Richard Rolle, De amore Dei contra amatores mundi

Toward a New Critical Edition

By Madeline Freeman, Kennice Leisk, Samantha Pynes, Grace Rea, and Claire Siewert, with the assistance of Andrew Kraebel 

Latin text on parchment

Oxford, St. John’s College MS 127, f. 7r (selection)

The purpose of a critical edition is to present an annotated reconstruction of a text, noting significant variants from the surviving manuscripts, that is, the handwritten copies in which all texts circulated before the invention of print. Our critical edition offers a version of the text that we, the editors, believe to be as close as possible to what our author wrote, while also providing the means to analyze our decisions by consulting the variants (scribal errors, substitutions, omissions, additions, etc.). The text in such an edition is the product of collation, the process of comparing and weighing all the variants – every point at which the manuscripts of the text present different readings. This process allows us to organize variants and begin to understand the relationships among manuscripts. After collation, we chose a “copy text” to serve as the basis for our edition, a manuscript with minimal substantive errors, and which we believe to be relatively close to the authorial text. Still, we have adopted readings from other manuscripts when we consider our copy text to be in error. 

In our edition, annotations are presented beneath this reconstructed text. The first section, the apparatus biblicus or “biblical apparatus,” lists biblical allusions and quotations used by the author. Below that comes the much bigger apparatus criticus or “critical apparatus,” which identifies the variants in the manuscripts of the text (more on this below). In both apparatuses, entries are keyed to the line number in the edited text, followed by the lemma (the word or words being referenced) and then the relevant information, the biblical citation or variants. So, for example, in Chapter 1 of our edition, in lines 40–41, the author alludes to a passage from the Gospel of Matthew, and the following note appears in the apparatus biblicus:

40–41 preciosa—uendit] Cf. Matth. 13. 45–46

The lemma (in this case, the entire phrase from preciosa to uendit) is set off from the citation with a square bracket, and the citation begins with Cf. (confer or “compare”) to indicate that the author is drawing on some of the language of the passage without quoting it directly.

The apparatus criticus works on the same principles, though it contains several abbreviations to allow the material to be presented with greater efficiency: 

Abbreviation Description
add.  addidit, the scribe adds text after the lemma
a.c. ante correctionem, the variant indicates what the manuscript read before correction
def. deficit, the manuscript is missing the lemma due to damage
del. deleuit, the lemma is present in the manuscript but has been deleted
in marg. in margine, the variant is in the margin
om. omisit, the manuscripts listed are missing the lemma
p.c. post correctionem, the variant indicates what the manuscript read after correction
sup. superlinearis, the variant is above the line
trans. transposuit, the order of the words is switched

The apparatus criticus also makes use of another form of abbreviation, the siglum (pl., sigla), to refer to each of the manuscripts used in the edition. If there are multiple sigla in an entry, then the variant occurs in all of those manuscripts. Some entries also include the name of the previous editor, Theiner, to record when the text he printed departs from ours. Further, we have used Greek letters to stand in for select groups of manuscripts which tend to move together (i.e., to have the same variants), again, to help declutter our apparatus and to demonstrate relationships among manuscripts. (For these sigla, see the next section.)

How does this work in practice? Consider the following example from Chapter 1, line 14, one of the most straightforward forms of entry in the apparatus:

14 senciet] scencient Bo

Here, very simply, we indicate that the manuscript Bo reads scencient in place of the lemma senciet, on the one hand an orthographic variant (or change in spelling the same word, sc- for s-) acceptable in medieval Latin and which would not on its own be recorded, but, on the other hand, also changing the number of the verb (plural rather than singular). Though not stated specifically, it should be inferred that every other manuscript of the text here reads senciet (or some tolerable spelling of the same). Other cases are more complicated. Consider, for example, this entry for Chapter 2, line 16:

16 sibi uidetur] sic. φ Ba D H R T W, trans. ζ J, illi uidetur P1, sibi uideretur C1, si uideretur L, uideretur sibi E

Here, because our copy text and several typically good manuscripts seem to be in error, we first indicate which manuscripts provide the reading we have printed (the long list of sigla following sic.). The abbreviated trans. means that in J and ζ (the constant group Bo and C2) the words of the lemma were flipped, i.e., uidetur sibi, and the rest of the entry presents four seemingly independent variants (though the agreement of C1 E L on uideretur seems significant: see further below). It can be hard to get used to reading entries like this one, but it is worth the effort, since the apparatus offers an abundance of information about the transmission of the text we have edited.

