Some years ago I met with a colleague briefly to discuss my plans to build an electronic edition of George Herbert’s English verse. I broached the topic with some trepidation, for Randy McCloud’s Herbert essays are exercises in un-editing, what he calls a “critique of editorial response” to the manuscripts and early print volumes of The Temple (Cloud, “fiat flux” 61).  I described my plans: that the Digital Temple would include high-resolution images of two manuscripts and a copy of the first edition (1633) alongside diplomatic transcriptions of all three. There would also be a user interface for viewing these materials in a number of ways: as individual documents; in a variety of frames formats; in modern-spelling versions; with links to the digital images; in some user-determined combination of these options. Insofar as the project would bring together all of the relevant primary documents, Randy’s (or Randall’s or Random’s) response was encouraging. But I strongly suspect he doubted the claim that my edition would escape falling prey to the mischief of his pseudonymous colleagues.
By permitting access to images of the source documents, the Digital Temple exposes its transcriptions to the scrutiny of the user. It also aims to preserve the peculiarities and ambiguities that, in Jerome McGann’s assessment, are “expelled” from the “output of TEI’s markup constraints” (200). The Digital Temple, then, is the kind of edition described by Martha Nell Smith. The “primary goal” of digital editing, she writes, “is not to dictate what can be seen but rather to open up ways of seeing. The disambiguating codes [of the TEI] are tools to understand texts, not to define them . . . In fact, the different perspectives offered via the encoding” and “image-based digital surrogates . . . create a climate of possibility for interpretation (315-16).” Such an interpretive climate for a Herbert edition is appropriate, for as I hope to demonstrate below, conventional print-based editions of this canonical poet have tended to obscure otherwise significant features of the source documents—features which highlight The Temple’s historical position at the convergence of manuscript and print cultures.
Many would endorse the poststructuralist and democratizing spirit of Smith’s and McGann’s remarks. It is crucial to note, however, that both acknowledge that definition and disambiguation are necessary: McGann implicitly in his “mark[ings],” however multiple, of a poem’s “poetic fields” (208-11); Smith more explicitly by allowing that codification is unavoidable. Disambiguation is not to be shunned, but rather viewed as the provisional and perhaps temporary effort to read: a marked transcription is the trace of a reading in a particular historical moment. Recall that Blake’s warning, “Expect poison from standing water,” is accompanied in The Marriage of Heaven and Hell by the prophet Isaiah’s observation (to Blake in a dream) that “many are not capable of a firm persuasion of anything” (Blake 37, 39). This healthy tension between definition and provisionality is a rhetorical feature of the kind of electronic edition Smith envisions and which I hope the Digital Temple to be.
However, when it comes to marking texts for digital storage, display, data retrieval, and computer-assisted text analysis, the democratizing ideal is somewhat untenable, even when the goal is to create and encode diplomatic transcriptions of all sources. The encoder makes decisions at every turn not only about which features of each text to mark and which to ignore, but frequently about what is actually there to begin with. Images of the original documents may be readily available, but unless the researcher has also encoded the transcriptions, the analysis will be constrained by an editor’s tagging. I can only imagine what McCloud might do with an encoded Herbert—how a multiply transcribed and displayed “Easter Wings,” for example, could turn the activity of “un-editing” into de-encoding or anti-tagging or (dare I say?) parallel fragmentation. For tagging is interpretation, is critical reading. Indeed, text encoding, the central activity of digital scholarly editing, is a discipline akin to literary analysis because equally burdened with the demands of interpretive and ideological responsibility.
Before exploring this problem more fully, I begin by noting several examples that demonstrate the advantages of a digitized Temple. The first is Herbert’s poem “The Altar.” The penultimate line of this shape poem in the 1633 edition is “O let thy bleſſed SACRIFICE be mine” (Grosart 18). The Bodleian manuscript, perhaps the copy text for 1633, also has “blessed” (Charles and Di Cesare 15v). In the Williams manuscript, transcribed by an amanuensis and emended by Herbert some fifteen years earlier, “bleſsed” replaces “onely” which has been struck, presumably by the author (Figure 1):
Figure 1: “The Altar.” Williams MS. Jones B62, 15v
From a prosodic or any other aesthetic point of view, the change is negligible. From a historical-theological perspective, however, it is significant. For “onely sacrifice” might have seemed a stridently Protestant emphasis on the uniqueness of the Atonement at the expense of more Catholic notions of Real Presence and the replication of sacrifice in the ritual of the Eucharist. Changing “onely” to “blessed” replaces a doctrinal assertion with a gesture of adoration. The notion of Christ’s sacrifice being “mine”—radically here, now, in this moment of communal worship—is no longer qualified by the restrictive “onely.” This is early evidence of what Anthony Milton calls “negative popery,” the tendency among some late Jacobean and early Caroline divines to introduce innovative or potentially controversial ideas minus the usual caveats affirming the English church’s distinct confessional identity vis-à-vis the Roman church (63-72, 471-72, 541-44). With “onely” removed, the Williams “Altar” changes from a poem both linguistically and visually evocative of an altar of sacrifice, yet conscious of and determined to restrain its popish sentiment, to a poem no longer marked in a way that would distinguish its avant-garde enthusiasms from actual popery. 
