<hi rendition="#times #sc">CWDA Documentation</hi>

The following provision accounts for early modern print and scribal characters not documented by the Unicode Standard when this resource was published. ct ſh ſi ſl sp ſſ ſs ſſi ſſl Scribal abbreviation for Latin -us. Unicode U+1DD2 is provided for ‘COMBINING US ABOVE’, but the character/glyph intended here always follows the characters with which it is combined, and its vertical alignment is more often linear or even sublinear than supralinear. 9 ff ος Looks like a circled sigma. œ Resembles small Greek epsilon with dasia. font-style:italic; text-decoration:none; font-variant:small-caps; text-align:center; font-weight:bold; font-family:times-new-roman; target-name:new; -+| +-| --+| +--| -+-| ++| --| unstressed syllable stressed syllable foot boundary metrical line boundary

The original purpose of this digital companion to the Oxford edition of George Herbert: Complete Works (GHCW) was to provide its editors with reliable access to all the source materials used to construct the edited texts and critical apparatus for print publication. Our goal now is to preserve the sources as digital surrogates and transcriptions and to make them freely available in perpetuity as the Complete Works Digital Archive (CWDA).

As part of that effort, and recognizing the inevitable instability of encoding languages over time, we provide here an overview of the project contents as expressed in TEI-XML. This documentation combines (in the present file, documentation.xml) the following:

a master teiHeader element, within which are nested all project metadata, including elements whose URIs (captured as xml:id attributes) are referenced by other elements in the project files

a text element containing the prose description of the project provided here

 It is hoped that this documentation will help to ensure the project’s long term viability—especially its translatability across the fullest possible range of future text technologies and platforms. To that end, the discussion below, which assumes minimal knowledge of TEI-XML, is aimed principally at the project’s future custodians.

The editorial policy is simple: to capture the intellectual content of the primary sources in TEI-conformant XML. The only exceptions to this focus are the inclusion of tags recording the expanded forms of abbreviations; tags for names, places, and dates; and additional markup for metre and rhyme in the files containing Herbert’s devotional masterpiece, The Temple. The capture and markup otherwise serve our main objective: to present diplomatic transcriptions of all witnesses alongside digital images of the corresponding manuscripts and print exemplars.

Still, this focus is not without its challenges, especially with respect to differences in content and layout among the many source documents. The following policy statement, though not exhaustive in scope, accounts for most of the TEI-XML markup used to capture that variety.

Markup of Transcriptions: Parallel Segmentation

A discrete XML file is provided for each sub-division of the major works—whether poem, chapter, letter, oration, or note (on Valdesso’s Considerations)—and for each of the works not subject to sub-division, i.e. A Treatise of Temperance and Sobrietie, Outlandish Proverbs / Jacula Prudentum, and Herbert’s Will. Each file contains diplomatic transcriptions of all witnesses, expressed in parallel using the TEI Critical Apparatus tag set: Quâ, Fabio quondā sub duce nata salus. quæ Fabio quondam ſub duce tuta fuit. Quæ Fabio quondam ſub duce tuta fuit. This simplified rendering of l. 10 from Herbert’s Latin epigram, ‘Roma Anagr.’ (Lucus 25), includes three of the poem’s nine witnesses. Each reading element (rdg) is accompanied by a witness attribute (wit) whose value is the siglum of the corresponding source. The hashtag indicates that the attribute points to a corresponding element elsewhere in the file, in this case a single item on a list of all nine witnesses: W Fj1 D Bla41 Blh21 Fv45 NroW Bla38 Bla75 The witness attribute (wit) on the reading element (rdg), in other words, points to a witness element (witness) with the matching siglum, expressed here as the value of an xml:id attribute, which is always an identifier unique to the file (i.e. a URI or ‘Uniform Resource Identifier’). These witness elements, in addition to providing information for output processing, point to still other elements by way of the ref attribute, in this case the essential bibliographical information associated with the nine sigla, four of which are provided below: Lucus United Kingdom London Dr Williams’s Library MS. Jones B62 <hi rend="italics">Ecclesiastes Solomonis</hi>. James Duport Cambridge 1662 United Kingdom Northamptonshire Northamptonshire Record Office W(A)/6/VI/1 United Kingdom London British Library Add. MS. 6038 This listBibl feature and its child elements are nested within the teiHeader element in the file documentation.xml.