We have prepared a translation, linked below, to aid in understanding the text. A reader can easily compare the translation to the Latin by line number, given at the start of each sentence in English. One might also want to focus on specific lines with variants and consider how the translation might be altered were a variant to be preferred to our text. [MF & KL]

One of the most influential religious writers of later medieval England, Richard Rolle was in some respects an unlikely spiritual authority. Leaving Oxford before taking a degree, Rolle first returned to his family home in Yorkshire, and apparently without warning he soon fled into the woods and took up the life of a hermit or religious solitary. (Typically, the hermit’s way of life was regarded as an advanced stage in the discipline of a monk or nun, but throughout the Middle Ages at least some people always seem to have adopted it without previously taking vows in religious orders.) After a few years as a hermit, Rolle began to have unusual experiences that he regarded as signs of divine favor, sensations reserved (he thought) for elite religious figures who had devoted their lives to contemplation and the love of God – that is, he says he was given glimpses into heaven, and he felt the fire of God’s love actually warming his chest, accompanied by the persistent taste of sweetness in his mouth. Not long thereafter, he says he began to hear the heavenly sound of angelic choirs, and that all of his thoughts, all of his prayers, all of his own singing, were thereafter in harmony with this celestial song. He died, presumably from the plague, on September 29, 1349, and some local followers seem to have pursued a frustrated attempt to have him canonized as a saint.

Rolle wrote voluminously across his career, and – significantly for his later influence – he worked in both Latin and the English vernacular. His writings include commentaries on books of the Bible, poetry, letters of spiritual advice for nuns and priests, and longer treatises on the ideal ordering of a life devoted to contemplating the love of God. The text edited here fits into that last category, and it seems above all to reflect Rolle’s desire to emulate the Christian spiritual texts of earlier centuries that were, in his day, considered standard authorities. Such aspirations are reflected in his title, De amore Dei contra amatores mundi (On the love of God against the lovers of the world), very clearly meant to allude to the title of St. Augustine’s magnum opus, De ciuitate Dei contra paganos (On the city of God against the pagans). Like the great monastic writers of the twelfth century, Rolle seeks to explain the nature of divine love, what it means for someone to love God, what it means to experience God’s love, and so on. Some of his discussion is positive – that is, articulating what love is – but much of it is negative, defining what Rolle understands to be properly called love in contrast to misguided ideas of love, especially love for things that are not God.

Near the end of his career, Rolle returned to this text, drawing some passages from it, and adapting and expanding others, to form part of his Latin Amendment of Life (Emendatio Vitae). This little work was by far his most successful, surviving today in over one hundred manuscript copies and translated into English several times in the century following Rolle’s death. In contrast, De amore Dei’s relatively limited circulation (discussed next) could indicate that it was tailored to suit the needs of a more specific audience. While Emendatio is a general religious manual purportedly appropriate to any and all readers, De amore Dei seems intended for those who are, like Rolle, leading the rarefied life of devout contemplation. It is meant to challenge these readers, but it also holds the author out as an example, and in so doing, it provides crucial insight into his life and thought. [AK]

Rolle’s De amore Dei survives today in nineteen extant manuscript copies, all but one of which were known by its earlier editor, Theiner. Since our edition includes only Chapters 1 and 2 of the text, we did not include two copies (Oxford, Bodleian Library MS Laud. misc 528, and New Haven, Beinecke Library MS Marston 243), both of which begin imperfectly at the start of Chapter 4. Further, earlier work by Andrew Kraebel indicates that one copy, Hereford, Cathedral Library MS O.VIII.1, was prepared directly from another surviving manuscript, Oxford, Bodleian Library MS Bodley 861, while another copy, Oxford, Bodleian Library MS Bodley 769, was prepared directly from the Hereford manuscript. Hereford and Bodley 769 therefore offer no independent witness to the text – they are what is typically termed codices descripti – and we have not included them in our sample. 