Seeing a transcription in which “onely” is struck, but not removed, and accompanied by the emendation “bleſsed” is a reading experience very different from that of an edition in which the emendation is relegated to a footnote and the emended version privileged because assumed to be the author’s final intention. The former derives from an editorial theory according to which final intention is not the only one, nor even the most significant. It is a theory that not only values readings rather than a reading (or the reading), but also the reading of readings—of corrections and emendations as telling a story about the world of the text, its ontogeny as a living artifact.
Take as further example a typographic feature exclusive to the same poem as it appears in the first edition (Figure 2). Here, the Cambridge printers Thomas Buck and Roger Daniel have rendered “heart,” “sacrifice,” and both occurrences of “Altar” in uppercase type with spaces between the letters—
H E A R T, A L T A R, S A C R I F I C E—perhaps visually evoking the broken body associated with the bread of the Eucharist:
Figure 2: “The Altar.” (Grosart 18).
Modern editions—including F. E. Hutchinson’s and, more recently, Helen Wilcox’s—reproduce the Buck/Daniel innovation even though there is no evidence supporting it in either of the two manuscripts (Hutchinson 26; Wilcox 92). An otherwise decorative feature, this typographic novelty assumes a special connection among certain words at the poem’s theological/ecclesiological centre—the relative prominence of ceremonial and devotional pieties—and is therefore ideological in effect. Again, seeing the innovation is one thing; seeing it as innovation is quite another, a perspective not only possible but more likely in a digital environment that displays the 1633 edition alongside the two manuscripts. 
Editors have inadequately represented the Williams manuscript in particular. They privilege instead the later Bodleian manuscript and 1633 edition not only because these contain almost all of Herbert’s total known oeuvre, but also because they are presumed to embody a final and thus pristine version of poems that first appeared in the Williams manuscript. As the foregoing example demonstrates, differences between earlier and later versions of his poems might tell us something significant about Herbert’s theological itinerary at a time when the struggle for the English church’s confessional identity had intensified (Lake, “Calvinism” 53-66; Milton 418-39). The Bodleian manuscript was transcribed by Herbert’s friends at Little Gidding shortly after the poet’s death in 1633. The source of that transcription, what Walton called a “little book” (Walton 286), presumably in Herbert’s hand, is now believed lost. Poems in the Williams manuscript (representing roughly half the total number in The Temple) were composed some fifteen to twenty years earlier, their order determined perhaps as early as 1618 and the volume transcribed by an amanuensis and corrected in Herbert’s hand shortly thereafter (Charles xv-xxviii). Herbert that year was appointed Reader in Rhetoric and two years later became Public Orator at Cambridge, a post he relinquished to enter the priesthood only in 1626. It is not unreasonable to speculate whether the younger man with secular aspirations might have held—or was more willing to express—views different from those of the ordained minister. At the very least, the correction in “The Altar” suggests a mind concerned about theological niceties at a time when James I’s centrist divines were busy anatomizing doctrinal matters along with their Protestant sister churches in Europe. Perhaps Herbert’s work might soon undergo the historicist scrutiny enjoyed by Donne’s in recent years, the Digital Temple enhancing efforts to see thispastor mansuetus as a centrist divine attuned to early Stuart religious controversy. 
In addition to a theological/political dimension, parallel display of the several sources reveals interesting aesthetic details and differences. The third line of W’s “Prayer ,” for example, reads “The soule in para’frase, Hart in pilgrimage” (34r), while the Bodleian manuscript (B) has “The soule in paraphrase, heart in pilgrimage” (33r). Both Hutchinson’s and Wilcox’s transcriptions follow B (and 1633) even though the apostrophe in W signifies an elision that maintains perfect pentametre and the other witnesses allow the tri-syllabic “paraphrase”  (Figure 3):
Figure 3: From “Prayer ,” Williams MS. Jones B62, 34r.