The foregoing describes in a general way the markup or ‘tagging’ protocol of the CWDA. Combined with images corresponding to all pertinent witness pages, the features documented here provide all the information necessary for tracing the origins of any given reading as captured by the transcriptions.

The first example above presents three witnesses to one line of verse in parallel rdg elements, each element containing the entirety of the line as it appears in the witness corresponding to its siglum on the wit attribute. This simple parallel rendering is representative of many of the verse lines, proverbs, and prose paragraphs as captured throughout the TEI-XML files. However, for practical reasons having to do with the handling of multiple witnesses, the introduction of new witnesses to files already extant, and to the archive’s historically varied approach to markup, other poems and prose works are presented in a manner that avoids repetition of the content. Recall the first example: Quâ, Fabio quondā sub duce nata salus. quæ Fabio quondam ſub duce tuta fuit. Quæ Fabio quondam ſub duce tuta fuit. Another way to capture the same witness data is as follows:   Quâ, quæ Quæ Fabio quond ā am s ſ ub duce nata salus tuta fuit . Here, content common to all witnesses, rather than repeated, is bounded by the opening and closing tags of a single rdg element with multiple sigla on its wit attribute. Additional app/rdg instances, nested inside the one that contains the line as a whole, capture at a more granular level all differences among the witnesses. Both versions of this example produce identical output when submitted to one and the same processing instruction—for example, one that extracts the witness data and arranges it in parallel:   [W] Quâ, Fabio quondā sub duce nata salus.  [NroW] quæ Fabio quondam ſub duce tuta fuit.  [Bla38] Quæ Fabio quondam ſub duce tuta fuit.

Note that it would also be possible, using another simple processing instruction, to produce as output from the second version of markup that of the first. This might be desirable in the interest of making the markup as humanly readable as possible, though no effort has been made at this time to apply such a transformation across the XML. Indeed, the actual markup of l. 10 from Lucus 25, including all nine witnesses, is as follows:   Quâ, Quâ Qua Quæ quæ F ff f abio quondam quond ā am quond ā am sub ſub d D uce nata salus. nata salus nata ſalus. tuta fuit. tuta fuit: tuta fuit   quæ Fabio quondam ſub duce tuta fuit.

This example also introduces several additional features of markup, the details of which are included in the following discussion.

Additional Markup Features (General)

Markup of abbreviations records both the abbreviated form and the expansion by way of abbr and expan elements nested inside a choice element. This feature allows a parser, whether for display purposes or a search-and-retrieval operation, to ‘choose’ one or the other form of the word (or both). The Latin adjective/adverb quondam, for example, though displayed in its diplomatic form, quondā (i.e. abbreviated, as it appears in witnesses Bla41 and Fj1), is captured also as an editorial expansion in the underlying codebase: quond ā am As with the verse-line example above, the same data can be captured with alternative markup: quondā quondam And as is true of the verse-line example, the markup method—in this case whole- versus partial-word—varies throughout.

Persons and places are tagged in a manner similar to the handling of source witnesses. In this case, ref attributes on persName and placeName elements point to URIs in the documentation.xml file, each of which provides information about the person or place marked by the persName or placeName element. All references to a person or place, whatever the orthographical variation or form of abbreviation, are resolved in the markup to one and the same identifier, thereby facilitating robust data search and retrieval. For example, both of the following— G: H: George Herbert and Georg Herbt —point to the same entry in the documentation: George Herbert 3 April 1593 Montgomery Castle Montgomeryshire Wales 1 March 1633 Bemerton Rectory Bemerton Montgomeryshire Wales This example does not exhaust the kinds of information that might be associated with the xml:id attribute value ‘herbertGeorge’. Nor is any of the information provided in this example required. It should also be noted that the tagging of references to persons and places is incomplete at this time, though sufficiently established as to be completed by anyone with access to the documentation provided here.

Dates in the source transcriptions are handled through simple standard-form references on the when attribute in the date element: A° Dō: Anno Domini 1632

The transcriptions, being diplomatic, rely heavily on the Unicode Standard to capture original orthography and special characters. These include Latin long-s and all ligatures, as well as scribal (and sometimes print) abbreviations that employ macrons on vowels signifying nasal consonants, and double macrons for instances signifying three or more characters. Every effort has been made to expand such abbreviations in the codebase through tags parallel to those capturing the original (i.e. the abbr/expan pairs cited in examples above).

