end of the text on folio 54v, rather than after its pictures on folio 55r might have been enough to cause the reversal of the two leaves.

Subsequently, if by Purchas’s time folios 55 and 53 were already artificially joined into a bifolium at the center of a quire, their further reversal would only have been a small step, pushing folio 55 one further step back to its present position, as in chart 11.

Original structure of Quire VI: hypothetical reconstruction

Chart 13. Quire VII

The change from Section 2 to Section 3 of the Codex Mendoza’s text coincides with the change from Quire VI to Quire VII (apparently deliberately created by the original elimination of the last five leaves of VI after folio 55) and with a change from “Pilgrim” to “Cross” paper. The structure of Quire VII could be accepted as quite regular were it not for traces of evidence between folios 64-5: escaped thread at the top and fragments of stub lower down the gutter. The appearance of one (though not both) of the standard quire signatures on folio 66 implies that folio 65 still belongs to Quire VII. At least two reconstructions are possible, which are detailed in the following sections.

Chart 14. Quire VII, reconstruction A

Quire VII, reconstruction A

This reconstruction accepts that the structure of folios 56-65 is a regular quire of ten leaves, but fails to explain the stub between 64 and 65 except through some unspecified effect of the later disturbances.

Quire VII, reconstruction B

Assuming that there is no textual disturbance between folios 64 and 65, this reconstruction assumes a similarly regular structure of bifolia in a quire of ten leaves, except that the original conjoint mate of folio 56 was eliminated after folio 64 and that the present folio 65 was an immediate replacement.

Chart 15. Quire VII, reconstruction B

Note: Folio 65 is creased and bunched at the tail of the gutter, while folio 56 shows no parallel disturbance in the equivalent area. However, the picture-cycle and text appear continuous from folio 64 to 65.

Quire VIII, state after the repairs of 1986: visible evidence only

By 1986, the last leaf of the Codex Mendoza (folio 71) and the following blank leaf (folio 72) were detached. In the repairs of 1986, repairing tissue was used to create an artificial bifolium out of these folios, which was subsequently sewn into the codex. The thread is surrounded by an additional free-standing fold of tissue. This necessitated setting folio 72 further out, making its old sewing holes and conjoint stub clearly visible. The stub at the gutter of folio 68r was released (but left in place), and the underlying surface was repaired with tissue.

Chart 16. Quire VIII, state after the repairs of 1986: visible evidence only.

Quire VIII, presumed earlier state, after the seventeenth-century rebinding

As in Quire VI, the seventeenth-century structure was artificial. Every leaf except folio 70 (and perhaps 72) was lined at the gutter with a stub of the familiar early repairing paper. The stub at folio 71v is visible both in the 1938 Cooper Clark facsimile and (though perhaps already slightly modified in shape) in the transparency made in 1985 for the 1992 facsimile. It was removed in 1986 and replaced with lighter tissue. The stub at folio 68r was released in 1986, but the rest are still stuck down. Like folios 55 + 53, folios 67 + 68 are artificially joined by a fold of the old repairing paper to form the central bifolium of the made-up quire with a thread at the gutter. However, it is now impossible to tell exactly how the stubs of folios 66v, 69r, and 71v were joined up to fit into the artificial structure. There is no sign of an old stub on either side of folio 70, but its recto is pasted at the gutter onto folio 69v.

The blank leaf, folio 72, carries old sewing holes which are similar in kind and partly in position to those of folios i-ii. The old sewing holes of folio 72 differ from the positions of those of the seventeenth-century binding. Although it carries a different watermark, folio 72 probably dates from the same period as folios i-ii and fulfilled the same function as a flyleaf at the end of the Codex Mendoza. It now serves as a blank separating the Codex Mendoza from the second part of the volume. Originally, it was set deeper into the sewing, with a fairly broad stub which perhaps folded around or into Quire VIII; the area of its former fold carries traces of discoloration which could indicate that it was once lined on the outside with a fold of repairing paper, though nothing of this now remains.

Chart 17. Quire VIII, presumed earlier state, after the seventeenth-century rebinding

It is striking that the seventeenth-century quire signature, properly “8” in the sequence, is not visible at the top right of folio 66r (there is a possible erasure at about the same place, but yielding nothing under ultra-violet light). Its absence or erasure could imply that, in that system, these final leaves were regarded as part of the preceding quire (Quire VII). If so, this might suggest that the seventeenth-century binder had somehow been manipulated to attach these leaves together to yield a large quire of folios 56-65 + 66-71 + ?72. Given the artificiality of such a structure (chart not attempted), this might still be possible even with two sets of threads, as now visible between folios 60-61 and 67-68. However, it is dangerous to base too much on the negative evidence of a missing quire signature.