We know that what appears to be an extra syllable is often easily elided by the ear. But whatever we might think or the Bodleian scribes and Cambridge printers thought, the Williams scribe (and, presumably, Herbert himself) felt it necessary to elide “parafrase” as “para’frase.” That such metrical precision mattered to Herbert is evident in, for example, “Affliction .” In W, the third line of the ninth stanza has a caret adding an e to “throwst,” thereby rounding out an otherwise nine-syllable line: “Turning my purge to food: thou throwest mee” (Figure 4):
Figure 4: From “Affliction ,” Williams MS. Jones B62, 49v
The B scribe apparently followed a corrected version, duplicated in Hutchinson and Wilcox.  But the experience of reading the Hutchinson, Wilcox, Bodleian, or Buck/Daniel rendering of the poem is not the same as reading W’s. The former show a finished poem, the latter a work in progress—and the emendation a meticulous attention to aural dynamics.
These few observations only begin to suggest the advantages of parallel display in a digital edition. To cite a more famous example, the two stanzas of “Easter Wings” in the 1633 edition are printed vertically down rather than horizontally across their respective verso and recto pages, the first line of each stanza running adjacent to the right margin. How did seventeenth-century readers experience this poem? Did they turn the book leftward 90º and read the recto page of the opening first, thus reversing the order of the stanzas read in the usual verso-to-recto manner (as would be the case with both the Williams and Bodleian manuscripts where the stanzas’ orientation is horizontal)? Like their uppercase renderings of sacramental words in “The Altar,” the Buck/Daniel innovation here is an editorial imposition—in this case reifying the title in the layout of the stanzas: each has become a symmetrical pair of wings. Hutchinson’s 1941 Oxford edition, the standard until recently, retains the manuscripts’ horizontal lineation. The more recent Cambridge edition includes images of B and 1633. Wilcox’s transcription, however, follows the latter’s vertical rendering, as well as its slight curvature at the stanzas’ top margins and more pronounced curve at the bottoms—which would be respectively the left and right margins if the poem were rendered horizontal. In B, on the other hand, not only are these curvatures reversed (rendered vertically, the winged stanzas would be flying downward rather than upward—hardly an image of heavenly ascent), but the slighter of the two curvatures is far slighter still. Moreover, in the W image, which Wilcox does not include, the right-margin alignment is total (Figure 5):
Figure 5: “Easter Wings,” Williams MS. Jones B62, 27v-28r
In discussing preference for the vertical versus horizontal arrangement of the poem, Wilcox cites those critics who surmise “Herbert’s conscious control over his materials” (Wilcox 146), even though the poet is unlikely to have had a hand in the novel layout because he died before work on the book began. Of the three artifacts, only W has evidence of Herbert’s hand—and its rendering of “Easter Wings” bears little resemblance to that of the first and most subsequent editions.
Notice too that in all three witnesses both stanzas are titled. Titles elsewhere in 1633 appear only at the beginnings of poems; there are no running headers other than those indicating the volume’s several sections (“The Church-porch,” “The Church,” and “The Church Militant”), though the last is also the title of that section’s single poem. Similarly in W and B, titles are underlined and not reproduced on succeeding pages (except where a second poem with the same title is clearly intended and has been read as such by most editors). Yet, a single title appearing over her transcription (and, albeit, acknowledging the treatment of the titles in all three sources), Wilcox avers that “in the unusual circumstances of this pattern poem spread across two pages, it seems more likely” that the title is “a running title” (Wilcox 147). This is a perfectly reasonable rationale; but reducing the two titles to one in her transcription, Wilcox tacitly (and in my view unnecessarily) disambiguates this aspect of the reading experience.
The foregoing descriptions are no substitute for the experience of seeing and navigating among the several transcriptions and source images. (Note, as further example, the extensive emendations in the W “Easter Wings” pictured above.) However, while multiple display eliminates the problem of having to decide among variant readings—and thus which to include in the main text and which in the apparatus—the digital editor is in no way absolved of critical responsibility when it comes to transcribing, encoding, and establishing the text of each artifact individually. This is particularly evident with respect to the manuscripts, of which Williams might stand as a representative example. My focus here is the interrelationships among certain paleographic features, the manuscript’s intellectual content, and the critical activities of transcription and encoding.