All fifteen remaining manuscripts were included in our collation, and they are presented in our apparatus criticus with the following sigla (and with links to digital facsimiles as they are publicly available):

Siglum Shelfmark
Ba Oxford, Balliol College MS 224a
Bo Oxford, Bodleian Library MS 861
C1 Cambridge, Corpus Christi College MS Parker 365
C2 Oxford, Corpus Christi College MS 193
D Durham Cathedral Library MS B.IV.35
E Cambridge, Emmanuel College MS 35
H New York, Pierpont Morgan Library MS M.872 (formerly Castle Howard)
J Oxford, St. John’s College MS 127
L Lincoln Cathedral Library MS 209
P1 Bloomington, Indiana University MS Poole 20
P2 Paris, Bibliothèque nationale de France MS lat. 15700
R Manchester, John Rylands University Library MS Lat. 395
S London, British Library, Sloane MS 2275
T Trier, Stadtbibliothek MS 685/247
W Leicester, Leicestershire PRO MS Wyggeston Hospital 10D34/15

As discussed further below, our collation indicates that a single scribal copy stands behind eight of these manuscripts, while other smaller groupings are also evident. We have used the following Greek letters to indicate a consensus reading of these manuscripts:

Siglum Consensus of...
ɑ Bo C2 H J P1 R W
β Ba C1 D E L P2 S T
ϵ Bo C2 R W
ζ Bo C2
φ P2 S

The earliest of the surviving manuscripts are likely W (c. 1350–75) and L (c. 1375–1400), while the latest are P1 and R, both from the final quarter of the fifteenth century. Virtually all of the remaining copies were prepared in the first half of the fifteenth century. With the exception of T (of Dutch or German Carthusian manufacture), all of the surviving manuscripts of the text were prepared in England, with a significant portion – Bo C2 D H L – coming from the north.

Many of the manuscripts present De amore Dei as part of a collection of religious or devotional texts by various authors. Indeed, in each of the fifteen manuscripts included in our collation, De amore Dei appears alongside other works by Rolle – though in some cases this is due to later binding of once-distinct books, rather than scribal planning from the outset. In some cases, our text was copied together with the writings of other important religious figures, including St. Isidore, St. Benedict and St. Augustine, which illustrates the high status given to the text, and to Rolle himself, by readers and scribes. 

Most of the scribes seem generally to have followed their exemplars closely, with only the occasional introduction of minor errors, such as transposition or short omissions. Significantly, E contains a major omission at the beginning of Chapter 1, lacking the start of the text before the word ardencia in l. 46, but this is due to missing leaves rather than scribal error. In contrast, the scribe of T stands out as making some very clearly intentional changes to the text, improving details he apparently found to be problematic. For example, in the sentence beginning “Amplius autem...” at l. 19 of Chapter 2, T contains two unique variants. Where his exemplar apparently read amicorum ... uitam (“the life of the friends”), T clarifies by substituting electorum ... uitam. That is, he takes these “friends” to be a euphemism for God’s chosen or “the elect.” (As it happens, we found this genitive to be most likely introduced to an antecedent copy in error, or perhaps as a gloss, and we do not print it in our text.) In another error in the same sentence, an antecedent copy of several manuscripts, including T, apparently transformed the adjective pauperam (which we print) to the nonsense word (but recognizably a noun) paupertariam, and, apparently recognizing that some error has crept into the transmission, T offers a plausible correction: paupertatem uoluntariam. This scribe clearly wanted to produce a clear, comprehensible copy of the text, and he used his understanding of Latin and his own good sense to do so. [CS]

Of the manuscripts included in our collation, the affiliation to emerge most immediately was the pair Bo C2, which agree on over fifty readings throughout our sample chapters – and, in many cases, these agreements (denoted ζ in our apparatus) are otherwise unattested and clearly scribal (e.g., ch. 1, l. 17: pastoris for pocioris, ch. 1, l. 38: poscit for possidet, ch. 2, l. 74: perfecerit perfecte for perfecte fecerit, etc.). In a smaller cluster of significant cases, Bo C2 are joined by W (e.g., ch. 2, l. 37: dulcissimum for dilectissimum), while, much more commonly, Bo C2 R W form a discrete group (denoted ϵ: see, e.g., ch. 1, l. 9: sapienciam for se ipsam, ch. 1, l. 37: Sapiencia added after est, ch. 2, l. 52: contradicentibus for contradiccionibus, etc.). Clearly, then, Bo C2 depended on a common exemplar, and another common exemplar seems likely to have been shared by R W, suggested by such unique agreements as Chapter 2, l. 11: desolacio uel temptacio for desolacionis temptacio, and Chapter 2, l. 63: melodiam for gloriam. All of this indicates that the exemplars for, on the one hand, Bo C2 and, on the other, R W, themselves descended from a single manuscript. Since, as noted above, two further manuscripts have already been identified as descending from Bo, this hypothetical ancestor accounts for six of the nineteen extant copies of the text. 