Original structure of Quire VIII: hypothetical reconstruction

The direction of mold and felt sides show that none of the five leaves from 66 to 70 could have been conjoint with each other and suggest that Quire VIII resembled earlier quires in its original structure (e.g. perhaps it was originally a ten-leaf quire with folios 66-70 as the first five leaves, all with their rectos as “mold” sides). The earlier system of quire signatures yields the “7”-shaped sign as usual at the top center of folio 66r (this time without the later quire signature, as explained above). The last original leaf, folio 71 (“mold” side at verso), reintroduces a pilgrim watermark (Pattern D), but in yet another subvariety. The only leaf with which folio 71 is at all likely to have been conjoint is folio 70, though there is some difference in texture (perhaps due to the later damage).

Chart 18. Original structure of Quire VIII: hypothetical reconstruction.

Part 2 (folios 73-85):

Monetary tables and lower endpapers

The tables illustrate the comparative value of Roman and Greek monetary standards against their English and French equivalents of the later sixteenth century. The headings are in English and English silver of 1563 is mentioned in the first table (folios 73v-74r).

Each of the nine tables was written across a single large sheet; the first six are numbered 1-6 in early ink in the lower right corner. The tables were then folded and pasted together, as below, to form a clumsy booklet within a narrow fold of parchment. A note about the tables, on a smaller bifolium (folios 83-4 with no watermark), was kept with them; in this note, John Greaves questions whether, in view of their errors, the tables were really those created by Sir Thomas Smith praised in Camden’s Elizabeth.

Chart 19. Part 2 (folios 73-85): Monetary tables and lower endpapers

Codex Mendoza: Summary reconstruction of binding history3

First stage (1540s-1560s?)

Jerónimo López reported having seen a volume “with covers of parchment”, similar to and perhaps identifiable as the Codex Mendoza in the home of the Indian master-painter Francisco Gualpuyogualcal around 1541 (Nicholson 1992a, 1–2).4 However, the identification of that book as the Codex Mendoza is disputed. The Codex Mendoza itself offers no physical evidence for the existence of a binding in the first three decades or so of its existence (before and immediately after its passage from Mexico to France), but instead may rather contain some clues indicating that it had survived in a disbound condition.

The disturbances in the collation structure (e.g. the apparent replacement of the “title page” at folio 2 and the cancellations of leaves at the ends of Quire VI after the end of Section 2 at folio 55 and Quire VIII after the end of Section 3 at folio 71) can all be explained as integral to the making of the book and do not necessarily represent later accidents or losses. However, another clue about the early state of the manuscript may lie in the rough creases which are still visible especially in the first two sections but perhaps throughout the manuscript (up to folio 71); these indicate that, at some point in its history, the leaves of the Codex Mendoza had been roughly folded over twice, perhaps first vertically and then horizontally, into a quarter of its size. The similar character and locations of the creases seem to indicate that the complete manuscript was thus folded at once or that at least a few quires were simultaneously folded. Such folding could only have been possible if the manuscript had been either disbound or, at most, covered only with a single layer of limp parchment.

Examination of the creases under a magnifying glass indicates that they were formed after painting and writing: there are no signs of the disturbances in paint or ink-flow which would have been inevitable if a brush or pen were applied to creases already present in the blank paper. Similarly, the creases do not seem to appear in the front endleaves (pastedown and folios i-ii): these show vertical pleats which do not match those of the main text-block and their faint hints of horizontal indentations look like sympathetic settling against the text-block rather than the direct effects of folding. The early flyleaf at the end (folio 72) shows even less indications of creasing, except again perhaps very slight “sympathy” marks. These observations indicate that the text block had been folded before these endleaves were in place.

The evidence of the folds may be considered alongside the observation that the outside pages of the Codex Mendoza’s text block (folios lr and 71v) are browner than the inside pages of the manuscript; such discoloration is most likely explained through the absence of a cover or any other endleaves for a considerable period of time. This could mean that the manuscript did not even have a limp parchment cover. Therefore, in conclusion, it would seem that the Codex Mendoza had survived disbound for some period of its early history and during that time was roughly folded over twice into a small package—perhaps during a pirate raid.