How, for example, is one to distinguish between majuscule and miniscule characters—and why does it matter? Take the poem “Love,” exclusive to Williams (Figure 6):
Figure 6: “Love,” Williams MS. Jones B62, 38v-39r
In their transcriptions both Hutchinson and Wilcox render as uppercase the initial l’s in the loves of the title, stanza 1 line 1, and stanza 4 line 5 (Hutchinson 201-02; Wilcox 12). Though the title’s initial is slightly larger, its form is indistinguishable from the others, so the editors’ decision to give uppercase initials to all occurrences of the poem’s titular word is sound. To the word lend in the final line of stanza 3, however, Hutchinson and Wilcox give a lowercase initial even though that l is closer in size to that of the title than to those of the other two occurrences of love. Again, the editors’ reading of the loves is perhaps appropriate given the poem’s title. And yet unlike the more famous “Love [III],” where the title refers to a personification addressed directly by the poet, this poem’s love is the abstract quality or attribute. I sympathize with Hutchinson’s/Wilcox’s reading, especially given the poem’s final line with its majuscule initials. The devout reader might be anxious that love (Love?) in the penultimate line be as prominent as possible and thus anticipate (or “prevent,” Herbert might say) the emphatic conclusion: “Whereas if I orecome thee & thy love [Love] / Hell, Death & Divel come short of mee” (39r). Surely the imposing and more obviously majuscule “Hell, Death & Divel” will “come short” of the speaker only if his qualifying bout has been with “Love” rather than merely “love.” Then again, an apparently timid but in fact victorious “love” would be consistent with the quiet strength Herbert typically ascribes to grace. In their transcriptions Hutchinson and Wilcox render uppercase the initials for all four words. For the following poem (“Sinne ”), however, they prefer B’s miniscule initials for apparitions, sinn, death, and sinnes in the second stanza (42v) to W’s majuscule renderings (Figure 6). Hutchinson’s apparatus ignores altogether this detail, and Wilcox goes so far as to aver that in B and 1633 “the text is unchanged” (Hutchinson 63; Wilcox 228-29).
But does this finger-pointing, one editor at another, really get us anywhere? If for the poem “Love” I designate all loves’ initials (excepting the title) miniscule based on the evidence of lend, am I not reading lend as a relatively insignificant word even as Hutchinson and Wilcox interpret love (with good cause) as significant? Well, I suppose I can point to the title’s L and that of Lord in the first stanza’s final line: these are more similar to each other than to the other occurrences of love. So my final decision is to go against Hutchinson/Wilcox and render all occurrences of love (excepting the title) lowercase. But the process of arriving at this decision disabuses me of the notion that a diplomatic transcription can be disinterested and objective. A key question, then, is whether the tagging should document such editorial doubt. The TEI provides guidelines for dealing with multiple readings within a single witness; but such instances are so numerous as to make detailed tagging highly prohibitive. But even from a theoretical perspective, the attempt to avoid interpretive decisions denies the fact that tagging or editing in general is a form of critical reading. Emphasizing ambiguity, moreover, is itself an ideologically invested interpretive strategy. One might question whether the status of the final word’s initial in the penultimate line of “Love” is in fact ambiguous—or even whether it matters.
Transcribing diplomatically is not as straightforward as it might seem. Transcription from manuscript is, after all, a kind of markup, an encoding of scribal forms as print fonts. Another challenge to the disambiguating rhetoric of any encoding protocol is tagging for structural and formal features—especially when the goal is a parallel-text edition. Consider, for example, “The British Church,” appearing in B and 1633 only. Here is a section of the poem, with an image from the first edition, rendered in the Digital Temple interface (Figure 7):
Figure 7: “The British Church,” B (left) and 1633 (right) transcriptions, with 1633 image (p.102) 
Wilcox’s transcription follows this arrangement of the poem as ten three-line stanzas. Hutchinson’s follows B (Hutchinson 109-10; Wilcox 390-91). Figure 8 shows the image and the Digital Temple’s diplomatic transcription (the first three lines are actually lines 4-6 of the first stanza interrupted by a page break):
Figure 8: “The British Church,” transcription and image from Bodleian MS. Tanner 307 (77r)
This arrangement of five six-line stanzas better reflects the poem’s rhyme scheme, AABCCB being the smallest repeatable unit, reinforced visually in B but not in 1633. Buck and Daniel (followed by Wilcox) might have determined that metrical pattern rather than rhyme scheme should determine stanzaic division—two tetrametre lines plus one dimetre line constituting each three-line unit. But look at the poem “Evensong,” also in B and 1633 only. The first edition has eight four-line stanzas; B has four eight-line stanzas. Figure 9 shows transcriptions of the first 2/4 stanzas in parallel display, sans images (you’ll have to trust me):
Figure 9: Transcriptions from “Evensong,” B (left) and 1633 (right)
Again, the formal criteria for determining stanzaic unit should include more than spatial arrangement. Here, however, it is not rhyme scheme but rather metre that is the primary determining factor: trimetre, pentametre, tetrametre, tetrametre, trimetre, tetrametre, tetrametre, pentametre—this pattern repeated three times for a total of thirty-two lines. The longer stanza, reinforced visually in B, is a subtle feature of the poem’s music. Given Herbert’s significant contribution to English verse forms, it is important to foreground this visual complement to aural effect if and where there is evidence to support it.