Here, to illustrate, is a complete transcription of the opener to Herbert’s Will (copy-text version), followed by the TEI-XML from which it is generated (both versions—the copy-text original, signed by Herbert, and the registered copy):      A° Dō: 1632 T: Georgii Herbert cli͞ci def͞.               A° Dō: Anno Domini 1632 T: Testamentum Georgii Herbert cli͞ci clerici def͞. defuncti This example, as it happens, illustrates many of the features described thus far: discrete witnesses in parallel (including whitespace); dates and names (the latter including the ref attribute referencing additional information in a separate file); Unicode characters capturing nasal consonants and other scribal abbreviations; and editorial expansions of abbreviations. As elsewhere, the choice element facilitates the suppression of one alternant in the rendering (here, the content of the expan element) and expression of the other (that of the abbr element). Again, one might choose the reverse scenario through a small adjustment in the stylesheet instruction:      Anno Domini 1632 Testamentum Georgii Herbert clerici defuncti      [‘In the Year of Our Lord 163[3], George Herbert, priest, dead’]

Other special characters in early modern books and manuscripts are retained in the diplomatic transcriptions. The Unicode Standard in its current iteration provides codepoints for some of these, including Latin long-s (ſ), the ſt and st ligatures, and the various f ligatures (i.e. ff, fl, ffi, etc.). These are included with no additional markup. Characters lacking such standardization are captured in the transcriptions by way of the gaiji provision, the TEI module whose name is drawn from the Japanese for ‘external characters’ (i.e. 外字). For these, we invoke the g and glyph elements, where the former is embedded in the transcription and points to the latter, stored under the teiHeader in the documentation.xml file: ſl ſſi ος 9 ſl ſſi ος Scribal abbreviation for Latin -us. Unicode U+1DD2 is provided for ‘COMBINING US ABOVE’, but the character/glyph intended here always follows the characters with which it is combined, and its vertical alignment is more often linear or even sublinear than supralinear. 9 Note that the mapping element nested within each glyph element corresponds to its counterpart in the transcription’s g element. The gaiji module thus serves our purpose of capturing the source contents diplomatically while also providing a placeholder for future Unicode entities that correspond to the glyphs described here.

Additional Markup for Major Divisions of the Text

Below are brief descriptions of the contents of the seventeen major divisions of the CWDA text, along with any features of the tagging protocol or file arrangement particular to a division and not already discussed above.

Volume I: English Prose
The Countrey Parson

CP consists of 40 discrete XML files: 37 chapters, The Authour to the Reader, The Authour’s Prayers, and the title pages from the 1652 and 1671 editions (including Herbert’s Remains). Each line break (lb) indicator (an empty element belonging to the TEI milestoneLike model class) is assigned a URI using the xml:id attribute whose value consists of the CP chapter number, witness, page number, and line number. Here, for example, is the tag for line 25 on page 2 of chapter 1 in the 1652 edition:

A Treatise of Temperance and Sobrietie

TTS consists of a single XML file documenting four witnesses to Herbert’s translation of Cornaro’s dietary autobiography: Ts34, Ts34a, Ts36, and Tm78. The scheme for line breaks is similar to the one for CP (see above), though here there are many instances in which multiple witnesses share the same break——i.e. line 8 on page 4 in three of the four witnesses.

Common to all witnesses are two footnotes corresponding to two asterisks in the text. These are handled using the anchor and note elements, e.g. * *[content of note 1], where the value of the target attribute on the note element points to that of the xml:id attribute on the anchor element.

Notes on Valdesso’s Considerations

Herbert’s commentaries on a selection of the Hundred and Ten Considerations of Juán de Valdés (‘Valdesso’) were published in two seventeenth-century English editions, here Vc (1638) and Dc (1646). In the first, Herbert’s notes appear together in the front of the book, followed by Nicholas Ferrar’s English edition of Valdés’s Considerations. In the later edition of 1646, they are printed as marginal commentaries alongside the passages to which they correspond. Each of the twenty Considerations, together with its corresponding note(s), is handled in a discrete XML file, with a separate div (i.e. division) element for each of four parts (two for each of two witnesses). Here, for example, is the organizational structure of the file for ‘Consideration 10’ (i.e. nv10.xml):