Some complicating factors must immediately be registered, however, since there is clear evidence within this group for at least one, and very likely two, instances of conflation – that is, of a scribe consulting, and taking corrections from, a second exemplar. The more clear-cut case involves R, which frequently departs from Bo C2 W, even when these manuscripts evidently present an anterior reading, and instead borrows what seem to be belated, scribal readings from the group, discussed below, identified in our collation as β. A sustained example of this conflation comes in the sentence beginning “Eternitatis utique” in Chapter 2, l. 33, but other smaller corrections appear throughout Chapter 2: ll. 5 (reducta for redacta), 46 (semel for simul) or 70 (antea for ante), and more are noted in the following paragraphs.

Less immediately apparent, but still quite likely, is the conflation of Bo C2’s exemplar with a β manuscript. (This would be, chronologically, a much earlier example of conflation, but the relatively early date of L indicates that the β hyparchetype was created early in the text’s transmission.) Here the clearest evidence comes in Chapter 2, l. 67, with the addition of the phrase ad Christum (or, in C2, Deum) graditur, added in β copies to resolve (and simplify) an unusually dense sentence. The various ways in which scribes incorporate this phrase into the authorial text (in the case of Bo C2, with an initial qua) suggest that it began as a marginal correction or gloss in the β hyparchetype, and was perhaps copied marginally in at least some subsequent manuscripts – this would make it a relatively easy bit of text for the scribe of the Bo C2 exemplar to notice when consulting a second copy. The same may be true of the added amicorum in Chapter 2, l. 19 (noted above), present in β manuscripts and, likely through conflation, Bo C2 R.

Of the remaining copies, three – H J P1 – are independent of both the Bo C2 R W group and the much larger β tradition (on which, more in a moment). Of these three, P1 is closest to Bo C2 R W (see, e.g., ch. 1, l. 26: the omission of iugiter, also dropped in Bo C2 W and presumably present in R through conflation, ch 1., l. 33: quippe for itaque, ch. 2, l. 59: postea for postmodum, etc.), though P1’s relatively late date (s. xv ex.) makes some variants less clear-cut. In Chapter 2, l. 33, for example, where we posit an anterior suauitatem, is suauitatem uel dulcedinem in P1 the source of the substituted dulcedinem in Bo C2 W (R conflated with β), or does it represent an exemplar corrected in light of the reading in Bo C2 W? Likewise, some confused abbreviation likely stands behind the different variants for idolatria in Chapter 1, l. 3 – and here we suspect that β preserves an anterior reading. While P1 shares the variant ydola ita with H W, only H P1 change the verb (inherent for inheret) to reflect the number of ydola. Is this convergent error, or does it indicate the correction of P1’s exemplar against a manuscript related to H?

At this point we have more questions than definitive answers. H J P1 are clearly very close to the archetype, and H J only once agree in a reading we have identified as an error (ch. 1, l. 79: quantum for quanto). Could H J share an exemplar? Are some minor errors in H, shared with β manuscripts, indications of conflation? Evidence for any of these problems gets very thin, and we think it better to adjudicate variants on a case-by-case basis, holding out the possibility that some more definitive evidence will emerge when these manuscripts are collated for the remaining chapters of the text. In our sample, we indicate agreement of all non-β manuscripts with the Greek letter ɑ, but we do not mean for this shorthand to indicate that (as in the β group) all of these manuscripts descended from a single hyparchetype. It seems just as likely that the exemplar(s) of J H and the ancestor of P1 + Bo C2 R W derived from the archetype independently. [GR]