Parallel display, of course, obviates the need to decide the issue either way. I’ve argued that Hutchinson’s 1941 transcription should be upheld on aesthetic grounds and that Wilcox’s 2007 adherence to 1633 should admit of exception where appropriate (this is poetry, after all—to be experienced first aurally rather than visually or even semantically). But the purpose of a digital edition is not to establish the text in this traditional sense—a methodology whereby literary-critical judgments constrain the object of study that is supposed to be the basis on which to form literary-critical judgments! Indeed, the foregoing observations are derived from experiencing “The British Church” and “Evensong” within the presentational rhetoric of the Digital Temple’s parallel-display interface. The new medium, with its limitless “space,” allows the editor to get out of the way.
But while the technology removes certain editorial burdens, it introduces others. Having committed to the TEI’s parallel segmentation protocol as the best way to handle difference among and simultaneous display of the three witnesses, I run into semantic difficulties not immediately apparent. Whereas the project’s user interface nicely represents the formal differences between multiple versions of the same poem, the code base from which it is derived has not in fact overcome the problems associated with difference of this sort. Here, briefly, is the markup structure for “The British Church” (see Figures 7 and 8) beginning with the first (and second) stanza(s). The Bodleian manuscript is represented by #b, 1633 by #p:
<milestone ed="#b" unit="stanza" n="1"/>
<milestone ed="#p" unit="stanza" n="1"/>
<milestone ed="#p" unit="stanza" n="2"/>
<milestone ed="#b" unit="stanza" n="2"/>
<milestone ed="#p" unit="stanza" n="3"/>
<!-- etc. -->
The <lg/> (line group) tag, absent here, typically “contains a group of verse lines functioning as a formal unit, e.g. a stanza, refrain, verse paragraph, etc” (TEI Consortium, “lg”). However, the combination of line-by-line parallel encoding and difference in stanzaic arrangement between the two versions introduces what the text-encoding community calls overlapping or multiple hierarchies. The <milestone/> element overcomes this difficulty by marking points in the text where new stanzas begin. Unlike the <lg/> element, which must have lines nested within its opening and closing tags, <milestone/> is an empty element and may appear at any point in the markup. Without <milestone/>, it would be impossible to mark both the three- and six-line stanzaic divisions of 1633 and B respectively while preserving line-by-line parallel transcription and encoding. So <lg/>’s usual function has been usurped (necessarily) by <milestone/>. But while <milestone/> solves a practical dilemma—the problem of multiple display formats—can it really be said to have replaced <lg/>’s semantic function? Because <lg/> and <l/> (line) are nesting elements, their relationship in the markup is clear: everything within a <lg></lg> frame belongs to, is some aspect of, a unit designated “line group,” whereas the empty <milestone/> has no obvious connection to the lines that follow it. A parser can be (indeed, has been) instructed to read the file as if there were such a connection; but the file itself, speaking the language of TEI-XML, makes no such claim and thereby loses some of its semantic precision. There is also loss in functionality: for whereas the <milestone/> solution facilitates a style sheet transformation that renders visually the desired stanzaic divisions, a parser searching for whole stanzas rather than boundary markers—individual units comprised of multiple lines rather than the interstitial points separating such units—will assume, unless instructed otherwise, that “The British Church,” “Evensong,” and other poems whose versions differ in this way simply do not consist of multiple stanzas. The <milestone/> displacement of <lg/> is perfectly valid and TEI-conformant XML. But is it correct?