Second stage France, between 1553 (after 1568?) and 1587

While the manuscript was in the hands of André Thevet between 1553 and 1587, it was either bound for the first time or rebound with new endleaves of probably French paper at both ends (upper pastedown, folios i-ii, 72). The parallels between the endleaves’ main watermark (Pattern G) and French watermarks of the 1570s together with the apparent lateness of the Thevet signature on what is now the upper pastedown may indicate that he had it bound later rather than earlier in his period of ownership, maybe only after 1568-9 when he assumed the title Cosmographe du roi. The date 1571, written in an inscription in mirror writing on the last page of the Codex Mendoza (folio 71v) may also be relevant, though the rest of it is too heavily crossed out to be legible. This binding must have dated from before the passage of the manuscript to England in 1587, since the inscription bearing that date is written on folio ii verso. The first system of “7”-shaped quire signatures most likely dates from the time of this binding.

Third stage (England, seventeenth century, between 1655 and 1665)

The present binding of parchment over pasteboards evidently constitutes a second (or third) binding, made in England in the seventeenth century as the watermarks of the lower endleaves indicate. It incorporates the separate English manuscript on monetary tables which constitutes Part 2 (folios 73-84). No author’s name is attached to the tables, but an attribution to Sir Thomas Smith (1513-1577) is assumed (and disputed) in the attached note at folio 83r by the Oxford mathematician John Greaves (1602-1652). It also appears in early catalogue-entries (see below). The tables are certainly on the same subject and of the same character as Sir Thomas Smith’s “tables of money,” published in an appendix to his Life by John Strype (1820, 263–73), but they are not identical. Furthermore, these tables are more elaborate and contain classical quotations which do not appear in Strype’s printed version. The watermarks and script of Part 2 suggest that it might be early seventeenth-century, somewhat later than Sir Thomas Smith’s own time.

As compared with the order of pictures reproduced in Purchas’s edition of 1625, the disturbed order of leaves in Quire VI of the Codex Mendoza may indicate that the rebinding took place after Purchas’s study of the manuscript. The second series of quire signatures and the foliation sequence almost certainly belong to this period of rebinding, as is probably the case of also most or all of the repairing guards of thick paper (stubs [a]-[d], etc.).

The volume itself seems to contain no direct evidence to allow closer dating of the present binding; dating from the watermarks of the lower endleaves is inconclusive. The possibilities would seem to be that either: (a) the binding was made between 1625 and 1654, most likely for John Selden (d. November 30, 1654) or, since his date of acquisition is uncertain, possibly for the previous owner or (b) the manuscript was bound after Selden’s death, perhaps soon after or maybe even before its arrival at the Bodleian between 1655 and 1659. Many of Selden’s printed books and manuscripts were rebound around this time; however, such bindings are usually recognizable as modest bindings covered in brown (perhaps calf) leather, with a simple fillet decoration (véase B. C. Barker-Benfield 1997, 31–34). The parchment binding, though modest and without decoration, does not seem entirely consistent with these.

The evidence of the seventeenth-century lists of Selden’s books would appear to confirm that the consolidation of Parts I and II took place after his death. A handwritten catalogue of his printed books and manuscripts (now MS. Selden Supra 111) was prepared soon after his death and before the arrival of the collection to Bodley, perhaps in 1655 while his collection was still housed at his London residence, Carmelite House (home of Elizabeth, Countess of Kent). The Codex Mendoza is listed in the section for books “In the Furthermost Roome but one Westwards:” (folios 121v-128r), in that room’s final bookpress headed “F.” Section “F” contains 88 numbered items, of which no. 31-88 appear under the subheading “Libri Arab:”. Most of F’s books are indeed oriental (if not Arabic).5 The Codex Mendoza appears at no. 51 (folio 127v): “Historia Mexicana Hispanica cum Figuris fol’. Ms’”. This corresponds almost exactly with the seventeenth-century inscriptions on folio iir of the Codex Mendoza, “MS. fo” and “Historia Mexicana, Hispanice. Cum figuris, quasi hieroglyphicis.”

This catalogue entry makes no mention of the monetary tables of Part 2, and indeed the heading at the start of “F” (folio 126v) reads, “Bookes unbound.” It seems likely that this heading may apply to all 88 books listed below. The word “Unbound,” if genuinely applicable to Codex Mendoza, nevertheless should have included at least the paper flyleaves of the sixteenth-century French binding, since they are still present today and may be stretched to cover at least a limp binding.

Some leaves earlier, the catalogue listed manuscripts “In the roome at the long end of the Librarie” (folio 103r ff.). At Shelf “C” presumably of that room, under the heading “English Manuscripts:” (folio 109r), Item 6 comprised “English & Roman Standards Compared, fo: M:S:’ ”. This title seems to be an appropriate description for Part 2 (the heading of the first table at folio 73v starts, “The first Romane Standerd of Silver…”) and the identification is confirmed by the appearance of the number “6”, boldly written in ink at the top center of folio 73r.6

The two parts are described together as one volume in an early list of John Selden’s manuscripts (Library Records b. 475), in a section headed “Libri MS Latini et Graeci in folio etc” (folios 20v-21r):

“Historia Mexicana cum figuris q(ua)si hieroglyphicis.

w(i)th Sr Th: Smith Tables of values of Roman & gr: coynes.”