Already mentioned above, the β group consists of eight manuscripts – Ba C1 D E L P2 S T –marked by a persistent pattern of minor but substantive variants, indicating their shared descent from a single antecedent copy. These variants tend to be omissions, with the scribe of this hypothesized antecedent dropping minor words that, while generally unnecessary to understand the text, offer clarity on the structure of sentences and the relationship between clauses. The omission of conjunctions (e.g., ch. 1, l. 11: et ... et, ch. 1, ll. 35–36: et ... quoque, ch. 2, l. 4: atque, etc.) or sentence adverbs (e.g., ch. 1, l. 20: utique, ch. 2, l. 43: namque, etc.) is especially common, while in other cases the scribe drops words that can readily be implied from context (e.g., ch. 1, l. 67: esse). Erroneous additions are less common, though we suspect that amicorum in ch. 2, l. 19 (mentioned above) would fit that description, and some substitutions seem to arise from a desire to clarify or improve the text (e.g., ch. 1, l. 36, deliciis, to avoid the repeated diuiciis, perceived as inelegant, or, more extensively, the sentence beginning “Eternitatis utique” in ch. 2, l. 32), or from simple scribal error (e.g., ch. 2, l. 46: semel for simul). Overall, then, it would seem that the scribe of the common antecedent for these manuscripts sought to prepare a relatively “trim” text, abbreviating his exemplar by eliminating words deemed unnecessary, and making what he may have considered corrections as he went.

The relationship of the various manuscripts within the β group is harder to determine, though some affiliations are clear enough. C1 E evidently shared an exemplar, indicated by their unique substantive agreements (e.g., ch. 2, l. 35: nequaquam for non, or, yet more tellingly, ch. 2, l. 77, retaining from this exemplar an erroneous perturbacione uel before temptatione), with many further errors unique to each manuscript. P2 S likewise vary as a pair with considerable frequency (denoted with φ in our apparatus), but this may reflect (as Ralph Hanna has argued) the use of P2 as exemplar for S. Then again, while many of the remaining unique errors in P2 may have been identified and corrected by the scribe of S, it seems difficult to imagine that scribe independently and of his own accord generating the word iam, in Chapter 1, l. 27, omitted in P2. Perhaps, then, this pair also shared an exemplar.

These two pairs – C1 E and P2 S – also show some signs of conflation against manuscripts in the ɑ group. Hence P2 S agree with ɑ manuscripts in Chapter 1, l. 18 querere against adquirere in other β copies, and they agree in the harder tanto Christi reading with H J (and, mutatis mutandis, W) against the various easier solutions given in other manuscripts at Chapter 2, l. 40. The case of C1 E is a bit different: here the two manuscripts at times agree with ɑ readings independently of one another (e.g., C1 including the enclitic -que in ch. 2, l. 42, where E agrees with other β copies; or E including mundi in ch. 2, l. 18, where in this case C1 agrees with other β copies). In all likelihood, then, the scribes of C1 E worked from a corrected exemplar, independently incorporating some, but not all, of its corrections into the texts they copied.

Of the remaining β manuscripts, Ba sometimes shares correct readings with ɑ copies (e.g., ch. 2, l. 13: queretur). While Ba may present another instance of conflation, then, it seems more likely simply to be a fairly early copy in the transmission of the β group, and it indeed contains a relatively low number of unique variants. L more frequently shares errors with other β manuscripts: see, e.g., Chapter 2, l. 16, where eleuat is given as eleuauit in D L and, through a further error, leuauit in C1 E. As suggested above, T often varies independently, reflecting its scribe’s desire to improve the text, and, in this case, it is hard to know whether T’s eleuat represents access to a relatively good exemplar, or the scribe’s correction of eleuauit to match the present tense of terret earlier in the sentence. D is yet more prone to independent variation and careless error, making its relationship to the rest of the group particularly difficult to ascertain.

Within this group, then, Ba and L seem to be the best witnesses to the text as it was transmitted by the β hyparchetype. [SP]

De amore Dei has been edited once before, by Paul Theiner, in an edition published in 1968 by the University of California Press, revising his Harvard dissertation (1962). In his introduction (57–62), Theiner offers a brief account of the relationship among the surviving manuscripts, ultimately determining to base his edition on the manuscript we have assigned the siglum R. His rationale, however, has several limitations. At the outset, Theiner divides the manuscripts on the basis of their arrangement of the text in six or seven chapters – though he incorrectly identifies Ba as a six-chapter text and fails to notice that C1 is, on closer inspection, divided into seven chapters, simply lacking a decorated initial at the start of Chapter 2 (to be supplied in the margin, with a guide letter visible). This leads him more or less to distinguish between, on the one hand, the constant group we have identified as ϵ (with the two Bo derivatives, noted above) and, on the other, all of the remaining copies. While he tentatively notes some of the constant groups likewise identified in our collation (C1 E, for example, and P2 S), his approach generally gives too much weight to the larger issue of the division of the text, at the expense of the patterns of shared error that can cast more light on the work’s transmission.