Markup clearly is the digital editor’s primary editorial activity. New technologies help to resolve some traditional problems, but they introduce new ones. Some of the foregoing issues are more or less troubling than others—but they are issues, if not with respect to transcription and layout for display purposes, then with respect to semantic precision and accurate encoding for the purposes of data retrieval and analysis. Return once again to W’s “The Altar” (Figure 1). Are there any stanzas here, any repeated units of verse arranged according to metre and rhyme? Well, there are couplets throughout. But beyond that? The middle section might be said to consist of two four-line stanzas. And the pedestal at the bottom is a mirror image of the table thingy at the top: two tetrametre followed by two pentametre lines reversing the pentametre-tetrametre order of the first section or “stanza” or “table thingy.” In a conventional print edition, the stanza issue is not. The editor might simply represent as accurately as possible the spacing and lineation of the source and leave this interpretive question to the reader. But the encoder of digital texts, concerned about semantic accuracy and automated retrieval, must determine what and how to label. And labeling (or tagging or encoding), like transcribing, is reading—an activity fraught with interpretive pitfalls and the exhilaration that entails.
Perhaps now is a good time for Confessions of a Digital Editing Novitiate. My initial decision to edit Herbert electronically stems from an interest in text analysis, in the poems’ intellectual content as explored with the aid of a computer (see Whalen 27-65). My research was limited to the first edition as transcribed from the Chadwyck-Healey database (now Early English Books Online). Substantial differences among the first edition and manuscripts as observed through conventional reading, however, led me to wonder whether it might be worthwhile to do some comparative computer-assisted analysis of all three sources. Having decided, then, to include the manuscripts, and discovered just how labour-intensive the project would be, I began seriously to consider whether my own critical focus is sufficient to justify it. So now I find myself presuming to provide a service to the community of Herbert scholars, as well as an accessible tool to teachers and students of seventeenth-century English verse. The encoding goal is a single source file that combines the several documents and all variants (by using the TEI’s parallel segmentation method), and that includes tags for both original and modern spellings, abbreviations and expansions, deletions and additions, marginal notes, differences in the order of poems among the several sources; and for stanzaic forms, metre, and rhyme. I once assumed that while time-consuming and tedious, the task would be simply a matter of gathering all the materials and transcribing and tagging them—as if I were some content scribe labouring happily under a lofty purpose in my electronic scriptorium. How wrong I was is the (by now obvious) subtext of this essay. Such a discovery can be unpleasant and even paralyzing; or it can be an opportunity to clarify the purposes and advantages of a digital edition.
Accurate and thoroughly encoded transcriptions of the manuscripts and first edition will be susceptible to computer-assisted analysis and thereby (potentially) provide information about Herbert’s verse that would be very difficult if not impossible to obtain without a computer. Having first generated a simple concordance, for example, one can create image clusters or word groupings whose contents are selected according to some semantic commonality. One might then generate lists of co-occurrences, measured for relative frequency, of words from within and/or among these clusters. Such data might tell us something new about, or support (or not) an otherwise conventional analysis of, say, the Bodleian manuscript. The results might then be compared to those derived from a similar analysis of the Williams manuscript to determine whether and in what ways Herbert’s theological itinerary, idiolect, or some other aspect of his poetic practice evolved or changed over time. 
But the text-analysis focus is not justified by the effort to produce a digital edition, nor would such functionality appeal to more than a handful of humanists (for reasons that include valid doubts about the critical legitimacy of electronic text analysis). Access to all the relevant documents in one easily navigable space, however, is of significant value to all Herbert scholars and students of seventeenth-century verse. The ability not only to perform searches or generate data but also to move freely among the transcriptions and images and to view and compare them simultaneously; to have access to high-resolution images of the Williams and Bodleian manuscripts and a copy of the first edition that are a vast improvement over currently available facsimiles; to determine whether a particular edition (including this one) accurately transcribes and represents the texts or sufficiently accounts for variants; to encourage more thorough examination of Herbert as a historically evolving religious thinker: enabling such exploration is the project’s primary goal.
The Digital Temple will also constitute a formidable teaching aid. In addition to providing both original- and modern-spelling texts along with a critical apparatus, annotations, and bibliography, the project will allow undergraduates full access to a domain hitherto reserved for the specialist scholar or graduate student. Instructors will have at their disposal the means to introduce their students to the world of the text: the relationship between manuscript and print cultures at a time when modern notions of the author were being conceived; poems as historically situated, revised, and evolving rather than pristine objects; the strange and largely alien beauty of early modern literature in its (albeit, digitally represented) material form.
One final advantage: access and preservation. David Wykes, Director of Dr. Williams’s Library, London, tells me that the repository frequently receives requests to view MS. Jones B62 for iconic purposes—to hold, admire, and revere the work of a much loved poet. As one himself inclined to venerate such relics—are not electronic editions shrines of a sort?—I share this desire even while recognizing that librarians understandably are nervous about acolytes, however respectful their devotional enthusiasms. And though a digital representation is not the thing itself, it might for the faithful be redolent of a kind of real presence.