Outlandish Proverbs and Jacula Prudentum

The collections of proverbs attributed to Herbert were printed in the seventeenth century under three titles: Outlandish Proverbs and Witt’s Recreations (1640), and Jacula Prudentum (1652). For a detailed account of the complex publication history of these collections, see the Textual Introduction for OP/JP in volume one of the Oxford edition. The 1184 proverbs are presented here as a continuous sequence spread across twenty-three files in segments of fifty items each. Twenty-one of these files cover items 1-1032 (op1-50.xml, op51-100.xml, etc.), originally published as Outlandish Proverbs; items 1033-1184, first published as part of the expanded edition Jacula Prudentum, are spread across the remaining two files (i.e. jp1033-1100.xml, jp1101-1184.xml). For reader convenience, items unnumbered in witnesses Jp52 and NLW are numbered here with surrounding square brackets; this numbering is understood to have been supplied by the CWDA editors (i.e. the tagging does not recognize its supplemental status). The result, at any rate, is a consistent cross-numbering of all three witnesses. An additional manuscript witness, Story Books of Little Gidding vol. IIa (LG), is currently untraced and not included here (though its variants, as recorded in Hutchinson, are included in the Oxford edition).

Letters

Five of Herbert’s nineteen English Letters, three of them autograph, come to us in seventeenth-century manuscripts, while the earliest extant witnesses to the remainder are found in print publications. Our diplomatic transcriptions in several instances resort to the use of the table element, together with its XML descendants, row and col, to indicate approximately the layout of information otherwise captured by the closer and signed elements alone. We do so recognizing the TEI provision that ‘it may often be difficult to make a clear distinction between details relating purely to the rendition of information and those relating to the information itself’ (chapter 15, Tables). Complementing the closer element and its child-element signed in the Letters files are the opener element and its child, salute. The head element is reserved for a letter’s title-like heading, whatever its contents, and if it appears at the beginning, prior to the opening salute (where present) or first paragraph. One exception is Letters 19, where this information appears alone on the verso of a two-page manuscript and is captured therefore using the byline element.

Herbert’s Will

A single file combines both the original (signed by the testator) and registered copies. To balance accommodation of differences in layout between the witnesses against the desire for elegant rendering in the HTML (the latter by way of the same set of stylesheets currently applied to all other members of the CWDA’s XML corpus), the Will tagging is structured as follows:

Division two contains an itemized list of what the will refers to as ‘legacyes’, identified in the encoding as ‘discharges’ (i.e. the value of an xml:id attribute). This division is sub-divided into two anonymous-block (ab) elements, one containing a brief preamble, the other containing the list itself. Unlike all but one other division of the text (postscript, discussed below) and their child elements in which the two witnesses are interwoven on a granular level, the list of discharges (ab #2) is divided into two broad rdg elements, each containing the whole of one witness or the other. This difference results from a need to facilitate the capture of complex divergences in the witnesses’ presentation of ab #2’s content.

For similar reasons, rather than using the p element, paragraphs throughout are marked at their starts by the empty milestone element, which includes in all cases the unit attribute, whose value is ‘paragraph’, together with the ed attribute, whose value is the siglum of the witness to which the paragraph break corresponds, e.g.

Division three, which begins with the Latin abbreviation ‘MS’ (i.e. memoriæ sua or ‘[for] his memory’), pertains to additional directions. This is followed by the postscript element containing the Latin text of probate, a legal requirement for entering the will into the official registry. This registered witness is presented under the siglum NAp95; the original-witness siglum is NAp36. (Additional information pertaining to these witnesses—and all others from across the edition—is provided under the sourceDesc element in the documentation.xml file, and more extensively in the Oxford edition.) A concluding closer element captures information pertaining to the notary public who examined and approved the will.

Volume II: Latin and Greek Verse and Prose
Musæ Responsoriæ

MR consists of one print and two manuscript witnesses captured across forty-three poem files, three of which are dedicatory (for James I, Prince Charles, and the bishop of Winchester, Lancelot Andrewes). A third manuscript witness is included for poem files 11, 14, and 40.

Also included here is a file capturing five witnesses to Andrew Melville’s Anti-Tami-Cami-Categoria, the 204-line poem to which Herbert’s polemical sequence is a ‘Reply’. These include the same three witnesses containing the whole of MR as well as the David Calderwood editions of Parasynagma Perthense (1620) and Altare Damascenum (1623). Rather than interwoven on a line-by-line basis, the five witnesses are captured in their entirety as separate sub-divisions of the main poem, with the corresp attribute on each div element pointing to its witness: Anti-tami-cami-categoria

Fj1 Bla41 Atc20 Atc23 D
ATCC witness Atc23, presented in two columns in the source, is captured by way of the table element: one-hundred rows in two columns for a total of two-hundred lines of verse (lines 5-8 are omitted in this witness):