Theiner’s favoring of R as copy text is, likewise, problematic. He observes that R, like other ϵ manuscripts, fails to break the text at the beginning of Chapter 2, but that, at the start of what is therefore presented as Chapter 2 (properly Chapter 3) in the manuscript, the scribe has added the label “capitulum iii” in red ink. He concludes that, in R’s exemplar, “the chapter divisions were somewhat uncertain” (59). Such an account, while possible, does not reckon with the clear evidence of the variants, which (as noted above) indicate that R is in fact conflated. Further, while comparing exemplars in search of a satisfactory form of the text, the scribe of R made some obvious improvements of his own, resulting in a significant number of easier readings (see, e.g., his thorough revision of the sentence beginning “Quoniam quidem inter mortales” in ch. 2, starting at l. 66).

The decision to use R as copy text would not in itself be problematic, if Theiner had been willing to emend the text where needed. But, like earlier editors of Rolle’s Latin, Theiner prefers “best text” procedures, and he therefore aims to print what he finds in R “without any changes beyond those necessary to make literal sense of the text” – though he “confess[es] succumbing in one or two places to the use of an alternative reading” (61). Our notes for the first two chapters identify 60 points at which we prefer a different reading than what he printed, 15 of these being places where he gives a reading found uniquely in R. His critical apparatus is, likewise, partial and often inaccurate in what it does report – though, commendably, we have found only one typographical error in his text (quit for quia in ch. 2 l. 71) and one incorrectly expanded abbreviation (pro for pre in ch. 2 l. 78).

A new edition is clearly needed, and our presentation of the first two chapters indicates the lines along with such work should proceed. We have used J as our copy text, preferred as a representative of the ɑ group (see above) with the fewest number of unique scribal variants – that is, places where J was the only manuscript that presented an otherwise unattested reading. To be sure, J is not without its errors, and there are places where we have adopted a reading found in other copies. In many cases, this has simply involved restoring the reading found in all or close to all of the other manuscripts. In others, we have had to weigh the variants with greater care. In Chapter 1, l. 35, for example, uncreated Wisdom (increata Sapiencia) is, uniquely in J, the subject of the verb erigit (she “raises up” the poor), which seems close to the β reading, eripit (“snatches away”), also found in Bo and, likely under the influence of β, R, while H gives the same verb in the perfect tense, eripuit. C2 P1, however, read excipit (“she receives” or “selects,” perhaps here in the sense of “she favors” or “prefers”), and W gives a variant spelling of the same, exipit. That last spelling could be responsible for the readings in β Bo H J R, with x misread as r, and J may then represent a further effort to make sense of a confusing word, one which the scribes of C2 P1 have, independently, written in its orthographically more familiar form. We have therefore printed excipit, a harder word, and one which builds on the sense of the preceding sentences. Further, as this example indicates, W and, though late, P1 have often proven useful in adjudicating between variants, and both of these manuscripts should be taken into account by future editors of the full text.

Apart from cases where they are significant (e.g., excipit), we have generally not included variant spellings in our apparatus, but we have otherwise aimed to be as inclusive as possible, preferring to give too much information to too little, with the intention of illustrating more fully the value of each of our fifteen manuscripts as witnesses to the text. On the basis of what we present here, we believe that a future edition could proceed with J as copy text, with variants drawn from H P1 W and, representing the β tradition, Ba L. An edition prepared along these lines will result in a version of the text that is quite close to what the author wrote, while also, in its variants from β, providing the major alternative form in which it circulated, especially outside the North of England. [MF & KL]


View the Edition  View the Translation 


We would like to thank Dr. Tim O’Sullivan in Trinity’s Classics Department for supporting this class and encouraging our work. Special thanks go to the other members of the Early Book and Manuscripts Lab, who have contributed to the transcription of manuscripts of De amore Dei, but who were not part of the class in which we edited the text: Sarah Henderson, Stephanie Gredell, Rebecca Kroger and Jenna Shultz. We are grateful to Dr. Liz Hebbard for supplying images of P1, and to Dr. Petra Hofmann for supplying images of J, and for giving us permission to reproduce the selection from J, above. Our thanks to the imaging services team at the Rylands Library for putting R in the queue for digitalization, and to Dr. Lauren Turek and Martina Mauldin for supporting our work through the Mellon Initiative, which covered the expenses of the remaining digitalization orders.