Burrows, John. “Textual Analysis.” A Companion to Digital Humanities. Ed. Susan Schreibman, Ray Siemens, and John Unsworth. Oxford: Blackwell, 2004. 323-47. Print.
Charles, Amy M., ed. The Williams Manuscript of George Herbert’s Poems: A Facsimile Reproduction. Delmar, NY: Scholars’ Facsimiles & Reprints, 1977.
Charles, Amy M. and Mario A. Di Cesare, eds. The Bodleian Manuscript of George Herbert’s Poems: A Facsimile of Tanner 307. By George Herbert. New York: Scholars’ Facsimiles & Reprints, 1984. Print.
Cloud, Random. “Enter Reader.” The Editorial Gaze: Mediating Texts in Literature and the Arts. Ed. Paul Eggert and Margaret Sankey. New York: Garland, 1998. 3-50. Print.
───. “fiat flux.” Crisis in Editing: Texts of the English Renaissance. Ed. Randall McLeod. New York: AMS, 1993. 61-172. Print.
Blake, William. The Complete Poetry and Prose of William Blake, rev. ed. Ed. David V. Erdman. New York: Doubleday, 1988. Print.
Grosart, Alexander B., ed. The Temple: Sacred Poems and Private Ejaculations, Being a Facsimile Reprint of the First Edition. By George Herbert. London: Elliot Stock, 1876. Print.
Guibbory, Achsah. Ceremony and Community from Herbert to Milton: Literature, Religion, and Cultural Conflict in Seventeenth-Century England. Cambridge: Cambridge UP, 1998. Print.
Hockey, Susan. Electronic Texts in the Humanities. Oxford: Oxford UP, 2000. Print.
Hutchinson, F.E., ed. The Works of George Herbert. By George Herbert. Oxford: Clarendon P, 1941. Print. Lake, Peter. “Calvinism and the English Church 1570-1635.” Past and Present 114 (1987): 53-66. Print.
───. “Lancelot Andrewes, John Buckeridge and Avant-Garde Conformity at the Court of James I.” The Mental World of the Jacobean Court. Ed. L. Levy Peck. Cambridge: Cambridge UP, 1991. 113-33. Print
Lancashire, Ian. “Empirically Determining Shakespeare's Ideolect.” Shakespeare Studies 25 (1997): 171-85. Print.
───. “Probing Shakespeare’s Idiolect in Troilus and Cressida, 1.3.1–29.” University of Toronto Quarterly 68.3 (1999): 728-67. Print.
McGann, Jerome. “Marking Texts of Many Dimensions.” A Companion to Digital Humanities. Ed. Susan Schreibman, Ray Siemens, and John Unsworth. Oxford: Blackwell, 2004. 198-217. Print.
Milton, Anthony. Catholic and Reformed: The Roman and Protestant Churches in English Protestant Thought 1600-1640. Cambridge: Cambridge UP, 1995. Print.
Smith, Martha Nell. “Electronic Scholarly Editing.” A Companion to Digital Humanities. Ed. Susan Schreibman, Ray Siemens, and John Unsworth. Oxford: Blackwell, 2004. 306-22. Print.
TEI Consortium. “lg.” Text Encoding Initiative P5 Guidelines. TEI Consortium. 1 Feb. 2009. Web. < http://www.tei-c.org/release/doc/tei-p5-doc/en/html/ref-lg.html>.
Tyacke, Nicholas. Anti-Calvinists: The Rise of English Arminianism c.1590-1640. Oxford: Clarendon P, 1987. Print.
Walton, Izaak. “The Life of Mr. George Herbert.” Seventeenth Century Prose and Poetry, 2nd ed. Ed. A.M. Witherspoon and F. J. Warnke. Orlando: Harcourt Brace Jovanovich, 1982. 271-88. Print
Whalen, Robert. “‘How shall I measure out thy bloud’ or ‘Weening is not measure’: TACT, Herbert, and Sacramental Devotion in the Electronic Temple.”Early Modern Literary Studies 5.3 (2000): 7.1-37. Web. <http:// extra.shu.ac.uk/emls/05-3/whalherb.html>. Jointly published in a special issue of Text Technology 9.3 (1999): 27-65. Print.
Wilcox, Helen, ed. The English Poems of George Herbert. By George Herbert. Cambridge UP, 2007. Print.
 See also Cloud's "Enter Reader."