Several poems in the MR sequence include special features as follows. The Duport edition (witness D) includes marginal notes—i.e. note elements whose content refers to specific lines in ATCC. Each of these is handled by way of an anchor element bearing an xml:id attribute whose value corresponds to the note element containing the Melville reference. The note in turn points to its corresponding anchor by way of the target attribute, e.g. 2 Inde ad verſ. 128. The anchor element is located at the precise place in the text to which the note refers. This arrangement allows for maximum encoding and processing flexibility—accommodating, for example, disparate locations in the file for the note element and its corresponding anchor; or, say, the creation of a list of all Melville references and their precise locations in the MR sequence. In many (but not all) such cases the anchor element is located immediately adjacent to an actual superscript footnote anchor in the source transcription, captured here by way of the hi element, which includes a corresp attribute that points to the anchor. The note itself in this case is located elsewhere in the file: Prædicat autores; 3tertia plena Deo eſt. 3 Inde 176. The content of the note element is a reference to Melville’s ATCC that in this way is tied directly to its corresponding anchor and the adjacent footnote anchor in the source. (We hasten to note that several of the D witness’s references to ATCC are inaccurate, errors first identified in Hutchinson; and that while it is possible to add a choice/sic/corr feature that would preserve the errors and provide corrections, we have not done so at this time—our objective for CWDA, as distinct from the Oxford critical edition, being limited to accurate capture of the primary-source content.)

Similar to the handling of Melville references in witness D is that of single-word glosses in the left and right margins of MR poem 33, witness Fj1, e.g. Ordo The word ‘Ordo’ in the seventh of such notes corresponds to its appearance in l. 29 of the poem. The anchor element is located at the end of the line, while its corresponding note is located elsewhere in the file. Where a poem’s note and corresponding anchor elements are located in disparate parts of the file, this is usually due to the need to capture the layout of the source using the table element. In all such cases the poem and its accompanying sets of marginal references/glosses are separated into discrete cell elements, one for the poem itself and one for each marginal grouping, whether left, right, or both. These cell elements in addition are numbered 1-3 using the n attribute: Ordo Nempe Novatores quis Veteranus amat? Militat et nobis, quem vos contemnitis, ordo: Asſecla

We note, finally, that all three of the complete witnesses to MR include, between numbered poems 30 and 31, an unnumbered copy of Herbert’s ‘Roma. Anag.’ (see Lucus, below).

Passio Discerpta

One of two autograph Latin verse sequences in W, its sole witness, PD consists of twenty-one poems in as many files.

Lucus

The second of two autograph verse sequences in witness W, the thirty-five poems in Lucus include three with additional witnesses: poems 10, 25, and 32. The file for Lucus 32 (variously titled ‘Triumphus Mortis’ and ‘Inventa Bellica’) includes, in addition to W, three seventeenth-century manuscript witnesses and a copy of the nineteenth-century Pickering edition of Herbert’s Works. Lucus 25, the notorious Rome anagram (‘Roma. Anagr.’), includes copies of the poem as it appears in the Duport edition of 1662 (D) and in seven manuscript witnesses additional to W. In two of these manuscripts (Fj1 and Bla41), and in D, the poem is situated, unnumbered, between numbered epigrams 30 and 31 of the Musæ Responsoriæ.

Memoriæ Matris Sacrum

The nineteen poems of MMS (fourteen in Latin, five in Greek) are transcribed from a single print witness published in 1627, with one exception, an incomplete manuscript copy of MMS2 (ll. 1-51). The XML does not include markup either acknowledging or correcting the many errors in the application of Greek diacritics. Here, and across the archive, the transcriptions are strictly diplomatic.

Alia Poemata

CWDA’s handling of Herbert’s miscellaneous Latin verse, eleven poems in all, is mostly unremarkable. We combine as a single item (AP1) two untitled poems ascribed to Herbert in Epicedium Cantabrigiense (1612), published on the occasion of the death of Henry, Prince of Wales. The title, given here in square brackets, is from the volume’s title page (i.e. ‘In obitum immaturum … Principis Walliæ’). The same title treatment is applied to AP4 (‘In obitum Serenissimæ Reginæ Annæ’), from the title page of the poem’s single print witness, Lacrymæ Cantabrigienses (1619).