(folio 20v, Col. I, third item in unnumbered list).

Parts of this catalogue, including folios 1r-7v and the list on folios 20v-21r, are in the handwriting of Thomas Lockey, who, as Bodley’s Librarian in 1660-5, was responsible for listing and arranging Selden’s books.7

A second catalogue of a similarly early date (Library Records b. 474) is also associated with Lockey’s librarianship, although it went through different hands. It lists the materials more definitely as shelved in Duke Humfrey’s Library, under the more explicit title The Catalogue of Mr: Seldens bookes at the West end of Sr: Th: Bodleys Library (folio i verso). The two parts of the manuscript are described together in similar terms under the heading “MS Graeci et lati. in folio” (folio 7v, not in Lockey’s hand): “Hist(ori)a Mexicana cum figuris ipsorum, with Sr Tho: Smith’s Tables of Roman & greek Coyns.” This catalogue is fragmented, but survives complete in another contemporary transcript (MS. Add. C. 40), where the Codex Mendoza is entered under the same heading and with the same wording on p. 35. The general title of MS. Add. C. 40 (p. xii) reads, “A Catalogue of the books given by Mr Seldens Executors to the Library of the University of Oxford”.

The catalogue evidence proves that the two parts now bound together as MS. Arch. Selden. A. 1 were still separate at the time of John Selden’s death in 1654, as catalogued in London circa 1655, but that they were listed together by Thomas Lockey during his period as Bodley’s Librarian before his resignation on November 29, 1665. However, within that period it is not clear where that binding took place, whether in London between 1655 and 1659 or in Oxford between 1655 and 1665. During that decade, a great deal of activity had taken place over Selden’s books to ensure that they were safely transferred to Oxford and made available on the shelves at the far end of Duke Humfrey’s Library. Part of that process involved binding together various thin manuscripts of booklet thickness into larger volumes, apparently on no other basis aside from approximately matching size. It was this process which left the fifteenth-century Selden Carol Book bound up not only with an eighth-century leaf of Gregory’s Pastoral Care, but also with William Cartwright’s 1630s play The Royall Slave and with other further pieces. These were assembled in one of the bindings of brown leather to form the composite volume now shelfmarked as MS. Arch. Selden. B. 26.

The exact dates of transfer for Selden’s manuscripts are problematic. It appears that some reached Oxford in 1655 and that others did not arrive until 1659 with the bulk of the library, but there are no separate listings to distinguish these; many others never reached Oxford. Because of problems with John Selden’s will, the disposal of his books became, in effect, the decision of his executors, amongst whom Sir Matthew Hale (1609-1676) took the most active role. It is not impossible that Hale could have ordered some “tidying” by means of binding up the loose materials before Oxford’s allocation of manuscripts was transferred. Otherwise, however, at least the brown leather bindings can be matched with those of non-Selden books which can be shown to have been bound at Oxford in between the 1650s and1660s, even if the style of these and of Codex Mendoza’s parchment binding are perhaps too plain to definitively determine where they were bound.

Fourth stage (Oxford, Bodleian Conservation Workshop, 1985-6)

Over the following three centuries or so of the manuscript’s period at Oxford, frequent handling may have resulted in various running repairs. A systematic program of minor repairs which did not involving major pulling or rebinding was carried out in 1985-6, as described above. Folios i and ii were added to provide further protection at the front in 1986.

Table 1. Codex Mendoza: Sample List of Beta X-Rays of Watermarks

References

Barker-Benfield, B.C. 1997. “Appendix.” In The Works of Geoffrey Chaucer and “The Kingis Quair”: A Facsimile of Bodleian Library, Oxford, MS Arch. Selden. B. 24, by Geoffrey Chaucer, edited by Julia Boffey and A. S. G. Edwards. Cambridge: D.S. Brewer.

———. 2000. “The Bindings of Codex Mendoza.” Bodleian Library Record 17 (2): 96–105.

Berdan, Frances, and Patricia Rieff Anawalt, eds. 1992. The Codex Mendoza. 4 vols. Berkeley, CA: University of California Press.

Briquet, C. M. 1968. Les filigranes: dictionnaire historique des marques du papier dès leur apparition vers 1282 jusqu’en 1600. 4 vols. Ámsterdam: Ed. A. Stevenson.