 In Manuscript, Print, and the English Renaissance Lyric (Ithaca: Cornell UP, 1995), Arthur Marotti emphasizes the 1633 publication of Herbert’s Temple and John Donne’s Poems as “a watershed event” in the development of early modern print culture (247). Though his analysis of The Temple’s publication is not nearly as extensive as his attention to Donne, Marotti argues that both authors’ personae were in large part created and subsequently reified by the publication process. Just as the presentation and especially the ordering of Donne’s Poems reflect “the need to protect the reputation of Dean Donne from moral taint” (252), so does The Temple “posthumously glorif[y] its humble author through the print medium” (257). Rightly identifying print culture as shaping the public perception of authors’ identities, Marotti also implies that the real Donne and Herbert lie somewhere beneath the ideological machinery of the publication process. The Digital Temple does not presume to reveal a poet hitherto obscured by some editorial agenda; it recognizes, rather, that we are dealing here with Herberts rather than a single Herbert: the Herbert of the Williams manuscript, of the Bodleian manuscript, of the 1633 edition and the Cambridge printers (Thomas Buck and Roger Daniel), and of Herbert’s modern editors (F.E. Hutchinson, Mario A. DiCesare, and Helen Wilcox). Neither do I mean merely to deny essentialist notions of the author. The Digital Temple, rather, deliberately seeks to record and foreground significant agencies in the production of Herbert’s verse up to and including the editio princeps: Herbert himself, an amanuensis in W, the scribes at Little Gidding, Buck and Daniel (and perhaps Nicholas Ferrar) at the Cambridge press—in short, what Jerome McGann in The Textual Condition (Princeton: Princeton UP, 1991) calls traces of a “social text” (21). Drawing a sharp distinction “between a work’s bibliographical and its linguistic codes” (52), McGann argues that “as the process of textual transmission expands, whether vertically (i.e., over time) or horizontally (in institutional space), the signifying processes of the work become increasingly collaborative and socialized.” This collaborative process includes both readers of texts and those who produce them: authors, amanuenses, printers, publishers, compositors, book designers, etc. One need not embrace Roland Barthes’s mort d’auteur— his Brechtian vision of the author “diminishing like a figurine at the far end of the literary stage” (“The Death of the Author,” Image-Music-Text, trans. Stephen Heath [London: Fontana, 1977], 145)—to acknowledge that Herbert’s poems are not pristine things, but artifacts with a history, an ontogeny that can enhance our understanding of their meaning.
 See Lake, “Lancelot Andrewes” 113-14. Peter Lake defines “avant-garde conformity” as a style of divinity favoring sacrament, ceremony, and discipline over preaching and emphasis on scripture, and which was positively hostile to the predestinarian separation of a godly elect from the ungodly.
 See Guibbory 47. Though otherwise acknowledging the complexity of confessional identity and its bearing on our understanding of “The Altar,” Achsah Guibbory mistakenly attributes this typographic emphasis to Herbert himself.
 The decidedly Calvinist outcome of the 1618 Synod of Dort, where James had sent an anti-Arminian delegation, might have had something to do with the emendation. If so, it suggests a defiant Herbert rather than Walton’s mild and eirenical cliché. See Tyacke 47; Milton 405-06.
 A historian, not a literary scholar, poignantly reminds us of Herbert’s potential in this regard. Anthony Milton concludes the final chapter of his magisterial study of the English church’s vexed confessional identity with a reference to Herbert’s provocative poem, “The British Church”: “Blessed be God, whose love it was / To double-moat thee with his grace, / And none but thee” (ll.28-30, quoted in Milton 528). Milton fails to mention a detail pertinent to his analysis: that “The British Church” is absent from the much earlier Williams manuscript, suggesting that the ecclesiastical developments Milton documents seem to have provoked Herbert toward (or more fully to disclose) this vision of special dispensation.
 Neither do Hutchinson and Wilcox bother to note the Williams elision in their apparatuses (Hutchinson 51, 493; Wilcox 178-79).
 Hutchinson notes the emendation (48); Wilcox does not (160, 166).
 Parallel display in the Digital Temple is handled using a customized iteration of the Versioning Machine 4.0. (As this article goes to press, v4.0 is not yet publicly available, but v3.2 is available at http://v-machine.org.)
 For an example of idiolect studies using text analysis software, see Lancashire “Empirically Determining” and “Probing Shakespeare” For co-occurrence studies and general discussions of computer-assisted text analysis see Hockey 63-84 and Burrows 323-47.