The four poems for Francis Bacon (AP5-7 and 10) are captured here from a variety of early witnesses, manuscript and print. The most complex of these XML files contains the eleven-witness AP7, ‘In honorem Illustriss. D. D. Verulamii, Vicecomitis Sti Albani’ (four print volumes and seven manuscripts). With one exception, these witnesses are interwoven on a line-by-line basis, the granular encoding resembling the sample provided above. The exception, because it presents the poem in two columns, is captured using the table element under a single instance of the rdg element whose wit value is the siglum Dd72a.

Somewhat complex too is AP11, whose seven print witnesses present a variety of layouts, particularly with respect to stanza groupings and line breaks. For this reason, the file forgoes the lg element, capturing stanza divisions instead with the milestone element, the value of whose unit attribute is ‘stanza’, while its ed attribute bears the siglum/sigla to which the stanza break applies.

Dubia

Each of the five poems here is based on a single known manuscript witness (which is part of the reason why it is deemed to be among poemata incerti auctoris). The five are numbered relative to their candidacy for inclusion in the Herbert canon (Dubia 1 having the strongest claim, Dubia 5 the weakest—see the Oxford edition for details).

Orationes

As with the English prose, the markup for Herbert’s Latin Orationes includes lb elements, each with its own URI, e.g. , here the thirty-third line on f. C4v of the sole witness to Oratione 1. For Oratione 3 only, the medial number on the xml:id value is a page rather than folio indicator, e.g. the fifth line on p. 11:

We also provide here the English version of Oratione 1 that was published alongside its counterpart in Latine Orations (1623)—though it is uncertain whether Herbert was responsible for the translation.

Epistolæ

Of Herbert’s Latin correspondence, nineteen letters survive, sixteen of them recorded in the official ‘Orator’s Book’ of the Epistolæ Academiæ (in a hand not Herbert’s, though the headings to all sixteen and the sign-offs to three are in fact autograph). All but two of the sixteen appear in one or more additional witnesses (sigla GBE, GBZ, and/or Bla73). Witnesses to three of the nineteen letters include Herbert’s autograph originals. Though lb elements mark all line breaks in the sources, they have yet to be assigned URIs. And, as is sometimes the case elsewhere in CWDA, the source witnesses here are transcribed in their entirety as separate divisions of the single XML file containing them.

Epistolæ 17’s sole witness (Wj57) includes notes and note anchors, as well as (perhaps oddly, for a manuscript) catchwords. The notes have target attributes pointing to corresponding anchor elements placed next to superscript symbols that appear in the text. These latter are captured by hi elements, each with a corresp attribute whose value points to that of the xml:id attribute on its adjacent anchor. Notes 1 and 2 are footnotes located at the bottom of the page, while note 3 is located in the left margin next to the line of text in which its anchor occurs. The place attribute on note captures this distinction. Wj57’s three catchwords are captured using the fw element—even though catchwords in manuscripts, strictly speaking, are not forme-work features, and the third catchword here (‘ornatu’) lacks its counterpart at the top of the following page.

Epistolæ 18 is one of the few surviving Herbert autographs. Neither here nor elsewhere in this resource, however, do we include the handShift element and scribe or scribeRef attribute to identify hands. Our transcription otherwise is diplomatic throughout, capturing all Latin abbreviations and inserting their expanded forms as expan elements. To accommodate the capture of layout, the table element, child to a single closer element, contains two cell elements, one for a postscript to the letter, the other for Herbert’s signature. These are followed by a salutation (salute) addressed to the letter’s recipient, the bishop of Winchester, Lancelot Andrewes.

Volume III: English Verse
The Temple

The three essential witnesses—Williams MS. Jones B62 (W), Bodleian MS. Tanner 307 (B), and the first edition of 1633 (T)—are captured here as diplomatic transcriptions in line-by-line parallel segmentation, with one exception: ‘Easter-wings’. This most famous of Herbert’s shape poems is encoded as two parallel div elements nested within a comprehensive third div. The first of these subdivisions captures the two facing-page stanzas of witnesses W and B as the two columns of a single table element. The second subdivision is devoted exclusively to witness T, where the poem’s lines are printed vertically—top to bottom and right to left. All twenty lines for this witness are nested within a single lg element, the value of whose type attribute is ‘doubleDecet’ and whose rend attribute carries the additional values ‘center’ and ‘vertical-text’:

Lord who createdst man in wealth & store Lord, who createdst Man in wealth & store, My tender age in sorrow did beginn My tender age in sorrow did beginne:
My tender age in ſorrow did beginne:

These specifications facilitate stylesheet processing whose display output presents the witnesses in parallel while retaining their distinctive layouts: the two stanzas or ‘wings’ of W and B as horizontal lines pressing rightward, those of T as ‘ascending’ vertical lines. The twenty lines are numbered in the encoding according to their traditional verso-recto order, even if their order for witness T in the XML is, first, lines 11-20 (the recto side of the two-page opening), followed by lines 1-10 (the verso).