Gómez Tejada, Jorge. 2012. “Making the Codex Mendoza, Constructing the Codex Mendoza: A Reconsideration of a 16th Century Mexican Manuscript.” Tesis doctoral, Yale University.

Hauptstaatsarchiv Stuttgart. 1983. Die Wasserzeichenkartei Piccard im Hauptstaatsarchiv Stuttgart (Veroffentlichunqen der staatlichen Arcbivverwaltung Baden-Wiirttemberg, Sonderreihe). Edited by Gerhard Piccard. Vol. 13. Kohlhammer.

Heawood, Edward. 1950. Watermarks, Mainly of the 17th and 18th Centuries (Monumenta Chartae Papyraceae Historiam Illustrantia). Hilversum: The Paper Publications Society.

Lestringant, Frank. 1991. André Thevet: cosmographe des derniers Valois. Ginebra: Librairie Droz.

Nicholson, Henri B. 1992a. “The History of the Codex Mendoza.” In The Codex Mendoza, edited by Frances Berdan and Patricia Rieff Anawalt, 1:1–11. Berkeley: University of California Press.

———. 1992b. “The History of the Codex Mendoza.” In The Codex Mendoza, edited by Frances Berdan and Patricia Rieff Anawalt, 1:1–11. Berkeley: University of California Press.

Piccard, Gerhard, ed. 1980. Die Wasserzeicbenkartei Piccard Im Hauptstaatsarthiv Stuttgart (Veroffentlichunqen Der Staatlichen Archivverwaltung Baden-Württemberg, Sonderreihe. Vol. 9.

Strype, John. 1820. The Life of the Learned Sir Thomas Smith: Principal Secretary of State to King Edward the Sixth, and Queen Elizabeth. Wherein Are Discovered Many Singular Matters Relating to the State of Learning, the Reformation of Religion, and the Transactions of the Kingdom, during His Time. In All Which He Had a Great and Happy Influence. Oxford: Clarendon Press.

Valls i Subira, Oriol. 1980. The History of Paper in Spain. Vol. 2. Madrid: Empresa Nacional de Celulosas.

1 In 1985-6, the recipe for the standard paste used for tissue-repair in the Bodleian Conservation Workshop was as follows:

Purified wheat starch paste: 50g wheat starch GPR [General Purpose Reagent] from BDH (now Merck) Chemical Company; 5g potato starch GPR; 400ml de-ionized water. This paste used to adhere Japanese tissue, first water-torn to shape.

[Ex inf. Alison McKay, Head of Conservation Workshop, 14 Aug. 1998, who observed that since 1985-6 this recipe has been slightly modified.]

2 On p. 43, Lestringant casts some doubts on the 1553 date, but ultimately accepts it.

3 An extended version of this section was published by B.C. Barker-Benfield (2000, 17:96-105) as “The Bindings of Codex Mendoza.”

4 Translation of the relevant passage quoted by H. B. Nicholson in his contribution to the 1992 facsimile.

5 Items 85-87 refer to a “Rotulus,” “Liber,” and “Volume(n)” in hieroglyphics, with which it is tempting to correlate the Selden Roll (MS. Arch. Selden. A. 72(3)) and (with the second or third entry) Codex Selden (MS. Arch. Selden. A. 2); but, what was the third item?

6 Also on folio 73r, but well away at the center left on the broad parchment guard, is the letter “M”[?] in ink. Although this does not correspond to the “C” of no. 6’s shelf, this could nevertheless be another shelf-letter: similar matches between volume numbers in the catalogue’s listings, alongside differing shelf-letters, occur with other Selden manuscripts, see B. C. Barker-Benfield, op cit. supra, p. 31 and n. 8. The explanation must be simply that the shelves were allocated different letters at some point, whilst their contents retained the same order.

7 A pencil note in the hand of Richard Hunt (1908-1979) at the front of Library Records b. 475 (folio 1r) reads “The first twelve pages are written by Sr Matthew Hale and p. [left blank]”. Although this plainly indicates the hand of folios 1r-7v (and hence also that of 20v-21r), the hand is not Hale’s, but Lockey’s. This emerges through comparison with a certain example of Lockey’s handwriting in his signed letter to Archbishop Sheldon of 15 July [1664] at MS. Tanner 338, folios 180r-181v. Lockey’s script is fairly large and clear, but slightly awkward; his ampersand is distinctive, as are his final “e” and the descenders of “p” and “y”. Whilst the lists are written somewhat less carefully than the formal letter, it is clearly the same hand.