Here, to illustrate, are renderings using a customized iteration of the Versioning Machine:

Williams MS. Jones B62 Bodleian MS. Tanner 307 The Temple
Poems in Williams MS. Jones B62 (W)

In addition to roughly half of the poems in what would come to be known as The Temple, together with the Latin sequences discussed above, W includes six English poems that are omitted in B and T. (It is perhaps notable that the title The Temple appears nowhere in W, and that while it does appear in B, the subtitle Sacred Poems and Private Ejaculations occurs in neither manuscript, appearing for the first time in the edition of 1633.) Included here in the folder that contains the Temple poems, the six additional W files are distinguished from the others by virtue of the text element, the value of whose type attribute is ‘williamsMS’.

Other English Poems

Ten of the eleven titles included here were known to Hutchinson and are now widely accepted as canonical. Two notable additions to the textual record are a copy of the English poem to Francis Bacon (siglum Bla41), accompanied, as in the other three witnesses, by a copy of ‘Æthiopissa ambit Cestum diversi coloris virum’; and three witnesses to a poem omitted in all modern Herbert editions, ‘On the death of Mr Barker of Hammon, and his wife who dyed both together’.

Rhyme and Metrical Analysis

Herbert’s lyrical and formal sophistication is well-known. CWDA aims to enhance the study of this aspect of his verse by invoking the TEI provisions for rhyme and metrical analysis across the entirety of The Temple, namely the met, real, and rhyme attributes, as well as the rhyme element. The met attribute specifies the conventional metre within which the poet is working; the real attribute specifies the ‘realized’ pattern of the line (often a subjective value); and the rhyme attribute specifies the rhyming pattern as based on the final words or syllables of successive lines. This latter feature is complemented by the rhyme element. Not to be confused with the rhyme attribute, the rhyme element marks the actual rhyming word or syllable, and includes a label attribute specifying its role in the rhyme pattern. (This aspect of the tagging, provided in the samples here, has yet to be applied across all Temple poems.)

Here is a typical example from ‘The Church-porch’ (with the text of W only, the witness that serves as copy-text for this poem in the Oxford edition):

Thou whose sweet youth & early hopes inhance The price of thee, & mark thee for a treasure: Harken vnto a Verser, who may chance Ryme thee to good & make a bait of pleasure.   A Verſe may find him, who a sermon flies   And turne delight into a sacrifice.
The value of the met attribute on the div element is a series of symbols whose meanings are documented under the metDecl element in the documentation.xml file: the unstressed and stressed syllables (- and +), the metrical foot boundary (|), and the metrical line boundary (/). The metrical pattern, when specified on the div element, as here, is understood to recurse through the entire poem, indicating in this case that iambic pentameter is the conventional metre for all of its 462 lines. The real attribute, applied to each line (i.e. l) individually, specifies the actual pattern ‘realized’ within the conventional one, indicated here by alphabetical symbols also documented under the metDecl element—i.e. iamb (I), trochee (T), spondee (S), amphibrach (H), anapest (A), pyrrhus (P), etc. These are interchangeable (and documented as such) with combinative strings of -, +, and | symbols (-+|, +-|, ++|, -+-|, etc.) representing these metrical-foot units: -+| +-| --+| +--| -+-| ++| --| unstressed syllable stressed syllable foot boundary metrical line boundary Where a l element omits the real attribute, its realized pattern is understood to be identical with the conventional one specified by the met attribute on the div element.

Just as the value of met on div is understood recursively, so too the value of the rhyme attribute; the ababcc pattern of the present example, in other words, describes all 77 stanzas or lgs —with one exception, namely, any lg that includes its own rhyme attribute, understood in such cases to supercede the rhyme attribute on the higher-level div element.

Software programmed with these parametres in mind should be able to retrieve and process all data pertinent to metrical analysis and rhyme. In order to ensure robust processing of such data across the Temple corpus, however, other poems require a more complex treatment.

For example, ‘Antiphon’ II (titled ‘Ode’ in W) consists of three sestets with identical met and rhyme schemes, followed by a cinquain that differs in both metre and rhyme. This structure is further complicated by the presence throughout of sp (i.e. speech) and speaker elements. To accommodate this (and similar structures elsewhere in the files), the poem’s stanzas are marked not by lgs, but rather by special milestone elements, the value of whose unit attribute is always ‘stanza’, while the value of its type attribute is either ‘sestet’ or ‘cinquain’. The met and rhyme attributes in this case are assigned to neither the div nor the lg element, but rather to this special milestone element whose unit value is ‘stanza’ (and which functions effectively as a replacement for the lg element). This project’s TEI schema (the works.rng file) has been modified to accommodate these specifications by adding met and rhyme to the list of permissible attributes included under the element element, the value of whose name attribute is ‘milestone’. Here is an excerpt from ‘Antiphon’ II:

Ode. [Antiphon II] Chorus. Praised be the God of Love,      Men     Heere below:      Angels.     And heere above Cho. Who hath dealt his mercies so      Ang.     To his friend      Men.     And to his foe. Cho. Lord thou dost deserve much more      Ang.     Wee have none,      Men.     Wee have no store. Cho. Praised be the God alone Which hath made of two folds one.
Note that the real attribute, whose value expresses the realized metre in counterpoint to the conventional metre expressed by the met attribute, assumes its usual position on the l (i.e. line) element.

Still other poems are distnguished by overlapping patterns of metre and rhyme. ‘The Church-floore’, for example, consists broadly of two sections of twelve and eight lines respectively. The octave, though innovative, is fairly straightforward in terms of the encoding (six tetrameter punctuated by two pentameter lines, with a rhyme scheme resembling the sestet of an Italian sonnet to which has been added a closing couplet):    Hither sometimes sinne steales & staines    The Marbles neat, & curious vaines: But all is cleansed, when the Marble weeps.    Sometimes death, puffing at ye doore,    Blows all the dust about the floore; But while he thinkes to spoile the roome, he sweeps.    Blest be the Architect, whose art    Could build so strong in a weake heart. The twelve-line first half, however, is slightly more complex: four metrically identical tercets, across which stanzaic units plays a more extended rhyme scheme. To accommodate this overlapping structure, those twelve lines are tagged as a single line group element (i.e. lg) within which are nested four tercet lgs, a scheme rare for this project but wholly in accordance with TEI-conformant XML: Mark you the floore? that square & speckel’d stone,         Which lookes so firme & strong,                Is Patience.

CWDA Infrastructure and User Interface

All preceding documentation pertains to the capture and expression of the Herbert corpus as TEI-XML. These XML files are stored in folders whose titles correspond to those listed above. (A copy of the Temple XML files, with tags for rhyme and metrical analysis as described above, is stored in a separate folder titled TempleProsody.) Nested within each of these titled folders is an images folder that contains images of all the source pages referenced in the XML by way of attributes on the graphic and pb (i.e. pagebreak) elements: The facs (i.e. facsimile) attribute on the pb element points to its corresponding xml:id attribute on a graphic element whose url attribute (i.e. unique record locator) references the location of the folder in which the image is stored, named for its corresponding siglum and folio/page number (e.g. ec65). Each of these images is based on an uncompressed TIFF file that is processed and arranged as a series of nested raster tiles, together with an XML file specifying zoom levels and related parametres.

A separate src (i.e. source) folder contains the following additional files:

works.rng

works.xsl

works.css

works.js

worksOptionlists.xsl

 The first of these src files is the project schema, which expresses the TEI-specific constraints on the structure and content of the XML. It was created using the TEI’s Roma editor for meta-schema documentation and local encoding guidelines. Linked to the XML files by way of XML editing software (e.g. Oxygen), the RNG schema is used to validate the XML content, ensuring that it follows a set of pre-determined ‘rules’ (i.e. the schema specifications); violations are flagged by the software.

The remaining four files are the components of a customized iteration of the Versioning Machine, open-source software used to display the project’s XML-encoded content. The works.xsl file, combined with the additional CSS and JS stylesheets it invokes, transforms the XML content into Versioning Machine output. An additional XSL file invoked by works.xsl is the worksOptionlists.xsl file, which provides the project’s various tables of contents.

A Saxon XSLT processor in the Oxygen XML editor runs the stylesheets and their instructions against the TEI-XML files to create the end-user output: source transcriptions, images, and tables of contents (i.e. drop-down menus) combined in a browser-based (i.e. HTML) interface that facilitates intuitive navigation of the contents.